Functional diversity of three different DsbA proteins from Neisseria meningitidis

Neisseria meningitidis (the meningococcus) is a Gram-negative organism that commonly colonizes the mucosa of the human upper respiratory tract, from where it occasionally invades the bloodstream and the cerebrospinal fluid to cause life-threatening septicaemia and meningitis. With an incidence of 110 per 100 000 in many countries of the industrialized world, meningococcal infection represents a significant and high-profile public health problem, for which an effective solution through vaccination is urgently sought. The interactive biology of the meningococcus with its human host is modulated by a diverse range of bacterial surface structures. Among these, outer-membrane proteins play an important role, and their correct presentation on the bacterial surface is dependent on protein-folding catalysts like the disulphide bond (dsb)-forming disulphide oxidoreductase family. These proteins have been correspondingly implicated in bacterial virulence (Yu & Kroll, 1999), and are widespread amongst Gram-negative bacteria. In Escherichia coli, the family consists of the periplasmic DsbA, B, C, D, E and G redox proteins, each possessing a Cys-X-X-Cys motif at the active site (Missiakas & Raina, 1997). These proteins catalyse the formation of intramolecular disulphide bonds in secreted proteins following one of two pathways: the oxidative pathway (which involves DsbA and DsbB) or the isomerization pathway (which involves the other Dsb proteins). In this way, the Dsb proteins ensure the correct folding and stability of a large number of exported bacterial proteins so that these can achieve their appropriate final structures (Bader et al., 1999). In the oxidative pathway, DsbA rapidly donates the disulphide bond formed between the cysteine residues in its active site to a cysteine pair in a folding polypeptide emerging into the periplasm. DsbA is then converted back to its active oxidized form by membrane-bound protein DsbB, itself reoxidized by quinones of the cytoplasmic membrane (Guilhot et al., 1995). The isomerases DsbC and DsbG act to resolve multiple or mispaired disulfide bonds (Missiakas & Raina, 1997). Their functional oxidation state is maintained by DsbD, which acts as an electron transporter (Chung et al., 2000) and also recycles DsbE, another reductant thioredoxin involved in the maturation of c-type cytochromes (Fabianek et al., 1998).

Examination of the published genome sequence of the N. meningitidis serogroup B strain MC58 suggests that there are three potential dsbA homologues on the chromosome: nmb0278, nmb0294 and nmb0407. While some Salmonella serotypes encode a second dsbA on the virulence plasmid (Bouwman et al., 2003), no other bacterial species has until now been found to carry more than one dsbA gene on its chromosome.

Single putative homologues of dsbB (nmb1649), dsbC (nmb0550) and dsbD (nmb1519) have also been identified in the genome of N. meningitidis MC58 but no homologues of dsbE or dsbG appear to be present. The presence of multiple DsbAs suggests the possibility of specialized functions for each protein, engaged in different aspects of bacterial biology including virulence. Klee et al. (2000) have suggested that the putative DsbA encoded by nmb0294 might modulate proteins important for bacterial survival in the bloodstream. Elsewhere, nmb0294 has been highlighted as a meningococcal-specific gene, absent from Neisseria gonorrhoeae and Neisseria lactamica, and involved in pathogenesis (Perrin et al., 2002).

In this work, we report the characterization of the products of all three putative meningococcal dsbA genes, establishing that they each possess DsbA activity, but differ in their affinity for a range of target proteins in E. coli.

Bacterial strains, plasmids and growth conditions.
The bacterial strains used in this study are described in Table 1. E. coli strain Top10 was used as the recipient for cloning experiments, while the other E. coli strains were used as host strains for complementation assays. The E. coli JCB strains were kindly provided by J. C. Bardwell (Dept of Microbiology and Molecular Genetics, Harvard Medical School, Boston, USA) (Bardwell et al., 1991) and the NLM strains by N. L. Martin (Dept of Microbiology and Immunology, Queen's University, Kingston, Canada) (Turcot et al., 2001). The Neisseria strains used in Southern blotting experiments consisted of strains from our collection and others kindly provided by M. Pizza (IRIS, Chiron SpA, Siena, Italy) (Pizza et al., 2000) and M. Maiden (University of Oxford, UK).

