Cell division factor ZapE regulates Pseudomonas aeruginosa biofilm formation by impacting the pqs quorum sensing system

Abstract Pseudomonas aeruginosa is one of the leading nosocomial pathogens that causes both severe acute and chronic infections. The strong capacity of P. aeruginosa to form biofilms can dramatically increase its antibiotic resistance and lead to treatment failure. The biofilm resident bacterial cells display distinct gene expression profiles and phenotypes compared to their free‐living counterparts. Elucidating the genetic determinants of biofilm formation is crucial for the development of antibiofilm drugs. In this study, a high‐throughput transposon‐insertion site sequencing (Tn‐seq) approach was employed to identify novel P. aeruginosa biofilm genetic determinants. When analyzing the novel biofilm regulatory genes, we found that the cell division factor ZapE (PA4438) controls the P. aeruginosa pqs quorum sensing system. The ∆zapE mutant lost fitness against the wild‐type PAO1 strain in biofilms and its production of 2‐heptyl‐3‐hydroxy‐4(1H)‐quinolone (PQS) had been reduced. Further biochemical analysis showed that ZapE interacts with PqsH, which encodes the synthase that converts 2‐heptyl‐4‐quinolone (HHQ) to PQS. In addition, site‐directed mutagenesis of the ATPase active site of ZapE (K72A) abolished the positive regulation of ZapE on PQS signaling. As ZapE is highly conserved among the Pseudomonas group, our study suggests that it is a potential drug target for the control of Pseudomonas infections.


INTRODUCTION
Pseudomonas aeruginosa is a notorious opportunistic human pathogen causing a wide range of infections, such as cystic fibrosis (CF) lung infections, ventilator-associated pneumonia, and chronic wound infections 1,2 .P. aeruginosa infections are difficult to treat due to their increasing resistance as well as their biofilm formation capacity.Biofilms impede the diffusion of antimicrobial agents and preserve virulence factors (e.g., rhamnolipids), which impairs host defense systems and leads to persistent infections 3,4 .Moreover, the high cell density within the biofilm also promotes an ideal environment for the adaptive evolution of novel bacterial lineages such as the small colony variants 5 .
Biofilm cells have a distinct physiology compared to their free-living counterparts 6 .Multiple signaling pathways participate in the regulation of biofilm development, such as synthesis of extracellular polymeric substances (EPS), downregulation of motility, and enhancement of adhesion.There have been extensive studies focusing on the synthesis, localization, and functions of P. aeruginosa exopolysaccharides during biofilm development 7 .The exopolysaccharide synthesis in P. aeruginosa as well as many other bacteria is under the control of bis-(3'-5')-cyclic diguanylic acid (c-di-GMP), a widely distributed bacterial intracellular secondary messenger that also regulates many other physiological processes 8 .C-di-GMP regulates its target pathways via specific effectors such as PilZ domain containing proteins, which can then directly or indirectly affect gene expression 8 .The bacterial cell-to-cell communication system (also known as quorum sensing, QS) is another essential regulatory mechanism for biofilm formation 9 .QS employs small diffusible signal molecules to coordinate the expression of genes in a densitydependent manner and is, therefore, an ideal strategy to regulate gene expression in the biofilm microcolonies.
P. aeruginosa has four QS systems, las/rhl/pqs/iqs, that regulate a large set of genes in its genome 10 .The las and rhl QS systems employ N-acyl-homoserine lactones (AHLs) as signaling molecules, which activate specific intracellular LuxRtype receptors to regulate gene expression.The third QS system of P. aeruginosa uses the Pseudomonas quinolone signal (pqs), heptyl-3-hydroxy-4-(1H)-quinolone (PQS) as a QS molecule, which is highly specific to Pseudomonas species and could serve as a biomarker for the diagnosis of P. aeruginosa infections [10][11][12] .The pqs QS system is more complicated than the AHL QS systems in that the chemical compound 2-heptyl-4-quinolone (HHQ) is synthesized by the pqsABCD gene cluster and converted to PQS via a monooxygenase PqsH 10,11,13 .pqs QS regulates the biosynthesis of virulence factors such as pyocyanin and the PQS could serve as an iron chelator.Iron deficiency condition stimulates PQS synthesis, whereas excessive production of PQS will lead to autolysis of P. aeruginosa and the release of extracellular DNA (eDNA), which is an important biofilm matrix component 14,15 .Currently, the induction kinetics and function of the pqs QS system are still not fully elucidated.
Transposon insertion site sequencing (Tn-seq), which combines the transposon mutagenesis method with highthroughput sequencing technology, is a global approach to provide comprehensive information on gene function on a genome-wide scale 16,17 .Himar1, a transposon that could insert randomly at TA dinucleotides 18 , has been previously used in P. aeruginosa investigations.By applying Tn-seq analysis on tubing-biofilms, five novel genetic determinants were identified for P. aeruginosa biofilm formation in this study.Among them, we systematically investigated the role of PA4438.PA4438 is predicted to be homologous to zapE, which encodes a cell division factor and AFG1 (ATPase family gene 1, which belongs to AAA+ protein) in Escherichia coli 19 .Moreover, we observed that the ΔPA4438 mutant had a similar elongated phenotype under oxygen limitation and high-temperature conditions as the E. coli ΔzapE mutant, hence, we named PA4438 as zapE.We provide evidence that ZapE is involved in the conversion of HHQ to PQS by PqsH, which is dependent on its ATP hydrolyzing activity.Based on our study, we speculate that the AAA+ protein ZapE is a potential drug target due to its multiple functions involved in both cell division and PQS signaling.

