phnE and glpT Genes Enhance Utilization of Organophosphates in Escherichia coli K-12

Abstract

Wild-type Escherichia coli K-12 strain JA221 grows poorly on low concentrations (≤1 mM) of diisopropyl fluorophosphate and its hydrolysis product, diisopropyl phosphate (DIPP), as sole phosphorus sources. Spontaneous organophosphate utilization (OPU) mutants were isolated that efficiently utilized these alternate sources of phosphate. A genomic library was constructed from one such OPU mutant, and two genes were isolated that conferred the OPU phenotype to strain JA221 upon transformation. These genes were identified as phnE and glpT. The original OPU mutation represented phnE gene activation and corresponded to the same 8-bp unit deletion from the cryptic wild-type E. coli K-12 phnE gene that has been shown previously to result in phnE activation. In comparison, sequence analysis revealed that the observed OPU phenotype conferred by the glpT gene was not the result of a mutation. PCR clones of glpT from both the mutant and the wild type were found to confer the OPU phenotype to JA221 when they were present on the high-copy-number pUC19 plasmid but not when they were present on the low-copy-number pWSK29 plasmid. This suggests that the OPU phenotype associated with the glpT gene is the result of amplification and overproduction of the glpT gene product. Both the active phnE and multicopy glpT genes facilitated effective metabolism of low concentrations of DIPP, whereas only the active phnE gene could confer the ability to break down a chromogenic substrate, 5-bromo-4-chloro-3-indoxyl phosphate-p-toluidine (X-P_i). This result indicates that in E. coli, X-P_i is transported exclusively by the Phn system, whereas DIPP (or its metabolite) may be transported by both Phn and Glp systems.

Organophosphorus compounds (OPCs) constitute the largest class of pesticides currently used in industrialized and developing nations. They are also produced for a variety of other applications in agriculture, industry, medicine, and chemical warfare (5, 10, 34). Many of these compounds are known to inhibit mammalian cholinesterases, including acetylcholinesterase, which leads to accumulation of acetylcholine at all cholinergic terminals and a resultant blockage of neural signal transmissions. The immediate and long-term effects that result from the impairment of the central nervous system and muscarinic and nicotinic receptors are broad and complex and vary from mild discomfort to coma and death (1, 21, 22, 45). The breadth and magnitude of OPC applications have resulted in high numbers of incidents of human poisoning and environmental concern (5, 28, 34).

Microbial degradation is a very important part of the metabolism of OPCs in the environment (5, 14, 18, 19, 34, 39, 42). Although microorganisms capable of degrading many of the OPC pesticides have been isolated, knowledge about the biochemical pathways and the genes involved in degradation is sparse (5, 19, 34).

We studied microbial utilization of the model OPC diisopropyl fluorophosphate (DFP) (used in ophthalmic medications and in the treatment of myasthenia gravis) and its breakdown product, diisopropyl phosphate (DIPP), in Escherichia coli K-12. It has been reported previously that E. coli K-12 possesses hydrolytic enzymes that detoxify OPCs, including DFP, soman, and paraoxon (16, 17, 54). Crude extracts of E. coli K-12 strain JA221 are capable of hydrolyzing DFP to DIPP and hydrofluoric acid (unpublished data). This hydrolysis can also take place at a slower rate spontaneously in an aqueous environment. Further utilization of DIPP as a phosphorus source is presumed to proceed through phosphoester hydrolysis, yielding monoester and finally inorganic phosphate (19). However, certain E. coli strains (e.g., ATCC 11775) do not grow well on phosphodiester (dimethyl phosphate) or phosphomonoester (methyl phosphate) substrates as sole phosphorus sources (52), indicating that the initial breakdown of DFP to its phosphodiester product, DIPP, may not be the rate-limiting step for its ultimate utilization.

In this study, we show that the wild-type E. coli strain JA221 does not effectively utilize low concentrations of DFP and DIPP as sole phosphorus sources. To understand the metabolism of organophosphates, we isolated mutants of JA221 that exhibit enhanced utilization of DFP and DIPP as sole phosphorus sources. Here we report characterization of these mutants and cloning and characterization of two genes involved in utilization of these OPCs by E. coli.

MATERIALS AND METHODS

Chemicals, media, and growth conditions.

DFP was obtained from Aldrich Chemical Company (Milwaukee, Wis.). The chromogenic substrate 5-bromo-4-chloro-3-indoxyl phosphate-p-toluidine (X-P_i) was obtained from Bachem BioScience Inc. (King of Prussia, Pa.). DIPP was a kind gift from William T. Ledford (Syntheco, Inc.). Unless otherwise specified, all other chemicals were obtained from Sigma Chemical Company (St. Louis, Mo.) and were of the highest purity available.

