The Role of Oxidoreductases in Determining the Function of the Neisserial Lipid A Phosphoethanolamine Transferase Required for Resistance to Polymyxin

The decoration of the lipid A headgroups of the lipooligosaccharide (LOS) by the LOS phosphoethanolamine (PEA) transferase (LptA) in Neisseria spp. is central for resistance to polymyxin. The structure of the globular domain of LptA shows that the protein has five disulphide bonds, indicating that it is a potential substrate of the protein oxidation pathway in the bacterial periplasm. When neisserial LptA was expressed in Escherichia coli in the presence of the oxidoreductase, EcDsbA, polymyxin resistance increased 30-fold. LptA decorated one position of the E. coli lipid A headgroups with PEA. In the absence of the EcDsbA, LptA was degraded in E. coli. Neisseria spp. express three oxidoreductases, DsbA1, DsbA2 and DsbA3, each of which appear to donate disulphide bonds to different targets. Inactivation of each oxidoreductase in N. meningitidis enhanced sensitivity to polymyxin with combinatorial mutants displaying an additive increase in sensitivity to polymyxin, indicating that the oxidoreductases were required for multiple pathways leading to polymyxin resistance. Correlates were sought between polymyxin sensitivity, LptA stability or activity and the presence of each of the neisserial oxidoreductases. Only meningococcal mutants lacking DsbA3 had a measurable decrease in the amount of PEA decoration on lipid A headgroups implying that LptA stability was supported by the presence of DsbA3 but did not require DsbA1/2 even though these oxidoreductases could oxidise the protein. This is the first indication that DsbA3 acts as an oxidoreductase in vivo and that multiple oxidoreductases may be involved in oxidising the one target in N. meningitidis. In conclusion, LptA is stabilised by disulphide bonds within the protein. This effect was more pronounced when neisserial LptA was expressed in E. coli than in N. meningitidis and may reflect that other factors in the neisserial periplasm have a role in LptA stability.


Introduction
The obligate human pathogen, N. meningitidis, is the causative agent of transmissible sepsis and epidemic meningitis [1]. Although penicillin and cephalosporins are recommended for treatment, incremental sporadic increases in resistances to penicillin [2], rifampycin [3] and fluoroquinilones [4] have been detected and can cause periodic treatment failure. N. meningitidis is genetically conserved with the obligate human pathogen N. gonorrhoeae which is the causative agent of sexually transmitted gonorrhoea. Unlike N. meningitidis, N. gonorrhoeae has developed multi-drug resistance (MDR) phenotypes [5] which result in antibiotic treatment failures [6]. Since the rate of gonococcal disease in developing nations is approximately 10 times that of developed nations with approximately 106 million cases reported per annum, the world-wide spread of antibiotic resistant gonococci is of increasing concern [7].
One class of drugs which is being increasingly used for the treatment of MDR-Gram negative bacteria (GNB) causing sepsis is the class of cationic antimicrobial peptides (CAMPs), which includes the two clinically relevant compounds, polymyxin B and colistin (polymyxin E) and natural defensins such as LL-37 and protegrins. These compounds are generally considered to cause killing by lysis or by exerting a lethal oxidative stress response [8]. CAMPs have a cationic charge which enables them to bind to the negatively charged phosphate headgroups of the lipid A of LPS and accumulate in the outer membrane [9]. The electrostatic interaction stabilises the complex, enabling the polar face of the CAMP to align with the polar head groups of the lipid A and the lipophilic tail of the CAMP to insert into the membrane. The CAMP migrates across the periplasm and integrates into the inner membrane to form a pore, disrupting the phospholipid bilayer which results in cell lysis [9]. CAMPs consist of different structural classes which influences antimicrobial resistance profiles [10]. Although both N. gonorrhoeae and N. meningitidis are sensitive to protegrin and LL-37 [11,12], they are both intrinsically resistant to polymyxin [13,14].
GNB that are resistant to polymyxin substitute the negatively charged lipid A phosphate headgroups with positively charged moieties such as 4-amino-4-deoxy-L-arabinose (Ara4N). These residues disrupt the electrostatic interactions of the lipid A headgroups with polymyxin, reducing accumulation and integration of the CAMP in the outer membrane [9]. However, the intrinsically polymyxin resistant Neisseria spp. do not possess the ability to synthesise Ara4N [15]. Gonococci and meningococci produce similar lipopolysaccharide (LOS) structures which have a conserved inner core region composed of heptose and 3-deoxy-Dmanno-2-octulosonic acid (KDO) attached to a lipid A moiety embedded in the outer-membrane [16]. A variable length a-chain that extends from the 49 position of heptose I residue and is often composed of lacto-N-neotetraose forms the outer core region of neisserial LOS. In both species, the lipid A headgroups are decorated with phosphoethanolamine (PEA) by the lipid A PEA transferase, LptA [16,17]. Both the 1 and 49 headgroups of lipid A can be substituted with PEA and both residues are lost upon insertional inactivation of lptA [17]. Not only does the loss of PEA residues from lipid A result in the loss of resistance to polymyxin [13], it also results in sensitivity to complement mediated lysis [18] and a loss of the ability to attach and invade host cells [19]. The inactivation of LptA in N. gonorrhoeae results in the abrogation of colonisation in mouse and human models of infection [20]. In addition, LptA can be considered an important virulence factor in neisserial pathogens as it is absent in most commensal species and this is associated with reduced inflammatory potential of the lipid A from these species [21].
