C. burnetii nine Mile phase II, clone 4 (strain RSA 439) were grown in acidified citrate cysteine medium 2 (ACCM-2; Omsland et al., 2011) using 0.2-μm-pore size filter-capped 500-ml Erlenmeyer flasks containing 80 ml of medium. C. burnetii was inoculated at a concentration of 1.6 × 10⁴ genome equivalents per ml (GE/ml) quantified by quantitative PCR (qPCR) with a primer set specific to the C. burnetii rpoS gene (Table S1; Coleman et al., 2004). Cultures were incubated at 37°C in a tri-gas incubator (2.5% O₂, 5% CO₂, 92.5% N₂) with continuous shaking at 75 RPM (Omsland et al., 2011). Following 1 week of growth in the tri-gas incubator, flasks were capped and moved to room temperature for 21 d for generation of SCVs, as previously described (Sandoz et al., 2014).

The C. burnetii RNase III gene (rnc, CBU_1503; Primers: CbuRNaseIII_F and _R; Table S1) was cloned and the encoded, recombinant protein purified as previously described for the RNA helicase (rhlE) gene (Hicks et al., 2011). An RNase III assay substrate was prepared by T7 promoter-mediated in vitro transcription of a PCR product (from primers IVSflank+T7_F and IVSflank_R; Table S1) using a MAXIscript kit (Ambion), as instructed. The resulting RNA was electrophoresed in a 4% acrylamide (w/v)—8M urea gel, and the RNA was excised, eluted overnight at 37°C into a buffer [0.5 M ammonium acetate, 1 mM EDTA, and 0.2% sodium dodecyl sulfate (SDS) at pH 7] and then precipitated with 100% ethanol. RNase III assays were performed for 30 min at 37°C using 200 nM substrate RNA in RNase assay buffer (50 mM NaCl, 10 mM Tris, pH 7.9, 10 mM MgCl₂, 1 mM DTT, 450 mM KCl) in the presence of recombinant C. burnetii RNase III (30 μM). Positive controls were done using E. coli RNase III (Ambion) as per manufacturer's instructions. Products of the RNase III assays were analyzed on 4% acrylamide (w/v)—8M urea gels.

Termini of the IVS RNA were determined by 5′ and 3′ rapid amplification of cDNA ends (RACE) using a SMARTer RACE 5′/3′ kit, as instructed (Clontech). RNase III assay products were analyzed on 4% acrylamide (w/v)—8M urea gels from which IVS was purified, before resuspending in RNase-free water. Prior to 3′ RACE, IVS RNA was poly-adenylated at the 3′ end with E. coli Poly(A) polymerase, as instructed (NEB). cDNA was generated from IVS RNA before amplification by PCR using gene-specific primers (Table S1; IVS_RACE_GSP1, IVS_RACE_GSP2) and universal primers (provided with the kit). Further, a second nested PCR was performed on the 5′ RACE primary PCR product using IVS_RACE_NGSP1 (Table S1) and a universal short primer (provided with the kit) for increased resolution. PCR products were visualized on 2% agarose (w/v) gels, purified with a PCR clean-up and gel extraction kit as instructed (Clontech) and cloned into pRACE with an In-Fusion HD Cloning kit per manufacturer's instructions (Clontech). The plasmid content of individual clones was analyzed by automated Sanger sequencing at the Genomics Core Facility at the University of Montana.

Translation of the IVS ORF encoding S23p was performed using an E. coli S30 Extract System for Linear Templates, as instructed by the manufacturer (Promega). Linear DNA templates were prepared by ligating a PCR-amplified tac promoter (Ptac) to the experimentally-determined 5′ end of the IVS of C. burnetii (RSA 439 and RSA 493) to generate S23p_439 and S23p_493, respectively (for Primers refer to Table S1). Additionally, an artificial template was engineered by including a ribosome binding site (RBS; AGGAGG) 7 bp upstream of the start codon of S23p before ligating the Ptac (for primers refer to Table S1). A FluroTech™ Green_Lys in vitro Translation Labeling system (Promega) was included in the reaction for fluorescent detection of translated proteins. Reactions were analyzed on 12.5% acrylamide (w/v) SDS-polyacrylamide gel electrophoresis (PAGE) gels and visualized using a FLA 3000G Fluorescent Imager (Fujifilm).

