XpsD, an outer membrane protein required for protein secretion by Xanthomonas campestris pv. campestris, forms a multimer.

XpsD is an outer membrane lipoprotein, required for the secretion of extracellular enzymes by Xanthomonas campestris pv. campestris. Our previous studies indicated that when the xpsD gene was interrupted by transposon Tn5, extracellular enzymes were accumulated in the periplasm (Hu, N.-T., Hung, M.-N., Chiou, S.-J., Tang, F., Chiang, D.-C., Huang, H.-Y. and Wu, C.-Y.(1992) J. Bacteriol. 174, 2679-2687). In this study, we constructed a series of substitutions and deletion mutant xpsD genes to investigate the roles of NH- and COOH-terminal halves of XpsD in protein secretory function. Among these secretion defective xpsD mutations, one group (encoded by pCD105, pYL4, pKdA6, and pKD2) caused secretion interference when co-expressed with wild type xpsD, but the other (encoded by pMH7, pKdPs, and pKDT) did not. Cross-linking studies and gel filtration chromatography analysis indicated that the wild type XpsD protein forms a multimer in its native state. Similar gel filtration analysis of xpsD mutants revealed positive correlations between multimer formation and secretion interfering properties exerted by the mutant XpsD proteins in the parental strain XC1701. Those mutant XpsD proteins (encoded by pCD105, pYL4, pKdA6, and pKD2) that caused secretion interference formed multimers that are similar to the wild type XpsD multimers and those (encoded by pMH7, pKdPs, and pKDT) that did not formed smaller ones. Furthermore, gel filtration and anion exchange chromatography analyses indicated that the wild type XpsD protein co-fractionated with XpsD(Δ29-428) or XpsD(Δ448-650) protein but not with XpsD(Δ74-303) or XpsD(Δ553-759) protein. We propose that the mutant XpsD(Δ29-428) protein caused secretion interference primarily by forming mixed nonfunctional multimers with the wild type XpsD protein in XC1701(pCD105), whereas the mutant XpsD(Δ74-303) did so by competing for unknown factor(s) in XC1701(pYL4).

XpsD is an outer membrane (OM) 1 protein of Xanthomonas campestris pv. campestris required for the secretion of extra-cellular proteins with a cleavable NH 2 -terminal signal peptide (1)(2)(3). It is likely to be involved in the second step of a two-step secretory pathway, the type II pathway (4 -6). Mutations in the xpsD gene caused the accumulation of extracellular enzymes in the periplasm (1). These enzymes are probably exported from cytoplasm in the first step via a Sec-like pathway (7). Homologues of XpsD are widespread in Gram-negative bacteria. Among them, the PulD protein of Klebsiella oxytoca (8), the OutD protein of Erwinia chrysanthemi (9,10), the OutD protein of Erwinia carotovora (11), the XcpQ protein of Pseudomonas aeruginosa (12),and the ExeD protein of Aeromonas hydrophila (13) have been shown to be involved in the same type of secretion pathway. In addition, between 11 to 13 more genes in each of these bacteria are required for the second step of secretion (5,6). Some other XpsD homologues, however, are involved in a different type of secretory pathway, the type III pathway, for proteins without a cleavable NH 2 -terminal signal peptide. These XpsD homologues, which include the YscC protein of Yersinia enterocolitica (14), the HrpH protein of Pseudomonas syringae (15), the MxiD protein of Shigella flexneri (16), and the InvG protein of Samonella typhimurium (17), work with a different set of accessory proteins (6). There are other proteins which share significant sequence homolgy with XpsD, such as the pIV proteins of the filamentous phages f1, I2-2, and Ike of Escherichia coli (18 -20), the pIV protein of the filamentous phage pf3 of P. aeruginosa (21), the PilQ protein of P. aeruginosa (22), and the OrfE protein of Haemophilus influenzae (23). These proteins are required, respectively, for filamentous phage release (24,25), pili assembly (26), and genetic transformation (23,26). Alignment of all these protein sequences revealed the strongest homology at their COOH termini for approximately 200 amino acid residues (22,27).
Recent studies by Ká zmierczak et al. (28) indicated that pIV, which is required for filamentous phage release, appeared to form a stable complex composed of 10 -12 monomers. Moreover, the OutD protein of E. chrysanthemi, an XpsD homologues required for type II secretion, was co-precipitated with pIV by antibody against pIV (28), suggesting a complex formation between the two proteins. OutD is also highly homologous to pIV at the COOH-terminal end, implicating the involvement of this region in the mixed complex formation (28).
