The Thioredoxin Domain of Neisseria gonorrhoeae PilB Can Use Electrons from DsbD to Reduce Downstream Methionine Sulfoxide Reductases*

The PilB protein from Neisseria gonorrhoeae is located in the periplasm and made up of three domains. The N-terminal, thioredoxin-like domain (NT domain) is fused to tandem methionine sulfoxide reductase A and B domains (MsrA/B). We show that the α domain of Escherichia coli DsbD is able to reduce the oxidized NT domain, which suggests that DsbD in Neisseria can transfer electrons from the cytoplasmic thioredoxin to the periplasm for the reduction of the MsrA/B domains. An analysis of the available complete genomes provides further evidence for this proposition in other bacteria where DsbD/CcdA, Trx, MsrA, and MsrB gene homologs are all located in a gene cluster with a common transcriptional direction. An examination of wild-type PilB and a panel of Cys to Ser mutants of the full-length protein and the individually expressed domains have also shown that the NT domain more efficiently reduces the MsrA/B domains when in the polyprotein context. Within this frame-work there does not appear to be a preference for the NT domain to reduce the proximal MsrA domain over MsrB domain. Finally, we report the 1.6Å crystal structure of the NT domain. This structure confirms the presence of a surface loop that makes it different from other membrane-tethered, Trx-like molecules, including TlpA, CcmG, and ResA. Subtle differences are observed in this loop when compared with the Neisseria meningitidis NT domain structure. The data taken together supports the formation of specific NT domain interactions with the MsrA/B domains and its in vivo recycling partner, DsbD.

Although many examples of the fusion of the MsrA and MsrB domains into a single, multifunctional polyprotein exist (1,2), the PilB proteins from Neisseria gonorrhoeae, Neisseria meningitidis, and Fusobacterium nucleatum are unique. The PilB proteins (also known as the MsrA/B polyprotein) are composed of three domains. The N-terminal (NT) 3 domain functions as a thioredoxin-like disulfide reductase (3). The middle methionine sulfoxide reductase A (MsrA) and C-terminal methionine sulfoxide reductase B (MsrB) domains function with opposite substrate stereospecificity (2,4,5). MsrA reduces or repairs the S-form of methionine sulfoxide, whereas the MsrB domain is specific for the R-form. The polyprotein is thought to help Neisseria survive the burst of reactive oxygen species generated as a protection mechanism within the genitourinary tract of the host. The N terminus of PilB contains a signal sequence that is thought to target the protein to the outer membrane, thus orienting the protein in the periplasmic space (6). Removal of the targeting signal or production of the protein from an internal ribosome binding site results in cytoplasmic localization. In this location the antioxidant properties of the MsrA-MsrB domains appear to be lost when the organism is challenged with exogenous hydrogen peroxide (6).
The three-dimensional structures and catalytic mechanisms of individual MsrA and MsrB domains from several organisms have been characterized and recently reviewed (2,5,(7)(8)(9)(10)(11)(12). The last step of the Msr reaction requires the reduction of the disulfide bond within the Msr molecule to regenerate the active site for additional rounds of catalysis. When the Msr protein is present in the cytoplasm, the reduction is carried out by thioredoxin (Trx). Trx is then recycled by thioredoxin reductase (TrxR) at the expense of NADPH. For PilB, which is periplasmic, it appears that the reduction of the MsrA and MsrB domains is performed by the NT domain. The structure of the NT domain from N. meningitidis exhibits the Trx fold (13). Like many other proteins with a Trx fold, the NT domain presents a Trp-Cys-Xaa-Xaa-Cys catalytic motif that undergoes oxidation-reduction cycling. The redox potential of this motif (Ϫ227 to Ϫ232 mV) from N. meningitidis PilB is slightly more oxidizing than the redox potential of Escherichia coli Trx (eTrx, Ϫ270 mV) (3).
