INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Phosphorus plays a central role in the metabolism of all living organisms and is a required nutrient. In addition to its role in innumerable metabolic pathways, it is a component of phospholipids, RNA, DNA, and the principal nucleotide cofactors involved in energy transfer and catalysis in the cell. Despite the ubiquitous role of P in metabolism, the biochemistry of P-containing compounds is generally considered to be quite simple, consisting almost entirely of phosphate-ester formation and hydrolysis. Thus, it not surprising that most P found in living systems is in the form of inorganic phosphate and its esters. However, there are an increasing number of studies showing biochemical reactions of P compounds that do not involve the formation or hydrolysis of phosphate-esters. Some of these reactions involve compounds in which the P is at a lower valence state, suggesting that previously unsuspected P redox reactions may be important in the metabolism of this element.

On earth virtually all known phosphorus exists in the +5 oxidation state. This is true for both inorganic phosphate and the organic phosphate-esters that play a role in the bulk of known P metabolism. Thus, the discovery of phosphonates (+3 valence), and phosphinates (+1 valence) in living systems was somewhat surprising (18). These reduced P compounds, which have direct carbon-phosphorus bonds, are clearly important for the organisms in which they are found. For example, Tetrahymena may possess as much as 30% of its membrane lipids in the form of phosphonolipids (22). It is now known that similar compounds are found in a wide range of organisms, including humans, plants, and bacteria (18). In addition to the discovery of these compounds, there is a variety of evidence suggesting that other reduced P compounds play a role in the metabolism of this element. A number of reports, some dating back centuries, have suggested that phosphine (H₃P) is produced during the decomposition of organic material. This toxic, spontaneously flammable gas is equivalent to the most-reduced forms of P (3 valence). Although these early reports must be considered anecdotal, recent reports have verified natural production of phosphine by using modern methods of chemical analysis. In these studies gas chromatography combined with mass spectroscopy was used to detect phosphine in the atmosphere, in anaerobic harbor sediments, and in sewage treatment facilities (8, 11-14). Other studies have demonstrated the reduction of phosphate in anaerobic soil and during corrosion of metals under anaerobic conditions (19, 38, 40). These data clearly demonstrate that reduction of P occurs in nature. In each case living organisms are suspected to be the causative agents. Recent studies have begun to explore the biochemistry of phosphonate and phosphinate production (34, 36); however, nothing is known about the metabolism of other reduced P compounds.

Other investigations have shown that biologically catalyzed oxidation of reduced P compounds can occur. A number of bacteria have been shown to be capable of oxidizing reduced P compounds when these are provided as the sole source of P. Inorganic phosphite (+3 valence) was oxidized to phosphate by numerous laboratory strains of microorganisms, including prokaryotes such as Escherichia coli, Agrobacterium tumefaciens, and several species of Pseudomonas and Rhizobium, as well as one eukaryote, Saccharomyces cerevisiae (1, 6, 26). Hypophosphite (+1 valence) can also be oxidized to phosphate by bacteria (9, 16). As is the case with P reduction, very little is known about the process of P oxidation by living organisms. Partial purification of the responsible enzymes has been achieved for both phosphite and hypophosphite oxidation, from Pseudomonas fluorescens and Bacillus caldolyticus, respectively (16, 27). However, nothing is known about the genetics of these processes, because molecular and genetic techniques for study of the responsible organisms were largely unavailable at the time of these investigations. It was recently shown by genetic methods that the enzyme C-P lyase from E. coli has phosphite oxidase activity (29, 30). However, due to the current inability to assay C-P lyase in vitro, this activity remains uncharacterized biochemically.

These data indicate that metabolic pathways involving redox reactions of P compounds may be quite common in microorganisms, yet the nature of these pathways remains largely unexplored. In no case are both genetic and biochemical data available for a P-oxidizing system. With this goal in mind, we isolated a variety of organisms capable of oxidizing the reduced P compounds phosphite and hypophosphite. In this report the first molecular and genetic characterization of a biochemical pathway for oxidation of these reduced P compounds is presented. Biochemical characterization of the enzymes involved in the pathway will be reported elsewhere.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and plasmids. The bacterial strains used in the study are shown in Table 1. In general, DH5 and DH5/pir were used as hosts for cloning experiments, while S17-1 and BW20767 were used as donor strains for conjugation experiments involving broad-host-range plasmids. Plasmids pMMB67EH and pMMB67HE (10) and pDN18 and pDN19 (35) were from David Nunn. Plasmid pLAFR5 (21) was from Stephen Farrand. Plasmid pBEND2 (23) was from Stanley Maloy. Plasmids pTZ18R, pUC4K, and pSL1180 (5) were from Pharmacia (Piscataway, N.J.). Plasmid pBluescript KS(+) was from Stratagene (La Jolla, Calif.).