Secondary structure prediction indicated that XpsD is rich in ␤-structures. At least 14 amphipathic ␤-strands were predicted based upon Jä hnig's algorism (29). Near the middle of XpsD, there is a region of more than 100 amino acids that is longest compared with other XpsD homologues. It is rich in Gly and Ser, which is observed in only one other XpsD homologue, i.e. the OutD protein of E. carotovora (11). This long loop divides the entire XpsD into two domains: the upstream NH 2 -terminal domain terminates at the 330th amino acid residue and the downstream COOH-terminal domain begins at the 440th residue.
In this report, we show that XpsD also forms multimeric complex as pIV does. In order to find out what roles do NH 2and COOH-terminal domains play in the complex formation, we constructed a series of deleted xpsD mutants, examined each for secretion interference in presence of wild type xpsD gene and conducted gel filtration and anion exchange chromatography analyses on them.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmid Constructs-All the X. campestris pv. campestris strains used in this study were from our previous work (1). The parental strain XC1701 is a spontaneous rifampicin-resistant mutant from a natural isolate XC17. Both XC1708 (xpsD::Tn5) and XC17433 (⌬xpsE,F,G,H,I,J,K,L,M,N,D) are Tn5-derived mutants of XC1701. The detailed structures of other xpsD mutations are depicted in Fig. 1. All of them were cloned in a broad host range plasmid pCPP30 (IncP replicon, tet r ), which was kindly provided by D. Bauer of Cornell University. The xpsD-phoA fusion gene on pKD20 plasmid was constructed by inserting the signal peptide-less phoA gene from pSWFII (a kind gift from J. Beckwith; Ref. 30) into the XbaI site of pKD2. Escherichia coli strain BMH71-18 (thi supE ⌬(lac-proAB) mutS::Tn10 (FЈproAB lacI q lacZ⌬M15)) used in site-directed mutagenesis was purchased from Promega Corp. All plasmids were maintained in E. coli Site-directed Mutagenesis-Site-directed in vitro mutagenesis with the Altered Sites System purchased from Promega Corp. was used to generate new XbaI sites in the xpsD gene. The xpsD⌬29 -428 and xpsD⌬414 -759 mutations were constructed in our previous work (3). For the construction of the xpsD⌬545-553 mutation, the following two primers were used: 5Ј-TTTTTCCAAGAAATTCTAGAGCGACGC-CATTGC-3Ј for the upstream XbaI site and 5Ј-CTAGCGCACTA-ATATCTAGAGCTGCATTTTTTCC-3Ј for the downstream XbaI site.
Electroporation-X. campestris pv. campestris cells grown to an OD 600 (optical density at 600 m wavelength) of 0.3-0.5 were washed with sterile water to remove exopolysaccharide. One and one-half ml of the washed cells were resuspended in 100 -200 l of sterile water, into which 100 ng of plasmid DNA (in 5-10 l) was added. Electroporation was performed three times with the Gene Pulser/Pulse Controller (Bio-Rad), at 200 ohms, 25 microfarads, and 2.5 kV in a cuvette with a 0.4-cm gap. The shocked cells were diluted in 1 ml of L broth and incubated at 28°C for 3-4 h before plating on L agar containing 100 g/ml rifampicin, 15 g/ml tetracycline, and 50 g/ml kanamycin.
Plate Assays for Protein Secretion-Cells grown on L agar were toothpicked onto XOL agar (1) plus 0.2% (w/v) starch and XOL agar plus 0.1% (w/v) skimmed milk for assaying extracellular ␣-amylase and protease, respectively. Subcellular Fractionation-The supernatant collected from centrifuging cultured cells was saved as extracellular fraction and concentrated by cold trichloroacetic acid precipitation (3). The other subcellular fractions, i.e. periplasmic, membranous, and cytoplasmic fractions, were prepared following the procedures of Hu et al. (3). Western blot and immunodetection performed on protein in each fraction was according to the procedures of Hu et al. (3).
