Immuno-Stimulatory Activity of Escherichia coli Mutants Producing Kdo2-Monophosphoryl-Lipid A or Kdo2-Pentaacyl-Monophosphoryl-Lipid A

Lipid A is the active center of lipopolysaccharide which also known as endotoxin. Monophosphoryl-lipid A (MPLA) has less toxicity but retains potent immunoadjuvant activity; therefore, it can be developed as adjuvant for improving the strength and duration of the immune response to antigens. However, MPLA cannot be chemically synthesized and can only be obtained by hydrolyzing lipopolysaccharide (LPS) purified from Gram-negative bacteria. Purifying LPS is difficult and time-consuming and can damage the structure of MPLA. In this study, Escherichia coli mutant strains HWB01 and HWB02 were constructed by deleting several genes and integrating Francisella novicida gene lpxE into the chromosome of E. coli wild type strain W3110. Compared with W3110, HWB01 and HWB02 synthesized very short LPS, Kdo2-monophosphoryl-lipid A (Kdo2-MPLA) and Kdo2-pentaacyl-monophosphoryl-lipid A (Kdo2-pentaacyl-MPLA), respectively. Structural changes of LPS in the outer membranes of HWB01 and HWB02 increased their membrane permeability, surface hydrophobicity, auto-aggregation ability and sensitivity to some antibiotics, but the abilities of these strains to activate the TLR4/MD-2 receptor of HKE-Blue hTLR4 cells were deceased. Importantly, purified Kdo2-MPLA and Kdo2-pentaacyl-MPLA differed from wild type LPS in their ability to stimulate the mammalian cell lines THP-1 and RAW264.7. The purification of Kdo2-MPLA and Kdo2-pentaacyl-MPLA from HWB01 and HWB02, respectively, is much easier than the purification of LPS from W3110, and these lipid A derivatives could be important tools for developing future vaccine adjuvants.


Introduction
Lipopolysaccharide (LPS), the major component of the outer layer of outer membrane in most Gram-negative bacteria, is also known as endotoxin [1,2] and is important for membrane stability [3]. LPS of wild type Escherichia coli usually consists of Kdo 2 -lipid A and a polysaccharide

Materials and Methods
Construction of E. coli mutant strains Table 1 lists all of the strains and plasmids used in this work. E. coli strains without plasmids were grown in LB medium (10 g/L tryptone, 5 g/L yeast extract and 10 g/L NaCl) at 37°C, and E. coli strains harboring plasmids pCP20 or pKD46 were grown at 30°C. When necessary, 30 μg/mL kanamycin or 100 μg/mL ampicillin was included in the medium.
E. coli W3110 is a nonpathogenic wild type K-12 strain and has been widely used in laboratory for gene engineering. Strain HW001 was constructed from W3110 by integrating F. novicida lpxE into lacZ in the chromosome, and HW002 was constructed from HW001 by deleting lpxM from the chromosome [19]. Because lpxE was integrated in the lacZ site, lacI was also removed in HW001 and HW002 to release the control to the lactose operon. E. coli strain WBB06 was derived from W3110 by inserting tet6 into the region of waaC and waaF; this strain can only synthesize Kdo 2 -lipid A [26]. HWB01 was constructed from HW001 by deleting waaC and waaF from the chromosome, and HWB02 was constructed from HW002 by deleting waaC and waaF from the chromosome (Fig 1). Briefly, the upstream fragment of waaC was PCR-amplified using primers WaaC-U-F (5'-CCGCTCGAGTAAATCAAGCAAGCC TAT-3') and WaaC-U-R (5'-AAAACTGCAGCTGCTTGCCCTGTATGGT-3'), and the downstream fragment of waaF was PCR amplified using primers WaaF-D-F (5'-CCCAAGCTT AGCTCTTATGCGTCGCGATTCAG-3') and WaaF-D-R (5'-AAAACTGCAGTGCTACGCTG GCTTATC-3'). A DNA fragment containing the kanamycin resistance gene, FRT-kan-FRT, was amplified from pKD13 using primers kan-FRT-F (5'-AACTGCAGGTGTAGGCTGGAGC TGCTTCG-3') and kan-FRT-R (5'-CCCAAGCTTACCTGCAGTTCGAAGTTCCT-3'). These three fragments were then ligated together to form plasmid pBS-CFkan carrying the knockout fragment waaCU-FRT-kan-FRT-waaFD. Genes waaC and waaF in the chromosomes of HW001 or HW002 were removed using Red recombination. Plasmid pKD46 was first transformed into HW001 or HW002, and then the knockout fragment waaCU-FRT-kan-FRT-waaFD was amplified and transformed into the cells. The waaC-waaF locus was deleted from the chromosome through the recombination catalyzed by Red enzymes expressed by pKD46. The correct transformants were selected by growing cells on LB plates containing kanamycin, and plasmid pKD46 was cured by growing the cells at 42°C. The mutagenesis frequencies were around 12%. Next, plasmid pCP20 was transformed into the cells, and FLP recombinase was expressed to remove the kan gene inserted in the chromosome. Then, plasmid pCP20 was also cured by growing cells at 42°C, resulting in strains HWB01 and HWB02 [27,28]. There were no selection markers left on the chromosomes, and thus they can grow in medium without the addition of antibiotics.
