Proteomic Profiling of Outer Membrane Vesicles Released by Escherichia coli LPS Mutants Defective in Heptose Biosynthesis

Escherichia coli releases outer membrane vesicles (OMVs) into the extracellular environment. OMVs, which contain the outer membrane protein, lipopolysaccharides (LPS), and genetic material, play an important role in immune response modulation. An isobaric tag for relative and absolute quantitation (iTRAQ) analysis was used to investigate OMV constituent proteins and their functions in burn trauma. OMV sizes ranged from 50 to 200 nm. Proteomics and Gene Ontology analysis revealed that ΔrfaC and ΔrfaG were likely involved in the upregulation of the structural constituent of ribosomes for the outer membrane and of proteins involved in protein binding and OMV synthesis. ΔrfaL was likely implicated in the downregulation of the structural constituent of the ribosome, translation, and cytosolic large ribosomal subunit. Kyoto Encyclopedia of Genes and Genomes analysis indicated that ΔrfaC and ΔrfaG downregulated ACP, ACEF, and ADHE genes; ΔrfaL upregulated ACP, ACEF, and ADHE genes. Heat map analysis demonstrated upregulation of galF, clpX, accA, fabB, and grpE and downregulation of pspA, ydiY, rpsT, and rpmB. These results suggest that RfaC, RfaG, and RfaL proteins were involved in outer membrane and LPS synthesis. Therefore, direct contact between wounds and LPS may lead to apoptosis, reduction in local cell proliferation, and delayed wound healing.


Introduction
Gram-negative bacteria release nanovesicles from their outer membrane into the extracellular milieu which are called outer membrane vesicles (OMVs) [1,2]. OMVs, which range from 20 to 200 nm in size, have a lipid-bilayered diameter and spherical proteolipids. They contain lipopolysaccharides (LPS), periplasmic proteins, outer membrane lipids, cytoplasmic proteins, DNA, RNA, outer membrane proteins, and other elements related to virulence [3][4][5][6][7]. OMVs play important roles in bacterial physiological and phthological aspects such as genetic material transfer, protein transport, nutrient acquisition, antibacterial activity, virulence factor delivery, interkingdom communication, immune response modulation, and neutralizing phage decoy activity [8][9][10][11][12][13][14]. As the major component of OMV, LPS contains lipid A, a core oligosaccharide, and the O-specific polysaccharide of the related to virulence [3][4][5][6][7]. OMVs play important roles in bacterial physiological and phthological aspects such as genetic material transfer, protein transport, nutrient acquisition, antibacterial activity, virulence factor delivery, interkingdom communication, immune response modulation, and neutralizing phage decoy activity [8][9][10][11][12][13][14]. As the major component of OMV, LPS contains lipid A, a core oligosaccharide, and the O-specific polysaccharide of the O-antigen (Figure 1b) [15]. LPS has been reported to be associated with infection. In burn injury and infection-related diseases, LPS of pathogens interact with membrane-bound or soluble CD14, lipopolysaccharide-binding protein (LBP), and Tolllike receptor 4 (TLR4) to initiate cellular production of pro-inflammatory cytokines, immune cell recruitment, and endotoxin clearance [15]. In Escherichia coli, the synthesis of LPS required rfa (also known as waa) operons that consist of many genes [16]. Three operons regulate the genes in the rfa locus. The first operon comprises rfaC (waaC), rfaD (or gmhD), rfaF (or waaF), and rfaL (or waaL) genes. Then, rfaB (or waaB), rfaG (or waaG), rfaI (or waaO), rfaJ (or warJ/waaJ), rfaK (or waaU), rfaP (or waaP), rfaQ (or waaQ), rfaS (or waaS), rfaY (or waaY), and rfaZ (or waaZ) are organized in the second operon. The remaining short kdtA operon consists of kdtA (or waaA) and kdtB (or coaD) [17][18][19][20]. In the LPS core biosynthesis pathway, rfaC has been proposed to play a crucial role in transferring heptose to the LPS core.  The destruction of the LPS structure would influence the microbiological features of the bacteria and result in the alternation of the OMVs they released. Nakao et al. have reported that mutation in rfaC produced defective LPS in OMV but still maintained membrane integrity. E. coli with rfaC mutation generated OMV comparable to the wild type but produced more extracellular DNA (eDNA) in the culture, which is involved in initial attachment and biofilm formation as well as enhancing cell wall hydrophobicity [21,22]. Upon formation in the wound site, biofilms can have harmful effects including impaired epithelialization, granulation tissue formation, and reduced inflammatory response, which delays the healing process [23,24]. Several studies also mentioned that mutations in the rfa locus lead to severely truncated LPS and result in elevated antibiotic resistance, multidrug resistance, and resistance to several bacteriophages used in therapy for bacterial infec-tions [25]. In burn injury patients, this phenomenon delays the wound healing process [26]. Another study reported that glycosyltransferase activity and O-antigen attachment to lipid A in E. coli was affected by mutations in the rfaL gene, thereby reducing biofilm formation [27]. Owing to the highly conserved inner-core composition of Gram-negative bacteria, the rfa gene may be a possible therapeutic target against infection.
Proteomic analysis of the entire set of proteins has been used to specify the composition of OMVs in several studies [1,15]. However, these studies remain limited, and only a few proteins have been recognized [6,7]. The main objective of this study was to elucidate the effect of LPS structure on OMV composition using the proteomics analysis. Based on the characteristics of the OMV genes above, we profiled rfaC, rfaG, and rfaL using quantitative proteomics to investigate E. coli outer membrane genes, and interpreted the effect of LPS structure on OMV composition. The novelty of this study is to provide proteomic data of the truncated RfaC, RfaG, and RfaL proteins in E. coli to support claims in previous publications that Rfa proteins could be a target for the treatment of the infected wounds of patients. An isobaric tag for relative and absolute quantitation (iTRAQ)-based proteomic analysis technique was employed to identify the expressed proteins between OMVs released by E. coli BW25113 and its mutant strains, including the rfaC-, rfaG-, and rfaL-defect strains. The functional classification of proteins and key pathways were analyzed by utilizing the Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, respectively. Our results provide essential information regarding the mechanism of the expressed genes and key pathways among proteins and E. coli strains.

