Formation of a membraneless compartment regulates bacterial virulence

Abstract

The RNA-binding protein CsrA regulates the expression of hundreds of genes in several bacterial species, thus controlling virulence and other processes. However, the outcome of the CsrA-mRNA interactions is modulated by competing small RNAs and other factors through mechanisms that are only partially understood. Here, we show that CsrA accumulates in a dynamic membraneless compartment in cells of E. coli and other pathogenic species. In addition to CsrA, the compartment contains components of the RNA-degrading complex (degradosome), regulatory small RNAs, and selected mRNAs. Formation of the compartment is associated with a switch between promoting and repressing virulence gene expression by CsrA. We suggest that similar CsrA switches may be widespread in diverse bacteria.

Fig. 1. CsrA forms distinct foci structures.

A Schematics of the full-length csrA gene, translationally (a) or transcriptionally (b) fused to gfp. Boxes represent translated genes, and the solid green line represents an untranslated region containing a synthetic ribosomal binding site (RBS). Both fusions are located at the native chromosomal locus of the csrA gene. Black arrows and lines represent the native regulatory region (PcsrA) and 5’UTR of csrA, respectively. B Formation of CsrA-GFP foci in EPEC. EPEC strains containing csrA-gfp and csrA-rbs-gfp (strains NE9077 and NE9121 shown in A) were grown in DMEM to OD₆₀₀ 0.5. Next, bacterial cells were fixed, washed, and visualized by fluorescence microscopy. Scale bar 2 µm. The two right panels show enlarged sections marked in the merge panels. C Quantification of CsrA foci. Bacteria were grown and fixed as in (B). The percentage of foci containing bacteria was determined by microscopy. For each experiment, n ≥ 200 bacteria were analyzed in each sample. Data presented as mean values and standard deviation of three independent experiments, with an unpaired two-tailed t-test p-value = 0.0023 (**) between the compared samples. ND not-detectable. D Distribution of focus sizes. Bacteria were grown and processed as described in (B). Analysis of CsrA-GFP foci area size was done using the NIS Elements software. In each biological repeat, n > 400 bacterial focus areas were measured. The largest focus area detected in the same biological repeat was set as 100%. The relative percentage range of the focus size is indicated in the X-axis. Data presented as mean values and standard deviation of three biological repeats. E Formation of CsrA foci by enteropathogens. Cultures of EPEC, C. rodentium (C.R.), Y. pseudotuberculosis (Y.P.), and V. cholerae (V.C.), expressing CsrA-GFP translational fusion from an inducible plasmid were grown to OD 0.45–0.6. Induction was done in the final hour of growth. In the case of S. typhimurium (S.T), csrA-gfp is translationally fused on the chromosome (strain LA11509). All samples were processed and visualized as in (B) and quantified as in (C). The Y axis represents the average percentage of focus containing bacteria and the error bars represent the standard deviation of 3 different images taken from each sample. n ≥ 200 bacteria. ND non-detectible. F Representative images of the experiment described in (E). Scale bar 2 µm.

Fig. 2. CsrA equilibrium between the focus and cytoplasm compartments.