Media. Most media used in the study have been previously reported (39). Minimal A medium was as described in reference 32. Antibiotics were used at the following concentrations for E. coli and Pseudomonas stutzeri WM88: kanamycin, 50 µg/ml; ampicillin, 100 µg/ml; tetracycline, 12 µg/ml; and streptomycin, 100 µg/ml. For Pseudomonas aeruginosa, antibiotics were used as follows: carbenicillin (instead of ampicillin), 200 µg/ml; tetracycline, 100 µg/ml; and rifampin, 25 µg/ml. P compounds were prepared fresh and filter sterilized prior to addition to media at a final concentration of 0.5 mM. Noble agar (1.6%) was used to solidify media used for testing P oxidation phenotypes (see below). Sucrose-resistant recombinants of strains carrying the Bacillus subtilis sacB gene as a counterselectable marker were selected on agar-solidified medium containing 10 g of tryptone, 5 g of yeast extract, and 50 g of sucrose per liter. Denitrification was tested in tightly closed screw cap tubes completely filled with Luria-Bertani broth with and without 0.1% NaNO₂ or 0.1% NaNO₃.

P oxidation phenotypes. P oxidation phenotypes were scored by growth on 0.4% glucose-MOPS (morpholinepropanesulfonic acid) medium with the compound under study supplied at 0.5 mM as the sole P source (29). The ability to oxidize a compound to phosphate allows growth on this medium. Because the amount of P required for growth is relatively small, the contaminating levels of phosphate found in many medium components, especially agar, allow slight background growth of all strains in these media. To control for this variable, the strains in question were always compared to suitable positive and negative controls streaked on the same plate.

NMR analysis of the P compounds used in the study. One of our concerns in using reduced P compounds as medium components was the known instability of these compounds under aerobic conditions. To address this issue, we examined the stability of phosphite and hypophosphite in stock solutions and in MOPS medium by ³¹P nuclear magnetic resonance (NMR). Spectra were obtained in 10-mm tubes at ambient temperature by using either a General Electric GN500-NB (pulse time, 55 µs; relaxation delay, 3.5 s) or a General Electric GN300-NB (pulse time, 24 µs; relaxation delay, 4 s) instrument. D₂O was added to allow deuterium signal locking to be used. For experiments in which the P concentration ranged from 250 to 1,000 µM, 512 or 1,012 scans were taken for each sample. Fewer scans were used for samples with high P concentrations. No detectable oxidation products of either phosphite or hypophosphite were seen after 2 weeks of incubation under the growth conditions used in this study. Phosphite stock solutions were stable for at least 1 year. However, prolonged storage of hypophosphite stock solutions led to accumulation of phosphite, approaching ca. 50% of the total P after 6 months of storage at 4°C. For this reason, all reduced P stock solutions were prepared fresh, and media containing reduced P compounds were used within 2 weeks of preparation.

DNA methods. Standard methods were used throughout for isolation and manipulation of plasmid DNA. Chromosomal DNA was isolated from P. stutzeri WM88 by the cetyltrimethylammonium bromide method (3). DNA hybridizations were performed as previously described (31). Probes used for hybridization experiments were labeled with [-³²P]dATP by using the Prime-a-Gene kit (Promega, Madison, Wis.) according to the manufacturer's specifications. DNA sequences were determined from double-stranded templates by automated dye terminator sequencing at the Genetic Engineering Facility, University of Illinois. The initial sequences of each clone were always determined by using standard lacZ forward and reverse primers. The remaining sequences were obtained either with internal primers or from nested deletions constructed with the ExoIII/Mung Bean deletion kit (Stratagene).

Cloning and analysis of 16S rDNA. 16S ribosomal DNA (rDNA) from P. stutzeri WM88 was amplified by PCR from genomic DNA with Vent DNA polymerase (New England Biolabs, Beverly, Mass.) by using the primers 5'-TTGGATCCAGAGTTTGATCMTGGCTCAG-3' and 5'-GTTGGATCCACGGYTACCTTGTTACGAYT-3'. The PCR products from separate reactions were cloned into pWM73 (28) to generate pWM206 and pWM207. The complete DNA sequences of both clones were determined, and these sequences are in complete agreement. To identify the species, this 16S rDNA sequence was compared to others in the Ribosomal Database Project (http://rdpwww.life.uiuc.edu) by utilizing the collection of analysis tools provided at this Internet site (25).