Bacterial Strains, Cultures, and Growth Conditions
E. coli strain BW25113 and its mutants with deleted rfaC, rfaG, and rfaL genes were used ( Figure 1, Table 1). All strains were purchased from Horizon Discovery (Cambridge, UK). The E. coli BW25113 strain is a K-12 derivative with a recA+ and hsdR genotype and the parent of the Keio collection of single-gene knockouts which has >100-fold higher transformation efficiency than the commonly used cloning E. coli hosts. The rfa genes in the mutant strains were deleted by using a one-step gene deletion method with the phage lambda Red recombinase. Bacteria were maintained and grown on agar at 37 • C with aeration.

OMV Sample Preparation
OMVs were isolated from the late log-phase (16 h) culture of E. coli. In brief, 200 mL LB was inoculated with 2 mL overnight culture medium, incubated at 37 • C, and agitated by shaking at 200 rpm for 6 h. The cells were pelleted by centrifugation (10,000 rpm for 15 min) and the supernatant was filtered through a 0.22 µm membrane filter (JET BIOFIL, Guangzhou, China) to remove cells and cellular debris. The filtrate was subjected to ultracentrifugation (35,000 rpm) for 2 h at 4 • C using a Type 45 Ti rotor (Beckman, CA, USA). For washing the OMVs, the pellet was suspended in phosphate-buffered saline (PBS) and then ultracentrifuged (42,000 rpm) for 2 h at 4 • C using the same rotor. The pellet was finally resuspended in PBS and stored on ice. Three biological replicates of each experiment were performed in the following analysis.

