|Home | About | Journals | Submit | Contact Us | Français|
We report a search for small RNAs (sRNAs) in the low-GC, Gram-positive human pathogen Streptococcus pneumoniae. Based on bioinformatic analyses by Livny et al. (J. Livny, A. Brencic, S. Lory, and M. K. Waldor, Nucleic Acids Res. 34:3484-3493, 2006), we tested 40 candidates by Northern blotting and confirmed the expression of nine new and one previously reported (CcnA) sRNAs in strain D39. CcnA is one of five redundant sRNAs reported by Halfmann et al. (A. Halfmann, M. Kovacs, R. Hakenbeck, and R. Bruckner, Mol. Microbiol. 66:110-126, 2007) that are positively controlled by the CiaR response regulator. We characterized 3 of these 14 sRNAs: Spd-sr17 (144 nucleotides [nt]; decreased in stationary phase), Spd-sr37 (80 nt; strongly expressed in all growth phases), and CcnA (93 nt; induced by competence stimulatory peptide). Spd-sr17 and CcnA likely fold into structures containing single-stranded regions between hairpin structures, whereas Spd-sr37 forms a base-paired structure. Primer extension mapping and ectopic expression in deletion/insertion mutants confirmed the independent expression of the three sRNAs. Microarray analyses indicated that insertion/deletion mutants in spd-sr37 and ccnA exerted strong cis-acting effects on the transcription of adjacent genes, indicating that these sRNA regions are also cotranscribed in operons. Deletion or overexpression of the three sRNAs did not cause changes in growth, certain stress responses, global transcription, or virulence. Constitutive ectopic expression of CcnA reversed some phenotypes of D39 ΔciaR mutants, but attempts to link CcnA to -E to comC as a target were inconclusive in ciaR+ strains. These results show that S. pneumoniae, which lacks known RNA chaperones, expresses numerous sRNAs, but three of these sRNAs do not strongly affect common phenotypes or transcription patterns.
A large number of noncoding small RNAs (sRNAs) 50 to 400 nucleotides (nt) in length have been detected and characterized recently in numerous bacterial species (reviewed in references 3, 18, and 78). Some abundant, stable sRNAs, such as RNase P (14), tmRNA (34), and scRNA (4.5S RNA) (23, 33), are highly conserved and play important housekeeping and stress-related functions in RNA metabolism, protein degradation, and secretion. But most regulatory sRNAs are conserved only among closely related species (42). Many sRNAs play key roles in responses to stress conditions, such as iron limitation, osmotic shock, temperature shift, stationary phase, and metabolic imbalance, in different bacterial species (3, 15, 17, 25, 26, 46, 76, 78, 79). Other sRNAs are expressed during growth or developmental phases that are specific for particular bacterial species (38, 64, 68, 75). In addition, sRNAs have been postulated to mediate virulence gene expression in several pathogenic bacteria and their survival in hosts (3, 6, 37, 55, 68, 73).
Little is known about RNA metabolism in Streptococcus pneumoniae (pneumococcus), which is a major human respiratory pathogen that causes several serious invasive diseases, including pneumonia, otitis media (ear infection), sinusitis, meningitis, and septicemia (49). Pneumococcus exists as a commensal bacterium that inhabits and colonizes the nasopharynx of up to 20 and 50% of healthy adults and children, respectively, at any time (10). The transition from commensal bacterium to opportunistic pathogen often occurs after a respiratory tract infection, and invasive pneumococcal diseases result in over 1.6 million deaths annually worldwide, especially among young, elderly, debilitated, and immunosuppressed individuals (reviewed in references 13 and 30). Clearly, S. pneumoniae has the ability to inhabit numerous niches in the human body (31, 32), and responses to these different environments likely play roles in colonization and disease progression.
Complete genome sequences of several serotypes of S. pneumoniae have significantly increased our understanding of pneumococcal physiology, pathogenesis, and evolution (27, 28, 39, 70). The three highly conserved sRNAs, RNase P, tmRNA, and scRNA, are contained in all pneumococcal genomes, but other potential sRNAs were not identified or annotated. Bioinformatic searches failed to identify pneumococcal homologues of RNA binding protein Hfq and endoribonuclease RNase E, which are important in sRNA functions and RNA metabolism in Escherichia coli (17, 40, 45, 47, 48, 65). Bioinformatic analyses indicate that Hfq homologues are also absent from several other bacterial pathogens, including Chlamydia and Mycoplasma species, Streptococcus pyogenes, Enterococcus faecium, Helicobacter pylori, and Campylobacter jejuni (66), some of which are known to produce sRNAs.
The only sRNAs identified in S. pneumoniae to date were discovered by serendipity in a search for genes directly regulated by the CiaRH two-component system (TCS) (22), which plays roles in resistance to β-lactam antibiotics, responding to stress, and competence (11, 16, 19, 56). Halfmann et al. defined a consensus CiaR binding site upstream of several genes known to be directly regulated by CiaRH, including lic (modification of teichoic acids), mal and man (sugar metabolism), htrA (stress response protease), parB (chromosomal segregation), and ppmA (protease maturation) (22). Scanning the pneumococcal genome revealed CiaR binding sites upstream of five genes (ccnA to -E) that encode redundant sRNAs (csRNA1 to -5 or CcnA to -E) containing segments of similar sequences and predicted secondary structures. CiaR was shown to positively regulate expression of the CcnA to -E sRNAs, which were undetectable in a ΔciaR mutant (22). However, the functions of CcnA to -E are not known, other than that CcnD and CcnE affect stationary-phase autolysis (22).
To understand regulatory mechanisms and RNA metabolism more fully in S. pneumoniae, we performed an empirical search for sRNAs in virulent serotype 2 strain D39 (39). We based this search on a previous bioinformatic analysis by Livny et al. (41), which predicted putative sRNAs in the genome of pneumococcal serotype 4 strain TIGR4 (41). This method (sRNAPredict2) predicted possible sRNAs based on appropriate spacing of putative promoters and terminators in intergenic regions and conservation of primary sequences and secondary structures among closely related species (42). We converted predictions of sRNAs in the TIGR4 genome to the genome of strain D39 and then experimentally tested for 40 of these predicted sRNAs by Northern analysis. We report here the identification of nine new pneumococcal sRNAs and the previously reported CcnA sRNA (22). We also report the operon structures of three of these pneumococcal sRNA genes and phenotypic characterizations of deletion mutants and ectopic constructs overexpressing the three sRNAs.
Strains used in this study are listed in Table S1 of the supplemental material. Strains were grown on plates containing trypticase soy agar II (modified; Becton-Dickinson [BD]) and 5% (vol/vol) defibrinated sheep blood (TSAII BA) and incubated at 37°C in an atmosphere of 5% CO2. For liquid cultures, strains were cultured statically in BD brain heart infusion (BHI) broth or a chemically defined medium (CDM) (69) at 37°C in an atmosphere of 5% CO2. The pH of the BHI broth before or after equilibration with 5% CO2 was 7.4 or 7.1, respectively. Unless indicated otherwise, ciaR+ strains were inoculated from frozen stocks into 4 ml of BHI broth in 17-mm plastic tubes, serially diluted, and grown for 10 to 16 h. Cultures with an optical density at 620 nm (OD620) of 0.1 to 0.4 were diluted to a starting OD620 between 0.002 and 0.005 in 5 ml of BHI broth in 16-mm glass tubes. For ΔciaR strains, overnight cultures were limited to 10 to 11 h, and cultures with an OD620 of 0.05 to 0.13 were diluted to an OD620 of ≈0.001 in 5 ml of BHI broth in 16-mm glass tubes. Culture tubes were gently inverted before the OD620 was monitored directly using a Spectronic 20 Genesys spectrophotometer. In some cases (see below), 30 ml of diluted starting culture was grown in a 250-ml bottle. At the times indicated, the bottles were swirled gently, and 0.75 ml of culture was removed to a 1-cm-path length cuvette for OD620 determination.