Plasmid constructions. Because of the large number of plasmids constructed during the course of this study, only the basic steps of these constructions are presented here. In many cases the restriction sites found within the polylinker of each vector were used for these constructions (Fig. 1).The first set of plasmids was used in subsequent constructions as vectors or as a source for antibiotic resistance cassettes. The broad-host-range IncQ plasmids pWM263 and pWM264 were constructed by replacement of the EcoRI-HindIII polylinkers of pMMB67HE and pMMB67EH, respectively, with the EcoRI-HindIII polylinker of pSL1180. Similarly, the broad-host-range IncP plasmids pWM265 and pWM266 were constructed by replacement of the EcoRI-HindIII polylinkers of pDN18 and pDN19, respectively, with the EcoRI-HindIII polylinker of pSL1180. Plasmid pJK25, carrying an aph cassette flanked by symmetrical polylinkers, was constructed by insertion of the 1.3-kbp SalI cassette of pUC4K into the SalI site of pBEND2. Plasmid pJK25 greatly simplifies in vitro construction of gene disruptions by allowing isolation of the aph gene cassette (encoding resistance to kanamycin) by digestion with a single restriction endonuclease, chosen from a variety of different possible enzymes.

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Structures of the broad-host-range plasmids pWM263 and pWM265. The physical maps of the cloning vectors pWM263 and pWM265 are shown. The large number of unique restriction sites in these plasmids greatly facilitates subcloning of DNA inserted into these vectors (see text for details). Only unique restriction sites are shown. Two additional plasmids, pWM264 and pWM266, are similar but with the polylinker in the orientation opposite to that in pWM263 and pWM265, respectively. Genes cloned into pWM263 and pWM264 can be expressed from the tac promoter (ptac) in an IPTG (isopropyl--D-thiogalactopyranoside)-dependent manner. The rrnB terminator (trrnB) in these plasmids terminates transcripts originating at ptac. All four plasmids can be mobilized to a variety of recipients from E. coli hosts that carry the tra genes of RP4 in trans. The bla gene encodes resistance to -lactam antibiotics in pWM263 and pWM264. The tetA gene encodes resistance to tetracycline in pWM265 and pWM266. The lacZ gene of pWM265 and pWM266 is not functional due to stop codons in the large polylinkers of these plasmids.

Genetic techniques. In general, conjugation between E. coli donors and P. aeruginosa or P. stutzeri recipients was performed by mixing donor and recipient cells in a 10:1 ratio and incubating overnight on TYE agar. Cells from the mating mixture were then scraped from the surface and resuspended in basal medium, and various aliquots were spread onto selective agar. The genomic library of P. stutzeri WM88 in pLAFR5 was moved into P. aeruginosa PAK en masse by replica plating master plates of the library in E. coli S17-1 onto a lawn of P. aeruginosa PAK. After overnight incubation, these plates were replica plated onto minimal A medium-tetracycline agar to select for exconjugates. In general, the P oxidation phenotypes of various plasmid subclones in P. aeruginosa were examined in strain P. aeruginosa PAK pil rif. Plasmids were moved into this strain by conjugation with E. coli BW20767 or S17-1 donors with selection on TYE agar with rifampin in combination with either tetracycline or carbenicillin, as appropriate. Exconjugants of E. coli donors and P. stutzeri WM567 recipients were selected on glucose-MOPS medium with an appropriate antibiotic. Exconjugates of E. coli donors and either WM581, WM688, or WM691 were selected on TYE agar with kanamycin and tetracycline.

Nucleotide sequence accession numbers. The GenBank accession numbers for the P. stutzeri WM88 DNA sequences determined in this study are AF038653 for 16S rDNA, AF061070 for the minimal region required for the oxidation of phosphite to phosphate, and AF061267 for the minimal region required for oxidation of hypophosphite to phosphite.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Isolation of hypophosphite-oxidizing organisms. All organisms require P in its most-oxidized form, phosphate, for growth. Because of this it is possible to select for organisms capable of oxidizing reduced P compounds to phosphate. Thus, in media containing reduced P compounds as the sole P source, only those organisms capable of oxidizing these reduced compounds to phosphate are able to grow. To enrich for hypophosphite-oxidizing organisms, we inoculated 0.4% glucose-MOPS medium containing 0.5 mM hypophosphite as the sole P source with soil and water samples from a variety of environments. The cultures were incubated aerobically with vigorous agitation at either 30 or 37°C. In every case these enrichments yielded hypophosphite-oxidizing organisms, usually within a few days. The organisms were obtained in pure culture by repeated single-colony isolation on agar-solidified 0.4% glucose-MOPS medium containing 0.5 mM hypophosphite.