Table 1. Bacterial strains

E. coli strains were grown at 37 °C in LuriaBertani (LB) broth with shaking (200 r.p.m.) or on LB containing 1·5 % agar in plate culture. The growth medium was supplemented with 50 µg ampicillin ml¹, 40 µg kanamycin ml¹, 0·003 % (w/v) X-Gal and 0·5 mM IPTG when appropriate. For use in alkaline phosphatase and β-galactosidase assays, E. coli strains were grown in medium A (10·5 g K₂HPO₄ l¹, 4·5 g KH₂PO₄ l¹, 1 g (NH₄)₂SO₄ l¹, 1 g sodium citrate l¹) supplemented with 0·4 % (w/v) glucose, 1 µg vitamin B1 ml¹ and 1 mM MgSO₄. Neisserial strains were cultured at 37 °C in 5 % CO₂ on GC medium agar supplemented with 1 % IsoVitaleX enrichment (both Becton-Dickinson).

Sequence analysis.
Sequences were compared using the Darwin (Data Analysis and Retrieval With Indexed Nucleotide/peptide Sequences) resource from the Computational Biochemistry Research Group () (Gonnet et al., 2000). InterProScan (), SignalP () and Psort () were used for examination of motifs in protein sequences. BLAST searches were completed at and an investigation of sequences in meningococcal genomes was carried out using the Comprehensive Microbial Resource database at .

Recombinant DNA techniques and nucleotide sequencing.
Unless otherwise stated, recombinant DNA techniques were carried out as described elsewhere (Sambrook et al., 1989). Chromosomal DNA, plasmids and RNA were extracted using the appropriate Qiagen kits.

The putative dsbA genes from N. meningitidis MC58 were amplified by PCR using the primer pairs for nmb0278, nmb0294 and nmb0407 shown in Table 2. E. coli dsbA was amplified from the genome of strain JCB570 with primers Ec_dsbAF and Ec_dsbAR (Table 2). Amplified fragments were cloned into the vector pGEM-T (Promega) to create plasmids pdsb1, pdsb2, pdsb3 and pEcdsbA, respectively. Inserts were sequenced on an ABI 3100 automated DNA sequencer (Applied Biosystems International) using the phage M13 forward- and reverse-sequencing primers (Table 2).

Table 2. Primers

Southern blotting.
ClaI-digested neisserial chromosomal DNA was separated by agarose gel electrophoresis, and the fragments transferred to a positively charged nylon membrane (Hybond-N; Amersham). DNA probes labelled with DIG-11-dUTP (Roche) were generated by PCR using primer pairs for nmb0278, nmb0294 and nmb0407 (Table 2). Hybridization was carried out at 37 °C overnight in DIG EasyHyb buffer (Roche) containing the denatured probe. After hybridization, the membrane was washed at high stringency. Hybridized probes were detected using a chemiluminescent detection system (Roche) according to the manufacturer's protocol.

Functional assays of DsbA activity
Motility on soft agar.
To assay motility, E. coli strains were grown overnight in broth. An aliquot (100 µl) of the overnight culture was spread on to an LB plate containing a sterile 0·5 cm diameter nitrocellulose filter (type HA; Millipore). A lawn of cells was allowed to grow and the filter was transferred to a soft agar LB plate containing 0·4 % agar and incubated at 30 °C overnight, after which the diameter of growth around the filter disc was measured.

Alkaline phosphatase (AP) assay.
AP activities were measured essentially as described by Brickman & Beckwith (1975). Briefly, bacteria were grown overnight in 20 ml medium A. Cells were incubated on ice for 20 min, pelleted by centrifugation for 10 min at 2350 g and washed twice with 20 ml Tris, pH 8·0. After the last centrifugation, cells were resuspended in 2 ml Tris pH 8·0, and the OD₆₀₀ was measured in a spectrophotometer (Cecil CE2021). The cells were then permeabilized by adding 50 µl 0·1 % (w/v) SDS and 100 µl chloroform and mixed by vortexing. Samples were equilibrated for 5 min at 28 °C after which 500 µl p-nitrophenol phosphate [0·4 % (w/v) in Tris, pH 8·0] was added and the time recorded. The reaction was allowed to proceed at 28 °C until a yellow colour developed. As soon as the colour change was noted, the reaction was stopped by adding 1 ml 1 M KH₂PO₄. Samples were vortexed before the A₄₂₀ and A₅₅₀ were measured. Units of AP activity were calculated from the absorbances as indicated by Brickman & Beckwith (1975).

