The advance of assembly of exopolysaccharide Psl biosynthesis machinery in Pseudomonas aeruginosa

Abstract Biofilms are microbial communities embedded in extracellular matrix. Exopolysaccharide Psl (ePsl) is a key biofilm matrix component that initiates attachment, maintains biofilms architecture, and protects bacteria within biofilms of Pseudomonas aeruginosa, an opportunistic pathogen. There are at least 12 Psl proteins involved in the biosynthesis of this exopolysaccharide. However, it remains unclear about the function of each Psl protein and how these proteins work together during the biosynthesis of ePsl. PslG has been characterized as a degrader of ePsl in extracellular or periplasm and PslD is predicted to be a transporter. In this study, we found that PslG and its glycoside hydrolytic activity were also involved in the biosynthesis of ePsl. PslG localized mainly in the inner membrane and some in the periplasm. The inner membrane association of PslG was critical for the biosynthesis of ePsl. The expression of PslA, PslD, and PslE helped PslG remain in the inner membrane. The bacterial two‐hybrid results suggested that PslE could interacted with either PslA, PslD, or PslG. The strongest interaction was found between PslE and PslD. Consistently, PslD was disabled to localize on the outer membrane in the ΔpslE strain, suggesting that the PslE‐PslD interaction affected the localization of PslD. Our results shed light on the assembly of ePsl biosynthesis machinery and suggested that the membrane‐associated PslG was a part of ePsl biosynthesis proteins complex.

widely accepted, the exact mechanism underlying their biosynthesis remains poorly understood. A better understanding of the molecular mechanisms of polysaccharide biosynthesis may provide strategies for the control of chronic infections and problems related to biofilm formation.
The ePsl is a neutral pentasaccharide repeat containing ᴅ-mannose, ᴅ-glucose, and ʟ-rhamnose (Byrd et al., 2009). The polysaccharide synthesis locus (psl) contains 15 genes, 11 of which (pslACDEFGHIJKL) are required for ePsl biosynthesis (Byrd et al., 2009). However, the function of each Psl protein remains largely unknown. It has been reported that PslB is a bifunctional enzyme and is involved in sugar-nucleotide precursor production for ePsl biosynthesis (Byrd et al., 2009;Lee, Chang, Venkatesan, & Peng, 2008). PslD is a secreted protein and may play a role in exopolysaccharide export (Campisano, Schroeder, Schemionek, Overhage, & Rehm, 2006). Our previous study (Yu et al., 2015) has demonstrated that PslG is an endoglycosidase mainly targeted ePsl and, the catalytic residues E165 and E276 are critical for the hydrolytic activity.
PslG can degrade ePsl to prevent biofilm formation and disassemble existing biofilm when supplied exogenously. While whether PslG is involved in the biosynthesis of ePsl remains controversial. Byrd et al. (2009) considered PslG was required for the biosynthesis of ePsl. On the contrary, Baker et al. (2015) found that neither PslG nor its enzymatic activity appeared to be required for ePsl biosynthesis and biofilm formation. Strain PAO1ΔpslG constructed by Byrd et al. (2009) has deleted a cis-acting element located in the 3' of pslG that altered the translation of pslH (Baker et al., 2015), while, the ΔpslG strain constructed by Baker et al. (2015) is in the background of a psl overexpression strain PAO1ΔpelFP BAD psl rather than wild type PAO1.
Bioinformatic analyses suggest that ePsl biosynthesis mechanism resembles the biosynthesis of Escherichia coli group 1 capsular polysaccharides, with PslA, PslD, and PslE similar to WbaP, Wza, and Wzc, respectively (Franklin et al., 2011). It is proposed that biosynthesis and translocation of ePsl is temporally and spatially coupled by multiprotein complex. Nevertheless, there has not been any investigation about the interaction and localization of Psl proteins that involved in the ePsl biosynthesis.
In this study, we further investigate the role of PslG and its hydrolytic activity on the biosynthesis of ePsl in P. aeruginosa PAO1.
Interactions among Psl proteins (PslA, PslD, PslG, and PslE) and their effects on the subcellular localization of Psl proteins have been examined. Our results shed light on the assembly of ePsl biosynthesis machinery.