The recombinant NT domain from N. meningitidis has been shown to reduce the individual, oxidized MsrA and MsrB domains similar to eTrx (3). A unique surface loop near the active site was identified (13). The observation that the NT domain cannot interact with the E. coli thioredoxin reductase (eTrxR) also suggested that this structural feature may be involved in specific interactions with NT-MsrA/B domains and its recycling partner. It is still unclear, however, if the NT domain of PilB can reduce the downstream MsrA and MsrB domains in the polyprotein context. Moreover, the source of electrons used by the NT domain to reduce the downstream Msr domains is unknown (10,11). In an effort to address these questions, we have compared the efficiency of the methionine sulfoxide reduction process for PilB from N. gonorrhoeae with constructs of differing lengths and Cys to Ser mutations. We have also determined the 1.6-Å crystal structure of the NT domain from N. gonorrhoeae, identified the DsbD protein as a potential in vivo electron donor, and found that proteins homologous to DsbD, the NT domain, and the Msrs are next to each other in the genomes of some organisms, suggesting that they are part of the same methionine sulfoxide repair pathway.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The full-length PilB protein from N. gonorrhoeae and domain variants were purified from E. coli using an N-terminal His tag within the pET-28b vector from Novagen as previously described (7). The constructs contained the following residues: full-length PilB, 23-522; NT domain, 23-182; MsrA domain, 183-362; MsrB domain, 375-522; and the MsrA-MsrB fusion, 183-522. Residue numbering is based on starting from the initiator methionine residue. For the constructs that contained the NT domain, the removal of the first 22 residues deleted the hydrophobic targeting signal. The proteins were treated with biotinylated thrombin to remove the His tag and purified via HiLoad Super-dex75 size-exclusion and Resource Q columns. The desired fractions were pooled, dialyzed against 20 mM Hepes, pH 7.5, 100 mM NaCl, concentrated with a Centriprep YM10 concentrator (Amicon), aliquoted, flash frozen with liquid N 2 , and stored at Ϫ80°C. The Cys to Ser variants of each of the PilB domains and the Leu double mutant described below were generated using the QuikChange site-directed mutagenesis kit from Stratagene. The L38M/L41M double mutant of the NT domain of PilB was used to generate the protein containing selenomethionine for independent phase determination. Selenomethionine was incorporated into the Met double mutant by the growth of cells in minimal media and the repression of endogenous Met synthesis prior to induction (14).
E. coli MsrA was expressed in E. coli as a glutathione S-transferase fusion protein and purified on a glutathione-agarose column as previously described (15). The N-terminal glutathione S-transferase was removed by treatment with thrombin. E. coli MsrB (yeaA) was also produced as a poly-His fusion and puri-fied in a similar manner to PilB proteins described above. In this case, however, the His tag was not removed (2). The ␣ and ␥ domains of E. coli DsbD were overexpressed, purified, and quantified as in Collet et al. (16) using the following extinction coefficients (⑀ 280 , in M Ϫ1 cm Ϫ1 ): 20,340 for ␣ and 8,370 for ␥. The remaining protein concentrations were determined using the Bio-Rad protein assay kit using bovine serum albumin as the protein standard.
Met(O) Reductase Activity-The specific activity for the reduction of the Met(O) was determined as previously described using tritium-labeled N-acetyl-L-[ 3 H]methionine sulfoxide, N-Ac-L-[ 3 H]MetRSO, a racemic mixture of the R-and S-forms of methionine sulfoxide (17). Briefly, the 30-l reactions con- Electron Flow from E. coli DsbD-The reduced forms of the ␣ and ␥ domains of DsbD were made just prior to the experiment by incubating the proteins with 10 mM DTT at 37°C for 30 min. The oxidized form of the NT domain was generated by incubating the protein with 20 mM oxidized glutathione at 37°C for 20 min. For both of the treatments above, the excess DTT and oxidized glutathione were removed by passing the proteins through a PD-10 desalting column equilibrated with 25 mM sodium/potassium phosphate, pH 8, 0.5 mM EDTA. To test the electron flow pathway, stoichiometric amounts (1 M) of the ␥ domain of DsbD in the reduced state and the NT domain of PilB in the oxidized state were incubated together at 37°C. Then a catalytic amount of the ␣ domain (100 nM) was added to the system. The fluorescence of the NT domain was monitored in a Hitachi F-4800 spectrofluorometer. The wavelength used for tryptophan excitation was 295 nm, and the emission was monitored at 330 nm.
