Present addresses: Laboratoire de Cristallographie et de Cristallogenese des Proteines, Institut de Biologie Structurale Jean Pierre Ebel, Grenoble, France. ‡Department of Hematology, Erasmus Medical Center, Rotterdam, The Netherlands.
The manganese-oxidizing factor of Pseudomonas putida strain GB-1 is associated with the outer membrane. One of the systems of protein transport across the outer membrane is the general secretory pathway (Gsp). The gsp genes are called xcp in Pseudomonas species. In a previous study, it was shown that mutation of the prepilin peptidase XcpA and of a homologue of the pseudopilin XcpT inhibited transport of the factor. In the present study, we describe the genomic region flanking the xcpT homologue (designated xcmT1). We show that xcmT1 is part of a two-gene operon that includes an xcpS homologue (designated xcmS). No other xcp-like genes are present in the regions flanking the xcmT1/xcmS cluster. We also characterized the site of transposon insertion of another transport mutant of P. putida GB-1. This insertion appeared to be located in a gene (designated xcmX) possibly encoding another pseudopilin-related protein. This xcmX is clustered with two other xcpT-related genes (designated xcmT2 and xcmT3) on one side and homologues of three csg genes (designated csmE, csmF and csmG) on the other side. The csg genes are involved in production of aggregative fibres in Escherichia coli and Salmonella typhimurium. A search for XcmX homologues revealed that the recently published genome of Ralstonia solanacearum and the unannotated genome of P. putida KT2440 contain comparable gene clusters with xcmX and xcp homologues that are different from the well-described ‘regular’xcp/gsp clusters. They do contain xcpR and xcpQ homologues but, for example, homologues of xcpP, Y and Z are lacking. The results suggest a novel Xcp-related system for the transport of manganese-oxidizing enzymes to the cell surface.
After iron, manganese is the second most abundant metal in the earth's crust. It occurs in fresh and marine waters and sediments and is a constituent of a large variety of minerals (Tebo et al., 1997). It is an essential trace element for all living organisms as a structural or functional constituent of many enzymes. In most natural environments, the metal undergoes transitions between soluble reduced and highly insoluble oxidized forms. The cycling and bioavailability of manganese thus largely depends on its redox chemistry.
The manganese is generally oxidized extracellularly, and the oxidizing enzymes are thought to be constituents of outer cell coverings, e.g. the sheath of Leptothrix (Emerson and Ghiorse, 1992), the exosporium of Bacillus spores (Tebo et al., 1997) or the outer membrane of Pseudomonas (Okazaki et al., 1997). Using transposon mutagenesis, several phenotypically non-oxidizing mutants of the manganese-oxidizing Pseudomonas putida strain GB-1 have been isolated (de Vrind et al., 1998). Among them were two transport mutants, strains GB-1-008 and -009, in which the manganese-oxidizing activity was accumulated intracellularly. Analysis of the mutation sites indicated that transport of the manganese-oxidizing enzyme across the outer membrane in P. putida GB-1 is mediated by a type II secretion system, also called the general secretory pathway (Gsp) (Brouwers et al., 1998). Interestingly, an enzyme involved in manganese (and iron) reduction in Shewanella putrefaciens also appeared to depend upon a type II secretion system for transport across and insertion into the outer membrane (DiChristina et al., 2002).
Type II secretion is one of several transport mechanisms used by Gram-negative bacteria for the secre-tion of exoproteins (Russel, 1998; Koster et al., 2000; Sandkvist, 2001). It is a two-step process in which the substrates are first translocated over the cytoplasmic membrane by the Sec apparatus (Meyer et al., 1999). This first step is signal sequence dependent and generally requires the substrates to be non-folded. In the periplasm, the exoproteins fold into their (near)-native conformation and are then recognized by proteins of the Gsp machinery, which translocates the substrates across the outer membrane. Recently, it was shown that proteins transported into the periplasm by the TAT (twin-arginine transport) system can also be substrates of the Gsp machinery (Voulhoux et al., 2001, cf. de Groot et al., 1996). The gsp genes and their products have homologues in a wide variety of Gram-negative bacteria and are called xcp in Pseudomonas species (Filloux et al., 1998).
In Pseudomonas aeruginosa, the type II pathway is used by exotoxin A and degradative enzymes such as lipases and proteases (Filloux et al., 1998 and references therein). The Xcp machinery is encoded by at least 12 genes, xcpA and xcpP–Z. The latter are present in a gene cluster, in which xcpP and Q have an opposite orientation with respect to xcpR–Z. The xcpA gene is present at a different locus. The xcpA gene, also called pilD, encodes a (leader) peptidase required for the processing and methylation of the precursor (prepilin) of the type IV pilus subunit PilA. The prepilin peptidase XcpA (PilD) also functions in the processing of the so-called pseudopilins, XcpT–W, which share N-terminal sequence similarity with the prepilin PilA (Nunn and Lory, 1992). XcpX, although lacking the conserved Phe(+1) and Glu(+5) characteristic of the prepilin processing site, is also considered to belong to the pseudopilins and is indeed processed by XcpA (Bleves et al., 1998). The function of the pseudopilins is not known. They have been suggested to assemble into a pilus-like structure, which might aid in pushing the exoprotein through the outer membrane channel (Filloux et al., 1998; Sauvonnet et al., 2000). The latter is thought to be formed by the outer membrane protein XcpQ, a member of the superfamily of secretins (Bitter et al., 1998). The role of the other Xcp proteins in secretion is also speculative. XcpR has a conserved nucleotide binding site (Walker box A; Walker et al., 1982) and may be an ATPase involved in energizing protein secretion (Filloux et al., 1998) or function as a kinase (Sandkvist, 2001).
