Patatin-like phospholipase CapV in Escherichia coli – morphological and physiological effects of one amino acid substitution | npj Biofilms and…

CapVQ329R inhibits swimming motility of E. coli MG1655

The dinucleotide cyclase DncV synthesizes cAMP-GMP to inhibit rdar biofilm formation and motility in the animal commensal strain E. coli ECOR3149. To assess whether the downstream cAMP-GMP receptor, the patatin-like phospholipase CapV, has a function on its own, we expressed CapV in the heterologous host MG1655, an E. coli K-12 derivative which is not known to synthesize cAMP-GMP nor to harbor DncV. Besides wild-type CapV, we overexpressed its variant CapVQ329R, which had been derived by cloning a respective mutated DNA fragment as described in Supplementary Results. To this end, we observed that expression of the variant CapVQ329R caused suppression of swimming motility, whereas overexpression of the wild-type protein CapV did not alter the apparent motility in semi-solid tryptone broth (TB) agar (as described in Supplementary Results; Fig. 1a and Supplementary Figs. 13). TB medium promotes motility of E. coli MG1655 compared to LB medium in the semi-solid agar plate assay50. To further characterize CapVQ329R-induced E. coli MG1655 swimming inhibition, we assessed the production of cell-associated flagella and flagellin upon overexpression of CapVQ329R compared to CapV. Visualization of bacterial cells by transmission electron microscopy (TEM) showed that overexpression of CapVQ329R compared to wild-type CapV dramatically reduced the total number of flagella-producing cells as well as the number of flagella per cell after 6h incubation at 37C (Fig. 1c, d). In agreement, visualization of flagella by Leifson staining upon CapVQ329R overexpression showed cells with intact flagella up to 4h incubation at 37C, but almost no cell with flagella after 6h (Fig. 1c, e). In congruence with the analysis by TEM and light microscopy examination of Leifson staining, we observed inhibition of production of cell-associated extracellular flagellin in a protein gel upon overexpression of CapVQ329R after 6h (Fig. 1f, g). Initial analysis of differential gene expression by qRT-PCR indicated 8% downregulation of expression of the class 1 flagella regulon gene flhD encoding a subunit of the major regulator FlhD4C2, and less than 50% downregulation for the representative genes of class 2 fliA encoding the flagella specific sigma factor and class 3 fliC encoding the subunit of flagella, upon CapVQ329R compared to CapV overexpression. Compared to the vector control, flhD, fliA and fliC were downregulated 28, 83, and 73%, respectively, upon overexpression of CapV. Thus, the expression of CapVQ329R interferes with flagella expression beyond the flagella regulon cascade and can affect, for example, depolymerization or degradation of flagella. Cumulatively, these results indicate that CapVQ329R suppressed swimming motility of MG1655 by post-translationally inhibiting the production of flagellar filaments gradually during the growth phase, while initially production of functional flagellar filaments had been observed.

a, b Flagella-dependent swimming motility of wild-type E. coli MG1655 vector control (VC) and upon overexpression of wild-type CapV, its mutants (red) and mutants of CapVQ329R (blue). In total, 3l of a OD600=5 cell suspension were inoculated into soft agar plates containing 1% tryptone, 0.5% NaCl, and 0.25% agar, and the swimming diameter was measured after 6h at 37C. c Flagella production of a representative E. coli MG1655 VC cell and upon overexpression of CapV, CapVQ329R (Q329R) and CapVQ329R/D197A (D197A) as observed by TEM. d Quantification of the number of flagella per cell upon overexpression of CapV and CapVQ329R after visualization by TEM (c). The number of evaluated cells n=20. Cells were grown in TB medium for 6h at 37C. e Production of surface-associated flagellin of E. coli MG1655 VC, upon overexpression of CapV and CapVQ329R. f Assessment of flagellin subunit FliC expression by colloidal Coomassie staining from E. coli MG1655 culture supernatants after shearing of flagella upon expression of CapV, CapVQ329R, CapVQ329R/D197A and CapVQ329R/S64A. Cells were grown in TB medium at 37C for 6h. g Assessment of flagellin subunit FliC expression over time in E. coli MG1655 overexpressing CapVQ329R. Samples were harvested at different time points in the growth phase for Western blot analysis of FliC. lc, loading control. h Proposed development of filamentation and flagella inhibition of E. coli MG1655 upon CapVQ329R expression over time. Bars represent mean values from three biologically independent replicates with error bars to represent SD. Differences between mean values were assessed by two-tailed Students t test: ns, not significant; *P<0.05, **P<0.01, and ***P<0.001 compared to E. coli MG1655 VC. Vector control VC=pBAD28. pCapV = CapV cloned in pBAD28; Q329R=CapVQ329R cloned in pBAD28. S64A=CapV Q329R/S64A cloned in pBAD28. D197A=CapV Q329R/D197A cloned in pBAD28. R27A=CapVQ329R/R27A cloned in pBAD28. G24A/G25A=CapVQ329R/G24A/G25A cloned in pBAD28.

