1. Microbiology and Infectious Disease

Intracellular glycosyl hydrolase PslG shapes bacterial cell fate, signaling, and the biofilm development of Pseudomonas aeruginosa

  1. Jingchao Zhang
  2. Huijun Wu
  3. Di Wang
  4. Lanxin Wang
  5. Yifan Cui
  6. Chenxi Zhang
  7. Kun Zhao  Is a corresponding author
  8. Luyan Ma  Is a corresponding author
  1. Frontiers Science Center for Synthetic Biology and Key Laboratory of Systems Bioengineering (Ministry of Education), School of Chemical Engineering and Technology, Tianjin University, China
  2. State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, China
  3. University of Chinese Academy of Sciences, China
https://doi.org/10.7554/eLife.72778
Share this article
Cite this article
  1. Jingchao Zhang
  2. Huijun Wu
  3. Di Wang
  4. Lanxin Wang
  5. Yifan Cui
  6. Chenxi Zhang
  7. Kun Zhao
  8. Luyan Ma
(2022)
Intracellular glycosyl hydrolase PslG shapes bacterial cell fate, signaling, and the biofilm development of Pseudomonas aeruginosa
eLife 11:e72778.
https://doi.org/10.7554/eLife.72778
  2. Download BibTeX
  3. Download .RIS

Abstract

Biofilm formation is one of most important causes leading to persistent infections. Exopolysaccharides are usually a main component of biofilm matrix. Genes encoding glycosyl hydrolases are often found in gene clusters that are involved in the exopolysaccharide synthesis. It remains elusive about the functions of intracellular glycosyl hydrolase and why a polysaccharide synthesis gene cluster requires a glycosyl hydrolase-encoding gene. Here, we systematically studied the physiologically relevant role of intracellular PslG, a glycosyl hydrolase whose encoding gene is co-transcribed with 15 psl genes, which is responsible for the synthesis of exopolysaccharide PSL, a key biofilm matrix polysaccharide in opportunistic pathogen Pseudomonas aeruginosa. We showed that lack of PslG or its hydrolytic activity in this opportunistic pathogen enhances the signaling function of PSL, changes the relative level of cyclic-di-GMP within daughter cells during cell division and shapes the localization of PSL on bacterial periphery, thus results in long chains of bacterial cells, fast-forming biofilm microcolonies. Our results reveal the important roles of intracellular PslG on the cell fate and biofilm development.

Editor's evaluation

One of the major findings related to the production of long chain polysaccharide polymers by bacteria is why there often is a hydrolytic enzyme encoded within the gene cluster encoding the biosynthetic proteins. At a basic level, these hydrolases cleave the chains to prevent toxic accumulation of intracellular polysaccharides, but it has now been discovered that a glycosyl hydrolase acting on the Pseudomonas aeruginosa PSL polysaccharide affects other important responses of this Gram-negative bacterium. The PslG hydrolase regulated cell length, PSL cell surface localization and signaling via cyclic-di-GMP, overall controlling accumulation of the cells into a protective biofilm.

https://doi.org/10.7554/eLife.72778.sa0

Introduction

Structured, surfaced-associated communities of microorganism known as biofilms are important life forms of bacteria prevailing in nature, industrial, and clinical settings (Costerton et al., 1995; Stoodley et al., 2002). In general, biofilm development involves four specific stages: attachment, microcolony formation, matured microcolonies, and dispersal. Bacteria within biofilms are embedded in an extracellular matrix that protects bacterial cells from antibiotics, host defenses, and environmental stresses. Even though the components of biofilm matrix differ from species to species, it generally composes of exopolysaccharides, proteins, and nucleic acids (Stoodley et al., 2002; Flemming and Wingender, 2010). Exopolysaccharides are critical biofilm matrix components for many bacteria, which often promote attachment to surfaces and other cells, act as a scaffold to help maintain biofilm structure, and provide protection (Stewart and Costerton, 2001; Häussler and Parsek, 2010; Stewart and Costerton, 2001). Gene encoding glycosyl hydrolase is often found in gene clusters that are involved in the synthesis of exopolysaccharide (Franklin et al., 2011). Beyond of exopolysaccharide degradation, little is known about whether these genes affect bacterial physiology and biofilm development.

Pseudomonas aeruginosa is an opportunistic human pathogen that can cause life-threatening infections in cystic fibrosis (CF) patients and immune-compromised individuals (Govan and Deretic, 1996; Lyczak et al., 2000; Ramsey and Wozniak, 2005). P. aeruginosa can produce at least three different types of exopolysaccharides: alginate, PEL, and PSL (also named as Pel/Psl polysaccharide previously). Alginate is not expressed at high levels in the majority of non-CF isolates (Colvin et al., 2012), whereas PSL is expressed by most P. aeruginosa natural and clinical isolates (Colvin et al., 2012; Ma et al., 2012; Jennings et al., 2021). PEL is significant for biofilm formation when PSL cannot be synthesized (Jennings et al., 2015). In P. aeruginosa PAO1, PSL is a primary scaffold matrix component that can form a fibre-like matrix to enmesh bacteria within a biofilm (Colvin et al., 2012; Ma et al., 2009; Wang et al., 2013). PSL has shown multiple functions in the biofilm formation of PAO1. For example, PSL can act as a ‘molecular glue’ to promote bacteria cell-cell and cell-surface interactions (Ma et al., 2006; Ma et al., 2009). PSL trails on a surface guide bacteria exploration and microcolony formation (Zhao et al., 2013). Moreover, PSL can also work as a barrier to protect bacteria from antibiotics and phagocytic cells (Mishra et al., 2012; Billings et al., 2013; Tseng et al., 2013). Interestingly, PSL can function as a signal to stimulate biofilm formation through affecting intracellular signal molecule cyclic-di-GMP (c-di-GMP) (Irie et al., 2012).

PSL is synthesized by psl operon, containing 15 co-transcribed genes (pslABCDEFGHIJKLMNO) (Byrd et al., 2009). PslG is a glycosyl hydrolase that encoded by pslG wthin psl operon. PslG has been shown to degrade PSL in vitro or within biofilm matrix which is released from dead bacteria (Yu et al., 2015; Zhao et al., 2018), hence it can inhibit biofilm formation and disrupt a formed biofilm at a nanomolar concentration (Yu et al., 2015; Baker et al., 2016). PslG was first thought to be essential for PSL synthesis, since deletion of pslG gene led to a loss of PSL production (Byrd et al., 2009). Later studies found that deletion of pslG in the earlier work has a polar effect on the expression of pslH and thus resulted in the loss of PSL production, and absence of pslG itself did not result in a complete loss of PSL production per se, but led to a less production of PSL and reduced bacterial initial attachment compared with PAO1 (Baker et al., 2015; Wu et al., 2019). PslG is localized mainly at the inner membrane and some in the periplasm. PslA, PslD, and PslE help PslG anchoring in the inner membrane, which is critical for PslG to be involved in the biosynthesis of PSL (Wu et al., 2019). In addition, the glycoside hydrolytic activity of PslG is also important for PSL production and the key amino acid residues for this activity are E165 and E276 (Wu et al., 2019; Yu et al., 2015).

C-di-GMP is an important second messenger controlling a wide range of cellular processes in many bacteria, such as motility, cell differentiation, biofilm formation and production of virulence factors (Römling et al., 2013). Reports have shown that c-di-GMP is asymmetrically distributed among daughter cells upon bacterial cell division and the asymmetric division on surfaces produces specialized cell types, a spreader for dissemination and a striker for local tissue damage (Christen et al., 2010; Laventie et al., 2019). It has not been investigated whether an intracellular glycoside hydrolase would affect the c-di-GMP level.

In this work, aiming to study the effect of PslG on the cell fate and biofilm development of P. aeruginosa at the single cell level, we systematically studied the pslG in-frame deletion mutants by employing a high-throughput bacterial tracking technique (Zhao et al., 2013). The morphology and motility behavior of bacterial cells in the course of biofilm development were analyzed at the single-cell level. Using pCdrA::gfp reporter, the c-di-GMP level of each cell was also monitored. Microscopically, the attachment behavior of cells on the microtiter surfaces and the pellicles formed at the air-liquid interface were also characterized. Our data suggest that lacking of pslG impacts cell morphology, the signaling function of PSL, the c-di-GMP distribution and bacteria distribution within a biofilm. Based on our results together with those in literature, a model is proposed to understand the role of pslG in the biofilm development.

Results

ΔpslG strains cannot form rings on microtiter dish wells even when PSL production is induced to the wild-type level

Our previous study showed that the pslG in-frame deletion mutant (was named as ΔpslG2 by Wu et al., 2019 hereafter termed as ΔpslG) decreased the production of PSL and bacterial initial attachment on the microtiter surface (Wu et al., 2019). To know whether the attachment defect of ΔpslG mutant is due to PSL reduction, we replaced the promoter of psl operon by PBAD promoter in ΔpslG background, resulting in a PAO1-derived Psl-inducible ΔpslG strain (named as PBAD-pslΔpslG), whose PSL production can be induced by the concentration of arabinose. When induced with 0.5% arabinose, PBAD-pslΔpslG strain produced similar amount of PSL as that of PAO1 (Figure 1B). However, it was not able to form a ring at the air-liquid interface as that seen in either PAO1 or PBAD-psl strain (referred as WFPA801 previously, Ma et al., 2006) under 0.5% arabinose induction (Figure 1B, rings are indicated by an arrow). The ring formation can be recovered by either PBAD-pslG inserted in chromosome attB site (PBAD-psl ΔpslG attB::pslG in Figure 1B) or pslG knocked into the ΔpslG mutants at the original location of pslG (PBAD-pslΔpslG::pslG in Figure 1B). Whereas PslGE165Q, E276Q (two key hydrolytic active sites of PslG were mutated) cannot complement this phenotype (Figure 1B). These results indicate that PslG and its glycoside hydrolytic activity are important for the ring formation at the air-liquid interface.

Figure 1 with 1 supplement see all
Download asset Open asset
Inducing PSL production in ΔpslG background cannot recover its defects on bacterial initial attachment and yet affects swimming motility.

(A): A schematic of the psl operon in PAO1 and its corresponding mutants used in this study. Genes pslA-O are shown in boxes (not to scale). Angled lines represent the extent of deleted sequence, and black arrows indicate transcriptional start sites (not to scale). (B): The Psl production of tested strains inducing with 0% or 0.5% arabinose. The amount of Psl was determined by immune-dot blotting and normalized to the level of PAO1. A representative of dot blotting as well as corresponding microtiter dish wells and their crystal violet (CV) reading (OD560) posted CV staining in attachment assay were shown under each corresponding column. Arrows indicate the ring at air-liquid interface. (C): The swimming motility of tested strains inducing with 0% or 0.5% arabinose. The corresponding image of swimming zone was shown under each column. Statistical significances were measured using student’s t-test (**, p < 0. 01).

Figure 1—source data 1

Figure 1B, C source data.

https://cdn.elifesciences.org/articles/72778/elife-72778-fig1-data1-v2.xlsx

Flagella and Type IV pili (T4P) are also important for the initial attachment of P. aeruginosa (O’Toole and Kolter, 1998; Bruzaud et al., 2015). We then tested the flagellum-driven swimming motility and T4P-mediated twitching motility of PBAD-pslΔpslG strain to evaluate the function of flagella and T4P. Without arabinose induction, PBAD-pslΔpslG strain exhibited similar swimming ability and twitching motility as did PAO1 and PBAD-psl (Figure 1C and Figure 1—figure supplement 1). However, with 0.5% arabinose induction, PBAD-pslΔpslG strain (having a wild type level of PSL production, Figure 1B) showed reduced swimming zone compared to that of either PAO1 or PBAD-psl (Figure 1C). T4P-mediated twitching motility of PBAD-pslΔpslG was not affected under conditions with or without arabinose (Figure 1—figure supplement 1C). These results demonstrate that pslG deletion does not affect the function of flagella or T4P directly, yet inducing PSL production in PBAD-pslΔpslG attenuates the swimming motility with no effect on bacterial growth (Figure 1—figure supplement 1D), which might impact its attachment phenotype. Taken together, our results show that increasing PSL production in PBAD-pslΔpslG could not rescue its defect on attachment, suggesting that ΔpslG might have multiple effects on bacterial physiology.

ΔpslG impacts the bacterial distribution and maximum thickness of pellicles

We then investigated the effect of ΔpslG on the biofilms formed at the air-liquid interface, termed as pellicles, by using confocal laser scanning microscopy. The total pellicle biomass of ΔpslG is similar to that of PAO1 after 24 hr growth, although ΔpslG produced much less PSL and had defect on initial attachment (Figures 1, 2A and C). However, ΔpslG has significant higher maximum thickness than that of PAO1 (Figure 2B). In addition, there are less bacteria in each section image of ΔpslG pellicles compared to that of PAO1 (Figure 2D, left and middle panel). The PSL matrix in ΔpslG pellicles shows weaker fluorescent intensities than that of PAO1 (Figure 2D, middle and right panels), which is consistent with their corresponding PSL production. In spite of that, the fibre-like PSL can be detected in the pellicles of ΔpslG, which have a radial pattern as previously described for PAO1 pellicles (Figure 2D, middle panels) (Wang et al., 2013). These results suggest that the pslG deletion might impact bacterial distribution within biofilms.

Figure 2
Download asset Open asset
Comparison of pellicles formed by PAO1 and ΔpslG mutant.

(A): Three-dimensional images of 24 hr air-liquid interface biofilms (pellicles) formed by PAO1 and ΔpslG. (B): Biofilm biomass and maximum thickness of PAO1 or ΔpslG strain. (C): PSL in pellicles of PAO1 and ΔpslG. (D): Typical section images of pellicles formed by PAO1 and ΔpslG. Left panel, section images showed the top-down view (square) and side view (rectangle) of corresponding pellicles. Middle panel, section images at the middle of corresponding pellicles. The distribution of bacteria (green), the fibre-like PSL matrix (red) and corresponding DIC images (grey) were shown. Right panel, PSL fluorescence intensity in corresponding biofilm images shown in the middle panel (the average intensity of PSL in per μm3 biofilm is shown in the upper right corner). Green, SYTO9 stained bacteria, Red, TRITC-HHA stained PSL. Statistical significances were measured using student’s t-test (***, p < 0.001 when compared to PAO1). Scale bar: 50 μm for A and the left panel in (D); 10 μm for the middle panel in D.

Figure 2—source data 1

Figure 2B, C source data.

https://cdn.elifesciences.org/articles/72778/elife-72778-fig2-data1-v2.xlsx

Single-cell tracking analysis indicates that loss of PslG or its glycoside hydrolytic activity promotes microcolony formation in flow-cell systems

To further understand the effect of ΔpslG, by employing bacterial tracking techniques, we observed ΔpslG cell behavior at the single-cell level. Figure 3A shows the surface coverage obtained by all tracked bacterial trajectories for a specific time period during microcolony formation. Red color indicates the surface area that has been visited by bacteria, while black color indicates a ‘fresh’ surface area that has never been visited. Bacterial cells are shown in blue. Under the same total bacterial visits (marked as N in Figure 3A), the difference in the surface coverage between ΔpslG and PAO1 is not obvious at N~10,000. As N increases, the surface coverage of ΔpslG is clearly less than that of PAO1 (Figure 3A). At N~100,000, ΔpslG has a surface coverage of 52% ± 6% while PAO1 has 81% ± 10%. Compared with PAO1, the less efficiency in covering the surface leads to a more non-uniform bacterial visit distribution for ΔpslG (Figure 3B). Correspondingly, by fitting the distribution of bacterial visits with a power law, different power law exponents were obtained (see one example in Figure 3C). The averaged power law exponents of three repeats are –2.9 ± 0.1 for PAO1 and –2.5 ± 0.1 for ΔpslG. Such differences in bacterial visits distribution resulted that the time required to observe a visible microcolony (defined as clusters of more than 30 cells in this study, marked by dash lines in Figure 3D) in the field of view is shorter for ΔpslG (5.9 ± 1.7 hr) than that for PAO1 (8.7 ± 2.4 hr). In addition, after 10 hr of growth, compared with PAO1, ΔpslG formed more microcolonies in the field of view (Figure 3E). The microcolony formation phenotype of ΔpslG mutants can be reverted back to WT-like when they are complemented with PslG by knocking pslG into the ΔpslG mutants at the original location of pslG, namely ΔpslG::pslG. As illustrated in Figure 3, at N~100,000, the surface coverage of ΔpslG::pslG reaches to 70% ± 7% and the power law exponent of the bacterial visit distribution is –3.2 ± 0.13 (averaged over three repeats), both are closer to those of PAO1 than those of ΔpslG. The number of microcolonies formed by ΔpslG::pslG in the field of view after 10 hr of growth is also similar to that of PAO1 and less than that of ΔpslG. The fast microcolony formation phenotype of ΔpslG mutant is also able to be complemented by PslG that is expressed from plasmid (Figure 3—figure supplement 1). However, ΔpslG::pslGE165Q,E276Q, in which the key glycoside hydrolytic activity sites are mutated cannot recover WT-like phenotype (Figure 3 and Figure 3—figure supplement 1), suggesting the importance of PslG hydrolytic activities. Complementation tests were also performed in strains from PBAD-psl background, and similar trends were observed (Figure 3—figure supplement 2). PEL has been shown to play a role in bacterial aggregation (Jennings et al., 2021), thus we also tested PEL-negative strains in wild type (ΔPpel) or the ΔpslG background (ΔpslGΔPpel). The phenotype of ΔpslGΔPpel was similar to that of ΔpslG, suggesting little contribution of PEL on the fast microcolony formation of ΔpslG (Figure 3 and Figure 3—figure supplement 3). Tracking GFP-tagged bacteria in a flow cell also shows that the PAO1 biofilms tend to spread on surfaces (Figure 3F and Video 1) while ΔpslG cells tend to accumulate and form microcolonies with strong fluorescence intensity (Figure 3F and Video 2). This is consistent with the results of bacterial visit distribution map shown in Figure 3B. Taken together, these results indicate that loss of pslG promotes microcolonies formation by changing the surface exploration of bacteria during microcolony formation, a phenomenon that has been reported mostly for strains producing a high level of PSL (Zhao et al., 2013).