LptA is a member of the phosphoethanolamine (PEA) transferase family, YhjW/YjdB/YijP, and the structure of the globular domain has been recently solved by Wanty et al. [22]. LptA was shown to possess five disulphide bonds suggesting that it could be a substrate for the protein oxidation pathway [23]. Neisseria spp. contain up to three specialised oxidoreductases, DsbA1, DsbA2 and DsbA3 [23] that introduce disulphide bonds into proteins in the periplasm. DsbA1/2 have overlapping specificities and are required for the introduction of disulphide bonds into the type IV pilin proteins, PilE and PilQ [24][25][26]. However, DsbA3 has no known substrate. When oxidoreductases are inactivated in GNB, the loss of disulphide bonds in the substrate proteins results in decreased stability of the protein in the periplasm due to sensitivity of the proteins to periplasmic proteases [27]. To further understand the mechanisms governing LptA expression in Neisseria spp., we examined whether LptA stability is dependent upon the periplasmic protein oxidation pathway and whether this affected polymyxin resistance.

Methods and Materials
Bacterial strains and growth conditions E. coli strains were grown either in Luria-Bertani (LB) broth or LB agar at 37uC. Meningococcal strains were grown with 5% CO 2 at 37uC unless specified otherwise. Gonococcal base agar (GCA, Oxoid) was supplemented with 0.4% glucose, 0.01% glutamine, 0.2 mg of co-carboxylase per litre, and 5 mg of Fe(NO 3 ) 3 per litre.
GC broth (GCB) consisted of 1.5% (w/v) special peptone, 0.4% (w/v) potassium phosphate dibasic (K 2 HPO 4 ), 0.1% (w/v) potassium phosphate monobasic (KH 2 PO 4 ) and 0.5% (w/v) sodium chloride (NaCl) with 0.043% NaHCO 3 . The antibiotic concentrations used for E. coli strains were as follows: ampicillin at 100 mg/ml, kanamycin at 50 mg/ml and chloramphenicol at 30 mg/ml. Those for N. meningitidis strains were as follows: kanamycin at 100 mg/ml, spectinomycin at 60 mg/ml, erythromycin at 3 mg/ml and tetracycline at 1 mg/ml. E. coli strains were transformed by chemical transformation [28]. Meningococci were transformed by the plate transformation procedure [29] or by chemical transformation. E. coli strain DH5a was used for the cloning and propagation of plasmids. Polymyxin minimum inhibitory concentrations (MICs) were determined using polymyxin B or E E-strips (bio Mérieux) as per manufacturer's instructions. For E. coli isolates, LB agar contained 0.4 mM isopropyl b-D-1thiogalactopyranoside and was inoculated with 4610 7 colony forming units for each strain. A polymyxin-E E-strip was overlayed on the plate which was incubated overnight at 37uC and the zone of inhibition was read after 18 hrs. For meningococcal isolates, GCB was inoculated with 4610 7 colony forming units for each strain. A polymyxin-B E-strip was overlayed on the plate which was incubated overnight at 37uC and the zone of inhibition was read after 18 hrs.
Construction of LptA::His x6 expression vectors for use in E. coli and N. meningitidis All bacterial strains and plasmids used in this study are described in Table 1 and Table 2. For the expression of neisserial LptA::His x6 in E. coli, the high copy expression vector pCMK526 was engineered to contain lptA with a C-terminal His tag under control of the P Trc promoter. Wild-type lptA gene was PCR amplified from N. meningitidis strain NMB genomic DNA with primer pair KAP70 (59-ATGTCTGCCGGACGTTTGAATGG-39) and KAP71 (59-ATTGCCGTGATCGGGAACTTGG-39) and cloned into the low copy expression vector pHSG576 [30] to form pCMK519. lptA was PCR amplified from pCMK519 with primer pairs KAP96 (59-AGCGGATAACAATTTCACA-CAGGA-39) and KAP193 (59-GGGAAGCTTTGCGCG-GACGGCGGCAGGCTGCCAATATATC-39), which replaced the stop codon with a HindIII site, then cloned into the HincII site of the low copy expression vector pHSG576 to form pCMK521. To enable directional cloning of lptA into pTrc99A and pCMK133, an NcoI site was included upstream of lptA. To do this, lptA gene was amplified from pCMK521 with KAP97 (59-GTTTTCCCAGTCACGAC-39) and KAP64 (59-CATGC-CATGGTAGGAGGTCCAAATGATAAAACCGAACCTGAG-GCCGAAGC-39) which introduces an NcoI site and E. coli consensus shine dalgarno site (italics) upstream of the lptA start codon (bold). The PCR product was treated with T4 DNA polymerase, ligated into the HincII site of pHSG576 to form pCMK522. To create a gene encoding LptA::His x6 , lptA was cloned into the vector pCMK133, which contains a HindIII site for in frame fusion with a His x6 -tag next to an aphA-3 marker in the polylinker of pENTR4 (Invitrogen). The vector pCMK133 was engineered to enable the expression of a C-terminal His-tag protein when a gene of interest containing a HindIII site replacing the stop codon is cloned into the NcoI/HindIII site replacing ccdB. Both vector pCMK522 and pCMK133 were digested with NcoI and HindIII, ligated and transformed into E. coli strain DH5a. A clone that contained lptA within pCMK133 with a successful Cterminal His-tag fusion was named pCMK524. The lptA::His x6 gene was excised from pCMK524 by restriction digest with NcoI and SmaI and ligated with pTrc99A [31] restricted with NcoI and SmaI to form pCMK526.