Significantly, upon overexpression of CapVQ329R in E. coli MG1655 TEM and light microscopy demonstrated not only the loss of flagella, but the concomitant development of long thin filamentous cells (Fig. 1c, e). In contrast, E. coli MG1655 cells overexpressing wild-type CapV were only slightly elongated compared to the control (Fig. 1c, e).

Assessment of the temporal development of cell filamentation throughout the growth phase upon induced expression of CapVQ329R by light microscopy after Leifson staining indicated that filamentation did not initiate before 2h after commencement of CapVQ329R expression (Figs. 1e and 2a, b). Subsequently, though the cell length and the frequency of filamentation dramatically increased, whereby after 3h almost all cells displayed as short filaments around 6 times the length of standard rod cells (Fig. 2a, b). After 4h of induction, CapVQ329R expressing E. coli MG1655 cells were on average 25 times longer than control cells. After 6h of induction, CapVQ329R expressing E. coli MG1655 cells were on average more than 50 times longer than the standard rod-shaped cell (Fig. 2, b). Those long cells did not show any movement, while shorter filaments up to approximately 20 times the length of standard E. coli cells, although rare, still showed active swimming motility (Supplementary Movie S1). At the opposite, overexpression of wild-type CapV only slightly increased the cell length compared to E. coli MG1655 control. On note, after 22h induction of CapVQ329R expression by 0.1% L-arabinose, short rod-shaped motile cells dominated again, which suggested filaments to restart cell division after CapVQ329R expression had diminished due to L-arabinose depletion. In line with this hypothesis, induction of CapVQ329R production by 0.2% l-arabinose did not cause reversion to rod-shaped cells nor showed the cells any movement after 22h (Fig. 2a, c and Supplementary Movie S2). Upon transfer of those filamentous cells to fresh TB medium without L-arabinose, however, the emergence of rod-shaped cells was again observed (Fig. 2c). A scheme of this developmental process leading to filamentation with consecutive loss of flagella upon expression of CapVQ329R is displayed in Fig. 1h.

a Light microscopy pictures of cell morphology of E. coli MG1655 VC and upon overexpression of CapV and CapVQ329R (Q329R) in TB medium at different time points at 37C. b Quantification of cell length of E. coli MG1655 VC and upon overexpression of CapV and CapVQ329R in TB medium at different time points. The quantification is based on results from at least three independent experiments with the assessment of 70 cells from each group. c Cell morphology 3h after addition of fresh TB medium to filamentous E. coli MG1655 cells overexpressing CapVQ329R. Arrowheads indicate invaginations at proposed future division sites. d, e, f Assessment of cell length upon overexpression of CapV, CapVQ329R and CapVQ329R derivatives in E. coli MG1655 upon induction with different l-arabinose concentrations. Light microscopy pictures (d), quantification of cell length (e) and protein expression level (lc=loading control) (f) upon induction with 0.01% and 0.1% l-arabinose in TB at 37C for 4h. The quantification is based on results from at least three independent experiments with the assessment of 70 cells from each group. Bar, 5m. Vector control VC=pBAD28. pCapV=CapV cloned in pBAD28; Q329R=CapVQ329R cloned in pBAD28. G24A/G25A=CapVQ329R/G24A/G25A cloned in pBAD28. R27A=CapVQ329R/R27A cloned in pBAD28. S64A=CapVQ329R/S64A cloned in pBAD28. D197A=CapVQ329R/D197A cloned in pBAD28.

A positive correlation of the filamentation phenotype of the E. coli MG1655 cells with CapVQ329R production level was demonstrated using increasing L-arabinose concentrations. When incubated for 4h with 0.01% l-arabinose, the low-level CapVQ329R expression created a heterogenous cell population displaying no or restricted cell filamentation (Fig. 2d, e). In contrast upon incubation with 0.1% l-arabinose all cells became filamentous as observed previously. Concomitantly, the CapVQ329R expression level was comparable to the CapV expression level excluding that the observed morphological changes were due to significantly different protein expression levels. As expected, CapV expression was higher upon induction with 0.1% l-arabinose than with 0.01% l-arabinose (Fig. 2f). The 6xHis-tag added to the C-terminus of the protein to detect protein production level did not alter the proficiency of CapVQ329R to induce filamentation (Supplementary Fig. 2g). Cumulatively, filamentation and motility repression are caused by the Q320R substitution in CapV.