Identification and characterization of hypophosphite-oxidizing organisms. Based on microscopic examination, colony morphology, and growth characteristics, at least 10 different hypophosphite-oxidizing organisms were obtained from these enrichments (data not shown). We chose to concentrate our efforts on one of these organisms that shows particularly robust growth in media with hypophosphite as the sole P source. The doubling time of this isolate is 97 min in succinate-MOPS medium at 37°C with hypophosphite as the sole P source, relative to 75 min in same medium with phosphate as the sole P source. The growth yield is identical in both media. This organism also grows in medium with phosphite as the sole P source and, thus, is also capable of phosphite oxidation (the doubling time in succinate-MOPS medium with phosphite as the sole P source is 120 min). The organism, which was obtained from an enrichment inoculated with local soil, is an oxidase-positive, gram-negative bacterium that forms wrinkled yellow-orange colonies on a variety of media. To identify this organism, we cloned and sequenced its 16S rDNA after PCR amplification. The rDNA sequence was analyzed by using a collection of phylogenetic tools provided by the Ribosomal Database Project (25). These data indicate that the organism is a strain of P. stutzeri, which we designated P. stutzeri WM88. The 16S RNA gene of P. stutzeri WM88 was identical (1,456 of 1,456 bp) to the previously reported sequence from P. stutzeri DSM 50227 (4). This species assignment is fully consistent with a number of other traits, including the classic P. stutzeri trait of denitrification (data not shown).

Oxidation of phosphite and hypophosphite by known bacterial species. In our original enrichments, P-oxidizing organisms were very similar to P. stutzeri WM88 were isolated from numerous inocula. Our finding that strain WM88 was P. stutzeri led to the question of whether other pseudomonads are capable of P oxidation. We tested the ability of a variety of known Pseudomonas species to oxidize phosphite and hypophosphite as shown by their ability to utilize these compounds as sole P sources. Surprisingly, none of the species tested are able to oxidize hypophosphite, including four known P. stutzeri strains, two strains of P. fluorescens, P. aeruginosa, P. mendocina, P. putida, and three strains of P. syringae (Table 1). Only two, P. stutzeri DSM50227 and P. mendocina CH50, are able to oxidize phosphite. In addition to testing pseudomonads, we screened a large number of common laboratory organisms for use of hypophosphite or phosphite as a source of P. None of these strains are able to utilize hypophosphite, although a few are able to utilize phosphite (data not shown). Among these, only the phenotype of E. coli is relevant to this study. E. coli S17-1 can oxidize phosphite but not hypophosphite. Our finding that E. coli S17-1 is incapable of hypophosphite oxidation is in contrast to that of Lauwers and Heinen, who showed that E. coli 2037 was capable of hypophosphite oxidation (24). These contrasting data may be due to differences in the two E. coli strains.

Cloning of the genes required for hypophosphite oxidation by P. stutzeri WM88. The genes required for oxidation of phosphite and hypophosphite by P. stutzeri WM88 were cloned by using a strategy similar to that used for the original isolation of the organism. Because E. coli S17-1 is not able to oxidize hypophosphite, it cannot grow on medium with hypophosphite as the sole P source. We constructed a genomic library of P. stutzeri WM88 in the broad-host-range, conjugal cosmid pLAFR5 and tested whether the plasmid clones conferred the ability to grow on medium with hypophosphite as the sole P source upon E. coli hosts. Because we were uncertain whether P. stutzeri genes would be expressed in E. coli, we tested this phenotype in P. aeruginosa PAK as well. Like E. coli, P. aeruginosa PAK is unable to oxidize hypophosphite. After transfection of the cosmid library into E. coli S17-1, individual clones were conjugally transferred to P. aeruginosa PAK by a replica mating technique, as described in Materials and Methods. In all, 2,400 plasmid clones were examined in the two host strains. Seven plasmid clones, pWM234, pWM235, pWM236, pWM237, pWM238, pWM239, and pWM240, that conferred the ability to oxidize hypophosphite upon E. coli S17-1 were isolated; however, 2 to 3 days of incubation on hypophosphite medium was required for this phenotype to be clearly displayed. Five of these seven clones, pWM235, pWM236, pWM237, pWM238, and pWM239, also conferred this trait upon P. aeruginosa PAK. In contrast to that in E. coli, hypophosphite oxidation in P. aeruginosa was quite rapid, with growth on hypophosphite medium occurring within 16 h. No clones that conferred hypophosphite oxidation upon P. aeruginosa and did not also do so for E. coli were obtained.

Subcloning of genes required for oxidation of phosphite and hypophosphite. To facilitate subcloning of the genes required for hypophosphite and phosphite oxidation, two sets of broad-host-range plasmids with the polylinker from pSL1180 were constructed (Fig. 1). The polylinker from pSL1180 is comprised of all possible 6-bp palindromes and, therefore, is cut by all known palindrome-recognizing restriction endonucleases. The IncQ plasmids pWM263 and pWM264 are derivatives of the widely used expression vectors pMMB67EH and pMMB67HE with the pSL1180 polylinker. The IncP plasmids pWM265 and pWM266 are derivatives of pDN18 and pDN19 with the pSL1180 polylinker. The large number of unique restriction sites present in these vectors allows removal of DNA from either end of the plasmid insert without the need for rigorous restriction mapping prior to deletion construction. Deletions are easily constructed by digestion with any enzyme that cuts uniquely within the polylinker and any number of times within the insert. After subsequent ligation and transformation, deletion plasmids that lack all DNA between a polylinker restriction site and the most-distal site within the cloned region are recovered. The extent of the deleted DNA fragment(s) is determined subsequently by restriction analysis of the remaining fragment.