For the expression of neisserial LptA::His x6 in N. meningitidis, the low copy expression vector pHSG576 [30] was modified to contain the GC cryptic backbone of pYT250 [15] to allow replication in Neisseria, to confer resistance to kanamycin instead of resistance to chloramphenicol and to contain lptA downstream of the neisserial P ompR promoter. Briefly, the aphA-3 gene was liberated from the vector pUC18K [32] by restriction digest with SmaI and cloned into ScaI and MscI disgested pHSG576 [30] to disrupt cat1 and form pCMK597. A 72 bp multiple cloning site (MCS) was engineered. Briefly, the primers KAP584 (59-GAATTCGGATCCACAGGAGGCAGACCATGGCCACTC-GAGAGCTCGGTACCCGGGCATGCATCTAGAGTCGAC- 39) and KAP585 (59-GTCGACTCTAGATGCATGCCCGGG-TACCGAGCTCTCGAGTGGCCATGGTCTGCCTCCTGTG GATCCGAATTC-39) were resuspended in 16T4 DNA ligase buffer (New England Biolabs) at equimolar concentrations and annealed by incubation at 90uC for 5 minutes followed by incubation at room temperature for 60 minutes. The resultant fragment was restricted with EcoRI and HincII and cloned into pCMK597, restricted with the same enzymes, to form pCMK598. The GC cryptic backbone from pYT250 was excised by restriction digest with HindIII and cloned into the HindIII site of pCMK598 to form the neisserial shuttle vector pCMK599. The neisserial P ompR promoter was liberated from pJSK411::ompR [33] by digest with BamHI and EcoRI and cloned into the BamHI and EcoRI restriction sites of the MCS of pCMK599 to form pCMK600. The lptA::His x6 construct was liberated from pCMK526 by restriction digest with NcoI and SmaI and cloned into the NcoI and SmaI restriction sites of the MCS of pCMK600 to form the neisserial LptA::His x6 expression vector, pCMK1001. To create an expression shuttle vector with a chloramphenicol marker, the GC cryptic plasmid from pYT250 was cloned into the HindIII site of pHSG576 to create pCMK940. The P ompR -lptA::His x6 cassette was excised from pCMK1001 with the restriction sites SmaI and EcoRI and ligated into these sites in pCMK940 to create pCMK946.
Construction of LptA::His x6 and oxidoreductase expressing E. coli The wild-type E. coli strain JCB570 and corresponding E. coliDdsbA mutant strain JCB571 were used as control strains [34]. EcDsbA was expressed from pCMK255 as described previously [35]. To complement JCB571 with EcDsbA, pCMK255 was transformed into this host to create CKEC272 [35]. To construct an E. coli strain expressing LptA::His x6 without any oxidoreductase, JCB571 was transformed with pCMK526 to form CKEC543. To create an E. coli strain co-expressing LptA::His x6 and EcDsbA, CKEC543 was transformed with pCMK255 to form CKEC564.

Construction of LptA and oxidoreductase mutants in N. meningitidis
To make a DlptA::aadA mutant in strain NMB, pKA314 [13] was transformed into strain NMB and was stored as CKNM204. CKNM101 (NMBDdsbA1), CKNM102 (NMBDdsbA2) and CKNM105 (NMBDdsbA1/dsbA2) were created as described in Kumar et. al. [36]. CKNM631 (NMBDdsbA3) was constructed by transformation of pCMK631 (pHSG576 containing dsbA3::aphA-3) into N. meningitidis strain NMB. The vector pCMK631 was constructed by SOE PCR to introduce an internal KpnI and XbaI into which an aphA-3 cassette was inserted. In the first round of SOE PCR, an internal region of dsbA3 from NMB genomic DNA Table 2. Plasmids used in this study.
Expression of LptA::His x6 in oxidoreductase mutants of N. meningitidis.
The neisserial LptA::His x6 expression vector pCMK1001 was transformed into N. meningitidis strain NMB to form strain CKNM216. CKNM219 (CKNM216DdsbA1) was constructed by transformation of pCMK109 [36] into CKNM216. CKNM221 (CKNM216DdsbA1/dsbA2) was constructed by transformation of the vector pJKD2641 [36] into CKNM219. CKNM222 (CKNM216DdsbA3) was constructed by chemical transformation of the vector pJKD2643 [36] into CKNM216. To create a neisserial oxidoreductase triple mutant expressing LptA::His x6 , pCMK946 was naturally transformed into CKNM212. The retention of the plasmid was confirmed by extraction of the plasmid and restriction mapping, followed by a Western immunoblot which confirmed that LptA::His x6 was expressed.
Purification of LptA::His x6 from CKNM216 The membrane fraction of CKNM216 was separated by resuspending the growth from four overnight plates in 30 ml binding buffer (20 mM Na 3 PO 4 , 500 mM NaCl, 20 mM imidazole, pH 7.4) and sonicated on ice for 30 mins (30 sec on, 30 sec off). The suspension was centrifuged at 30006g, 20 mins and the supernatant further centrifuged at 100,0006g for 120 mins at 4uC. The supernatant was discarded and the pellet re-suspended in 20 ml binding buffer containing 2% triton X-100 at room temperature for 3 hrs. The suspension was centrifuged at 50,0006g for 30 mins at 4uC. LptA::His x6 was purified from the supernatant on a HisTrap FF column (GE Helathcare Life Sciences). Following equilibration and application of the supernatant, the column was washed with 10 volumes of binding buffer containing 2% triton X-100 and the protein eluted in 10 volumes of elution buffer (20 mM sodium phosphate, 0.5 M NaCl and 500 mM imidazole, pH 7.4) containing 2% triton X-100 as per manufacturer9s instructions. The eluate was dialysed three times against 10% glycerol, 20 mM Tris-HCl, pH 7.9), and the protein concentrated to 0.5 ml using a Centricon 10 column as per manufacturer9s instructions (Millipore).