This assay was performed essentially as previously described (Raghavan et al., 2008). Briefly, linear DNA templates (S23p+RBS, S23p_439, and S23p_493) used in IVTT reactions were cloned into pCR2.1-TOPO (Invitrogen), as per manufacturer's instructions, to generate pS23p+RBS, pS23p_439, and pS23p_493, respectively (see Table S2). E. coli (TOP10F′) was transformed with these plasmids to generate strains IRW201, IRW202, IRW203, respectively, or an empty vector (strain IRW204; Table S2) as instructed by the manufacturer (Invitrogen) and were grown overnight at 37°C in lysogeny broth (LB) broth containing 100 μg/ml ampicillin (LBamp). These cultures were used to inoculate fresh LBamp at a 1:10 (v/v) dilution, and the mixture was grown to mid-logarithmic phase (2 h) at 37°C. Isopropyl-beta-D-thiogalactopyranoside (IPTG) was added to 1 mM and growth was assayed spectrophotometrically at 600 nm at hourly intervals for 7 h.

The half-life of IVS RNA in E. coli was determined using a RPA. The IVS element and ~400 bp of flanking sequences were PCR amplified by standard protocol using IVSflank_F and _R primers (Table S1) and cloned into pCR2.1-TOPO per manufacturer's instructions (Invitrogen) to generate pIVS1. E. coli TOP10F' was transformed with pIVS1 to generate strain IRW205, which was cultured in LBamp for 16 h at 37°C and then used to inoculate LBamp at a 1:20 dilution. Following growth for 2 h at 37°C, IPTG was added to 1 mM and cultures were allowed to grow an additional 2 h. Rifampin was added to a final concentration of 160 μg/ml and 1 ml samples were taken at 5 min intervals for 35 min. RNA was purified from samples using TRI Reagent (Ambion) and processed using a RPA III kit (Ambion) [3 μg total RNA; 800 pg IVS probe (Table S3)], as previously described (Hicks et al., 2011). Biotinylated IVS RPA probes were prepared using a MEGAScript kit as instructed (Thermo Fisher), biotin-labeled UTP (Bio 16-UTP; Ambion), and corresponding IVS primers (Table S1; IVSprobe_F and IVSprobe_R+T7).

In vivo stability of IVS RNA in C. burnetii was determined by qRT-PCR. C. burnetii was cultured in ACCM-2, and DNA and RNA were isolated from the same flask at day 0, 1, 7, 10, 13, 16, 19, and 22 using a DNeasy Blood and Tissue kit (Qiagen) and Ribopure kit (Amibion), respectively, as per manufacturer's instructions. Resulting RNA was treated with TURBO DNase (Ambion), precipitated with 100% ethanol and quantified by spectrophotometry (260 nm). RNA (20 ng) from each growth time point was converted to cDNA using an iScript cDNA synthesis kit (Bio-Rad), as instructed by the manufacturer. Genome numbers were determined by qPCR as above. cDNA (diluted 100-fold) was used to perform qRT-PCR with corresponding primer sets (Table S1), as previously described (Raghavan et al., 2008). Amplified cDNA was normalized to respective genome numbers. In addition, total RNA was isolated from C. burnetii cultured in ACCM-2 (0–21 d) and processed as described above, except that 500 ng total RNA was converted to cDNA. qRT-PCR was performed on cDNA samples (diluted 10⁴-fold) using two unique primer sets specific to C. burnetii's 16S, and IVS (Table S1). The qRT-PCR primers were designed using Accelrys gene software (Biovia).