View larger version (36K): [in this window] [in a new window]   FIG. 2.   Deletion analysis of the hypophosphite- and phosphite-oxidizing functions encoded by the plasmid pWM239. A series of deletion derivatives of the plasmid pWM239 were constructed and tested for expression of the hypophosphite and phosphite oxidation phenotypes in E. coli S17-1 and P. aeruginosa PAK pil rif, as described in the text. The ability to confer growth in 0.4% glucose-MOPS medium containing hypophosphite (Hpt) or phosphite (Pt) as the sole phosphorus source is indicative of the ability to oxidize the indicated compound to phosphate. Examination of the P oxidation phenotypes displayed by P. aeruginosa carrying the various deletion plasmids indicates that the shaded region between the KpnI2 and AseI sites is required for Pt oxidation. Further, these data suggest that oxidation of hypophosphite proceeds via a phosphite intermediate. Thus, P. aeruginosa strains carrying plasmids lacking the Pt region are also defective in hypophosphite oxidation. The ability to oxidize Pt in E. coli hosts, indicated by (+), is not related to the plasmids, because E. coli is a natural phosphite oxidizer. Therefore, deletions of the Pt region do not affect hypophosphite oxidation in E. coli, and the minimal region required for this phenotype can be determined by examination of the complementation pattern in this host. Accordingly, the shaded region between the SstI and NheI sites is required for hypophosphite oxidation in E. coli. The thick line at the top represents the cloned region of the P. stutzeri chromosome, with the restriction sites used for construction of individual deletions shown. The flanking restriction sites used for construction of each deletion were provided by the polylinker of the vector pWM265 (Fig. 1). The thin lines represent the remaining insert region of each deletion plasmid. Not all restriction sites within the insert were mapped for each enzyme; therefore, the sites shown are not necessarily unique.

Construction and phenotypic characterization of P. stutzeri WM88 mutants unable to oxidize phosphite and hypophosphite. To verify that the cloned genes described above are responsible for the observed traits of phosphite and hypophosphite oxidation by P. stutzeri WM88, we constructed a variety of deletion and insertion mutations within these genes and recombined them into the original host as described in Materials and Methods (Fig. 3). The observed hypophosphite and phosphite oxidation phenotypes of these mutants are completely consistent with the hypothesis that hypophosphite is oxidized through a phosphite intermediate and verify that the cloned genes were indeed responsible for the ability of P. stutzeri WM88 to oxidize these compounds. These phenotypes also fully support the conclusion from the subcloning experiments as to the locations of the genes required for each trait.

View larger version (19K): [in this window] [in a new window]   FIG. 3.   P. stutzeri WM88 chromosomal mutations in the region linked to P oxidation phenotypes. The regions of the P. stutzeri WM88 chromosome putatively required for oxidation of hypophosphite (Hpt) and phosphite (Pt) (arrows) were identified by complementation of heterologous hosts (see the text and Fig. 2). Deletion and insertion mutations in clones of this genomic region were constructed in vitro and recombined onto the P. stutzeri WM88 chromosome as described in the text. The P oxidation phenotypes of these mutants were examined by scoring the ability to grow on media containing either Hpt or Pt as the sole P source as described in the text. These phenotypes confirm that the genes carried in this region are responsible for the oxidation of phosphite and hypophosphite by the original isolate (see text for discussion). The thin line at the top indicates the chromosomal region of P. stutzeri under study, with relevant restriction sites shown. Not all sites for each enzyme have been mapped; therefore, the sites shown are not necessarily unique. The thick lines below this represent the extents of the in vitro-constructed deletion mutations. The triangles show the sites of insertion mutations.

Complementation of P. stutzeri deletion mutants with nested plasmid deletions. The final verification of the DNA regions required for oxidation of hypophosphite and phosphite was provided by complementation of the del1, del2, and del3 mutations (Fig. 3) in the native P. stutzeri host with selected plasmids from the deletion series (Fig. 2) described above. With the exception of the size of the fragment required for hypophosphite oxidation by P. stutzeri, these data are completely consistent with the results obtained from expression of the genes in heterologous hosts (Table 2).The data with respect to phosphite oxidation are essentially identical for the native and heterologous hosts. Thus, both the del1 mutation, which removes the entire region, and the del3 mutation, which removes only the region implicated in phosphite oxidation, can be complemented for the phosphite oxidation phenotype by pWM288. This plasmid carries the KpnI-to-AseI fragment previously shown to be the minimal DNA fragment capable of conferring phosphite oxidation upon the heterologous host P. aeruginosa. The reason for the slower growth of del1 hosts carrying pWM239, pWM276, and pWM275 in phosphite medium is unclear; however, these strains are clearly able to oxidize phosphite. This effect is not seen with the same plasmids in the del3 host.