Tn-seq analysis of P. aeruginosa biofilm formation
To systematically investigate genes that contribute to P. aeruginosa biofilm formation, Tn-seq analysis was employed to study P. aeruginosa tubular biofilms.A pooled library of nearly 500,000 mutants was constructed in the wild-type strain PAO1 by using the Himar1 transposon.In this PAO1 mutant library, the coverage rate of TA insertion sites was about 55%, which was higher than the ideal coverage of 30%, suggesting that coverage of TA insertion sites reached sufficient saturation.After the bottom 10% read counts were removed, there were 5158 insertion genes in this library (compared to the 5570 genes in the PAO1 genome), which further confirmed that the library contained almost all nonessential gene mutants.
As shown in the schematic drawing in Figure S1A, the PAO1 mutant library was cultivated in the tubular biofilm reactors for 3 days, giving sufficient time to enrich the mutants with good fitness in the biofilm mode of growth, while low fitness biofilm mutants would be selected out.Pearson's correlation analysis indicated that there were significant differences between the input pools and output pools (Figure 1A).Furthermore, the excellent correspondence between technical replicates (R 2 of two input mutant pools was 0.9791) (Figure 1B) and between biological replicates (R 2 of two output mutant pools was 0.9159) (Figure 1C) suggested that the cultivation of Tninsertion mutants in tubular biofilms with the fluid flow was reproducible.

Identification of genes required for P. aeruginosa biofilm formation
We normalized the abundance of Tn insertion mutants in each mutant pool by DESeq2 and used the normalized values to calculate the input-to-output ratio.A high ratio suggests that the corresponding mutated gene is important for biofilm formation.In total, 924 mutants had a statistically different abundance after biofilm cultivation in the tubular system compared to that before cultivation, with 902 mutants having an increase in the input-to-output ratio after biofilm cultivation (Figure S1B, Table S1).About 71.0% (640 mutants) of the above mutants had a fivefold or greater reduction in output mutant pools compared to the input mutant pool (Figure 1D), suggesting that the mutated genes might be biofilm determinants.Next, we categorized mutants with high input-to-output ratios by gene function and found that the mutated genes were involved in amino acid metabolism, energy metabolism, cell membrane formation, and wellknown biofilm formation mechanisms (Figure 1E).In addition, mutants with insertions in exopolysaccharide synthesis and motility genes also had high input-to-output ratios (Figure S2A-E), suggesting that our Tn-seq results reasonably reflected the previous biofilm studies.Interestingly, besides the above mutants, we also found that mutants with insertions in hypothetical genes had high input-to-output ratios, indicating that genes with unknown functions might play important roles in P. aeruginosa biofilm formation.