Transmission Electron Microscopy (TEM)
Vesicles that were obtained were placed on a glow-discharged carbon-coated EM grid and allowed to rest briefly. They were then gently washed with deionized water twice and the grids were stained with 1% uranyl formate and dried at room temperature overnight. The grids were then observed using a transmission electron microscope (FEI Tecnai G2 F20 S-TWIN, FEI Company, Hillsboro, OR, USA) at 120 kV. TEM figures were visualized using the ImageJ software (National Institutes of Health, Bethesda, MD, USA) and analyzed using Student's t-test. Diagrams were generated using the GraphPad Prism software (version 8.0, GraphPad software, San Diego, CA, USA) [28,29].

Nanoparticle Tracking Analysis (NTA) of OMVs
The quantities and sizes of OMVs were measured using the NanoSight NS500 nanoparticle tracking system (NTA) (Malvern Instruments Ltd., Malvern, Worcestershire, UK) supplied with a 488 nm blue laser and camera with a complementary metal-oxide semiconductor (CMOS) image sensor. Frozen OMV samples were thawed at room temperature and diluted 1/1000 in 50 mM HEPES buffer at pH 7.4 before analysis. Polystyrene beads (100 nm diameter) were used as the positive control and HEPES buffer alone served as the negative control. Samples were pumped into a NanoSight instrument using a syringe and set at the '20 speed setting (in arbitrary units) on the NS500. The quantification was captured in five 60 s reads at room temperature (at approximately 23.9-25.2 • C), and the instrument was optimized at automatic setting (for 'blur,' minimum track length,' and 'minimum expected size' setting), whereas viscosity was set to 'water' (0.883-0.911 cP). Camera level and focus for automated image setup were chosen for video enhancement, and the NTA software version 2.3 (Malvern Instruments Ltd., Malvern, Worcestershire, UK) was used to determine a total of 1498 frames per sample with a threshold of 5 (in arbitrary units). The data, including mean size (nm), mode size (nm), and concentration (particles/mL), were arranged, and an average of five reads was measured and plotted as particle size versus number of particles per Ml using Origin85 and the average size of OMVs was calculated using GraphPad Prism software [30].

Protein Extraction
The E. coli cultures were incubated at 37 • C for 16 h to acquire the total protein of OMVs. The cultures were centrifugated at 5000× g at 4 • C for 10 min, followed by protein extraction as previously described [24]. Trypsin digestion was then performed for 100 g of total cellular protein from each sample using Trypsin Gold (Promega, Madison, WI, USA) at a mass ratio of 30:1 trypsin-to-protein extract. Reactions were incubated at 37 • C for 16 h. Afterward, the digested proteins were dried using vacuum centrifugation [31].

iTRAQ Labeling and SCX Fractionation
First, 0.5 M triethylammonium bicarbonate (TEAB) was used to dissolve the dried peptide samples. Then, the peptide samples from E. coli cultures were labeled with iTRAQ reagents 114 and 116, respectively. The labeled proteins were incubated at room temperature for 2 h, then collected and dried using vacuum centrifugation. The protein labeled by iTRAQ was dissolved in 4 mL buffer A (25 mM NaH 2 PO 4 in 25% ACN, pH 2.7) and separated using an LC-20AB HPLC pump system (Shimadzu, Kyoto, Japan) with Ultramex SCX column (4.6 mm × 250 mm, 5 µm, Phenomenex, Torrance, CA, USA). The sample was eluted by the linear gradient of buffer A for 10 min, 50-60% buffer B (25 mM NaH 2 PO 4 and 1 M KCl in 25% ACN, pH 2.7) for 27 min, and the 60-100% buffer B for 1 min. Subsequently, the peptide was eluted every minute to 20 fractions at a flow rate of 1 mL/min and an absorbance wavelength of 214 nm. Finally, the protein fractions were desalted using Strata X C18 (Phenomenex, Torrance, CA, USA) and dried using vacuum centrifugation [32].