DNA sequence analysis of the minimal region required for phosphite oxidation. The data presented above demonstrate that the 5.6-kbp KpnI fragment localized to the right end of the pWM239 insert carries the functions required for oxidation of phosphite to phosphate by P. stutzeri WM88. To gain further insight into the nature of these functions, the complete DNA sequence of this region, carried in pWM294, was determined as described in Materials and Methods (Fig. 4A and Table 3). Seven open reading frames (ORFs) encoding products with significant homology to proteins in the sequence databases were identified in this sequence. Only six of these ORFs reside within the AseI-to-KpnI fragment shown to carry the functions required for oxidation of phosphite to phosphate. Five of these, designated ptxA to ptxE (for phosphite oxidation) are likely to be involved in oxidation of phosphite to phosphate, because their products exhibit homology to proteins of known functions that suggest functions consistent with roles in P oxidation. Accordingly, PtxA, PtxB, and PtxC are likely to comprise a binding-protein-dependent phosphite transporter. PtxD is homologous to the family of D-isomer-specific 2-ketoacid dehydrogenases (15), suggesting that it may be an NAD-dependent phosphite dehydrogenase. PtxD exhibits 27 to 33% identity to various members of the family, including conservation of the NAD binding site and important catalytic residues (data not shown). The ptxE gene (truncated in this sequence) appears to encode a lysR family transcriptional regulator. The ptxABCDE' genes probably comprise a transcriptional unit. The genes overlap each other by a few bases (in the case of ptxA-ptxB and ptxD-ptxE) or are separated by at most 8 bp (ptxB-ptxC and ptxC-ptxD). The remaining two ORFs, orf86 and orf117, are homologous to a family of site-specific recombinases (2) and to a protein of unknown function in E. coli, respectively. Orf117 is not required for phosphite oxidation, because it lies outside the required region as defined by heterologous expression. A role for Orf86 in phosphite oxidation is formally possible but seems unlikely.

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Physical structures of DNA fragments shown to be required for oxidation of phosphite and hypophosphite by P. stutzeri WM88. The complete DNA sequences of both fragments were determined as described in the text. (GenBank accession no. AF061070 and AF061267). (A) Structure of a 5.6-kbp KpnI fragment encoding functions required for oxidation of phosphite to phosphate in P. stutzeri WM88. Seven ORFs, indicated by arrows, were identified within this sequence. The ptxE gene is truncated in this clone, as indicated by the partially shaded arrow. Five of these genes, designated ptxA through ptxE, are likely to be involved in oxidation of phosphite and probably form a single transcriptional unit. PtxABC probably comprise a binding-protein-dependent transport system for the uptake of phosphite. PtxD is probably an NAD+-dependent phosphite dehydrogenase, and PtxE is probably a transcriptional regulator for the ptxABCDE operon. (B) Structure of an 8.9-kbp SstI-to-NheI fragment encoding functions required for oxidation of hypophosphite to phosphite in P. stutzeri WM88. Nine ORFs, indicated by arrows and designated htxA through htxI, are likely to form a single transcriptional unit. Relevant restriction sites used for various plasmid constructions are shown. The BglII and AgeI sites shown in boldface were used as insertion sites for gene disruption experiments (Fig. 3). HtxA is a putative -ketoglutarate-dependent hypophosphite dioxygenase. HtxBCDE comprise a putative binding-protein-dependent hypophosphite transporter. The remaining genes encode subunits of a putative C-P lyase but are not required for oxidation of hypophosphite. The partially shaded arrow indicates that the htxI gene is truncated in this sequence. See the text and Table 3 for details. Approximately 15 kbp separates the two regions. This 15-kbp region is not required for oxidation of either compound and was not characterized.