MOPS modified medium (MMM) (modified from the medium of Neidhardt et al. [40]) contained (per liter) 8.372 g of 3-(N-morpholino)propanesulfonic acid (MOPS), 0.717 g of N-tris(hydroxymethyl)methyl glycine (Tricine), 2.92 g of NaCl, 0.51 g of NH₄Cl, 6 mg of MgSO₄ · 7H₂O, 3 mg of nitrilotriacetic acid, 48 mg of K₂SO₄, 102 mg of MgCl₂ · 6H₂O, 1 mg of MnSO₄ · H₂O, 2.8 mg of FeSO₄ · 7H₂O, 0.1 mg of CaCl₂ · 2H₂O, 0.1 mg of CoCl₂ · 6H₂O, 0.1 mg of ZnSO₄ · 7H₂O, 0.02 mg of H₃BO₃, 0.01 mg of Na₂MoO₄ · 2H₂O, 0.01 mg of CuSO₄, 30 mg of leucine, 20 mg of tryptophan, and 20 g of glucose, and the pH was adjusted to 7.4 with KOH. Purified agar (15 g; catalog no. A 7921; Sigma Chemical Co.) and 12 g of Na₃ citrate · 2H₂O per liter were used to prepare solid media. MOPS rich medium (MRM) was MMM fortified with 10 ml of a vitamin mixture (catalog no. B6891; Sigma Chemical Co.) and 20 ml of an amino acid mixture (catalog no. R7131; Sigma Chemical Co.) per liter.

Organisms were routinely grown overnight at 37°C in an incubator-shaker at 200 rpm in broth or at 34°C on plates containing Luria-Bertani (LB) medium (41) supplemented with ampicillin.

Utilization of alternate phosphorus sources.

When utilization of alternate phosphorus sources was tested, MMM or MRM was used. Unless otherwise indicated, MMM was used since it gave results similar to the results obtained with MRM but required slightly longer growth periods. Ampicillin was used at a concentration of 100 μg/ml. Unless otherwise indicated, phosphorus sources were added (after the pH values of stock solutions were adjusted) as follows: 4 mM potassium phosphate for P_i-rich medium, 1 mM DFP (from a 1 M stock solution in isopropanol), and 1 mM DIPP. The medium containing DFP was used immediately after preparation. The concentration of the chromogenic substrate X-P_i was 0.1 mM in liquid X-P_i-containing medium, and 1 μmol (100 μl of a 10 mM stock solution) was spread onto P-deficient 90-mm plates 1 h prior to use. No phosphorus source was added to P-deficient media. To test for phosphorus source-dependent growth, bacteria were routinely grown in LB broth overnight, washed three times with P-deficient MRM broth, appropriately diluted, and used to inoculate broth or solid medium plates. Unless otherwise indicated, organisms were grown on solid media at 34°C for 34 to 40 h (for MRM) or 36 to 42 h (for MMM).

To ascertain the lag periods and generation times, bacteria were grown in MMM broth supplemented with 0.2 mM P_i (a growth-limiting concentration) for 18 h so that all of the P_i in the medium would be exhausted, appropriately diluted, and used as inocula. The generation times were calculated by using exponentially growing cells and regression analysis (20).

Plasmids and strains.

The low-copy-number plasmid vector pWSK29 was a kind gift from Sidney R. Kushner (48). The high-copy-number plasmid vector pUC19 was obtained from Boehringer Mannheim Biochemicals (Indianapolis, Ind.). E. coli K-12 strain JA221 (F⁻ hsdM⁺ hsdR lacY leuB6 ΔtrpE5 recA1 λ⁻) (ATCC 33875) and the library of Alteromonas haloplanktis 214 variant 3 in E. coli K-12 strain JA221 (ATCC 37436) (32) were obtained from the American Type Culture Collection (Rockville, Md.). Wild-type E. coli JA221 transformed with pBR322 or pUC19 was used appropriately as a control with test organisms containing these plasmids. E. coli DH10B [F⁻ mcrA Δ(mrr-hsdRMS-mcrBC) φ80dlacZΔM15 ΔlacX74 endA1 recA1 deoR Δ(ara leu)7697 araD139 galU galK nupG rpsL] was obtained from Life Technologies (Gaithersburg, Md.).

The pFΔRL construct was made by deleting the EcoRI fragment from the EcoRI sites in the pF1 genomic DNA insert and the polylinker of pUC19; the excised fragment was ligated into a pUC19 EcoRI site to generate pFΔRR. A third construct (pFΔF) was made by excising the EcoRV fragment from the pF1 clone, which deleted phnF completely (also phnG and part of phnH) without affecting the phnE gene (Fig. 1A).

FIG. 1.