In the P. putida strain WCS358, the presence of an xcp operon containing xcpP–Z except for an xcpX homologue has been reported (de Groot et al., 1999). In P. putida also, the xcpA gene is present at a different locus in the genome (de Groot et al., 1994). So far, a substrate for the Xcp machinery in P. putida WCS358 has not been identified, but one might be encoded by the first gene of the xcp operon, designated uxpA (de Groot et al., 1999).
In both transport mutants of P. putida GB-1 described above, homologues of xcp genes were shown to be mutated by transposon insertion (Brouwers et al., 1998). The insertion in mutant GB-1-008 appeared to be located in the xcpA gene, based on strong similarity of its product to that of P. putida WCS358 xcpA. The genomic organization flanking both xcpA genes was highly similar, representing part of the pil operon involved in pilus biogenesis (de Groot et al., 1994; Brouwers et al., 1998). The insertion site of mutant GB-1-009 was identified as an xcpT-like gene, based on homology of its product with the XcpT proteins of P. putida WCS358 and P. aeruginosa (Brouwers et al., 1998). The encoded protein product is characterized by a typical pseudopilin N-terminal sequence, which includes a short leader peptide. The location of the xcpT homologue in the genome was not determined.
In the present study, we analysed the genomic location of the xcpT homologue and characterized the mutation site of a third transport mutant of P. putida GB-1. The results strongly suggest that secretion of the manganese-oxidizing factor depends on an Xcp-related transport system other than the regular Xcp machinery. Analysis of the genome sequences of Ralstonia solanacearum and P. putida strain KT2440 indicates that the presence of such putative alternative Xcp-related secretion systems is a general trait in these related bacterial species.
Heterologous complementation of P. putida secretion mutants GB-1-008 and -009
In a previous study (Brouwers et al., 1998), we showed that transposon (Tn5) insertion in a homologue of the xcpA gene in the manganese-oxidizing P. putida strain GB-1 resulted in inhibited transport of the manganese-oxidizing factor across the outer membrane. These data suggest that extracellular manganese oxidation depends on the Xcp-mediated type II protein secretion mechanism, which has been characterized in detail in P. aeruginosa (Bally et al., 1992; Filloux et al., 1998). Clusters of xcp genes have been described in P. putida strain WCS358, P. aeruginosa PAO1 and Pseudomonas alcaligenes Ps93 (de Groot et al., 1999), but their involvement in manganese oxidation could not be tested because this activity could not be detected in these strains (results not shown). Therefore, we performed heterologous complementation experiments to test the involvement of the Xcp machinery in secretion of the manganese-oxidizing factor in P. putida GB-1.
The xcpA product of strain GB-1 is very similar to the corresponding proteins from P. aeruginosa and P. putida strain WCS358 (77% and 82% identity respectively), and it is assumed to function as a prepilin peptidase that processes, among others, the precursors of pseudopilins involved in type II secretion. To confirm the functional homology between the xcpA product of P. putida GB-1 and that of P. putida WCS358, cosmid pPX851 carrying xcpA of P. putida WCS358 was introduced in mutant GB-1-008. Extracellular manganese oxidation appeared to be restored (not shown), indicating that the XcpA proteins from both P. putida strains function in protein secretion.
In contrast, the mutation in strain GB-1-009, carrying a Tn5 insertion in an xcpT homologue (43% and 42% identity between the GB-1 product and the P. putida WCS358 and P. aeruginosa PAO1 XcpTs), was not complemented by cosmids pPX135 and pAX24, containing the complete xcp gene clusters of P. putida WCS358 and P. aeruginosa PAO1 respectively (results not shown). In addition, plasmid pAG115, carrying xcpT of P. putida WCS358 under the control of the inducible tac promoter, which partially complemented a P. aeruginosa xcpT mutation (de Groot et al., 1999), did not restore secretion of manganese-oxidizing activity in mutant GB-1-009 (results not shown). Lack of complementation suggests that the XcpT proteins from the xcp clusters of both P. putida WCS358 and P. aeruginosa PAO1 are not functionally equivalent to the XcpT homologue of P. putida GB-1. Possibly, the latter is part of a secretion system different from the ‘regular’ Xcp machinery. This assumption was tested by localization of the encoding gene and analysis of its surroundings.
Localization of the xcpT homologue involved in transport of the manganese-oxidizing factor of P. putida strain GB-1
To characterize the genetic location of the xcpT homologue identified in mutant GB-1-009, a genomic library (de Vrind et al., 1998) was screened by hybridization with digoxygenin (DIG)-labelled plasmid pPLH6, carrying the 18 kb Tn5-containing EcoRI fragment from mutant GB-1-009 (Brouwers et al., 1998 de Vrind et al., 1998). Hybridizing cosmids were introduced in mutant GB-1-009, and several of them were able to restore the secretion of manganese-oxidizing activity (not shown). The insert of one of the complementing cosmids, pCOS9-20, contained EcoRI fragments of 12, 6.5, 2.3 and 0.7 kb (not shown). Only the 2.3 kb fragment hybridized with pPLH6 and, subcloned in pLAFR3, it was able to complement the GB-1-009 mutation fully (results not shown).