DNA sequence analysis of the minimal region required for hypophosphite oxidation. The subcloning and complementation experiments outlined above demonstrate that the SstI-to-NheI fragment at the left end of pWM239 encodes genes involved in the oxidation of hypophosphite to phosphite. Therefore, the complete DNA sequence of this fragment, carried in pWM360, was also determined. Nine ORFs, designated htxA, htxB, htxC, htxD, htxE, htxF, htxG, htxH, and htxI (for hypophosphite oxidation), were identified in this sequence (Fig. 4B and Table 3). The nine genes are all transcribed in the same direction and probably form a transcriptional unit. The putative protein products of each of these ORFs display homology to proteins of known functions, suggesting possible roles for these genes in P. stutzeri WM88. The htxA gene is probably required for hypophosphite oxidation, because plasmids lacking this gene (deletion of the HpaI-to-NheI fragment) fail to confer hypophosphite oxidation upon E. coli or P. stutzeri del2 (Fig. 2 and Table 2). Further, this is the only clearly identifiable ORF within the HpaI-to-NheI fragment. HtxA is homologous to a family of -ketoglutarate-dependent dioxygenases that catalyze the oxidation of their respective substrates by using O₂ as the immediate electron acceptor (7). Most notably HtxA is 26.2% identical and 33.4% similar to proline hydroxylase from Dactylosporangium. Importantly, the sequence conservation between the two proteins includes regions shown to be important for substrate binding in other members of the family (data not shown). Based on its homology to this family of proteins and on the demonstration that the gene encoding this protein is required for hypophosphite oxidation, it is probable that HtxA is an -ketoglutarate-dependent hypophosphite dioxygenase.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Relatively few biologically mediated reactions involving oxidation or reduction of P are known. It seems likely, however, that this dearth of knowledge does not reflect the absence of such reactions in nature. On the contrary, reduced P compounds (phosphonates and phosphinates) are known to be common in many organisms (18), and production of the most-reduced P compound, phosphine, in natural systems has been clearly demonstrated (8, 11-14). Further, P-oxidizing organisms have been previously isolated (6, 9, 16). Our own studies suggest that P-oxidizing organisms may, in fact, be quite common. We inoculated numerous enrichments designed to select organisms capable of oxidizing hypophosphite to phosphate by use of this reduced compound as the sole P source. In every case these enrichments yielded P-oxidizing organisms, usually within a few days. Preliminary characterization of these organisms indicated that many species were represented. It must be noted, however, that none of the laboratory strains were tested displayed this phenotype, including several strains known to be of the same species as one of our isolates, P. stutzeri WM88. One of these, P. stutzeri DSM50227, is very similar to P. stutzeri WM88 as shown by its identical 16S RNA sequence. Interestingly, P. stutzeri DSM50227 was one of only two pseudomonads tested that was able to oxidize phosphite, a trait that we showed was essential for oxidation of hypophosphite by P. stutzeri WM88. We are uncertain of the reason that so many laboratory strains were unable to oxidize hypophosphite, while at the same time it was so simple to isolate related strains from nature that could. This is probably a reflection of the power of the enrichment technique used for isolation of these organisms; however, it is not uncommon for laboratory strains to differ from their wild counterparts.

The data presented indicate that P. stutzeri WM88 oxidizes hypophosphite to phosphate in a two-step pathway via a phosphite intermediate. Two lines of evidence support this conclusion. First, all plasmid clones that confer hypophosphite oxidation upon the heterologous host P. aeruginosa also confer the ability to oxidize phosphite. Conversely, all subclones that lack the ability to confer phosphite oxidation also lack the ability to confer hypophosphite oxidation. Second, similar results were obtained for mutants of P. stutzeri WM88. Thus, mutants unable to oxidize phosphite are always unable to oxidize hypophosphite, while mutants unable to oxidize hypophosphite but still able to oxidize phosphite are readily obtained. Complementation of these P. stutzeri mutants with the same plasmids further supports this conclusion.

The available evidence suggests that two novel P-oxidizing enzymes are involved in the process. First, the HtxA protein is likely to be an -ketoglutarate-dependent hypophosphite dioxygenase, based on its similarity to other known proteins of this type (7). Such an enzyme would use molecular oxygen as the direct electron acceptor for oxidation of hypophosphite by using -ketoglutarate as a cosubstrate and producing phosphite, succinate, and CO₂ as products. Although we have no direct evidence that this reaction occurs, we have recently shown that P. stutzeri is incapable of hypophosphite oxidation when grown anaerobically with nitrate as an electron acceptor. Phosphite oxidation was not impaired under similar growth conditions (data not shown). This datum suggests that oxygen may be required for hypophosphite oxidation and supports the conclusion that HtxA may be a dioxygenase. Oxidation of phosphite is likely to occur by the action of an NAD-dependent phosphite dehydrogenase activity encoded by the ptxD gene. PtxD is highly homologous to a large number of proteins of this general type, most notably those involved in the oxidation of 2-ketoacids (15). This enzyme may be similar to a previously described enzyme from P. fluorescens; however, that enzyme was never characterized in detail with purified protein, nor was its gene identified (27). We have recently demonstrated that PtxD possesses the predicted enzymatic activity and are currently purifying the protein to allow detailed biochemical characterization of this enzyme. These data will be reported elsewhere.