Figure 3 with 3 supplements see all
Download asset Open asset
Effects of pslG on the formation of microcolonies in flow-cell systems.

(A): Surface coverage maps at a total of 10,000, 50,000, and 100,000 bacterial visits for PAO1, ΔpslG, ΔpslG::pslG, and ΔpslG::pslGE165Q,E276Q cells. Red color indicates the surface area that has been visited or contaminated, while black color indicates a ‘fresh’ surface area. Bacteria in the current frame are shown in blue. The surface coverage in the figure is the result of an experiment. (B): The intensity map of bacterial visits at N ~ 100,000. The color scale of black to cyan represents bacterial visits of 0–1,000. (C): The graph displays one measurement result for the visit frequency distributions of PAO1, ΔpslG, ΔpslG::pslG and ΔpslG::pslGE165Q,E276Q at N ~ 100,000. The slope in the figure is the fitting result of an experiment. (D): Examples of microcolonies (enclosed by dash lines) formed by PAO1 and ΔpslG cells, respectively, cultured in a flow cell for about 8 hr. (E): The number of microcolonies in the field of view formed by PAO1, ΔpslG, ΔpslG::pslG and ΔpslG::pslGE165Q,E276Q at 10 hr after inoculation in a flow cell. The number (N) of frames analyzed are 14, 43, 52, and 51 for PAO1, ΔpslG, ΔpslG::pslG, and ΔpslG::pslGE165Q,E276Q, respectively. Error bars represent standard deviations of the means. Statistical significances were measured using one-way ANOVE. n.s., not significant; *p < 0.05; **p < 0.001; ***p < 0.0001. F. Snapshots taken at 10 hr after inoculation in a flow cell, showing the microcolonies formed by ΔpslG and PAO1. Bacteria were tagged by GFP. Fluorescence images and corresponding bright-field images were shown. Scale bar, 5 μm.

Figure 3—source data 1

Figure 3C source data.

https://cdn.elifesciences.org/articles/72778/elife-72778-fig3-data1-v2.xlsx
Figure 3—source data 2

Figure 3E source data.

https://cdn.elifesciences.org/articles/72778/elife-72778-fig3-data2-v2.xlsx
Video 1
Download asset
An example video of tracking the biofilm formation of Gfp-tagged PAO1 cells.

The video was taken at a frame interval of 10 min for 10 hr and was played back at 5 fps. Scale bar, 5 μm.

Video 2
Download asset
An example video of tracking the biofilm formation of Gfp-tagged ΔpslG cells.

The video was taken at a frame interval of 10 min for 10 hr and was played back at 5 fps. Scale bar, 5 μm.

T4P-driven motility is not the main factor in promoting the microcolony formation of ΔpslG strains in flow-cell systems

T4P-driven motilities on surface, such as walking and crawling, affect microcolonies formation in flow-cell systems (Conrad et al., 2011). To investigate why ΔpslG promotes microcolonies formation, we calculated the twitching speed of bacterial cells. To minimize the effect due to possible different production of PSL, we compared the measurements between PAO1 and PBAD-pslΔpslG under 0.5% arabinose, under which both strains show relatively similar production of PSL (Figure 1). The results show a slightly reduced average speed and a higher crawling percentage of PBAD-pslΔpslG cells compared with PAO1 (Figure 1—figure supplement 1A, B). But such differences are not statistically significant (P = 0.29), indicating that the twitching motility may not be the main factor in promoting the microcolony formation.

ΔpslG shapes the localization of PSL on bacterial periphery, leading to long chains of bacterial cells that are connected by PSL

During bacterial tracking in flow cells, we frequently observed long chains of bacterial cells in the ΔpslG strain or PBAD-pslΔpslG strain when PSL production was induced with arabinose (Figure 4A), which typically started to appear 1~2 hr after inoculation of bacteria into a flow cell under tested conditions and could reach to about 50% of cell population at a later time (Figure 4B). Such long chains of cells were not observed in strains that have intact pslG or when PSL production was not induced in PBAD-pslΔpslG (Figure 4—figure supplement 1). In a typical cell division process, two daughter cells will be disconnected from each other when the formation of septum is completed. However, in pslG deletion mutants, the two daughter cells could not separate into physically disconnected progenies, leading to a cell chain (See one example of PBAD-pslΔpslG in Video 3). The length of chains varied. Among all the observed chains, the chains consisting of 4 cells were observed most frequently, which is true both in ΔpslG and in PBAD-pslΔpslG strains (Figure 4C). These bacterial cell chains can grow as cells continue to divide, yet some can also be broken by bending of chains (Video 3), suggesting that the bacterial chains are not connected by septum. In addition, bacterial cell chains were also observed in liquid culture of ΔpslG mutants (data not shown), indicating that bacterial adhering on surfaces is not required for the formation of bacterial cell chains.

Figure 4 with 1 supplement see all
Download asset Open asset
Characterization of long bacterial chains of ΔpslG strains.

(A): Examples of long bacterial chains (indicated by black dotted outlines) formed by ΔpslG and PBAD-pslΔpslG cells. (B): The faction of single isolated bacterial cell and cell in chains at different time points after inoculation in a flow cell. The number of analyzed picture in each strain is n = 88 (about 3200 cells) for PBAD-pslΔpslG and n = 87 (about 3500 cells) for ΔpslG. (C): The number distribution of cells consisted in a chain. The number of analyzed cells is n = 301 for PBAD-pslΔpslG and n = 322 for ΔpslG. Scale bar, 2 μm.

Figure 4—source data 1

Figure 4B source data.

https://cdn.elifesciences.org/articles/72778/elife-72778-fig4-data1-v2.xlsx
Figure 4—source data 2

Figure 4C source data.

https://cdn.elifesciences.org/articles/72778/elife-72778-fig4-data2-v2.xlsx
Video 3
Download asset
An example video of the formation of PBAD-pslΔpslG long cell-chain.

The video was taken at a frame interval of 5 min for 3.5 hr and was played back at 5 fps. Scale bar, 5 μm.

To investigate whether PSL has any contribution on the formation of such long bacterial chains, we stained PSL in the biofilm formed in flow cells by FITC-HHA (green fluorescence dyes FITC labeled lectin HHA). The PSL of ΔpslG strains is tightly associated with bacteria compared to PAO1 and PBAD-psl with arabinose (Figure 5A and Figure 5—figure supplement 1). We also used cell membrane stain FM4-64 to help locate cell periphery and septum. Strikingly, strong PSL signal is often found around septa in ΔpslG strains and strains with catalytic site mutation (PBAD-pslΔpslG::pslGE165Q, E276Q or ΔpslG::pslGE165Q, E276Q), which barely observed in strains with wild type pslG (such as PAO1 and PBAD-psl) under the same growth conditions (Figure 5A and Figure 5—figure supplement 1).

Figure 5 with 3 supplements see all
Download asset Open asset
Cells in bacterial chains are connected by PSL and can be disassembled by PslG supplied exogenously.

(A): Fluorescence staining of long bacterial chains formed by PBAD-pslΔpslG cells. Staining of control samples of PBAD-psl is also shown. Green shows PSL stained by FITC-HHA, and red shows the bacterial cell membranes stained by FM4-64. Arrows in the PSL stain image indicate the bright spotted PSL locations, and arrows in the membrane stain image show the septum locations. (B): Time–lapse images show the break-up process of long bacterial chains of PBAD-pslΔpslG when PslG was supplied. Scale bar, 5 μm.

From the PSL staining results, we speculate that PSL might help to connect bacterial cells together to form long bacterial chains. To test this hypothesis, we first tested whether the PBAD-pslΔpslG could form long bacterial chains when PSL is not produced. Under a culture condition without arabinose, the transcription of psl operon in PBAD-pslΔpslG is not induced, no bacterial chains were observed as shown in Figure 4—figure supplement 1, suggesting that PSL production is required for the formation of long bacterial chains. Next, we treated the long bacterial chains with purified PslG, which has been shown to be able to disaggregate microcolonies and matured biofilms by hydrolyzing PSL (Yu et al., 2015). At 4 min after addition of exogenous PslG, the long chains of bacterial cells were seen clearly to start to be broken up, and they were completely disconnected into single cells after 12 min of PslG treatment (Figure 5B) (Videos 4 and 5). Bacterial chains can be separated by exogenous PslG within a few minutes further confirmed that the bacterial chains were not connected by septum. These results together with the fact that PSL is a ‘sticky’ exopolysaccharide suggest that bacterial cells are frequently connected by PSL in ΔpslG mutant strains, leading to the long bacterial cell chains.

Video 4
Download asset
An example video of the degradation of PBAD-pslΔpslG long cell-chain.

The video was taken at a frame interval of 1 min for 0.5 hr and was played back at 5 fps. Scale bar, 5 μm.

Video 5
Download asset
An example video of the degradation of ΔpslG long cell-chain.

The video was taken at a frame interval of 1 s for 10 min and was played back at 5 fps. Scale bar, 5 μm.

This long-cell-chain phenotype can be complemented by wild type PslG, but not hydrolytic activity sites mutation PslGE165Q, E276Q (Figure 5—figure supplement 2), suggesting that the hydrolytic activity of PslG plays a key role on the formation of long bacterial cell chains. To check whether exopolysaccharide PEL and alginate play roles on the observed phenotype induced by pslG deletion, we also tested PAO1-derived alginate-negative strain ΔalgD as well as PEL-negative strains ΔPpel and ΔpslGΔPpel. The phenotype of either ΔPpel or ΔalgD is similar to that of PAO1, whereas ΔpslGΔPpel exhibits a similar phenotype as ΔpslG strains, for bacterial chains formation, microcolonies formation, and initial attachment (Figure 5—figure supplement 3, Figure 3—figure supplement 3, Figure 1—figure supplement 1C). These results indicate that PEL and alginate may have little effect on the observed phenotype induced by pslG deletion in this study. This data also imply that the long bacterial chains might be a contributor for ΔpslG strains to promote microcolony formation.

Lack of PslG or its hydrolytic activity has effects on the fate and C-di-GMP distribution of daughter cells during cell division

C-di-GMP is a critical intracellular signal molecule that affects a variety of cell activities including cell motility, cell fate after division and biofilm formation. We then monitored the c-di-GMP level of cells for each cell division event by employing pCdrA::gfp as a reporter (Irie et al., 2012). Cells with high c-di-GMP levels show strong fluorescence intensity as previously described (Irie et al., 2012). We focused on the first bacterial division events after cells attached on the surface and analyzed the fluorescence intensity (corresponding to c-di-GMP levels) in two daughter cells right after division (Figure 6A). By comparing the fluorescence intensity of each daughter cell relative to its mother cell, the division events can be classified into three types: none of the daughter cells becomes bright (none-bright), one daughter cell becomes bright (one-bright) and both of daughter cells become bright (two-bright). The results show that none-bright type is observed most frequently in all tested strains, which has an occurrence probability of ~ 60% for PAO1, ~ 58% for ΔpslG, and ~ 52% for PBAD-pslΔpslG. Interestingly, both ΔpslG mutants show a relatively higher probability of two-bright type (~17% for PBAD-pslΔpslG strain, ~ 16% for ΔpslG strain) than that of PAO1 (~11%) (Figure 6B and Figure 6—figure supplement 1). In addition, the catalytic site mutated strains show a phenotype similar to ΔpslG strains, have a relatively higher probability of two-bright type (~19% for both ΔpslG::pslGE165Q, E276Q and PBAD-pslΔpslG::pslGE165Q, E276Q strain) (Figure 6B). Thus, compared with PAO1, ΔpslG cells and bacterial cells with the catalytic site mutated PslG would have a higher probability to have both daughter cells with high c-di-GMP levels. During asymmetric divisions, daughter cells with high c-di-GMP levels keep staying on the surface and daughter cells with low c-di-GMP levels tend to move away (Christen et al., 2010; Laventie et al., 2019). Therefore, both daughter cells with high c-di-GMP levels might enhance bacterial stay on the surface and alter their movement pattern. In addition, a high level of c-di-GMP enhances PSL production. An earlier work has shown that cells form a microcolony through a PSL-based rich-get-richer mechanism (Zhao et al., 2013), during which founder cells can be very important. The cells of two-bright cases have high c-di-GMP levels and thus can act as founder cells to promote microcolony formation. Altogether, the slight change in c-di-GMP levels of daughter cells in pslG mutants can be likely one of reasons to promote the formation of microcolony and long bacterial chains.

Figure 6 with 1 supplement see all
Download asset Open asset
Lacking of PslG or its hydrolytic activity has effects on the fate and c-di-GMP distribution of daughter cells.

(A): Three types of cell division based on fluorescence intensity changes of daughter cells relative to that of their mother cell: none of daughter cells becomes bright (none-bright), one daughter cell becomes bright (one-bright), and both daughter cells become bright (two-bright). Examples given are PAO1 cells. The fluorescence is from pCdrA::gfp, which acts as a reporter for the c-di-GMP level of cells. (B): The measured probability of three types of division in PAO1, ΔpslG, PBAD-pslΔpslG, ΔpslG::pslGE165Q,E276Q, and PBAD-pslΔpslG::pslGE165Q,E276Q. The total number of analyzed division events from more than three repeats is n = 174 for PAO1, n = 168 for ΔpslG, n = 109 for PBAD-pslΔpslG, n = 131 for ΔpslG::pslGE165Q,E276Q, and n = 115 for PBAD-pslΔpslG::pslGE165Q,E276Q. Statistical significances were measured using one-way ANOVA. n.s., not significant; **p < 0. 01. Scale bar, 5 μm.

Figure 6—source data 1

Figure 6B source data.

https://cdn.elifesciences.org/articles/72778/elife-72778-fig6-data1-v2.xlsx

PSL produced by pslG mutants has a stronger signaling function

As shown in Figure 5, PSL is often localized around septa in ΔpslG strains. Since PSL can have a signaling function by stimulating intracellular c-di-GMP production (Irie et al., 2012), we speculated that PSL produced from ΔpslG might have different signaling properties. To test the signaling functions of PSL, we set up a co-culture system, which contained a PSL donor strain and a reporter strain. PAO1 harboring the plasmid pCdrA::gfp was utilized as an intracellular c-di-GMP reporter strain while PAO1, ΔPpsl, ΔpslG, and ΔpslG:: pslGE165Q,E276Q strains as PSL donors respectively, in which ΔPpsl (named as WFPA800 previously) was used as a negative control because it does not produce PSL due to deletion of the promoter of psl operon (Figure 1A). As shown in Figure 7, PAO1 that produces wild type level of PSL can induce a stronger GFP fluorescence signal in the reporter stain compared to ΔPpsl (Figure 7A,B). Strikingly, ΔpslG strain and the hydrolytic activity sites mutant (ΔpslG:: pslGE165Q,E276Q) stimulated a higher fluorescence signal than that of PAO1 (Figure 7B) although both of which produce less PSL (about 30% of PAO1 level, Figure 1A), suggesting that PSL synthesized from these two pslG mutants has a stronger signaling effect on stimulating the intracellular c-di-GMP production than that of wild type. We then examined the impact of exogenous PslG on the fluorescence signal of PAO1/pCdrA::gfp. After 10 hr of PslG treatment, the fluorescence intensity of PAO1/pCdrA::gfp was significantly reduced compared to the non-treatment control and it had a dose-dependent manner ( Figure 7C, D). This further suggests that the hydrolysis of PslG might reduce the signaling property of PSL.