Detection of the expression of LptA::His x6 was determined by western immunoblotting. Whole cell lysates (1 mg) were separated by 4% stacking and 12% separating sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) by standard methods and transferred to nitrocellulose membranes. The membranes were blocked for 1 hr at room temperature with 2% BSA in TBS. The polyclonal rabbit anti-LptA IgG [19] and the conjugated monoclonal mouse anti-His-HRP (Sigma) primary antibodies were used at 1:500 and 1:1,000 in blocking buffer respectively, and the membranes incubated overnight at room temperature. Horse radish peroxidase-conjugated anti-rabbit IgG secondary antibody (Santa Cruz Biotechnology) was used for detection of rabbit anti-LptA IgG at a concentration of 1:1,000 in blocking buffer for 3 hrs at room temperature and the membranes were developed by colormetric analysis using with 30 mg 4-chloronaphthol (Sigma) dissolved in 10 ml methanol plus 30 ml H 2 O 2 (Univar) in 40 ml Tris-NaCl, pH 7.4.
Determination of the redox status of LptA::His x6 The redox states of LptA::His x6 in vivo were determined by the same method as described previously for DsbA [37]. Briefly, growth from overnight plates was resuspended in broth to an OD 600 = 1.2 for E. coli cultures, and an OD 560 = 2.0 for N. meningitidis cultures. Trichloroacetic acid (final concentration 5% v/v) was added directly to 1 ml of culture to denature and precipitate whole-cell proteins, which were then collected by centrifugation, washed with acetone and dissolved in 100 ml 1% SDS-100 mM Tris-HCl (pH 7.5) containing 20 mM 4-acetamido-49-maleimidylstilbene-2,29-disulfonic acid (AMS). For a reduced control, 1 ml of culture was treated with 17 mM dithiothreitol (DTT) for 10 minutes at 37uC with shaking prior to treatment with TCA (5% v/v) and AMS alkylation as described above. In addition, 1 ml of culture was centrifuged and resuspended in 100 ml sterile H 2 O. All samples were boiled for 10 minutes after the addition of 25 ml of 56 sample buffer without reducing agent. Samples were loaded onto a 4% stacking/12.5% bis/acrylamide SDS-PAGE which was run at 100 V in the cold room. The proteins were transferred to a Hybond C nitrocellulose membrane (Amersham Life Sciences) at 100 V for 1.5 hr using a Bio-Rad PROTEAN 3 Western blot apparatus (Bio-Rad Laboratories). The membrane was probed with mouse anti-His x6 antibody (Sigma) diluted 1:1000 in blocking buffer (2% BSA in Tris-NaCl, pH 7.4), incubated at 37uC with shaking, for 3-4 hours. The membrane was then probed with the HRP-linked donkey anti-mouse IgG (Santa Cruz Biotechnology) diluted 1:1000 in blocking buffer and incubated at 37uC for 2 hrs with shaking. The membranes were developed with either the ECL detection kit Western Blotting Analysis System (Amersham Life Sciences) or by colorimetric solutions as described previously. Purified truncated NmLptADMA which migrates as a 31 kDa protein was prepared according to Anandan et al. [38].

Purification of LOS/LPS and analysis by mass spectroscopy
The LOS/LPSs were prepared by hot phenol/water extraction [39]. The crude LOS/LPS from the aqueous phase was dialyzed against deionized water using the aqueous phase 1000 MWCO dialysis tubing. After dialysis, the samples were freeze-dried and resuspended in a solution of 20 mM Tris-HCl and 2 mM MgCl 2 at pH 8.0. DNase I (100 ml of 7 mg/ml in 20 mM Tris-HCl, pH 8.0), and RNase A (100 ml of 17 mg/ml) were added and the solution was incubated for 3 hrs at 37uC. After adding 400 mg of proteinase K and adding CaCl 2 to a final concentration of 2 mM, the sample was incubated at 37uC overnight. The LOS was pelleted by ultracentrifugation at 100,0006g for 18 hrs. The pellets were re-suspended in water and re-centrifuged at 100,0006g for 4 hrs. The resulting pellets were suspended in water, freeze-dried analysed as described.