Two binding-protein-dependent transporters are likely to be involved in the oxidation of phosphite and hypophosphite in P. stutzeri WM88. The PtxABC transporter is probably involved in the uptake of phosphite, whereas the HtxBCDE transporter is probably involved in the uptake of hypophosphite. Each of these transporters is homologous to the E. coli PhnCDE transporter, which is responsible for uptake of phosphonates. This is relevant in that phosphonates are structurally similar to both phosphite and hypophosphite. Further, the E. coli PhnCDE transporter is known to transport phosphite in addition to phosphonates (29). Our evidence directly supports the conclusion that the HtxBCDE transporter is involved in uptake of hypophosphite in E. coli and P. aeruginosa. The observation that the genes encoding the transporter are not required for hypophosphite oxidation in P. stutzeri demonstrates that an additional hypophosphite transporter is present in the genome of this organism.

One possible explanation for the ability to oxidize phosphite and hypophosphite is that a broad-specificity C-P lyase may be responsible for the oxidation reactions. Because E. coli C-P lyase has been shown to oxidize phosphite to phosphate (30), it seems possible that other C-P lyases may be capable of oxidizing either phosphite or hypophosphite. Indeed, we demonstrated that genes in the same operon as htxA are highly homologous to E. coli and R. meliloti genes which encode C-P lyase subunits. Although these genes are not required for hypophosphite oxidation in P. stutzeri WM88, a role for C-P lyase cannot be excluded. Despite removal of the htxF, htxG, and htxH genes and part of the htxI gene, the deletion mutants we constructed retain C-P lyase activity and thus carry a second C-P lyase locus elsewhere on the genome. Further, the E. coli and P. aeruginosa hosts used for our complementation studies both contain the genes encoding C-P lyase. Regardless of this possibility, the predicted roles for HtxA and PtxD should be sufficient for hypophosphite and phosphite oxidation without invoking a role for C-P lyase. Unambiguous demonstration of the function of the genes isolated from P. stutzeri WM88 will await the results of studies now under way on the biochemical characterization of the enzymes they encode.

Whether or not C-P lyase is involved in oxidation of reduced P compounds in P. stutzeri WM88, the observed linkage of genes encoding a putative C-P lyase and the P-oxidizing enzymes found in this system is intriguing. C-P lyase is one of the only enzymes known to catalyze oxidation of reduced P compounds. It is capable of catalyzing the production of phosphate from either phosphonates or phosphite (30). Further, phosphonate and phosphinate production are among the few studied examples of biological production of a reduced P compound (34, 36). There is a known linkage between phosphonate production and the production of one of the reduced P substrates oxidized by P. stutzeri WM88, namely, phosphite. The common soil bacterium Streptomyces hygroscopicus produces the phosphinate antibiotic bialaphos. One of the known intermediates in the biosynthesis of this compound is phosphonoformate, an unstable compound that spontaneously decomposes to CO₂ and phosphite (34). Finally, although P. stutzeri PtxD is homologous to a large number of dehydrogenases at the protein level, its gene is not generally homologous to the respective genes at the DNA level (data not shown). One key exception is the homology of ptxD to a gene within a cluster of S. hygroscopicus genes involved in the biosynthesis of bialaphos, encoding a putative dehydrogenase thought to catalyze the oxidation of hydroxymethylphosphonate to phosphonoformate (17). Taken together, these data suggest a connection, albeit of uncertain nature, between the biosynthesis and catabolism of C-P bonds and the oxidation of the reduced P compounds phosphite and hypophosphite.

The isolation of numerous organisms capable of oxidizing reduced P compounds provides new evidence that a P redox cycle exists in nature. The large genomic fragment we isolated from P. stutzeri carries genes that allow assimilation of P from two or three types of reduced P compounds, i.e., phosphite, hypophosphite, and possibly phosphonates. Thus, it represents a chromosomal region largely dedicated to metabolism of reduced P compounds. The finding of genes that appear to be specifically dedicated to this process is a clear indication that these compounds are present in the environment and that it is advantageous for organisms to be able to utilize them. Further, our NMR studies demonstrate that hypophosphite spontaneously oxidizes to phosphite within a few months under aerobic conditions. Therefore, there is a continuous, or at least repetitive, production of these compounds in nature. Oxidation of reduced P compounds may serve two purposes. First, there is a great deal of energy available from these reactions. The ability to harness this energy would certainly be advantageous to an organism that possessed it. Second, phosphate is a limiting nutrient in many ecosystems. Thus, the ability to oxidize any reduced P compounds to phosphate is also of obvious merit.

The source of these reduced P compounds is of great interest but remains unclear at this time. Numerous reports demonstrating the presence of the most-reduced form of P, phosphine, in the environment now exist. We, and others, have shown the production of such compounds under anaerobic conditions; however, these experiments do not always result in the production of reduced P compounds (40). Thus, the environmental conditions resulting in P reduction remain to be determined. The overall importance of P redox cycling in nature is also unclear at this time, although the flux of P through this redox cycle can apparently be quite high, as shown by the loss of up to 45% of the incoming P as volatile compounds in some sewage treatment facilities (8). Further study of this intriguing process is clearly required.