Strains containing antibiotic markers were constructed by transforming linear DNA amplicons synthesized by overlapping fusion PCR into competent pneumococcal cells as described previously (60). For antibiotic selections, TSAII BA plates were supplemented with 200 μg kanamycin per ml, 100 μg spectinomycin per ml, 150 μg streptomycin per ml, 0.3 μg erythromycin per ml, or 2.5 μg chloramphenicol per ml. Strains and plasmids used and primers synthesized for this study are listed in Tables S1 and S2, respectively, of the supplemental material. All constructs were confirmed by DNA sequencing of the amplicon region used for transformation. PCRs, purification of amplicons for transformation and sequencing, and sequencing reactions were similar to those described previously (58). D39 Δspd-sr17::Pc-ermAM, D39 Δspd-sr37::Pc-ermAM, and D39 ΔccnA::Pc-ermAM deletion/insertion mutants (see Table S1 and Fig. S1 in the supplemental material) were constructed by replacing DNA sequences from bp −5 to +71, bp −32 to +39, and bp +46 to +83 in spd-sr17, spd-sr37, and ccnA, respectively, relative to transcription start sites, with the Pc-ermAM cassette amplified from strain IU1547 (51). A ΔciaR::P-ermB mutant was constructed by replacing bp 186 to 591 of ciaR (intact ciaR is 741 bp) with the P-ermB cassette derived from pAMβ1 (7).
Spd-sr17, Spd-sr37, and CcnA were expressed ectopically from the chromosomal expression platform (CEP) site (20) driven from their native promoters by replacing the Pmal (PM) region of CEP with fragments that extended from 100 bp upstream to 20 bp downstream of spd-sr17, 85 bp upstream to 28 bp downstream of spd-sr37, or 153 bp upstream to 27 bp downstream of ccnA. For the CEP::Pmal(c)-spd-sr17, CEP::Pmal(c)-spd-sr37, and CEP::Pmal(c)-ccnA constructs, the respective sRNA gene sequence extending 20, 28, or 27 bp downstream was placed at the +1 site of the Pmal promoter of CEP. These constructions deleted the MalR operator site (53) and thereby created the constitutive Pmal(C) promoter (see Fig. Fig.2D,2D, below; see also Table S2 in the supplemental material). The CEP::PspxB ccnA construct (see Table S2) contains 150 bp upstream of the spxB transcription start site (58). CEP::PrRNA-spd-sr17 contains 104 bp of the rRNA promoter region identified by comparison to the Bacillus subtilis rrnB and rrnO P2 promoters (36). The construct was designed to retain a G at the +1 position to maximize promoter strength, and the expressed spd-sr17 contains an extra G at its 5′ end. CEP::PHspac-spd-sr37 and CEP::PHspac(−1)-spd-sr37 contain 128 and 127 bp, respectively, of the PHyperspank promoter of pDR111 (4) driving spd-sr37 from the +1 and the −1 positions, respectively.
A control CEP::P-kan (kanamycin-resistant) construct lacking sRNA inserts was constructed by transforming an amplicon synthesized from pCEP (20) as template and primers VR49 and VR54 (see Tables S1 and S2 in the supplemental material). A CEP::P-aad9 construct was made by replacing kan in CEP::P-kan with aad9 (spectinomycin resistant) (50). Microarray control analysis of strains IU1690 (D39 parent) versus IU 2586 (D39 CEP::P-kan) grown in BHI broth to mid-exponential phase showed no changes in the relative transcript amounts from amiF and treR, the genes flanking the CEP region, or from any other genes (data not shown).
Markerless comC3 and comC6 alleles (see Fig. S2 in the supplemental material) were generated using the kanr-rpsL+ (Janus cassette) allele replacement method described in reference 67. D39 ΔcomC::[Pc-kanr-rpsL+] was constructed by replacing DNA sequence from bp −122 to +122 relative to the comC coding region with a [Pc-kanr-rpsL+] cassette. The rpsL1 allele in strain IU1781 was repaired to rpsL+ by transformation with a fusion amplicon containing a rpsL+-rpsG+-cat, which contains a chloramphenicol resistance marker (cat), its ribosomal binding site, and four upstream nucleotides (see Table S1 in the supplemental material) inserted immediately downstream of the rpsG stop codon. One transformant (IU3373) of 79 patched was Cmr and Strs and contained the expected rpsL+ sequence. Amplicon from IU3373 was used to transform IU3598 (D39 comC6 rpsL1). Nine of 75 transformants were Cmr and Strs. One transformant was stored as IU3631(D39 comC6 rpsL+-rpsG+-cat).
CFU per ml were determined for strains grown to an OD620 of 0.075 to 0.09 by serially diluting cultures in phosphate-buffered saline (PBS) and spreading aliquots of dilutions onto TSAII BA plates. CFU per ml obtained on plates for strains IU2678 or IU2841 were similar to the number of chains per ml counted manually using a Petroff-Hausser chamber on a phase-contrast microscope (data not shown).
After cultures reached an OD620 of ≈0.06 and ≈0.2, 500 μl was removed and centrifuged at 16,000 × g for 2 min at room temperature. Pellets were suspended in 100 μl of BHI broth. Cells were examined using a Nikon E-400 phase-contrast microscope, and images were captured using a cooled digital SPOT camera (1). At least 25 to 30 chains from each of two independent cultures of each strain were counted to determine distributions of numbers of cells per chain.
Strains were diluted into 30 ml of BHI broth in 250-ml bottles to a starting OD620 of ≈0.001, and separate parallel, unperturbed cultures were used for each time point. Starting from the initial inoculation and at 1-h intervals thereafter, 1 ml of cell suspension was removed and mixed with 40 ng of amplicon DNA carrying a novobiocin resistance marker (see Table S1 in the supplemental material), which was obtained by PCR of CP1500 genomic DNA using primers XXV-F-036 and XXV-R-040 (see Table S2 in the supplemental material). After incubation for 55 min at 37°C in an atmosphere of 5% CO2, 800-μl aliquots of the cell suspensions were mixed with 3 ml of melted 42°C 0.8% nutrient broth plus 0.7% (wt/vol) Bacto Agar containing 2.5 μg novobiocin per ml and poured onto TSAII BA plates containing the same concentration of novobiocin. The transformation frequency was determined as the ratio of Novr CFU to total CFU per unit volume of cell suspension. The detection limits for this assay was ≈10−5 and ≈10−8 at early and late time points in growth curves, respectively.
