Bacterial Argonaute Proteins Aid Cell Division in the Presence of Topoisomerase Inhibitors in Escherichia coli

ABSTRACT Prokaryotic Argonaute (pAgo) proteins are guide-dependent nucleases that function in host defense against invaders. Recently, it was shown that TtAgo from Thermus thermophilus also participates in the completion of DNA replication by decatenating chromosomal DNA. Here, we show that two pAgos from cyanobacteria Synechococcus elongatus (SeAgo) and Limnothrix rosea (LrAgo) are active in heterologous Escherichia coli and aid cell division in the presence of the gyrase inhibitor ciprofloxacin, depending on the host double-strand break repair machinery. Both pAgos are preferentially loaded with small guide DNAs (smDNAs) derived from the sites of replication termination. Ciprofloxacin increases the amounts of smDNAs from the termination region and from the sites of genomic DNA cleavage by gyrase, suggesting that smDNA biogenesis depends on DNA replication and is stimulated by gyrase inhibition. Ciprofloxacin enhances asymmetry in the distribution of smDNAs around Chi sites, indicating that it induces double-strand breaks that serve as a source of smDNA during their processing by RecBCD. While active in E. coli, SeAgo does not protect its native host S. elongatus from ciprofloxacin. These results suggest that pAgo nucleases may help to complete replication of chromosomal DNA by promoting chromosome decatenation or participating in the processing of gyrase cleavage sites, and may switch their functional activities depending on the host species. IMPORTANCE Prokaryotic Argonautes (pAgos) are programmable nucleases with incompletely understood functions in vivo. In contrast to eukaryotic Argonautes, most studied pAgos recognize DNA targets. Recent studies suggested that pAgos can protect bacteria from invader DNA and counteract phage infection and may also have other functions including possible roles in DNA replication, repair, and gene regulation. Here, we have demonstrated that two cyanobacterial pAgos, SeAgo and LrAgo, can assist DNA replication and facilitate cell division in the presence of topoisomerase inhibitors in Escherichia coli. They are specifically loaded with small guide DNAs from the region of replication termination and protect the cells from the action of the gyrase inhibitor ciprofloxacin, suggesting that they help to complete DNA replication and/or repair gyrase-induced breaks. The results show that pAgo proteins may serve as a backup to topoisomerases under conditions unfavorable for DNA replication and may modulate the resistance of host bacterial strains to antibiotics.

. Expression of SeAgo and LrAgo in E. coli. (A) Analysis of pAgo expression in E. coli by Western blotting. Protein samples were obtained from the wild-type or rec-minus E. coli strains at 4.5 hours of growth. The proteins were separated by 4-20% gradient denaturing SDS-PAGE, transferred to a nitrocellulose membrane and visualized by Western-blotting with His-tag specific antibodies (see Materials and Methods for details). The turquoise and pink arrows show positions of SeAgo and LrAgo, respectively. (B) Growth of E. coli strains in the absence or presence of Cfx depending on expression of SeAgo (uniduced, 0% Ara; induced with 0.01% or 0.1% of Ara). E. coli cells were transformed with empty pBAD or pBAD encoding SeAgo and were grown at 30 °C in a plate reader without Cfx or with Cfx (0.5 ng/mL) and indicated concentrations of arabinose. Averages from 3 technical replicates are shown.

Fig. S2
. Analysis of the CFU numbers for E. coli strains lacking or containing pAgos and grown in the absence and in the presence of Cfx. The samples were taken from E. coli cultures grown for 4.5 hours in the absence or in the presence of Cfx (Fig. 3A), and CFU numbers were determined by plating their serial dilutions on LB agar plates without Cfx. Means from three biological replicates are shown, individual data points from each measurement are indicated (colored pairwise for each independent experiment).