Functional map of phnE and glpT genes. All clones and constructs were used to transform the wild-type strain E. coli JA221 and were tested for the ability to confer vigorous growth on plates containing 1 mM DIPP as described in Materials and Methods in order to determine the functional genes. The locations and directions of transcription of the ORFs are indicated by arrows. A plus sign indicates positive growth on DIPP-containing plates, corresponding to the growth of E. coli mutant DB. A minus sign indicates negative growth on DIPP-containing plates, equivalent to the growth of the wild-type strain E. coli JA221 transformed with the control vector. Restriction enzyme cutting sites are indicated as follows: B, BglI; P, PvuII; S, PstI; N, NsiI; H, HindIII; R, EcoRI; and V, EcoRV. (A) phnE: five overlapping clones of genomic DNA obtained from E. coli mutant DB (pB1, pB3, pB12, pE2, and pF1) and three constructs derived from the pF1 clone (pFΔRL, pFΔRR, pFΔF). No sites were found for BamHI, HindIII, and PstI. (B) glpT: genomic clone (pB9) obtained from the DB mutant, a derivative construct (pqTa), and PCR-amplified regions of the glpT gene (PCRpT). No sites were found for BamHI and EcoRI.

The pqTa construct was made by deleting a fragment from an NsiI site in the insert of the genomic (pB9) clone and a PstI site in the polylinker of pUC19, which removed glpBC and all but 259 bp of glpA (Fig. 1B).

PCR.

The 336- and 228-bp PCR constructs that contained 8-bp repeats in phnE were obtained for sequencing by using the 16-mer IEF1 forward primer (5′-GTTTACCAGCCCGTTC-3′) and the 16-mer IER1 reverse primer (5′-CTCTTCGAGCTTGTTG-3′) from JA221 and the DB mutant, respectively.

Using the 16-mer primers T621F (5′-GTGCTCATCATGATCG-3′) and T2206R (5′-CGTGATTTCATGCGTC-3′), we obtained the 1,586-bp PCRpT constructs, which contained the glpT open reading frame (ORF), 176 bp of upstream flanking DNA, and 52 bp of downstream flanking DNA, by PCR amplification of the glpT gene from the following three sources: (i) plasmid pB9; (ii) genomic DNA from the original OPU isolate (DB); and (iii) genomic DNA from wild-type E. coli JA221. The inserts were excised by digestion with XhoI and HindIII and were ligated to SalI- and HindIII-digested pUC19 and pWSK29 vectors. The resultant PCRpT-pB9, PCRpT-DB, and PCRpT-JA constructs were used to chemically transform E. coli wild-type strain JA221.

DNA sequencing.

Universal primers were used to sequence flanking DNA regions of the insert in the pUC19 vector. To obtain DNA sequences of the genomic DNA regions, the sections of interest were amplified by PCR. The products were ligated to the pCRII vector by using the Invitrogen protocol (Invitrogen Corporation, San Diego, Calif.). Ligated plasmids were grown in E. coli DH10B, and both strands of the PCR-amplified region were sequenced. Eight 16-mer oligonucleotide primers, T621F, T1023F (5′-AGTCGCGATGGTCACC-3′), T1419F (5′-GGCGAATAATGCCACC-3′), T1805F (5′-AAGCAATCGAGATCCC-3′), T998R (5′GGTAACCCAACCGTCG-3′), T1393R (5′-AATCCTGTGGCTTGCC-3′), T1790R (5′-TTCATCATGGGTTCGG-3′), and T2206R, were used to sequence the entire PCRpT region that contained glpT. DNA sequencing was conducted by the dideoxy chain termination method (44).

Genomic DNA purification.

Genomic DNA was purified with a Qiagen kit (Qiagen Inc., Valencia, Calif.) by using a modified protocol. E. coli genomic DNA was precipitated with isopropyl alcohol (prior to application to the column), spooled, rinsed with 70% ethanol, and dissolved in a minimum volume of TE buffer (43). The concentrations of salts and the pH were adjusted (to values comparable to the values for buffer G2 of the Qiagen kit), and the DNA was purified on the Qiagen column by using the manufacturer’s recommendations.

Electroporation.

Electrocompetent E. coli DH10B was transformed by plasmids by electroporation, using a Bio-Rad Gene Pulser electroporation system, a Pulse Controller, and 0.1-cm cuvettes at 2,500 V. Pulsed cells were suspended in SOC (43), incubated at 37°C for 60 min, and plated onto LB plates supplemented with ampicillin, and incubated at 34°C for 13 h.

Cloning.