Figure 7
Download asset Open asset
PSL produced by pslG mutant has a stronger signaling function.

(A): A coculture system using pCdrA::gfp as a reporter plasmid to evaluate the intracellular c-di-GMP. The reporter strain (PAO1/pCdrA::gfp) was mixed with PSL provider strain (PAO1, ΔpslG, or ΔPpsl: a PSL-negative mutant) at a ratio of 1: 1, and the GFP fluorescence value and OD600 in each co-culture system was recorded once per hour for 24 hr. (B): The fluorescence intensity per OD600 of PAO1/pCdrA::gfp after 24 hr co-culture with PSL provider strain PAO1, ΔPpsl, ΔpslG, or ΔpslG::pslGE165Q,E276Q. PSL production of each donor strain was shown under corresponding column. Statistical significances were measured using student’s t-test (**, p < 0. 01 when compared to PAO1). (C): The fluorescence intensity and corresponding OD600 of PAO1/pCdrA::gfp during 24 hr of PslG treatment. (D): The fluorescence intensity per OD600 of PAO1/pCdrA::gfp post 24 hr of treatment with different concentration of PslG.

Taken together, our results show that lack of intracellular PslG shapes the localization of PSL on bacterial periphery, enhances the signaling function of PSL, which might further affect the c-di-GMP distribution in daughter cells during cell division, leads to changing of bacterial surface exploration, daughter cells being connected together after cell division and thus the formation of bacterial chains and microcolonies. Strikingly, all these phenotypes depend on the hydrolytic activity of PslG and PSL production, implying that PslG, as a glycosyl hydrolase, can modulate the signaling property of PSL by its hydrolytic activity to affect P. aeruginosa biofilm development.

Discussion

PslG has been shown to be an efficient PSL degrader in vitro and in biofilm matrix, yet its physiologically relevant role within a bacterium or biofilm is ill-defined. In this work, by systematically studying the pslG knock-out mutants both at a single cell and community level, the effects of pslG on the bacterial physiology and surface behavior have been illustrated comprehensively. Based on our results, we propose a model for the role of pslG in the biofilm development of P. aeruginosa as the following (Figure 8).

Figure 8
Download asset Open asset
A schematic to show the effects of pslG in the biofilm development of P. aeruginosa.

The structural analysis of PslG has shown that it has a catalytic domain for the hydrolysis of PSL, and is typically considered to modify or hydrolyze the PSL polymer before PSL are secreted out of the bacterial cell (Yu et al., 2015; Baker et al., 2015). Earlier studies have shown that PslG is localized mainly on the inner membrane and a little in the periplasm (Wu et al., 2019; Baker et al., 2015). It has been proposed that PslG on membrane hydrolyzes PSL polymer during biosynthesis to help the release of PSL at right time (Wu et al., 2019) and the portion in periplasm is to degrade any PSL accumulated in the periplasm space (Baker et al., 2015). In this study, our results revealed that it is very important for P. aeruginosa and its communities to have a glycosyl hydrolase encoding gene, pslG within the polysaccharide synthesis gene cluster. Loss of PslG or its catalytic activity affects the properties and functions of PSL, including its localization and signaling. The consequence of these changes has two aspects. One is that phenotypically long chains of cells are formed. The long chains are not observed in strains with intact pslG. Such long chains seem not induced by the adherence on a surface as they can also be observed in liquid cultures. Rather, there are multiple lines of evidences to support that PSL is the main factor for the formation of long chains. Firstly, by adding PslG externally into cell cultures, long chains were observed to disassemble into single cells, presumably due to the hydrolysis of PSL by PslG as shown in literature (Yu et al., 2015; Baker et al., 2015; Zhang et al., 2018). Secondly, such long chains are not observed in PBAD-pslΔpslG when PSL production is not induced (under 0% arabinose). Thirdly, the fluorescence staining experiments also show that PSL are often localized around septa of cells, where daughter cells are supposed to be disconnected after division.

The other aspect as a consequence of the change in PSL due to the loss of PslG is that, at the molecular level, the probability of both daughter cells having a higher level of c-di-GMP than that of their mother cell in a division event is increased. The increased level of c-di-GMP then would result in reduced cell motility and promote cells to transit to biofilm style (Römling et al., 2013). This may also help cells to form long chains by reducing the breakage of chains due to reduced cell motility and increased PSL production.

As it is widely known that PSL plays a very important role in the biofilms of P. aeruginosa, then both aforementioned aspects would have an effect on the biofilm development at different stages. At the attachment stage, reduced swimming motility, which can be caused by the higher level of c-di-GMP, can contribute to the reduced surface attachment observed on microtiter surfaces. Long chains of cells may also contribute to the reduced swimming motility. However, under our experimental setup, we cannot measure the swimming of long chains in liquid cultures when they move toward to the air-liquid interface. After cells attach on a surface, both the formation of long chains and the increased level of c-di-GMP in two daughter cells will help both daughter cells to stay on the surface, and result in a more hierarchical bacterial visit distribution (Figure 3), which then lead to an earlier formation of microcolony. As cells continue to grow and proliferate, such differences in cell behavior between ΔpslG and PAO1 finally result in matured biofilms with different structures as shown in pellicles formed at the air-liquid interface, where pellicles of ΔpslG are rougher than those of PAO1 although their biomasses are similar. We note that the aforementioned effect of pslG on the biofilm development is also dependent on the synthesis of PSL.

Thus, by revealing the important roles of PslG on bacterial physiology and the biofilm development, our results further expand our understandings of PslG functions. To the best of our knowledge, the ability of a glycosyl hydrolase to affect bacterial c-di-GMP levels has not been reported until this work. It would be also very interesting to see in the future whether other glycosyl hydrolases in different bacterial species can have similar functions.

Similarly, glycosyl hydrolase has also been found to be encoded in gene clusters that are involved in the synthesis of other exopolysaccharides. For example, algL, a gene within the alginate synthesis operon of P. aeruginosa, also encodes a lyase to degrade alginate (Franklin et al., 2011). An early work showed that ΔalgL in a CF isolate mucoid strain FRD1 results in cell death, due to accumulation of alginate in the periplasm (Jain and Ohman, 2005). Interestingly, a later work showed that ΔalgL in a PAO1-derived mucoid strain PDO300 does not cause cell death, yet increases alginate production (Wang et al., 2016). A recent study further revealed that the function of AlgL in PAO1 is to clear the periplasmic space of accumulated alginate during polymer biosynthesis, while lack of AlgL might inhibit bacterial growth under certain conditions (Gheorghita et al., 2022). pelA is another example of encoding a hydrolase in the PEL synthesis operon (Franklin et al., 2011), which has been shown to inhibit biofilm formation as that of PslG (Baker et al., 2016). It would be interesting to investigate whether algL or pelA affects bacterial intracellular c-di-GMP levels.

In summary, in this study, we have provided a comprehensive analysis on the effect of pslG on the biofilm development of P. aeruginosa. Our results indicate that although pslG is not essential for synthesis of PSL, it plays an important role in regulating the proper functions of PSL, and loss of pslG or its hydrolytic activity results in malfunction of PSL, which then cause changes in both morphology and surface behavior of bacterial cells through PSL-mediated interactions. This work shed light on better understanding the role of PslG and would be helpful in developing new ways for biofilm control through pslG-based PSL regulation.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Pseudomonas aeruginosa)PAO1Holloway, 1955Prototroph.
Strain, strain background (Pseudomonas aeruginosa)ΔpslGWu et al., 2019In-frame deletion of pslG.Ma’s lab, strain No. P539.
Strain, strain background (Pseudomonas aeruginosa)ΔPpslMa et al., 2006PSL-negative, the promoter of psl operon deletion mutant.Previous name is WFPA80.
Strain, strain background (Pseudomonas aeruginosa)PBAD-pslMa et al., 2006PSL-inducible strain, the promoter of psl operon is replaced by araC-PBAD.Previous name is WFPA801
Strain, strain background (Pseudomonas aeruginosa)ΔPpelMa et al., 2012PEL-negative, the promoter of pel operon deletion mutant.Previous name is WFPA830.
Strain, strain background (Pseudomonas aeruginosa)PBAD-pelMa et al., 2012PEL-inducible strain, the promoter of pel operon is replaced by araC-PBAD.Previous name is WFPA831.
Strain, strain background (Pseudomonas aeruginosa)PBAD-psl ΔpslGThis studyIn-frame deletion of pslG in PBAD-psl background.Ma’s lab, strain No. P977.
Strain, strain background (Pseudomonas aeruginosa)ΔpslG ΔPpelThis studyThe promoter of pel operon deletion strain in ΔpslG background.Ma’s lab, strain No. P1717.
Strain, strain background (Pseudomonas aeruginosa)PBAD-psl ΔpslG ΔPpelThis studyThe promoter of pel operon deletion strain in PBAD-psl ΔpslG background.Ma’s lab, strain No. P1711.
Strain, strain background (Pseudomonas aeruginosa)ΔalgDWhitchurch et al., 2002Alginate-negative, the algD:: tet deletion mutant of PAO1.Previous name is WFPA1.
Strain, strain background (Pseudomonas aeruginosa)ΔpslG::pslGWu et al., 2019pslG was inserted into pslG deletion mutant at chromosome pslG locus.Ma’s lab, strain No. P963.
Strain, strain background (Pseudomonas aeruginosa)ΔpslG::pslGE165Q, E276QWu et al., 2019pslG was replaced by the active sites mutated pslG (E165Q + E276 Q).Ma’s lab, strain No. P964.
Strain, strain background (Pseudomonas aeruginosa)ΔpslG attB::PBAD-pslGWu et al., 2019PBAD-pslG was inserted into pslG deletion mutant at chromosome attB site.Ma’s lab, strain No. P1716.
Strain, strain background (Pseudomonas aeruginosa)PBAD-psl ΔpslG::pslGThis studypslG was inserted into PBAD-pslΔpslG strain at chromosome pslG locus.Ma’s lab, strain No. P967.
Strain, strain background (Pseudomonas aeruginosa)PBAD-pslΔpslG::pslGE165Q,E276QWu et al., 2019pslG was replaced by the active sites mutated pslG (E165Q, E276Q) in PBAD-psl strain.Ma’s lab, strain No. P966.
Strain, strain background (Pseudomonas aeruginosa)PBAD-psl ΔpslG attB::PBAD-pslGThis studyPBAD-pslG was inserted into PBAD-psl ΔpslG strain at chromosome attB site.Ma’s lab, strain No. P1715.
Strain, strain background (Pseudomonas aeruginosa)PAO1/pCdrA::gfpRybtke et al., 2012PAO1 strain carrying plasmid pCdrA::gfp. AmpR, GmR.
Strain, strain background (Pseudomonas aeruginosa)PAO1/pHERD20TWu et al., 2019PAO1 strain carrying plasmid pHERD20T
Strain, strain background (Pseudomonas aeruginosa)ΔpslG /pHERD20TWu et al., 2019ΔpslG strain carrying plasmid pHERD20T
Strain, strain background (Pseudomonas aeruginosa)ΔpslG /pGYu et al., 2015ΔpslG strain carrying plasmid pHERD20T-pslG. AmpR
Strain, strain background (Pseudomonas aeruginosa)ΔpslG /pGDMWu et al., 2019ΔpslG strain carrying plasmid pHERD20T-pslGDM, pHERD20T with double active sites mutated pslG(E165Q + E276 Q), AmpR
Recombinant DNA reagentpCdrA::gfp (plasmid)Rybtke et al., 2012pUCP22Not-RNase III-gfp (ASV)-T0-T1, a cyclic di-GMP level reporter consisting of the cyclic di-GMP-responsive cdrA promoter fused to gfp gene, AmpR, GmR
Recombinant DNA reagentpHERD20T (plasmid)Qiu et al., 2008pUCP20T Plac replaced with 1.3 kb AflII-EcoRI fragment of araC-PBAD cassette. AmpR
Recombinant DNA reagentpG(plasmid)Yu et al., 2015pHERD20T-pslG. AmpR
Recombinant DNA reagentpGDM (plasmid)Wu et al., 2019pHERD20T with double active sites mutated pslG (E165Q + E276 Q), AmpR
Recombinant DNA reagentpEX18Gm (plasmid)Hoang et al., 1998Allelic exchange vector, GmR
Recombinant DNA reagentpSW196 (plasmid)Baynham et al., 2006Modified from mini-CTX with pBAD30-based vector, for inserting an arabinose-inducible gene at the neutral attB site. TcR
Recombinant DNA reagentpFLP2 (plasmid)Hoang et al., 1998Source of Flp recombinase, AmpR
Antibodyanti-ePsl (Rabbit polyclonal)Byrd et al., 2009Exopolysaccharide Psl specific antibody.IF(1:1667)
OtherTRITC-HHAEY-lab, INCFluorescent labeled lectin HHA.
OtherFITC-HHAEY-lab, INCFluorescent labeled lectin HHA.
OtherSyto9Invitrogen, Molecular probesGreen-fluorescent nucleic acid stain.
OtherFM4-64Invitrogen,Molecular probesLipophilic Styryl Dye.

Bacterial strains and growth conditions

Request a detailed protocol

All P. aeruginosa stains used in study were listed in the key resources table. P. aeruginosa stains were grown at 37 °C in LB without sodium chloride (LBNS) or Jensen’s, a chemically defined media (Jensen et al., 1980). Biofilms of P. aeruginosa were grown in Jensen’s medium at 30 °C. L-arabinose (Sigma) was used as inducer for genes transcribed from PBAD promoter in P. aeruginosa. Antibiotics for P. aeruginosa were added at the following concentrations: gentamicin 30 μg/mL; ampicillin 100 μg/mL; carbenicillin 300 μg/mL. For Pseudomonas selection media, irgasan at 25 μg/mL was used.

The psl-inducible strains PBAD-pslΔpslG was constructed in accordance with PBAD-psl (the promoter of psl operon in PAO1 was replaced by araC-PBAD-psl) (Ma et al., 2006). Briefly, plasmid pMA9 (Ma et al., 2006) was transferred into pslG deletion mutant by conjugation (Wu et al., 2019). All deletion mutants were constructed by the similar in-frame deletion strategy (Ma et al., 2006). For single recombination mutant selection, LBNS plates with 30 μg/mL gentamycin and 25 μg/mL irgasan were used; for double recombination mutant selection, LBNS plates containing 10% sucrose were used. Gene insertion at the attB site of P. aeruginosa was performed as described previously (Hoang et al., 2000).

Bacterial attachment on microtiter dish

Request a detailed protocol

The assay was done as described previously with modifications (Ma et al., 2006; O’Toole, 2011). Overnight culture was 1/100 diluted into Jensen’s media (with or without arabinose) and incubated at 37 °C with shaking until the OD600 reached 0.5. The 100 μL of such culture was inoculated into 96-well PVC microtiter dish (BD Falcon), and incubated at 30 °C for 30 min. Then the planktonic and loosely adherent bacteria cells were washed off by rinsing the plate in water. The remaining surface-attached cells were stained by 0.1% crystal violet, solubilized in 30% acetic acid, and finally the value of OD560 was measured.

Motility assay

Request a detailed protocol

Swimming motility assay was performed as preciously described (Zhao et al., 2018). Briefly, strains were grown overnight on LBNS plates. Single colony was stab-inoculated with a sterile toothpick on the surface of Jensen’s plates (0.3% BD Bacto Agar). Plates were incubated upright at 37 °C overnight. Swimming zones were measured accordingly. For the twitching motility assay, the strains were stab inoculated with a sterile toothpick into the bottom of thin Jensen’s plates, cultivated at 30 °C for 2–3 days, and the twitching motility zones were visualized at the agar plate interface (Wang et al., 2013).