Fig. S3.
Effects of pAgo expression on E. coli cell morphology. E. coli transformed with a control empty pBAD plasmid or pBAD ncoding pAgos were grown in the absence (top three rows) or in the presence (bottom three rows) of ciprofloxacin. The expression of pAgos was induced with 0.01% Ara. The samples were taken at 4.5 hours from the cultures shown in Fig. 3A. Fluorescence microscopy after acridine orange staining. The scale bar is 10 µm. The first column corresponds to images presented in Fig. 4, the other four images in each row are additional microscopic fields of view for the same samples. coli strains lacking of containing pAgos. All cultures were grown in 0.5 liter of LB in the absence and in the presence of Cfx (0.3 ng/mL) and 0.01% Ara, and OD 600 was monitored each 30 minutes. The dashed lines indicate 5.5 h and 12.5 h time points used for purification of pAgo-associated smDNAs. (B) Analysis of smDNAs purified from pAgos. SmDNAs isolated from pAgos after one-step purification (using Co 2+ -affinity resin) were treated with alkaline phosphatase to remove pre-existing 5'-phosphates, labeled with γ-P 32 -ATP and polynucleotide kinase and separated by 19% denaturing urea PAGE. The marker lane (M) contains 5'-labeled DNA oligonucleotides of indicated lengths.

Fig. S5. Asymmetry of smDNA distribution around Chi sites at the stationary phase of growth.
SmDNA analysis was performed as described in Fig. 7, but for the stationary cultures (12.5 hours of growth). Averages from two replicate experiments are shown. (A) Metaplots of the densities of smDNAs around Chi sites. SmDNA numbers were independently calculated for the DNA strands co-oriented (green, F) and oppositely oriented (gray, R) relative to the Chi sequence (5'-GCTGGTGG-3') for all Chi sites in both genomic strands and smoothened with a 400 bp sliding window. (B) Metaplots of the normalized densities of smDNAs around cooriented Chi sites for E. coli strains grown in the absence and in the presence of Cfx (the data correspond to green smDNA profiles in panel A). Arrows indicate relative differences between the background and minimal densities of smDNAs around the Chi sites. Metaplots of the relative densities of smDNAs were calculated for 1833 GCSs identified in the reference genome (averages from two replicate experiments, calculated independently for the 5.5 h and 12.5 h time points). SmDNA densities were calculated for both DNA strands around GCSs for +Cfx and -Cfx conditions independently, and then the +Cfx density was divided by the -Cfx density. The resultant relative density was smoothened with a 2 kb sliding window. Metaplots of the relative densities of smDNAs around Chi sites adjacent to a downstream GCS (dark gray curves, 192 sites in total) and Chi sites adjacent to an upstream GCS (yellow curves, 191 sites in total) (averaged for two replicate experiments). A Chi site and a GCS were considered adjacent if there were no co-oriented Chi sites in between. SmDNA density was calculated for DNA strands co-oriented with Chi sites for the +Cfx and -Cfx conditions independently, and then the +Cfx density was divided by the -Cfx density. Resultant relative density was smoothened with a 1 kb sliding window. Blue rectangles mark the region (from the Chi sequence to +5 kb) used to quantify the relative enrichments of smDNAs in panel B. (B) Quantification of relative enrichments of smDNAs at the 3'-sides of Chi sites for Chi sites adjacent to a downstream GCS (DS, carmine) and Chi sites adjacent to an upstream GCS (US, violet). The error bars represent mean values ± STD for the same sets of Chi sites as in panel A. Enrichments were compared by a two-sided t-test. P-values of <0.05, <0.005 and <0.0005 are indicated with one, two, and three asterisks, respectively. The most significant differences in the smDNA enrichment are observed for SeAgo at the logarithmic phase of growth (5.5 h, pvalue 1.1e-4) and for LrAgo at the stationary phase of growth (12.5 h, p-value 5.9e-6).

Fig. S8. Analysis of smDNA distribution relative to transcription units in E. coli. (A)
Densities of smDNAs within genes co-directed or reversely directed relative to replication. For each gene, the amounts of smDNAs were calculated independently for the sense and antisense DNA strands, normalized by the gene length and expressed as RPKM (reads per kilobase per million aligned reads in the library). The data were independently averaged for each of the four types of orientation (sense and antisense strands for co-directed and reversely directed genes). (B) Densities of smDNAs in intergenic regions for convergent, divergent, and co-oriented gene pairs (located either in the plus or in the minus genomic strands). Only <500 bp intergenic regions were taken into account. SmDNA numbers were independently averaged for each of the four types of gene orientation (convergent, divergent, and two directions of co-oriented genes), normalized by the length of each intergenic region and expressed in RPKM. The data are shown as box plots (center line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range).   * The enrichment of plasmid-derived smDNAs is calculated as a ratio of the observed number of plasmid-mapped reads to the expected number of plasmid mapped reads. The expected number of reads = (genome-mapped reads + plasmid-mapped reads)×(plasmid size × copy number)/(genome size + plasmid size × copy number)