LC-MS/MS Analysis
The labeled protein fractions that were diluted in 40 µL 0.1% (v/v) trifluoroacetic acid were introduced to a nanoLC-MS/MS for analysis (Q Exactive mass spectrometer, Thermo Fisher Scientific, Waltham, MA, USA), performed in positive ion mode coupled with Easy nLC (Thermo Fisher Scientific, Waltham, MA, USA) for 60 min. MS data were obtained using a data-independent top 10 method. Furthermore, automatic gain control (ACG) target was set to 1e6 (1e6 = 1,000,000), the maximum inject time to 50 ms, and the duration of Dynamic exclusion was 60 s. Survey scans were obtained at a resolution of 70,000 and m/z 200, and isolation width was 2 m/z. Normalized collision energy was 30 eV, and underfill ratio, which determines the minimum percentage of the target value possible to be gained at the maximum fill time, was determined as 0.1%. The instrument was performed with peptide identification mode permitted [33].

Proteomic and Bioinformatic Analysis
Protein characterization was conducted using Mascot ® (version 2.2; Matrix Science, MA, USA) and the Proteome Discoverer™ software (version 1.4; Thermo Scientific, Waltham, MA, USA) using the sequences from the UniProt Human Database (133,549 sequences, downloaded on 3 March 2013). In this method, the parameters used were mass tolerance = 20 ppm, MS/MS tolerance = 0.1 Da, enzyme = trypsin, missed cleavage-2, oxidation (M), iTRAQ 8plex (Y) as the possible variable modifications, and carbomidomethyl (C), iTRAQ 8plex (N-term), iTRAQ 8plex (K) as the permanent modifications. The calculation for false discovery rate (FDR) of peptide characterization used a bait database search at a filtering basis of FDR ≤0.01. The iTRAQ ratio between the two groups of >1.2 or <0.83 defined the differential protein expression, and all of the diversely expressed RfaC, RfaG, and RfaL proteins were examined using UniProt (http://www.uniprot.org/; accessed on 28 December 2021) [34].
A Venn diagram was generated to elaborate the general diversely expressed proteins between the OMVs released by E. coli wild-type (BW25113) and its mutant strains. The cross-comparison of the gene names generated Venn diagrams and sets of gene lists are shown in Table 2. GO analysis (version go_201608.obo; www.geneontology.org; accessed on 28 December 2021) was used to examine the biological importance of the distinct expressed proteins. Furthermore, the distinct expressed proteins were entangled in the identical process; function and components were distributed into corresponding clusters. KEGG pathway analysis was carried out to explore the potential of the biological pathways using the online software (KEGG Automatic Annotation Server (KAAAS)) [33].

Statistical Analysis
All data were statistically analyzed using IBM ® SPSS ® Software Version 18.0 (IBM Corp., Armonk, NY, USA) and expressed as mean ± standard error. Data were analyzed using one-way analysis of variance or a two-tailed paired t-test. Significant differences between groups were detected using * p < 0.05 and ** p < 0.01 to indicate statistical significance.

TEM Analysis of OMVs Released from E. Coli BW25113 and Mutants
Generally, in Figure 2a-d, OMVs were not uniform in size. High magnification TEM images of OMVs showed spherical particles. However, the overall sample quantification data showed that the size of the OMVs released by the ∆rfaG strain was significantly larger than that of E. coli BW25113, whereas OMVs released by the ∆rfaC and ∆rfaL strains were significantly smaller than that of E. coli BW25113.

Statistical Analysis
All data were statistically analyzed using IBM ® SPSS ® Software Version 18.0 (IBM Corp., Armonk, NY, USA) and expressed as mean ± standard error. Data were analyzed using one-way analysis of variance or a two-tailed paired t-test. Significant differences between groups were detected using * p < 0.05 and ** p < 0.01 to indicate statistical significance.

TEM Analysis of OMVs Released from E. Coli BW25113 and Mutants
Generally, in Figure 2a-d, OMVs were not uniform in size. High magnification TEM images of OMVs showed spherical particles. However, the overall sample quantification data showed that the size of the OMVs released by the ΔrfaG strain was significantly larger than that of E. coli BW25113, whereas OMVs released by the ΔrfaC and ΔrfaL strains were significantly smaller than that of E. coli BW25113.