β-Galactosidase (β-gal) assays.
β-Gal activity was assayed essentially as described by Miller (1972). The identical protocol to the AP activity assays was followed, except that Tris, pH 8·0, was replaced with chilled buffer Z (90 mM Na₂HPO₄, 35 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, 2·7 ml β-mercaptoethanol l¹; pH 7·0), p-nitrophenol phosphate with ONPG [0·4 % (w/v) in buffer Z] and KH₂PO₄ with 1 M sodium carbonate. Units of activity were calculated from the absorbance values as indicated by Miller (1972).

DTT sensitivity.
DTT sensitivity assays were performed essentially as in Sardesai et al. (2003). Complementation of dsbA mutation was detected by the ability to grow on LB agar plates containing 12 mM DTT, a concentration that totally inhibits growth of the E. coli dsbA strain JCB571 but not the wild-type strain JCB570. DTT was freshly dissolved in molten LB agar and the plates were used within 30 min of pouring to prevent oxidation of DTT by air. Single colonies were streaked and growth was observed after an overnight incubation at 37 °C.

Dsb protein family in N. meningitidis
The multiple sequence comparison program PhyloTree (within the Darwin package) was used to put the differences between peptide sequences on a quantitative graphical basis. In the resulting unrooted dendrogram (Fig. 1), pairwise relationships are displayed according to the principle that the length of the branch path joining two peptide sequences is in proportion to their degree of homology, calculated as the PAM (percentage accepted mutations) distance separating two sequences (Gonnet, 1994; Gonnet et al., 1992). Short PAM distances thus signify close similarity of sequence; long distances denote correspondingly divergent sequence. NMB0278, NMB0294 and NMB0407 are more similar to each other in sequence than they are to any other meningococcal putative Dsb protein (NMB1649, putative DsbB homologue; NMB0550, putative DsbC homologue; and NMB1519, putative DsbD homologue), and, as shown graphically in Fig. 1, the three sequences clustered close to E. coli DsbA rather than other E. coli Dsb proteins. This suggests that, at least with respect to primary sequence, NMB0278, NMB0294 and NMB0407 are indeed DsbA-like (and individually less likely to represent straightforward homologues of DsbE or DsbG proteins, both apparently absent from the meningococcal genome).

(15K): Fig. 1. Graphical comparison of Dsb protein sequences. NMB signifies meningococcal protein sequences; E.c. signifies E. coli protein sequences. All sequences were obtained from GenBank.

Comparison of the protein sequences of NMB0278, NMB0294 and NMB0407
An alignment of the three putative meningococcal DsbA sequences with E. coli DsbA is shown in Fig. 2. All three open reading frames commence with N-terminal sequences typical of cleavable signal peptides, suggesting export beyond the cytoplasm. NMB0278 and NMB0294 contain predicted lipoprotein attachment sites (dotted underlined in Fig. 2), suggesting a possible membrane location, while NMB0407, lacking such a motif, is predicted to be periplasmic (Fig. 2). These localizations have since been confirmed in N. meningitidis (Tinsley et al., 2004). All four possess the crucial disulphide oxidoreductase active site motif, Cys-X-X-Cys (shown in bold in Fig. 2). A *VLEFF**F motif directly upstream of this active site (underlined in Fig. 2), characteristic of DsbA proteins, is also present (Martin et al., 1993). The active site of DsbA is made up of two domains separated by 113 aa in the linear sequence (Martin et al., 1993). The second part of the active site, distant from the Cys-X-X-Cys in sequence, is found between residues 166 and 170 in the E. coli DsbA protein. DsbA and thioredoxin sequences are also conserved in this region (Martin et al., 1993) and all possess a proline (at residue 170 in E. coli DsbA), which has been suggested to pack against the Cys-X-X-Cys active site. All three meningococcal proteins also contain this conserved proline residue, and NMB0278, NMB0294 and NMB0407 are identical around this region of the sequence.