Crystallization, Structure Determination, and Refinement-Crystals of the L38M/L41M mutant of the NT domain containing selenomethionine were obtained by the vapor diffusion method. Equal volumes of protein (30 mg ml Ϫ1 in 20 mM Hepes, pH 7.5, 100 mM NaCl) and well solutions (0.1 M Mes, pH 6.5, 0.2 M NH 4 SO 4 , 26% polyethylene glycol 2000 MME) were mixed and incubated at 4°C. The crystals were transferred to a solution containing a cryoprotectant (0.1 M Hepes, pH 7.0, 0.2 M NH 4 SO 4 , 32% polyethylene glycol 2000 MME, 25% glycerol) prior to data collection at Ϫ170°C. A three-wavelength multiwavelength anomalous dispersion dataset (Table 1) (18). The phases were improved by solvent flattening and maximum likelihood density modification with RESOLVE (18). The resulting 1.6-Å electron density maps were unambiguous, and the autobuild feature of RESOLVE was able to generate 94% (150/159 residues) of the starting model. A comparison of the experimental and composite omit electron density maps facilitated the modification of the model using O (19). The model was initially refined with CNS using alternating cycles of simulated-annealing, positional, and B-factor refinement (20). The last rounds of refinement were performed with REFMAC5 (21). The final model (R work /R free ϭ 18.8/20.9, Table 1) contains residues 32-182 and 90 solvent molecules.

NT Domain More Efficiently Recycles the Downstream MsrA/B Domains When Present in the Polyprotein Context-
Because the NT domain does not interact with eTrxR, one is not able to monitor the change in NADPH absorbance to measure the reduction of Met(O). Therefore, we chose to follow the Msr reaction by monitoring the reduction of tritiated N-acetyl-Met(O) (Fig. 1) (17). DTT, as has been known for some time, stimulates the MsrA and MsrB reactions at concentrations Ͼ1 mM by reducing the active site disulfide bond following the release of product. This finding is substantiated by the increase in activity of the individual MsrA and MsrB domains without the addition of the NT domain (Fig. 1A). The addition of the NT domain, however, enables a dramatic increase in reductase activity even at low DTT levels (0.1 mM, 20-to 50-fold molar excess to NT domain). We rationalize that this stimulation of activity is a consequence of the ability of the NT domain to catalyze the reduction of MsrA or MsrB by DTT. Thus, we hypothesize that the electrons flow in a series of thiol-disulfide exchange reactions from DTT to the NT domain, and finally to the oxidized MsrA or MsrB domains. This idea is supported by an increase of Msr activity upon the addition of a 2-fold molar excess of the NT domain to the fused MsrA-MsrB domain construct (Fig. 1B). The further stimulation of the Msr activity when the NT domain is present in the fulllength protein context suggests that the NT domain may reduce the downstream domains with greater efficacy. In contrast, the activity of the E. coli MsrA and MsrB (Fig. 1C) are not appreciably (ϳ1.5-fold) stimulated by the addition of the NT domain. This latter observation implies that the NT domain has specific molecular interactions with its downstream MsrA and MsrB domains. Moreover, these interactions may not be made as effectively when the NT domain acts intermolecularly due in part to the loss of local concentration effects. One would expect, however, that by increasing the NT domain concentration Ͼ2-fold, the Msr activity of the MsrA/B construct would asymptotically approach that of full-length PilB. The data presented support that the three domains of PilB function independently and that the NT domain appears not to be pre-associated with the MsrA/B domains. The presence of the NT domain in the polyprotein context greatly stimulates the reduction of the downstream Msr domains. As described in more detail below, the linker region between each domain appears to be of sufficient length to allow the free association of the domains during catalysis.
