A second RNA-binding protein is essential for ethanol tolerance provided by the bacterial OLE ribonucleoprotein complex

Significance

Although large bacterial noncoding RNAs (ncRNAs) are rare, those whose functions have been experimentally established perform fundamental roles in genetic information transfer, RNA processing, and protein production and localization. OLE (ornate, large, extremophilic) RNAs represent one of the most widespread and well-conserved classes of bacterial ncRNAs whose activities remain unknown. We have identified mutations in an OLE-associated protein (OapA), an essential partner for OLE RNA, which cause more severe inhibition of host bacterial growth under cold or ethanol stress conditions compared with knockout strains lacking OLE or OapA. A genetic screen using a bacterial strain carrying the mutant OapA protein revealed another protein partner that also forms a complex with OLE RNA, and is essential for the biological function of this mysterious bacterial ncRNA.

Abstract

OLE (ornate, large, extremophilic) RNAs comprise a class of structured noncoding RNAs (ncRNAs) found in many extremophilic bacteria species. OLE RNAs constitute one of the longest and most widespread bacterial ncRNA classes whose major biochemical function remains unknown. In the Gram-positive alkaliphile Bacillus halodurans, OLE RNA is abundant, and localizes to the cell membrane by association with the transmembrane OLE-associated protein called OapA (formerly OAP). These characteristics, along with the well-conserved sequence and structural features of OLE RNAs, suggest that the OLE ribonucleoprotein (RNP) complex performs important biological functions. B. halodurans strains lacking OLE RNA (∆ole) or OapA (∆oapA) are less tolerant of cold (20 °C) and short-chain alcohols (e.g., ethanol). Here, we describe the effects of a mutant OapA (called PM1) that more strongly inhibits growth under cold or ethanol stress compared with strains lacking the oapA gene, even when wild-type OapA is present. This dominant-negative effect of PM1 is reversed by mutations that render OLE RNA nonfunctional. This finding demonstrates that the deleterious PM1 phenotype requires an intact RNP complex, and suggests that the complex has one or more additional undiscovered components. A genetic screen uncovered PM1 phenotype suppressor mutations in the ybzG gene, which codes for a putative RNA-binding protein of unknown biological function. We observe that YbzG protein (also called OapB) selectively binds OLE RNA in vitro, whereas a mutant version of the protein is not observed to bind OLE RNA. Thus, YbzG/OapB is an important component of the functional OLE RNP complex in B. halodurans.

Noncoding RNAs (ncRNAs) are involved in numerous, diverse cellular processes in all domains of life (1). In bacteria, bioinformatic search strategies that use sequence and structure conservation have revealed several possible classes of novel large ncRNAs whose functions remain unknown (2–5). Each of these classes is remarkable for its size, complex structural features, and sequence conservation. Moreover, these classes are distinct from all other known RNA classes, suggesting that their biological and biochemical roles are likely to be novel.

Of these, OLE (ornate, large, extremophilic) RNAs (Fig. 1A), named for their ornate structure, large size, and prevalence in extremophilic bacteria, represent one of the longest ncRNA classes to be discovered in bacteria (2). OLE RNAs are typically ∼600 nucleotides long, have highly conserved and complex structures, and are widespread in anaerobic, Gram-positive bacteria. To date, we have identified 795 nonidentical representatives from various species in the Firmicutes phylum. Intriguingly, OLE RNAs represent the most widespread and most conserved bacterial ncRNA class whose function remains unknown.

Fig. 1.

OLE RNA sequence, structure, and genome context. (A) Updated consensus sequence and secondary structure model for OLE RNAs based on 795 unique representatives from bacterial genomic and metagenomic DNA sequences. (B) Conservation of the genomic region carrying the gene for OLE RNA. The most common gene arrangement adjacent to ole is depicted at the top of the bar graph. The colored section of each bar represents the proportion of all representatives that carry the identified gene in this position relative to ole. Gray represents all other genes found at that position.

The ole gene commonly resides in tandem with a gene encoding a transmembrane protein of unknown function (Fig. 1B). This protein was originally named OAP (for “OLE-associated protein”) but hereafter we call this protein OapA for reasons described below. The presence of the oapA gene is exclusive to the organisms that harbor ole (2, 6, 7), suggesting that the function of OapA is intimately tied to the function of OLE RNA. OapA is a 21-kDa protein predicted to have four transmembrane helices containing several conserved amino acid motifs (Fig. 2A). We have previously shown that OapA from the bacterium Bacillus halodurans binds OLE RNA in vitro, and that this RNA localizes to cell membranes in an OapA-dependent fashion (6). Specifically, the protein appears to exhibit at least a 2:1 (OapA:OLE RNA) binding interaction with a K_D of ∼50 nM. The protein requires portions of the P3 and P3.1 stems (Fig. 1A) for maximal binding affinity (6).

Fig. 2.

Analysis of the OLE-associated protein OapA. (A) Amino acid sequence and predicted topology of B. halodurans OapA. The sequence conservation reflects both the conservation of the identity and physicochemical properties of the amino acid at each position (see Materials and Methods for details). Black circles represent 100% conservation of amino acid identity among all 735 unique OapA sequences examined. (B) Location of OapA mutations examined in this study. (C) Growth of each OapA mutant strain after 48 h incubation in media containing 5% ethanol. Each bar represents the average of at least three experiments normalized to WT, and error bars are the SEM.

Because OLE RNAs are found in many species of extremophilic bacteria and localize to the cell membrane, it was previously speculated that the OLE ribonucleoprotein (RNP) complex might play a role in a pathway protecting these species from their extreme environments. Notably, ∆ole, ∆oapA, and ∆ole-oapA strains of B. halodurans have phenotypes consistent with membrane stress. These strains exhibit growth defects when exposed to media containing short-chain alcohols, such as ethanol, or when grown in cold temperatures (7).