Blast search with CapV from E. coli ECOR31 showed that CapV homologs with >60% identity are not only found in individual E. coli and V. cholerae strains, but are widely distributed among gamma-proteobacteria, including Yersinia, Salmonella, Pseudomonas, Shewanella, and Klebsiella species (Supplementary Fig. 4a). Phylogenetic analysis of representative CapV homologs supported classification into four different subgroups (Supplementary Fig. 4b). Alignment of the amino acid sequences of those CapV homologs showed that Q329 is a nearly invariant amino acid even among distantly related CapV proteins (Supplementary Fig. 4a). Q329 is, though, not required for catalytic activity and not part of other characteristic patatin-like phospholipase A2 (PNPLA) consensus motifs and the PNPLA core domain which is restricted to aa 19210 (Prosite) in the 361 aa long protein. In order to clarify the position of R329 in the protein, we generated a structural model of CapV using the closest structural homolog from the PDB database as a template, the lysophospholipase-like protein FabD from Solanum cardiophyllum (PDB: 1oxwC; Fig. 3a) which shows 22% amino acid identity with CapV. According to this model, R329 is located within helix 12, the second last helix of CapVQ329R in the context of the RARGRR329 sequence pointing outward with no obvious change in the overall structure of the monomer or potential oligomer assembly to be observed.

a Predicted structural model of the CapV from E. coli ECOR31 shown as ribbon representation. The structural model was built with the I-TASSER server, the result was processed with SWISS-MODEL. The model was based on the coordinates of the 22% identical protein FabD from Solanum cardiophyllum (PDB: 1oxwC). b The graphical representation and schematic indication of the positions of the conserved motifs (indicated by the green bar) and putative active site residues S64 and D197 (marked by red stars) in the PNPLA domain of the 361 aa CapV from E. coli ECOR31 (from L19 to F210). Black arrow, Q329. The graph was assessed by ExPASy_Prosite. c Sequence alignment of CapV from E. coli ECOR31 and selected known phospholipases from other species establishes the conserved motifs of the PNPLA domain, GG-G-x-[K/R]-G, G-x-S-x-G, and DG-[A/G], boxed in black, green, and purple, respectively. Entirely conserved residues are shown in white on a red background. Conserved residues are boxed. Putative catalytic residues of CapV are indicated with filled red triangles. The residues in CapVQ329R mutated to alanine are marked with red asterisks above the sequence. The consensus sequence at the bottom indicates in uppercase letter residues with 100% identity and in lowercase letter residues with higher than 70% conservation. Alignment was performed using CLUSTALW, and the result was processed with ESPript 3.0. Sequence identity as in the Methods section. d 32P-cAMP-GMP-DRaCALA of E. coli cell lysates expressing CapV, CapVQ329R, and CapVQ329R/D197A. VC=pBAD28; pCapV=CapV cloned in pBAD28; Q329R=CapVQ329R cloned in pBAD28. D197A=CapVQ329R/D197A cloned in pBAD28. e Assessment of affinity for cAMP-GMP of CapV, CapVQ329R, and CapVQ329R/D197A expressed in E. coli MG1655. 32P-cAMP-GMP was mixed with twofold dilutions of cell extracts starting at a fourfold dilution.

As CapV from V. cholerae46, CapV of E. coli ECOR31 contains a N-terminal canonical PNPLA domain with three main characteristic conserved signature motifs48,51, the phosphate or anion binding motif GG-G-x-[K/R]-G, the esterase box G-x-S-x-G, and the DG-[A/G] motif as part of the catalytic dyad (Fig. 3b, c). The G-x-S-x-G motif includes the conserved nucleophilic serine 64 of the active site characteristic for the phospholipase A2 (PLA2) superfamily48,52.

To investigate if catalysis is required for swimming inhibition and filamentation upon CapVQ329R overexpression, a catalytically inactive S64A variant of the protein (CapVQ329R/S64A) was generated. Compared with CapVQ329R, overexpression of CapVQ329R/S64A equally inhibited swimming motility and induced filamentation (Figs. 1a and 2d, e), demonstrating that the G-x-S-x-G motif of CapVQ329R is not required for the phenotype. However, the substitution of arginine in the GG-G-x-[K/R]-G motif (CapVQ329R/R27A) and aspartic acid of the DG-[A/G] motif (CapVQ329R/D197A) by alanine relieved both the repression of swimming motility and induction of filamentation. Substitution of the two structural glycine residues (CapVQ329R/G24A/G25A) of the GG-G-x-[K/R]-G also partially suppressed swimming motility and induced only a mild filamentous phenotype upon overexpression of the protein (Figs. 1a, b and 2d, e). In summary, the GG-G-x-[K/R]-G and DG-[A/G] motifs of CapVQ329R are required for repression of the swimming phenotype and cell filamentation.

As substitution of the nucleophilic serine 64 did not relieve motility and cell filamentation, we were wondering whether alternative serine residues are involved in the physiological activity of CapVQ329R. To this end, we substituted S33, S113/114, S146, S177, and S206 by alanine residues. Most of these serine residues were selected as they are located close to the catalytic site (Supplementary Fig. 2f). Only serine 206 was required to induce motility inhibition and promote cell filamentation by CapVQ329R (Supplementary Fig. 2f).