The NT Domain Reduces the MsrA and MsrB Domains Equivalently-The NT, MsrA, and MsrB domains each contain two Cys residues. These residues undergo disulfide bond formation during different stages of the catalytic cycle. To test the efficiency with which the NT domain reduces the downstream Msr domains, both Cys residues of each domain and of different domain combinations as indicated in Fig. 2, were mutated to Ser. The specific activity for each construct was monitored at a low DTT concentration (0.1 mM), a high DTT concentration (10 mM), and with the NADPH-eTrxR-eTrx reducing system (Fig. 2). The domains that are still active are indicated. When only the MsrA and MsrB domains are active (variants C68S/ C71S/C440S/C495S and C68S/C71S/C207S/C349S, respectively), a similar stimulation in activity is observed with 10 mM DTT and the eTrx-based system. The presence of an active NT domain (variants C440S/C495S and C207S/C349S), however, results in a significant increase in specific activity particularly at 0.1 mM DTT, similar to data presented in Fig. 1. The further increase in MsrA and MsrB activity at 10 mM is most likely the result of the direct reduction of the MsrA/B domains by DTT  (Fig. 1A). The lack of an increase in the eTrxR-stimulated activity supports the inability of the NT to interact with eTrxR. The activity of wild-type, full-length PilB (labeled "WT ") at 0.1 mM DTT is essentially the sum of the activities for the Msr domains in the presence of the active NT domain. Therefore, it appears that the NT domain reduces the downstream Msr domains with the same relative efficiency, i.e. the NT domain does not preferentially reduce one Msr domain over another when both are present. The additive relationship for the Msr activity is reasonable given that the NT domain can only reduce one Msr domain at a time and must await reduction by DTT or its in vivo partner.
The ␣ Domain of E. coli DsbD Is Able to Shuttle Electrons from DsbD-␥ to the NT Domain-The PilB protein has been shown to be targeted to the periplasmic space in N. gonorrhoeae (6). This localization prevents the NT and Msr domains of the protein from interacting with the cytosolic NADPH-TrxR-Trx and other reducing systems. Therefore, a source of electrons for the reduction of the NT domain disulfide bond resulting from the reduction of the downstream Msr domains must be present in the periplasm. The periplasm is rich in proteins that exhibit thioredoxin folds and either catalyze disulfide bond formation (e.g. DsbA in E. coli) or disulfide bond isomerization (e.g. DsbC and DsbG) (22)(23)(24)(25). The source of electrons for the latter proteins is the three domain protein DsbD (16,26,27). The membranous middle or ␤ domain shuttles the electrons from the cytosolic NADPH-TrxR-Trx system to the Trx-like ␥ domain present in the periplasm. The electrons are then passed to the periplasmic, Ig-like ␣ domain, which in turn reduces DsbC and DsbG. Therefore, the ␣ domain of DsbD is a likely candidate for the electron donor to the NT domain of PilB.
To test the above hypothesis, we employed the E. coli ␣ and ␥ domains of DsbD instead of the N. gonorrhoeae DsbD domains for the following reasons: (i) the individual DsbD domains from E. coli are well characterized biochemically and structurally (22)(23)(24)(25)(26)(27), (ii) the ␣ and ␥ domains from E. coli show no change in fluorescence upon oxidation-reduction (16), (iii) there is high sequence identity (32%) for the entire DsbD protein between these organisms, and (iv) the N. gonorrhoeae DsbD domains have not been cloned or characterized in any way. Thus, the use of the readily available E. coli DsbD domains allows an evaluation of the ability of the homologous ␣ domain to reduce the NT domain of PilB.