Analysis of the transcriptome of B. halodurans revealed that OLE RNA represents one of the most abundant transcripts under normal growth conditions (6, 7). These RNAs are relatively stable for a bacterial transcript, with a half-life approaching 3 h (7). Interestingly, OLE RNA abundance increases fivefold when cells are exposed to ethanol (7). Thus, OLE RNAs are up-regulated under particular cellular stresses, and might possess a function that helps cells respond to these extreme conditions.

To further investigate the biological and biochemical roles of the OLE RNP complex, herein we examined the growth of B. halodurans strains carrying mutations in either OLE RNA or OapA under ethanol or cold stress conditions. These studies uncovered a mutant OapA (called protein mutant 1 or PM1) that causes a more severe version of the same phenotype resulting from OLE RNA or OapA knockouts. Moreover, the strain producing PM1 OapA exhibits a dominant-negative phenotype, wherein cells are exceptionally sensitive to cold and ethanol stresses even when WT OapA is also present. A series of OLE RNA mutations were examined in cells carrying PM1 OapA, and were found to rescue the dominant-negative phenotype. These findings suggest that PM1 OapA requires a functional OLE RNA for its negative effect, and that the OLE RNP complex requires one or more additional components to form its active state.

We conducted a genetic screen with a B. halodurans strain carrying the oapA-pm1 gene to search for additional components of the RNP complex. It was speculated that extragenic suppressors for the extreme cold- and ethanol-sensitivity phenotypes caused by PM1 could accrue in genes that were essential for formation of the complete OLE RNP complex structure. From this genetic screen, a gene for a protein of unknown function, YbzG (AYT26_RS00915), was identified. B. halodurans YbzG was then shown to specifically bind B. halodurans OLE RNA in vitro. These findings demonstrate that the OLE RNP complex requires YbzG to form its active complex. Therefore, we recommend naming the OLE-associated proteins “OapA” (formerly OAP) and “OapB” (YbzG), and will use these names in this report.

Results and Discussion

Updated Sequence and Consensus Model for OLE RNAs.

Only 15 examples of OLE RNA from fully sequenced bacterial genomes were described in the initial report of the discovery of this ncRNA class (2). Subsequently, the consensus sequence and structural model was revised based on ∼200 unique OLE RNA representatives (7). We have now uncovered a total of 795 unique examples, mostly due to the tremendous expansion of bacterial sequence databases. All examples found in genomic DNA sequences from phylogenetically assigned bacterial species (as opposed to metagenomic sequences) are predominantly restricted to the Firmicutes phylum, and are divided among three classes: Bacilli, Clostridia, and Tissierellia. Indeed, all newly found OLE RNAs are in species of bacteria characterized as extremophilic, spore-forming, Gram-positive, and anaerobic (either facultative or obligate).

Even with the addition of newly uncovered sequences, the predicted secondary structure for OLE RNAs reported previously (7) remains extraordinarily well conserved, and includes numerous base-paired substructures and multistem junctions (Fig. 1A). Covariation between nucleotides that results in retention of Watson–Crick base-pairing between nucleotides is used as strong evidence for the formation of stem regions. Strikingly, about half of the nucleotides are conserved in more than 75% of the representatives. This high conservation of nucleotide identities and secondary structure indicate that OLE RNAs form an intricate tertiary structure that is critical for its biochemical function.

Genome Contexts of OLE RNA Genes Between Distal Bacterial Species Are Similar.

The ole gene of B. halodurans is typically found in a large operon (2), although a separate promoter was also identified immediately upstream of the ole gene in this organism (8). This distinctive genetic context is similar even between distantly related bacterial species (Fig. 1B). In over 96% of the organisms that carry OLE RNA, its gene is located directly upstream of oapA, the gene coding for the OLE-associated protein OapA. Typically, these two genes are immediately flanked by the isoprenoid biosynthesis genes ispA (upstream) and dxs (downstream). In many archaeal species, isoprenoid compounds are produced to modulate the fluidity and integrity of cell membranes (9), which is likely also a major challenge for bacteria that live under extreme environmental conditions.

Additional genes, predicted to be involved in DNA repair and other activities, are arranged in a common order on both sides of this core gene arrangement (Fig. 1B). In roughly half of the representatives noted above, ole and oapA are located in a genomic region with the following gene order: nusB, folD, xseA, xseB, ispA, dxs, yqxC, nadF, argR, and recN. However, in the analogous region present in B. halodurans (the organism examined in the present study), nadF is absent. Otherwise, the remaining genes are present in the order listed. Previous reverse-transcription and PCR experiments in B. halodurans revealed that the 11 genes in this cluster indeed can yield a single transcriptional unit (2), despite the activity of the promoter immediately upstream of the ole gene (7, 8).

This common assembly of genes flanking ole and oapA among diverse bacterial species presumably is relevant to the biological function of the RNA and protein products. These flanking genes are involved in several major but distinct cellular processes. As noted above, IspA (geranyltranstransferase) and Dxs (1-deoxy-d-xylulose-5-phosphate synthase) are involved in the biosynthesis of isoprenoids (10). XseA (exoDNase VII large subunit), XseB (exoDNase VII small subunit), and RecN (recombination protein N) function in DNA repair (11, 12). FolD (bifunctional methylenetetrahydrofolate dehydrogenase and methenyltetrahydrofolate cyclohydrolase) and NadF (ATP-dependent NAD kinase) are related to coenzyme metabolism (13, 14). Both NusB (transcription antitermination protein), required for transcription of rRNA genes (15), and YqxC (rRNA methylase), predicted to methylate rRNA due to the presence of S4 and FtsJ domains (16), are related to the production of ribosomes. Finally, ArgR (arginine repressor) down-regulates arginine biosynthesis (17). Unfortunately, at this time we cannot draw a connection between all of these genes to infer a specific biological process that might involve the function of OLE RNA.