PSL dot-blotting

Request a detailed protocol

Strains were incubated in Jensen’s medium with shaking at 30 °C for 24 hr. Cells of an OD600 of 4 were collected by centrifugation to extract crude bacterial surface-bound exopolysaccharides. Pellet was re-suspended in 100 μL of 0.5 M EDTA, and boiled at 100 °C for 5 min. After centrifugation at 13,000 g for 10 min, the supernatant fraction was treated with 0.5 mg/mL proteinase K at 60 °C for 1 hr and proteinase K was then inactivated at 80 °C for 30 min. PSL immunoblotting was performed as previously described using PSL antibody (Byrd et al., 2009). ImageJ software was used to quantify the immunoblot data. The protein concentration of each sample culture was measured by a BCA protein assay kit (Thermo) to ensure the same amount of cell lysate was used in each experiment.

Flow cell assembly, sterilization, and washing of the system

Request a detailed protocol

Flow cells made of polycarbonate were purchased from the Department of Systems Biology, Technical University of Denmark. Each flow cell has three identical rectangle channels (40 × 4 × 1 mm3) and was assembled by attaching a cover glass as substratum as previously described (Sternberg and Tolker-Nielsen, 2006). The assembled flow cell was connected to a syringe through a 0.22 μm filter (Millipore) using silicon tubing. Then the whole system was sterilized overnight with 3% H2O2 at 3 mL/hr using a syringe pump (Harvard Apparatus). After sterilization, autoclaved, deionized water was used to wash the whole system overnight. Before inoculation of bacteria into the flow cell, the system was flushed for 5 min at a flow rate of 30 mL/hr by Jensen’s medium using a syringe pump (Harvard Apparatus). Then the medium flow was stopped and 1 mL of a diluted bacteria culture (OD600 ~ 0.01) were injected directly into the channel of the flow cell using a 1 mL syringe equipped with a needle. A 5-min incubation period was allowed after inoculation to let cells attaching to the surface, which was then followed by a medium flow with a large flow rate of 30 mL/hr for 5 min to wash out floating cells. After that the flow rate was set to 3 mL/hr, and image recording was started. In this work, the flow cell experiments were conducted at 30 ℃.

Biofilms and image acquisition

Request a detailed protocol

Pellicles (air-liquid interface biofilms) were grown in glass chambers (Chambered # 1.5 German Coverglass System, Nunc) with a glass coverslip at the bottom of each chamber as described previously (Wang et al., 2013). 1/100 dilution of a saturated (overnight) culture in Jensen’s media for P. aeruginosa was inoculated into the chamber, and incubated at 30 °C for 24 hr. The PSL was stained with lectin TRITC-HHA (EY lab, INC) at 100 μg/mL for 2 hr in the dark. Then bacteria were strained with SYTO9 (5 μM final concentration, Molecular Probes, Invitrogen) for 15 min. Fluorescent images were obtained using a FV1000 CLSM (Olympus, Japan). The excitation/emission parameters for TRITC-HHA and SYTO9 were 554 nm/570 nm and 480 nm/500 nm, respectively. CLSM-captured images were analyzed using COMSTAT software (Heydorn et al., 2000).

For flow cell experiments, the flow was stopped before staining. PSL was stained with lectin FITC-HHA (EY lab, INC) at 100 μg/mL for 20 min in the dark, and then the flow was running for a short time to flush out the non-binding dye. Subsequently, bacteria were tagged by Gfp or stained by cell membrane stain FM4-64 (10 μM final concentration, Molecular Probes) for 2 min in the dark (without flow). Next, the flow was resumed to flush out the dye and ready for examination under microscope. Images were captured using an EMCCD camera (Andor iXon) on a Leica DMi8 microscope equipped with Zero Drift autofocus system. The image size is 66.5 μm × 66.5 μm (1,024 × 1,024 pixels). The images were recorded with a 100 × oil objective (plus 2 × magnifier).

Detection of PSL signaling function on stimulating bacterial intracellular C-di-GMP

Request a detailed protocol

The c-di-GMP levels were determined using pCdrA::gfp as a reporter as described previously (Rybtke et al., 2012). The growth curve and green fluorescent signal of PAO1 /pCdrA::gfp were measured via recording the OD600 values and the corresponding GFP fluorescence (Ex/Em 488/520) by a Synergy H4 hybrid reader (BioTek). The promoter activity was calibrated as the relative fluorescence divided by the OD600. In co-culture system, PAO1 harboring plasmid pCdrA::gfp was the reporter strain to indicate the level of intracellular c-di-GMP. PAO1, PBAD-psl, ΔpslG, and ΔpslG::pslGE165Q,E276Q strains were PSL donor strains, respectively. In PslG treatment assay, PslG was added into cultures when inoculating PAO1 /pCdrA::gfp, the OD600 values and the corresponding GFP fluorescence were tracked for 24 hr.

Single-cell tracking image analysis

Request a detailed protocol

Images were processed and analyzed in the same way as described in reference (Zhang et al., 2018). Simply, 16-bit greyscale images were first converted to binary images for the detection of bacteria with a standard image processing algorithm. Geometry information of cells such as center position, size and aspect ratio etc. were then collected. Bacterial trajectories were obtained by connecting cell positions in all frames of a time series, from which bacterial motion can be measured and analyzed. Specifically, the twitching speed of each tracked cell at frame n was calculated by the displacement of the cell between nth and (n + 1)th frames divided by the corresponding time interval.

For quantitatively comparing the fluorescence intensity of cells containing pCdrA::gfp reporter between mother cell and daughter cells, first the fluorescence intensity of each cell I was measured by the averaged fluorescence intensity value within the area enclosed by the cell envelope. The mother cell was measured when it irreversibly attached to the surface (typically 40~50 min before the division completed), and the daughter cells were measured right after the division (i.e. the two daughter cells are completely separated. In practice, the daughter cells were measured within 10 min right after the division completed due to the 10-min time interval for the fluorescent image recording). Then the ratio of fluorescence intensity between each daughter cell (Idau) and its mother cell (Imot) was calculated, γ = Idau/Imot. The relative standard deviation was estimated by SDγγ=(SDIdauIdau)2+(SDImotImot)2. Here, SDγ refers to standard deviation of γ. Similar for SDIdau and SDImot. We define a daughter cell to be fluorescent brighter than its mother cell if γ>1+(SDγγ)¯, here (SDγγ)¯ is the averaged value of the relative standard deviation for all analyzed division events.

A cluster is an aggregation of multiple cells. We used a minimum distance criterion to judge whether a cell belonged to a cluster or not. If the minimum distance between any point of the scrutinized cell body and any point of any cell body of the cluster, is smaller than 0.5 μm (i.e. about one width of a bacterial cell), then the scrutinized cell is considered to belong to the cluster, otherwise not.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting file. Source Data files have been provided for Figure 1,2, 3,4,6, and 7.

References

Decision letter

  1. Gerald Pier
    Reviewing Editor; Brigham and Women's Hospital, United States
  2. Wendy S Garrett
    Senior Editor; Harvard T.H. Chan School of Public Health, United States
  3. Joe Jonathan Harrison
    Reviewer; University of Calgary, Canada

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Intracellular glycosyl hydrolase PslG shapes bacterial cell fate, signaling, and the biofilm development of Pseudomonas aeruginosa" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Joe Jonathan Harrison (Reviewer #2).

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

While there was clear interest in the overall subject and the direction of the experimental approaches, the reviewers found multiple deficiencies in the overall experimental design that diminished the strength of the conclusions based on the data presented. Thus, the ability to complete numerous experiments needed to strengthen the data within 2 months is highly unlikely and serves as the basis for the decision. In particular the lack of trans-complemented strains or use of strains with point mutations in PslG raised issues about pleiotropic effects and specificity. Lack of building on prior work in the field and citing prior results on the alginate lyase, AlgL, which led to a similar phenotype, diminishes the claims to novelty and primacy. Also, the impact of two other surface polysaccharides, Pel and alginate was not explored. Most Pseudomonas aeruginosa strains make alginate at a low level, but PAO1 does not do so under aerobic conditions in vitro, so use of this strain as the major one to test the hypotheses and draw conclusions could represent findings not generally applicable to P. aeruginosa. Additional concerns about the interpretation of the Raman spectra and weaknesses in the data that support the conclusions regarding the basis for the effect of loss of PslG on the cellular structures further diminish the impact of the findings.

Reviewer #1:

The manuscript primarily presents phenotypic results to shape their story and conclusions. Given the importance of both Psl and Pel to exopolysaccharide-dependent behaviors in P. aeruginosa PAO1 (and many other strains), the potential importance and contribution of Pel is essentially ignored here. Because both Pel and Psl can generally influence several of the phenotypes explored in this manuscript, the authors' desire to draw specific conclusions from the effect of PslG cannot be separated from potential contributions of Pel polysaccharide in these experiments. Another difficulty in assessing the results and conclusions lies in the fact that some experiments utilize only the WFPA801ΔpslG or ΔpslG strains, which are inherently different (Figure 3 shows ΔpslG with PAO1 as a control while Figure 4 shows WFPA801ΔpslG with PAO1 as a control). Thus, there are two parallel lines of evidence that do not fully converge. In addition, as the pslG mutation is not complemented in any set of experiments, the specificity of pslG to the effects detailed here is not definitely demonstrated. There may be much to be learned by probing an arabinose-inducible pslG construct as the authors have done with overall Psl production.

The most compelling data in the manuscript involves the impact of pslG upon divided cells. Both the c-di-GMP reporter and cell-chain data point to a role of PslG in disposition of progeny. These results included in Figures 4, 6, and 7 are fascinating but require further probing to actually explain the role of PslG.

A deeper interrogation of cell division may uncover a direct connection between PslG and the cell cycle. Several other types of experiments may be useful to discern this connection.

The manuscript misses the opportunity to reference and build upon recent work by the Wozniak and Parsek labs on Pel and Psl (from just the last 6 years).

Line 116. From all of the descriptions here, the methods, and the figure 1 legend, it is not clear what has been quantified in Figure 1A as the measure of Psl.

Line 131. Even though the WFPA801ΔpslG strain shows a "statistically significant" decrease in swim zone diameter in comparison to WT or un-induced conditions, the difference is minor. Certainly, this strain exhibits flagellar motility. Given the small difference in flagellar motility, which is used here as an indirect assessment of flagellar function that may apply to surface attachment, it would be more prudent to conclude "…the defect of ΔpslG on initial attachment MAY BE the result of multiple contributing factors."

Line 173. "…by changing the surface exploration at early stages of biofilm formation…" I disagree this is what the data show here and suggest this contradicts the data discussed lines 157-164 in the same paragraph where surface coverage differences are not apparent between pslG and PAO1 at N<10,000. Indeed, it would be the impact of pslG upon the 10,000-100,000 range of visits that begs for further discernment. Or what is the working definition of "early stages" here?

Line 228. The Raman spectra of the three strains clearly show many differences in presence and intensity of numerous peaks between 800-1800 cm-1. In the absence of more specific and supportive references and other controls, it is not clear that the 865 cm-1 peak is the definitive feature between these three spectra nor is it clear this peak allows for the specific discernment of a 1,4-glycosidic linkage polysaccharide. Lastly, it is not clear how the authors can even discern Psl from Pel polysaccharide by Raman given the data presented here. There may be no "structural change in Psl" observed here but rather a shift from Psl to Pel.

Line 238. These results are difficult to interpret. If the ΔpslG strain makes less Psl, then how is the chain-forming phenotype dependent upon increased Psl production in the WFPA801ΔpslG strain?

Reviewer #2:Pseudomonas aeruginosa is an opportunistic pathogen that is one of the leading causes of bacterial infection worldwide. The ability of this bacterium to form biofilms, which are aggregates of these microorganisms encased in a protective layer of polymers, is key for its ability to survive in different environments and cause disease. In this report, the role of a glycoside hydrolase enzyme, PslG, which is co-produced with more than a dozen other enzymes used by the bacteria to build their protective biofilm polymeric matrix, is investigated. The authors provide evidence that this enzyme determines cell fate in the early stages of P. aeruginosa biofilm formation by enabling cell detachment post-division. Understanding the natural functions of these enzymes is important because they comprise an emerging class of experimental therapeutics that target biofilms.The data presented by the authors in the manuscript are intriguing. However, the paper needs to be more carefully written because the impact will be tempered by the presentation. Understanding the natural role of glycoside hydrolases in biofilm development is a fundamental problem for the field that microbiologists have been pondering for more than a decade. Most of the conclusions are supported by data, but findings are not communicated as effectively as they could be, which detracts from what otherwise is a very nice dataset. There are some controls that are missing as well.

Key constructive criticisms

The authors should complete a complementation analysis for pslG in all their assays. As the authors note, the pslG gene is in an operon and so polar effects are a concern, and because pslG mutations appear pleiotropic, ensuring that phenotypes do not result from second site mutations elsewhere in the chromosome of engineered strains seems especially important.

To ensure that the hydrolase function of PslG is responsible for the observed phenotypes, it would be prudent to repeat some key assays in which the active site residues of PslG have been disrupted by site-directed mutagenesis.

Line 110-121. It is not clear what the authors are measuring and reporting in Figure 1A. In the text, the authors make statements about the quantities of ePSL produced by the different strains. However, the data in Figure 1 has no units and it is not clear whether the authors are measuring ePSL directly, or if the data represents something else. Could the authors quantify biofilm formation shown in the pictures below Figure 1A? Also, it would be highly beneficial if a semi-quantitation of the dot blots shown was provided too. Corresponding text should be added to the figure legend to succinctly explain the data presented in the figure.

Line 134-135. Could the authors better explain how the data led to their conclusion that PslG modulates surface attachment through multiple mechanisms? It is not clear from the text why the authors conclude that changes in motility are causally linked to defects in bacterial surface attachment, because the flagellum can be dispensable for biofilm formation depending on the assay. Also, the evidence here indicates that pslG mutations are pleiotropic.

Lines 228-234 and Figure 5 C. Caution needs to be applied using Raman spectroscopy on whole biofilms because the composition of biofilms is complex. How can the authors be certain that changes at 865 cm-1 correlate specifically with a glycosidic link in ePSL? There has been a demonstration that deletion of pslG affects intracellular c-di-GMP which influences many things in physiology including multiple components of the P. aeruginosa biofilm matrix. The authors would need to purify PSL from the WFPA801 and WFPA801 pslG strains and analyze that material to make these conclusions with confidence.

Lines 212-227 and Figure 5A and B. Why isn't data for the pslG mutant also included in Figure 5A?

Lines 276-277. In this case, "data not shown" is not acceptable because the WFPA801 and WFPA801 pslG are engineered to have artificially high levels of PSL production. The data that are not shown correspond to key information corroborating physiological relevance to the wild type background. Please add these data to the manuscript.

Reviewer #3:

In this manuscript, the authors investigated the role of PslG, a Pseudomonas aeruginosa glycosyl hydrolase, in biofilm formation and cell physiology of a P. aeruginosa. PslG is encoded on the 15 gene psl operon, and a previous study indicated that deletion of pslG caused reduced ePSL (extracellular PSL polysaccharide) production. However, in the previous study, the pslG deletion may have been polar on other downstream psl genes and may have affected ePSL production. Therefore, in this study the investigators constructed a pslG deletion in a strain where the entire psl operon is controlled by the arabinose-inducible promoter. In this way, ePSL is produced but the operon lacks pslG. The investigators showed that in this strain, ePSL is produced at levels similar to the wild-type strain. However, the pslG deletion strain had several phenotypic defects that suggested a functional role of the glycosyl hydrolase. Among the defects was a reduction in biofilm formation, even when ePSL was produced, and a reduction in swimming motility. The investigators also used microscopic tracking of the cells, and noticed a reduction in surface area covered, likely due to clumping of the cells. The clumping of the cells was due to their formation of cell chains, which were not observed in the wild-type strain or in the new strain, when ePSL production was not induced with arabinose. Interestingly, the investigators noted that in the cells that were unable to separate following division (cell chains) had ePSL accumulated at the boarders between the cells. The chains could be released by adding exogenous PslG enzyme to the cultures. The investigators suggested that in the absence of PslG, the ePSL may accumulate at the cell boarders, holding the daughter cells together. In addition to these phenotypes, the investigators also noticed a difference in the production of the cyclic-di-GMP signaling molecule in the pslG mutant strain. Since c-di-GMP is an important signaling molecule for biofilm production, the investigators put forth a model for the role of PslG in c-di_GMP signaling in the development of P. aeruginosa biofilms.