| Bacterial strains and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 1. Unless indicated, E. coli strains were grown at 37°C in Luria Bertani Broth (LB, Becton Dickinson), P. aeruginosa stains at 37°C in LB without sodium chloride (LBNS) or Jensen's, a chemically defined medium (Jensen, Fecycz, & Campbell, 1980). L-arabinose (Sigma) was used as inducer for genes transcribed from P BAD promoter in P.

| Microtiter dish biofilm assay
In the biofilm attachment assay, 1/100 dilution of a saturated (overnight) culture in Jensen's media for P. aeruginosa was inoculated into glass tubes. When the OD 600 reached 0.5, the 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 measured (OD 560 ) as described previously (Ma et al., 2006;O'Toole, 2011).

| ePsl immuno-dot blotting and cell extract western blotting analysis
P. aeruginosa cell surface associated polysaccharide extracts were obtained from culture that equivalents approximately 4 OD 600 , and examined by immunoblotting using anti-ePsl antiserum as previously described (Byrd et al., 2009). To induce the transcription of the pslG in the recombinant plasmid, arabinose was added to Jensen's media.
The immunoblotting data were analyzed using Image Lab software. shanghai China). The software Image Lab was used to analyze the immune-blotting data.
Then 20 ml of ddH 2 O was gently added. The sample was placed on ice for 20 min, and then centrifuged at 45,000 rpm for 45 min at 4°C. The supernatant fraction was collected as periplasmic sample. The pellet was resuspended in 50 ml buffer II (10 mM Tris-HCl pH 7.5, 5 mM EDTA, 1 mM DTT, 10 μg/ml DNase I), and then applied to sonication.

| Protein expression and purification
PslG 31-442 was expressed and purified as previously described (Yu et al., 2015). The first 30 residues of PslG were truncated because they were predicted to be a signal peptide by the Signal P4.1 server. Briefly,

| Bacterial two-hybrid system
Bacterial two-hybrid experiments were conducted as described (Zhang et al., 2009 Kan and used as a positive control. Positives were verified by using the aadA gene, which confers streptomycin resistance, as a second reporter. Cells harboring weaker interactors grew more slowly, requiring longer incubation time for colony development.

| Statistical analyses
All the experiments were performed in at least three triplicates. The results are presented as the mean ± SD. Student's t-tests were used to evaluate significance.