Moreover, no hypotheses have been derived from the remaining ∼50% of neighboring genes because they vary widely and do not exhibit a pattern indicative of a common biological process. It is possible that this genomic region carries genes related to diverse processes that all broadly respond to certain stress conditions. However, the levels of mRNAs for genes in the B. halodurans operon do not change substantially during exposure to ethanol stress compared with the increase in OLE RNA abundance, which is detected at five times the level compared with cultures without ethanol (7). Thus, the OLE RNP complex could function independently from the other gene products in this operon, but contribute to the overall stress response through its involvement in a biological process that is different from those influenced by the neighboring genes.

Interestingly, genes rarely associated with ole might still provide valuable clues regarding the function of OLE RNA. For the ∼4% of the ole representatives that reside outside the genomic region described above, oapA still remains between ispA and dxs. In these instances, the displaced ole gene is often located directly upstream of a c-di-AMP riboswitch. Riboswitches for c-di-AMP frequently regulate genes related to cell wall metabolism and the transport of ions and other osmoprotectants (18). Indeed, each c-di-AMP riboswitch adjacent to a displaced ole gene is predicted to regulate a downstream gene encoding an NlpC/P60 family protein. Such proteins are known to function as cell wall hydrolase enzymes (19, 20). This genomic arrangement of displaced ole genes again suggests that the OLE RNP complex might be related to the integrity of the barriers between the cell contents and the environment, namely the functional status of the cell wall or cell membrane.

OapA Mutations Cause Reduced Cell Growth Under Ethanol-Mediated Stresses.

Our updated bioinformatics data also yielded a large increase in the number of nonidentical OapA sequences. We identified 735 unique OapA representatives. The amino acid sequence alignment revealed that approximately half of the positions exhibit at least modest conservation (Fig. 2A). All of the most strongly conserved amino acids reside in the putative transmembrane and intracellular regions, suggesting that the function of OapA likely does not involve interactions with the peptidoglycan matrix of the cell wall or with extracellular molecules.

Six amino acids whose identities are strictly conserved reside in or near transmembrane helices 3 and 4. Each of these invariant residues, which presumably are essential for OapA function, corresponds to known consensus protein motifs, as previously reported (6). Specifically, there are two aspartic acid residues (D100, D104) in a DxxxD motif (21) located in transmembrane helix 3 (Fig. 2A). This sequence (or a similar peptide motif) is present in some enzymes involved in the biosynthesis of certain isoprenoid compounds (21, 22). Also in helix 3, two strictly conserved glycine residues (G103, G111) and another glycine that is not universally conserved (G107) appear to form a GxxxGxxxG motif. Peptide sequences with this periodic “glycine zipper” arrangement are known to promote docking between helices (23). Finally, at the base of helix 4, two invariant lysine residues (K153, K157) are consistent with a KxxxK motif, which has been shown to bind RNA in other proteins (24).

To investigate the functions of these conserved residues in OapA, alanine substitution mutations were made to each key residue (SI Appendix, Table S1). Specifically, three mutant B. halodurans oapA constructs were generated that encode translation products named PM1, PM2, and PM3. PM1 represents an OapA protein that carries D100A/D104A mutations in the DxxxD motif. Similarly, the OapA construct PM2 carries G103A/G107A/G111A mutations in the GxxxGxxxG motif, and the OapA construct PM3 carries K153A/K157A mutations in the KxxxK motif (Fig. 2B and SI Appendix, Table S1). Plasmid constructs carrying an unaltered ole gene plus one of the mutant versions of the oapA gene were transformed into a B. halodurans strain whose genome lacks ole and oapA (∆ole-oapA). All OapA mutant strains exhibit reduced growth in the presence of 5% ethanol compared with WT B. halodurans cells (Fig. 2C). This ethanol-dependent reduction in cell growth caused by OapA mutations is mostly analogous to that observed for B. halodurans strains wherein one or two components of the OLE RNP complex have been deleted (∆ole, ∆oapA, and ∆ole-oapA) (7). For example, strains lacking genomic ole and oapA genes, but carrying WT ole and either PM2 or PM3 oapA genes on a plasmid exhibit reductions in cell growth in 5% ethanol (Fig. 2C). However, growth of the PM2 and PM3 strains is not readily classified as representative of the strain carrying only the genomic ∆ole-oapA deletion or the same strain rescued by delivery of WT ole and oapA genes on a plasmid. Thus, we cannot yet determine if the strictly conserved amino acids in the GxxxGxxxG and KxxxK motifs are critically relevant for ethanol resistance.

Most importantly, the PM1 strain grew far worse than the knockout or other OapA mutant strains upon exposure to ethanol (Fig. 2C). This same trend revealing the special sensitivity of the PM1 strain to ethanol exposure is also observed when the number of viable cells is quantitated by colony forming units (CFU). Plots of the CFU μL⁻¹ values (SI Appendix, Fig. S1A) for WT cells and most other Δole-oapA strains remain relatively constant, despite the fact that their cultures increase in OD₆₀₀ (Fig. 2C). This suggests that cell division for these strains is counterbalanced by rapid cell death in 5% ethanol. However, more than two orders-of-magnitude difference in the CFU density curve between the PM1 strain and all other strains is evident after 48-h incubation. These observations indicate that, for the PM1 strain, the production of new viable cells cannot offset ethanol-triggered cell death, which results in a rapid decline in CFU μL⁻¹ values. However, we have not yet determined if ethanol stress decreases the rate of cell replication, increases the rate of cell death, or both.

The strong deleterious effect caused by PM1 indicates that D100 and D104 of the DxxxD motif are essential for the mechanism of the OLE RNP complex in a manner that is distinct from the PM2 and PM3 mutations, and thus also distinct even from deletion of the entire oapA gene. This unusual effect is also observed when cells are cultured in the presence of other short-chain alcohols (SI Appendix, Fig. S1B). However, the PM1 strain described above grows normally when the culture medium lacks added alcohols (SI Appendix, Fig. S1C).