Overall, I think this is a thorough study and the results are well presented. A weakness of this study (in my opinion) is that while adequately exploring the phenotypes of the PslG mutant in an ePSL producing strain, a functional role for PslG in polysaccharide degradation during ePSL biosynthesis is not adequately developed. The model in Figure 8 (and the abstract) implies that the role of PslG is through c-di-GMP signaling molecule. However, the data in Figure 4B and Figure 5B show only a minor effect of the pslG deletion on c-di-GMP signaling. In Figure 4B, most cells are classified as "none-bright" in both the experimental and control strain. Only a minor percentage of cells are classified as "two-bright". The difference in "two-bright" between the control and experimental in Figure 4B, although statistically different, is not a high percentage of the total cells. Figure 5A shows a moderate difference in c-di-GMP production as indicated by a reporter gene. However, Figure 5A does not include a growth curve, so the difference seen in fluorescence could be due to a difference in cell numbers/growth. This would not be surprising, given the many physiological defects of the pslG mutant. (a growth curve needs to be shown to determine if the fluorescence difference is due to cell growth differences). The model in Figure 8 implies that c-di-GMP signaling (and the "two-bright" daughter cells) is the primary signaling event for biofilm formation in the pslG mutant. In my opinion, this is overstated, given the minor effect of the pslG mutation on c-di-GMP production. As far as I know, PslG does not have a c-di-GMP binding domain, so it is difficult to draw a molecular link between the pslG mutant and c-di-GFP. The slight increase in c-di-GMP in the mutant cells may be due to an indirect effect of an overall stress response in the mutant cell, and not a direct signaling mechanism.

I recommend that the authors read and cite the publication, "Role of an alginate lyase for alginate transport in mucoid Pseudomonas aeruginosa" Infect Immun. 2005 Oct;73(10):6429-36.doi: 10.1128/IAI.73.10.6429-6436.2005.

In the discussion, the authors state, "To the best of our knowledge, the ability of a

glycosyl hydrolase to affect bacterial phenotype and c-di-GMP levels has not been

reported until this work. It would be also very interesting to see in the future whether

other glycosyl hydrolases in different bacterial species can have similar functions."

In the paper above, the authors characterized another glycosyl hydrolase from Pseudomonas aeruginosa, that degrades alginate, and is encoded on the alginate biosynthetic operon (so, a very similar model system). In that study, the authors found that deletion of algL in an alginate producing strain resulted in cell death, due to accumulation of alginate in the periplasm. There is a precedence for these types of studies. The mechanism for PslG may be different from AlgL (and the PslG deletion obviously doesn't kill the bacteria), but the AlgL study could provide a hypothesis on the molecular role of PslG. In addition, both AlgL and PslG reside in the periplasm (assuming the model in reference: Frontiers in Microbiology 2:167 is correct). The authors of the current study only mention the periplasmic location for PslG in the Discussion section. The implication in the present study is that PslG is secreted. If PslG is secreted, there should be evidence for that (or a citation that shows that it is secreted). The localization of this enzyme could be crucial in understanding its function.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Intracellular glycosyl hydrolase PslG shapes bacterial cell fate, signaling, and the biofilm development of Pseudomonas aeruginosa" for further consideration by eLife. Your revised article has been evaluated by Wendy Garrett (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1:

The manuscript seeks to delineate the role of pslG, a glycosyl hydrolase encoded within an operon that enables the production of the Psl polysaccharide of the bacterium Pseudomonas aeruginosa. The results in the manuscript show numerous effects using a combination of pslG mutants and over-expression strains upon P. aeruginosa behavior important to colonization and biofilm development. This adds to our understanding of the multi-functional enzymes involved in regulating polysaccharide production and the overall behavior of P. aeruginosa.

The manuscript primarily presents phenotypic results to shape their story and conclusions. Given the importance of both Psl and Pel to exopolysaccharide-dependent behaviors in P. aeruginosa PAO1 (and many other strains), the potential importance and contribution of Pel remains essentially ignored in this manuscript. Because both Pel and Psl can generally influence several of the phenotypes explored in this manuscript, the authors' desire to draw specific conclusions from the effect of PslG cannot be separated from potential contributions of Pel polysaccharide in these experiments.

The story is certainly complicated as study of pslG mutants does affect levels of Psl polysaccharide. The most compelling data in the manuscript involves the impact of pslG upon divided cells. Truly fascinating. However, it is not clear that the authors demonstrate the importance of PslG to "normal" cell division and/or how Psl polysaccharide might contribute to cell division in the wild-type. There is no basis in the literature to suggest that pslG mutations would be common in select environments.

Several new lines of experimental evidence have been added to the manuscript in this revised form and these changes are a great collective improvement-but the true role(s) and effect of PslG are still not clear from the evidence presented.

There are several sentences where the English syntax and grammar require editing.

Line 159. There does not appear to be fewer bacterial cells in the pslG biofilms-and no quantification is given to support this statement.

Line 178. "…a more hierarchical bacterial visit distribution for pslG…" It is not clear how this shows hierarchical visitors or a broad range of visit frequencies. How is this description evident based upon what is shown in Figure 3B?

Line 184. There appear to be 27-28 cells in the PAO1 cluster detailed in Figure 3D.

Line 188. "…microcolonies of ΔpslG are more compact". This representation of the data in Figure 3F does not necessarily support this statement. Is this a projection view showing a 2D image of the thicker 3D biofilms as described in Figure 2? How was this quantified?

Line 221. How were these twitching speed experiments performed? These are not described in the supplemental Figure legend or the methods.

Line 240. "…were observed most frequently in either ΔpslG or PBAD-pslΔpslG strain…" The purpose of this statement is not clear. These are the only strains shown in Figure 4. Is this a comparison?

Line 241. What is the support for the statement "…yet some can also be broken by bending of chains"? This is from separate experimental evidence? Or just an observation?

Line 246. This Figure 5 data is the most intriguing of the manuscript.

Line 309. "…which might enhance their stay on the surface, reduce their surface motility, and promote the microcolony formation". Given the small variation in the data shown in Figure 6B, support for this statement really requires specific testing. Further, not all previous data in the manuscript support it. The text in an earlier section (starting line 213) effectively concludes that there is no quantifiable motility difference between these strains.

Line 497. The Psl staining with TRITC-HHA is specific to Psl? Is there binding to Pel or other polysaccharides?

Line 770. The Figure 1 title is not the best description of what is shown here. No "increasing" levels of Psl are shown. There are two. The data show adding of Psl into a pslG background.

Line 821. It is not stated anywhere in the legend or methods from how many frames the data analyzed for Figure 3E were obtained. There are error bars, but it is not clear what these represent.

Line 831. The criteria for "microcolony" is not entirely clear. Some of these cells clearly have different spacing but are outlined with dashed lines.

Reviewer #2:

The manuscript by Zhang and colleagues carefully describes the phenotypes of pslG mutants of P. aeruginosa. PslG encodes a glyocoside hydrolase. While the biochemistry of this enzyme has been understood for many years, its physiologically relevant role in P. aeruginosa biofilm formation has remained ill-defined.

The experiments in the manuscript have been meticulously executed. There are controls and complementation analyses that provide confidence in the results obtained. The technical proficiency with microscopy is commendable. However, while the authors provide a data-rich manuscript, an understanding of the consequences of PslG expression appears lacking beyond phenotyping.

Perhaps this criticism is most pertinent to the observed changes in the power-law distribution for bacteria during the earliest stages of biofilm formation for pslG mutants. Interpretation/experimentation is absent that connects these observations to social biology. Such connections, which are front and center in prior work published by one of the co-authors (Dr. Khun Zhao), could help to explain conservation of glycoside hydrolases among synthase-dependent exopolysaccharide secretion systems like the Psl synthase. For example, beyond the careful phenotyping presented in this paper, co-culture of Pbad-Psl with Pbad-Psl-PslG strains, or perhaps wild type and pslG strains, that have been uniquely labelled with fluorescent proteins and tracked using Dr. Kun Zhao's elegant single-cell methods could directly demonstrate fitness changes for pslG mutants in surface exploration or colonization relative to wild type. In principle, wouldn't such a fitness cost provide an explanation for PslG function that is rooted in social evolutionary theory? Perhaps there are some trade-offs that aren't yet apparent. The link to c-di-GMP signaling provides some molecular insight even the sensory perception and signal transduction pathway is not yet fully known. Such analyses could take the work assembled here to the next level with little additional experimental effort, and as such, strikes me as a missed opportunity to provide significant, additional understanding of some really nice data.

https://doi.org/10.7554/eLife.72778.sa1

Author response

[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

While there was clear interest in the overall subject and the direction of the experimental approaches, the reviewers found multiple deficiencies in the overall experimental design that diminished the strength of the conclusions based on the data presented. Thus, the ability to complete numerous experiments needed to strengthen the data within 2 months is highly unlikely and serves as the basis for the decision. In particular the lack of trans-complemented strains or use of strains with point mutations in PslG raised issues about pleiotropic effects and specificity.

We have collected some complementation data. Moreover, since we have trans-complemented strains and point mutations in PslG on hand, it would not take very long to perform more experiments if it is necessary.

Lack of building on prior work in the field and citing prior results on the alginate lyase, AlgL, which led to a similar phenotype, diminishes the claims to novelty and primacy.

We apologize for missing this paper. We will certainly add this reference and its associated information in the introduction. As pointed by the Reviewer 3, lacking of AlgL leads to cell death. Thus, the phenotype of AlgL deletion is certainly different from what we have found in PslG mutant.

Also, the impact of two other surface polysaccharides, Pel and alginate was not explored. Most Pseudomonas aeruginosa strains make alginate at a low level, but PAO1 does not do so under aerobic conditions in vitro, so use of this strain as the major one to test the hypotheses and draw conclusions could represent findings not generally applicable to P. aeruginosa.

We will perform more experiments involving strains that lack of Pel and alginate production to explore whether Pel and alginate have any contribution on the phenotype of PslG mutant. Fortunately, we have also Pel-negative strain and alginate-negative strain on hand too. So, it will not take much time to construct strains that we need for such experiments.

In addition, regarding on the alginate production of our lab PAO1 strain, in some of our earlier studies, we have measured the alginate production of our PAO1 under our lab aerobic growth conditions, and the results show that the production of alginate is relatively low, only 0.3% of PAO1-drevied mucoid strain (Ma, L.*, Wang, J., Wang, S., Anderson E.M., Lam J.S., Parsek, M.R., Wozniak, D.J. 2012, Synthesis of multiple Pseudomonas aeruginosa biofilm matrix exopolysaccharides is post-transcriptionally regulated. Environ. Microbiol. 14 (8):1995-2005). We will add corresponding discussion in the revision

Additional concerns about the interpretation of the Raman spectra and weaknesses in the data that support the conclusions regarding the basis for the effect of loss of PslG on the cellular structures further diminish the impact of the findings.

We will perform more controls to confirm Raman spectra of PslG mutant. Right now we have already Pel-negative strain, PBAD-Pel strain, and PAO1 algD deletion strain on hand, which can be used as the controls. In addition, we will construct an in-frame deletion mutant that has pel and algD deletion in δ PslG background in order to further diminish the possible impact of Pel and alginate on the Raman spectra. Since the plasmids for the construction of mutants are ready, we expect the construction of new mutants as well as new controls to be finished within a couple of months.

Reviewer #1:

The manuscript primarily presents phenotypic results to shape their story and conclusions. Given the importance of both Psl and Pel to exopolysaccharide-dependent behaviors in P. aeruginosa PAO1 (and many other strains), the potential importance and contribution of Pel is essentially ignored here. Because both Pel and Psl can generally influence several of the phenotypes explored in this manuscript, the authors' desire to draw specific conclusions from the effect of PslG cannot be separated from potential contributions of Pel polysaccharide in these experiments.

We agree with Reviewer #1 that both PSL and PEL are important to exopolysaccharide-dependent behavior of PAO1. In the revised manuscript, we have added more text about PEL and alginate in the introduction part (See L60-L66).

Based on literature results (for example, Colvin et al., 2012, Environmental Microbiology), PAO1 is a PSL-dominant strain, whose biofilm formation is mainly dependent on PSL. Thus, we speculate that PEL may have little contribution to the phenotypes of pslG mutants observed in this work. Following suggestions of reviewers, we have added the results of PEL-negative strain as well as pslG deletion in PEL-negative background in the revised manuscript (see Figure 3—figure supplement 3 and Figure 5—figure supplement 3). As expected, all phenotypes of pslG mutants are still observed in PEL-negative strains, indicating that the results of this study are not relied on PEL expression.

Another difficulty in assessing the results and conclusions lies in the fact that some experiments utilize only the WFPA801ΔpslG or ΔpslG strains, which are inherently different (Figure 3 shows ΔpslG with PAO1 as a control while Figure 4 shows WFPA801ΔpslG with PAO1 as a control). Thus, there are two parallel lines of evidence that do not fully converge.

Thanks for the comment. In the revised manuscript, we have added more datasets including those for ΔpslG and WFPA801ΔpslG strain (WFPA801 is referred as PBAD-psl in the revised manuscript as Reviewer #2 suggested), which are shown in the revised figures (Figure 1, Figure 3, Figure 3—figure supplement 1, Figure 3—figure supplement 2, Figure 3—figure supplement 3, Figure 5—figure supplement 2, Figure 5—figure supplement 3, Figure 6, Figure 6—figure supplement 1, Figure 7).

In addition, as the pslG mutation is not complemented in any set of experiments, the specificity of pslG to the effects detailed here is not definitely demonstrated. There may be much to be learned by probing an arabinose-inducible pslG construct as the authors have done with overall Psl production.

We thank the Reviewer for the comment. Following reviewers’ suggestions, we have performed more experiments, including those for complementation analysis of pslG by constructing complemented strains. We have tested not only ΔpslG::pslG and PBAD-psl∆pslG::pslG strains, where pslG was inserted into pslG deletion mutant at chromosome pslG locus, but also ΔpslG attB::PBAD-pslG and PBAD-psl ΔpslG attB::PBAD-pslG strains, where PBAD-pslG was inserted into pslG deletion mutant at chromosome attB site. The results show that they all can complement the phenotypes observed in this work. In the revised manuscript, we have added these results in Figure 1, Figure 3—figure supplement 1, and Figure 5—figure supplement 2.

In addition, results in an earlier work (Wu et al. 2019, MicrobiologyOpen) also showed that arabinose-inducible pslG in plasmid vector pHerd20T can complement PSL production of the pslG deletion mutant (same strain used in this study).

The most compelling data in the manuscript involves the impact of pslG upon divided cells. Both the c-di-GMP reporter and cell-chain data point to a role of PslG in disposition of progeny. These results included in Figures 4, 6, and 7 are fascinating but require further probing to actually explain the role of PslG.

We thank the Reviewer for the comment and suggestion. To further understand the role of PslG, we have performed more experiments on the PslG with mutations at the glycoside hydrolytic activity sites. The results indicate that the glycoside hydrolytic activity of PslG is critical for all the phenotypes we have shown in this manuscript, suggesting an important role of the hydrolytic activity of PslG in determining the localization of PSL and its signaling function, which affects the intracellular c-di-GMP level. In the revised manuscript, corresponding results have been added in Figure 1, Figures 3-7 and associated supplement figures. We have also added several sentences in the discussion at L338 –L348.

A deeper interrogation of cell division may uncover a direct connection between PslG and the cell cycle. Several other types of experiments may be useful to discern this connection.

We thank the Reviewer for the comment. Our results show that the long bacterial chains can be disconnected within a few minutes by addition of purified PslG in media (Movie S2 and S3), suggesting that bacteria cells in the chains are actually completely divided daughter cells which are connected by PSL on bacterial periphery. In addition, Measurements of growth curves as well as the measured total biomass of matured biofilms are all similar between WT and pslG deletion mutants. Taken together these results, it seems that the cell division in long chains is very likely not directly connected with PslG. PslG may rather indirectly affect the intracellular c-di-GMP level and the cell cycle through changing the PSL signaling and localization.

In the revised manuscript, we have modified text accordingly (L260-268 and L339-346).