Northern blots were carried out using a NorthernMax kit (Ambion) as per manufacturer's instructions. C. burnetii was cultured in ACCM-2, and total RNA was isolated at 1, 2, 3, 4, 5 d using TRI Reagent Solution (Ambion) following manufacturer's instructions. One hundred nanograms total RNA from each time point was electrophoresed through 1.5% agarose-formaldehyde gels and blotted onto positively-charged BrightStar-Plus nylon membrane (Ambion). Membranes were UV-cross-linked and hybridization were carried out using single-stranded rRNA probes specific to 16S, 23S F1, and 23S F2. rRNA probes were generated as previously described (Warrier et al., 2014). Membranes were developed with Chemiluminescent Nucleic Acid Detection Module Kit (Thermo Fisher) following manufacturer's protocol, and visualized using a LAS-3000 imaging system (Fujifilm). Densitometry was performed using ImageJ software (Schneider et al., 2012; Please see Tables S1 and S3, for primers and probe details).

The IVS of C. burnetii was first described by Afseth et al. as a 444-nt element that is excised to yield a fragmented yet functional 23S rRNA (Afseth et al., 1995). The IVS element is located in helix 45 of C. burnetii's precursor 23S rRNA. IVS RNA of C. burnetii (RSA 493) forms a stable stem-loop secondary structure (ΔG = −143.26 kCal/mol) due to the presence of complementary sequences at its termini (Figure 1). The terminal inverted repeats form a 28-bp stem with complete complementarity except for a single mismatch and a bulge. IVS elements of other bacteria are also associated with 23S rRNA genes and excised by RNase III (Burgin et al., 1990; Ralph and McClelland, 1993; Evguenieva-Hackenberg and Klug, 2000); an endoribonuclease that cleaves rRNA precursors during maturation (Dunn and Studier, 1973). To determine if RNase III is responsible for excision of IVS from the 23S rRNA precursor of C. burnetii, we performed an in vitro assay with T7-transcribed substrate RNA containing IVS and ~400 nt of flanking sequences (Figure 2A) and either recombinant C. burnetii RNase III or commercial E. coli RNase III. Since RNase III cleavage is nonspecific at low ionic strength (Dunn, 1976), we tested cleavage efficiencies at various ionic strengths (data not shown), and a final ionic concentration of 0.5 M (50 mM NaCl plus 450 mM KCl) was used in the reactions. When products of the RNase III assay were analyzed on polyacrylamide gels, three distinct RNA bands corresponding to IVS (444 nt) and its flanking sequences (393 and 364 nts) were observed (Figure 2B). These results indicate that C. burnetii RNase III is capable of cleaving IVS in vitro and most likely does so in vivo. Indistinguishable RNA fragments were observed in reactions containing E. coli RNase III but were absent in the negative control (Figure 2B).

Afseth et al. characterized the IVS length by identifying a sequence that was not a portion of the 23S rRNA by sequence homology (Afseth et al., 1995). To determine actual termini of the element, we performed 5′ and 3′ RACE analyses on IVS RNA isolated from the Coxiella RNase III in vitro assay. At least seven clones were isolated from each reaction and sequenced. The sequences of the 5′ and 3′ ends of IVS RNA are shown with black dots indicating experimentally-determined locations of the RNase III cleavage sites (Figure 3A). While the 3′ end of the IVS was uniform, the 5′ end showed a preferred cleavage site plus three minor, additional sites. When the sites of cleavage were mapped on the predicted secondary structure of IVS, they were found to be in the middle of the stem structure at about 9 bases from the 5′ end and about 15 bases from the 3′ end of the predicted IVS RNA sequence (Figure 3B). RACE analysis indicates the actual size of the C. burnetii IVS as 428 vs. 444 nt, as originally predicted (Afseth et al., 1995).

