Silencing of natural transformation by an RNA chaperone and a multitarget small RNA

Abstract

A highly conserved DNA uptake system allows many bacteria to actively import and integrate exogenous DNA. This process, called natural transformation, represents a major mechanism of horizontal gene transfer (HGT) involved in the acquisition of virulence and antibiotic resistance determinants. Despite evidence of HGT and the high level of conservation of the genes coding the DNA uptake system, most bacterial species appear non-transformable under laboratory conditions. In naturally transformable species, the DNA uptake system is only expressed when bacteria enter a physiological state called competence, which develops under specific conditions. Here, we investigated the mechanism that controls expression of the DNA uptake system in the human pathogen Legionella pneumophila We found that a repressor of this system displays a conserved ProQ/FinO domain and interacts with a newly characterized trans-acting sRNA, RocR. Together, they target mRNAs of the genes coding the DNA uptake system to control natural transformation. This RNA-based silencing represents a previously unknown regulatory means to control this major mechanism of HGT. Importantly, these findings also show that chromosome-encoded ProQ/FinO domain-containing proteins can assist trans-acting sRNAs and that this class of RNA chaperones could play key roles in post-transcriptional gene regulation throughout bacterial species.

Keywords: Legionella pneumophila; ProQ/FinO; RNA chaperone; natural transformation; non-coding RNA.

Fig. S1.

Complementation of L. pneumophila Paris lpp0148_TAA mutant. (A) Northern blot analysis of comEA expression in the Paris wild-type strain and lpp0148_TAA mutant transformed with the empty plasmid pX3, the pX3 plasmid carrying the lpp0148 gene alone (pX3-0148), or the putative lpp0148-0149 operon (pX3-0148-9). (B) Determination of the transformation frequency of the same strains. Two clones for each of the complementing plasmids were tested (#1 and #2). Bacteria were exposed to transforming DNA and then plated on selective and nonselective media. The transformation frequency is the ratio of the number of cfus counted on selective medium divided by the number of cfus counted on nonselective medium. Transformation frequency for the Paris strain was below the detection limit (<1.10⁻⁹). Error bars represent SE of experimental determination of cfu counts.

Fig. 1.

The intergenic sRNA RocR directly interacts with the ProQ/FinO domain of Lpp0148 and represses competence. (A) RNA fold predicted secondary structure of the intergenic 66-nt-long sRNA RocR. Bases are colored according to base-pairing probabilities (see color scale). See also Fig. S2. (B) Northern blot analyses of comEA and RocR expression in L. pneumophilla Paris wild-type, lpp0148_TAA, and ΔrocR strains at an OD₆₀₀ of 0.8 at 37 °C. (C) Natural transformability of the same strains as in B. Error bars represent SD from the mean of three independent experiments. (D) EMSAs of RocR–Lpp0148 complexes. RocR full size, 5′ (SL1 + 2), or 3′ (SL3) parts were incubated with increasing concentrations of Lpp0148 full size or its ProQ/FinO domain in the presence of unlabeled tRNA in excess and run in a native acrylamide gel.

Fig. S2.