Validation of candidate biofilm determinants with unknown function
We selected five hypothetical genes (PA0222, PA1112, PA2345, PA3797, and PA4438) with the highest fold changes between input and output mutant pools for experimental validation (Table 1).Clean knockout mutants for these five genes were constructed, and the biofilm morphology of each mutant was examined by using a confocal laser scanning microscope (CLSM).Compared to the PAO1 wild-type strain, ΔPA0222 showed similar biofilm structures with slightly smaller microcolonies; ΔPA1112, ΔPA2345, and ΔPA3797 formed tiny microcolonies with significantly less biofilm biomass; and ΔPA4438 formed loose biofilm aggregates with decreased amounts of biomass (Figures 2A  and S2F).To further compare the biofilm formation ability of different strains, mutants (tagged with GFP) and PAO1 (tagged with mCherry) were mixed at a 1:1 ratio to cultivate biofilms, and flow cytometry was then used to analyze the proportion of fluorescent cells isolated from coculture biofilms.The results show that all five mutants lost fitness against PAO1 in coculture biofilms, and ΔPA4438 had the lowest fitness among the five mutants (Figure 2B).Furthermore, the growth rates of these five mutants showed no significant difference compared to PAO1 (Figure 2C), suggesting that the reduced biofilm biomass and fitness of these mutants in mixed biofilms were not due to growth deficiency.Taken together, these results suggest that each of the five genes, particularly PA4438, plays an important role in P. aeruginosa biofilm formation and are novel genetic determinants.
PA4438 is a homologue of the E. coli cell division protein ZapE PA4438 is predicted to encode an AAA+ AFG1 ATPase and its predicted amino acid sequence has regions of high conservation to the cell division protein ZapE of E. coli (EcZapE) (Figure S3).A previous study reported that the loss of Ec-ZapE would result in cell elongation under oxygen-limited and high-temperature conditions 19 .We examined the morphology of PAO1, ΔPA4438, and its complementation strains under 37°C aerobic, 42°C aerobic, and 37°C anaerobic environment conditions.Significant elongation of ΔPA4438 cell size was observed under 42°C aerobic and 37°C anaerobic conditions (Figure 3A).Complementation by PA4438 or ec-zapE (zapE from E. coli) to the ΔPA4438 mutant (ΔPA4438/ pPA4438 and ΔPA4438/peczapE) reverted its cell size back to a similar length of PAO1 under anaerobiosis and hightemperature conditions (Figure 3A).These results confirmed that PA4438 encodes a functional homologue of EcZapE, hereafter, we name PA4438 as zapE.
Next, we evaluated the biofilm formation ability of PAO1, ΔzapE, ΔzapE/pzapE, and ΔzapE/peczapE (Figure 3B).The biofilm phenotype of the two complementation strains showed no significant difference with PAO1, implying that EcZapE might play a regulatory role in E. coli biofilm formation.This suggests that zapE (PA4438) is not only important for regulating biofilm formation but also for proper cell division under oxygen-limited and high-temperature environments.

ZapE regulates the pqs QS and pyoverdine synthesis
To further investigate the regulatory mechanism of zapE in P. aeruginosa biofilm formation, the EPS production and motility of the ΔzapE mutant were examined.The results showed that exopolysaccharides and motility of ΔzapE were indistinguishable from PAO1 (Figure S4).Next, RNAsequencing-based transcriptomic analysis was employed to elucidate the global effect of zapE on P. aeruginosa gene expression.This analysis showed that 277 genes were differently expressed in the ΔzapE compared to PAO1 (log 2 |Fold-change| > 2, p < 0.05), including 37 upregulated genes and 240 downregulated genes (Figure S5A).Among these 277 significant genes, there were 111 genes with annotated functions.The expression levels of genes that encode well-known biofilm structural factors such as  exopolysaccharides, pili, and flagellar were similar between ΔzapE and PAO1 (Figure S5B-E), which is consistent with the results from phenotypic analysis (Figure S4).Strikingly, the expression levels of genes involved in phenazine synthesis (14.4% of the total differentially expressed genes with annotated function), pyoverdine synthesis (14.4% of the total differentially expressed genes with annotated function), inorganic ion transport and metabolism (12.6% of the total differentially expressed genes with annotated function), intracellular trafficking and secretion (9.0% of the total differentially expressed genes with annotated function), energy metabolism (8.1% of the total differentially expressed genes with annotated function), and QS (8.1% of the total differentially expressed genes with annotated function) showed the biggest difference between ΔzapE and PAO1 (Figure 4A).Almost all pqs QS regulated genes were expressed at a lower level in ΔzapE compared to PAO1 (Figure 4B).Since the pqs QS system is involved in autolysis and the release of extracellular DNA, it is activated within a narrow window during the planktonic growth 20 ; we thus examined the activation of pqs QS in ΔzapE and PAO1 by introducing the bioreporter P pqsA -gfp fusion 20 .The results showed that P pqsAgfp expression was significantly decreased in ΔzapE during the entire exponential phase (Figure S6A).We also quantified the production of pyocyanin (a derivative of phenazine) and pyoverdine by PAO1, ΔzapE, and the complementation strain, and the results were consistent with the RNA-seq analysis, the production of pyocyanin and pyoverdine were decreased in the absence of zapE (Figures 4B and 5A,B).
Pyoverdine is a siderophore and signal molecule that regulates biofilm formation of P. aeruginosa 21,22 .As the expression of fpvA and fpvB, which are essential for the uptake of ferric-pyoverdine complexes by P. aeruginosa 23,24 , was also downregulated in ΔzapE compared to PAO1 (Table S2), we suspected that iron limitation is a factor causing biofilm reduction of the ΔzapE mutant.To test this hypothesis, we added an iron chelator (Dipy, 20 μg/ml) to the biofilm culture to cause an iron-limited environment.PAO1 formed less robust biofilm structures with smaller colonies in the iron-limiting medium compared to the normal growth medium, which was similar to the biofilm morphology of ΔzapE grown in the normal growth medium (Figure S7A).As expected, ΔzapE could rarely form regular biofilm structures in iron-limiting conditions (Figure S7A).To further confirm the effect of iron on the ΔzapE mutant biofilm formation, ferric iron was exogenously added into the cultures of PAO1 and ΔzapE.The results showed that iron promoted the biofilm formation of the ΔzapE, although the overall amount of biomass did not reach the level obtained by PAO1 (Figure S7B).