The manuscript misses the opportunity to reference and build upon recent work by the Wozniak and Parsek labs on Pel and Psl (from just the last 6 years).

Following the reviewer’s suggestion, in the revised manuscript, we have added text on PEL and alginate in the introduction part (See L61-L66) and cited four more references, which are listed below:

Jennings LK, Dreifus JE, Reichhardt C, Storek KM, Secor PR, et al. 2021. Pseudomonas aeruginosa aggregates in cystic fibrosis sputum produce exopolysaccharides that likely impede current therapies. Cell Rep 34: 108782.

Jennings LK, Storek KM, Ledvina HE, Coulon C, Marmont LS, et al. 2015. Pel is a cationic exopolysaccharide that cross-links extracellular DNA in the Pseudomonas aeruginosa biofilm matrix. Proc Natl Acad Sci U S A 112: 11353-58

Baker P, Hill PJ, Snarr BD, Alnabelseya N, Pestrak MJ, Lee MJ, Jennings LK, Tam J, Melnyk RA, Parsek MR, Sheppard DC, Wozniak DJ, Howell PL. 2016. Exopolysaccharide biosynthetic glycoside hydrolases can be utilized to disrupt and prevent Pseudomonas aeruginosa biofilms. Science Advances, 2:e1501632. DOI: 10.1126/sciadv.1501632, PMID: 27386527

Ma L, Wang J, Wang S, Anderson EM, Lam JS, Parsek MR, Wozniak DJ. 2012. Synthesis of multiple Pseudomonas aeruginosa biofilm matrix exopolysaccharides is post-transcriptionally regulated. Environmental Microbiology 14:1995-2005. DOI: 10.1111/j.1462-2920.2012.02753.x, PMID: 22513190

Line 116. From all of the descriptions here, the methods, and the figure 1 legend, it is not clear what has been quantified in Figure 1A as the measure of Psl.

We apologize for not expressing the legend clearly in our previous manuscript. What shown in Figure 1A are PSL production measured by dot blotting and the results of initial attachment assay in microtiter dish well of tested strains.

In the revised manuscript, we have modified both the figure legend of Figure 1 and description text (L118-L134) in the revised manuscript.

Line 131. Even though the WFPA801ΔpslG strain shows a "statistically significant" decrease in swim zone diameter in comparison to WT or un-induced conditions, the difference is minor. Certainly, this strain exhibits flagellar motility. Given the small difference in flagellar motility, which is used here as an indirect assessment of flagellar function that may apply to surface attachment, it would be more prudent to conclude "…the defect of ΔpslG on initial attachment MAY BE the result of multiple contributING FACTORS."

We thank the reviewer for the comment. We agree with what the reviewer suggested. In the revised manuscript, the corresponding text has been modified to be: “These results showed that pslG deletion did not affect the function of flagella and T4P directly, yet increasing PSL production in PBAD-pslΔpslG attenuated the swimming ability, which might impact its attachment phenotype.” (L45-L148).

Line 173. "…by changing the surface exploration at early stages of biofilm formation…" I disagree this is what the data show here and suggest this contradicts the data discussed lines 157-164 in the same paragraph where surface coverage differences are not apparent between pslG and PAO1 at N<10,000. Indeed, it would be the impact of pslG upon the 10,000-100,000 range of visits that begs for further discernment. Or, what is the working definition of "early stages" here?

We apologize for not expressing our points clearly. Here, “early stages” is not referred to the period of N<10,000, but referred to the time period from the beginning of bacterial tracking (just after the inoculation of bacterial cells) to the time point when first microcolonies in the field of view appear, which is the same as that defined in the reference (Zhao et al., 2013). Following the five stages of biofilm development defined in an earlier work (D. Monroe, PLoS Biol. 2007), which includes (I) initial attachment, (II) irreversible attachment, (III) maturation I, (IV) maturation II and (V) dispersion, the time period defined by “early stages” here covers stages (I), (II) and (III).For the data shown in Figure 3A,B,C,in terms of number of bacterial visits, “early stages” cover the whole range of N ≤ ∼100,000.

In the revised manuscript, in order to make our point more clearly and to avoid possible misunderstandings as the Reviewer pointed out, we have modified the text "…by changing the surface exploration at early stages of biofilm formation…" to "…by changing the surface exploration during microcolony formation" (L204-L205).

Line 228. The Raman spectra of the three strains clearly show many differences in presence and intensity of numerous peaks between 800-1800 cm-1. In the absence of more specific and supportive references and other controls, it is not clear that the 865 cm-1 peak is the definitive feature between these three spectra nor is it clear this peak allows for the specific discernment of a 1,4-glycosidic linkage polysaccharide. Lastly, it is not clear how the authors can even discern Psl from Pel polysaccharide by Raman given the data presented here. There may be no "structural change in Psl" observed here but rather a shift from Psl to Pel.

We agree with what the reviewer pointed out. To make a definitive conclusion from Raman data, we do need more controls and more specific and supportive references, which are beyond the scope of this study. Given such considerations as well as to be more focused on the role of PslG, in the revised manuscript, we have removed the Raman spectra data to leave them for a future work and added more data on the effect of PslG on PSL signaling (Figure 7 and corresponding description at L310-332).

Line 238. These results are difficult to interpret. If the ΔpslG strain makes less Psl, then how is the chain-forming phenotype dependent upon increased Psl production in the WFPA801ΔpslG strain?

The chain-forming phenotype of ΔpslG requires PSL production but does not need increased PSL production. Therefore, ΔpslG can form long chains even if it produces less PSL than PAO1. In addition, PSL produced by WFPA801 ΔpslG (named as PBAD-psl ΔpslG in the revised manuscript) under 0.5% arabinose is similar to the level of PAO1 as shown in Figure 1B. But under a condition of 0.1% arabinose where PSL production of PBAD-psl ΔpslG strain is presumably less than that under 0.5% arabinose, long chains were also observed (See Figure 5—figure supplement 2). However, under a condition of 0% arabinose where there is no PSL production, we didn’t observe long cell-chains (see Figure 4—figure supplement 1).

Reviewer #2:

Pseudomonas aeruginosa is an opportunistic pathogen that is one of the leading causes of bacterial infection worldwide. The ability of this bacterium to form biofilms, which are aggregates of these microorganisms encased in a protective layer of polymers, is key for its ability to survive in different environments and cause disease. In this report, the role of a glycoside hydrolase enzyme, PslG, which is co-produced with more than a dozen other enzymes used by the bacteria to build their protective biofilm polymeric matrix, is investigated. The authors provide evidence that this enzyme determines cell fate in the early stages of P. aeruginosa biofilm formation by enabling cell detachment post-division. Understanding the natural functions of these enzymes is important because they comprise an emerging class of experimental therapeutics that target biofilms.

The data presented by the authors in the manuscript are intriguing. However, the paper needs to be more carefully written because the impact will be tempered by the presentation. Understanding the natural role of glycoside hydrolases in biofilm development is a fundamental problem for the field that microbiologists have been pondering for more than a decade. Most of the conclusions are supported by data, but findings are not communicated as effectively as they could be, which detracts from what otherwise is a very nice dataset. There are some controls that are missing as well.

We thank Reviewer #2 for constructive comments. We have modified both texts and figures to improve the manuscript. More experimental data sets including controls are also added.

Key constructive criticisms

The authors should complete a complementation analysis for pslG in all their assays. As the authors note, the pslG gene is in an operon and so polar effects are a concern, and because pslG mutations appear pleiotropic, ensuring that phenotypes do not result from second site mutations elsewhere in the chromosome of engineered strains seems especially important.

We thank the Reviewer for suggestions. We have added several datsets of complementation analysis for the pslG deletion mutants, which include PslG expressed from plasmid under PBAD promoter, PBAD-pslG inserted in chromosome attB site (ΔpslG attB::PBAD-pslG) and pslG knocked into the ΔpslG mutants at the original location of pslG (ΔpslG:: pslG). The results show that all phenotypes of ΔpslG can be complemented. These results have been added in Figures 1-6 and corresponding supplement figures in the revised manuscript.

To ensure that the hydrolase function of PslG is responsible for the observed phenotypes, it would be prudent to repeat some key assays in which the active site residues of PslG have been disrupted by site-directed mutagenesis.

Following reviewers’ suggestion, using strains in which the active sites of PslG are mutated (E65Q, E276Q), we have performed more tests on assays of PSL production, attachment, swimming motility, the formation of microcolony and long bacterial chains. The results indicate that the hydrolytic activity of PslG is critical for all these phenotypes (see Figures 1-7 and corresponding supplement figures).

Line 110-121. It is not clear what the authors are measuring and reporting in Figure 1A. In the text, the authors make statements about the quantities of ePSL produced by the different strains. However, the data in Figure 1 has no units and it is not clear whether the authors are measuring ePSL directly, or if the data represents something else. Could the authors quantify biofilm formation shown in the pictures below Figure 1A? Also, it would be highly beneficial if a semi-quantitation of the dot blots shown was provided too. Corresponding text should be added to the figure legend to succinctly explain the data presented in the figure.

Thanks for the suggestions. We apologize for confusing. What shown in Figure 1B are PSL production and attachment of tested strains. Results of PSL production shown in Figure 1B are semi-quantitation of the dot blots that are normalized to the level of PAO1. In the revised manuscript, we have modified the legend of Figure 1B (L774-786). We have also added the measurement results of corresponding biofilm biomass (OD560) in the revised Figure 1B.

Line 134-135. Could the authors better explain how the data led to their conclusion that PslG modulates surface attachment through multiple mechanisms? It is not clear from the text why the authors conclude that changes in motility are causally linked to defects in bacterial surface attachment, because the flagellum can be dispensable for biofilm formation depending on the assay. Also, the evidence here indicates that pslG mutations are pleiotropic.

The reviewer is right that the flagellum can be dispensable for biofilm formation depending on the assay. But based on our results and literature results, flagellum can also help the surface attachment of cells. At attachment stage, loss of flagella (such as ΔfliC mutant) can have a biofilm phenotype as that of a PSL-negative strain (Ma et al. 2006 J of Bacteriol). Given such considerations, since pslG deletion mutants show defected attachment, we think it is reasonably to check whether the swimming motility has been affected or not in these strains.

To make our point clearer, in the revised manuscript, we have modified the sentence to be: “These results show that pslG deletion does not affect the function of flagella and T4P directly, yet increasing PSL production in PBAD-pslΔpslG attenuated the swimming ability, which might impact its attachment phenotype. ”

Lines 228-234 and Figure 5 C. Caution needs to be applied using Raman spectroscopy on whole biofilms because the composition of biofilms is complex. How can the authors be certain that changes at 865 cm-1 correlate specifically with a glycosidic link in ePSL? There has been a demonstration that deletion of pslG affects intracellular c-di-GMP which influences many things in physiology including multiple components of the P. aeruginosa biofilm matrix. The authors would need to purify PSL from the WFPA801 and WFPA801 pslG strains and analyze that material to make these conclusions with confidence.

We thank Reviewer #2 for the comment. As in our reply to a similar comment of Reviewer #1, we agree with reviewers that to make a definitive conclusion from Raman data, more controls and more specific and supportive references are needed, which are beyond the scope of this study. Reviewer #2 also suggested to use purified PSL for the Raman examination. While we totally agree this is worth to try, it is not clear for us to what extent that such measurements using purified PSL could resemble those using fresh biofilms in which PSL is in its natural state, as the purification process may or may not change the properties of PSL. That is partially the reason why we examined the Raman spectra of biofilms directly in this work. In a future work, to better reveal the properties of PSL, measurements of PSL in both ways (i.e., either in biofilms or in purified state) may need to be performed.

Given such considerations as well as to be more focused on the role of PslG, in the revised manuscript, we have removed the Raman spectra data to leave them for a future work and added more data on the effect of PslG on PSL signaling (Figure 7 and corresponding description at L310-332).

Lines 212-227 and Figure 5A and B. Why isn't data for the pslG mutant also included in Figure 5A?

Thanks for the suggestion. In the revised manuscript, we have included the data of pslG deletion mutant and pslG with hydrolytic active site mutation (see Figure 6 in revision, referred to figure 5 in the previous version).

Lines 276-277. In this case, "data not shown" is not acceptable because the WFPA801 and WFPA801 pslG are engineered to have artificially high levels of PSL production. The data that are not shown correspond to key information corroborating physiological relevance to the wild type background. Please add these data to the manuscript.

Following the reviewer’s suggestion, we have added the data in the revised manuscript as Video 3.

Reviewer #3:

In this manuscript, the authors investigated the role of PslG, a Pseudomonas aeruginosa glycosyl hydrolase, in biofilm formation and cell physiology of a P. aeruginosa. PslG is encoded on the 15 gene psl operon, and a previous study indicated that deletion of pslG caused reduced ePSL (extracellular PSL polysaccharide) production. However, in the previous study, the pslG deletion may have been polar on other downstream psl genes and may have affected ePSL production. Therefore, in this study the investigators constructed a pslG deletion in a strain where the entire psl operon is controlled by the arabinose-inducible promoter. In this way, ePSL is produced but the operon lacks pslG. The investigators showed that in this strain, ePSL is produced at levels similar to the wild-type strain. However, the pslG deletion strain had several phenotypic defects that suggested a functional role of the glycosyl hydrolase. Among the defects was a reduction in biofilm formation, even when ePSL was produced, and a reduction in swimming motility. The investigators also used microscopic tracking of the cells, and noticed a reduction in surface area covered, likely due to clumping of the cells. The clumping of the cells was due to their formation of cell chains, which were not observed in the wild-type strain or in the new strain, when ePSL production was not induced with arabinose. Interestingly, the investigators noted that in the cells that were unable to separate following division (cell chains) had ePSL accumulated at the boarders between the cells. The chains could be released by adding exogenous PslG enzyme to the cultures. The investigators suggested that in the absence of PslG, the ePSL may accumulate at the cell boarders, holding the daughter cells together. In addition to these phenotypes, the investigators also noticed a difference in the production of the cyclic-di-GMP signaling molecule in the pslG mutant strain. Since c-di-GMP is an important signaling molecule for biofilm production, the investigators put forth a model for the role of PslG in c-di_GMP signaling in the development of P. aeruginosa biofilms.

Overall, I think this is a thorough study and the results are well presented.

We thank Reviewer #3 for his/her positive comments.

A weakness of this study (in my opinion) is that while adequately exploring the phenotypes of the PslG mutant in an ePSL producing strain, a functional role for PslG in polysaccharide degradation during ePSL biosynthesis is not adequately developed. The model in Figure 8 (and the abstract) implies that the role of PslG is through c-di-GMP signaling molecule. However, the data in Figure 4B and Figure 5B show only a minor effect of the pslG deletion on c-di-GMP signaling. In Figure 4B, most cells are classified as "none-bright" in both the experimental and control strain. Only a minor percentage of cells are classified as "two-bright". The difference in "two-bright" between the control and experimental in Figure 4B, although statistically different, is not a high percentage of the total cells. Figure 5A shows a moderate difference in c-di-GMP production as indicated by a reporter gene. However, Figure 5A does not include a growth curve, so the difference seen in fluorescence could be due to a difference in cell numbers/growth. This would not be surprising, given the many physiological defects of the pslG mutant. (a growth curve needs to be shown to determine if the fluorescence difference is due to cell growth differences). The model in Figure 8 implies that c-di-GMP signaling (and the "two-bright" daughter cells) is the primary signaling event for biofilm formation in the pslG mutant. In my opinion, this is overstated, given the minor effect of the pslG mutation on c-di-GMP production. As far as I know, PslG does not have a c-di-GMP binding domain, so it is difficult to draw a molecular link between the pslG mutant and c-di-GFP. The slight increase in c-di-GMP in the mutant cells may be due to an indirect effect of an overall stress response in the mutant cell, and not a direct signaling mechanism.

We thank the reviewer for his/her comments.

– We have measured the growth of strains, and the results show a similar growth of WT, ∆pslG, and ΔPpsl. This is also consistent with the result that ∆pslG has similar biomass as that of WT strain shown in Figure 2. The results of growth curves have been added in Figure 7A and C in the revised manuscript (Figure 5 in the original version has been modified and renumbered to be Figure 7 in the revised manuscript). Based on the growth measurement, we can conclude that the observed differences in c-di-GMP in Figure 7 is not due to the growth difference of strains. In addition, we have also examined the total protein of stains in WFPA801 (which has renamed as PBAD-psl in this revision) background, the results indicated that PBAD-psl, PBAD-pslΔpslG, and PBAD-pslΔpslG::pslGE165Q, E276Q have similar growth rate at the condition with or without arabinose.