RocR is the cognate sRNA of Lpp0148 and represses competence (related to Fig. 1). (A) Determination of the principal RNA partner of Lpp0148 by the RIP-seq method in L. pneumophila Paris. Exponentially growing bacterial culture was fixed with 1% formaldehyde for 30 min. Lpp0148 was immunoprecipitated using specific affinity-purified antibodies, and bound RNAs were extracted and purified. cDNA libraries were prepared following 3′-end polyadenylation and 5′-end RNA adapter ligation and sequenced. The graph shows the number of reads per nucleotide obtained in two biological replicates at the lppnc0692 locus. The locus is schematized under the graph, with the enriched 66 nt highlighted in red and the position of the stem-loop structures predicted by RNA-fold (SL1, SL2, and SL3). (Right) The RNA fold structure prediction of this enriched 66-nt sRNA. (B) Complementation of the ΔrocR mutant in the L. pneumophila Paris strain. The wild-type, lpp0148_TAA, and ΔrocR strains were transformed with the empty plasmid pX3 or the pX3 plasmid carrying the rocR gene (pLLA21). The transformability of the resulting strains was determined and the quantity of comEA and RocR was analyzed by Northern blot at an OD₆₀₀ of 0.8 at 37 °C. For transformation experiments, bacteria were exposed to transforming DNA and then plated on selective and nonselective media. The transformation frequency is the ratio of the number of cfus counted on selective medium divided by the number of cfus counted on nonselective medium. Transformation frequency for the Paris strain was below the detection limit (<1.10⁻⁹). Error bars represent SD from the mean of three independent experiments. Differences in transformation frequencies were considered significant when P values of Welch’s t tests on log-transformed data were below 0.05.

Fig. S3.

Conservation of RocR in all sequenced species of Legionellales (related to Fig. 1). The 10 sequenced species of Legionellales were aligned at the rocR locus with the MAUVE software (http://darlinglab.org/mauve/mauve.html). Bottom shows a zoom-in on the rocR sequence from the L. pneumophila Paris strain with the percentage of sequence identity as well as the position of the three stem loops of RocR.

Fig. 2.

Mutations of the ProQ/FinO domain of Lpp0148 impair its ability to repress competence and to stabilize RocR. (A) Diagram of the L. pneumophila Lpp0148 protein. The ProQ/FinO PFAM domain (PF04352) is shown in red. Loss-of-function mutations are indicated as downward-facing black triangles or hatched box for mutations resulting in a frameshift. See also Fig. S4. (B) Western blot analysis of Lpp0148 and Northern blot analysis of the competence-induced comEA gene in the L. pneumophila JR32 wild-type strain and its mutant derivatives (Δlpp0148 and mutated lpp0148 alleles). A cross-reacting band and the 5S rRNA were used as loading controls for Western blot and Northern blot, respectively. (C) Northern blot analysis of the decay of RocR following transcription inhibition with rifampicin (100 µg/mL) at an OD₆₀₀ of 0.8 in the L. pneumophila JR32 wild-type strain and its mutant derivatives.

Fig. S4.

Hypercompetent mutants obtained by random mutagenesis of lpp0148 in the L. pneumophila JR32, pXDC91 strain (related to Fig. 2). (A) The two diagrams above the table show the comEA JR32 wild-type locus and the transcriptional fusion comEA-gfp borne by the pXDC91 (CmR). The Inset at the top right corner explains the symbols used. (B) The table shows the different mutants obtained. For each single mutant, 10-fold dilution of a cell suspension at an OD₆₀₀ of 1 was spotted on CYE + Cm and incubated for 3 d at 37 °C. The colonies under UV exposure are shown on the Right. (C) Effect of different lpp0148 mutations on the transformation efficiency of the JR32 pXDC91 strain. Bacteria were exposed to transforming DNA and then plated on selective and nonselective media. The transformation frequency is the ratio of the number of cfus counted on selective medium divided by the number of cfus counted on nonselective medium. Transformation frequency for the Paris strain was below the detection limit (<1.10⁻⁹). Error bars represent SD from the mean of three independent experiments.

Fig. 3.