Sequence analysis of the 2.3 kb fragment (subcloned in pUC19 resulting in pHL3) revealed two complete and two partial open reading frames (ORFs; Fig. 1). One complete ORF represents the xcpT homologue, which was previously identified by the analysis of the sequences adjacent to the transposon insertion in mutant GB-1-009 (Brouwers et al., 1998). In this paper, it will be designated xcmT1 (xcm: xcp-related system involved in transport of the manganese-oxidizing factor) to discriminate it from other xcp(T)-like genes (see below). The protein encoded by xcmT1 appears to be very similar to products of many xcpT (gspG)-related genes, of which rsc2310 and rsp0150 of the recently published genome of R. solanacearum (Salanoubat et al., 2002) yielded the highest similarity scores (Table 1).
Table 1. . Comparison of the organization of gsp -related clusters in R. solanacearum chromosome ( rsc ) and megaplasmid ( rsp ) and in P. putida GB-1 and similarities of encoded proteins.
rsc gene number and orientation
Homologous P.putida protein, % similarity
rsp gene number and orientation
Homologous P.putida protein, % similarity
. Putative protein shows 50% similarity to MofC from the mof operon (manganese-oxidizing factor) of L. discophora.
. Putative proteins show homology to multicopper oxidases such as laccases.
. Possibly belonging to one operon; start and stop codons overlap or located close together.
xcmT1 is closely followed (11 bp apart) by the 5′ part of an ORF, which is probably part of the same operon. As only part of this ORF is present on the complementing 2.3 kb fragment, it is possible to conclude that the inactivation of xcmT1 rather than a polar effect on downstream genes caused inhibition of secretion in strain GB-1-009. The 3′ end of the ORF was identified by partial sequence analysis of pPLH6. The gene, designated xcmS , encodes a protein of 43.5 kDa with homology to the GspF proteins of the type II secretion systems, examples of which include P. aeruginosa XcpS (24% identity, 41% similarity; Bally et al., 1992 ) and P. putida WCS358 XcpS (26% identity, 41% similarity; de Groot et al., 1996 ). Highest similarity was found with the putative products of the xcpS ( gspF )-related genes rsc2309 and rsp0149 from R. solanacearum , which are clustered with the aforementioned xcmT1 gene homologues rsc2310 and rsp0150 respectively ( Table 1 ).
Two additional ORFs in the xcmT1/xcmS region appeared to be unrelated to xcp (gsp) genes (Fig. 1). The potential product of the ORF upstream of the xcp cluster is a protein of ≈ 15 kDa, showing homology to the protein family of lytic transglycosylases. Examples of members of this family are the lytic transglycosylase from bacteriophage PRD1 (31% identity, 48% similarity; Koonin and Rudd, 1994) and from Escherichia coli (35% identity, 42% similarity; Engel et al., 1991). The potential lytic transglycosylase from P. putida GB-1 contains the prokaryotic lytic transglycosylase signature (LIKAESGYNPKARSRAGAVCLMQLMPDTA), including the conserved active site glutamate and serine residues (indicated in bold) (Koonin and Rudd, 1994). Interestingly, the R. solanacearum chromosome contains a highly homologous lytic transglycosylase gene (rsc2291), located not far from the xcmT1/xcmS homologues rsc2310/rsc2309 (Table 1). Downstream of the P. putida GB-1 ltg gene, the 5′ end of an ORF (117 codons) was identified. Its (partial) product did not reveal any homology in database searches, but a highly similar ORF appears to be present in P. putida KT2440 (see below). On the basis of this homology, the P. putida GB-1 ORF is designated psk1 (see below and Fig. 4A). The stop codon of the (partial) ORF downstream of the xcm cluster overlapped with that of xcmS. Its partial product showed homology with the NTAA/SNAA/SOXA family of monooxygenases, for example the nitrilotriacatate monooxygenase (33% identity, 49% similarity in 180 C-terminal amino acids; Uetz et al., 1992).
It should be noted that xcmS is located downstream of xcmT1, whereas in most described regular xcp (gsp) operons, the xcpT (gspG) gene is preceded by the xcpS (gspF) homologue (de Groot et al., 1999). Interestingly, the aforementioned xcmT1 and xcmS homologues of R. solanacearum are in the same order as in P. putida GB-1 (Table 1, cf. Fig. 4B). They are present on both the R. solanacearum chromosome and megaplasmid, where they are clustered and probably in one operon with other putative gsp-related genes, including gspD (xcpQ)-, gspE (xcpR)- and additional gspG (xcpT)-related genes (Table 1, cf. Fig. 4B). In contrast, the small xcmT1/xcmS cluster of P. putida GB-1 appears to be located separately in a different genomic region (cf. Fig. 4A, see below).
Identification of a second cluster containing xcpT-related genes in P. putida GB-1
Transposon mutant GB-1-006 displayed reduced manganese-oxidizing activity compared with the P. putida GB-1 parent strain (≈ 30%; de Vrind et al., 1998). This activity was recovered only after cell disruption, indicating that strain GB-1-006 represents another mutant affected in transport of the manganese-oxidizing factor. The mutation could not be complemented by pPX135 and pAX24 carrying the xcp gene cluster of P. putida WCS358 and P. aeruginosa respectively. In addition, the cosmids complementing xcmT1 mutant GB-1-009 did not restore manganese oxidation in mutant GB-1-006, indicating that the mutation is localized in a different region of the chromosome. However, plasmid pPX949, containing a P. putida WCS358 genomic fragment hybridizing with xcpPRSTUVW of P. aeruginosa, was able to complement the GB-1-006 mutation. Limited sequence analysis of this fragment revealed the presence of at least a second xcpR homologue, not identical to the xcpR gene from the previously characterized xcp cluster from this strain (de Groot et al., 1996). Plasmid pPX949 did not complement the xcmT1 mutation of GB-1-009.