The oxidized form of the separate NT domain was generated by incubation with 20 mM oxidized glutathione (see "Experimental Procedures"). As has been previously shown for the NT domain from N. meningitidis and other Trx-like molecules, the NT domain of PilB exhibits a marked increase in fluorescence upon reduction of the disulfide bond (3). This property was used to monitor the reduction of the disulfide bond of the N. gonorrhoeae NT domain (excitation, 295 nm; emission, 330 nm; ϳ1.9-fold increase in fluorescence, data not shown). The oxidized NT domain (1 M) was pre-mixed with an equivalent amount of the reduced form of the ␥ domain of DsbD and monitored for 250 s (Fig. 3A). The flat baseline that was observed is consistent with the inability of the ␥ domain to donate electrons to the NT domain. The addition of a catalytic amount (0.1 M) of the reduced ␣ domain, however, resulted in   (16). Thus, the small amount of the ␣ domain was able to accept the electrons from the ␥ domain to reduce the NT domain of PilB. The reduction, however, was not complete (ϳ50%), i.e. the fluorescence did not plateau. This observation is most likely due to the similarity in the reduction potentials for all three proteins: Ϫ227 to Ϫ232 mV for the NT domain from N. meningitidis, Ϫ229 mV for the ␣ domain of E. coli DsbD, and Ϫ241 mV for the ␥ domain (3,16). The addition of 0.5 M more of the ␥ domain displaced the equilibrium and did result in complete reduction of the NT domain (data not shown). Moreover, the addition of the reduced ␣ domain resulted in the formation of N-acetyl-Met, the product of the reaction (Fig. 3B). Similar to the ability of the NT domain and eTrx to reduce individual Msr domains, the Trx-like ␥ domain appears to be able to directly reduce the MsrA and MsrB domains of PilB resulting in a low level of Met(O) reduction.
NT Domain Is Different from Other Membrane-tethered Thioredoxins-The 1.6 Å structure of the NT domain from N. gonorrhoeae (residues 32-182 visible) was determined using the multiwavelength anomalous dispersion method via selenomethionine incorporation (Table 1). Because the NT domain, however, does not naturally contain any Met residues, except for the N terminus, Leu 38 and Leu 41 were mutated to Met, an approach often used for structure solution (14,28). The Cys 68 -Pro 69 -Leu 70 -Cys 71 motif was found to be in the reduced state (Fig. 4A) and located near Trp 67 , Tyr 140 , and Pro 141 within the canonical Trx fold (Fig. 4B). In contrast to the N. meningitidis structure (13), the orientation of the Leu 70 side chain as defined by the electron density was unambiguous. The proximity of Cys 68 to Trp 67 is consistent with the observed increase in fluorescence for the reduced state of the NT domains from N. meningitidis (3) and N. gonorrhoeae (data not shown).
A superposition of the N. gonorrhoeae NT domain with E. coli Trx (eTrx) (Fig. 4C) reveals that the NT domain contains an additional ␣-helix and ␤-strand (␣4 -␤4). This insert is found in other membrane-tethered, Trx-like proteins, including TlpA (Fig. 4D), CcmG, and ResA (29 -31). These added features are particularly evident when a sequence alignment is made on a structural basis (supplemental Fig. S1). Moreover, the N. gonorrhoeae NT domain contains an additional eightresidue insertion, Phe 100 -Gly 107 , similar to the NT domain from N. meningitidis (Fig. 4E) (13). The conformation of this loop sequence is, however, subtlety different in the structures. For example, Leu 101 and His 102 at the apex of the loop are further away from Cys 68 and Tyr 67 in the N. gonorrhoeae structure. The side chain of Lys 105 is also present in a completely different rotamer position. In addition to these differences, the N. gonorrhoeae structure also delineated the positions and geometries of additional residues relative to the N. meningitidis structure: residues 32-33 and 177-182. The lack of significant molecular interactions from these residues to the rest of the domain suggests that the core of the NT domain contains residues 23-170 (Fig. 4B). The remaining residues, 171-182, most likely function as a flexible linker to the adjacent MsrA domain.