Antisera-Antisera against XpsD and ␣-amylase were from our previous work (2,3). For preparing antiserum against the XpsD(⌬29 -428) protein, an XbaI-EcoRI fragment from pCD105 was inserted into pUC19 (31). A protein with an apparent molecular mass of 44 kDa overexpressed in the isopropyl-␤-D-thiogalactopyrnoside-induced E. coli culture was collected from Coomassie Blue-stained SDS-polyacrylamide gels and injected into rabbit for raising antibody, following the standard procedures (32).
Protein Cross-linking Reaction in Vitro-Water-washed XC1701 cells resuspended in 10 mM Tris-HCl, pH 7.5, were disrupted by twice passing through French press (at 18,000 -20,000 lb/in 2 pressure and 1 drop/2-4-s flow rate), followed by centrifugation at 5,000 rpm for 15 min at 4°C to remove unbroken cells. The membrane collected as pellet from centrifugation at 30,000 rpm for 50 min at 4°C was resuspended in buffer A (20 mM Tris-HCl, pH 8.0, 10 mM EDTA, 2% Triton X-100) and incubated at 4°C for 30 min. The supernatant collected from centrifugation at 11,000 rpm for 10 min at 4°C was saved as the Triton X-100 membrane extract, into which glutaraldehyde was added to a final concentration of 2.5 mM and incubated for 10 min at 37°C.
Protein Cross-linking in Vivo-Glutaraldehyde in various concentrations was added to a stationary phase culture of XC17433(pKC118) in L broth and incubated shaken for another 10 min at 28°C. The treated cells were collected by centrifugation at 12,000 rpm for 10 min at 4°C and resuspended in 20 l of SDS-polyacrylamide gel electrophoresis sample buffer and analyzed by immunoblotting.
DE52 and Mono Q Anion Exchange Chromatography-XC17433 or XC1701 harboring various mutant xpsD genes on pCPP30 was grown in L broth to stationary phase, washed in water to remove exopolysaccharide, and resuspended in 10 mM Tris-HCl, pH 7.5. Triton X-100 membrane extract was chromatographed by passing through a DE52 column (Whatman, 1-ml size) or an FPLC Mono Q column (Pharmacia Biotech Inc., HR 5/5, 1-ml size) pre-equilibrated with Buffer A. After extensively washing with Buffer A, the XpsD protein was eluted with Buffer A containing 0.2 M NaCl or a 0 -1 M NaCl linear gradient. The collected fractions (0.5 ml each) were analyzed by immunoblotting.
Gel Filtration Chromatography-Fractions containing the XpsD protein eluted from DE52 column were collected and precipitated by the addition of solid ammonium sulfate to a final concentration of 70%. The pellet collected from centrifugation was dissolved in a small volume of Buffer B (20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.2 M NaCl, 1% deoxycholic acid) and chromatographed on FPLC Superdex HR-200 (Pharmacia Biotech Inc., 25-ml size) which was pre-equilibrated with Buffer B. Fractions were collected and analyzed by SDS-polyacrylamide gel electrophoresis followed by immunoblotting. The column was calibrated with the following molecular mass standards: thyroglobulin (669 kDa), apoferritin (443 kDa), ␤-amylase (200 kDa), bovine serum albumin (66 kDa), and carbonic anhydrase (29 kDa).
Complementation Test of the xpsD Mutations-Each mutant xpsD gene was inserted downstream to the lac promoter on pCPP30 plasmid and introduced into XC1708 (xpsD::Tn5) via electroporation. The transformants selected on tetracycline plate were examined for ␣-amylase and protease secretion (data not shown). Moreover, the fate of ␣-amylase in all these transformants was investigated by immunoblot analysis of the extracellular, periplasmic, cytoplasmic, and membranous fractions with antibody against ␣-amylase. Results from both experiments indicated that only XC1708(pKU2) and XC1708-(pCD103) secreted ␣-amylase extracellularly as well as the parental strain XC1701 and XC1708(pKC118) (Figs. 1 and 2). All of the other XC1708 transformants and XC1708 itself accumulated ␣-amylase in the periplasm.