To identify similar complementing fragments in P. putida GB-1, plasmid pPLH4, containing the 9 kb Tn5-containing EcoRI fragment from strain GB-1-006 (de Vrind et al., 1998), was used as a probe in screening the P. putida GB-1 genomic library. Several cosmids with ≈ 20 kb genomic GB-1 inserts were isolated that were able to complement the GB-1-006 mutation (but not the xcmT1 mutation in GB-1-009). One of these cosmids, pCOS6-5, contained EcoRI fragments of ≈ 0.7, 1.2, 2.2, 3.0, 6.0 and 8.0 kb. Two probes were used to characterize these fragments in hybridization experiments: pHL3, containing xcmT1–xcmS, and a 2.3 EcoRI fragment from pPX949, containing the second xcpR homologue from P. putida WCS358 cloned in pUC19 (called pAG600). Even though pCOS6-5 did not complement the xcmT1 mutation of GB-1-009, the pHL3 probe reacted with the 3.0 kb EcoRI fragment of pCOS6-5, whereas the pAG600 probe reacted with the 2.2 kb fragment (results not shown). Interestingly, plasmid pPX949 contained a 7.0 kb EcoRI fragment hybridizing with the xcmT1/xcmS probe in addition to the 2.3 kb xcpR-containing fragment.
The 2.2 and 3.0 kb EcoRI fragments from pCOS6-5 were subcloned in pUC19, and the resulting plasmids were called pHL12 and pHL13 respectively. Both inserts were also cloned in pLAFR3 in two orientations for complementation experiments. Only the pHL13 insert was able to complement the GB-1-006 mutation. Complementation probably depended on a pLAFR3 promoter, as the insert was only effective in one orientation (cf. de Vrind et al., 1998). Also, the 7.0 kb fragment from pPX949, cloned in pLAFR3 in the correct orientation, restored extracellular manganese oxidation in mutant GB-1-006.
Sequence analysis of pHL13 revealed five complete and two partial ORFs, all oriented in the same direction. They appear to be part of an operon as, in most cases, the start and stop codons of the potential genes are located close together or overlap (Fig. 2). Two of the ORFs appear to represent xcpT(gspG)-related genes, designated xcmT2 and xcmT3 (Fig. 3A). Their products show high similarity to those encoded by gspG-like genes from both the R. solanacearum chromosome and megaplasmid (Table 1, cf. Fig. 4B). Partial sequence analysis of the 7.0 kb EcoRI fragment from pPX949 (see above) also revealed the presence of an xcmT2 homologue in P. putida WCS358 (Fig. 3A). The xcmT2 gene of P. putida GB-1 is preceded by the 3′ end of a short (46 codons) partial ORF (xcmQ). Although this short fragment did not reveal any homology in database searches, it is postulated to represent the 3′ end of an xcpQ-related gene in view of its location, compared with the gsp clusters of R. solanacearum (see also Fig. 4B). Closely following xcmT3, an ORF designated xcmX is present, for which no homology with a known xcp (gsp)-related gene was found in data banks. However, the xcmX product is very similar to the R. solanacearum putative proteins RSc2300 and RSp0140 (Table 1). The potential protein XcmX and its homologues from R. solanacearum have an N-terminal sequence strongly resembling that of pseudopilins except that the conserved glutamate residue at position +5 with respect to the putative processing site is missing (Fig. 3B, cf. Filloux et al., 1998). The pseudopilin XcpX of P. aeruginosa also has an atypical N-terminal sequence lacking the glutamate at +5, but has been shown to be processed by the prepilin peptidase XcpA (Bleves et al., 1998; cf. Fig. 3B).
Two complete ORFs and the 5′ end of a partial ORF are present downstream of xcmX. Their potential products show homology to CsgE, F and G, respectively, known to play a role in the production and/or transport of aggregative fibres, designated curli, in E. coli and Salmonella typhimurium (Römling et al., 1998). The encoding genes will be called csmE, csmF and csmG (csm: csg-related, possibly involved in manganese oxidation) respectively. The similarities of CsmE, F and G compared with the respective E. coli proteins are 47%, 62% and 66%. No homology was found to any other proteins from data banks.
Restriction analysis of plasmid pPLH4 localized the transposon insertion in mutant GB-1-006 in xcmX. To narrow down the complementing fragment of pHL13, deletions were made using either SfiI or XmaI (Fig. 2B and C). In both cases, no complementation was obtained, which could be explained by complete (SfiI) or at least partial (XmaI) deletion of xcmX. Then, polymerase chain reaction (PCR) primers including relevant restriction sites were designed to clone all three csm genes plus xcmX, or both xcmT genes plus xcmX in the correct (complementing) orientation in pLAFR3 using PCR (cf. Fig. 2D and E). Both constructs restored manganese oxidation in mutant GB-1-006, indicating that the product of xcmX was able to complement the GB-1-006 mutation.
As expected, pHL12, carrying the GB-1 genomic fragment hybridizing with the second xcpR homologue from P. putida WCS358 (see above), appeared to contain the 3′ end of an ORF showing strong homology with gspE(xcpR)-related genes. It will be called xcmR. Its encoded product strongly resembled RSc2308 and RSp0148 from R. solanacearum (Table 1) and showed 60% similarity to PA2677, an XcpR-like protein from P. aeruginosa. Protein PA2677 is not identical to the XcpR protein from the regular xcp operon, but belongs to a separate xcp-like cluster called hpl and containing homologues of xcpR–X. The (partial) XcmR protein contains the bacterial type II secretion system protein E signa-ture (aspartate box 1: TIEDPVE; and aspartate box 2: LRHDPDKILVGEIRD) (cf. Filloux et al., 1998) and the nucleotide binding site Walker motif A (GPTGSGKT) (Walker et al., 1982; cf. Filloux et al., 1998). Downstream of xcmR, an ORF is present that encodes a protein with slight homology to RSp0147 (cf. Table 1). This suggests that the xcmR gene is organized in a similar gsp-like cluster to that present in R. solanacearum.