The various lipid A samples were prepared by subjecting each LOS preparation to 1% SDS in 10 mM sodium acetate buffer, pH 4.5 at 100uC for 1 hr, followed by lyophilisation. The SDS was removed by washing the dried residue with 100 ml of distilled water and 500 ml of acidified ethanol (prepared by combining 100 ml of 4 M HCl with 20 ml of 95% ethanol), followed by centrifugation (3,0006g, 20 min). The precipitate was then washed with 500 ml of 95% ethanol and centrifuged. The centrifugation and washing steps were then repeated. The final precipitate was lyophilized to give purified lipid A. Lipid A fractions were analysed by matrix assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF-MS) using an AB SCIEX TOF/TOF 5800 (Applied Biosystems). Lipid

Results
Expression of neisserial LptA in E. coli increases resistance to polymyxin and is dependent upon the co-expression of an oxidoreductase To examine whether neisserial LptA stability was dependent upon oxidoreductases, we first established a model to detect neisserial LptA expression in E. coli ( Table 1). Expression of LptA::His x6 in E. coli JM109 (CKEC526, 1.5 mg/ml) resulted in a 16-fold increase in polymyxin MIC when compared to JM109 carrying the empty vector (EXEC94, 0.094 mg/ml). To test for the dependency of neisserial LptA stability on the presence of an oxidoreductase, LptA::His x6 was expressed in JCB571 (CKEC543) in which the chromosomal copy of dsbA had been insertionally inactivated [34]. Expression of LptA in E coli lacking a functional DsbA had no effect on polymyxin resistance and the MIC (0.094 mg/ml) was the same in both CKEC543 and JCB571. When JCB571 was complemented with a vector expressing EcDsbA (CKEC272), no difference in MIC from the parental wild-type JCB571 was detected. However, co-expression of neisserial LptA::His x6 and EcDsbA in JCB571 (CKEC564) resulted in an increase of polymyxin MIC to 3 mg/ml, representing a 32-fold increase in MIC over parental wild-type JCB571. Therefore, neisserial LptA::His x6 increases resistance of E. coli to polymyxin only when expressed in the presence of an oxidoreductase.

LptA catalyses the addition of PEA groups to E. coli lipid A
To determine the effects of LptA::His x6 expression on modifications to lipid A in E. coli, samples of lipid A were prepared from E. coliDdsbA JCB571 strains expressing LptA::His x6 and/or EcDsbA and analysed by mass spectrometry (see methods). Figure 1. The lipid A from strains E. coliDdsbA JCB571 and JCB570 (wild-type parent) gave the same spectra as that shown for the lipid A from JCB571 carrying a vector expressing EcDsbA (CKEC272). The observed ions were consistent with the normal E. coli lipid A molecules; namely the bis-phosphorylated hexaacylated structure (m/z = 1796), its mono-phosphorylated derivative (m/z = 1716), and its heptaacylated version due to the addition of a palmitic acid residue (m/z = 2034). The lipid A preparations from strains JCB571 carrying LptA::His x6 (CKEC543) and JCB571 co-expressing LptA::His x6 and EcDsbA (CKEC564) showed these ions plus additional ions due to the addition of PEA to the mono-phosphorylated structure (m/z 1839, i.e. 1716+123) and the bis-phosphorylated structure (m/z 1919; i.e. 1796+123). In addition, some minor lipid A molecules with a single PEA (m/z 1693, 1891, 1947, and 2157) were also detected along with those without PEA (m/z 1570, 1768, 1824 and 2034) in both CKEC543 and CKEC564. However, the ratio of total peak area of the lipid A molecules with PEA (i.e. the peak area of the m/ z 1839+1919+1693+1891+1947+2157 ions) to those without PEA (i.e. the peak area of the m/z 1716+1796+1570+1768+1824+2034 ions) was ten-fold less in strain JCB571 with LptA::His x6 alone than for JCB571 co-expressing LptA::His x6 and EcDsbA (0.036 versus 0.36). Thus, a single PEA addition to lipid A of LPS of JCB571 strains expressing LptA::His x6 (CKEC543 and CKEC564) was detected that is not present in control strains JCB570 (wild type), JCB571 (DdsbA) and JCB571 expressing EcDsbA alone (Figure 1). This addition is more prevalent in JCB571 co-expressing LptA::His x6 and EcDsbA than in JCB571 with LptA::His x6 alone and correlates with increased resistance to polymyxin of the former strain. This experiment confirms that neisserial LptA::His x6 is able to transfer PEA to lipid A of E. coli LPS resulting in increased resistance to polymyxin when coexpressed with the oxidoreductase EcDsbA.

MALDI-TOF MS analysis of the various lipid A samples is shown in
The stability of neisserial LptA::His x6 is dependent upon the presence of EcDsbA in an E. coli host Alkylation with 4-acetamido-49-maleimidylstilbene-2,29-disulfonic acid (AMS) is a technique used to determine the in vivo redox states of proteins [40]. AMS forms covalent adducts with free thiol groups of reduced cysteine residues adding ,500 a.m.u. to the mass of a protein for each reduced cysteine residue modified which can be detected by mobility shift on SDS-PAGE. Therefore, an immunoblot using anti-His x6 antibody was used for the detection of LptA::His x6 in AMS alkylated and untreated cell lysates from E. coli JCB571 co-expressing LptA::His x6 and EcDsbA (CKEC564). Isolates JCB571, JCB571 containing pTrc99A (CKEC288), JCB571 expressing EcDsbA, alone and JCB571 expressing LptA::His x6 alone were used as controls. A protein of molecular weight of 48 kDa was detected in JCB571 co-expressing LptA::His x6 and EcDsbA but not in any other strain (Figure 2). Intact LptA is predicted to have a molecular weight of 60 kDa. A comparison of the whole cell lysates with the purified LptA::His x6 which has been shown to be intact by crystallisation (data not shown) indicated that the protein in the lysates was the same mass as the purified sample ( Figure S1). Thus it appears that the stained molecular weight markers were inaccurate as a standard for determining mass under the conditions used for these SDS-PAGE gels.