A Formation and enlargement of CsrA foci over time. EPEC csrA-gfp (strain NE9077) was grown in DMEM to the indicated densities (OD₆₀₀), fixed, washed, and visualized by fluorescence microscopy. Scale bar 2 µm. B Distribution of foci intensity over time. Bacteria were grown as in (A) and images were analyzed using Image J software. Bacteria lacking CsrA foci were eliminated from this analysis. In the subpopulation of foci-containing bacteria, the sum of focus intensities was calculated. For each sample, n > 400 bacterial foci were analyzed. The X-axis represents the relative intensity (%) compared to the focus containing the highest intensity in a single experiment (set as 100%). The Y-axis represents the percentage of foci counted for each intensity range. Data presented as mean values and standard deviation of three biological repeats, with multiple t-tests (unpaired two-tailed) done between OD 0.2 and OD 1 samples, indicating significant differences with a t-test p-value < 0.01 (**) and t-test p-value < 0.001 (***). C The relative CsrA-GFP intensity in the foci and cytoplasm. EPEC expressing CsrA-GFP (strain NE9077) was grown to the indicated densities (OD₆₀₀) and the sum intensity per constant area (correlates with relative concentration) of cytoplasmic and CsrA-GFP focus was determined for each bacterium (n ≥ 100 bacteria for each measurement), using Image J. The X-axis shows the average foci/cytoplasm ratio of relative CsrA-GFP intensities. The Y-axis indicates the percentage of bacteria belonging to the different categories. Data presented as mean values and standard deviation of three independent biological repeats, with multiple t-tests done between OD 0.2 and OD 1 samples, indicating significant differences with an unpaired two-tailed t-test p-value = 0.0002 (***), 0.0001 (***), and 0.0005 (***) for foci/cytoplasm intensity ratios of: 1-5, 6-10, and >10, respectively. D The cytoplasmic CsrA-GFP relative intensities. EPEC was grown to the indicated densities and the relative GFP intensity in the cytoplasm was determined as in (C). The results were normalized to the cytoplasmic GFP signal obtained at OD 0.2. Data presented as mean values and standard deviation of three independent biological repeats, n ≥ 100 bacteria were measured from each sample, ns- not significant (Unpaired two-tailed t-test between OD 0.2 and OD 0.6 p-value = 0.189 and between OD 0.2 and OD 1 p-value = 0.205).

Fig. 3. RNA binding by CsrA drives focus formation.

A Schematics of merodiploid csrA genes carried by EPEC strains. These include csrA wild type (a), or csrA mutant (csrA_R44A) deficient in RNA binding (b), fused to gfp (csrA-gfp and csrA_R44A-gfp, respectively). The second allele is FLAG tagged csrA (csrA-flag). All of the alleles are localized to the native chromosomal site and expressed under the native csrA regulatory region. Boxes represent translated genes, and black lines and arrows represent the native promoters and 5’UTR of csrA, respectively. B CsrA-GFP and CsrA_R44A-GFP and CsrA-FLAG form heterodimers. EPEC strains with the indicated genotypes (strains NE9361, NE9362, NE9142, NE9133) were grown in DMEM to OD 0.5. Next, Bacteria were lysed and subjected to immunoprecipitation with anti-FLAG antibodies, followed by Western blotting using anti-GFP and anti-FLAG antibodies. Stain-free gel of total protein was used as loading control (Control). Cleared lysate (Input), unbound (UB) and immunoprecipitated (IP) fractions were analyzed. C The two strains indicated in (A) were grown as in (B), fixed, and visualized by fluorescence microscopy, Scale bar 2.5 µm. The two right panels show the enlarged sections marked in the middle panels. D Merodiploid EPEC strains expressing wild-type CsrA as well as GFP fused to either wild-type CsrA, or CsrA mutants (L4A and R44A), deficient in RNA binding (NE9133, NE9141, and NE9142), were grown and processed as in (A). Scale bar 1 µm. E Quantification of the CsrA-GFP foci produced by the three merodiploid EPEC strains as described in (D). Data presented as mean values and standard deviation of three independent biological repeats, with an unpaired two-tailed t-test p-value = 0.0011 (**) between the WT vs L4A and p-value = 0.002 (**) between WT and R44A. ND not detectable. F The amount of CsrA-GFP, CsrA_R44A-GFP, and Csra_L4A-GFP in the three merodiploid EPEC strains shown in (D) was determined by Western blot using an anti-GFP antibody. Stain-free gel of total protein was used as loading control (Control).

Fig. 4. The involvement of CsrB and CsrC in CsrA foci formation.