Localization and organization of gsp-related clusters in P. putida KT2440
At the moment, the genome sequence of P. putida strain KT2440, a close relative of P. putida GB-1 as well as of P. putida WCS358, is being sequenced. Comparison of sequences and gene clusters identified in strain GB-1 (and WCS358) with those of strain KT2440 revealed a very high homology, suggesting that the nucleotide sequences and genome organization in general are very similar in these P. putida strains.
Like P. putida GB-1, P. putida KT2440 contains clustered xcmT1/xcmS genes, which are highly homologous to the corresponding R. solanacearum chromosomal and megaplasmid genes rsc2310/rsc2309 and rsp0150/rsp0149 respectively (Fig. 4A). The xcmT1/xcmS cluster is preceded by a potential ltg gene in the opposite orientation (Fig. 4A). Analysis of the adjacent sequences showed that the ltg gene is followed at ≈ 175 bp distance by two ORFs in the same orientation encoding two very similar proteins belonging to the family of histidine protein kinases (Stock et al., 2000). They are tentatively designated Psk1 and Psk2 (Psk: protein sensor kinase) and show amino acid similarity to, e.g. the ZraS protein of E. coli (primary accession number Q8X614), especially in the C-terminal part (46% and 51% respectively). ZraS (also called HydH) forms a two-component regulatory system with ZraR (HydG, see also below) regulating hydrogenase formation in response to elevated Zn2+ or Pb2+ concentrations (Leonhartsberger et al., 2001). The first 117 N-terminal amino acids of Psk1 are 90% identical (99% similar) to the product of the P. putida GB-1 ORF downstream of the ltg gene (Fig. 1, see above). Generally, the C-terminal part of histidine kinases represents the conserved phosphorylation domain, whereas the N-terminal domain is variable and responds to external stimuli (Stock et al., 2000). This can explain why the partial P. putida GB-1 Psk1 did not show any homology with proteins in databases. The ORFs psk1 and psk2 are followed by a gene encoding a potential response regulator (designated Ptr, phosphorylation transcription regulator, 55% similarity to E. coli ZraR, primary accession number Q8X613). Interestingly, the chromosome of R. solanacearum contains a set of genes encoding a putative sensor kinase and a response regulator directly upstream of the gsp-related cluster (Table 1). However, their products do not show sequence similarity to the potential sensor kinases and response regulator of P. putidaKT2440.
Pseudomonas putida KT2440 also contains a gsp -related cluster containing xcmT2 , xcmT3 , xcmX and csmE , F and G homologues ( Fig. 4B ). This cluster is localized in a different region of the genome from the xcmT1 / xcmS cluster (≈ 60 kb distance). The region immediately upstream of the xcmT2 homologue is organized in a similar way to the gsp -related clusters of the R. solanacearum chromosome and megaplasmid, including the presence of a gspD ( xcpQ )-related gene immediately preceding the xcmT2 homologue ( Fig. 4B , cf. Table 1 ). The 46 C-terminal amino acid residues of the encoded protein were 96% identical to those of the product of xcmQ of strain GB-1, confirming the identification of the latter as the 3′ end of an xcpQ -related gene (see above). The four ORFs between xcmQ and xcmR of P. putida KT2440 show significant homology to the corresponding R. solanacearum genes only ( rsc2304–2307 and rsp0144–0147 , cf. Fig. 4B ). In contrast to the R. solanacearum chromosome and megaplasmid, the P. putida KT2440 genome does not contain xcp ( gsp )-related genes immediately upstream of the xcmR homologue ( rsc2308 and rsp0148 in R. solanacearum , Fig. 4B ). Instead, some of the upstream genes (all with the same orientation as the xcp -related cluster) show homology to genes from the mof operon of Leptothrix discophora ( mofC in particular, Fig. 4B , cf. a similar homologue in R. solanacearum , Table 1 , cf. Brouwers et al., 2000 ), and others show homology to genes from the mnx operon of Bacillus SG1 ( mnxC and mnxG in particular, Fig. 4B , cf. van Waasbergen et al., 1996 ). Both the mof operon and the mnx operon have been proposed to be involved in manganese oxidation in the respective bacterial species ( van Waasbergen et al., 1996 ; Corstjens et al., 1997 ; Brouwers et al., 2000 ). Note that the region immediately downstream of the csgG homologue does not contain other csg -related genes ( Fig. 4B ). Homologues of genes putatively encoding structural subunits of curli fibres (CsgA and B) are present in P. putida KT2440 starting at nucleotide 3321560. Homologues of csg genes could not be found in R. solanacearum .