Furthermore, we wanted to clarify whether specifically the Q329R mutation is required to induce filamentation. Replacement of Q329 by the other positively charged amino acid lysine still partially repressed the apparent swimming motility and induced filamentation (Fig. 1b), while replacement of Q329 by asparagine retained the wild-type CapV phenotype (Supplementary Fig. 2f, g).

The binding site for cAMP-GMP in CapV has not been identified. The Q329R substitution might alter the binding of cAMP-GMP or other cyclic dinucleotides to CapV. To this end, binding of cAMP-GMP to CapV and CapVQ329R was analyzed by the differential radial capillary action of ligand (DRaCALA) assay. DRaCALA is based on the retention of small molecular compounds at the protein application spot on a nitrocellulose membrane upon binding. The experiment showed that both proteins bound cAMP-GMP with approximately equal affinity while the CapVQ329R/D197A mutant showed diminished binding (Fig. 3d).

Patatin-like phospholipases hydrolyze the sn-2 acyl ester bond of neutral and phospholipids48,53. In order to assess whether overexpression of CapV and CapVQ329R caused significant changes in the lipid profile concomitant with filamentation, we extracted lipids after 4 h of growth extracts to mass spectrometry (Fig. 4). Based on untargeted charged surface hybrid column quadrupole time-of-flight mass spectrometry (CSH-QTOF MS) analysis a total of 326 lipid species were identified (Fig. 4a). Principle component analysis showed distinct classification of samples into groups correlating with the overexpression of the wild-type CapV and CapVQ329R variant protein (Fig. 4b), indicating that the lipid profiles are significantly altered. Subsequently, we applied hierarchal clustering analysis to segregate the samples cumulatively according to overall changes in the individual lipid compounds. Based on the changes in the lipid compounds, the samples can again be classified into distinct groups according to the expressed proteins (Fig. 4f). Lipids known to be most abundant in the E. coli membrane, phosphatidylethanolamines (PE), phosphatidyl-glycerols (PG) and fatty acids (FAs), displayed the highest relative peak intensity (Fig. 4ce and Supplementary Fig. 5). We observed significant changes in the peak intensity of members of phospholipid classes, most abundant membrane components of E. coli such as PEs and PGs, but also of free FAs, lysophospholipids (LPE and LPG), phosphatidylcholines (PCs), ceramides (Cer) and sphingomyelins (SM), although the peak intensity of the latter three classes was at least 100-fold lower (Fig. 4ce and Supplementary Fig. 5). The peak intensity was, however, in none of the cases on the average more than sixfold different. Notably, among the top 50 most significantly altered lipids, PE and PG derivatives with distinct FAs profiles are predominantly represented (Fig. 4f). While CapV overexpression showed downregulation of a restricted group of lipids, a significantly higher number of lipids species were downregulated upon CapVQ329R overexpression. Notably, lipids species were also upregulated upon overexpression of CapV and CapVQ329R suggesting a role of the patatin-like phospholipases in membrane reorganization and/or signaling. Only a few lipid species were distinctively upregulated upon overexpression of CapVQ329R, PE32:2 (16:1 16:1) with two monounsaturated fatty acids and the monounsaturated fatty acid FA18:1. Thus CapV and CapVQ329R might have unique substrate profiles. Whether the observed alterations in the lipid profile are based on distinct residual catalytic activities of the two proteins or indirectly associated with the expression of CapV and CapVQ329R needs to be investigated further. In summary, these results indicate that amino acids in the catalytic motifs are required to induce filamentation and to repress motility. Not or only partially activated CapV and CapVQ329R can alter the lipid profile indicating that CapV is not only a receptor for a second messenger molecule, but might also be involved in alternative second messenger signaling in its nonactivated state.

a The number of identified lipid species in CapV and CapVQ329R induced filamentous cells compared to E. coli MG1655 vector control (VC) by untargeted charged surface hybrid column-quadrupole time-of-flight mass spectrometry (CSH-QTOF MS) analysis. For abbreviation of lipid compounds consult Methods. b Principle component analysis of lipid abundance upon overexpression of wild-type CapV and CapVQ329R variant proteins in E. coli MG1655 compared to VC. Of six samples each, outliers have been removed. ce Alternation and relative abundance of PE (c), PG (d), and FA (e) derivatives by untargeted CSH-QTOF MS analysis. Bars represent mean values from five independent replicates with error bars to represent SD. f Heatmap of selected 50 most significantly altered lipid species built based on hierarchical clustering. Each square represents one sample of each group. The color scale presenting the difference of each log2 transformed peak intensity value to the log2 transformed mean for each lipid species and the percentage of each lipid class is indicated on the right of the heatmap. Heatmap analysis was performed on the Tutools platform (https://www.cloudtutu.com), a free online data analysis website. Differences between mean values were assessed by two-tailed Students t test: ns, not significant; *P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001; black stars in c, d, e and f: compared to MG1655 VC; red stars in f: statistical significance between E. coli MG1655 pCapV and E. coli MG1655 pCapVQ329R. VC=pBAD28. pCapV=CapV cloned in pBAD28; Q329R=CapVQ329R cloned in pBAD28.