| PslG and its glycoside hydrolytic activity are involved in the biosynthesis of ePsl in P. aeruginosa
Our previous data indicated that overproduced PslG in wild type strain PAO1 reduced the production of ePsl and biofilm biomass, yet overproduced catalytically inactive PslG E165Q + E276Q did not affect the ePsl production and slightly increased biofilm biomass (Yu et al., 2015).
These results suggested that PslG might be involved in the biosynthe- Flagellum and type IV pili (T4P) also influence the initial attachment of P. aeruginosa (Klausen et al., 2003;O'Toole & Kolter, 1998). Therefore, we evaluated the flagellum-mediated swimming motility and the T4Pmediated twitching motility, the ΔpslG2 mutant showed similar levels of swimming and twitching motilities as wild type strain PAO1 (Appendix Figure A1), indicating the normal function of flagellum and T4P in ΔpslG2. The biofilm biomass of ΔpslG2 was slightly higher than WFPA800 in a 2-hr biofilm assay (Appendix Figure A2), indicating the ePsl synthesized from ΔpslG2 is functional. These results further suggest that PslG is involved in ePsl biosynthesis.
We then further investigated whether the glycoside hydrolytic activity of PslG is important for ePsl production. We constructed a chromosomal site-mutation strain ΔpslG2::pslG E165Q + E276Q with E165Q and E276Q mutation within PslG. This pslG mutant strain showed little ePsl production as that of ΔpslG2 mutant (Figure 1a).
Although the attachment ability of ΔpslG2::pslG E165Q + E276Q was higher than ΔpslG2, it was still significantly less than that of PAO1 (fourfold lower than PAO1, Figure 1b). The ePsl production of ΔpslG2 could be restored by a baseline level expression of PslG (grown without inducer arabinose) from the plasmid pG (PslG was cloned in F I G U R E 1 The contribution of PslG and its glycoside hydrolytic activity on the production of ePsl and initial attachment of P. aeruginosa. (a) The relative ePsl production of PAO1, ePsl-negative strain WFPA800, the pslG in-frame deletion mutant ΔpslG2, ΔpslG2::pslG, and the PslG catalytic residues mutant ΔpslG2::pslG E165Q + E276Q . The amount of ePsl is normalized to the level of PAO1. The corresponding anti-ePsl immune-dot blot is shown under each bar. (b) Shown is the corresponding initial attachment of the five strains. Values are means from two independent experiments, each with three replicates. The image under each bar is a representative microtiter dish well from corresponding crystal violet biofilm assay. (C) The ePsl production of ∆pslG2 that complemented by plasmid expressing wild type PslG (pG) or PslG E165Q + E276Q (pGDM). The amount of ePsl is normalized to the level of PAO1/pHERD20T. The corresponding anti-ePsl immune-dot blot and arabinose concentration are listed below each bar. The corresponding value of attachment assay for each strain shown under is normalized to the level of PAO1/pHERD20T, the superscript letter "a" indicates a significant difference compared to PAO1/pHERD20T of p < 0.01, as determined by Student's t test. **p < 0.01, Student's t test pHERD20T, Table 1), but it could not be restored by plasmid pGDM (PslG E165Q + E276Q in pHERD20T), regardless of the inducer level applied (0%, 0.5%, or 1%) (Figure 1c). The corresponding attachment was also consistent with the ePsl production ( Figure 1c, the value shown under each column). These results suggested the importance of PslG glycoside hydrolytic activity in ePsl production and implied that the hydrolytic activity was not only required for degradation of ePsl, but also involved in the biosynthesis of ePsl. Taken together, these results suggested that the PslG and its hydrolytic activity contributed on ePsl production and initial attachment in PAO1.

| Inner membrane fraction of PslG is critical for the biosynthesis of ePsl
The results of Baker et al. (2015) indicated that PslG could localize to both the inner membrane and the periplasm. We further in- The previous publication showed that SadC was localized in the inner membrane (Zhu et al., 2016). Therefore, we have transferred a plasmid pSadC-GFP (carrying the sadC-gfp gene, Table 1) into all tested strains in order to use the SadC-Gfp as a loading control for membrane fraction. In addition, RNA polymerase was used as a loading control for the cytoplasmic fraction. The results of loading controls indicated that the same amount of cell fractions was loaded for each experiment, and each fraction was well separated.
We then further studied whether the expression level of PslG affected its localization. The pG could restore ePsl production of In the ePsl-inducible strain WFPA801, the transcription of entire psl locus was induced by arabinose, its PP/IM value of PslG was 0.37 with 1% arabinose (Figure 2b). For ΔpslG2/pG, arabinose only induced the expression of PslG, there was more PslG localized in the periplasm, the PP/IM value of PslG was 0.97, 1.93, and 2.14 while induced with 0.01%, 0.1%, and 1% arabinose, respectively (Figure 2b).

| The localization of PslG is affected by PslA, PslD, and PslE
To figure out any Psl protein affecting the localization of PslG, we focused on proteins PslA, PslD, and PslE, which were predicted to be localized on the inner membrane and possessed periplasmic domains (Franklin et al., 2011). WFPA801ΔpslA, WFPA801ΔpslD, and WFPA801ΔpslE containing the plasmid pSadC-GFP were constructed to examine the effect of Psl proteins on the localization of PslG. Western blot results showed that more PslG localized in the periplasm than in the inner membrane in above PslA, PslD, or PslEdeleted strains (Figure 3a). The ratio of PslG in periplasm to inner membrane was 1.46, 1.77, and 1.42 in PslA, PslD, and PslE mutants ( Figure 3b), indicating that these three proteins are important to maintain PslG in the inner membrane.