A regular xcp operon highly similar to that of P. putida WCS358 was found starting at nucleotide 666548 with the gene encoding XcpP (87% similarity, including a GTG start codon, cf. de Groot et al., 1996). As in P. putida WCS358, an xcpX homologue is absent. The xcp cluster is preceded by a homologue of P. putida WCS358 uxpA (97% similarity between the protein products), encoding a putative member of the SoxB protein family and proposed to be the substrate of the Xcp machinery in P. putida WCS358 (de Groot et al., 1996). It can be assumed that P. putida GB-1 also contains an xcp operon organized in a similar manner. Attempts to isolate xcp genes of P. putida GB-1 by PCR with primers based on sequences of P. putida WCS358 and KT2440 failed. Also, attempts to create insertion mutants in the putative xcp operon of GB-1 with suicide plasmids, previously used to construct xcp mutants of P. putida WCS358 (de Groot et al., 1996), were not successful. Possibly, the homology between the xcp genes of the three P. putida strains, although apparently high, was insufficient to result in efficient annealing or recombination.
Potential involvement of two-component signal transduction in manganese oxidation in P. putida GB-1
The assumption that the genomic organization of P. putida GB-1 strongly resembles that of P. putida KT2440 led us to re-examine some data on a previously isolated non-oxidizing transposon mutant, strain GB-1-005. The transposon insertion in this strain appeared to be located in a genomic EcoRI fragment of ≈ 12 kb (de Vrind et al., 1998). Preliminary sequence analysis of parts of this fragment (cloned in pUC19 and called pPLH12) did not yield clear homologies (de Vrind et al., 1998). Plasmid pPLH12 was used as a probe to isolate hybridizing clones from the genomic P. putida GB-1 library. We observed that some of these clones also hybridized with pPLH6, and it appeared that the cosmids isolated from these clones complemented both mutant GB-1-005 and xcmT1 mutant GB-1-009. This indicates that the affected genes are located within 20 kb of each other. One of the complementing cosmids, pCOS9-21, contained a 3.5 kb EcoRI fragment hybridizing with pPLH12. Complementation of mutant GB-1-005 with this subfragment did not succeed. Sequences of both ends of this fragment cloned in pUC19 (pHL5, Table 2) were obtained, and the sequence adjacent to Tn5 in strain GB-1-005 was determined in an EcoRI–BamHI subclone of pPLH12 with a Tn5-specific primer. The obtained sequences were mapped on the sequence of P. putida KT2440, which localized the Tn5 insertion in mutant GB-1-005 within the potential psk2 gene (Fig. 4A).
20 kb genomic fragment from GB-1 in pLAFR3 complementing GB-1-006
20 kb genomic fragment from GB-1 in pLAFR3 complementing GB-1-009 and -005
20 kb genomic fragment from GB-1 in pLAFR3 complementing GB-1-009 and -005
2.3 kb EcoRI fragment from pCOS9-20 in pUC19
3.5 kb EcoRI fragment from pCOS9-21 in pUC19
2.2 kb EcoRI fragment hybridizing with pAG600 from pCOS6-5 in pUC19
3.0 kb EcoRI fragment hybridizing with pHL3 from pCOS6-5 in pUC19
Bacterial manganese oxidation is postulated to be catalysed by enzymes or enzyme complexes, containing Cu ions in the active sites (Tebo et al., 1997). As in almost all bacterial species studied, manganese is oxidized extracellularly, the manganese-oxidizing factor or at least part of it has to be transported and secreted across the outer cell wall and, in natural conditions, incorporated in outer cell coverings such as the outer membrane, sheath or spore coat. The results presented here indicate that, in Gram-negative bacteria, the transport of the manganese-oxidizing factor is mediated by a protein transport system related to the gsp, also called xcp, machinery. Mutations in genes (xcmT1 and xcmX) encoding pseudopilin-like proteins inhibited the secretion of manganese-oxidizing activity in P. putida GB-1. The identified xcp-like genes resemble, but are clearly not identical to, those of the xcp operon first identified in P. aeruginosa and P. putida WCS358 (Bally et al., 1992; de Groot et al., 1996). The elucidation of the complete genome sequence of P. aeruginosa PAO1 revealed the presence of many xcp homologues outside the xcp operon (PA3095–PA3105), including several homologues of xcpQ and xcpR (Stover et al., 2000). Recently, it was demonstrated that one such cluster, called the hxc cluster and possessing a complete set of xcp-like genes, functions in the transport of the alkaline phosphatase LapA (Ball et al., 2002). The presence of multiple functional type II secretion systems within one organism was related to the high adaptation potential of P. aeruginosa (Ball et al., 2002). Similarly, R. solanacearum appears to contain several xcp(gsp)-like gene clusters and many of the related gene clusters involved in type IV pilus biogenesis (Salanoubat et al., 2002). In this case also, the presence of so many homologues potentially functioning in protein secretion and the production of attachment factors (depending on protein secretion systems for transport across the outer membrane) is suggested to broaden the adaptive ability of the organism (Salanoubat et al., 2002).
Apparently, P. putida also possesses several Xcp-related transport systems. We tentatively called the system involved in transport of the manganese-oxidizing factor Xcm. As judged from the composition of homologous gene clusters in R. solanacearum and P. putida KT2440, and based on direct identification of xcp-related genes, the putative xcm system contains at least an xcpQ, xcpR and xcpS homologue and four different pseudopilin-encoding genes, including three xcpT-like genes. Also, the ‘regular’xcp(gsp) clusters contain several pseudopilin-encoding genes, called xcpT–W (gspG–J). It should be noted that the three XcmT proteins all show a higher degree of similarity to XcpT than to XcpU, V or W. The putative XcmT pseudopilins are likely to be processed by the leader peptidase XcpA. The latter was shown previously to be involved in secretion of the manganese-oxidizing factor of P. putida GB-1 (Brouwers et al., 1998).