Strains IU1690 (D39 wild type [WT]), IU2083 (D39 ΔccnA::Pc-ermAM), IU2084 (D39 Δspd-sr17::Pc-ermAM), and IU2086 (D39 Δspd-sr37::Pc-ermAM) were grown exponentially in 9 ml of BHI broth to an OD620 of 0.1. Three milliliters of cultures was transferred to tubes containing the following chemicals to give the final concentrations indicated: 0.1 or 0.2 M NaCl, 25 or 50 ng mupirocin per ml, or 0.1 or 0.03 μg ampicillin per ml. Growth was further monitored and compared to that of control cultures lacking additives. For further antibiotic resistance testing, the same set of strains was grown exponentially in BHI broth to an OD620 of 0.1 to 0.2. One milliliter of culture was added to molten soft agar (see above) and poured onto TSAII BA plates. After the soft agar solidified, Sensi-Disks (BD BBL) containing 10 IU penicillin, 10 μg ampicillin, 75 μg ticarcillin, 1 μg oxacillin, 30 μg vancomycin, or 10 μg amdinocillin were placed on the surface of the plates. Plates were incubated for 24 h at 37°C in an atmosphere of 5% CO2, and the diameters of inhibition zones were determined. Virulence studies were carried out on this set of strains using a murine pneumonia model of infection with intranasal inoculation (≈2 × 108 CFU) of 4- to 5-week-old, male BALB/c mice as described elsewhere (58) All procedures were approved in advance by the Institutional Animal Care and Use Committee and were performed according to the guidelines of the National Research Council.
Total RNA was prepared by a method that retained RNA species of all sizes. Thirty-milliliter BHI broth cultures of the D39 parent contained in 50-ml conical centrifuge tubes were grown from a starting OD620 of 0.005 to a final OD620 of 0.2 (exponential phase), for an additional 1 h after reaching an OD620 of 0.6 (early stationary phase), or from a starting OD620 of 0.002 to a final OD620 of 0.1, after which synthetic competence stimulatory peptide 1 (CSP1) was added to a final concentration of 100 ng per ml for 12 min. Ten milliliters of exponential-phase and CSP-treated cultures and 3 ml of stationary-phase cultures were chilled in precooled flasks on ice, transferred to precooled 50-ml centrifuge tubes, and centrifuged at 13,000 × g for 6 min at 4°C. Pellets were resuspended in 0.5 ml of boiling solution A (1% [wt/vol] sodium dodecyl sulfate [SDS], 20 mM sodium acetate, 8 mM EDTA, pH 5.5) and heated in a boiling water bath for 2 min. The 0.5 ml of lysate was extracted with acidified phenol (pH 4.3) at 65°C with shaking, followed by extraction with phenol-chloroform-isoamyl alcohol and chloroform at room temperature as described previously (74). Nucleic acids were precipitated with ethanol, dried, treated with DNase (Promega RQ1 DNase; 12.5 units per 250 μg total RNA), and the phenol-chloroform extraction, ethanol precipitation, and drying steps were repeated.
Total RNA (5 μg per lane) and size markers (RNA Century-plus [Ambion] and a 49-nt DNA oligomer) were fractionated by polyacrylamide gel electrophoresis (PAGE) on Tris-buffered EDTA (TBE)-8% polyacrylamide-7 M urea gels (5) and transferred electrophoretically to BrightStar-Plus positively charged nylon membranes (Ambion) using a semidry blot apparatus (Bio-Rad). RNA was cross-linked to membranes with UV light using the optical cross-linking setting on an XL-1000 UV Spectro linker (Spectronics Corp.). Hybridizations were conducted according to the protocol accompanying the Ultrahyb-oligo buffer (Ambion) at 37°C for 16 to 20 h with rotation. Blots were washed twice with 2× SSC (1× is 0.15 M NaCl plus 0.015 M sodium citrate) and 0.5% (wt/vol) SDS for 30 min each at 37°C with rotation.
DNA oligonucleotides used as probes for Northern blots are listed in Table S3 of the supplemental material. Two probes that target different regions were designed for some of the predicted sRNA sequences: 12(1)-12(2); 17(1)-17(2); 37(1)-37(2); 52(1)-52(2). T4 polynucleotide kinase (New England Biolabs) was used to end label 2 pmol of each synthetic DNA oligonucleotide with 1.67 pmol of [γ-32P]ATP (6,000 Ci per mmol; Perkin-Elmer). Radiolabeled oligonucleotides were purified using Sephadex G-25 quick spin columns (Roche). Labeled Northern blots were exposed to X-ray film to obtain an image and to a phosphor screen (Amersham) for 10 to 30 min for quantitation. The phosphor screen was scanned with a Typhoon 9200 variable mode imager (Amersham), and quantitation of bands was performed with ImageQuant software (Molecular Dynamics). sRNA amounts were normalized to 5S rRNA amounts by stripping blots (5) and rehybridizing them with a radiolabled 5S rRNA probe (see Table S2 in the supplemental material). To allow comparisons, normalized amounts of sRNA from CSP-treated and stationary-phase samples were divided by the normalized amounts of sRNA from exponential-phase samples, which were set to 1.
Primer extension using the primers in Table S2 of the supplemental material was performed as described before (58) to map the 5′ ends of the spd-sr17, spd-sr37, and ccnA sRNA transcripts in strain D39 (IU1690) grown exponentially in BHI broth (OD620, 0.2). The comC transcription start point (see Fig. S2 in the supplemental material) was mapped for strains D39 (IU1690) and IU2678 (D39 ΔciaR::P-ermB CEP::pnative-ccnA). The 5′ U residue determined by this method was 1 nt shorter than the 5′-G start reported elsewhere (21).
To determine relative amounts of comD transcript by quantitative reverse transcription-PCR (qRT-PCR) at different growth points and for microarray analyses of exponential cultures (OD620, 0.06 to 0.1), total RNA was prepared from cultures growing at 37°C in BHI broth in separate bottles as described above for the natural transformation assays. RNA from 8 to 12 ml of culture was prepared by a rapid lysis procedure followed by purification using an RNAeasy minikit (Qiagen) as described previously, including on-column treatment with DNase I (Qiagen) (58, 59). This procedure excludes RNA species smaller than 200 nt. For qRT-PCR and RT-PCR assays, 5 μg of total RNA was further digested with DNase using a DNA-free kit (Ambion).
Synthesis, labeling, and hybridization to S. pneumoniae microarrays (Ocimum Biosolutions) covering 2,018 open reading frames (ORFs) of the R6/D39 genome, scanning, and analysis using the Cyber-T web interface were performed as described previously (32, 39, 52, 58). Data were normalized without background subtraction by the global Lowess method using BASE (BioArray Software Environment; http://iubase.cgb.indiana.edu), excluding empty wells and Arabidopsis thaliana control spots. Intensity and expression ratio data for all transcripts have been deposited in the GEO database (accession number GSE14688).
qRT-PCR analysis was performed as described before (58) with the exception that cDNA was prepared with a qScript Flex cDNA synthesis kit (Quanta Biosciences) according to the manufacturer's protocol. PCR primers (see Table S2 in the supplemental material) covering the center regions of spd_0239 (F1 and R1 [see Fig. Fig.2C,2C, below]), spd_0240 (F2 and R2 [see Fig. Fig.2C]),2C]), or the 3′ region of the comD transcript were used for RNA quantitation. All primer sets showed standard curves with R2 values of >0.985, 90 to 110% reaction efficiencies, and only one peak in dissociation curves. spd_0240 and comD transcript amounts were normalized to 16S rRNA amounts by using primers in Table S2 of the supplemental material as described previously (58). The 16S-normalized comD transcript amounts were then expressed as ratios relative to the amount of comD transcript in the D39 parent strain (IU1690) at mid-exponential phase (OD620 of 0.1 to 0.2), which was set to 1.