The PM1 Strain Exhibits a Dominant-Negative Sensitivity to Ethanol and Cold.

Strikingly, the PM1 construct yields a dominant-negative trait, whereby the mutant protein causes a deleterious effect on cells in the presence of ethanol even when they carry a normal copy of the oapA gene. Specifically, WT B. halodurans cells that carry natural ole and oapA genes plus a plasmid carrying both the unaltered ole gene and a mutated oapA-pm1 gene are adversely affected by 5% ethanol nearly as strongly as B. halodurans PM1 cells that lack the WT oapA gene (Fig. 2C). In these cells, both the WT and the PM1 versions of OapA should be present, but the WT protein is not able to overcome the deleterious effects of PM1. Similar results are observed when these cells are cultured at reduced temperature (20 °C) (SI Appendix, Fig. S2), which is another stress that was previously known to more strongly affect the growth of ∆ole, ∆oap, and ∆ole-oapA cells (7). However, the strength of the deleterious PM1-mediated phenotype at 20 °C is somewhat diminished in the presence of WT OapA compared with the results with ethanol toxicity.

These surprising findings suggest that the amino acid changes in PM1 might cause OapA to have a deleterious interaction with another member of the pathway in which the OLE RNP complex functions. Although the severe PM1 phenotype could be due to a problem that is unrelated to the function of the RNP complex, this is unlikely to be the case based on our additional findings. First, neither WT cells carrying the PM1 construct nor ∆ole-oapA cells carrying the PM1 construct exhibit growth defects when grown under normal culture conditions (SI Appendix, Fig. S1B). This finding suggests that PM1 does not simply interfere with cell growth. Second, the same stress conditions that are necessary to inhibit growth of mutants expressing defective OLE RNA or OapA protein are required for the strong lethal phenotype to become apparent. Indeed, PM1 appears to cause the same ethanol- and cold-sensitivity phenotypes observed with ∆ole-oapA cells to become much worse.

These observations suggest that the strong distress caused by PM1 might be due to the disruption of a factor associated with the OapA component of the RNP complex. Moreover, this hypothetical OapA partner also most likely functions in the same biological pathway to yield stronger versions of the same phenotypes observed for the ∆ole-oapA strain. Perhaps the normal status of this factor yields partial resistance to stresses, such as ethanol and cold, whereas its disruption by the defective PM1 OapA causes a more severe response to these same stresses.

Mutations to OLE RNA also Can Reduce Cell Growth Under Ethanol Stress.

To further evaluate the strong sensitivity phenotype caused by the PM1 OapA construct, we sought to establish the importance of the OLE RNA to this effect. A series of OLE RNA mutants were constructed (SI Appendix, Table S1) and evaluated in bacterial strains carrying either the WT or PM1 OapA constructs.

One RNA region we examined is the base-paired substructures P12 through P15 (Fig. 3A, Upper). The OLE RNA mutant 1 (RM1) construct includes two mutations that disrupt base pairs within the P14 stem, whereas the RM2 construct carries the same mutations as RM1, but also includes two additional mutations that completely restore P14 stem base-pairing. B. halodurans ∆ole-oapA cells transformed with a plasmid either carrying the RM1 or RM2 ole gene variant plus a WT oapA gene were tested for growth in the presence of 5% ethanol (Fig. 3B). The results reveal that destabilization of the P14 stem in the RM1 construct yields an OLE RNA sequence that fails to improve growth of cells lacking a WT OLE RNA. In comparison, the RM2 construct that carries compensatory mutations to restore P14 base-pairing improves growth to a level similar to the WT rescue strain (Fig. 2C). These results demonstrate that even the subtle disruption of an OLE RNA substructure in the RM1 construct likely causes a loss of the biological function of the RNP complex.

Fig. 3.

Disruptive OLE RNA mutations preclude the function of the OLE RNP complex and overcome the strongly deleterious phenotype caused by PM1. (A) Sequence and secondary structure of B. halodurans OLE RNA regions that carry mutations examined in this study. (B) Growth of B. halodurans strains carrying WT OapA with OLE RNA mutations as annotated, in media containing 5% ethanol and incubated for 48 h. (C) Growth of strains with PM1 OapA and OLE RNA mutations as annotated, in media containing 5% ethanol and incubated for 48 h. Growth data are normalized to the analogous data for WT cells depicted in B. Additional details for B and C are as described for Fig. 2C.

Similar results are observed with additional OLE RNA mutant constructs. The RM3 construct carries mutations converting a GNRA tetraloop to a UNCG tetraloop (Fig. 3A). These are types of RNA loop families that are commonly found in natural structured ncRNAs (25, 26). However, OLE RNA function appears to require the presence of this GNRA tetraloop because the RM3 strain has a phenotype that is indistinguishable from an OLE RNA deletion strain (Fig. 3B). Another RNA construct, RM4, was prepared to examine the effects of two mutations that reside in highly conserved regions encompassed by the P2 through P5 substructures (Fig. 3A, Lower), which are known to be important for OapA binding (6). The nucleotide changes carried by RM4 should diminish binding between OLE RNA and OapA, thereby preventing RNP complex formation. Again, B. halodurans ∆ole-oapA cells carrying the RM4 construct in place of WT OAP RNA do not show improved growth in the presence of 5% ethanol (Fig. 3B), suggesting that these mutations indeed disrupt proper formation and function of the RNP complex. However, we have not measured the effects of the mutations on the levels of these OLE RNA mutants in cells. Thus, the loss of OLE RNA, rather than a disruption of its function, cannot yet be ruled out.

The PM1 Phenotype Requires a Structurally Intact OLE RNA.