Isolation and analysis of crude Kdo 2 -lipid A, Kdo 2 -MPLA and Kdo 2 -P-MPLA from different E. coli strains Lipids were isolated from 400 mL cultures of E. coli WBB06, HWB01, HWB02 cells at an OD 600 of 1.5 using the Bligh-Dyer method [29]. Cells were harvested, washed once with phosphate-buffered saline (PBS) and stirred in a 76 mL-single-phase mixture of chloroform, methanol and water (1:2:0.8, v/v/v) for 1 h at room temperature. Lipids including Kdo 2 -lipid A or its derivatives were extracted from the cells. The supernatant was collected and converted into a two-phase Bligh-Dyer mixture of chloroform, methanol and water (2:2:1.8, v/v/v) by adding 20 mL chloroform and 20 mL water. Crude Kdo 2 -lipid A, Kdo 2 -MPLA or Kdo 2 -P-MPLA was isolated from the lower phase, dried with a rotary evaporator and stored at −20°C. For TLC analysis, crude Kdo 2 -lipid A, Kdo 2 -MPLA or Kdo 2 -P-MPLA were dissolved in a mixture of chloroform and methanol (2:1, v/v), applied on silica gel 60 TLC plates, and separated in a glass chamber containing a solvent of chloroform, methanol, acetic acid and water (25:15:4:4, v/v/v/v). Then, the plates were dried, sprayed with 10% sulfuric acid in ethanol, and charred at 175°C to visualize the lipid bands [19,30,31]. The Rf values for the lipid bands on each TLC were determined.
For ESI/MS analysis, crude Kdo 2 -lipid A, Kdo 2 -MPLA or Kdo 2 -P-MPLA samples were dissolved in a mixture of chloroform and methanol (2:1, v/v) and analyzed using a Waters SYNAPT Q-TOF mass spectrometer equipped with an ESI source (Waters Corp., Milford, MA, USA) in the negative-ion mode [19,23]. Sodium formate solution was used for calibration. ESI/MS was carried out at -100 V, and the collisional activation of ions was performed at -6 V. Data acquisition and analysis were performed using MassLynx V4.1 software.
To evaluate membrane permeability, overnight cultures were harvested, washed twice with PBS, and resuspended in PBS to an OD 600 of 0.5. The, 1.92 mL of the cell suspension was mixed with 80 μL of 1 mM N-phenylnaphthylamine (NPN, Sigma-Aldrich), and the fluorescence was immediately measured by a spectrofluorometer (Hitachi, Tokyo, Japan). The excitation wavelength, emission wavelength and slits used were 420 nm, 350 nm, and 5 nm, respectively. The permeability was indicated by the fluorescence absorption per OD 600 value of the sample [33].
To measure the surface hydrophobicity, overnight cultures were harvested, washed twice with PBS (pH 7.4), and resuspended in PBS to an OD 600 of approximately 1.0, which was recorded as A 0 . Then, 2 mL of the suspension was mixed with 800 μL of xylene and incubated for 3 h at 4°C. Then, the OD 600 of the aqueous phase was measured and recorded as A. The value of [(A 0 -A)/A 0 ]×100 represents the hydrophobicity.