For the detection of the spd_0239-spd_0240-ccnA cotranscript, cDNA synthesis was performed with a StrataScript first-strand synthesis system (Stratgene) using 242 ng of total RNA from IU1690 and 700 nM ccnA Northern probe (see Fig. S1C in the supplemental material) as the reverse primer. PCRs were performed with 0.25 μl of the cDNA reaction mixture, forward primers F1 or F2 (see Fig. Fig.2C),2C), ccnA Northern probe as the reverse primer, and PCR enzyme GoTag (Promega). Amplicons of 1,096 bp and 456 bp from F1 and F2 to the ccnA probe, respectively, were predicted and detected (data not shown). Detection of trmU-spd-sr37-spd_0128-gidA cotranscript was performed similarly with R3 as the reverse primer for the RT reaction and primers F3 and R3 for the PCRs (see Fig. Fig.2B).2B). A 2,109-bp amplicon was predicted and detected (data not shown). Control amplifications carried out by omitting reverse transcriptase in the first step of the procedure did not show any product indicative of DNA contamination in the RNA preparations (data not shown).
Livny, Waldor, and coworkers predicted 63 putative sRNAs with lengths of 66 to 365 nt in the intergenic regions of the TIGR4 strain (41). We numbered these candidate sRNAs from 1 to 63 according to the order in Table S10 of reference 41. BLAST searches were run for the 63 sRNAs against the genome of laboratory strain R6. Forty candidate sRNAs were located in similar regions of the TIGR4 and R6 genomes (coordinates can be found in Table S3 of the supplemental material) and showed more than 90% sequence identity between the two genomes. Twelve of the 40 sequences were identical in the TIGR4, R6, and D39 genomes (Table (Table1;1; see also Table S3 in the supplemental material), and all 40 predicted sRNAs were identical between related strains R6 and D39 (39).
Probes were designed for the 40 putative sRNAs (see Table S3 in the supplemental material), and Northern analyses were performed on total RNA prepared from strain D39 growing in exponential phase, at early stationary phase, and after treatment with CSP for 12 min in BHI broth (see Materials and Methods). Of the 40 sRNA candidates analyzed, 22 probes showed no signal, and 6 probes showed multiple bands on the Northern blots (data not shown). Distinct bands corresponding to sRNA species of approximately 80 to 400 nt were found for 12 probes (Fig. 1A and B; Table Table1).1). We designated these sRNAs temporarily as Spd-sr (S. pneumoniae D39 sRNA) followed by the numbers assigned above. The genomic coordinates in the R6 and D39 strains, flanking genes, predicted lengths, and approximate sizes of these detected sRNAs are shown in Table Table11.
In the course of our study, Spd-sr56 was reported as the ccnA-transcribed csRNA1 (22), and we refer to Spd-sr56 (csRNA1) as CcnA sRNA here. One candidate, sRNA (Spd-sr7), was considerably longer (≈400 nt) than its predicted size of 63 nt (data not shown). This longer transcript likely corresponded to a contiguous transcript with the upstream gene, spd_0803, which contains an open reading frame of 303 bp. BLAST searches of the probes used for the Northern blots showed that four candidate sRNAs, Spd-sr10, Spd-38, Spd-39, and Spd-52, may be redundant and transcribed from multiple sites in the D39 chromosome (Table (Table1).1). There is a second putative gene with a nearly identical sequence to that of spd-sr10 at a different chromosomal location. spd-sr38, spd-sr39, and possibly seven other similar sRNAs are transcribed divergently from the transposase gene within one end of complete and truncated IS1167 elements located around the chromosome. Another candidate, Spd-sr52, may arise from multiple intergenic regions with similar sequences (>85% identity) (Table (Table1).1). Probes 52(1) and 52(2), which were designed to detect a predicted 135-nt sRNA (Table (Table1;1; see also Table S3 in the supplemental material), hybridized instead to two slightly different sized transcripts of ≈60 and ≈65 nt, respectively (Fig. (Fig.1A).1A). Primer extension assays showed a 5′ end consistent with the ≈60-nt transcript that hybridized to probe 52(1) (data not shown). The origin of the ≈65-nt transcript detected by probe 52(2) was not studied further.
Five species of sRNAs showed differential expression in response to growth phase or CSP addition in the parent D39 strain (Fig. (Fig.1;1; Table Table1).1). The relative amounts of Spd-sr12, Spd-sr14, Spd-sr17, and Spd-sr54 decreased in stationary phase and that of CcnA (Spd-sr56) increased upon CSP addition. We initially focused on studying the Spd-sr17, Spd-sr37, and CcnA sRNAs, because they produced single, strong bands (Fig. (Fig.1B),1B), and Spd-sr17 and CcnA showed strong differential regulation (5-fold [P < 0.01] and 3.6-fold [P < 0.005], respectively) (Fig. (Fig.1C)1C) under the limited number of conditions tested here.
spd-sr17 is located in the intergenic region between the gene for tRNAThr1 and asd, which encodes aspartate β-semialdehyde dehydrogenase from the threonine-lysine biosynthetic pathway (Fig. (Fig.2A).2A). Spd-sr17 was predicted to be 164 nt in length (Fig. (Fig.2A).2A). Primer extension experiments mapped the 5′ end of spd-sr17 downstream from an extended −10 region (TCTAGTAATAT) (see Fig. S1A in the supplemental material), characteristic of pneumococcal promoters (61). Therefore, the length of Spd-sr17 is 144 nt, assuming transcription termination after the last T of the predicted strong terminator (see Fig. S1A). Pneumococcus lacks Rho factor homologues, and this estimation of the 3′ end is likely to be accurate to within a couple of nucleotides. The end of this transcription terminator was 94 bp upstream of the start codon of asd, and Spd-sr17 does not contain an apparent antiterminator RNA segment, nor does it encode a threonine-lysine-rich peptide, making it unlikely that Spd-sr17 acts as a leader RNA for asd under the conditions tested so far.
A band consistent with a 144-nt Spd-sr17 species was detected in Northern blots using probes 17(1) (Fig. (Fig.1B1B andC; see also Fig. S1A in the supplemental material) and 17(2) (data not shown) (see Fig. S1A), and this band was missing in a D39 Δspd-sr17 mutant deleted for part of spd-sr17 complementary to the two probes (Fig. (Fig.1C).1C). Upon longer exposure, faint bands were visible above 750 nt on blots; however, these bands were still present for the Δspd-sr17 mutant (Fig. (Fig.1C),1C), indicating that they were not specific to the spd-sr17 genomic region. To confirm that spd-sr17 is an independent transcription unit, Northern blot assays were performed with RNA prepared from a D39 Δspd-sr17 CEP::Pnative-spd-sr17 construct, which contains a copy of the spd-sr17 region (from −100 to +164) inserted into the ectopic CEP site (Fig. (Fig.2D;2D; see also Table S1 in the supplemental material) (20). This strain produced the same 144-nt band as the IU1690 parent strain, consistent with inclusion of the spd-sr17 promoter and terminator in the ectopic copy (Fig. (Fig.3A).3A). Likewise, ectopic expression of Spd-sr17 from the constitutive Pmal(c) promoter or from the PrRNA promoter in the parent strain caused overexpression of the same band (Fig. (Fig.3A).3A). Microarray analyses of Δspd-sr17 mutants did not show changes in the relative transcript amounts of the flanking genes (data not shown), consistent with spd-sr17 as a single-gene operon. Spd-sr17 likely folds into a compact, stable structure that has stem-loop structures at the 5′ and 3′ ends and a moderate-sized, unpaired internal loop (Fig. (Fig.4A4A).
spd-sr37 is in the intergenic region between trmU encoding tRNA (5-methylaminomethyl-2-thiouridylate)-methyltransferase and spd_0128 encoding a conserved hypothetical protein in the MutT/Nudix family. Spd-sr37 was predicted to be 92 nt in length (41). Primer extension assays mapped a transcript 5′ end downstream from an extended −10 promoter box (see Fig. S1B in the supplemental material); therefore, the length of the spd-sr37 transcript is 80 nt, assuming transcription termination at the last T of the second of two relatively weak predicted terminators downstream of spd-sr37 (see Fig. S1B).