During cell division, positioning of the FtsZ cytokinetic ring at the site of constriction between nucleoids is coordinated with chromosome replication, nucleoid segregation and cell elongation31,54,55. Impairment of this process leads to cell division arrest and filamentation, which can be induced by DNA damage and nucleoid occlusion15,27. After completion of cell segregation, the nucleoid subsequently becomes more compact56. To determine the effect of CapV and CapVQ329R overexpression during the cell division process, we analyzed FtsZ-ring positioning in an E. coli K-12 MG1655 derivative with a chromosomally encoded FtsZ-GFP fusion protein57 and the position and shape of the nucleoid with DAPI staining. After induction of CapV and CapVQ329R in liquid medium at 37C for 4h, we immediately subjected the cells to fluorescence microscopy on agarose pads to visualize cell shape, septum, and FtsZ-ring formation and nucleoid location. Overexpressing wild-type CapV in the E. coli K-12 MG1655 FtsZ-GFP strain, we observed clearly visible constrictions that corresponded with a correctly positioned FtsZ ring and a single nucleoid in nondividing cells and two fully replicated and/or segregated nucleoids in dividing cells, respectively (Fig. 5a). Of note, nucleoids upon CapV expression appeared slightly more compact than those in the control.

a Phase-contrast and fluorescence images of FtsZ-GFP expressing cells (E. coli MG1655 derivative BS001 harboring vector control (VC), pCapV and pCapVQ329R). Cells were cultured in TB medium at 37C for 4h, stained with DAPI, and assessed immediately under fluorescence microscopy. Large fragments of unsegregated nucleoids are indicated by white arrows. Bar, 3m. b Quantification of the average distance between two adjacent FtsZ rings upon overexpression of CapV and CapVQ329R, refers to Table 1. c Time-lapse analysis of mCherry-MinC expressing cells (E. coli MG1655 derivative PB318 harboring VC, pCapV and pCapVQ329R). A representative elongated cell is displayed. Graphs on the right of fluorescence images display the line profiles of fluorescent signals emanating from the cell. Arbitrary fluorescent units are obtained, analyzed by the Fiji ImageJ 1.8.0 software, and are plotted on the y axis; cell length (in m) is plotted on the x axis. Bar, 3m. VC=pBAD28. pCapV=CapV cloned in pBAD28; Q329R=CapVQ329R cloned in pBAD28.

In contrast, upon CapVQ329R-induced cell elongation, most filamentous cells displayed smooth contours with no visible septa (Fig. 5a and Supplementary Fig. 6a, b), suggesting that the block in cell division occurs prior to the development of constriction. However, one or two constrictions were occasionally observed in a few filaments (Supplementary Fig. 6a). We found that most of the CapVQ329R-induced filamentous cells were polynuclear and contained multiple abnormally shaped or positioned nucleoids. Discrete patches of DNA staining could be compact, asymmetrically positioned in the filament or displayed an extended or decondensed shape occupying an extensive part of the filament (Fig. 5a and Supplementary Fig. 6b). We also observed a few filamentous cells containing nucleoids, which were evenly positioned throughout the filament with larger interchromosomal spaces (Supplementary Fig. 6b). Collectively, these observations suggest filamentous cells induced by CapVQ329R overexpression to be (partially) defective in chromosome segregation and nucleoid condensation.

Distinct FtsZ rings were observed in most of the filaments suggesting that the block in cell division occurred after FtsZ-ring formation. Large variations in the distance between two FtsZ rings lead to an unequal number of FtsZ rings in filamentous cells of similar length. CapVQ329R promoted filamentation with an average cell length of 31.2m (n=102), which corresponds to >15 times the standard cell length. The majority of the filamentous cells contained three FtsZ rings with multiple segregated and/or unsegregated nucleoids distributed between the rings. Of note, the distance between two adjacent Z-rings varied dramatically in the population of filamentous cells, with on average fourfold longer spacing in CapVQ329R expressing cells compared to cells expressing CapV and the control (Fig. 5b and Table 1). Equally variable was the number of partitioned nucleoids among filamentous cells.

Abnormal FtsZ expression levels induce cell filamentation58. Comparative immunoblot analysis observed slightly lower FtsZ protein production levels in CapVQ329R and CapV overexpressing cells, while the ftsZ mRNA steady-state level was with 91 and 97% of vector control expression only slightly changed (Supplementary Fig. 6c). Whether this (partial) decrease in FtsZ production levels is responsible for CapVQ329R-induced filamentation needs to be further investigated. In summary, CapVQ329R inhibits cell division at the level of constriction initiation and interferes with nucleoid segregation and condensation.