| Protein-protein interaction among PslE with PslA, PslD, and PslG
We utilized bacterial two-hybrid system to determine whether there are direct interactions among PslA, PslD, PslE, and PslG (Table 2). These results were consistent with the results of proteins interactions ( Table 2), suggesting that PslE might help PslD to span to the outer membrane by direct PslE-PslD interaction.

| D ISCUSS I ON
The ePsl is a key biofilm matrix component of the life-threaten pathogen P. aeruginosa. It promotes bacteria cell-cell and cellsurface interaction by acting as a "molecular glue" (Ma et al., 2009(Ma et al., , 2006; it forms a fiber-like matrix to protect bacteria from antibiotics and phagocytic cells (Billings et al., 2013;Mishra et al., 2012); and it can function as a signal to stimulate biofilm formation . However, the molecular mechanism of ePsl biosynthesis remains unknown. In this study, we focused on the role of glycoside hydrolase PslG in the biosynthesis of ePsl. We Glycoside hydrolases are common in many bacterial exopolysaccharide biosynthesis operons, such as PssZ in Listeria monocytogenes (Koseoglu et al., 2015), PgaB and BcsZ in E. coli (Mazur & Zimmer, 2011;Wang, Preston, & Romeo, 2004), and WssD and AlgL in Pseudomonas fluorescence (Bakkevig et al., 2005;Spiers, Bohannon, Gehrig, & Rainey, 2003). Our previous study demonstrated the structure of glycoside hydrolase PslG and its effects on biofilm when applied exogenously (Yu et al., 2015), while little is known about its function in the process of ePsl biosynthesis. Baker et al. (2015) had studied the role of pslG in a psl overexpression strain PAO1ΔpelFP BAD psl. They concluded that pslG had no involvement in the biosynthesis of ePsl. However, in a psl overexpression system, only a huge change on ePsl production could be find. Therefore, to determine the role of PslG and its endoglycosidase activity in the biosynthesis of ePsl in P. aeruginosa PAO1, we constructed strain ΔpslG2 and ΔpslG2::pslG E165Q + E276Q , and found that PslG and its hydrolytic activity were important for initial attachment and ePsl production. Monday and Schiller (1996) and Penaloza Vazquez, (1997)   . Though the differences in ePsl production between WFPA800, ΔpslG2, and ΔpslG2::pslG E165Q + E276Q were not enough to make significant differences in a 30 min attachment assay, the differences of biofilm biomass could be found in a biofilm assay post 2 hr incubation (Appendix Figure A2), in which the biofilm biomass of ΔpslG2, and ΔpslG2::pslG E165Q + E276Q were slightly higher than WFPA800, suggesting the ePsl synthesized from pslG mutants is functional.
PslG localizes in the inner membrane and periplasm (Baker et al., 2015). We are interested in whether the specific localization of PslG plays different role in the biosynthesis of ePsl. We found PslG in wild type PAO1 mainly localized in the inner membrane. When PslG was overexpressed alone, more PslG localized in the periplasm with a decrease in ePsl production. These results suggest inner membrane association of PslG helps synthesize ePsl polymer, while PslG in the periplasm may degrade ePsl polymer randomly.
As the localization of PslG is critical to ePsl production, we have further investigated other Psl proteins that might modulate the localization of PslG. We focus on the predicted periplasmic proteins (PslA, PslD, and PslE) that may interact with PslG in the ePsl assembly apparatus. We found that more PslG localized to the periplasm in the absence of PslA, PslD, or PslE. Interaction of PslE with PslG was further confirmed via bacterial two-hybrid assay.
These results suggested that the membrane-associated PslG was a part of ePsl biosynthesis machinery, in which PslA, PslD, and PslE might help control or delay the release of PslG into periplasmic space. Our data have also shown that the hydrolytic activity of PslG is important for the synthesis of ePsl, implying that the ePsl biosynthesis machinery may allow PslG in an optimal localization to control the degradation of ePsl polymer at certain length ( Figure 5).