The product encoded by xcmX has an N-terminal sequence resembling that of the pseudopilins but lacking the conserved glutamate at position +5 with respect to the processing site. Probably, it is also processed by XcpA, as has been shown for the atypical pseudopilin XcpX of P. aeruginosa (Bleves et al., 1998). It should be stressed, however, that XcmX shows no sequence similarity to (P. aeruginosa) XcpX outside the N-terminal segment (Fig. 3) and is also much shorter than XcpX (175 versus 413 residues). With respect to its length, XcmX more closely resembles the atypical minor pilin PilX of the type IV pili system of P. aeruginosa (243 residues), which also lacks the glutamate at position +5. The +5 glutamate residue in the pseudopilins has been suggested to function in subunit recognition during the formation of pseudopili in the secretion process (Bleves et al., 1998; Filloux et al., 1998). The atypical pseudopilins lacking this residue may function in the termination of pseudopilus formation (Bleves et al., 1998; Filloux et al., 1998). Mutation of xcmX not only inhibited transport of the manganese-oxidizing factor, but also reduced its activity. Possibly, the interference with the secretion process left the manganese-oxidizing factor incorrectly folded and/or the active site (partly) unexposed.
Like P. putida KT2440, the P. putida GB-1 and WCS358 genomes very probably contain homologues of the R. solanacearum genes rsp0144–0147 (or the corresponding chromosomal ones) in between xcmQ and xcmR. The encoded products do not show homology to proteins from databases, including known Xcp(Gsp)-related proteins. However, some of them have characteristics of transmembrane proteins (Table 1), and the regular Xcp machinery is known to involve (inner) membrane proteins. We propose that the four putative proteins contribute to the Xcm secretion apparatus. Interestingly, homologues of xcpP, Y and Z cannot be found in P. putida KT2440 outside the regular xcp/gsp cluster. The XcpP, Y and Z proteins are characteristic of the type II secretion systems, and such proteins are not involved in type IV pili biogenesis. The absence of these homologues in the genome of P. putida KT2440 outside the regular xcp/gsp cluster suggests that such proteins do not contribute to Xcm-dependent secretion, discriminating the latter from the classical Xcp/Gsp transport systems. Possibly, the Xcm system has evolved more recently from the Pil machinery than the Xcp/Gsm systems. Consistent with this hypothesis, blast searches for XcmS homologues revealed higher similarity to PilC-related proteins than to XcpS-related proteins (results not shown). Similarly, the GspD-related protein Rsp0143 showed more homology to PilQ-related proteins than to XcpQ-like molecules. This was less apparent for XcmQ of P. putida KT2440, however.
Apparently, the pseudopilins encoded by the putative regular xcp operon of P. putida GB-1, if active, cannot substitute for those encoded by the xcm cluster. Neither mutant GB-1-009 nor mutant GB-1-006 could be complemented by the xcp operons from P. aeruginosa or P. putida WCS358, whereas the P. putida WCS358 XcpA protein was completely functional in P. putida GB-1. In future research, the isolation and mutagenesis of the P. putida GB-1 xcp operon will be attempted to study the possible involvement of its components in the transport of the manganese-oxidizing factor.
A special feature of the putative Xcm secretion system of P. putida GB-1 (and P. putida KT2440) is the clustering of xcm genes with csgE, csgF and csgG homologues. The start codon of csgE overlaps with the termination codon of xcmX, suggesting that the csg homologues form one operon with the xcmX, xcmT2 and xcmT3 genes. In E. coli and in Salmonella species, the csg genes are responsible for the production of aggregative fibres, called curli fibres in E. coli (Römling et al., 1998). The csg genes are organized in two divergently transcribed operons. CsgAB encode the major curli structural subunit CsgA and the minor putative nucleator and branching subunit CsgB. Generally, csgD, E, F and G are clustered in a second operon. CsgD functions as a transcriptional activator. The roles of CsgE, F and G are not fully understood, but the proteins are thought to function in curli assembly and transport. CsgG appears to be an outer membrane lipoprotein assisting in CsgA and B secretion and protecting them from degradation (Loferer et al., 1997). Recently, CsgE has been assigned the role of chaperone and CsgF that of additional nucleator in curli polymerization (Chapman et al., 2002). Presently, it is not known whether P. putida GB-1 is able to form curli fibres. Like P. putida KT2440, P. putida GB-1 probably contains homologues of csgA and B at a different location in the genome. However, curli fibres cannot be observed in electron micrographs of cells grown under our experimental conditions (Okazaki et al., 1997; cf. Chapman et al., 2002). In addition, neither wild-type P. putida cells nor any of the mutants could be stained with Congo red, known to bind to curli fibres (unpublished results, cf. Chapman et al., 2002). The possibility that the csm genes encode components of the Xcm secretion machinery seems unlikely, considering the absence of these genes in R. solanacearum. Nevertheless, this possibility or the alternative possibility that the Xcm system is involved in secretion of curli fibres produced under special conditions needs to be analysed in future experiments.
An interesting preliminary finding is the involvement of one or both of a set of two putative sensor kinases belonging to a two-component signal transduction system (Stock et al., 2000) in the production and/or transport of the manganese-oxidizing factor. Manganese oxidation in P. putida GB-1 is first observed in the early stationary phase, and the oxygen concentration in the cultures during growth was shown to influence the manganese-oxidizing activity (Okazaki et al., 1997). Possibly, extracellular signals can affect the production and/or transport of the manganese-oxidizing factor. In P. aeruginosa, the xcp secretion machinery was shown to be regulated by multiple quorum-sensing modulons (Chapon-Hervéet al., 1997). Future research will involve the identification of the target(s) of the putative response regulator of the signal transduction machinery of P. putida GB-1.