(53K): Fig. 2. Comparison between the protein sequences of N. meningitidis NMB0278, NMB0294, NMB0407 and E. coli DsbA. Asterisks represent amino acids that are identical in meningococcal proteins. The conserved active site residues are shown in bold and larger, around both the Cys-X-X-Cys thiol disulphide oxidoreductase site and the second part of the active site. Putative membrane lipoprotein attachment sites detected with InterProScan in the sequences of NMB0278 and NMB0294 are dotted underlined. The cytochrome-c haem-binding site in the NMB0407 sequence is circled.

NMB0278 and NMB0294 both encode a protein with a Cys-Pro-His-Cys motif at the putative active site, as do DsbA sequences from many other organisms (Missiakas & Raina, 1997). They are very similar in size and sequence (231 and 232 aa, respectively; 73 % identity), suggesting that the two genes arose through gene duplication. This is supported by an examination of G+C content of the corresponding genes: nmb0278 and nmb0294 have identical G+C contents (56·47 mol%), higher than the mean for the MC58 genome (53·7 mol%) and considerably higher than that of nmb0407 (49·22 mol%). The NMB0407 sequence is slightly shorter (214 aa) and encodes a Cys-Val-His-Cys motif at the putative active site. It is only 50 % identical in sequence to the other meningococcal proteins, although extended domains of similarity to the other two meningococcal sequences are obvious, particularly around the active site regions (Fig. 2).

Distribution of the putative meningococcal dsbA genes in different neisseriae
Homologues of all three putative dsbA genes from strain MC58 are present in the sequenced genome of strains Z2491 (serogroup A) and FAM18 (serogroup C). However, only homologues of nmb0278 and nmb0407 were found in the genome of N. gonorrhoeae strain FA1090; no nmb0294 homologue is present. The chromosomal context of each gene in MC58 is different and there is no indication as to any specific substrate target(s).

The presence of each putative meningococcal dsbA gene in the genomes of various neisserial strains was further investigated by Southern hybridization. Meningococcal genes were used as probes for hybridization to ClaI-digested chromosomal DNA from a wide range of strains, chosen to represent the diverse serogroup B meningococcal population (Pizza et al., 2000), other serogroups, and the commensal neisseriae. All N. meningitidis strains tested were found to contain one copy of each putative dsbA gene. As found with N. gonorrhoeae, only the nmb0278 and nmb0407 probes hybridized to DNA of the N. lactamica, Neisseria cinerea and Neisseria polysaccharea strains examined. The nmb0294 probe was the only one to hybridize to chromosomal DNA of Neisseria subflava, Neisseria flava and Neisseria flavescens strains. None of the probes bound to chromosomal DNA prepared from strains of Neisseria sicca or Neisseria mucosa (Table 3).

Table 3. Results of Southern blotting experiments ClaI-digested chromosomal DNA was probed with DIG-labelled DNA fragments for nmb0278, nmb0294 and nmb0407. Reactivity to each probe was visualized using a chemiluminescent detection system. E. coli HB101 was used as a negative control.

Functional studies
E. coli strain JCB571 is defective in disulphide bond formation because of the replacement of the dsbA gene by a kanamycin cassette (Bardwell et al., 1991). Plasmids pdsb1, pdsb2, pdbs3 and pEcdsbA were transformed into E. coli JCB571 and tested for the ability to complement the dsbA mutation. As others have reported, growth rates of each strain used in complementation experiments were similar. To confirm that the genes were being expressed, RNA was prepared from E. coli JCB571 harbouring plasmids pdsb1, pdsb2 and pdsb3 and RT-PCR performed using oligonucleotide primers for the nmb0278, nmb0294 and nmb0407 genes, respectively. These experiments indicated that all three meningococcal genes were being expressed in the E. coli dsbA strain (Fig. 3) and, in confirmation, the corresponding proteins were detected in abundance by SDS-PAGE (Fig. 4). Each construct was then examined for DsbA-typical activity and compared with both the wild-type E. coli strain (JCB570) and the dsbA mutant strain carrying no plasmid (JCB571).