For evaluate the auto-aggregation abilities of the tested strains, overnight cultures were harvested, resuspended in fresh liquid LB and adjusted to an OD 600 of approximately 2.0, which was recorded as A 0 . Then, 12 mL of this suspension was added to a test tube and incubated at 22°C. After 24 h, the OD 600 of the supernatant was monitored and recorded as A i . The value of [(A 0 -A i )/A 0 ]×100 represents the auto-aggregation ability.

Measurement of minimum inhibitory concentration of E. coli strains to antibiotics
Minimum inhibitory concentration (MIC) is defined as the concentration of antibiotics that significantly reduced the metabolic activity and inhibited visible cell growth (OD 600 0.15). MIC of novobiocin, erythromycin, and clarithromycin for strains W3110, HW001, HW002, WBB06, HWB01 and HWB02 was determined in sterile 96-well plates using microdilution method [32,34,35]. The mixture was composed of 100 μL bacterial suspension diluted in LB (OD 600 around 0.02) and 100 μL antibiotics with different concentrations (0.98, 1.95, 3.91, 7.81, 15.6, 31.2, 62.5, 125, 250, 500, 1000 μg/mL), using LB as blank control. After incubation at 37°C for 24 h, the MIC and colony forming unit (CFU) values of bacterial cultures before and after treatment with antibiotics were determined. The experiments were performed twice in triplicate.

Purification of different structures of Kdo 2 -MPLA and LPS for cell stimulation
About 32.8, 34.9, 29.7, 18.6, 20.3 and 19.5 mg purified LPS samples were obtained from WBB06, HWB01, HWB02, W3110, HW001 and HW002 cells, respectively. The purified samples were dissolved in PBS to the concentration of 1 mg/mL. To confirm the purity, DNA, RNA and proteins in the samples were measured by using NanoDrop 2000 Spectrophotometer (Thermo Scientific) [40,41], and 1 mg/mL of standard LPS (Sigma-Aldrich, Prod. No. L2630) was used as control.
For this assay, 200 μL of RAW264.7 cells (1×10 5 cells/mL) were prepared by incubating them in 96-well plates for 24 h, then the medium was replaced with fresh medium containing sonically dispersed purified LPS ligands diluted from the 1mg/mL stock suspensions to concentrations of 0.1, 1, 10, 100 ng/mL. The THP-1 cells (200 μL, 1×10 5 cells/ml) were prepared by adding phorbol 12-myristate 13-acetate (PMA, Sigma-Aldrich) to differentiate the cells so that they could adhere to 96-well plastic plates [44]. After 48 h incubation, the medium was replaced with fresh medium containing sonically dispersed LPS ligands. After 24 h incubation in medium containing LPS ligands, the supernatants of both cell cultures were collected and stored at -80°C. Growth medium without any LPS ligands was used as blank control.

Statistical analysis
All statistical analyses were performed using GraphPad Prism 5.0 software. Student's t-test was used to analyze the difference between wild-type W3110 and E. coli mutants, and Ã p <0.05 and ÃÃ p<0.01 were be considered as significant and statistically significant.

Results
Construction of E. coli mutants HWB01 and HWB02 that produce Kdo 2 -MPLA and Kdo 2 -pentaacyl-MPLA, respectively Mutant strains HWB01 and HWB02 were constructed by deletion and integration of certain key genes related to LPS biosynthesis in the chromosome of E. coli W3110. HWB01 was constructed by integrating the F. novicida lpxE gene and deleting lacI, waaC, and waaF in the chromosome of wild type E. coli W3110 [27,28]; HWB02 was constructed by further deletion of lpxM in the chromosome of HWB01 [22]. The gene lpxE is found in F. novicida but not in E. coli [26], and the phosphatase LpxE that is encoded by lpxE can efficiently remove the 1-phosphate of lipid A in E. coli. Genes waaC and waaF encode heptosyltransferase WaaC and WaaF, respectively, and sequentially add two L-D-heptoses to the inner Kdo residue of Kdo 2 -lipid A in E. coli [26]. The acyltransferase LpxM, encoded by lpxM, adds a secondary acyl chain at the 3'-position of Kdo 2 -lipid A. Because lpxE was integrated in the lacZ site of the chromosome [18,19,26], lacI was also removed to release the control to the lactose operon. The correct deletion and replacement of genes in HWB01 and HWB02 were confirmed by using DNA sequencing.