NTA Analysis of OMVs from WT and Mutant E. Coli
To further determine the distribution of OMV size released by the E. coli BW25113 and its mutant strains, we analyzed the heterogenous population of OMVs using NTA. The results revealed that the size distribution of OMVs in each group ranged from 50 to 200 nm in diameter, but the majority were 100 nm (Figure 3a). Furthermore, the histogram data showed that OMVs released by the ΔrfaC and ΔrfaG strains were significantly larger in diameter compared with those of the ΔrfaL strain (Figure 3b).

NTA Analysis of OMVs from WT and Mutant E. Coli
To further determine the distribution of OMV size released by the E. coli BW25113 and its mutant strains, we analyzed the heterogenous population of OMVs using NTA. The results revealed that the size distribution of OMVs in each group ranged from 50 to 200 nm in diameter, but the majority were 100 nm (Figure 3a). Furthermore, the histogram data showed that OMVs released by the ∆rfaC and ∆rfaG strains were significantly larger in diameter compared with those of the ∆rfaL strain (Figure 3b).

Heat Map Analysis of Mutant Strains Compared to E. Coli BW25113 Strain
Heat map analysis showed that GalF, ClpX, AccA, FabB, and GrpE proteins were upregulated among the OMVs released by the mutant strains compared with the E. coli BW25113 ( Figure 5). In contrast, PspA, YdiY, RpsT, and RpmB proteins were found to be downregulated in mutant strains compared with E. coli BW25113.

GO Bar Chart of Mutants Versus E. Coli BW25113 strains
GO classification revealed that protein binding was the most shared in the OMVs of ∆rfaC and that of BW25113 strain, whereas structural constituent of ribosome and rRNA binding were the least shared in molecular function (MF). Cytosol and translation clusters of the ∆rfaC strain were downregulated in the cellular component (CC) and biological process (BP), respectively (Figure 6a). Between ∆rfaG and BW25113 strains, protein binding and structural constituent of ribosome were upregulated in MF, and cytosol was abundant in CC ( Figure 6b). Additionally, for the ∆rfaL strain versus wild-type BW25113, cytosol and identical protein binding were the most prevalent in CC and MF, respectively. In addition, the structural constituent of ribosome, translation, and cytosolic large ribosomal subunit were strongly downregulated between ∆rfaL and BW25113 strains (Figure 6c).

Discussion
Vesicles obtained from the outer membrane of Gram-negative bacteria are known as OMVs and are highly diverse in size, composition, and function [35]. In this study, TEM was used to determine the appearance and size of OMVs released by E. coli BW25113 and rfaC, rfaG, and rfaL knockout strains (Figure 2a-d). Moreover, using NTA, the diameter of