(110K): Fig. 3. RT-PCRs on the RNA of E. coli strain JCB571 carrying various plasmids. Reactions were performed on 1 µg RNA of E. coli JCB571 carrying plasmids pdsb1 (lanes 2 and 3), pdsb2 (lanes 4 and 5) and pdsb3 (lanes 6 and 7), with primer pairs nmb0278_5'/nmb0278_3', nmb0294_5'/nmb0294_3' and nmb0407_5'/nmb0407_3', respectively. As a control for DNA contamination, the RNA was treated with RNase prior to the RT-PCR cycle. As expected, these reactions generated no product (lanes 3, 5 and 7). Lane 1, 1 kb DNA ladder marker.

(57K): Fig. 4. Periplasmic and membrane fractions of E. coli JCB571 (dsbA) containing various plasmids. Lanes: 1, pEcdsbA; 2, pdsb1; 3, pdsb3. Bands representing DsbA are indicated by arrows with molecular masses in kDa. Proteins were isolated using the Tris/EDTA method (Willis et al., 1974); 30 µg ml1 protein was loaded on a 12 % SDS-PAGE gel.

NMB0278 restores motility to an E. coli dsbA strain
DsbA is important for flagellar assembly in E. coli, as it catalyses formation of critical disulphide bonds in a protein of the P-ring (FglI) (Dailey & Berg, 1993). The dsbA mutant JCB571 is not able to move on soft agar whereas wild-type JCB570 is motile under the same conditions. We tested the ability of individually cloned genes in plasmids to restore the motile phenotype to JCB571. Only the E. coli DsbA and meningococcal NMB0278 proteins were able to re-establish motility fully comparable to the wild-type strain in our assay (Fig. 5). A partial effect was seen with JCB571 carrying pdsb2 (nmb0294), which was slightly more motile than JCB571.

(64K): Fig. 5. Motility of E. coli strains carrying cloned meningococcal and E. coli dsbA genes on 0·3 % LB agar. Clockwise from top: plate 1 (left): JCB571 carrying pEcdsbA, JCB571, JCB570. Plate 2 (right): JCB571 carrying pdsb1, pdsb2 and pdsb3.

NMB0278 and, to a lesser extent, NMB0294 assist folding of E. coli AP
AP is a periplasmic enzyme that contains two disulphide bonds in its folded conformation, essential for its activity (Kim & Wyckoff, 1991). JCB571 is defective in AP activity because of the general defect in disulphide bond formation (Bardwell et al., 1991). This can be exploited to assess DsbA activity encoded by cloned genes. High and low levels of AP activity were confirmed in wild-type JCB570 and dsbA JCB571, respectively. JCB571 transformed with plasmids encoding E. coli DsbA or meningococcal NMB0278 had an AP activity close to that of wild-type E. coli strain JCB570. NMB0294 partially complemented the defect in E. coli JCB571 but the AP activity of the E. coli dsbA strain was not increased by carriage of pdsb3 (nmb0407) (Table 4).

Table 4. Results of three functional assays Units of AP and β-gal were determined as described previously from cells grown in medium A. For AP and β-gal assays, experiments were performed on three separate occasions (in duplicate) and values given represent the mean of the separate assays±standard deviation from the mean.

Only NMB0294 can catalyse the formation of disulphide bonds in the MalFLacZλ102 fusion protein
A second pair of E. coli wild-type and dsbA strains (NLM142 and NLM117, respectively) was used to investigate a different protein folding function of DsbA. These strains produce the engineered fusion protein MalFLacZλ102, which, in the presence of DsbA (strain NLM142), folds so as to confer a Lac phenotype. A strain lacking DsbA (strain NLM117) is by contrast Lac⁺ (Turcot et al., 2001). β-gal activities of the various E. coli strains are shown in Table 4. The low β-Gal activity of wild-type NLM142 and high activity of the dsbA strain NLM117 were confirmed. Only pEcdsbA and pdsb2 were able to complement the mutation, decreasing levels of β-gal activity of the fusion protein to levels comparable to the dsbA-wild-type strain. Plasmids expressing the other two meningococcal genes poorly complemented the dsbA mutation in this assay.