To determine whether the strongly deleterious effect of PM1 is dependent on functional versions of its OLE RNA partner, we examined the growth of the B. halodurans ∆ole-oapA strains carrying PM1 and the OLE RNA mutants RM1 through RM4 (SI Appendix, Table S1) in the presence of ethanol. When OLE RNA is not present, cells carrying the PM1 construct exhibit growth that is substantially better than when OLE RNA is present (Fig. 3C). Similar results are observed when OLE RNA is supplied in the form of the mutants RM1, RM3, and RM4. Only RM2, which carries mutations that retain the original OLE RNA structure, permits strong ethanol toxicity of PM1 to persist (Fig. 3C). These results are consistent with our hypothesis that PM1 OapA retains its interaction with OLE RNA, and that this interaction is required for the dominant-negative phenotype.

A Genetic Screen Reveals Another Protein Necessary for OLE RNP Complex Function.

We recognized that two characteristics of the OLE RNP complex provide a strong basis for conducting a genetic screen. First, B. halodurans strains carrying PM1 OapA can be exploited to identify genetic mutants that relieve the strong inhibitory effects of ethanol. Second, the disruption of OLE RNA is one mechanism to overcome the PM1 phenotype, and so we reasoned that a genetic screen might reveal additional partners necessary for the function of the OLE RNP complex. Thus, suppressor mutations might occur in genes in the biochemical pathway in which the OLE RNP complex participates, or they might occur in genes that code for additional protein or RNA factors that directly interact with the complex.

Initial attempts using the B. halodurans ∆ole-oapA strain transformed with a plasmid carrying ole and the oapA-pm1 genes (∼10⁸ cells) failed to yield spontaneously occurring suppressor mutants under either ethanol- or cold-stress conditions. Consequently, cells were first treated with ethyl methanesulfonate (EMS) to introduce mutations into the starting population (Materials and Methods). After EMS treatment, cells were cultured at 20 °C and growth was observed after 1 wk of incubation. Single colonies were isolated, verified to grow at 20 °C, and plasmids from each strain were sequenced to confirm that the transcriptional promoter and the original sequences for the ole and oapA-pm1 genes remained unaltered.

The genomic DNA from each of 14 verified strains was isolated and sequenced. Ten unique genomes emerged from this effort, along with four isolates that were duplicates. To confirm that these 10 strains carried suppressor mutations that broadly overcome the PM1 phenotypic defect, each was successively tested for growth at 20 °C and in the presence of 5% ethanol (SI Appendix, Fig. S3). Each strain exhibits growth under both stress conditions, but yields cell densities that more closely approximate those obtained with the ∆ole, ∆oap, and ∆ole-oapA strains, rather than with WT cells. These phenotypes suggest that the genomic mutations acquired in the genetic screen likely disrupt the RNP complex to overcome the more severe deleterious effects caused by the PM1 mutation.

Analyses of the DNA sequences of the 10 suppressor strains revealed several unique mutations per genome (SI Appendix, Table S2). However, only one gene, ybzG (annotated as AYT26_RS00915 or BH0157), was mutated in multiple strains (Fig. 4A). The ybzG gene encodes a protein of unknown function that contains a KOW RNA-binding motif (conserved domain cd06088) commonly found in ribosomal RNA-binding proteins and transcription factors (27). Specifically, 6 of the 10 strains carried four distinct mutations to ybzG, all of which occur in highly conserved regions of the gene. Two of these mutations (G42V and H57Y) reside at the only two invariant amino acids in the natural protein (Fig. 4A). One of the two remaining mutations (M1I) occurs in the start codon, which presumably disrupts protein translation. The ybzG gene appears to be nonessential in Bacillus subtilis (28), a species related to B. halodurans. Therefore, it is possible that this mutation at the start codon acts as a knockout. The final mutation results in substitution of a highly conserved amino acid (G19S) located immediately adjacent to amino acids of the KOW motif. This mutation, like the others, probably causes substantial disruption of the protein’s natural function.

Fig. 4.

YbzG/OapB binds OLE RNA. (A) The YbzG/OapB protein sequence from B. halodurans (WP_010896340.1) is depicted with annotations denoting sequence and structural features. Amino acid conservation is indicated with colored shading as described for Fig. 2A: black, invariant; magenta, high; cyan, moderate; yellow, low. Underlined residues identify amino acids forming the putative KOW motif (26). Mutations identified in the genetic screen (SI Appendix, Table S2) are colored red and reside directly above the altered amino acid of the WT protein. Numbers in parentheses identify the strains carrying the mutation indicated. (B) Sequence and secondary structure of OLE_449–608 from B. halodurans, which serves as a truncated RNA binding target for YbzG/OapB. (C) EMSA results using 5′ ³²P-labeled full-length (OLE_1–637) or truncated (OLE_449–608) OLE RNA with purified WT YbzG/OapB. (D) Plot of the fraction of RNA bound by protein versus the concentration (c) of YbzG/OapB. Fraction bound values are derived from the EMSA results in C, and the K_D value is an estimate based on visual inspection of the plot. Additional experiments are necessary to precisely determine K_D values and the stoichiometry between the RNA and protein components. (E) EMSA results using 5′ ³²P-labeled OLE_1–637 with mutant H57Y YbzG/OapB.

It is unclear how the remaining four strains confer resistance against cold and ethanol through the mutations in their genomes. Other than ybzG, there were no other genes that were identified as mutated in more than one strain. Additional screening efforts, currently underway, should reveal the mutation targets in these other strains that allow cells to overcome the PM1 phenotype.

YbzG Selectively Binds OLE RNA.

Because YbzG is a protein of unproven function, the mechanism by which it could participate in the biological pathway related OLE RNP complex was not certain. However, we hypothesized that the protein might directly interact with OLE RNA due to the presence of the putative RNA-binding motif (27). To evaluate the RNA-binding ability of YbzG, EMSAs were performed with the full-length OLE RNA construct (OLE_1–637) and various truncated OLE RNA constructs (Fig. 4B and SI Appendix, Fig. S4).