ZapE interacts with PqsH
To further elucidate the link between ZapE and pqs QS, immunoprecipitation and mass spectrometry (IP-MS) were used to identify potential ZapE-interacting proteins.Remarkably, four proteins with well-known functions in relation to PQS and phenazine synthesis, including PqsH, PhzD1, PhzD2, and PhzM, were identified as potential interacting proteins of ZapE, besides the only known interacting target FtsZ 19 (Table 2).A surface plasmon resonance (SPR) assay was performed to verify the interaction between ZapE and PqsH, and the results suggested that ZapE could directly interact with PqsH (Figure S6B).

ZapE is an ATPase that is required for the synthesis of PQS
PqsH is an NADH-dependent monooxygenase, which converts HHQ to PQS 13 .As ZapE was shown to interact with PqsH and there was no difference in expression of pqsH in ΔzapE compared to PAO1 according to RNA-seq analysis (Figure 4B), we hypothesized that the low level of pqs QS of ΔzapE might be caused by the reduced enzyme activity of PqsH.To verify this, liquid chromatography-tandem mass spectrometry (LC-MS/ MS) was used to quantify the amount of HHQ and PQS in PAO1/pUCP20, ΔzapE/pUCP20, and ΔzapE/pzapE.As expected, the results confirmed that ΔzapE/pUCP20 produced less PQS compared to the wild-type PAO1/pUCP20 (Figure S6C), while there was no statistical difference in HHQ production between these three strains (Figure S6D).To further test PqsH enzyme activity, HHQ was exogenously added into cultures of PAO1 and ΔzapE, and the PQS bioreporter pqsAgfp was used to assess the pqs QS levels of these two strains.In the absence of zapE, exogenous HHQ did not restore pqs QS activity to the level of PAO1 (Figure 5C).Furthermore, ΔzapE cell lysates with exogenously added recombinant ZapE at 1 μM produced a higher amount of PQS compared to PAO1 cell lysates (Figure S6E), suggesting that ZapE promotes the conversion of HHQ to PQS.To further confirm whether higher amounts of ZapE increase the activity of PqsH in vivo, the pqs QS levels of a zapE chromosomal complementation strain ΔzapE::zapE and a zapE overexpression strain PAO1::zapE were measured.The results showed that the chromosomal complementation strain ΔzapE::zapE restored pqs QS to PAO1 levels; however, the overexpression strain PAO1::zapE produced a lower amount of PQS compared to PAO1, suggesting that the higher amounts of ZapE did not increase the activity of PqsH (Figure S6F).
Since ZapE is predicted to be an ATPase belonging to AAA+ proteins, we measured ATP hydrolysis activity of ZapE.The results showed that ZapE is a Ca 2+ /Mg 2+ ATPase (Figure S6G).An alanine substitution was introduced into the predicted ATP binding site of ZapE (K72A), which decreased the ATPase activity of ZapE (K72A) 10-fold compared to ZapE (Figure S6H).To explore whether the ATPase activity of ZapE is involved in PQS synthesis and pqs QS, we constructed the ΔzapE/K72A mutant for the phenotypic assays.The ΔzapE/K72A mutant did not produce PQS, pyocyanin, or pyoverdine to levels reached by wild-type PAO1 (Figures S6C and 5A-C).This suggests that the ATPase enzyme activity of ZapE is required for the synthesis of PQS and the pqs QS.

ZapE is highly conserved among the Pseudomonas species
The amino acid sequence of ZapE appeared to be highly conserved not only in P. aeruginosa strains (PAO1, PA14, and PAK), but also in other members of the Pseudomonas group, such as Pseudomonas stutzeri, Pseudomonas putida, Pseudomonas chloritidismutans, and Pseudomonas fluorescens (Figure S3).To examine whether ZapE plays a key role in the biofilm formation and cell division in other P. aeruginosa strains, zapE in-frame deletion mutants were constructed in PA14 and PAK.The cell shape and biofilm morphology of zapE mutants (PA14ΔzapE and PAKΔzapE), wild-type strains (PA14 and PAK), and their complementation strains (PA14ΔzapE/ pzapE and PAKΔzapE/pzapE) were observed by CLSM.Both PA14ΔzapE and PAKΔzapE had elongated cells under anaerobic and high-temperature conditions (Figure 6A).And the cell size of their complementation strains restored to the level of their wild type.More importantly, biofilms formed by PA14Δ-zapE and PAKΔzapE had thinner layers and were less aggregated in comparison to their wild-type strains and complementation strains (Figure 6B).