The IVS of C. burnetii contains an ORF that encodes a predicted 14.2-kDa protein that belongs to the ribosomal S23 family of hypothetical proteins (S23p). While analyzing the nucleotide sequence of the S23p ORF, we found a true stop codon 63 nt downstream from the stop codon identified by Afseth et al. (1995). This observation was consistent with the NCBI database entry for the S23p ORF (CBU_2096; GeneID: 2820718). Phylogenetic analyses performed using the predicted amino acid sequence of Coxiella's S23p, showed that the protein is most closely related to a homolog encoded in the IVS of Haemophilus sp. FF7, with 71% identity (Figure 4A). The predicted structure of C. burnetii's S23p was determined using PHYRE2 (Kelley et al., 2015) and was found to consist of four α-helices (Figure 4B), similar to the S23p of X. campestris (Lin et al., 2006). As previously reported, a number of C. burnetii S23p homologs were also found in pathogenic Leptospira (Afseth et al., 1995). These results prompted us to determine whether the C. burnetii S23p protein could be expressed using an IVTT system. Linear DNA templates were prepared by engineering Ptac upstream of the experimentally-determined IVS sequences of C. burnetii RSA 439 and RSA 493 to produce S23p_439 and S23p_493, respectively. Additionally, an artificial template was constructed with a ribosome binding site (RBS; AGGAGG) upstream of the start codon in addition to Ptac to produce S23p+RBS. Following IVTT, protein products were resolved by SDS-PAGE and visualized by fluorescent imaging. An apparent ~13-kDa protein band was observed when S23p+RBS was used as template but was not detected in IVTT reactions using S23p_439 or S23p_493 (Figure 5A). A corresponding silver-stained SDS-PAGE (Figure 5B) shows equal loading of IVTT reactions in the lanes. These results suggest that the IVS ORF has the potential to synthesize a protein, but only in the presence of a strong RBS. In fact, the putative AG-rich SD region (AGAAGA; Figure 6) is removed from the IVS during excision by RNase III (see Figure 3A) implying that this protein is probably not expressed in vivo.

In silico analysis of the IVS regions of ~80 C. burnetii strains showed that some possess deletions in the sequence immediately upstream of the S23p ORF (Frangoulidis and Walter, unpublished data). As a rule, the 5′ region of the IVS consisted of a potential SD sequence (AGAAGA) followed by three, tandem heptameric repeats (GACAGAT) upstream of the start codon of the S23p ORF (Figure 6). An additional heptameric repeat with a degenerate nucleotide at its 3′ end (GACAGAA) occurred prior to the predicted ATG of the S23p ORF (Figure 6). An unequal number of heptameric repeats was observed upon analysis of different strains of C. burnetii. In Dugway, RSA 331, Q212, and RSA 439 strains, three sets of the heptameric repeat units were found, whereas in strains RSA 493, Q154, Q177, and AuQ01 only two sets of the repeat were present (Figure 6). Interestingly, two non-synonymous nucleotide substitutions were also observed within the S23p ORF in C. burnetii strains Dugway and Q212 at positions 203 and 291, respectively. In addition, the C. burnetii strain AuQ01 was found to have a large, 176-bp deletion in its S23p ORF (data not shown). Taken as a whole, the non-synonymous substitutions and large deletion suggest that the protein-coding segment of the IVS is undergoing negative selection and is expendable.

In order to determine if expression of S23p affects the growth of E. coli, as previously observed with two Group I introns encoded in C. burnetii's 23S rRNA gene (Raghavan et al., 2008), we cloned the linear templates used in the IVTT assay into pCR2.1-TOPO to produce plasmids pS23p+RBS, pS23p_439, and pS23p_493, respectively. Respective E. coli strains harboring these plasmids (i.e., IRW201, IRW202, IRW203) or the vector alone (IRW204) (see Table S2) were induced with IPTG and their growth rates monitored spectrophotometrically for 7 h. As shown in Figure 7, growth rates of E. coli expressing S23p (e.g., strain IRW201) were not significantly different from E. coli containing the pCR2.1-TOPO cloning vector alone (strain IRW204). This observation is consistent with a previous report showing that E. coli expressing the IVS of S. typhimurium did not display any growth defects (Gregory et al., 1996).