A CsrB and CsrC are localized to the CsrA foci. EPEC expressing native CsrA fused to mScarlet (CsrA-mScarlet, strain LA10114) was transformed with two compatible plasmids: i) a plasmid encoding CsrB, or CsrC, sRNAs tagged with MS2 binding sites (CsrB_MS2-tag, CsrC_MS2-tag, plasmids pLA10176 and pLA10177), and ii) a plasmid expressing MS2-GFP under tet promoter (pLA11940), and constitutively expressed lacI gene. Bacteria were grown in infection-mimicking conditions and treated with IPTG to induce CsrB_MS2-tag and CsrC_MS2-tag. The MS2-GFP protein expression via a TetR-regulated promoter was induced in the last hour of growth. Bacterial cells were then fixed and visualized by fluorescence microscopy. Scale bar 2 µm. B CsrB is essential for the formation of CsrA foci in the early logarithmic growth phase. EPEC expressing CsrA-GFP (strain NE9077), or isogenic mutants deleted of csrB, or csrC, or both (NE9134, NE9135 and NE9137 respectively) were grown in DMEM to the indicated densities. Then, bacteria were fixed and visualized by fluorescence microscopy. Scale bar 1 µm. C Quantification of the experiment described in (B). Data presented as mean values and standard deviation from two independent experiments, t-test (unpaired two-tailed) was used to compare samples of OD 0.2, with a p-value = 0.0019 (**) between wt and ΔcsrB strains, p-value = 0.231 (ns) between wt and ΔcsrC, and p-value = 0.03 (*) between wt and ΔcsrB, ΔcsrC strains. n ≥ 200 bacteria were measured from each sample in all independent experiments. D CsrA levels in the strains used in (B). Proteins were extracted from the cultures shown in (B) and the levels of CsrA-GFP were evaluated by western blot using anti-GFP antibody. The culture densities and genotypes are indicated above the lanes. Total protein staining was used as loading control (LC) and a segment of the gel is shown. E CsrA-GFP levels in wild-type EPEC grown to different growth phases in DMEM. Bacteria were grown to different ODs and levels of CsrA-GFP were assessed by western blot (as in D). Total protein staining was used as loading control (Control) and a segment of the gel is shown. The relative average amount of CsrA-GFP in three biological repeats is shown below a representative blot. F CsrB levels in wild-type EPEC grown to different growth phases in DMEM. Bacteria were grown to different ODs, the total RNA was extracted and the relative amounts of CsrB was measured using RT-PCR. Data presented as mean values and standard deviation of four biological repeats (bars) and P-value (unpaired two-tailed t-test) is indicated.

Fig. 5. Colocalization of CsrA and RNase E condensates.

A Co-visualization of RNase E puncta and CsrA foci. EPEC strain expressing native CsrA fused to mScarlet (CsrA-mScarlet) and native RNase E fused to mNeon (RNase E-mNeon) (strain LA10288) were grown in DMEM to OD 0.3. Next, bacterial cells were fixed and visualized by fluorescence microscopy. Scale bar 1.0 µm. Arrows indicate colocalization (blue) or lack of colocalization (orange) of RNase E puncta and CsrA foci. B Visualization of CsrA-mScarlet focus and RNase E-mNeon punctum in a single bacterium. The experiment was done as in A, Scale bar 1.0 µm. The blue line indicates the location of the CsrA focus overlapping one of the RNase E-mNeon puncta. The pink line indicates the location of an RNase E-mNeon punctum that does not colocalize with the CsrA focus. C Colocalization pattern of the RNase E punctum and CsrA focus (marked with blue and pink lines in (B)). Fluorescence intensity patterns across the lines were determined using Image J. Correlations across the blue line (left panel) and pink line (right panel) are shown. For each panel, the left Y axis refers to CsrA-mScarlet and the right Y axis to RNase E-mNeon. D Quantification of CsrA-RNase E colocalization. Using the methodology shown in (B) and (C) 30 bacteria were analyzed in three biological repeats (10 bacteria in each repeat). The data was plotted and the Y-axis for CsrA shows the Pearson correlation coefficient between CsrA foci and RNase E puncta, while the Y-axis for RNase E shows the correlation between RNase E puncta and CsrA. Data are presented as box and whisker plot (median, box: first and third quartiles, and whisker: minimum and maximum). Statistics were done using an unpaired two-tailed t-test, p-value < 0.0001 (****). E Colocalization of natively distributed CsrA foci with native RNase E puncta (Sample), versus colocalization of simulated random CsrA foci distribution and native RNase E (Random) was determined as described in Fig. S11. The graph compiles the data of four biological repeats. In each repeat, 400–2000 bacteria were analyzed. Data presented as mean values and standard deviation of the four biological repeats, with an unpaired two-tailed t-test p-value = 0.02 (*).

Fig. 6. Analysis of the CsrA foci associated RNAs.