Indeed, WT YbzG binds full-length OLE RNA and a truncated transcript (OLE_449–608) spanning regions P12 through P15 (Fig. 4 B and C). In contrast, two other OLE RNA segments that exclude these regions fail to be bound by YbzG. Similarly, YbzG does not bind unrelated RNAs from B. halodurans, including a transfer-messenger RNA or an adenosylcobalamin riboswitch (SI Appendix, Fig. S5). These results demonstrate that YbzG functions as an RNA-binding protein and that its binding site is likely to be selective for its target sequence and structure.

When the fraction of bound RNA versus free RNA was plotted against the protein concentration, we observed an apparent dissociation constant (K_D) of no poorer than 60 nM (Fig. 4D). This K_D value is sufficient for YbzG to form a stable interaction with OLE RNA even if the two molecular species are not abundant in cells. Moreover, this K_D is similar to that observed for the OapA protein (6). Therefore, we speculate that YbzG also might be an important component of a functional OLE-OapA-YbzG RNP complex, although we have not examined the system for simultaneous binding by both proteins.

Importantly, we observed that a YbzG variant isolated in our genetic screen, wherein the histidine at position 57 has been mutated to a tyrosine (H57Y) (Fig. 4A), does not bind OLE RNA under the conditions tested (Fig. 4E). The H57Y mutant was chosen for examination because it appears in two strains from our genetic screen and it alters an invariant residue in the KOW RNA-binding motif. This suggests that it is the failure of the H57Y YbzG variant to bind OLE RNA that yields the suppressor phenotype in the B. halodurans PM1 strain. When YbzG is unable to bind, it is equivalent to the loss of functional OLE RNA. These findings are consistent with our hypothesis that YbzG is a necessary partner for the function of the RNP complex.

Given that OLE RNA is the only known natural RNA target for YbzG, and that this interaction has a biological purpose, we propose renaming the ybzG gene “oapB” and its protein product OapB. We have detected the ybzG/oapB gene in all species that contain ole. However, it is also present in additional organisms in the Firmicutes phylum that do not carry the ole gene. Thus, it seems certain that OLE RNA is not the only natural binding partner for YbzG/OapB. Efforts to further define the RNA binding site for this identified protein partner might aid in identifying additional targets that have sequence or structural homology with its binding site in OLE RNA.

Concluding Remarks

The identification of PM1 OapA as a dominant-negative mutation that produces strong ethanol- and cold-sensitivity phenotypes is consistent with the hypothesis that the OLE RNP complex has a role in responding to these stresses. The PM1 phenotype is more severe than the level of sensitivity caused by complete elimination of OLE RNA, OapA, or both. The data indicate that these components of the RNP complex likely only partially contribute to alleviating these stresses. One possible explanation for our findings is that OapA has a hypothetical, and yet unknown, partner that also contributes to ethanol- and cold-stress responses (Fig. 5). In this scenario, PM1 OapA interferes with the status of this unknown factor, which causes a phenotype worse than that observed when oapA is deleted. This factor could be another protein partner, a second RNA molecule, or perhaps a small signaling molecule. However, other explanations for the current observations are also possible.

Fig. 5.

Schematic representation of the RNP complex formed by OLE RNAs. Depicted are the known components of the RNP complex drawn to scale based on molecular mass, except for the hypothetical involvement of an additional factor. An OapA dimer is embedded in a lipid bilayer and in contact with OLE RNA in a 2:1 complex (6). Notable features of each OapA protein include the DxxxD motif (site of the PM1 mutations), GxxxGxxxG motif or glycine zipper (G zipper), and a region enriched with positive-charged amino acids. Although one YbzG/OapB molecule is depicted, the true stoichiometry remains to be determined. We predict that an additional factor or factors interact with OapA and/or OLE RNA, but the nature, size, stoichiometry, and physical locations of any additional factors remain unknown.

Notably, the amino acid changes creating PM1 reside in the DxxxD motif that overlaps part of the GxxxGxxxG motif (Fig. 3B). Previously, it was proposed that OapA interacts with OLE RNA as a protein homodimer (6). If G103, G107, and G111 comprise a helix-docking motif, then a helix–helix interaction might occur between two OapA molecules. Alternatively, this GxxxGxxxG motif might be used to interact with another protein. Regardless, the mutations to the DxxxD motif in PM1 likely do not disrupt proper formation of the OLE–OapA interaction, or the strong deleterious phenotype for PM1 would not be observed.

To search for additional molecules that interact with OapA and OLE RNA, we conducted a genetic screen for mutations in the bacterial genome that could overcome the severe PM1 phenotype. This screen yielded another protein partner of the OLE RNP complex that appears to interact independently with OLE RNA. This mutated protein, YbzG/OapB, might alleviate the PM1 phenotype only by interfering with OLE RNA folding and function, rather than serving as the critical factor that causes the dominant-negative effect. We know this is a possible mechanism for overcoming the PM1 phenotype because the mutation or deletion of OLE RNA eliminates this severe PM1 effect. Perhaps YbzG/OapB simply serves as an RNA folding chaperone that helps OLE RNA adopt its active conformation as the RNP complex assembles.

The genetic screen generated four additional strains that exhibit reduced sensitivity to both ethanol and cold, but that lack changes to the ybzG/oapB gene. This finding suggests that there are other genetic loci whose mutation can affect the function of the OLE RNP complex, or contribute in some other manner to its biological pathway. However, each strain carries about a dozen mutations, and without multiple strains that carry mutations in the same gene, we currently do not know which of these individual mutations is necessary for overcoming the PM1 phenotype.