Previous work by our laboratory showed that Coxiella's Group I introns, Cbu_L1917, and Cbu_L1951, were spliced from the precursor 23S rRNA and demonstrated remarkable stability (~11 min) in an E. coli background, suggesting a possible adaptive role in Coxiella's biology (Hicks et al., 2011). Since the IVS is uniformly present and highly conserved among strains of C. burnetii, an adaptive role might also be possible for IVS RNA. To determine the intrinsic stability of the IVS RNA, we performed an in vivo stability assay in an E. coli model. E. coli IRW205 was cultured, expression induced by IPTG, and cells sampled after transcription was stopped with rifampicin. RPAs performed on RNAs isolated from these samples showed the half-life of IVS RNA to be ~4 min (Figure 8). Considering that the global, average half-life of mRNA in E. coli is ~5.7 min (Bernstein et al., 2002), these results suggest that IVS RNA does not display extraordinary intrinsic stability like Coxiella's introns. However, since the RNase activities of E. coli may differ from C. burnetii, we also quantified levels of IVS RNA in C. burnetii. C. burnetii was cultured in axenic medium for 22 d and IVS RNA levels were measured by performing qRT-PCR on total RNA purified from the bacterium at 72-h intervals, using primers specific to IVS (Primers: IVS_qPCR_F and _R; Table S1). qPCR was also performed on genomic DNA isolated at the same time points (Figure 9A) to normalize the amount of IVS RNA on a per-genome basis. Results showed that IVS RNA was highest at 1 d (Figure 9B), i.e., the beginning of exponential phase (Omsland et al., 2011). Since rRNAs are highly transcribed during this growth phase (e.g., see the commensurate increase in 16S rRNA in Figure 9B) and mature 23S rRNA is formed following excision of IVS, the increased level correlates with increased synthesis of 23S rRNA. As the bacterium approached stationary phase (~4 d; Omsland et al., 2011), levels of IVS significantly declined and subsequently became insignificant (Figure 9B). Therefore, these results suggest that soon after excision from the 23S rRNA, IVS RNA is degraded in Coxiella, as observed in E. coli (Figure 8).

The 23S rRNA of eubacteria is highly conserved, however, fragmentation caused by IVS elements does not appear to affect ribosome function (Evguenieva-Hackenberg and Klug, 2000). In fact, IVS elements occur within the 23S rRNA genes of several bacteria, including Leptospira (Ralph and McClelland, 1993), Salmonella (Burgin et al., 1990), and Yersinia (Skurnik and Toivanen, 1991). Although fragmentation does not affect the viability of these bacteria, it has been shown to affect the rate of 23S rRNA degradation in Salmonella as cells reach stationary phase (Hsu et al., 1994). Indeed, there appears to be a direct correlation between the degree of IVS-mediated fragmentation and the rate of 23S rRNA degradation, conferring a selective advantage to the bacterium. A similar adaptation could conceivably be adaptive to C. burnetii as it transitions to the small, spore-like, SCV. To investigate this possibility and to help elucidate the biological function of IVS, we performed Northern blot analyses to compare quantities of 16S rRNA to the two fragments of 23S rRNA from C. burnetii cultured in ACCM-2. As a result of RNase III-mediated IVS excision, the 23S rRNA of C. burnetii is cleaved into fragment 1 (F1, ~1.2 kb) and fragment 2 (F2, ~1.7 kb). Since C. burnetii's structural rRNAs are all encoded by a single-copy operon, their concentrations would be expected to initially be at parity following transcription. However, we found that quantities of 23S F1 were consistently and significantly higher (P < 0.043) than 16S rRNA while the levels of 23S F2 were not significantly different from those of the 16S rRNA (Figure 10).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.