As stated above, bacterial manganese oxidation appears to be mediated by Cu-containing proteins, multicopper oxidases in particular (Tebo et al., 1997). In P. putida GB-1, a potential multicopper oxidase called CumA was shown to be involved in manganese oxidation (Brouwers et al., 1999). It is possible that CumA is the manganese-oxidizing enzyme secreted by the xcm secretion machinery. In the signal sequence of CumA (cf. Brouwers et al., 1999), a typical twin-arginine motif can be distinguished, suggesting that the protein is transported across the inner membrane via the TAT (twin-arginine transport; Berks et al., 2000) system rather than via the Sec machinery. The latter has always been thought to precede the gsp/xcp-dependent outer membrane translocation step, but recent data suggest that, in P. aeruginosa, the xcp system can be fed by the TAT system as well (Voulhoux et al., 2001). Similarly, a twin-arginine motif can be recognized in the signal sequence of UxpA, the putative substrate of the Xcp system of P. putida WCS358 (de Groot et al., 1996). The overexpression of CumA in E. coli and production of specific antibodies is under way to study the localization of CumA in transport mutants compared with wild-type cells.
Finally, the analysis of the sequence of P. putida KT2440 upstream of the csg/xcm cluster (Fig. 4B) revealed potentially interesting ORFs, some of them showing homology to genes involved in manganese oxidation in L. discophora SS-1 (Brouwers et al., 2000) and Bacillus strain SG-1 (van Waasbergen et al., 1996) respectively. It is worthwhile to attempt the isolation of these potential genes from P. putida GB-1 and to study the effect of mutations on manganese-oxidizing activity.
Bacterial strains, plasmids and culture conditions
Bacterial strains and plasmids used are summarized in Table 2. When cultured on streptomycin-containing media, P. putida strain GB-1 yielded a spontaneous streptomycin-resistant subclone designated strain GB-1-002 (Table 2). This strain has been used as the parent strain throughout this study to have an additional phenotypical marker. However, for the sake of simplicity, only the name GB-1 will be used for the parent strain in the text.
Pseudomonas putida GB-1 was grown at room temperature on LD medium as described for Leptothrix discophora ( Boogerd and de Vrind, 1987 ). Solid media contained 1.8% Bacto agar and were supplemented with MnCl 2 (final concentration 100 mM). E. coli strains were grown routinely in LB medium ( Sambrook et al., 1989 ) at 37°C. Antibiotics were used at the following concentrations in mg ml −1 : P. putida : tetracycline (Tc) 75, kanamycin (Km) 40, streptomycin (Sm) 100, piperacillin 75; E. coli : Tc 15, Km 25, ampicillin (Ap) 100.
Complementation experiments were performed by electroporation of early stationary non-oxidizing P. putida cells with pLAFR3 constructs (Bio-Rad gene pulser, 2.5 KV, 200 W, 25 mF, 109 cells ml−1 H2O). Cells were adapted for 1 h in LD medium, after which they were transferred to solid media with relevant antibiotics. After growth, colonies were transferred to solid or liquid media supplemented with MnCl2 (100 mM) without antibiotics (de Vrind et al., 1998). Alternatively, cosmids were mobilized to P. putida cells by triparental mating using helper plasmid pRK2013 as described previously (Brouwers et al., 1998; de Vrind et al., 1998)
Southern hybridization and library screening
Hybridization experiments were performed with digoxygenin (DIG)-labelled probes. Hybridization with plasmid digests or digests of genomic DNA (Corstjens and Muyzer, 1993) was performed according to standard methods (Boehringer Mannheim GmbH, 1989) at a hybridization temperature of 65°C. Final washing was done at 65°C in 0.5× SSC (75 mM NaCl, 7.5 mM Na citrate, pH 7.0). The genomic library (de Vrind et al., 1998) was screened using standard protocols (Boehringer Mannheim GmbH, 1989) after transfer and growth of colonies on a nylon membrane covering solid LB with Tc (25 mg ml−1).
Sequences were obtained commercially (Eurogentec) by automated dideoxy chain-termination technology using pUC19 universal and reverse primers and primer walking. The partial sequence of pPLH6 was obtained using a custom-made primer based on the pHL3 sequence (5′-TCAGC CAGGTATTGCAAG-3′) and primer walking. Subclones of pHL13 introducing EcoRI and HindIII restriction sites (cf. Fig. 2) were made by PCR using HotStarTaq polymerase (Qiagen). The primers 5′-TCGTGGGAATTCTGCGAGATG GTGTTGTCGACG-3′ (EcoRI site) and 5′-CCGAAGCTTG GTCGTGGACAGCCAAATGGCG-3′ (HindIII site) were used to amplify the xcmT2,3–xcmX cluster, and the pUC19 reverse primer (EcoRI site) was used with primer 5′-CGAAAGCTTC CATTACCCGGACTCGCTGG-3′ (HindIII site) to amplify the xcmX–csmG cluster.
Sequences were analysed using the Wisconsin Sequence Analysis Package (Wisconsin Genetics Computer Group, version 8.1) and the blast program of the Swiss Institute of Bioinformatics; patterns and profiles were analysed with prosite. Alignment was performed with clustalw of the Pôle Bio-Informatique Lyonnais (PBIL). Preliminary sequence data of P. putida KT2440 were obtained from The Institute for Genomic Research at http:www.tigr.org. Sequences were deposited in GenBank under nucleotide accession numbers AF06532 and AF515666.
Preliminary sequence data of P. putida KT2440 were obtained from The Institute for Genomic Research at http:www.tigr.org.