DISCUSSION
P. aeruginosa is a leading nosocomial human pathogen with the capability to form biofilms.Its biofilm mode of growth is associated with chronic infections, which are difficult to eradicate due to their resilience to antibiotic therapy and host immune clearance 25 .Currently, approaches to understand biofilm formation mechanisms rely on the characterization of single gene knockout mutants of the model strains, which may not resemble natural conditions.In natural ecological systems, bacteria are nested in communities with highly intraspecific diversity, where variants with different biofilm formation capacities evolve.In this study, we applied the high-throughput Tn-seq method in combination with the robust tubular biofilm growth system to mimic the clinical biofilms where a considerable number of variants coexist and compete with each other.This integrated approach enables us to identify new genes that are vital for P. aeruginosa biofilm formation and maintenance, some of which have no predicted function (Figure 1E).In this study, we characterized a new protein involved in biofilm development, ZapE (PA4438), which was shown to be a cell division factor and a Ca 2+ /Mg 2+ ATPase belonging to the AFG1 (ATPase family gene 1) AAA+ proteins 26 .The AAA+ (AT-Pases associated with diverse cellular activities) proteins can generate energy from adenosine triphosphate (ATP) hydrolysis and are known to play significant roles in many biochemical processes, such as gene expression, chaperoning protein folding or unfolding 27 .AAA+ proteins are recognized by their Walker A and Walker B motifs and an extra β-strand between these two motifs [26][27][28][29] .So far, little is known about whether AAA+ proteins are involved in bacterial biofilm formation.
As ZapE is essential for bacterial cells to survive and maintain proper cell shape under anaerobic conditions (Figure 3), we initially speculated that the reduced biofilm biomass and loss of fitness of the ΔzapE mutant compared to the wild-type PAO1 strain were due to the lack of oxygen as a thick biofilm developed.However, we noticed that the ΔzapE mutant lost fitness against the PAO1 strain in static biofilm cocultures where only thin-layered biofilms developed.RNAseq analysis revealed that the ΔzapE mutant turned down the expression of most pqs QS regulated genes, which are critical for P. aeruginosa biofilm formation.The pqs QS regulates P. aeruginosa via several different mechanisms.First, pqs QS regulates the release of extracellular DNA and membrane vesicles, which serve as EPS components of P. aeruginosa biofilms 30,31 .Second, pqs QS regulates the synthesis of pyocyanin, which not only promotes the release of extracellular DNA 32 , but also facilitates the binding of extracellular DNA to the cell surface of P. aeruginosa 33 .Third, pqs QS positively regulates the synthesis of pyoverdine 34 , known to enhance subpopulation interactions and formation of large microcolonies of P. aeruginosa biofilms 35 .PQS is also an iron chelator, which helps P. aeruginosa to overcome iron limitation during infections 34 .Several efficient P. aeruginosa biofilminhibiting compounds have been developed by targeting pqs QS 36,37 .Recent studies have revealed the immune-modulating functions of the PQS molecules, which again might enhance the progress of P. aeruginosa infections 38 .Downregulation of pqs QS genes and decreased production of pqs-associated compounds for biofilm formation in the ZapE mutant suggest that ZapE regulates biofilm formation by involvement with pqs QS networks.
We identified that PqsH is a target protein of ZapE (Table 2 and Figure S6B), which provides a direct link between ZapE and pqs QS.To explore the binding interactions of PqsH with ZapE, protein-protein docking simulations were carried out.Since the three-dimensional (3D) structure of ZapE is unknown, we constructed its 3D structure by AlphaFold (Figure S8A,B), which is a novel machine learning approach that incorporates physical and biological knowledge about protein structure 39 .The results suggested that the PqsH residues Arg168, Arg332, Lys333, Trp336, Arg340, Ser364, Arg367, and Arg371 are involved in binding with the ZapE residues Pro14, Phe16, Asp19, Ser197, Gly18, Asp200, Arg204, Gln208, Glu351, Gln353, Glu356, and Thr359 through salt bridges and hydrogen bond interactions (Figure S8C,D and Table S3).PqsH is an NADH-dependent flavin monooxygenase that oxidizes HHQ to PQS 13 .Our results showed that exogenous HHQ could not restore the pqs QS of ΔzapE to PAO1 levels.The zapE chromosomal complementation strain ΔzapE::zapE (Figure 5D) and ΔzapE cell lysates with exogenously added recombinant ZapE at 1 μM (Figure S6E) had higher pqs QS levels than ΔzapE.We also showed that complementation of the ΔzapE mutant using an ATPase activity-deficient K72A ZapE variant failed to restore pqs QS to PAO1 levels.Our work provides strong evidence for the involvement of the ATPase activity of ZapE in the synthesis of pqs QS molecules (Figure 7).
In summary, our work showed that ZapE, a cell division factor and an AAA+ protein, is a novel biofilm regulator through modulating the pqs QS.To our knowledge, this is the first report of an AAA+ protein as a biofilm determinant and pqs QS regulator.As the ZapE amino acid sequence is highly conserved among the Pseudomonas group, this study could also provide a novel target for the development of anti-Pseudomonas compounds.