It seems unlikely that the dominant-negative phenotype is caused by PM1 OapA interfering with the function of YbzG/OapB. Mutant strain #8 from the genetic screen (Fig. 4A) lacks a start codon for the ybzG gene, which presumably precludes translation. If nonfunctional YbzG/OapB is the cause of the severe ethanol toxicity phenotype, then elimination of YbzG/OapB should also yield the same severe phenotype. Thus, we are continuing the genetic screening effort to identify additional factors related to the function of the OLE RNP complex to determine the cause of the dominant-negative phenotype. Given the large size of OLE RNA relative to its protein binding partners OapA and YbzG/OapB (Fig. 5), there are plenty of available conserved sequences and substructures for additional factors to bind the RNA. Biochemical methods to purify cell-derived OLE RNA or its protein factors by affinity tagging might also be an effective route to identifying additional factors of the RNP complex. If such additional binding partners for OLE RNA have known biochemical or biological functions, their identification might aid in assigning a function to this widespread and well-conserved ncRNA.

Materials and Methods

Bioinformatic Analyses.

Recently uncovered OLE RNA representatives were identified based on homology searches and sequence alignments with Infernal v1.1 (29) using the previous alignment of ∼200 OLE RNAs (7) as a seed file. Sequences were compiled using the bacterial and archaeal section of RefSEq v76 (30), and environmental DNA sequences as previously described (31). After manual removal of duplicate and truncated sequences, the resulting alignment of 795 unique OLE RNA sequences was used to establish a revised consensus sequence and structural model by exploiting nucleotide proportions and covariation data using the R2R computer algorithm (32).

A collection of OapA sequences was compiled from RefSeq v76 and environmental DNA sequences. Additional OapA sequences were identified using BLAST (33). Alignment was completed using Clustal Omega (34) and manually edited to remove duplicate and truncated sequences. Amino acid conservation was determined with JalView (35) using a scoring algorithm wherein residue identity is the highest contributor; consideration is also given when amino acid changes are within the same physicochemical class. Transmembrane topology was predicted using the TMHMM Server (36).

YbzG/OapB protein sequence representatives were obtained using BLAST, aligned with Clustal Omega, and analyzed using JalView as described for OapA.

Bacterial Strains, Plasmids, and Cultures.

B. halodurans C-125 was purchased from the ATCC (Catalog #BAA-125). The pHCMC05 plasmid was obtained from The Bacillus Genetic Stock Center (The Ohio State University). Specific OLE RNA and OapA mutants were generated by site-directed mutagenesis of the WT pHCMC05::ole-oapA plasmid described previously (7) using the QuikChange II XL kit (Agilent) according to the manufacturer’s instructions. Primers are listed in SI Appendix, Table S3. Transformation of plasmids into B. halodurans was completed following established protocols (7, 37).

Unless otherwise specified, B. halodurans was grown in LB broth (USB Corp.) that was prepared at 90% volume, autoclaved, and adjusted to full volume and pH 10 with 10% (wt/vol) filter-sterilized Na₂CO₃ [1% (wt/vol) final concentration]. Media was solidified with 1.5% (wt/vol) agar. Unless otherwise indicated, all cells were grown aerobically at 37 °C and shaken at 200 rpm.

Bacterial Growth Assays.

Experiments assessing the growth of B. halodurans strains in the presence of ethanol or at 20 °C were conducted in 14-mL round-bottom Falcon culture tubes using 3-mL culture volumes. B. halodurans cells in exponential growth phase were used for inoculation, according to the method previously described (7). Cells were first grown overnight in 3 mL of LB (pH 10) with 3 µg mL⁻¹ chloramphenicol and 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). These cultures were then diluted to 0.01 OD₆₀₀ in fresh media and incubated for 3 h. Then, for ethanol stress, cells were diluted again to 0.01 OD₆₀₀ in media containing a final concentration of 5% ethanol (vol/vol). Tube caps were sealed and the cultures shaken at 37 °C for the indicated times. For cold stress, cells were diluted to 0.01 OD₆₀₀ in fresh medium, then shaken at 20 °C for the indicated times. At each time point, OD₆₀₀ values were determined using a Cary 60 UV-Vis Spectrophotometer (Agilent Technologies).

Genetic Screen for Suppressor Mutations.

Mutagenesis was conducted as described previously (38). In brief, B. halodurans ∆ole-oapA cells carrying a plasmid with ole and oapA-pm1 genes were cultured to exponential phase (OD₆₀₀ = 0.4–0.8) and EMS was added to a final concentration of 1% (vol/vol). Cultures were incubated at 37 °C with shaking for 2 h. Cells were then pelleted and washed with Davis Salt [0.7% K₂HPO₄, 0.3% KH₂PO_4, 0.1% (NH₄)₂SO₄, 0.005% MgSO₄•7H₂O], then resuspended in LB (pH 10) and incubated for 3 h at 37 °C for recovery.

For selection, EMS-treated cells were cultured at 20 °C for 7 d. Cells grown from the mutagenized parental strain were collected and grown in separate cultures either at 20 °C or in the presence of 5% ethanol to confirm the stability of the suppressor phenotype. Once confirmed, these cells were plated and single colonies were isolated. Each single colony was tested for the ability to grow at 20 °C and in the presence of 5% ethanol as described above. If confirmed, plasmids were extracted from these strains and sequenced to confirm that the promoter region and ole and oapA-pm1 gene inserts were not mutated. Genomic DNA was isolated from each confirmed strain using the ZR Fungal/Bacterial DNA Miniprep kit (Zymo Research) according to the manufacturer’s instructions. The DNA was further purified with the gDNA Clean & Concentrator kit (Zymo Research) according to the manufacturer’s instructions. Whole-genome sequencing with single-nucleotide mutation detection was completed with data analysis by Genewiz, Inc.

Overexpression and Purification of YbzG/OapB.