– We agree with the reviewer on that “the increase in c-di-GMP in the mutant cells may be due to an indirect effect … and not a direct signaling mechanism.”, as the reviewer pointed out that “PslG does not have a c-di-GMP binding domain”. In fact, in Figure 8, we didn’t mean what the reviewer commented that “the role of PslG is through c-di-GMP signaling molecule.” nor mean that “c-di-GMP signaling (and the "two-bright" daughter cells) is the primary signaling event for biofilm formation in the pslG mutant.” What we would like to convey in Figure 8 is that PslG can affect the phenotype and c-di-GMP levels through PSL-mediated pathways, which then affect biofilm development.

– Psl has been shown to be able to act as a signaling substance to stimulate biofilm formation through affecting c-di-GMP (Irie et al., 2012). So by modifying Psl, PslG can indirectly affect c-di-GMP levels, which we believe is what has been happening in this study. There are multiple lines of evidence to support this. First, PslG can degrade PSL, which has been shown in literature. Second, in this study, the results of ∆pslG and pslG with hydrolytic active site mutation indicate that PSL without the cutting of PslG shows a stronger signaling property (see Figure 6 and Figure 7A and B). In addition, we also examined the signaling property of PSL with or without PslG treatment. As shown in Figure 7C and D, PslG treatment significantly reduces the signaling property of PSL.

– In the revised manuscript, to better convey our point, we have modified Figure 8 to be clearer on the role of PslG by modifying the text to be “shapes PSL localization” and “enhances PSL signaling function” in the figure.

– In this study, after we checked the c-di-GMP levels of two daughter cells after cell division, the results show that there is an increase in the proportion of the case of “two-bright”, although the value of two-bright case is still minor compared to that of none-bright case. But even the number of two-bright case is small, they can play an important role in the microcolony formation. Earlier work has shown that cells form a microcolony through a PSL-based rich-get-richer mechanism (Zhao et al., 2013), during which founder cells can be very important. The cells of two-bright case have high c-di-GMP levels and thus can act as founder cells to promote microcolony formation. Therefore, the slight change in c-di-GMP levels of daughter cells in ∆pslG can affect the microcolony formation.

I recommend that the authors read and cite the publication, "Role of an alginate lyase for alginate transport in mucoid Pseudomonas aeruginosa" Infect Immun. 2005 Oct;73(10):6429-36.doi: 10.1128/IAI.73.10.6429-6436.2005.

In the discussion, the authors state, "To the best of our knowledge, the ability of a

glycosyl hydrolase to affect bacterial phenotype and c-di-GMP levels has not been

reported until this work. It would be also very interesting to see in the future whether

other glycosyl hydrolases in different bacterial species can have similar functions."

We thank the Reviewer for providing the reference. In the revised manuscript, we have added this reference. We have modified the sentence as “To the best of our knowledge, the ability of a glycosyl hydrolase to affect c-di-GMP levels has not been reported until this work”.

In the paper above, the authors characterized another glycosyl hydrolase from Pseudomonas aeruginosa, that degrades alginate, and is encoded on the alginate biosynthetic operon (so, a very similar model system). In that study, the authors found that deletion of algL in an alginate producing strain resulted in cell death, due to accumulation of alginate in the periplasm. There is a precedence for these types of studies. The mechanism for PslG may be different from AlgL (and the PslG deletion obviously doesn't kill the bacteria), but the AlgL study could provide a hypothesis on the molecular role of PslG. In addition, both AlgL and PslG reside in the periplasm (assuming the model in reference: Frontiers in Microbiology 2:167 is correct). The authors of the current study only mention the periplasmic location for PslG in the Discussion section. The implication in the present study is that PslG is secreted. If PslG is secreted, there should be evidence for that (or a citation that shows that it is secreted). The localization of this enzyme could be crucial in understanding its function.

We thank the reviewer for his/her thoughtful comments. In the revised manuscript, we have added more discussions on algL. Please find at L405 – L414: “algL is a gene within the alginate synthesis operon of P. aeruginosa (Franklin et al., 2011). Interestingly, algL also encode a lyase to degrade alginate. An early work showed that ΔalgL in a CF isolate mucoid strain FRD1 resulted in cell death, due to accumulation of alginate in the periplasm (Jain et al. 2005). However, a later work showed that ΔalgL in a PAO1-derived mucoid strain PDO300 did not cause cell death, yet increase alginate production (Wang et al. 2016). pelA is the first gene in PEL synthesis operon, which also encodes a hydrolase (Franklin et al., 2011) that has been shown to inhibit biofilm formation as that of PslG (Baker et al., 2016). It would be interesting to investigate whether algL or pelA affects bacterial intracellular c-di-GMP levels”.

Regarding on the localization of PslG, in an earlier work of Wu et al. 2019 published in MicrobiologyOpen, results have shown that PslG is localized mainly in the inner membrane and some in the periplasm. PslG is only released into extracellular when bacteria are dead. Such information has been added in the introduction part in the revised manuscript (L78-80 and L88-93).

[Editors’ note: what follows is the authors’ response to the second round of review.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1:

The manuscript seeks to delineate the role of pslG, a glycosyl hydrolase encoded within an operon that enables the production of the Psl polysaccharide of the bacterium Pseudomonas aeruginosa. The results in the manuscript show numerous effects using a combination of pslG mutants and over-expression strains upon P. aeruginosa behavior important to colonization and biofilm development. This adds to our understanding of the multi-functional enzymes involved in regulating polysaccharide production and the overall behavior of P. aeruginosa.

The manuscript primarily presents phenotypic results to shape their story and conclusions. Given the importance of both Psl and Pel to exopolysaccharide-dependent behaviors in P. aeruginosa PAO1 (and many other strains), the potential importance and contribution of Pel remains essentially ignored in this manuscript. Because both Pel and Psl can generally influence several of the phenotypes explored in this manuscript, the authors' desire to draw specific conclusions from the effect of PslG cannot be separated from potential contributions of Pel polysaccharide in these experiments.

This comment is the same as the one in the previous round. As we stated in our replies in the previous round “Based on literature results (for example, Colvin et al, 2012, Environmental Microbiology), PAO1 is a PSL-dominant strain, whose biofilm formation is mainly dependent on PSL. Thus, we speculate that PEL may have little contribution to the phenotypes of pslG mutants observed in this work.” More importantly, we constructed and tested PEL negative strains in either PAO1 or PslG mutant background, and found that all phenotypes of pslG mutants are still observed in PEL-negative strains. These results have already been shown in Figure 1-Figure supplement 1C, Figure 3-Figure supplement 3, and Figure 5-Figure supplement 3 in the previous version of manuscript. Taken together, all these results suggest that Pel polysaccharide has little contribution on the phenotypes observed in this study (the corresponding text can be found at L277-282 in the previous revision, L280-288 in this revision). Therefore, we could not agree with the comment about “the potential importance and contribution of Pel remains essentially ignored in this manuscript.”.

The story is certainly complicated as study of pslG mutants does affect levels of Psl polysaccharide.

Our very first experiment shown in Figure 1A is to tell that the level of PSL polysaccharide is not likely the reason to cause those phenotypes of pslG mutants. In the revised manuscript, to make this point clearer, we have revised the subtitle as “ΔpslG strains cannot form rings on microtiter dish wells even when PSL production is induced to the wild type level.” (L116-117).

The most compelling data in the manuscript involves the impact of pslG upon divided cells. Truly fascinating.

We thank the reviewer for the positive comment!

However, it is not clear that the authors demonstrate the importance of PslG to "normal" cell division and/or how Psl polysaccharide might contribute to cell division in the wild-type. There is no basis in the literature to suggest that pslG mutations would be common in select environments.

We think PSL polysaccharide might not interfere with “normal” cell division in wild type. It may affect cell division only in some circumstances where PSL could not be hydrolyzed by PslG and hence cannot be released properly, such as in pslG deletion background.

The reviewer is right that there has been no literature so far (to the best of our knowledge) to suggest that pslG mutations would be common in select environments. The E165Q, E276Q mutation of PslG is a mutation that causes PslG to loss its hydrolase activity. We used this mutant to further confirm the importance of the cutting of PslG during PSL polysaccharide synthesis.

Several new lines of experimental evidence have been added to the manuscript in this revised form and these changes are a great collective improvement-but the true role(s) and effect of PslG are still not clear from the evidence presented.

We thank the reviewer for the comment. Here it is not clear for us what the reviewer means by saying “true role(s)”.

In this study, we showed that lacking of PslG or its hydrolytic activity in PAO1 enhances the signaling function of PSL, changes the relative level of cyclic-di-GMP within daughter cells during cell division and shapes the localization of PSL on bacterial periphery, thus results in long chains of bacterial cells, fast-forming biofilm microcolonies. We believe these results have revealed the important roles and effects of PslG on the cell morphology and cell activities such as biofilm development. However, to reveal the exact mechanisms at the molecular level, more studies are needed, and we hope our work can inspire more future studies in this direction.

There are several sentences where the English syntax and grammar require editing.

We thank the reviewer for the comment. We have checked through the manuscript and corrected all grammatical errors that we found.

Line 47: add “an”;

Line 47: “protect” change to “protects”;

Line 147: “increasing” change to “inducing”;

Line 180: “hierarchical” change to “non-uniform”;

Line 188: “more microcolonies of ΔpslG are formed in the field of view” change to “ΔpslG formed more microcolonies in the field of view”;

Line 192: add “namely”;

Line 194: delete “more”;

Line 219: “affects” change to “affect”;

Line 233-Line 234: “with arabinose induction” change to “when PSL production was induced with arabinose”; Line 295: delete “a”;

Line 296: delete “a”;

Line 295: “level” change to “levels”;

Line 521: add “and”;

Line 571: “abacterial” change to “a bacterial”;

Line 572: “. Otherwise not” change to “, otherwise not”;

Line 159. There does not appear to be fewer bacterial cells in the pslG biofilms-and no quantification is given to support this statement.

From Figure 2D, it can clearly be seen that there are less green fluorescence signals in pslG mutant biofilms than in PAO1. As the fluorescence signal is from each bacterial cell, less fluorescence signal in pslG mutant biofilms corresponds to fewer bacterial cells in it. As this is qualitatively clear, we don’t think it is necessary to do the quantitative measurement for this particular statement. In addition, for quantitative measurements, we have shown the total biomass and thickness in Figure 2B. The results show that pslG mutant has a higher maximum thickness, yet a similar total biomass compared with PAO1. Thus, we can deduce that on average pslG mutant would have less number of bacterial cells than PAO1 for each section of biofilms (such as shown in Figure 2D), which is consistent with the qualitative observation of Figure 2D.

Line 178. "…a more hierarchical bacterial visit distribution for pslG…" It is not clear how this shows hierarchical visitors or a broad range of visit frequencies. How is this description evident based upon what is shown in Figure 3B?

We thank the reviewer for the comment. In Figure 3B, the number of bacterial visits is color-coded, with cyan color standing for 1000 bacterial visits and black color standing for 0 bacterial visit. Comparing the bacterial visit distribution between PAO1 and DpslG mutant, we can see that the bacterial visit distribution is more non-uniform in DpslG than in PAO1; Under the same total bacterial visits (N~ 100000), DpslG displays several concentrated patches with green-yellowish to cyan-like which stand for about 200~1000 bacterial visits and large blackish regions which have near zero bacterial visits, while PAO1 displays a large portion of surface with red-like colors which stand for ~ 100 bacterial visits. Quantitatively, the maximum of bacterial visit number for DpslG is higher than that in PAO1, and the power law exponent of bacterial visit distribution is also less negative for DpslG than for PAO1, so DpslG has a more hierarchical bacterial visit distribution. The same method has been used in earlier work (Kun, Zhao, Boo, Shan, Tseng, & Bernard, et al. (2013). Psl trails guide exploration and microcolony formation in Pseudomonas aeruginosa biofilms. Nature, 497(7449), 388-391).

In the revised manuscript, we have modified the text to be: “…a more non-uniform bacterial visit distribution for ΔpslG (L180)”. In the legend of Figure 3B, we also added the maximum bacterial visit number for each bacterial visit distribution map.

Line 184. There appear to be 27-28 cells in the PAO1 cluster detailed in Figure 3D.

We thank the reviewer for the comment. In this study, microcolonies are defined as clusters of more than 30 cells. For clusters, we used a minimum distance criterion to judge whether a cell belonged to a cluster or not. If the minimum distance between any point of the scrutinized cell body and any point of any cell body of the cluster, is smaller than 0.5µm (i.e., about one width of a bacterial cell), then the scrutinized cell is considered to belong to the cluster, otherwise not. Based on these definitions, the PAO1 cluster shown in Figure 3D has 30 cells and thus is a microcolony.

To illustrate these cells clearly, in Author response image 1, the cells are colored manually in a way that neighboring cells don’t have the same color so that we can easily count how many cells that have been involved in this cluster.

Author response image 1
Download asset Open asset

Line 188. "…microcolonies of ΔpslG are more compact". This representation of the data in Figure 3F does not necessarily support this statement. Is this a projection view showing a 2D image of the thicker 3D biofilms as described in Figure 2? How was this quantified?

We thank the reviewer for the comment. The images shown in Figure 3F were taken after 10 hours’ growth in a flow cell, while those shown in Figure 2 are confocal laser scanning microscope images of pellicles grown on air-liquid interface. So they are different. Regarding on the images shown in Figure 3F, they are representative images taken by a regular fluorescence microscope to give a general view of microcolonies formed by PAO1 and ΔpslG. They can be considered as projection views of a slab sample (the thickness of the slab is closely related to the depth of field of the equipment, which is about 0.54 micron under our set-up). We do agree with the reviewer that this statement may not appropriate. In the revised manuscript, we have modified the sentence to be “Tracking GFP-tagged bacteria in a flow cell also showed that the PAO1 biofilms covered more surface (Video 1 and Figure 3F) and ΔpslG cells tended to form microcolonies with strong fluorescence intensity (Video 2 and Figure 3F), suggesting that bacteria were packed within microcolonies. This is consistent with the results visit distribution map shown in Figure 3B.” (L208-L212). We also included representative videos (Video 1 and 2) tracking the microcolony formation of PAO1 and ΔpslG in this revision.

Line 221. How were these twitching speed experiments performed? These are not described in the supplemental Figure legend or the methods.

We thank the reviewer for the comment. The twitching speed of each tracked cell at frame n was calculated by the displacement of the cell between nth and (n+1) th frames divided by the corresponding time interval. In the methods, we described how the experimental data were collected (lines 499-508) and briefly how the single cell tracking image analysis was performed (lines 544-549). As such imaging analysis methods have been described in detail in earlier studies (Zhang et al., 2018, and associated references therein), we did not describe it in details in the original version of the manuscript.

In the revised manuscript, we have added the above description on the measurement of twitching speed: “Specifically, the twitching speed of each tracked cell at frame n was calculated by the displacement of the cell between nth and (n+1)th frames divided by the corresponding time interval” (Lines 550-552).

Line 240. "…were observed most frequently in either ΔpslG or PBAD-pslΔpslG strain…" The purpose of this statement is not clear. These are the only strains shown in Figure 4. Is this a comparison?

We apologize for not expressing our point clearly. Here the purpose of this statement is to characterize the number of cells in a chain (i.e. the chain length in terms of the number of cells). Since WT cells don’t show long chains, we just show the distribution of cell chains in two pslG mutants. But we didn’t mean to compare between different strains. From the distribution of cell chains, we can see that the cell chains consisting of four cells are observed most frequently among all observed chains. This is true in both ΔpslG and PBAD-pslΔpslG strains.

In the revised manuscript, to make our point clearer, we have modified the text to be “The length of chains varied. Among all the observed chains, the chains consisting of 4 cells were observed most frequently, which is true both in ΔpslG and in PBAD-pslΔpslG strains (Figure 4C) (L242-244).”

Line 241. What is the support for the statement "…yet some can also be broken by bending of chains"? This is from separate experimental evidence? Or just an observation?

This can be seen clearly from the Video 3. In the revised manuscript, we have cited the corresponding video in the text (L246).

Line 246. This Figure 5 data is the most intriguing of the manuscript.