Transposon insertion mutant library generation
The pBT20 was chosen to construct the mutant library in this study, following the standardized method reported previously 40 .In brief, triparental conjugation was used to introduce pBT20 into P. aeruginosa PAO1.The transposon mutants were harvested and mixed thoroughly, which were adjusted to a bacterial density at OD 600 ~2.The transposon mutant suspension was preserved in 50% glycerol at −80°C with 1 ml per cryotube.The transposon insertion was examined by arbitrary PCR using primers Pa-Rnd1-F and BT20-Rnd1-R, Pa-Rnd2-F, and BT20-Rnd2-R (Table S5).

Biofilm cultivation of Tn-seq mutant libraries
Tubing-biofilms of the PAO1-Tn insertion mutant library were cultivated at 37°C following the standard method described previously 41,42 .In brief, the preculture for biofilm inoculation was prepared by enrichment of triplicate, parallel PAO1-Tn mutant library stocks in LB medium (1:9 vol/ vol) for 6 h.The precultures were mixed thoroughly followed by adjusting the cell concentration to OD 600 = 1.Thereafter, 1 ml of mixed preculture (input) was inoculated into replicate tubular biofilm systems and allowed initial attachment of bacteria for 2 h before starting the flow.The feeding medium was 1/10 LB medium, and the flow rate was set at 4 ml/h.Biofilms in the tubular systems were harvested after 3 days of cultivation (output) for genome DNA extraction.

Sequencing and analysis of transposon mutant libraries
The genomic DNA of both input and output transposon mutant libraries were extracted by the QIAamp DNA Mini Kit (Qiagen).The transposon sequencing library was prepared using a modified method reported previously 43 .Genomic DNA was sheared to 200-500 bp fragments, before end repair and dAtailing using the TIANSeq fragmentation/end repair/dA-tailing module (Tiangen).Illumina multiplexing adaptors (Tn-AD-1 and Tn-AD-2) were ligated onto the repaired genomic DNA by the TIANSeq Fast Ligation module (Tiangen).Thereafter, transposon junction sites were amplified with primers (Tn-Rnd1-F and Tn-Rnd1-R, Tn-Rnd2-F, and Tn-Rnd2-R), which recognize the end of the transposon and adaptor by Phusion Hot Start II High-Fidelity PCR Master Mix (Thermofisher).All adaptors and primers used in this study are listed in Table S5.The raw data of transposon mutant library sequencing were trimmed and mapped to the PAO1 reference genome.Unique gene reads were recorded and normalized based on the DESeq2 normalization method.Statistical analysis was performed by Origin9 (OriginLab Corporation), GraphpadPrism9 (Graphpad), and R-packages (https://www.r-project.org/).Significance was determined by criteria: (a) log 2 |Fold-change| > 2; (b) p value via multiple t test < 0.05.

In-frame deletion mutants, site-directed mutants, and complementation strain construction
To construct P. aeruginosa mutants, pK18-Gm-mobsacB was used as the vector of the knockout plasmid 44 , and all primers used in this study are listed in Table S5.First, upstream and downstream fragments of the target gene were amplified (Uprimer-F and Uprimer-R, Dprimer-F, and Dprimer-R) and fused into the knockout plasmid.Then, the target gene deletion mutant was constructed by in-frame deletion 45 .The complementation strain was constructed using the complementation plasmid pUCP20.The target gene was amplified and fused into pUCP20, which was transformed into the target gene deletion mutant by electroporation.To construct the ZapE K72A site-directed mutant, the complementation plasmid pUCP20-zapE was used as the template, and K72A-F and K72A-R were applied as PCR primers.To construct the zapE chromosomal complementation strain, an integration plasmid miniCTX-lacZ was used.The promoter region and the open reading frame of zapE were cloned into miniCTX-lacZ, and CTX-zapE-F and CTX-zapE-R were applied as PCR primers.The resultant plasmid was transformed into ΔzapE and PAO1 to construct the zapE chromosomal complementation strain ΔzapE::zapE and zapE overexpression strain PAO1::zapE, respectively.Reagents and kits used in this section included the QIAamp DNA Mini Kit (Qiagen), HiPure Gel Pure DNA Mini Kit (Magen), HiPure Plasmid Micro Kit (Magen), Q5 High-Fidelity 2× Master Mix (NEB), PrimeSTAR® HS DNA Polymerase with GC Buffer (Takara), and Gibson Assembly Master Mix (NEB).