DISCUSSION
In most organisms the Msr proteins are expressed as separate domains and are reduced by the cytoplasmic NAPDH-TrxR-Trx reduction system. In contrast, the Msr domains of PilB are fused in tandem to an N-terminal, Trx-like domain and located in the periplasm (6). This gene arrangement suggests that the role of the NT domain is to reduce the downstream Msr domains in the polyprotein context. Moreover, this proposition suggests that there is a specific electron donor within the periplasm to the NT domain.
The crystal structures of the NT domain from N. gonorrhoeae and N. meningitidis (Fig. 4E) reveal a unique loop insertion that differentiates it from other membrane-tethered Trx-like proteins, including TlpA, CcmG, and ResA (13, 29 -31). The presence of this loop and undoubtedly other surface feature differences to E. coli Trx support the inability of the NT domain to interact with E. coli TrxR and to activate eMsrA and eMsrB (Fig. 1C). These observations also suggest that the structure of the NT domain may enable specific protein-protein interactions with the downstream Msr domains.
The observed higher specific activity of full-length PilB with the addition of a low concentration of DTT than when the NADPH-eTrxR-eTrx reducing system is present (Figs. 1 and 2) further supports this notion. This idea of restrictive protein partners (i.e. substrate specificity) is reminiscent of the specific interactions between CcmG and CcmH in the cytochrome c maturation process and other periplasmic protein-protein interactions (23,24,30,32).
In order for the NT domain to reduce the downstream Msr domains there must be sufficient flexibility in the linker regions between the domains. The crystal structures of the NT, MsrA, and MsrB domains were examined to estimate the length of the linker between each of the domains. For example, the NT domain core is made up of residues 32-170 ( Fig. 4 and supplemental Fig. S1) based on the comparison to eTrx. The addi- where F o and F c are the observed and calculated structure factors, respectively, for the 95% of the data used in refinement. e R free is calculated as for R work for 5% of the data excluded from refinement. tional residues present in the structure (residues 171-182) exhibit coil and short helical structures that do not directly contact the rest of the protein. Based on the hydrogen-bonding patterns of the N termini of the E. coli and bovine MsrA proteins, the core of the MsrA domain consists of residues 35-228 (corresponds to residues 197-360 for N. gonorrhoeae PilB) (8,9). This core assignment is in accord with the near wild-type activity of the eMsrA protein when residues 1-41 have been truncated (33). For the N. gonorrhoeae MsrB domain, the core structure contains residues 405-506 with residues 377-404 also present in loosely associated coil and ␣-helical structures (2). Based on these considerations the linker between the NT and MsrA domains and the MsrA and MsrB domains could range from 15-27 and 14 -44 residues, respectively. Therefore, it appears that the linker segments between the domains are long enough to impart considerable flexibility to the full-length PilB protein so that the NT domain is not hindered from accessing the Msr domains.
The PilB protein from N. gonorrhoeae has been localized to the outer membrane using diagnostic markers for this compartment (6). Because anchoring to the membrane would restrict the movement of the NT domain, the MsrA and MsrB domains would have to "bend back" to the NT domain to be reduced. The long linker lengths between the domains should make these movements possible. Thus, membrane attachment of PilB could be functionally important as has been observed for CcmG and CcmE (34). It is not clear at this time, however, whether an intermolecular interaction between two PilB proteins predominates in the physiological environment.  periplasm to keep some periplasmic proteins in a reduced state for type I and type II cytochrome c maturation and other disulfide bond shuffling processes (24, 25,31,35). The key protein of this electron transport system is DsbD, a membrane protein with one integral and two periplasmic domains (Fig. 5A). Some bacteria possess a stripped-down version of DsbD, called CcdA, which only has the trans-membrane domain. For DsbD the electrons from the cytoplasmic Trx are shuttled between the three domains in the following progression: the central, transmembrane ␤ domain; the C-terminal, periplasmic ␥ domain, which contains a Trx fold; and the N-terminal ␣ domain, which exhibits an immunoglobulin-like fold (16, 32, 36 -39). Each of the domains contains two Cys thiols. The ␣ domain can transfer its electrons through the reduction of disulfide bonds to DsbC, DsbG, and CcmG. CcdA is able to reduce other proteins such as ResA and HelX, even though it does not contain an ␣ or ␥ domain (31,40). These observations prompted us to test whether the well characterized ␣ domain of E. coli DsbD could reduce the oxidized form of the NT domain of PilB.