A band consistent with a length of 80 nt was detected in Northern blots using probe 37(1) (Fig. 1B and C; see also Fig. S1B in the supplemental material) but not probe 37(2), which corresponded to a region upstream of the spd-sr37 5′ end (data not shown) (see Fig. S1B). Upon longer exposures, bands corresponding to transcripts of ≈750 nt and >1,000 nt were detected on Northern blots (Fig. 1B and C). The three bands (80, ≈750, and >1,000 nt) were absent in a D39 Δspd-sr37 mutant (Fig. (Fig.1C),1C), indicating that these transcripts contained sequences that paired with probe 37(1). Cotranscription of spd-sr37 with upstream trmU and downstream spd_0128 and gidA was confirmed by RT-PCR experiments (data not shown) (see Materials and Methods).
Independent transcription of Spd-sr37 was confirmed in constructs containing the Pnative-spd-sr37 gene expressed from the CEP site in the D39 Δspd-sr37 mutant (Fig. (Fig.2D2D and and3B).3B). Consistent with this interpretation, Spd-sr37 was overexpressed when driven from the constitutive Pmal(c) and PHspac(−1) promoters in the CEP site in the D39 parent strain (Fig. (Fig.3B).3B). The apparently identical size of Spd-sr37 expressed from the native and ectopic sites argues against processing of a larger transcript as the origin of the Spd-sr37 sRNA. It seems likely that the terminator after spd-sr37 (Fig. (Fig.2B)2B) serves as an attenuator that terminates the sRNA transcript but also allows transcription readthrough. Spd-Sr37 likely folds into a stable, highly paired structure lacking unpaired regions (Fig. (Fig.4B4B).
Microarray analyses of the D39 Δspd-sr37 mutant showed large (≈12-fold) increases in the relative transcript amounts from both upstream trmU and downstream spd_0128 compared to the D39 parent (data not shown). In addition, the relative transcript amount from the next gene downstream, gidA (tRNA uridine 5-carboxymethylaminomethyl modification enzyme) increased about 5-fold in the D39 Δspd-sr37 mutant. Expression of Spd-sr37 at about the wild-type level from the CEP::Pnative-spd-sr37 construct (Fig. (Fig.3B)3B) did not reduce the relative transcript amounts of trmU, spd_0128, and gidA in the D39 Δspd-sr37 mutant (data not shown). This result indicates that the Δspd-sr37 mutation was exerting a cis-acting effect on expression of the cotranscribed flanking genes, rather than a trans-acting effect caused by lack of expression of the sRNA.
Of the five similar ccnA to -E genes dependent on CiaR activation (22), ccnA and ccnB are located in tandem downstream from spd_0239 and spd_0240, which encode conserved hypothetical proteins similar to GNAT family acetyltransferases (Fig. (Fig.2C).2C). No predicted terminator sequences are located between spd_0239 and spd_0240 or between spd_0240 and ccnA. Probe 56, which corresponded to the nt 50 to 84 region of CcnA that was dissimilar to the other four CcnB to -E sRNAs (see Fig. S1C in the supplemental material), hybridized to the distinct, strong 93-nt CcnA band and to a weak >1,000-nt band (Fig. (Fig.1C).1C). Both bands were absent in blots of a D39 ΔccnA mutant, suggesting possible cotranscription of the ccnA region with adjacent genes, and probe 56 did not cross-hybridize with CcnB to -E (nt 92 to 151) from the D39 ΔccnA mutant (Fig. (Fig.1C).1C). Cotranscription of spd_0239-spd_0240 and the ccnA region was confirmed by RT-PCR experiments (see Materials and Methods) in which forward primer F1 or F2 and reverse primer (probe) 56 (Fig. (Fig.2C)2C) yielded a PCR product consistent with a size of 1,096 or 456 bp, respectively, but did not yield bands in controls lacking reverse transcriptase (data not shown). Thus, the ccnA region acts as a 3′-untranslated region of a spd_0239-spd_0240 cotranscript.
We confirmed that the 93-nt CcnA sRNA was still independently transcribed (20) by ectopic expression from the CEP site (Fig. (Fig.2D).2D). Expression of CcnA from its native promoter or from the constitutive Pmal(c) restored the 93-nt band in the D39 ΔccnA mutant to about the level found in the D39 parent strain (Fig. (Fig.5B).5B). Likewise, expression of the ectopic constructs in the D39 parent led to 2.6- to 4.0-fold-greater expression of CcnA (Fig. (Fig.3A),3A), with the highest overexpression from the constitutive Pmal(c) promoter. Together, these results confirm the previous conclusion of Halfmann et al. (22) that CcnA is transcribed independently from its own promoter to a transcription terminator. As reported before, CcnA likely folds into a structure with 5′ and 3′ stem-loop structures flanking a lengthy unpaired region (Fig. (Fig.4C4C).
Microarray analyses of the D39 ΔccnA mutant compared to the D39 parent showed that the relative transcript amounts from the upstream spd_0239 and spd_0240 genes increased 4- and 10-fold, respectively (data not shown). qRT-PCR analyses confirmed overexpression of these two transcripts (data not shown). These results are consistent with the Northern blot and RT-PCR data mentioned above and further confirm that the ccnA region is cotranscribed with the two upstream genes. Ectopic expression of CcnA at a nearly wild-type level from the Pnative or Pmal(c) promoters in a ΔccnA mutant failed to reduce the relative amounts of the spd_0239 and spd_0240 transcripts as determined by qRT-PCR and microarray assays (data not shown). Thus, the increase in relative amounts of the spd_0239 and spd_0240 transcripts in the ΔccnA mutant was a cis-acting effect of the construct, rather than a trans-acting effect of the sRNA. Similar to the spd-sr37 gene described above (Fig. (Fig.2B;2B; see also Fig. S1B in the supplemental material), the region transcribed independently into the CcnA sRNA is also cotranscribed with adjacent genes, and certain insertions in the ccnA region perturb the expression of these other genes, possibly by stabilizing the cotranscript.