The structures and functions of PslA, PslD, and PslE have not been experimentally determined. PslA might likely play a similar role to WbaP in providing a site for the assembly of the oligosaccharide repeating unit onto the isoprenoid lipid at the cytoplasmic face of the inner membrane (Franklin et al., 2011;Whitfield, 2006).
PslE has characteristic domains of a Wzz (or Wzc) homolog and is therefore predicted to act as the polysaccharide co-polymerase (PCP) component in this system (Franklin et al., 2011;Larue, Kimber, Ford, & Whitfield, 2009). The periplasmic domain of PCPs is proposed to affect polysaccharide chain length (Tocilj et al., 2008) and is thought to form critical interactions with the CPS/EPS export component thereby completing a complex that facilitates transfer of the polymer through the periplasm (Cuthbertson, Mainprize, Naismith, & Whitfield, 2009). PslD is predicted to be the polysaccharide exporter with structural similarity to the E. coli K30 capsule translocase, Wza, an integral outer membrane lipoprotein (Dong et al., 2006;Franklin et al., 2011). Predicted PslD 3-dimensional structure (Appendix Figure A3) has indicated that most of PslD can be structurally modeled onto Wza (PDB ID 2J58), but there is a clear dif- ference, PslD appears to lack the outer membrane barrel and large periplasmic domain. Therefore, it is difficult to understand how the Psl polymer is translocated across the outer membrane. In this study, we found that PslD had a strong interaction with PslE and it could not localize to the outer membrane without PslE, which suggest that PslE, the Wzc homolog, interacts with PslD and helps PslD localize to the outer membrane. In addition, our data also suggest PslE is likely to act as the periplasmic scaffold and recruit proteins to form a polysaccharide biosynthetic complex because PslE can interact with PslA, PslD, and PslG (Table 2, Figure 5). More PslD was detected in PAO1 than in pslA, pslE, or pslG deletion mutant, implying that PslD integrated into the ePsl biosynthetic complex is more stable than free PslD.
To the best of our knowledge, this is the first study to investigate the connection between protein interactions and their localizations during ePsl biosynthesis of P. aeruginosa. Our data showed the glycoside hydrolase PslG and its hydrolytic activity were important to ePsl production of P. aeruginosa. The inner membrane association of PslG might be involved in the biosynthesis of ePsl, while PslG localized in the periplasm may degrade ePsl. We have experimentally proved the PslE interacted with PslA, PslD, and PslG in vivo. All the three proteins, PslA, PslD, and PslE, had an impact on PslG localization, which was critical to ePsl biosynthesis. PslE helped PslD localize the outer membrane, these two proteins might form a complex to help transport Psl across the outer membrane. In summary, we have shown in this study that ePsl biosynthesis is a complex processing with dynamic protein-protein interactions, leading to the assembly of ePsl biosynthesis machinery.

ACK N OWLED G M ENTS
We thank Prof. Lichuan Gu at the Shandong University for pro- F I G U R E 5 A schematic view of Psl proteins in the ePsl biosynthesis machinery. Membrane associated PslG is a part of the ePsl biosynthesis machinery, in which PslA, PslD, and PslE allow PslG in an optimal localization to control the degradation of ePsl polymer at certain length. PslE interacts with PslA, PslD, and PslG, it helps PslD to localize in the outer membrane to export the ePsl

CO N FLI C T O F I NTE R E S T S
The authors declare that there is no conflict of interest.

AUTH O R CO NTR I B UTI O N S
H.W., D.W., and L.Z.M. conceived and designed experiments, and contributed to the writing of the manuscript. H.W. and M.T. conducted experiments.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
All data are provided in full in this paper.