Previous studies showed that SulA blocks cell division by direct interaction with cell division core component FtsZ during the bacterial SOS response19,32. To investigate whether CapVQ329R induced cell filamentation of E. coli MG1655 is caused by the activation of sulA, we analyzed cell morphology upon CapVQ329R overexpression in a sulA mutant59,60. We found that CapVQ329R induced the same filamentous cell phenotype in the sulA mutant as in wild-type E. coli MG1655 (Supplementary Fig. 6d), suggesting that CapVQ329R-induced cell filamentation is sulA-independent consistent with the observed FtsZ-ring formation.

Multiple factors coordinate the regulation of cell division29. Among them is MinC, which oscillates from pole to pole in order to prevent FtsZ-ring formation at the cellular poles34. Upon overexpression, MinC causes cell filamentation. We investigated the dynamics of MinC mobility by time-lapse fluorescence microscopy (Fig. 5c and Supplementary Movie 3). In rod-shaped wild-type cells, we observed that MinC oscillates between cell poles, moving a large fraction of the total fluorescence signals along the long axis of the cell as described previously61. A complete oscillation cycle lasted about 50s under our experimental conditions. Of note, we found that oscillation of the FtsZ-ring inhibitor MinC is not affected in CapVQ329R-induced filaments (Fig. 5c) (PB318,61). MinC displayed various fluorescence spots ranging from one up to 5 depending on the length of the filament (Fig. 5c and Supplementary Fig. 6e). These fluorescent spots oscillated within multiple invisible cell borders in the filaments suggesting physical restrictions by septa. Within one single filamentous cell, the number of the apparent fluorescence fractions followed thereby a three-step principle: n n1 n (n 2, located at both poles) or n n+1 n (n 1, located within the cell), during an oscillation periodicity of about 50s, which is similar as for rod-shaped control cells and wild-type CapV overexpressing cells (Fig. 5c and Supplementary Fig. 6e). It will be relevant to investigate how long-distance MinC oscillation is maintained in filamentous E. coli cells. In summary, MinC oscillation might remain the effective watchdog for the prevention of FtsZ-ring formation in filamentous cells.

CapV has been identified as a patatin-like phospholipase that causes growth retardation upon activation by cAMP-GMP in V. cholerae El Tor46. In E. coli MG1655, though overexpression of CapV of E. coli ECOR31 wild-type did not affect cell division during the entire growth phase. Though no effect was observed during the first 3h of cell growth in liquid culture, overexpression of CapVQ329R induced a mild decrease in optical density after 6h (Fig. 6a). As extensive filamentation might not permit a direct correlation between OD and cell number, we tested cell viability by spotting E. coli MG1655 cells on agar plates for counting of viable individual cells. Compared to the control and cells expressing CapV, ~50% of the colonies were recovered from E. coli MG1655 cell cultures expressing CapVQ329R (Fig. 6b). Consistently, cells stained with the nucleic acid stain SYTO 9 for viability and propidium iodide (PI) for cell death showed that CapVQ329R production induced approximately 50% E. coli MG1655 cells to stain selectively with PI after 6h, while no DNA staining with PI was observed after 3h (Fig. 6c and Supplementary Fig. 7a, b). Of note, cells in some filaments took up both dyes seemingly live and dead (Supplementary Fig. 7). Interestingly, the uptake of SYTO 9 into individual cells varied widely in particular in CapV expressing cells, which showed no decrease in cell viability. These data indicate a highly heterogeneous nucleic acid content in individual cells. Thus, expression of not or only partially activated CapV and CapVQ329R might differentially affect membrane permeability and cell viability.

a Growth curves of E. coli MG1655 upon CapV and CapVQ329R overexpression induced by 0.1% l-arabinose in TB at 37C. Each data point represents the meanSD of six biological replicates. tb = TB medium. b Colony-spotting assay on agar plates. Cells were grown at 37C and harvested at different time points. Cell viability determined by spotting serial dilutions (100106) on LB plates to assess colony-forming units. c Quantification of Live/Dead staining of E. coli MG1655 cells after 3h and 6h in TB medium at 37C. n=1200. VC=pBAD28. pCapV=CapV cloned in pBAD28; Q329R=CapVQ329R cloned in pBAD28.

Filamentation has been shown to be affected by environmental conditions26. We found that filamentation was particularly pronounced during cell growth in TB medium (1% tryptone, 0.5% NaCl), while it was restricted when cultured in LB medium (1% tryptone, 1% NaCl, 0.5% yeast extract) (Fig. 7a). Supplementation of TB medium with 0.5% yeast extract (YE) repressed the filamentous phenotype dramatically, while 5% restored the rod shape of all cells. Supplementation with 0.5% NaCl did not affect filamentation.

a Light microscopy pictures of E. coli MG1655 cell morphology and, b quantification of cell length in LB and TB medium supplemented with 0.5% NaCl, 0.5% YE, 5% YE, and VB6 (pyridoxine, 5mg/ml), respectively. The quantification is based on results from at least three independent experiments with the assessment of 70 cells from each group. Bar, 5m. VC=pBAD28. Q329R=CapVQ329R cloned in pBAD28.