Transcriptomic sequencing analysis
The total RNA was extracted with a QIAmp RNA kit from liquid culture of PAO1 and its mutant strain at the late exponential growth stage.The mRNA was purified using the NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB), and the sequencing library was prepared using the NEBNext Ultra II mRNA Library Prep Kit for Illumina (NEB).The library quality was examined by Qubit dsDNA highsensitivity assay (Thermofisher) and D1000 Screen Tape (Agilent) for concentration and fragmentation.Transcriptomic sequencing was performed on NovaSeq.6000 (Illumina Inc.) with a NovaSeq S4 reagent kit.The transcriptomic raw data were mapped to the PAO1 reference genome by the CLC genomics workbench (CLC bio, Qiagen) and were analyzed by DESeq statistical analysis.Significance was determined by criteria: (a) log 2 |Fold-change| > 2; (b) p value via multiple t test < 0.05.

Biofilm assay and imaging analysis
Biofilm formation and structures were evaluated by CLSM (Zeiss LSM 900).Static biofilms were prepared by inoculating overnight bacterial cultures at a final concentration of OD 600 ~0.01 and were cultivated for 24 h at 37°C.Competition biofilm assay was prepared by mixing PAO1 with PAO1 and its mutant at 1:1 vol:vol with final bacterial concentration at OD 600 ~0.01, followed by cultivation.CLSM 3D images were captured by LSM900 and analyzed by Imaris (Imaris, bitplane), ImageJ (https://imagej.nih.gov/ij/), and Zen (Zeiss).

Evaluation of growth and cell size of bacteria
The growth of PAO1 and its mutant as well as complementation strains were continuously monitored by a microplate reader (Tecan Spark).Overnight strain cultures were adjusted to a concentration of OD 600 ~0.01 in LB broth, which were allowed to grow at 37°C for 16 h with periodic shaking.Growth was monitored by measuring absorbance at 600 nm every 15 min.Bacterial cell sizes under four different cultivation conditions were evaluated using CLSM imaging.The four cultivation conditions include: (I) liquid culture in a 37°C shaker; (II) liquid culture in a 42°C shaker; (III) liquid culture in an anaerobic bag in a 37°C incubator; and (IV) liquid culture in an aerobic 37°C incubator.ImageJ (https://imagej.nih.gov/ij/) was used to measure the cell size of bacteria from CLSM images.

Motility assays
Motility assays were carried out as previously described with minor modifications 46,47 .Twitching motility was conducted by stabbing an isolate to the bottom of the LB plate (1.2% agar).Swimming motility was performed by stabbing an isolate on the surface of the LB plate (0.3% agar).After 24 h of incubation at 37°C, motility zones were measured.At least three replicates were measured for each sample.

Congo red assay
To assay colony morphology, Congo red plates (10 g/L tryptone, 150 μg/ml Congo red, 1% agar) were used.Overnight strain cultures were adjusted to a concentration of OD 600 ~0.01 in PBS (phosphate-buffered saline) buffer, and 5 μL of the culture was pipetted on plates.The plates were incubated at room temperature.

Bioreporter quantification of c-di-GMP and PQS
The c-di-GMP and PQS levels in PAO1 wild-type and its mutants were measured by the fluorescent intensity of a green-fluorescent-protein tagged reporter PcdrA-gfp 48 and PpqsA-gfp 49 , respectively.The fluorescent intensity was recorded by a Tecan Spark microplate reader with an excitation wavelength of 485 nm and an emission wavelength of 535 nm.
Pyoverdine and pyocyanin assays PAO1, its mutants, and complementation strains were incubated for 24 h in LB broth.The pyoverdine production was quantified by reading the absorbance of the supernatant at 405 nm and then normalized by OD 600 of the culture (Tecan Spark) 34,50 .For phenazine pigment-pyocyanin assay, cultures were adjusted to a concentration of OD 600 ~0.1 in PBS buffer, and then, 100 μL of culture was spread on PIA plates (Pseudomonas Isolation Agar; BD Biosciences).Pyocyanin was extracted with chloroform and 0.5 M HCl and the absorbance was measured at 520 nm (Tecan Spark) as previously described 51 .

LC-MS/MS analysis of HHQ and PQS
One milliliter of culture fluid was centrifuged at 12,000 rpm for 2 min; the supernatant was filtered through a 0.2 μm Millex Syring Filter and then mixed with an equal volume of methanol.The crude extract was subjected to LC-MS/MS as previously reported 52,53 .All MS experiments in this study were performed on an AB SCIEX QTRAP 4500 (Applied Biosystem).Analyst software was applied for data acquisition and processing.