In the absence of oxidoreductases, proteins without disulphide bonds will either accumulate in the periplasm as reduced proteins, or will be more sensitive to proteolytic degradation and be removed entirely. Since LptA::His x6 expression was detected in JCB571 in which EcDsbA was also present but not where LptA::His x6 is expressed in the absence of a functional oxidoreductase, we hypothesised that DsbA catalysed disulfide bond formation in LptA::His x6 was required for stability of LptA. In the absence of DsbA, LptA::His x6 would become sensitive to proteolytic degradation in the periplasm and therefore was absent in JCB571 expressing LptA::His x6 alone. This hypothesis is supported by the observation that LptA::His x6 was partially degraded (fainter band) in the presence of EcDsbA when the cell lysate was pre-treated with the reducing agent DTT which would have reduced the disulphide bonds in periplasmic proteins ( Figure 2). Finally, AMS alkylation of JCB571 co-expressing LptA::His x6 and EcDsbA did not result in a mobility shift in the size of LptA::His x6 . This result indicates that there were no free thiol groups for alkylation in LptA::His x6 because it was oxidised by EcDsbA. In conclusion, LptA::His x6 contains EcDsbA catalysed disulfide bonds that are necessary for protein stability in the periplasm. In the absence of EcDsbA, LptA::His x6 is rapidly removed via proteolytic degradation.
All three neisserial oxidoreductases contribute to resistance to polymyxin Unlike E. coli, which possesses a single oxidoreductase, Neisseria spp. contain up to three oxidoreductases, termed DsbA1, DsbA2 and DsbA3. To examine whether any or all of the neisserial oxidoreductases were required for the stability of LptA in neisseria, we examined the polymyxin resistance profile of a series of meningococcal oxidoreductase mutants. N. meningitidis strain NMB contains all three oxidoreductases. The MIC of wild-type strain NMB to polymyxin is 384 mg/ml while NMBDlptA is 1000fold more sensitive with an MIC of 0.38 mg/ml. Mutation of each of the individual oxidoreductases in turn resulted in a 3-fold decease in MIC to 128 mg/ml. All combinations of double mutations were tested and only a double mutation in both dsbA1 and dsbA2 resulted in a further 2-fold decrease in MIC to 64 mg/ ml (Table 1). However, since the level of polymyxin resistance of the NMBDdsbA1/dsbA2 double mutant was still 164-fold above the level obtained for the DlptA mutant, the NMBDdsbA1/dsbA2/ dsbA3 triple mutant was examined. Since the NMBDdsbA1/ dsbA2/dsbA3 triple mutant is temperature sensitive, all of the strains were compared at 30uC. The MICs for the single and double mutants remained unchanged at 30uC (data not shown). Interestingly, the NMBDdsbA1/dsbA2/dsbA3 triple mutant was sensitive to a MIC of 32 mg/ml of polymyxin (Table 1) which is 2fold less than that of NMBDdsbA1/dsbA2 but still 84-fold greater than the MIC for the NMBDlptA mutant.

LptA is reduced but remains stable in the absence of disulphide bonds in the neisserial periplasm
To test the redox status of LptA in N. meningitidis, LptA::His x6 was expressed from a neisserial shuttle vector pCMK964 (chloramphenicol resistant) or pCMK1001 (kanamycin resistant) in a variety of mutant backgrounds in which different combinations of oxidoreductases were removed by insertional inactivation of the chromosomal loci. To confirm that the LptA::His x6 was expressed and intact in N. meningitidis hosts, whole cell lysates of N. meningitidis strain NMB carrying pCMK1001 (CKNM216) were compared to whole cell lysates prepared from NMBDlptA, E. coli JCB571 co-expressing LptA::His x6 with EcDsbA, and E. coli JCB571 expressing EcDsbA alone. Immunoblots using anti-Histag antibody ( Figure S2, Panel A) and anti-LptA antibody ( Figure  S2, Panel B) detected a protein of the correct mass which was consistent in size with the protein detected in the E. coli cell lysates. To confirm that lptA::His x6 gene was intact, pCMK1001 was extracted from strain NMB and fully sequenced. In addition, LptA::His x6 was purified using the nickel columns from preparations of the neisseria membrane and was shown to react with both the anti-His-tag and anti-LptA antibodies ( Figure S2). To determine the effect of oxidoreductase expression on LptA::His x6 stability, pCMK1001 was introduced into the oxidoreducatase mutant backgrounds. Whole cell lysates of these strains were alkylated, separated by SDS-PAGE, transferred to a membrane and immunobloted using an anti-His HRP conjugate antibody to detect LptA::His x6 (Figure 3). LptA::His x6 is oxidised in each of the alkylated samples from NMBDdsbA1 expressing LptA::His x6 (CKNM219), NMBDdsbA1/dsbA2 expressing LptA::His x6 (CKNM221) and NMBDdsbA3 expressing LptA::His x6 (CKNM222). Strains expressing either DsbA1 or DsbA2 alone also produced oxidised LptA::His x6 (data not shown). However, in NMBDdsbA1/dsbA2/dsbA3 expressing LptA::His x6 (CKNM755) reduced LptA::His x6 accumulated. In conclusion, these alkylating assays indicate that DsbA1, DsbA2, and DsbA3 donate disulphide bonds to LptA. In addition, since reduced LptA::His x6 accumulated in the triple oxidoreductase mutant, this protein remained stable and was not immediately degraded in the absence of disulphide bonds.