Discussion
Vesicles obtained from the outer membrane of Gram-negative bacteria are known as OMVs and are highly diverse in size, composition, and function [35]. In this study, TEM was used to determine the appearance and size of OMVs released by E. coli BW25113 and rfaC, rfaG, and rfaL knockout strains (Figure 2a-d). Moreover, using NTA, the diameter of OMVs in all strains was shown to range from 50-200 nm (Figure 2a), with the highest mostly found in E. coli lacking rfaC, and rfaG, significantly different from OMVs released by E. coli BW25113 and E. coli lacking rfaL (Figure 2b). In addition, our NTA results showed that OMVs released by E. coli BW25113 and its mutant strains ranged between 50-200 nm in diameter, with the majority being 100 nm (Figure 3a). Additionally, the OMVs of ∆rfaC and ∆rfaG were found to be larger than those of BW25113 and ∆rfaL (Figure 3b). These results indicate that secreted OMVs have diverse functions. Turner et al. mentioned that OMV size is dependent on protein content and composition [30]. The smaller OMV fraction contained less protein than the larger OMV fraction and the heterogenous population of OMVs. Generally, the larger OMVs contained more adhesion protein for virulence, whereas small OMVs contained protein predominantly for metabolism. The size of OMVs also determined the mechanism and efficiency of cellular entry of the OMV into the host cell. Smaller OMVs are known to enter the cell through the micropinocytosis pathway, whereas larger OMVs enter the cell preferentially via the clathrin and dynamin-mediated pathway. Inhibition of clathrin-mediated endocytosis has no effect on enterotoxigenicity of E. coli OMVs [30,36].
Furthermore, in the proteomics study, the genes rpsT, galF, and rpmB were identified as shared among cluster mutants and E. coli BW25113 (Figure 4). Our results revealed that rpsT, which defines the 30S ribosomal protein S20 OS, and 50S ribosomal protein L28 OS that is encoded by the rpmB gene, were downregulated in both mutant and E. coli BW25113 strains. On the other hand, galF upregulated UTP-glucose-1-phosphate uridylyl transferase OS in E. coli BW25113 mutants. These proteomic data were consistent with the heat map analysis results. GalF, ClpX, AccA, FabB, and GrpE appeared to be upregulated among all OMV proteins in the E. coli BW25113 and mutant strains ( Figure 5). This result likely suggests that rfaC, rfaG, and rfaL are not directly responsible for fatty acid biosynthesis and elongation together with AccA and FabB, respectively [37,38]. Furthermore, GalF proteins for cellular UDP-glucose formation and GrpE for protein folding and thermos-sensing demonstrate a high protein expression following OMV protein knockout (∆rfaC, ∆rfaG, ∆rfaL) in mutant and E. coli BW25113 strains, indicating that UDP-glucose formation and heat shock response genes continue to function well without the presence of OMV proteins [39,40]. However, the downregulation of stress-related proteins such as PspA and YdiY [41,42] is likely attributable to their relationship with OMV protein construction, and the suppression of the aforementioned gene transcripts or protein expression by OMV protein knockout. Moreover, upregulation of the ClpX protein and downregulation of ribosomal proteins such as RpsT and RpmB are suggested to occur because of the involvement of these OMV genes in ribosomal 30S and 50S synthesis, respectively [43,44].
The results of GO exhibited that the binding protein is the most widely shared protein in ∆rfaC and E. coli BW25113 strains, whereas the structural constituent of the ribosome and rRNA binding are the least conserved. Moreover, cytosol and translation clusters of ∆rfaC in CC and BP, respectively are downregulated (Figure 6a). These data are reasonable considering the importance of OMVs in supporting the unique architectures corresponding to protein transport, genetic material transfer, interkingdom communication, antibacterial activity, neutralizing phage decoy activity, virulence factor delivery, and immune response modulation that involve binding activity [8][9][10][11][12][13][14]. Furthermore, the expression of protein binding and structural constituent of ribosomes is upregulated in MF and cytosol was abundant in CC between ∆rfaG and E. coli BW25113 strains (Figure 6b). These results indicate that rfaG is not directly involved in outer membrane and OMV synthesis, and possibly acts in other pathways owing to its glycosyltransferase activity [45]. However, the ∆rfaL strain versus E. coli BW25113 strain data showed substantial downregulation of the structural constituent of the ribosome, translation, and cytosolic large ribosomal subunit (Figure 6c). RfaL is known to encode a component of the O ligase, which transfers the completed O-antigen from the ACL to the core of a suitable LPS acceptor. Furthermore, it is involved in core modification, and plays a significant role in producing core heterogeneity, as well as O-antigen attachment [46]. The O-antigen is important for Gram-negative bacteria as it functions in targeting both the innate and adaptive immune systems in pathogenicity [47].