There were no overt phenotypic differences between the D39 parent strain and sRNA deletion strains IU2083 (D39 ΔccnA::Pc-ermAM), IU2084 (D39 Δspd-sr17::Pc-ermAM), and IU2086 (D39 Δspd-sr37::Pc-ermAM) in the following tests (see Materials and Methods): growth in BHI broth or CDM at 30°C or 37°C; growth in BHI broth adjusted to a pH between 6.8 and 7.4 at 37°C; growth on TSAII BA plates at 30°, 37°, or 40°C; cell morphology or chaining of the cells grown exponentially in BHI broth (OD620, ≈0.2) at 37°C; sensitivity to 0.1 or 0.2 M NaCl, 25 or 50 ng per ml mupirocin, or 0.1 or 0.03 μg per ml ampicillin in BHI broth culture; sensitivity on TSAII BA plates at 37°C to Sensi-Disks containing penicillin, ampicillin, ticarcillin, oxacillin, vancomycin, or amdinocillin. No statistically significant difference in mean survival time was detected between the D39 parent and the three sRNA deletion mutants in a BALB/c mouse model of pneumonia (see Materials and Methods). Likewise, with the exception of the genes flanking spd-sr37 and ccnA mentioned above, only small changes around the 2.0-fold cutoff were observed for relative transcript amounts in microarray profiles of the D39 ΔccnA, D39 Δspd-sr17, and D39 Δspd-sr37 mutants grown exponentially in BHI broth at 37°C (data not shown). Finally, overexpression of Spd-17 (≈7×), Spd-37 (≈5×), or CcnA (≈4×) in D39 merodiploids containing ectopic constructs driven by the PrRNA, PHspac(−1), or Pmal(c) promoters, respectively (Fig. (Fig.33 and and5)5) did not cause strong changes in growth or microarray profiles in bacteria grown exponentially in BHI broth at 37°C (data not shown).
Lack of phenotypes caused by deletion or overexpression of CcnA is consistent with redundancy with the other four highly similar CcnB-E sRNAs (22). In addition, the above results show that mutations in ccnA can lead to complicated cis-acting effects. To circumvent these problems and study the functions of CcnA, we tried to take advantage of the finding that expression of CcnA to -E is strictly dependent on the CiaR response regulator (22). Therefore, a ΔciaR mutation eliminates CcnA to -E redundancy, and ectopic expression of CcnA from the CEP site avoids disrupting the ccnA region (Fig. (Fig.2D).2D). No CcnA transcript was detected in the D39 ΔciaR CEP::Pnative-ccnA strain, indicating that the ectopic Pnative promoter was still fully dependent on CiaR for expression (Fig. (Fig.5C).5C). In strains expressing CcnA constitutively from different promoters in the CEP site, the amount of CcnA detected was uniformly lower (10% to 60%) in the ΔciaR mutant than in the ΔccnA mutant (Fig. 5B and C). We do not know why the CcnA amount is reduced in the ΔciaR mutant compared to the ciaR+ strain (Fig. 5B and C), but the expression level of CcnA from the Pmal(c) promoter in CEP was still sufficient to reverse some phenotypes compared to the nonexpressing ΔciaR CEP::Pnative-ccnA strain.
We observed that growth of the D39 ΔciaR mutant was substantially defective and extremely sensitive to aeration under these culture conditions. Growth of the D39 ΔciaR mutant in static BHI broth at 37°C in an atmosphere of 5% CO2 depended on the volume of medium relative to the size of the vessel and aeration of the cultures during growth (Fig. (Fig.6A).6A). In tubes, growth of the D39 ΔciaR mutant was moderately biphasic and usually reached a lower growth yield than that of the D39 parent strain (Fig. (Fig.6A),6A), as reported previously (29). In tubes without CO2, growth of the ΔciaR mutant was delayed for hours (data not shown). In bottle cultures, the extent of growth of the ΔciaR mutant was affected by how often cultures were perturbed to take density readings and was again reduced compared to that of the D39 parent strain (Fig. (Fig.6B).6B). We did not distinguish in these experiments whether these growth differences were caused by slight differences in oxygen content, pH, or both.
Because of this extreme sensitivity to growth conditions, experiments involving the ΔciaR mutant were somewhat variable and were repeated numerous times (>16) with five independent constructs. Greater than 80% of ΔciaR CEP::Pnative-ccnA cultures showed reduced growth, similar to that of the ΔciaR mutant (Fig. (Fig.6).6). In contrast, constitutive expression of CcnA from the Pmal(c) promoter in the CEP site improved growth compared to the ΔciaR and ΔciaR CEP::Pnative-ccnA mutants in greater than 85% of the cultures.
We noticed that the CFU per ml per OD620 unit was consistently reduced in the D39 ΔciaR mutant compared to the D39 parent at low cell densities (OD620, ≈0.08) in early exponential phase (Fig. (Fig.7,7, bars 1 and 2). ΔciaR strains containing CEP::Pc-kan (control) or CEP::Pnative-ccnA showed a lowered CFU per ml per OD620 unit, similar to that of the ΔciaR mutant (Fig. (Fig.7,7, bars 3 and 4). In contrast, ectopic expression of CcnA from Pmal(c) restored the CFU per ml per OD620 unit to that of the D39 parent strain (Fig. (Fig.7,7, bar 5).
This phenotype suggested that the chaining properties were different between the strains expressing or not expressing CcnA, because each chain of cells will form 1 CFU on plates. Microscopic examination of low-density cultures (OD620, 0.05 to 0.06) confirmed that considerably longer chains of cells were present in the ΔciaR, ΔciaR CEP::Pc-kan (control), and ΔciaR CEP::Pnative-ccnA mutants compared to the chaining distribution of the ciaR+ parent (Fig. 8A and B and data not shown). Constitutive ectopic expression of CcnA restored the chain length distribution of the ΔciaR mutant to that of the ciaR+ parent (Fig. (Fig.8C).8C). Direct counting of chains per unit volume with a counting chamber (see Materials and Methods) confirmed that at the same low culture density (OD620, ≈0.06), the ΔciaR CEP::Pnative-ccnA mutant produced fewer, but longer, chains whereas the D39 parent and ΔciaR CEP::Pmal(c)-ccnA construct produced shorter but more chains per volume of culture (data not shown). We confirmed these growth and chaining phenotypes with an independently constructed set of strains (IU3067 and IU3036) (data not shown; see Table S1 in the supplemental material). As a control, we replaced the ectopic CEP::Pmal(c)-ccnA locus in strain IU2841 (Fig. (Fig.8C)8C) with a CEP::P-aad9 marker to give strain IU3100 (see Table S1 in the supplemental material). Strain IU3100 formed long chains of cells, similar to those of the ΔciaR CEP::Pnative-ccnA mutant (Fig. (Fig.8B8B and data not shown). Finally, this apparent difference in chaining lessened as cultures continued to grow. At an OD620 of ≈0.2, all of the strains produced short or moderate-length chains of cells, similar to the D39 parent, but lysis was microscopically visible in cultures of the ΔciaR CEP::Pnative-ccnA mutant that was not present for the D39 parent and ΔciaR CEP::Pmal(c)-ccnA strains (data not shown). This chain shortening and lysis seemed to coincide with the biphasic shift in growth rate of the ΔciaR and ΔciaR CEP::Pnative-ccnA mutants (Fig. (Fig.6A6A).
Previous studies of comCDE operon expression and competence development in ΔciaR mutants showed dependencies on growth conditions, such as medium composition, degree of oxygenation, and pH (11, 12, 35, 44). We performed qRT-PCR analyses of relative comD transcript amounts and measured natural transformation efficiencies at different times during growth in bottles that were sampled only once (Fig. (Fig.9)9) (see Materials and Methods). The growth curves for these conditions (Fig. (Fig.9A)9A) were similar to those of cultures in tubes (Fig. (Fig.6A)6A) and different from those of cultures in bottles that were sampled every hour (Fig. (Fig.6B6B).