All three meningococcal gene products re-establish resistance to DTT in an E. coli dsbA strain
The role of thiol disulphide oxidoreductases in maintaining the redox balance of the bacterial periplasm can be demonstrated by challenge with reducing agents such as DTT. E. coli JCB571, lacking DsbA, is incapable of growth on LB agar containing 12 mM DTT. All of E. coli DsbA, N. meningitidis NMB0278, NMB0294 and NMB0407 expressed from plasmids were able to complement the dsbA-null E. coli strain JCB571 (Table 4).

The meningococcus uniquely harbours three genes encoding homologues of the protein-folding enzyme DsbA nmb0278, nmb0294 and nmb0407 in the serogroup B strain MC58. Bioinformatic and biochemical evidence indicates that while all three genes encode DsbA activity, complementing for example the DTT-sensitive phenotype of an E. coli dsbA strain, their protein-folding target specificities are likely to differ. Assays carried out in an E. coli dsbA-null background revealed clear differences between NMB0278 (hereafter DsbA-1) and NMB0294 (hereafter DsbA-2) otherwise by sequence comparison the most alike of the three meningococcal genes. NMB0278 complemented defects in motility and in the folding of AP, while NMB0294 restored the folding of the MalFLacZλ102 fusion protein. In all the activity assays performed, NMB0407 (hereafter DsbA-3) was less effective at complementing the E. coli dsbA mutation than the other two meningococcal proteins. Although by bioinformatic criteria of overall and critical motif sequence similarity (Figs 1 and 2) this gene product can be identified with some confidence as DsbA, differences between DsbA-3 and the other two suggest that their targets may be particularly different. The active site of DsbA-3 lies within the sequence motif Cys-Val-His-Cys-His-His, different from meningococcal DsbA-1/2 or the E. coli DsbA (Cys-Pro-His-Cys-Ala/Tyr-His), and the nature of residues between the active site cysteines influences redox potential and substrate specificity (Grauschopf et al., 1995). Further, the presence of His adjacent to the DsbA-3 active site tetrapeptide uniquely confers on this protein a potential cytochrome-c haem-binding property (Mathews, 1985), suggesting a potential role in the maturation of c-type cytochromes, a process catalysed by DsbE in E. coli and other organisms (Fabianek et al., 1998). Because there is no apparent direct homologue of dsbE in N. meningitidis, DsbA-3 may assume this role with respect to redox-dependent cytochrome-c maturation. It is improbable that NMB0407 is simply a DsbE rather than a DsbA homologue, both on bioinformatic grounds and as DsbE in other organisms does not have DsbA activity (L. Thöny-Meyer, personal communication). Indeed, cloned E. coli dsbE does not complement DTT sensitivity of our E. coli dsbA-null mutant (data not shown). The capacity of nmb0407 to encode other DsbE activities is currently under investigation.

Southern blotting experiments have shown that although N. meningitidis from all pathogenic serogroups carry copies of dsbA-1, dsbA-2 and dsbA-3, the distribution of these three genes varies in other neisserial strains. No single homologue appears essential as the commensal neisseriae carry either dsbA-1 and dsbA-3 or dsbA-2 on its own. In N. flava, N. flavescens and N. subflava, a copy of dsbA-2 alone appears sufficient, suggesting that in these strains, DsbA has a smaller set of substrates compared with the meningococcus or that the target specificity of the version of DsbA-2 found in these organisms is broader. Similarly, the existence of neisserial species that carry only versions of dsbA-1 and dsbA-3 (N. gonorrhoeae, N. lactamica, N. polysaccharea and N. cinerea) suggests that proteins folded by meningococcal DsbA-2 may be absent from these organisms or folded by a different mechanism. It is provocative that N. meningitidis, the neisserial species with maximal invasive pathogenic potential in humans, alone carries three versions of dsbA suggested by our functional assays not to be redundant in their function. We speculate that their action on different target proteins might contribute to the unique virulence of the meningococcus, and we are carrying this forward by transferring single and multiple dsbA mutations into the meningococcal genome and studying their effect on hostpathogen interactive biology.

This work was supported by a generous donation to the Imperial College Trust from the Ralph Sutcliffe Fund.