Yeast extract is the water-soluble portion of autolyzed yeast cells used to prepare microbiological culture media62. As a nutrient source, it provides nitrogen, amino acids, peptides, carbohydrates, vitamin B complex, and other components that promote microbial growth63. In particular, yeast extract contains B vitamins, water-soluble precursors of structurally unrelated enzyme cofactors including thiamine (B1), riboflavin (B2), nicotinamide (B3), pantothenate (B5), pyridoxine (B6), biotin (B7), folic acid (B9), and cobalamin (B12). To identify the component(s) in YE that contribute(s) to the repression of cell filamentation upon CapVQ329R overexpression, we supplemented TB medium with these individual B vitamins. While supplementation with thiamine, riboflavin, nicotinamide, pantothenate, biotin, folic acid, and cobalamin showed no effect, supplementation with the vitamin B6 precursor pyridoxine at 5mg/ml decreased the length of filaments by 50% (Fig. 7a, b and Supplementary Fig. 8). It will be relevant to investigate the molecular mechanism of pyridoxine to inhibit CapVQ329R-induced cell filamentation. Further, the functionality of interconvertible pyridoxal and pyridoxamine vitamin B6 complex compounds and their 5 phosphate biologically active counterparts can be investigated individually and in combination. In summary, pyridoxine is one component in LB medium, which is effectively inhibiting cell filamentation upon overexpression of CapVQ329R.

To determine if the effect of CapVQ329R is restricted to E. coli K-12 MG1655, we overexpressed CapVQ329R in commensal and UPEC E. coli strains64,65 and the gastrointestinal pathogen S. typhimurium UMR1. The NCBI database (accessed latest December 15, 2021) indicates CapV homologs to be encoded predominantly by human fecal E. coli strains and strains derived from animals and animal and plant products. UPEC E. coli can develop extended filamentation upon host cell escape20. In all cases, CapVQ329R overexpression inhibited apparent swimming motility (Supplementary Fig. 9a, b).

Furthermore, CapVQ329R induced different degrees of cell filamentation in the E. coli strains and S. typhimurium UMR1 (Supplementary Fig. 9c). CapVQ329R production induced mild cell filamentation in commensal E. coli Fec32 and Fec89, moderate cell filamentation was observed in commensal Fec67 and extensive filamentation occurred in ECOR31, UPEC CFT073, and S. typhimurium UMR1. A heterogeneous population of moderately filamented cells as well as individual rod-shaped cells was observed in the UPEC strain E. coli No. 12 and the commensal E. coli strain Tob1 upon CapVQ329R overexpression (Supplementary Fig. 9c). In summary, our results showed that the phenotypes observed upon CapVQ329R expression were not restricted to E. coli MG1655, but were common to genetically unrelated commensal and UPEC E. coli strains and S. typhimurium, indicating a general effect of CapVQ329R on bacterial cell morphology, regulation of flagella-mediated swimming motility and biofilm formation.

We were subsequently wondering, whether CapVQ329R affects phenotypes other than cell filamentation and flagella expression. To this end, we investigated the colony morphotype on agar plates. A significant number of E. coli isolates display a rdar biofilm morphotype on Congo red agar plates characterized by the expression of the extracellular matrix components amyloid curli and the exopolysaccharide cellulose64.

Congruent with the pronounced temperature-dependent effect of CapVQ329R, we investigated the effect of CapVQ329R in strains that displayed the rdar morphotype at 37C. We choose the UPEC strain No. 12, which expresses a semi-constitutive csgD-dependent rdar morphotype65. While overexpression of wild-type CapV had no major effect compared to the No. 12 control, overexpression of CapVQ329R dramatically disrupted the rdar morphotype (Fig. 8a). Observations of colonies by scanning electron microscopy and TEM demonstrated that CapVQ329R induced even more extensive cell filamentation on agar plates compared to liquid culture, without affecting cell arrangement (Fig. 8b). While nonfilamented cells produced a pronounced extracellular matrix, the filaments produced only little or no matrix (Fig. 8b).

a Rdar morphotype in wild-type E. coli No. 12 vector control (VC) and upon overexpression of CapV and its mutant CapVQ329R. Cells were grown on a salt-free LB agar plate for 24h at 37C. b Scanning electron microscopy of plate-grown colonies. Colony morphotypes grown on a salt-free LB agar plate were fixed after 24h of growth at 37C. c CsgD production upon overexpression of CapV and CapVQ329R in E. coli strain No. 12 compared to VC. Only colony morphotypes from the same plate and signals from the same western blot are compared. d LC-MS/MS quantification of in vivo amounts of c-di-GMP, cAMP, and cAMP-GMP upon overexpression of CapV and CapVQ329R in E. coli strain No. 12 compared to VC. Data are displayed as absolute amounts referred to the original cell suspension. VC=pBAD28. pCapV=CapV cloned in pBAD28; Q329R=CapVQ329R cloned in pBAD28.