Relative comD transcript levels normalized to 16S rRNA were determined at different stages of growth 2.5 to 6.5 h after inoculation as described in Materials and Methods (Fig. 9A and B). Under these culture conditions, relative comD transcript amounts were substantially higher (>100-fold) in the ΔciaR CEP::Pnative-ccnA mutant compared to the D39 parent or the ΔciaR CEP::Pmal(c)-ccnA strain in >88% of the cultures tested (Fig. (Fig.9B).9B). Thus, ectopic CcnA expression normally reduced the amount of the comD transcript, and by inference the comCDE cotranscript, strongly at all stages of growth. These results were confirmed in a set of independently constructed strains with the same genotypes as those shown in Fig. Fig.9B9B (strains IU2677 and IU2752, three isolates of IU3036, and IU3067) (see also Table S1 in the supplemental material). As a control, we determined the relative comD transcript amount in strain IU3100, in which the ectopic CEP::Pmal(c)-ccnA locus of strain IU2841 was replaced by a CEP::P-aad9 marker (see Table S1). The relative comD transcript amount in IU3100 increased to that of the ΔciaR CEP::Pnative-ccnA strain (Fig. (Fig.9B9B and data not shown). Therefore, even if a suppressor had arisen in strain IU2841, its ability to lower comD transcription would still be dependent on ectopic expression of CcnA. Finally, microarray experiments showed that the rest of the competence regulon was induced in the ΔciaR CEP::Pnative-ccnA strain compared to the ΔciaR CEP::Pmal(c)-ccnA strain (see Table S4 in the supplemental material).
We determined whether the natural transformation frequency followed the relative transcript amounts of comD and the competence regulon in these strains. Under these culture conditions (BHI at 37°C in a 5% CO2 atmosphere), the D39 parent was not naturally competent (asterisks) within the limits of detection of this assay (Fig. (Fig.9C).9C). In contrast, the ΔciaR CEP::Pnative-ccnA mutant became measurably competent, consistent with the increased competence regulon transcript amounts detected in the qRT-PCR and microarray experiments (Fig. (Fig.9B;9B; see also Table S4 in the supplemental material). The natural transformation frequency of the ΔciaR CEP::Pnative-ccnA mutant followed a reproducible pattern with time in culture, with a high peak (5 × 10−3 to 10−4) about 2.5 h after inoculation, followed by a drop to about 10−7 in the next hour and then a rise to 10−6 for the next 2 h. Ectopic expression of CcnA in the ΔciaR CEP::Pmal(c)-ccnA strain abolished or greatly reduced this natural competence (Fig. (Fig.9C),9C), again consistent with the results from transcript analyses. Similar results were obtained in three biological replicates of this experiment (data not shown).
Finally, to test whether the putative pairing between CcnA to -E and comC mRNA plays a role in negatively regulating competence regulon transcription, we introduced point mutations into the comC leader region that should reduce pairing to CcnA to -E (see Fig. S2 in the supplemental material). Unfortunately, the rpsL1 (streptomycin-resistant) mutation required by the allele exchange method (67) to make the base change mutations in comC altered the phenotypes of the D39 ΔciaR mutant (data not shown). Therefore, we confined these experiments to the ciaR+ background, and we repaired the rpsL1 mutation in one of the mutants. Mutants IU3167 (D39 comC3 [5 base changes] rpsL1) and IU3631 (D39 comC6 [14 base changes] rpsL+ linked to cat) (see Fig. S2) did not affect relative growth or comD transcript amount compared to the D39 ciaR+ strains during exponential growth in BHI broth (data not shown). The implications of this negative result and possible mechanisms for the reversal of ΔciaR mutant phenotypes by CcnA expression are considered in the Discussion section.
A relatively limited number of sRNAs have been identified so far in most Gram-positive bacteria, and considerably less is known about the functions of sRNAs in Gram-positive than in Gram-negative species (3, 6, 43, 54, 57, 62, 71, 73). In addition, the functions of RNA chaperones, such as homologues of Hfq, have been elusive in many of these Gram-positive species (2, 3, 15, 26), with the exception of Listeria monocytogenes, where Hfq binds to sRNAs and plays roles in mRNA stability (6, 43). We report here that S. pneumoniae expresses at least nine sRNAs besides the five related CcnA to -E sRNAs that were reported previously under the control of the CiaR response regulator (22). Besides CcnA to -E, three of these sRNAs (Spd-sr10, Spd-sr38/39, and Spd-sr52) may be transcribed from redundant genes, some of which are within transposon IS1167 elements (Table (Table1).1). Our initial attempts to identify pneumococcal sRNAs by cDNA cloning were not successful and cataloged numerous rRNA and tRNA degradation products (data not shown). Subsequently, we turned to bioinformatic predictions and found that 28% (11 of 40) of the sRNAs predicted by one of these methods (41) could be detected by Northern blotting (Fig. 1A and B and Table Table1;1; see also Table S3 in the supplemental material). Analogous approaches have shown similar success rates recently in identifying sRNAs in other Gram-positive bacteria (43, 55, 62, 71). CcnB to -E were not picked up in the bioinformatic analysis (41) or included among the 40 candidate sRNAs tested in this study. The inclusion of CcnA, but not CcnB to -E, in this set probably occurred, because a short segment of 34 nt in ccnA matched a corresponding segment in Streptococcus agalactiae, which was used as a BLAST partner but lacked intact ccnA to -E genes. Thus, pneumococcus expresses at least 14 sRNAs, despite an absence of homologues to Hfq and other known RNA chaperones (see the introduction). Hfq homologues are also absent in other Streptococcus species in which sRNAs have been detected, such as Streptococcus pyogenes (55, 66).
Of the 11 pneumococcal sRNAs analyzed here, 5 showed differential expression: relative amounts of Spd-sr12, Spd-sr14, Spd-sr17, and Spd-sr54 decreased as cells entered stationary phase, and the CcnA (Spd-sr56) amount increased in response to CSP (Fig. (Fig.1).1). The mechanisms and physiological function for these changes, if any, remain to be determined. Of the three sRNAs validated as independently expressed by 5′-end determinations and ectopic expression (Fig. (Fig.2,2, ,3,3, and and5),5), Spd-sr17 was transcribed as a monocistronic operon, whereas Spd-sr37 and CcnA were transcribed as independent sRNAs and the regions specifying these sRNAs were also cotranscribed as parts of operons with adjacent genes (Fig. (Fig.2).2). Consequently, deletion/insertion mutations in the spd-sr37 or ccnA region increased the relative transcript amounts of the upstream and downstream tRNA modification genes or upstream genes encoding putative GNAT family acetyltransferases, respectively (Fig. (Fig.2).2). Ectopic expression of Spd-sr37 and CcnA failed to reduce relative transcript amounts of these adjacent genes, indicating that the mutations were cis-acting, rather than trans-acting, through absence of sRNA expression (see Results). These results indicate that disruption of some pneumococcal sRNA genes could lead to phenotypes caused by changes in transcription of adjacent genes. Whether the genetic linkage of these sRNA genes indicates some functional relationship to the cotranscribed genes remains to be determined. On the other hand, the observation that Spd-sr37 or CcnA had apparently identical sizes whether expressed chromosomally or ectopically (Fig. (Fig.3B3B and and5B)5B) argues against these sRNAs arising solely through processing of larger transcripts.