Protein expression and purification
To express ZapE and K72A, the expression vector pET28a was used.The zapE and zapE K72A each were amplified from PAO1 genomic DNA and pUCP20-K72A with 28a-zapE-F and 28a-zapE-R.For PqsH, the expression vector pMAL-C5x was chosen, and the pqsH was amplified with MAL-pqsH-F and MAL-pqsH-R.Gibson Assembly Master Mix (NEB) was applied to fuse the amplicons with the expression vectors.The fusion gene constructs were then transformed into E. coli strain BL21 (DE3).HIS-ZapE and HIS-K72A were purified on Ni NTA beads (Smart Lifesciences) independently 19 , and MBP-PqsH was purified on Dextrin Beads 6FF (Smart Lifesciences) as previously reported 13 .The fusion proteins were further purified by ion exchange on a Mono Q 5/50 GL column (GE Biosciences) and then followed by a gel filtration on a HiLoad 16/60 Superdex 200 size exclusion column (GE Biosciences).

SPR analysis
The PqsH protein in 20 mM HEPES (pH 8.0) and 150 mM NaCl (HEPES buffered Steinberg's Solution, HBST buffer) was fixed on a CM 5 chip, and then, ZapE protein was serially diluted in HBST buffer.The binding affinities of PqsH and ZapE were measured by the single cycle mode of BIAcore T200 at 25°C.Biacore T200 software version 3.0 with a 1:1 Langmuir binding model was applied to analyze the binding kinetics of proteins.

IP-MS
Genomic loci of zapE without the stop codon and gfp were cloned into pHerd20T, and the gene locus of gfp was cloned into pHerd20T as a control.All primers used in this section are listed in Table S5.The resulting vectors p20T-zapE-gfp and p20T-gfp were transformed into the ΔzapE separately.The total proteins were incubated with GFP-trap Agarose (Chromotek) as previously described 54 .LC-MS/MS analysis was carried out by Wininnovate Bio Co. Ltd.

ATPase activity assay
A ultramicro ATPase activity assay kit for Na + /K + ATPase, Ca 2+ /Mg 2+ , and all types of ATPase (Nanjing Jiancheng) was used to test the activity of the purified ZapE and K72A in this study.The ATPase activity of samples is calculated by the formula B/(t × V) × D (B is the phosphate amount, t is the reaction time, V is the sample volume added into the reaction well, and D is the sample dilution factor).

Protein-protein docking
The 3D structure of ZapE was constructed by Alpha-Fold 39 .The protein-protein docking module in ClusPro 55 was used to predict the molecular docking of ZapE and PqsH (ID: AF-Q9I0Q0-F1-model_v2, AlphaFold Protein Structure Database) as previously reported 56 .Molecular operating environment (MOE) was chosen to predict the contacting residues of interacting proteins.

Statistical analysis
The results in this article are presented as the mean ± SD of at least three independent replicates.Student's unpaired t test was used to evaluate significance.***p < 0.001; **p < 0.01; *p < 0.05; ns, no significant.

Figure 1 .
Figure 1.Transposon-insertion site sequencing analysis of genetic determinants for Pseudomonas aeruginosa biofilm formation.(A) Correlation analysis of input and output samples.(B) Quality control analysis of two inputs from the same batch of experiments.Each point represents the number of unique gene reads (UGRs) of one specific mutagenized gene in the duplicate samples.(C) Quality control analysis of two outputs from different batches of experiments.(D) Dot plot showing the significantly reduced colonization of mutants during biofilm formation.Each dot represents the normalized UGRs of each gene in input and output.Fold change (FC) is colorized to show the differences.(E) Functional categories of essential genes during biofilm formation and maintenance.Genes were deemed as essential factors during P. aeruginosa biofilm formation, in the scenario that all mutants impart no fitness cost, when log 2 |FC (input/ output)| > 2 and p < 0.05.

Figure 7 .
Figure 7. Model of the ZapE regulation in the pqs quorum sensing system in P. aeruginosa.HHQ is synthesized by PqsABCD and converted to PQS by PqsH.ZapE is a cell division factor, which can directly interact with PqsH and positively regulate the synthesis of PQS.Both HHQ and PQS can bind to the transcriptional regulator PqsR; however, the affinity of PQS is approximately 100-fold more potent than HHQ at stimulating PqsR activity.Autoinduction occurs when either HHQ or PQS binds to PqsR, and then the expression of pqsA-E operon is activated.PqsE is a putative metallohydrolase protein, which positively regulates pyocyanin biosynthesis as well as biofilm formation.The biosynthesis of pyoverdine can be positively impacted by the biofilm and the iron chelator PQS.In brief, ZapE regulates P. aeruginosa biofilm formation by impacting the pqs quorum sensing.

Table 1 .
Hypothetical genes chosen in this study.

Table 2 .
List of identified ZapE-interacting proteins related to cell division, PQS, and phenazine synthesis.