To analyze the structure of LPS synthesized by the E. coli mutants HWB01 and HWB02, LPS was isolated from HWB01 and HWB02 and analyzed by TLC and ESI/MS (Fig 2) using WBB06 as a control. WBB06 can only synthesize Kdo 2 -lipid A because waaC and waaF have been inactivated in its chromosome (17). To determine the structural difference of LPS synthesized in HWB01 and HWB02, the total lipids were extracted from WBB06, HWB01 and HWB02 cells and separated by TLC [30] (Fig 2A). Two bands appeared at the top of the TLC plate for all samples. These bands represent phospholipids and migrated at similar rates, suggesting that the composition of the phospholipids in these three strains are similar. At the lower part of the TLC, however, there were bands that ran at different speeds for each sample. These bands may have been derived from Kdo 2 -lipid A, which usually migrates slower than phospholipids on TLC. Their different running speeds on TLC suggest that different structures of Kdo 2 -lipid A are produced by the WBB06, HWB01 and HWB02 cells. Rf values of Kdo 2lipid A molecules from HWB01, HWB02 and WBB06 were 0.416, 0.301 and 0.173, respectively, suggesting that Kdo 2 -lipid A molecules from HWB01, HWB02 are more hydrophobic due to the loss of the phosphate at the 1-position of lipid A. Kdo 2 -lipid A from HWB01 ran relatively faster than that from HWB02, suggesting the latter is more hydrophilic due to the loss of the second acyl chain at the 3'-position of lipid A.
To confirm their structures, Kdo 2 -lipid A samples extracted from E. coli WBB06, HWB01 and HWB02 were further analyzed by ESI/MS (Fig 2B). HWB01 and HWB02 cells have higher membrane permeability, higher surface hydrophobicity and higher auto-aggregation ability than wild type E. coli cells LPS forms the major component of the outer membrane inmost Gram-negative bacteria, covering approximately 75% of the cell surface area. Therefore, LPS helps stabilize the outer membrane and protects it from chemical attack [46]. The outer membrane permeability, cell surface hydrophobicity, and auto-aggregation of HWB01 and HWB02 cells were evaluated, using W3110, HW001, HW002 and WBB06 as controls (Fig 3). Compared with W3110, membranes of HW001 and HW002 were only slightly more permeable, but the membrane permeability of HWB01 and HWB02 cells increased by 3-and 5-fold, respectively (Fig 3A). The sensitivity of W3110, WBB06, HW001, HW002, HWB01 and HWB02 cells to the antibiotics novobiocin, erythromycin, and clarithromycin was tested, and MIC values were listed in Table 2. At MIC,  (Table 3). Based on MIC and CFU values, the bacterial cell sensitivity to these antibiotics increased according to the order of W3110, HW001, HW002, WBB06, HWB01 and HWB02. The results indicate that LPS structure is an important determinant of antibiotic susceptibility. This is consistent with the results of membrane permeability, suggesting that the increased membrane permeability of HWB01 and HWB02 cells might lead to their increased sensitivity to antibiotics.
The cell surface hydrophobicity of HW001 was almost the same as W3110 cells, but that of HW002 was slightly lower. However, the cell surface hydrophobicity of WBB06, HWB01 and HWB02 cells was significantly increased (Fig 3B). Loss of the hydrophilic polysaccharide of LPS in HWB01 and HWB02 cells might make these major molecules in the outer membrane more hydrophobic and thus increase the surface hydrophobicity of cells. The increased cell surface hydrophobicity might subsequently enhance the auto-aggregation of cells [47]. This is confirmed by the increased auto-aggregation of WBB06, HWB01 and HWB02 cells compared to that of W3110, HW001 and HW002 cells (Fig 3C). This increased aggregation may benefit the large-scale production of Kdo 2 -MPLA or Kdo 2 -pentaacyl-MPLA because aggregated cells are easier to collect. HWB01 and HWB02 cells showed less stimulating activities to HEK-Blue hTLR4 than wild type E. coli W3110 HEK-Blue hTLR4 cells expressing TLR4, MD-2, and CD14 were challenged with a range of CFU of W3110, HW001, HW002, WBB06, HWB01 or HWB02 cells, and the stimulating activities of these bacterial cells were determined by measuring the levels of the activator protein-1-dependent reporter SEAP in the mixtures. The stimulating activities of all of the strains were similar when less than 10 3 CFU/mL bacteria were used, but quite different stimulatory activities were observed when more than 10 3 CFU/mL bacteria were used (Fig 4). When more than 10 3 CFU/mL bacteria were used, the stimulating activities of HWB01 and HWB02 cells were less than that of W3110 and WBB06. Overall, the stimulating activities of W3110 and WBB06 were similar, those of HW001 and HWB01 were similar, and those of HW002 and HWB02 were also similar. For example, in a mixture containing 10 6 CFU/mL bacteria, the A 630 reached 1.21 for W3110 cells, 1.07 for WBB06, 0.80 for HW001 cells, 0.85 for HWB01, 0.36 for HW002 and 0.30 for HWB02 cells. These results indicate that the stimulating activity of E. coli cells to   5). Levels of the cytokines TNFα, IL-6, IL-8 or RANTES in the reaction mixtures were determined by ELISA. The cytokine levels were different because of the differences in the structure and concentration of LPS included in the mixtures, but they all increased as the concentrations of LPS ligands increased (Fig 5).  Kdo 2 -MPLA and Kdo 2 -Pentaacyl-MPLA induced the least levels of TNF-α, IL-6 and RANTES (Fig 5D-5F).