In conjunction with rdar morphotype downregulation, its major transcriptional activator CsgD was downregulated (Fig. 8c), demonstrating that CapVQ329R expression affects the central regulatory hub for rdar biofilm formation in strain No. 12. Furthermore, CapVQ329R overexpression equally downregulated the rdar morphotype in the commensal E. coli strains ECOR31, Fec67, Fec89, and Tob1 (Supplementary Fig. 9d). Downregulation of CsgD expression upon expression of CapVQ329R was exemplarily observed for strain ECOR31 (Supplementary Fig. 9e).

We then investigated the effect of CapVQ329R overexpression in commensal E. coli MG1655 and UPEC CFT073 which display a smooth and white morphotype when grown at 37C, indicative for the lack of rdar biofilm expression under these experimental conditions. After 48h of growth, though, a distinctly structured brown colony morphology with dye uptake was displayed upon CapVQ329R, but not upon CapV overexpression (Supplementary Fig. 9d). In the same line, S. typhimurium strain UMR1 displayed a smooth and white morphotype at 37C66, with its colony to develop a similar brown morphotype after 48h of CapVQ329R overexpression (Supplementary Fig. 9d). Again, this colony morphotype seem to be distinct from the rdar biofilm in coloration with no or below detection limit production of the biofilm regulator CsgD in E. coli MG1655 (Supplementary Fig. 9). Semi-constitutive rdar morphotype expressing ECOR31 and No. 12 displaying reduced morphology and coloration retained low CsgD levels upon CapVQ329R production (Fig. 8c and Supplementary Fig. 9c). Thus, CapVQ329R production affects biofilm-associated colony morphotype formation in a complex way in different E. coli strains.

In Enterobacteriaceae, cyclic di-GMP is a ubiquitous bacterial second messenger, which stimulates the development of the rdar biofilm morphotype via csgD expression40,67,68. We investigated the cyclic (di) nucleotide levels exemplarily in the UPEC strain E. coli No. 12. Along with inhibition of csgD expression, the in vivo cyclic di-GMP level was concomitantly decreased upon CapVQ329R overexpression (Fig. 8d). Besides cyclic di-GMP, both cAMP and cAMP-GMP regulate E. coli biofilm formation43,69. Consistent with a downregulated rdar biofilm phenotype, a reduction in cAMP and an increase in cAMP-GMP levels were observed upon CapVQ329R overexpression (Fig. 8d). Thereby, CapVQ329R might, for example, inhibit a diguanylate cyclase or promote phosphodiesterase activity to downregulate the cyclic di-GMP level. Taken together, these results indicate that CapVQ329R inhibits rdar biofilm formation potentially through regulation of the intracellular level of various cyclic (di)-nucleotide signals.

Recently, the patatin-like phospholipase CapV and the dinucleotide cyclase DncV have been shown to take part in bacterial antiphage defense70. Since CapVQ329R showed a physiological function independent of DncV, we wondered whether CapVQ329R still contributes to antiphage defense. While overexpression of wild-type CapV had no effect, interestingly, an approximately tenfold higher plaque formation was observed upon CapVQ329R overexpression compared to the control when MG1655 cells were infected with P1 phage (Supplementary Fig. 10a), indicating that CapVQ329R enhances susceptibility to bacteriophage P1 infection.

We also observed that overexpression of CapVQ329R, but not CapV renders the laboratory strain E. coli MG1655, the commensal strain ECOR31, and UPEC No. 12 more susceptible or sensitive against the beta-lactam antibiotic cephalexin (Supplementary Fig. 10b). A systematic investigation of altered susceptibility to various antibiotic classes awaits to be performed.

We were also wondering whether CapVQ329R-induced cell elongation and concomitant physiological changes impair the interaction of E. coli with host cells. To this end, we exposed the bladder epithelial cell line T24 to UPEC E. coli strain No. 12 overexpressing CapV and CapVQ329R. CapVQ329R expressing E. coli strain No. 12 associated significantly less with the T24 bladder epithelial cell line than the control and CapV expressing cells. Although a trend to lower induction of mRNA steady-state levels of genes coding for the pro-inflammatory markers IL-1 and CXCL-8 was observed for T24 cells exposed to CapVQ329R expressing bacteria, the obtained results were not statistically significant (Supplementary Fig. 10c). Thus, in summary, CapVQ329R production causes substantial physiological alterations in E. coli MG1655.

Read more:

Patatin-like phospholipase CapV in Escherichia coli - morphological and physiological effects of one amino acid substitution | npj Biofilms and...