Constructs for expression of wild-type YbzG/OapB and mutant YbzG/OapB-H57Y proteins were designed with N-terminus hexahistidine tags with Factor Xa protease cleavage sites. Inserts of His-Xa-ybzG/oapB (WT) (NC_002570.2/180746–181054) and His-Xa-ybzG/oapB (H57Y) were synthesized and cloned into a pMK vector and transformed into Escherichia coli K12 OmniMAX 2 T1R cells by GeneArt (Thermo Fisher). The insert was amplified by PCR using Phusion high-fidelity polymerase (New England BioLabs) and standard M13 primers. The PCR product was purified using the Qiagen PCR purification kit, verified by sequencing, digested with NhaI-HF and BamHI-HF to obtain the ybzG/oapB inserts, and subcloned into a pET11a vector. Ligation was conducted using the Quick Ligation Kit (New England BioLabs) according to the manufacturer’s instructions.

BL21(DE3) Chemically Competent E. coli (New England BioLabs) were transformed according to the manufacturer’s instructions with the pET11a plasmids containing the WT or H57Y His-Xa-ybzG/oapB constructs. Transformations were confirmed by colony PCR with primers flanking the multiple cloning sites, and subsequently verified by DNA sequencing.

For protein expression, each strain was plated, and a single colony was used to inoculate a 10-mL culture of LB with 100 μg mL⁻¹ ampicillin. Cultures were incubated overnight with shaking at 37 °C. The resulting culture was used to inoculate 1 L of LB with 100 μg mL⁻¹ ampicillin. The culture was incubated with shaking at 37 °C until the OD₆₀₀ reached 0.4–0.6. Protein expression was then induced by addition of IPTG to a final concentration of 1 mM. After growth at 37 °C for an additional 3 h, cells were harvested by centrifugation for 15 min at 5,000 × g and 4 °C. Cells were resuspended in 50-mL cell wash buffer [20 mM Hepes (pH 7.5 at 23 °C), 150 mM NaCl], and harvested by centrifugation for 15 min at 5,000 × g and 4 °C. Each cell pellet was resuspended in ∼5 mL of cell lysis buffer [20 mM Hepes (pH 7.5 at 23 °C), 300 mM NaCl, 4 mM β-mercaptoethanol, EDTA-free protease inhibitors (1 tablet per 50 mL; Thermo Fisher)] per gram of cell pellet. Cells, kept on ice, were lysed by sonication. Nonlysed cells and debris were pelleted by centrifugation for 30 min at 15,000 × g and 4 °C. The supernatant was transferred to a clean tube chilled on ice.

For purification, HisPur Ni-NTA Resin (Thermo Fisher) was first equilibrated with equilibration buffer (20 mM sodium phosphate, 300 mM NaCl, pH 7.8 at 23 °C) at 25 °C. The cell lysate was loaded onto the column and run through twice. The column was washed with buffer (20 mM sodium phosphate, 300 mM NaCl, 25 mM imidazole, pH 7.8 at 23 °C) until signal of the eluant, measured with Bradford reagent, was baseline. His-tagged constructs were eluted with buffer containing imidazole (20 mM sodium phosphate, 300 mM NaCl, 250 mM imidazole, pH 7.8 at 23 °C). The protein was dialyzed in a Slide-A-Lyzer Dialysis Cassette (Thermo Fisher) with a 10-kDa molecular mass cut-off against the equilibration buffer to remove imidazole. Purified protein was concentrated using an Amicon Ultra-15 Centrifugal Filter Unit (EMD Millipore) with a 10-kDa molecular mass cut-off. Samples were aliquoted to prevent more than one freeze–thaw cycle when used. Aliquots were flash-frozen with liquid nitrogen and stored at −80 °C. The purity of YbzG/OapB was assessed by SDS/PAGE with Coomassie staining, and only one band was detected. Protein concentrations were determined using a standard Bradford assay.

Binding Assays with YbzG/OapB Proteins and OLE RNA Constructs.

The DNA template for the full-length OLE RNA (OLE_1–637) was amplified from B. halodurans genomic DNA with incorporation of a T7 promoter using primers BhOLE-F and BhOLE-R (SI Appendix, Table S3). The OLE RNA truncation (OLE_449–608) template was created similarly using primers BhOLE449-F and BhOLE608-R (SI Appendix, Table S3). RNAs were generated by in vitro transcription reactions, purified, and 5′ ³²P-labeled as previously described (6). All other RNA constructs tested were prepared in the same fashion.

Protein samples were initially diluted in dialysis buffer to 10 µM to prepare a stock solution. A series of diluted protein samples was then prepared, where each dilution was fivefold more concentrated than the protein concentration in the respective final binding reaction. Each 10-μL binding reaction contained 2.5 μL protein solution, 2 μL 5ˊ ³²P-labeled RNA (∼10 nM), 5 μL 2× binding buffer [40 mM Hepes (pH 7.5 at 23 °C), 60 mM MgCl₂, 160 mM KCl, 10% glycerol, 0.6 mg mL⁻¹ E. coli tRNA (Sigma-Aldrich), 2 U μL⁻¹ SUPERase-In RNase Inhibitor (Thermo Fisher)], and deionized H₂O. Reactions were incubated for 5 min at 23 °C, at which point the unbound and bound products were directly separated by nondenaturing 6% PAGE. Nondenaturing PAGE was prepared with 19:1 acrylamide:bisacrylamide (Bio-Rad), 90 mM Tris-base, 90 mM boric acid, 5 mM Mg(OAc)₂, pH 8 at 23 °C (6). Electrophoresis was conducted at 4 °C using 40 W for various times depending on the length of the RNA (6.5 h for full-length OLE RNA, 3 h for truncations). The resulting gels were imaged and quantified using a Typhoon PhosphorImager (GE Healthcare). The fraction of bound RNA was derived from the ratios of band intensities using ImageQuant software (GE Healthcare). The values were plotted versus the logarithm of the protein concentration. The apparent dissociation constant (K_D) value was determined using a sigmoidal dose-response equation in GraphPad Prism 7.