Under all tested circumstances, the stimulating activity of Kdo 2 -lipid A was similar to that of LPS, the stimulating activity of Kdo 2 -MPLA was similar to that of PS-MPLA, and the stimulating activity of Kdo 2 -pentaacyl-MPLA was similar to that of PS-pentaacyl-MPLA. Kdo 2 -pentaacyl-MPLA induced the least stimulation in all mixtures. The stimulating activity of Kdo 2 -MPLA or Kdo 2 -pentaacyl-MPLA was always lower than LPS, suggesting that Kdo 2 -MPLA or Kdo 2 -pentaacyl-MPLA may be good candidates for vaccine adjuvant development [49].

Discussion
In this work, E. coli mutants HWB01 and HWB02 were constructed by deleting and integrating key genes related to LPS biosynthesis so that they were able to synthesize Kdo 2 -MPLA and Kdo 2 -pentaacyl-MPLA, respectively. With the loss of hydrophilic polysaccharide, the 1-phosphate group or the 3'-secondery acyl chain, the structural changes to LPS not only influenced the membrane permeability and cell surface hydrophobicity of HWB01 and HWB02 cells but also decreased the ability of LPS to activate the TLR4/MD-2 receptor of mammalian cells. The stimulating ability of Kdo 2 -pentaacyl-MPLA was lower than that Kdo 2 -MPLA, suggesting the importance of the secondary acyl chain at 3'-position of lipid A. The secondary acyl chain at the 3'-position is bound deeply into the MD-2 component of the TLR4/MD-2 complex [50], therefore, secondary deacylation at the 3'-position of Kdo 2 -MPLA might alter the dimerization of TLR4/MD-2 and consequently inhibit its activation. Kdo 2 -MPLA and Kdo 2 -pentaacyl-MPLA could moderately stimulate RAW264.7 and THP-1 cells and had biological activities comparable to MPLA [51]. Therefore, Kdo 2 -MPLA and Kdo 2 -pentaacyl-MPLA may be good candidates for immune-pharmacological exploitations, vaccine adjuvant engineering, and antiinflammatory intervention investigation (38).
In cells, MPLA is usually connected to polysaccharides. To prepare MPLA, LPS has to be first isolated and then hydrolyzed. The isolation and quantification of LPS are difficult because of the large size and micro-heterogeneity of the molecule, and hydrolyzing LPS can damage the structure of MPLA; thus, the efficiency and quality of MPLA prepared from LPS are limited. Kdo 2 -MPLA and Kdo 2 -pentaacyl-MPLA, however, can be directly isolated from HWB01 and HWB02 cells without hydrolysis, and their small size and micro-homogeneity make them easier to purify. HWB01 and HWB02 were constructed by marker-less deletion and integration into the chromosome to facilitate the production of Kdo 2 -MPLA and Kdo 2 -pentaacyl-MPLA by fermentation. Future studies should focus on optimizing the growth conditions of HWB01 and HWB02 or genetically modifying the strains for large-scale industrial fermentation. The findings in this study may have important implications for the development of future vaccine adjuvants.