The neisserial oxidoreductase, DsbA3, contributes to the activity of LptA Although mutation of the oxidoreductases resulted in decreased resistance to polymyxin in neisseria, it was not certain whether this phenotype correlated with a change in the substitution profile of the lipid A with PEA headgroups in the oxidoreductase mutants. To examine this further, the PEA substitution profile of lipid A was examined by MS from the oxidoreductase mutants and NMBDlptA ( Figure 4). The lipid A preparations from oxidore-  ductase mutants and the control wild-type strain NMB possessed ions consistent with the addition of PEA to the mono-phosphorylated structure (m/z = 1755, i.e. 1632+123), the bis-phosphorylated structure (m/z = 1835; i.e. 1712+123) and the tri-phosphorylated structure (m/z = 1915; i.e. 1792+123) (Figure 4). The NMBDlptA mutant lacked all of these ions, consistent with the loss of LptA activity. To compare the PEA level in all of these strains, all of the lipid A molecules with or without PEA were included in the calculations. The ratio of total peak area of the lipid A molecules with PEA (i.e. m/z 1557+1653+1755+1807+ 1835+1857+1889+1897+1915 ions) to those without PEA (i.e. the total peak area of the m/z 1434+1530+1632+1684+1712+1734+ 1766+1774+1792 ions) was 2:1 for wild-type strain NMB and 0.006:1 for NMBDlptA in which almost all PEA was absent. In comparison, NMBDdsbA1/dsbA2 had a ratio of 2:1 similar to wildtype while NMBDdsbA3 and NMBDdsbA1/dsbA2/dsbA3 each had a reduced ratio of 1.7:1. Therefore, it appears that DsbA3 contributes to the activity of LptA in vivo, which is consistent with the 3-fold decrease in polymyxin MIC observed for the DsbA3 mutants. Since there was no change in the ratio of PEA -bearing lipid A in the NMBDdsbA1/dsbA2 relative to the parental isolate, this would suggest that the disulphide bonds catalysed by DsbA3 alone are sufficient to ensure full activity of LptA.

Discussion
Neisserial LptA is an integral membrane protein which contains five disulphide bonds and has a predicted molecular weight of approximately 60 kDa. In E. coli, the stability and activity of neisserial LptA::His x6 was entirely dependent upon the coexpression of the EcDsbA oxidoreductase. Although expression of LptA::His x6 and EcDsbA in E. coli resulted in a 32-fold increase in polymyxin resistance, the proportion of lipid A headgroups substituted with PEA rose to 26%. It is unclear why expression of LptA did not result in 100% substitution of lipid A headgroups, but this may reflect the ability of neisserial LptA to recognise E. coli lipid A as a substrate. MALDI-TOF MS analysis of the E. coli lipid A confirmed that LptA modified lipid A headgroups with PEA, albeit only on one site. Due to the low prevalence of the ion substituted with PEA in these samples, an unambiguous allocation to the 1 or 49 position could not be made by NMR (data not shown). However, in meningococci, lipid A is substituted at both the 1 and 49 positions with PEA and both substitutions are lost in the absence of LptA [17]. Therefore, the attachment of PEA to one site of E. coli lipid A may be the result of specificity for the acyl-chain distribution on the lipid A. While the meningococcal lipid A possesses a lauroyl residue at the 29 position of each glucosamine [41], E. coli LPS contains a lauroyl and myristoyl residue at the 29 and 39 positions of the distal glucosamine residue respectively, resulting in an asymmetrical acylation pattern of lipid A [42]. Therefore, it is possible that the proximal glucosamine of E. coli lipid A which resembles the lipid A of meningococci could act as an acceptor resulting in substitution of the 49 position only. Alternatively, previous studies by Reynolds et al. [43] using EtpB in E. coli have demonstrated that acyl chain length and saturation of phosphatidylethanolamine determine which isoform is used as a substrate for the transfer reaction of PEA to lipid A. E. coli EtpB preferred phosphatydylethanolamine with C16 and C18 acyl chains, while phosphatydylethanolamine with C12 and C14 chains were not used by this transferase. Not surprisingly, in E. coli phosphatydylethanolamine isoforms with C12 and C14 acyl chains represents less than 5% of phosphatydylethanolamine present in the membranes [43]. In comparison, neisseria have membranes enriched with phosphatydylethanolamine isoforms carrying C12:0, C14:1 and C14:0 acyl chains which correspond to the length of the acyl chains decorating lipid A [44]. Therefore, if neisserial LptA::His x6 has a substrate preference for phosphatydylethanolamine isoforms with C12 and C14 acyl chains, the availability of this substrate would be a limiting factor for LptA::His x6 activity in E. coli.
Unlike E. coli, Neisseria spp. possesses three oxidoreductases which are considered to be specialist catalysts with different, although largely uncharacterised substrate specificity. DsbA1 and DsbA2 have been demonstrated to have some overlapping substrate specificity and are known to be responsible for stabilising the PilQ secretin and the PilE shaft proteins of Type IV pili [26]. DsbA3 which shares 57% and 51% amino acid identity with DsbA1 and DsbA2, respectively, is unable to oxidise the DsbA1/2 substrates and possesses a variant catalytic motif of CVHC rather than the canonical CPHC motif in DsbA1 and DsbA2 [24][25][26], [45]. Regardless of their functional and structural differences, all three oxidoreductases have similar redox potentials of 280 mV [46]. To determine whether any of these oxidoreductases contributed to the polymyxin resistance phenotype, each locus was inactivated in meningococcal strain NMB and led to a 3-fold increase in sensitivity to polymyxin for each mutant. Combinatorial mutations suggest that the defect in polymyxin resistance was additive, indicating that either each oxidoreductase affected a different pathway contributing to polymyxin resistance or they were co-operatively involved in stabilising proteins in a common pathway necessary for resistance. Apart from modification of lipid A with PEA groups, polymyxin resistance in meningococci and gonococci is also determined by the activity of resistance/ nodulation/division (RND)-type efflux pump [47,48] and indirectly by mutations affecting membrane barrier function [13]. Future analysis of these pathways will determine if there are oxidoreductase-dependent substrates involved in these mechanisms of resistance to polymyxin.