KEGG was also used to predict the pathway of glycolysis/glucogenesis since rfaC, rfaG, and rfaL were involved in LPS core glycosylation. The data revealed that ∆rfaC and ∆rfaG downregulate ACP, ACEF, and ADHE genes (Figure 7a). These results were likely attributable to the roles of ACP, ACEF, and ADHE in the transfer of acyl fatty acid during phospholipid synthesis, pyruvate dehydrogenase, and aldehyde dehydrogenase, respectively [48][49][50]. These proteins are possibly involved in LPS construction, which is implicated in the regulation of rfaC, rfaG, and rfaL genes. While the misfolded protein produced by a truncated set of genes will be retained in the ER and processed for ERrelated degradation [51], the deletion of rfa genes may lead to the further interference of the membrane's subsequent function [50]. Moreover, glycosylation deficiencies in the LPS protein production also disrupt protein localization in ER, which affects the functional membrane of protein [52]. In addition, the KEGG prediction showed high expression of PFKA, GAPA, and PYKA in E. coli by rfaL gene knockout (Figure 7b). PFKA, GAPA, and PYKA are involved in fructose phosphorylation, pyruvate formation, and GAPDH production (glycolysis pathway), respectively [53][54][55]. From these results, it is suggested that rfaL gene is not involved in energy production because of its role in E. coli LPS synthesis.
Regarding the involvement of rfaC, rfaG, and rfaL genes in E. coli LPS synthesis, a recent study mentioned that the small amount of clinically relevant Gram-negative human pathogen bacterial inoculum may cause bacteremia and eventually lead to death. E. coli infection in burn wound has been shown to lead to bacteremia at 24 to 48 h and death after 3 to 4 days [56]. This finding was also supported by Crompton et al., who found that in the early phase of healing, wounds treated by LPS exhibited apoptosis and reduction in local proliferation, thereby showing that contact between LPS and the wound delayed the healing process [57]. In addition, some previous studies mentioned links between glycolysis and immune system response. OMV LPS was capable of inducing macrophage metabolism shift from oxidative phosphorylation (OXPHOS) to glycolysis, which was supported by decreased mitochondrial oxygen consumption and reduced respiration activity, as well as increased mitochondrial reactive oxygen species production [58]. Thus, LPS-induced glycolysis at the wound site heightened infection by inhibiting dendritic cell maturation and inducing inflammasome activation [59,60].
Considering bacterial involvement in infection, biofilm formation could lead to further wound chronicity and delayed healing [61,62]. A previous study reported that absence of the genes responsible for generating OMV LPS, such as rfaC, leads to an increase in biofilm production and bacterial pathogenicity [21]. Biofilm formation may inhibit wound healing owing to a 10-fold increase in interleukin-1β (IL-1β), interleukin-6 (IL-6), and matrix metalloprotease-10 (MMP-10) expression, indicative of persistent inflammatory response and delayed healing process [63,64]. Furthermore, chronic wound biofilms can be highly tolerant and resistant to antibiotics owing to the formation of a shield to protect bacteria from the phagocytic activity of invading polymorphonuclear neutrophils (PMNs). rfaG gene mutation also resulted in OMV destabilization due to reduced core phosphorylation and LPS length [65,66]. This mutation also leads to a L-glycero-α-D-manno heptose (heptose) deficit that impairs the growth of E. coli [67]. Furthermore, due to the function of the rfaL gene in O-antigen biosynthesis, mutation of rfaL gene interferes with the transfer process of polymerized O-antigen to the lipid A core to form LPS [68]. Moreover, Oglycosylation-defective Gram-negative bacteria did not show any growth deficiency, but a tremendously diminished capacity to generate biofilm, thereby reducing its resistance against antibiotics, and affecting its survival ability in the host [69]. Hence, by deleting rfa genes in E. coli, we identified their association with OMV LPS biosynthesis, O-antigen glycosylation and biofilm formation ability. These findings can provide excellent targets for the identification of rfa gene inhibitors for the development of new antibiotics to enhance the wound healing process.

Conclusions
This study concludes that the OMV defects of constituent proteins RfaC, RfaG, and RfaL were upregulated, and downregulated important cellular proteins, and that they may play important roles in the glycolysis/glucogenesis pathway in E. coli. Hence, all findings in this study highlight that rfaG, rfaC, and rfaL genes are responsible for LPS synthesis as the component of OMV. Our findings also suggest that OMV contact with the injury area may inhibit the wound healing process by hindering the inflammatory response at the wound site.