Lpp0148 and RocR destabilize the comEA mRNA by base pairing. (A) Strand-specific read coverage of three competence loci obtained by RIP-seq with anti-Lpp0148 antibodies in L. pneumophila Paris wild-type (green), lpp0148_TAA (red), and ΔrocR (pink) strains. (B) Predicted duplex formation between RocR and the mRNA of Lpp0148-repressed genes. (C) comEA mRNA half-life determination by RT-qPCR. Decay of comEA was followed after transcription was stopped with rifampicin at an OD₆₀₀ of 0.8. Data, expressed as the relative amount of mRNA before the addition of rifampicin (t = 0), were fit to a first-order exponential decay and half-lives were calculated from three quantifications. (D) Cultures of different JR32 strains of L. pneumophila harboring the plasmid carrying the comEA-gfp fusion with wild-type (pXDC91) or mutated (pLLA27-28-29) RocR box were analyzed by flow cytometry. GFP levels were measured in 5.10⁵ cells per sample; error bars represent the SD. The strains are JR32 pXDC91 (RocR box WT), JR32 Δlpp0148 pXDC91, JR32 pLLA27 (RocR box m3), JR32 rocRm3 pLLA27, JR32 rocRm4 pXDC91, JR32 rocRm4 pLLA28 (RocR box m4), JR32 rocRm5 pXDC91, and JR32 rocRm5 pLLA29 (RocR box m5).

Fig. 4.

Lpp0148 and RocR control natural competence development. (A) The L. pneumophila Paris WT strain transiently develops competence during growth at 30 °C. Expression of comEA was followed using a comEA-gfp transcriptional fusion carried by pXDC91 (GFP/OD₆₀₀, green line) and natural transformability (red circle, triangle, and square) determined at different time points during growth (OD₆₀₀, blue line). Error bars on natural transformation efficiencies data represent SE. (B) Natural transformability during growth at 30 °C of the L. pneumophila Paris WT, lpp0148_TAA, and ΔrocR strains. (C) Expression of Lpp0148 and RocR decrease at the onset of the transformability phase at 30 °C. Expression of Lpp0148 was analyzed by Western blot, and expression of comEA and RocR was determined by Northern blot analysis. A cross-reacting band and the 5S rRNA were used as loading controls for the Western blot and Northern blot, respectively. (D) comEA mRNA half-life determination by Northern blot analysis before (OD₆₀₀ of 0.9) and during (OD₆₀₀ of 2.5) the competence phase. Decay of comEA was followed after transcription was stopped with rifampicin at the indicated OD. The 5S rRNA was used as a loading control.

Fig. 5.

The ProQ/FinO domain-containing proteins are widespread in Proteobacteria. Shown are the phylogenetic analysis of ProQ/FinO domain-containing proteins (see also Dataset S1 and Fig. S5) and the maximum likelihood phylogenetic tree of the 674 protein sequences containing a ProQ/FinO domain found in 2,775 complete prokaryotic proteomes (80 amino acid positions were used). Colors correspond to taxonomic main lineages.

Fig. S5.

Phylogenetic analyses of ProQ/FinO domain-containing proteins (related to Fig. 5 and Dataset S1). (A) Taxonomic repartition of the 2,775 complete prokaryotic proteomes available at NCBI regarding the ProQ/FinO domain-containing proteins. (B) Number of ProQ/FinO domain-containing proteins per proteome in the 2,775 complete prokaryotic proteomes available at NCBI. (C) Distribution according to the encoding gene location. ProQ/FinO domain-containing proteins are colored in black, orange, or purple when their coding gene is located on the chromosome, a plasmid, or a secondary chromosome, respectively. (D) Architecture of the different ProQ/FinO domain-containing proteins. For each cluster (numbers on the tree), multiple members were aligned using BlastP, and conserved domains are shown. The C-terminal part of the proteins (after the ProQ/FinO domain) was aligned to extract the possible region of high similarity.

Fig. S6.

Model of regulation of natural transformability in L. pneumophila. When cells are noncompetent, RocC stabilizes RocR and this ribonucleoprotein complex can bind the mRNAs of genes of the competence regulon via a conserved 6-nt sequence called RocR box and promote their degradation. Under competence-inducing conditions (for example, at the end of the exponential phase at 30 °C), RocC expression decreases. This triggers the destabilization of RocR and thus the stabilization of the mRNAs of the competence genes. These are then translated, the DNA uptake system is assembled, and horizontal gene transfer by natural genetic transformation can occur. IM, inner membrane; OM, outer membrane.