Stability of Mutant xpsD Gene Products-In order to determine that the negative results of the complementation tests were not due to protein instability, we performed immunoblot analysis with antibody against XpsD on total cell extracts of XC1708 (data not shown) and on partially purified proteins made in XC17433 (eluted from DE52 chromatography). XC17433 is a mutant strain which lacks the entire xps gene cluster on its chromosome. It is apparent that all mutant xpsD gene products examined, except XpsD(⌬29 -428) (data not shown), were detected, as well as the wild type XpsD synthesized from pKC118 (Fig. 3). However, the major protein bands observed for XpsD(⌬414 -759) (with estimated molecular mass of 43 kDa) and XpsD(⌬553-759) (with estimated molecular mass of 58 kDa) migrated faster than expected and appeared broader than the other XpsD proteins. Both observations indicated that these two proteins are probably less stable than the others. Only small amounts of degradation products were noted for XpsD(⌬74 -303) and XpsD(A553L/I554D) (Fig. 3). The weak intensity of the protein band observed for XpsD(⌬545-553) was probably due to low sample amount, not representative of protein instability. This interpretation is supported by the absence of degradation products from XpsD(⌬545-553) analyzed on gel filtration chromatography (Fig. 5).
Interference of the Parental Strain Protein Secretion by the Presence of Mutated xpsD Gene-When these mutated xpsD genes were introduced into the parental strain XC1701 and tested for extracellular protein secretion, it was observed that some mutant genes exerted negative dominance over the chromosomally located xpsD ϩ gene (Fig. 1). They include those encoded by pCD105, pYL4, pKdA6, and pKD2. On the other hand, no apparent secretion interference was observed with three other nonfunctional mutant XpsD proteins encoded by pMH7, pKdPs, and pKDT.
In Vivo and In Vitro Cross-linking of Wild Type XpsD-The negative dominance of some xpsD mutations suggested possible involvement of protein-protein interactions between wild type and mutant XpsD proteins. We performed protein cross-linking on the cell membrane prepared from XC1701 cells (in vitro) as well as on growing cultures of XC17433(pKC118) (in vivo). XpsD could be cross-linked in both cases by glutaraldehyde as detected on immunoblots (Fig. 4, arrow). While the former result suggested that XpsD is likely to be part of a multimeric complex, the latter further suggested that XpsD could be crosslinked in absence of all the other Xps proteins encoded by the xps gene cluster.
Next, we wanted to find out if those mutant XpsD proteins that did not exhibit negative dominance could not form multimeric complex with the wild type XpsD protein. As expected, the XpsD(⌬553-759) protein appeared between 21 and 24 min on FPLC Superdex HR-200 column when it was co-expressed in XC1701(pKDT) culture with the wild type XpsD protein, which appeared in near void volume (15-16 min) (Fig. 6A). This result indicated that the XpsD(⌬553-759) protein probably did not form a complex with the wild type XpsD protein, consistent with our model of negative dominance based on protein-protein interaction (Fig. 8B). However, the XpsD(⌬448 -650) protein synthesized in XC1701(pKdPs) co-migrated with the wild type XpsD protein (Fig. 6B). This result suggested that the XpsD(⌬448 -650) protein formed a complex, either with the wild type XpsD protein, or with other protein (including itself). The former possibility contradicts with the observation that the XpsD(⌬448 -650) protein did not interfere with secretion in XC1701, unless we assume that the heteromultimer of wild type XpsD and XpsD(⌬448 -650) is functional in protein secretion or enough wild type XpsD homomultimers were formed to support normal protein secretion in XC1701(pKdPs) (Fig. 8D).