Correlates were sought between polymyxin sensitivity, LptA stability or activity and the presence of each of the neisserial oxidoreductases. Due to the low levels of natural expression of LptA in meningococci, the amount of LptA in the oxidoreductase mutants could not be detected by western immunoblot (data not shown). Expression of LptA::His x6 in meningococci enabled the detection of the oxidation of the protein in the presence of natural levels of each oxidoreductase. LptA remained oxidised in the presence of DsbA1, DsbA2 or DsbA3 alone. The presence of PEA on the lipid A headgroups was marginally reduced in the absence of DsbA3 but remained unaffected in the absence of DsbA1 and DsbA2 together. This implies that DsbA3 contributed to the activity of LptA, possibly through the donation of disulphide bonds to LptA. The disulphide bonds in the globular domain of LptA were found to create five small loops across the surface of the protein and are not involved in linking any of the core structures together that surround the active site [22]. Therefore, it is more likely that these disulphide bonds may be involved in protecting LptA from proteolytic degradation. It appears from this study that multiple oxidoreductases can donate disulphide bonds to LptA but only those donated by DsbA3 have some effect on LptA activity suggesting that each oxidoreductase may be responsible for different bonds.
Although the effect of disulphide bonds on LptA stability was accentuated in E. coli, the disulphide bonds did not appear to be essential for LptA stability in the neisserial periplasm. This is demonstrated by the observation that the triple DdsbA1/dsbA2/ dsbA3 knockout in N. meningitidis possessing none of these oxidoreductases is still 84-fold more resistant to polymyxin than a meningococcal strain in which LptA was inactivated. The analysis  = 1712), the mono-phosphorylated (m/z = 1632) and the tri-of the lipid A substitution profile of NMBDdsbA1/dsbA2/dsbA3 confirmed that PEA addition to lipid A headgroups was present at approximately 63% of the wild-type levels and was similar to that of a DsbA3 mutant. This data suggests that even when LptA lacks the disulphide bonds donated by the oxidoreductases, the protein remains stable and mostly active. This was confirmed by the alkylation immunoblot of the NMBDdsbA1/dsbA2/dsbA3 expressing LptA which revealed that reduced LptA was accumulated in the absence of all three oxidoreductases and was not completely degraded as it was in E. coli. Currently, it is not clear whether there is a difference in the relative abundance of proteases in the periplasm of each species. However, if the neisserial periplasm had reduced proteolytic activity relative to the periplasm of E. coli, the reduced LptA would appear to accumulate in the neisserial host background. Other factors which may affect protein stability include the ability to form part of a protein complex which could protect LptA from degradation in N. meningitidis but not E. coli as these partners would be absent in the heterologous host.
In conclusion, this study has shown that post-translational modification of LptA with disulphide bonds is an important mechanism governing stability and activity of the enzyme in the E. coli periplasm. Although neisserial LptA is also oxidised in the presence of all three specialised neisserial oxidoreductase catalysts, DsbA1, DsbA2 and DsbA3, the absence of these enzymes does not result in the complete loss of LptA stability and minimally reduces activity. This suggests that LptA is reasonably stable in the absence of any of these known oxidoreductases and may suggest other factors could be involved in determining the stability of LptA in the periplasm. Although the oxidoreductases were not essential for LptA activity or stability in N. meningitidis, they appear to contribute to other mechanisms that determine polymyxin resistance. Figure S1 LptA::His x6 expression in whole cell lysates results in an intact protein of 60 kDa. Standardised cell lysates were separated by SDS-PAGE, followed by transfer to a membrane and western immunoblot using anti-His x6 HRP conjugate antibody to detect the presence of LptA::His x6 . Lane 1 contains a whole cell lysate prepared from CKNM216 (strain NMB expressing LptA::His x6 from pCMK1001); Lane 2 contains 250 ng of LptA::His x6 purified from E. coli JCB571 expressing LptA::-His x6 and EcDsbA (CKEC564) and Lane 3 contains the New England Biolabs 2-212 protein standard. When compared to this standard, LptA::His x6 was detected as protein of approximately 60 kDa, consistent with the purified LptA::His x6 which has been confirmed to be full-length by solving the crystal structure (data not shown). (TIF) Figure S2 LptA::His x6 is expressed from the shuttle vector pCMK1001 in Neisseria meningitidis. Standardised cell lysates were separated by SDS-PAGE, followed by transfer to a membrane and the Western immunoblots were developed using anti-His x6 HRP conjugate antibody (Panel A) or rabbit anti-LptA antibody (Panel B) to detect the presence of LptA::His x6 in the extracts. Lanes were: Lane 1: NEB ColorPlus prestained protein molecular weight standard; Lane 2: N. meningitidis strain NMB; Lane 3: NMB expressing LptA::His x6 from pCMK1001 (CKNM216); Lane 4: NMBDlptA::aadA; Lane 5: E. coli JCB571 expressing LptA::His x6 and EcDsbA; Lane 6: E. coli JCB571 expressing EcDsbA, and Lane 7: LptA::His x6 purified from neisserial membranes prepared from CKNM216 using nickel columns. NOTE: ColorPlus prestained protein molecular weight standard migrates aberrantly such that the intact 60 kDa LptA::His x6 appears smaller (refer to Figure S1). (TIF)