We thank the Reviewer for the positive comment!

Line 309. "…which might enhance their stay on the surface, reduce their surface motility, and promote the microcolony formation". Given the small variation in the data shown in Figure 6B, support for this statement really requires specific testing. Further, not all previous data in the manuscript support it. The text in an earlier section (starting line 213) effectively concludes that there is no quantifiable motility difference between these strains.

We thank the reviewer for the insightful comment. Following that, we have modified the text to make it be more appropriate given the results that we obtained in this study. In the revised manuscript, the sentence is now to be: “During asymmetric divisions, daughter cells with high c-di-GMP levels keep staying on the surface and daughter cells with low c-di-GMP levels tend to move away (Christen et al., 2010; Laventie et al., 2019). Therefore, both daughter cells with high c-di-GMP levels might enhance bacterial stay on the surface and alter their movement pattern. In addition, a high level of c-di-GMP enhances PSL production. An earlier work has shown that cells form a microcolony through a PSL-based rich-get-richer mechanism (Zhao et al., 2013), during which founder cells can be very important. The cells of two-bright cases have high c-di-GMP levels and thus can act as founder cells to promote microcolony formation. Altogether, the slight change in c-di-GMP levels of daughter cells in pslG mutants can be likely one of reasons to promote the formation of microcolony and long bacterial chains.” (L312-323).

Line 497. The Psl staining with TRITC-HHA is specific to Psl? Is there binding to Pel or other polysaccharides?

Yes, the Psl staining with TRITC-HHA is specific to Psl. No binding to Pel or alginate. This can be found in the papers of Ma et al. PLoS Pathogens 2009 and Ma et al. FEMS Immunolology & Medical Microbiol. 2012.

Line 770. The Figure 1 title is not the best description of what is shown here. No "increasing" levels of Psl are shown. There are two. The data show adding of Psl into a pslG background.

Thanks for the comment. We have revised the text as “Inducing PSL production in ΔpslG background cannot recover its defects on bacterial initial attachment and yet affects swimming motility.”

Line 821. It is not stated anywhere in the legend or methods from how many frames the data analyzed for Figure 3E were obtained. There are error bars, but it is not clear what these represent.

We thank the reviewer for pointing this out. In the revised manuscript, we have added the corresponding information in the legend of Figure 3E: “The number of microcolonies in the field of view formed by PAO1, ΔpslG, ΔpslG::pslG and ΔpslG::pslGE165Q,E276Q at 10 hrs after inoculation in a flow cell. The number (N) of frames analyzed are 14, 43, 52, 51 for PAO1, ΔpslG, ΔpslG::pslG and ΔpslG::pslGE165Q,E276Q, respectively. Error bars represent standard deviation of the mean. Statistical significances are measured using one-way ANOVE. n.s., not significant; *p < 0.05; **p < 0.001; ***p < 0.0001” (L848-851).

Line 831. The criteria for "microcolony" is not entirely clear. Some of these cells clearly have different spacing but are outlined with dashed lines.

We thank the reviewer for the comment. The definitions of microcolony (lines 185-186) and cluster (lines 568-572) have been stated in the text. The microcolonies outlined with dashed lines in Figure 3 are all determined based on these definitions. Particularly, in this study whether or not a cell belongs to a cluster is judged by a minimum distance criterion. We note that this criterion does not mean that cells need to be most densely packed inside a cluster. For example, for the outlined microcolony of PBAD-psl in the Figure 3—figure supplement 1A, each of cells along the dashed outline has at least one neighboring cell that is within the distance of 0.5µm of the cell (i.e. meets the requirement of minimum distance criterion) and belongs to the cluster. But there is a clear empty (free of cells) region enclosed by cells inside the microcolony. How dense a microcolony can be will depend on multiple factors including cell motility, cell growth and the incubation time etc.

Reviewer #2:

The manuscript by Zhang and colleagues carefully describes the phenotypes of pslG mutants of P. aeruginosa. PslG encodes a glyocoside hydrolase. While the biochemistry of this enzyme has been understood for many years, its physiologically relevant role in P. aeruginosa biofilm formation has remained ill-defined.

The experiments in the manuscript have been meticulously executed. There are controls and complementation analyses that provide confidence in the results obtained. The technical proficiency with microscopy is commendable.

We thank reviewer #2 for the positive comments.

However, while the authors provide a data-rich manuscript, an understanding of the consequences of PslG expression appears lacking beyond phenotyping.

We thank reviewer #2 for the comment. Our manuscript reveals at least one mechanism beyond phenotyping, which is the hydrolysis of PslG on PSL during biosynthesis can change the signaling function of PSL and shape the localization of PSL on bacterial periphery, leading to a higher portion of bacterial cells that are not able to go asymmetric divisions, which may become the founder cells for microcolonies formation, and finally resulting in long bacterial chains as well as the change of bacterial surface explore patterns. We have added several sentences to describe this point (See L321-323). We have also revised the last paragraph of result section (L312-323).

Perhaps this criticism is most pertinent to the observed changes in the power-law distribution for bacteria during the earliest stages of biofilm formation for pslG mutants. Interpretation/experimentation is absent that connects these observations to social biology. Such connections, which are front and center in prior work published by one of the co-authors (Dr. Khun Zhao), could help to explain conservation of glycoside hydrolases among synthase-dependent exopolysaccharide secretion systems like the Psl synthase. For example, beyond the careful phenotyping presented in this paper, co-culture of Pbad-Psl with Pbad-Psl-PslG strains, or perhaps wild type and pslG strains, that have been uniquely labelled with fluorescent proteins and tracked using Dr. Kun Zhao's elegant single-cell methods could directly demonstrate fitness changes for pslG mutants in surface exploration or colonization relative to wild type. In principle, wouldn't such a fitness cost provide an explanation for PslG function that is rooted in social evolutionary theory? Perhaps there are some trade-offs that aren't yet apparent. The link to c-di-GMP signaling provides some molecular insight even the sensory perception and signal transduction pathway is not yet fully known. Such analyses could take the work assembled here to the next level with little additional experimental effort, and as such, strikes me as a missed opportunity to provide significant, additional understanding of some really nice data.

We thank Reviewer #2 for the suggestions. It is certainly a great idea, which might lead to a new story. Our manuscript is already data-rich as reviewers stated above, thus adding more data may just dilute the main information.

Following the suggestion, we have tried once the co-culture experiment of PAO1 and pslG mutant. Some preliminary results are shown in Author response image 2. PAO1 are GFP-tagged while ΔpslG is not, so that they can be differentiated from each other.

Author response image 2
Download asset Open asset
(A): The biofilm microcolonies formation by PAO1 with ΔpslG in flow cell.

PAO1 are GFP-tagged while ΔpslG is not. (B): Surface coverage maps at a total of 10000, 50000 and 100000 bacterial visits for PAO1 with ΔpslG, PAO1 in co-culture and ΔpslG in co-culture. (C): The visit frequency distributions of PAO1 with ΔpslG, PAO1 in co-culture and ΔpslG in co-culture.

However, due to the phototoxicity caused by fluorescence illumination, recording of cell behavior under fluorescence illumination cannot be performed at a frame rate that is fast enough to meet the requirement of cell tracking (currently, the fluorescence recording is at a frame rate of 1 frame per 15minutes). We have also tried to record the cell behavior through both bright-field channel and fluorescence channel with a recording frame rate of 1 frame per 3 second for bright-field channel and 1 frame per 15 minutes for fluorescence channel. However, using this method, we cannot distinguish which type of cell is for those cells whose appearance and disappearance both happen during a bright-field recording period between two consecutive fluorescence recording time points. So, new methods for distinguishing two strains and/or new analysis techniques are needed, which are beyond the scope of this manuscript.

In addition, to get conclusive results of co-culture experiments, more control experiments are also needed including the single-strain experiments under the same fluorescence illumination conditions. Considering our manuscript is already data-rich as reviewers stated above, adding more data may just dilute the main point of this manuscript.

Given these considerations, we believe the study on co-culture experiments deserves a complete and separate treatment, and it is more appropriate to leave it for future work.

https://doi.org/10.7554/eLife.72778.sa2

Article and author information

Author details

  1. Jingchao Zhang

    Frontiers Science Center for Synthetic Biology and Key Laboratory of Systems Bioengineering (Ministry of Education), School of Chemical Engineering and Technology, Tianjin University, Tianjin, China
    Contribution
    Investigation, Writing – original draft, Writing - review and editing
    Contributed equally with
    Huijun Wu and Di Wang
    Competing interests
    No competing interests declared

  2. Huijun Wu

    1. State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China
    2. University of Chinese Academy of Sciences, Beijing, China
    Present address
    State Key Laboratory of Petroleum Pollution Control, China National Petroleum Corporation Research Institute of Safety and Environment Technology, Beijing, China
    Contribution
    Investigation, Writing – original draft
    Contributed equally with
    Jingchao Zhang and Di Wang
    Competing interests
    No competing interests declared

  3. Di Wang

    State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China
    Contribution
    Investigation, Project administration, Writing – original draft, Writing - review and editing
    Contributed equally with
    Jingchao Zhang and Huijun Wu
    Competing interests
    No competing interests declared

  4. Lanxin Wang

    1. State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China
    2. University of Chinese Academy of Sciences, Beijing, China
    Contribution
    Investigation
    Competing interests
    No competing interests declared

  5. Yifan Cui

    1. State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China
    2. University of Chinese Academy of Sciences, Beijing, China
    Contribution
    Investigation
    Competing interests
    No competing interests declared

  6. Chenxi Zhang

    Frontiers Science Center for Synthetic Biology and Key Laboratory of Systems Bioengineering (Ministry of Education), School of Chemical Engineering and Technology, Tianjin University, Tianjin, China
    Contribution
    Investigation
    Competing interests
    No competing interests declared

  7. Kun Zhao

    Frontiers Science Center for Synthetic Biology and Key Laboratory of Systems Bioengineering (Ministry of Education), School of Chemical Engineering and Technology, Tianjin University, Tianjin, China
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Supervision, Validation, Writing – original draft, Writing - review and editing
    For correspondence
    kunzhao@tju.edu.cn
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3928-1981

  8. Luyan Ma

    1. State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China
    2. University of Chinese Academy of Sciences, Beijing, China
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing – original draft, Writing - review and editing
    For correspondence
    luyanma27@im.ac.cn
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3837-6682

Funding

National Key Research and Development Program of China (2018YFA0902102)

  • Kun Zhao

National Natural Science Foundation of China (91951204)

  • Luyan Ma

National Natural Science Foundation of China (32070033)

  • Di Wang

National Natural Science Foundation of China (21621004)

  • Luyan Ma

National Key Research and Development Program of China (2019YFA0905501)

  • Di Wang

National Key Research and Development Program of China (2019YFC804104)

  • Luyan Ma

National Key Research and Development Program of China (2021YFA0909500)

  • Luyan Ma

PetroChina Company Limited

  • Huijun Wu

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work is supported by the National Key R&D Program of China (2018YFA0902102, 2021YFA0909500, 2019YFC804104, and 2019YFA0905501), the National Natural Science Foundation of China (91951204, 21621004, 32070033), and Research on basic science and technology of the strategic reserve fund projects of PetroChina Company Limited (No. 2020D-5008). The funders had no role in the study design, data collection and interpretation, or the decision to submit the work for publication.

Senior Editor

  1. Wendy S Garrett, Harvard T.H. Chan School of Public Health, United States

Reviewing Editor

  1. Gerald Pier, Brigham and Women's Hospital, United States

Reviewer

  1. Joe Jonathan Harrison, University of Calgary, Canada

Version history

  1. Received: August 4, 2021
  2. Preprint posted: August 18, 2021 (view preprint)
  3. Accepted: April 17, 2022
  4. Accepted Manuscript published: April 19, 2022 (version 1)
  5. Version of Record published: May 6, 2022 (version 2)

Copyright

© 2022, Zhang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 816
    Page views
  • 204
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Jingchao Zhang
  2. Huijun Wu
  3. Di Wang
  4. Lanxin Wang
  5. Yifan Cui
  6. Chenxi Zhang
  7. Kun Zhao
  8. Luyan Ma
(2022)
Intracellular glycosyl hydrolase PslG shapes bacterial cell fate, signaling, and the biofilm development of Pseudomonas aeruginosa
eLife 11:e72778.
https://doi.org/10.7554/eLife.72778

Categories and tags

Further reading

    1. Cell Biology
    2. Microbiology and Infectious Disease

    A cell wall synthase accelerates plasma membrane partitioning in mycobacteria

    Takehiro Kado, Zarina Akbary ... M Sloan Siegrist
    Research Article Updated

    Lateral partitioning of proteins and lipids shapes membrane function. In model membranes, partitioning can be influenced both by bilayer-intrinsic factors like molecular composition and by bilayer-extrinsic factors such as interactions with other membranes and solid supports. While cellular membranes can departition in response to bilayer-intrinsic or -extrinsic disruptions, the mechanisms by which they partition de novo are largely unknown. The plasma membrane of Mycobacterium smegmatis spatially and biochemically departitions in response to the fluidizing agent benzyl alcohol, then repartitions upon fluidizer washout. By screening for mutants that are sensitive to benzyl alcohol, we show that the bifunctional cell wall synthase PonA2 promotes membrane partitioning and cell growth during recovery from benzyl alcohol exposure. PonA2’s role in membrane repartitioning and regrowth depends solely on its conserved transglycosylase domain. Active cell wall polymerization promotes de novo membrane partitioning and the completed cell wall polymer helps to maintain membrane partitioning. Our work highlights the complexity of membrane–cell wall interactions and establishes a facile model system for departitioning and repartitioning cellular membranes.

    1. Cell Biology
    2. Microbiology and Infectious Disease

    The chemorepellent, SLIT2, bolsters innate immunity against Staphylococcus aureus

    Vikrant K Bhosle, Chunxiang Sun ... Lisa A Robinson
    Research Article

    Neutrophils are essential for host defense against Staphylococcus aureus (S. aureus). The neuro-repellent, SLIT2, potently inhibits neutrophil chemotaxis, and might, therefore, be expected to impair antibacterial responses. We report here that, unexpectedly, neutrophils exposed to the N-terminal SLIT2 (N-SLIT2) fragment kill extracellular S. aureus more efficiently. N-SLIT2 amplifies reactive oxygen species production in response to the bacteria by activating p38 mitogen-activated protein kinase that in turn phosphorylates NCF1, an essential subunit of the NADPH oxidase complex. N-SLIT2 also enhances the exocytosis of neutrophil secondary granules. In a murine model of S. aureus skin and soft tissue infection (SSTI), local SLIT2 levels fall initially but increase subsequently, peaking at 3 days after infection. Of note, the neutralization of endogenous SLIT2 worsens SSTI. Temporal fluctuations in local SLIT2 levels may promote neutrophil recruitment and retention at the infection site and hasten bacterial clearance by augmenting neutrophil oxidative burst and degranulation. Collectively, these actions of SLIT2 coordinate innate immune responses to limit susceptibility to S. aureus.

    1. Biochemistry and Chemical Biology
    2. Microbiology and Infectious Disease

    A tRNA modification in Mycobacterium tuberculosis facilitates optimal intracellular growth

    Francesca G Tomasi, Satoshi Kimura ... Matthew K Waldor
    Research Article

    Diverse chemical modifications fine-tune the function and metabolism of tRNA. Although tRNA modification is universal in all kingdoms of life, profiles of modifications, their functions, and physiological roles have not been elucidated in most organisms including the human pathogen, Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis. To identify physiologically important modifications, we surveyed the tRNA of Mtb, using tRNA sequencing (tRNA-seq) and genome-mining. Homology searches identified 23 candidate tRNA modifying enzymes that are predicted to create 16 tRNA modifications across all tRNA species. Reverse transcription-derived error signatures in tRNA-seq predicted the sites and presence of nine modifications. Several chemical treatments prior to tRNA-seq expanded the number of predictable modifications. Deletion of Mtb genes encoding two modifying enzymes, TruB and MnmA, eliminated their respective tRNA modifications, validating the presence of modified sites in tRNA species. Furthermore, the absence of mnmA attenuated Mtb growth in macrophages, suggesting that MnmA-dependent tRNA uridine sulfation contributes to Mtb intracellular growth. Our results lay the foundation for unveiling the roles of tRNA modifications in Mtb pathogenesis and developing new therapeutics against tuberculosis.