Analysis of XpsD and Its Secretion Interfering Mutants on Anion Exchange Chromatography-FPLC Mono Q chromatography was used to further investigate possible interactions between the wild type and mutant XpsD proteins. Triton X-100 membrane extract was passed through a Mono Q column and eluted with a 0 -1 M NaCl linear gradient and analyzed by immunoblots. The wild type XpsD protein was eluted in frac-tions 11-12 (Fig. 7A), whereas XpsD(⌬29 -428) and XpsD(⌬74 -303) proteins were eluted in fractions 9 and 10 (Fig. 7A). When the membrane extract prepared from XC1701(pCD105), which contained both the wild type XpsD and XpsD(⌬29 -428), was analyzed, the two proteins exhibited the same elution profile (Fig. 7B). On the other hand, the XpsD(⌬74 -303) protein produced by XC1701(pYL4) was separated from the wild type protein (Fig. 7C). This results suggested the complex formation of the wild type XpsD with XpsD(⌬29 -428), not with XpsD(⌬74 -303). The formation of nonfunctional heteromultimers of the wild type XpsD and the former mutant protein could explain why pCD105 interfered with protein secretion in XC1701 (Fig. 8A). Apparently, the secretion interference exerted by pYL4 could not be mediated by the formation of nonfunctional heteromultimers of XpsD(⌬74 -303) and the wild type XpsD, unless we assume that such heteromultimers are not stable on the anion exchange column. Or else, non-functional homomultimers of the XpsD(⌬74 -303) protein are in competition with the functional wild type XpsD homomultimers for some unknown factor(s) (Fig. 8C). DISCUSSION All the xpsD truncated mutants inspected in this study were defective in the secretion of ␣-amylase and protease, causing accumulation of ␣-amylase in the periplasm. These XpsD proteins can be divided into two groups: Group I, including XpsD(⌬29 -428) (pCD105), XpsD(⌬74 -303) (pYL4), and XpsD(⌬545-553) (pKdA6), displayed negative dominance over the wild type XpsD protein, and Group II, including XpsD(⌬414 -759) (pMH7), XpsD(⌬448 -650) (pKdPs), and XpsD(⌬553-759) (pKDT), did not. In this study we found good correlation between negative dominance and the formation of XpsD multimers: all Group I proteins formed multimers that appeared in the near void volume, whereas all Group II proteins did not. This indicates that Group I proteins exerted negative dominance by interacting with the wild type XpsD protein. This was supported by our observations that the XpsD(⌬29 -428) protein co-fractionated with the wild type XpsD protein, whereas the XpsD(⌬553-759) protein did not.
The amino acid sequences common in the Group I proteins are residues 1-28, 429 -544, and 554 -759. At the NH 2 terminus, a conserved lipoprotein signal peptidase cleavage site (--LLAG2C--) (33) is located between residues 21 and 22. Fatty acylation of XpsD has been demonstrated with [ 3 H]palmitate labeling of wild type XpsD (3). This suggested that wild type XpsD is probably cleaved between residues 21 and 22, followed by fatty acylation at the NH 2 -terminal cysteine residue. It is thus unlikely that the NH 2 -terminal amino acid residues 1-28 play any significant role in multimer formation. On the other hand, in the amino acid sequences spanning residues 429 -544 and 554 -759, which comprise almost half of the entire XpsD, eight amphipathic ␤-strands were predicted. The region between residues 545 and 553 was predicted to be within a loop based upon Jä hnig's algorism (29). The former two regions are either absent entirely in the XpsD(⌬414 -759) protein or partially in XpsD(⌬448 -650) and Xps(D⌬553-759) proteins, none of which formed functional multimers by themselves or affect protein secretion in XC1701. We propose that these two regions (429 -544 and 554 -759) are required for multimer formation. Although the mutant XpsD(⌬29 -428), XpsD(⌬74 -303), XpsD(⌬545-553), and XpsD(A553L/I554D) proteins formed multimers by themselves, all of them are not functional in protein secretion. We suggest that the regions covered within residues 74 -303 and the amino acid residues 553A, 554I, albeit unrelated with multimer formation, is important for protein secretion.
Due to the small size of deoxycholate micells (28,34), we carried out gel filtration chromatography in the presence of deoxycholate. Wild type XpsD eluted in the near void volume with an estimated molecular mass of approximately 1,000 kDa, suggesting that it forms stable multimeric complexes. At this stage we do not know whether XpsD forms homomultimers or heteromultimers. A homomultimer of XpsD of this size would contain at least 12 mature XpsD polypeptides (77 kDa each). This is comparable with the filamentous phage pIV multimers, estimated to be 10 -12-mers (28).
A gated channel formed by pIV was proposed for releasing of assembled filamentous phage particles (25). The COOH-terminal 200 amino acid sequence of XpsD are highly homologous to pIV and other analogous proteins (22,27). Moreover, XpsD co-expressed with pIV was precipitated with antibody against pIV, which does not cross-react with XpsD 2 . When we introduced pIV gene into XC1701, we clearly observed secretion interference. 3 Both results suggested complex formation between XpsD and pIV. In this study we provide evidences for the first time to show that an OM component of one of the terminal branches of general secretion pathway (5) probably forms stable complex of a minimal 12-mer. An OM secretion channel is anticipated for secreted proteins that are translocated across OM as disulfide-bonded molecules (35)(36)(37)(38)(39).