A The RNA content of the CsrA foci. EPEC expressing CsrA-FLAGx3-GFP or wild-type EPEC expressing untagged CsrA (negative control) were grown to OD 0.6 and subjected to the foci-enrichment protocol (Fig. S9), followed by RNA extraction and library preparation. RNA-seq was performed and the sequenced RNA was mapped to the EPEC genome. The experiment was performed in three biological repeats and gene-assigned read counts from the six sequenced libraries were normalized and compared by DESeq2. DESeq2 reports for each of the compared genes the Log₂ fold change (Log₂FC) of the normalized read counts between the CsrA foci and the wt control along with the p-value corrected for multiple hypothesis testing (padj) (Supplementary Data 3). For a gene to be considered statistically significantly enriched in the CsrA foci we applied stringent filtering requiring Log₂FC > 2 and padj<10⁻⁴. The genes that passed this stringent filtering are shown here with their respective Log₂FC (x-axis) and –Log₁₀ (padj) (y-axis) values. The identity of some of the genes and their biological function (manually curated from the literature) is indicated by color coding. RNAs belonging to the LEE1 (Ler, EscE) and LEE7 (GrlA) mRNAs are highlighted in bold fonts. B Representation of the different categories within the foci. A pie chart displays the number of genes detected in the CsrA foci assigned to the different categories. Genes were categorized as in (A). C mRNAs associated with the CsrA foci of EPEC compared to CsrA-associated RNAs in E. coli K12. A Venn diagram showing the genes obtained from the EPEC foci transcriptome (green circle) and E. coli K12 CsrA CLIP-seq analysis (red circle). Only genes common to the two strains were included in this comparison. The mRNAs were defined by their CDS boundaries. D sRNAs associated with the CsrA foci of EPEC compared to CsrA-associated sRNAs in E. coli K12. Like in (C), only sRNAs common to EPEC and E. coli K12 are compared. All of the sRNAs detected in the CsrA focus, but AspX, are known to contain two or more verified or putative CsrA binding sites (i.e., GGA triplet sequence). All of the sRNAs detected only in the CLIP-seq, but GcvB, contain one or no CsrA binding site.

Fig. 7. Switching of CsrA to virulence-repressive mode upon foci expansion.