The genes encoding Spd-sr17, Spd-sr37, and CcnA are present only in Streptococcus species. BLAST searches against the nr/nt database (E values of <10−4) showed that single copies of homologues of spd-sr17 and spd-sr37 are present in the genomes of S. pneumoniae, S. sanquis, S. gordonii, and S. suis. Homologues of spd-sr17, but not spd-sr37, are present in S. agalactiae, S. thermophilus, S. mutans, and S. equi. In species other than S. agalactiae and S. suis, spd-sr17 is flanked by the asd gene (Fig. (Fig.2A).2A). Homologues of spd-sr37, but not spd-sr17, are also present in S. pyogenes, where spd-sr37 is flanked by trmU and mutT(Nudix)-like genes similar to those in S. pneumoniae (Fig. (Fig.2B).2B). The conservation of spd-sr17 and spd-sr37 in different Streptococcus species is in agreement with the BLAST partners listed by Livny et al. (41). In contrast, BLAST searches revealed that multiple genes encoding homologues of CcnA to -E are present only in Streptococcus species of the S. mitis group (S. pneumoniae, S. mitis, S. oralis, S. gondonii, and S. sanguis), which also encode homologues of the comCDE competence operon (24).
In E. coli, sRNA pairing sometimes regulates the stability of target mRNAs directly and often modulates translation initiation, which indirectly can lead to changes in the stability of mRNA targets (17, 18, 78); hence, changes in relative transcript amounts in deletion mutants lacking sRNAs or constructs overexpressing sRNAs can lead to target identification (77). In S. pneumoniae D39, we did not detect overt phenotypes or changes in microarray analyses of deletion mutants or strains overexpressing Spd-sr17, Spd-sr37, or CcnA (Fig. (Fig.33 and and5).5). These negative results suggest that we may have failed to test the right growth or stress conditions, and only a single animal model of infection has been evaluated so far for the deletion mutants (see Results). It is also possible that these pneumococcal sRNAs may function largely at the translational level or by directly binding to proteins other than RNA chaperones. Little is known about RNA metabolism in S. pneumoniae, despite its importance as a human pathogen, and a link between translation efficiency and mRNA stability has not been established.
sRNAs structures can be classified into two groups. sRNAs that bind target mRNAs often have 5′ and 3′ stem-loop structures flanking central unpaired regions, whereas sRNAs that bind to proteins other than RNA chaperones often fold into highly paired, extended hairpin structures (reviewed in references 3 and 78). Based on its predicted structure, it seems likely that CcnA may bind to an mRNA target (Fig. (Fig.4C)4C) (22). TargetRNA analysis (72) of CcnA suggests strong possible pairing to leader regions of several mRNAs, including that of comC (Fig. 10), which encodes CSP and is the first gene of the comCDE operon (8, 9). In fact, the comC leader is the only common putative target of each of the CcnA to -E sRNAs. It is more difficult to classify Spd-sr17, because its central region is likely a mixture of paired and unpaired regions (Fig. (Fig.4A).4A). Target RNA analysis (72) turned up very few strong putative targets for Spd-sr17 (data not shown). On the other hand, Spd-sr37 likely forms a highly paired structure lacking unpaired regions, and it seems likely that Spd-sr37 may function by binding to a target protein. An attractive hypothesis is that Spd-sr37 may interact with the tRNA modification or MutT(Nudix)-like enzymes with which the spd-sr37 region is cotranscribed.
The positive regulation of ccnA transcription by the CiaR response regulator (22), the effects of ΔccnA mutations on upstream transcription (see Results), and the redundancy of the CcnA to -E sRNAs prompted us to determine the effects of ectopic CcnA expression in a D39 ΔciaR mutant. We confirmed here some phenotypes previously reported for D39 ΔciaR mutants, including biphasic growth and lower yield (29) and competence induction (35). In addition, we observed additional D39 ΔciaR phenotypes, such as chaining (Fig. (Fig.8).8). Our results confirmed the previous finding (22) that CcnA expression is absolutely dependent on the CiaR response regulator, and we observed that the level of constitutively expressed CcnA was reduced in a ΔciaR mutant compared to a ΔccnA mutant (Fig. 5B and C). This decrease in relative CcnA amount was not due to CiaR regulation, since the Pmal(c) promoter lacks the second, upstream CiaR-dependent promoter P1malM (20, 22). Despite this reduction, constitutive expression of CcnA consistently reversed several phenotypes of the D39 ΔciaR mutant, including reduced growth and a tendency to lyse (Fig. (Fig.6),6), reduced CFU per ml per OD620 unit (Fig. (Fig.7),7), increased chaining at low OD620 values (Fig. (Fig.8),8), and increased competence gene expression and natural competence induction (Fig. (Fig.9).9). Preliminary experiments confirmed that constitutive expression of CcnE also restored growth and reduced chaining of a D39 ΔciaR mutant (data not shown). Together, these results suggest that some of the complex phenotypes of ΔciaR mutants may be caused or exaggerated by the absence of the CcnA to -E sRNAs.
Previously, it was reported that simultaneous knockout of the five ccnA to -E genes did not increase natural competence development in a ciaR+ strain (22). Therefore, the negative regulation of competence by CcnA reported here (Fig. (Fig.9)9) may depend on additional changes in the ΔciaR mutant that are possibly masked in a ciaR+ strain. It is also unknown whether this negative regulation of competence was linked to or independent of the other phenotypic changes in the D39 ΔciaR mutant that were reversed by constitutive CcnA expression (Fig. (Fig.66 to to8).8). Multiple mechanisms were hinted at by microarray analyses, where constitutive CcnA expression in the ΔciaR mutant decreased the relative transcript amounts of 21 genes outside of the normal competence regulon, including five putative transcription regulators (see Table S4 in the supplemental material).
The lack of effect of comC leader mutations on relative comD transcript amount (see Results) is consistent with the previous conclusion that CcnA to -E do not strongly affect competence induction in ciaR+ strains under the conditions tested so far (22). These results also suggest that ComC mRNA may not be a binding target of CcnA to -E. However, additional experiments are needed to determine whether these comC leader mutants are affected at the translational level or under other conditions that affect competence induction. It also remains to be determined whether negative regulators of competence that are in the CiaR regulon, notably the HtrA protease (63), may override effects of CcnA to -E in ciaR+ strains. But the general lack of effects of deletion or overexpression of the three pneumococcal sRNAs studied so far on transcription patterns raises the issue that these sRNAs may operate primarily at the translational level, in response to specific stress conditions that have not yet been tested, or by modulating the activities of specific target proteins.
We thank Kyle Wayne for performing the animal experiments, Krystyna Kazmierczak for the construction of strain IU3373, Donald Morrison (University of Illinois, Chicago) for providing strains and synthetic CSP, and Jean-Pierre Claverys (CNRS-Université Paul Sabatier, Toulouse, France) for the pCEP plasmid. We thank the Center for Genomics and Bioinformatics at Indiana University Bloomington for access to microarray analysis equipment and programs.
This project was supported by grant numbers GM075068 (to A.L.F. and M.E.W.) and AI060744 (to M.E.W.) from the National Institute of General Medical Sciences and the National Institute of Allergy and Infectious Diseases, respectively.
The contents of this report are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.
Published ahead of print on 23 October 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.