As shown in Fig. 3A, the ␥ domain of E. coli DsbD is not able to reduce the oxidized NT domain. Only upon the addition of a catalytic amount of the ␣ domain does reduction occur as illustrated by the increase in tryptophan fluorescence in the active site. It is not surprising that the reduction was not 100% complete, because the reduction potentials of all three proteins are quite similar. Moreover, the addition of the ␣ domain enables the reduction of N-acetyl-Met(O) (Fig. 3B). The low level of activity observed with the addition of the Trx-like ␥ domain is most likely a result of direct reduction of the Msr domains, similar to the ability of the NT domain and eTrx. Thus by inference, the DsbD homolog from N. gonorrhoeae (32% sequence identity) should be able to provide reducing equivalents to the NT domain for the reduction of the downstream Msr domains (Fig. 5A). This finding is not consistent with the conclusions from the superposition of the N. meningitidis NT domain onto the crystal structure of E. coli CcmG/DsbE complexed with the ␣-domain of DsbD (13,19). Because DsbD and its homologs are the only known source of electrons to the periplasm and all homologs of the NT domain seem to be anchored to the inner membrane, the outer membrane localization of the N. gonorrhoeae PilB remains a conundrum.
It is important to note that future studies will be necessary to verify that the ␣ and ␥ domains of the N. gonorrhoeae DsbD protein function in a similar manner to E. coli DsbD. In particular, it will be interesting to know whether the N. gonorrhoeae DsbD domains couple better to the NT domain and to assess the specificity of the DsbD-NT domain interaction by comparison to other known targets of DsbD reduction in E. coli. Currently nothing is known about the Neisseria periplasmic reduction system, including DsbD. Further investigation is clearly warranted. It is tempting to speculate that components of the periplasmic reduction system and PilB in N. gonorrhoeae may be attractive targets for new antimicrobials (13).
These observations encouraged us to further investigate the genomic relationships of Trx, DsbD, MsrA, and MsrB homologs in the available genome databases. Gene cluster analysis was performed using the SEED server from the Fellowship for Interpretation of Genomes (www.theseed.org) (41). The four common cluster motifs are shown in Fig. 5B. As mentioned above, the N. gonorrhoeae and N. meningitidis PilB proteins contain all functional domains fused into one polyprotein. The DsbD gene for these bacteria is not near the PilB gene. In contrast, the F. nucleatum PilB gene is downstream of a CcdA homolog. A similar gene arrangement is seen in Streptococcus pneumoniae and Streptococcus pyogenes except that the Trx-like domain is not fused to the MsrA/B domains. Another variation in this latter theme is the transposition of the MsrA/B fusion protein in front of the CcdA-and Trx-like genes in Actinobacillus actinomycetemcomitans and Haemophilus influenzae. These gene clusters support the functional linkage between the transfer of electrons from a DsbD/CcdA homolog to a Trx-like protein followed by reduction of MsrA and MsrB domains within the periplasm of each representative organism (Fig. 5A). Moreover, all the Trx-like proteins and the MsrA/MsrB fusions mentioned above contain an N-terminal targeting signal for the periplasm as judged by the SignalP server (42).