A Schematics of positive and negative regulation of the LEE operons by CsrA. For simplicity, only the main relevant details are shown. The LEE1 (solid green arrow) encodes for components that form the T3SS export apparatus (green components within the T3SS complex) and for Ler, which positively regulates all the other LEE operons (black, brown, and blue solid arrows). LEE2 and LEE3 encode many subunits of the T3SS basal body (indicated in brown shades), and LEE4 encodes for EscF, EspA, EspB, and EspD, which form the T3SS needle, filament, and translocation pore (blue shades). LEE7 encodes for GrlA (not shown in this scheme), a regulator that activates the LEE1 promoter and thus Ler expression. CsrA directly binds to the LEE4 transcript, promoting the expression of EspD and other LEE4-encoded proteins. Overexpressed CsrA binds to the LEE1 and LEE7 mRNAs, thus blocking the production of GrlA and Ler. The bacterial inner and outer membranes and host cell membrane are indicated (IM, OM and HM, respectively). Dashed arrows represent transcriptional regulation (red), post-transcriptional regulation (purple), and protein production (black). Created using BioRender. Rosenshine, I. (2025) https://BioRender.com/k60n247. B CsrA switches to ler repressive mode at the late logarithmic growth phase. Wild-type EPEC or ΔcsrA (strain NN5898) mutants were grown in DMEM to OD 0.3, or 0.9. Proteins were then extracted and subjected to western blot analysis using anti-Ler and anti-EspD antibodies. EspD is used here as a typical LEE4-encoded T3SS component. When indicated, the csrA mutation was complemented by plasmid expressing CsrA (pCsrA). A Δler mutant was used as a negative control. Total protein was used as loading control and a segment of the gel is shown. C Repression of infectivity at the late logarithmic growth phase. EPEC grown in DMEM to the indicated densities were used to infect HeLa cells. At 30 min post-infection, the cells were fixed and stained with phalloidin-rhodamine (actin, Red) and anti-EPEC antibody (green). Images were recorded using fluorescence microscopy. To score for infectivity, cells containing more than five T3SS-dependent actin pedestals were labeled as infected. In each sample of three independent biological repeats, n ≥ 60 HeLa cells were assessed. Data presented as mean values and standard deviation of three biological repeats, with an unpaired two-tailed t-test, p-value = 0.01 between OD 0.6 and 0.8 (*), and p-value = 0.001 between OD 0.6 and OD 1.0 (**). D Representative images from the experiment described in (C). White arrows indicate attached bacterial microcolonies associated with actin pedestals. Scale bar 10 µm. E CsrB overexpression represses Ler production. Wild-type EPEC harboring csrA-gfp (wt, strain NE9198), or isogenic mutant deleted of csrB (ΔcsrB, strain NE9542) were supplemented when indicated with a plasmid expressing CsrB (pZE-CsrB). Bacteria were grown with or without IPTG (0.1 mM), and upon reaching OD 0.3, proteins were extracted and subjected to western blot analysis using an anti-Ler antibody. A non-specific band (NS) was used as the loading control. The relative amount of Ler in percentage is indicated below the lanes (Expression). This was calculated using Image J, setting the wild-type levels as 100%, and normalized using the intensity of the nonspecific bands. F CsrB overproduction suppresses EPEC infectivity. Hela cells were infected with EPEC ΔcsrB mutant (strain NE9542) supplemented with a plasmid expressing CsrB under IPTG regulated promoter (ΔcsrB/pCsrB). IPTG was added (0.1 mM), or not, to the infection media as indicated. The infected cells were fixed and stained with phalloidin-rhodamine (actin, red) and anti-EPEC (green). Infectivity was scored as described in (C). In each biological repeat, n ≥ 60 HeLa cells were assessed for infection. Data presented as mean values and standard deviation of three independent experiments, with an unpaired two-tailed t-test p-value = 0.007 (**). G Representative images from the experiment described in (F). Arrows indicate clusters of actin pedestals. Scale bar is 10 µm.

Fig. 8. Schematic of the proposed role of CsrA focus in virulence regulation.

A CsrA switches from supporting to repressing T3SS expression. During the early exponential growth phase (left half of the illustrated bacterium), most bacteria either lack CsrA foci or have small foci. In these bacteria, the cytoplasmic CsrA levels are relatively low (represented by the light green shade), allowing Ler expression, which triggers a positive feedback loop leading to the transcription of the LEE operons (blue arrow). The transcripts of these operons are shown collectively in a blue transparent rectangle. In these bacteria, the CsrA levels are sufficient to support the translation of key T3SS proteins in the LEE4 transcript (Pink arrow), leading to T3SS assembly and virulence ON state. When reaching the late exponential growth phase (right half of the illustrated bacterium), the LEE1 and LEE7 mRNAs are localized in the expanding focus. In this compartment, the higher CsrA concentration acts to repress Ler production (pink blunt arrow). This breaks the positive feedback loop, thus leading to a decay in T3SS expression and diminishing infectivity. Created using BioRender. Rosenshine, I. (2025) https://BioRender.com/k60n247. B Model of CsrA focus composition. CsrA dimers bridged by CsrB form a network constituting the focus backbone. The CsrB function is complex, as in addition to its structural function and inhibition of CsrA, it surprisingly also collaborates with CsrA to achieve repression of ler, and possibly other genes. Additional mRNAs and regulatory sRNAs are distributed between the focus and cytoplasmic compartments, and localized to the different compartments based on their affinity to CsrA, number of CsrA binding sites, and abundance. Other RNAs might be recruited to the foci via base pairing with foci-associated RNAs. Some regulatory processes may be executed within the condensates, such as post-transcriptional regulation by CsrA or/and foci-associated sRNAs. Alternatively, these compartments may store or sequester molecules, or function as RNA and/or protein degradation centers. The decay rate and processing of focus-associated RNAs might be executed by the foci-associated RNase E, which is also anchored to the inner membrane. The association of RNase E with the foci might be driven by RNAs that bridge CsrA and RNase E. Created using BioRender. Rosenshine, I. (2025) https://BioRender.com/k60n247.