To isolate the OPU mutated gene(s) that conferred robust growth to E. coli JA221 on DFP-containing and DIPP-containing plates, we used standard molecular cloning techniques (3, 41, 43). The OPU mutant E. coli DB, which contained a ca. 5.6-kb library fragment on pBR322, was cured of the plasmid by using coumermycin A₁ (8). Genomic DNA was purified from the plasmid-cured DB mutant and was partially digested with restriction endonuclease Sau3AI. DNA fragments (lengths, 4 to 20 kb) were purified by centrifugation (30,000 rpm for 28 h in an SW41 rotor at 4°C) in a sucrose density gradient (10 to 40% sucrose in 2 mM Tris–1 mM EDTA, pH 8.2) (41) and were ligated with DNA ligase to pUC19 that previously had been digested with restriction endonuclease BamHI and dephosphorylated with bacterial alkaline phosphatase (Sigma Chemical Co.). The ligated plasmids were used to transform electrocompetent E. coli DH10B by electroporation. Approximately 83,000 transformed colonies were pooled, and plasmids were purified by the alkaline lysis method (43). The resultant genomic DNA library was used to chemically transform E. coli wild-type strain JA221. Approximately 60,000 transformed cells were plated onto MRM supplemented with 1 mM DIPP as the sole phosphorus source.

Fifty colonies were identified that exhibited good growth on DIPP-containing plates. Plasmids were individually purified from 36 of these colonies and used to chemically transform E. coli wild-type strain JA221. The plasmids that conferred good growth to the bacterium on DIPP-containing plates were selected for further study.

RESULTS

Mutant isolation.

The goal of this study was to identify genes involved in the bacterial metabolism of DFP or its hydrolysis product, DIPP. In particular, we searched for genes that significantly enhanced the utilization of these substrates as sole phosphorus sources.

Originally, we screened a genomic library of A. haloplanktis in E. coli K-12 strain JA221 (32) for genes that enhanced utilization of DFP, since the library was expected to contain the gene for a highly active diisopropyl fluorophosphatase (EC 3.1.8.2) (9, 13). The library was plated onto medium containing DFP as the sole phosphorus source. Cells were also plated onto DIPP-containing, P-deficient, and phosphate-rich media. The assumption was that clones carrying the plasmid with the diisopropyl fluorophosphatase gene would exhibit more vigorous growth on the DFP-containing plates than on the DIPP-containing plates. However, no clone was identified that exhibited such preferential growth on DFP-containing plates.

Nevertheless, we were able to isolate a few spontaneous mutant bacterial clones that consistently grew well on both DIPP-containing (Fig. 2A) and DFP-containing plates compared with E. coli JA221. These isolates (DB, DC, FB, FC, and FL) were designated organophosphate utilization (OPU) clones. To test whether the OPU phenotype resulted from the A. haloplanktis library plasmid or from the mutation in the E. coli genome, we obtained the plasmids from the OPU isolates. However, after retransformation, these plasmids failed to confer robust growth to E. coli JA221 on DFP- or DIPP-containing plates. Furthermore, plasmid curing of the OPU isolates did not affect the OPU phenotype, indicating that DFP and DIPP utilization was the result of a spontaneous mutation on the chromosomes of the OPU strains and was not due to a plasmid-borne A. haloplanktis gene.

FIG. 2.

E. coli growth with different DIPP concentrations as the sole P sources. Plates were photographed following 40 h of incubation at 34°C. (A and B) Enhanced growth of OPU mutants with DIPP as the sole P source. Five mutant strains (DB, DC, FB, FC, and FL) and the wild-type strain E. coli JA221 were used. (A) Medium containing 1 mM DIPP. (B) P-deficient and P_i-rich media. The tiny colonies of the wild-type strain on DIPP-containing plates (A) and of both the wild-type strain and mutant DB on P-deficient plates (B) could not be reproduced by photography. (C) OPU phenotype conferred by the multicopy glpT gene is independent of the gene source. The wild-type strain was transformed with the pUC19 vector carrying the PCR-amplified glpT gene from the wild-type JA221 genome (JA), the DB mutant genome (DB), the pB9 clone (pB9), or the pUC19 vector alone (control). Following transformations cells were plated onto medium containing 1 mM DIPP.

Some mutants (FB and FC) exhibited enhanced growth on all plates, including both P_i-rich and P-deficient plates (unpublished data), which indicated that these mutants resulted from mutations in genes that affect general growth. However, other mutants (DB, FL, and DC) grew like wild-type strain JA221 on P_i-rich and P-deficient plates (unpublished data) but exhibited enhanced growth compared with the wild type on DIPP-containing plates (Fig. 2A). One of these mutants (DB) was selected for further study. The DB mutant and wild-type colonies grew to similar sizes on P-deficient medium (tiny colonies that could not be reproduced by photography) and P_i-rich medium (Fig. 2B). In comparison, the DB mutant grew significantly better than the wild type on DIPP-containing plates (Fig. 2A) and on DFP-containing plates (unpublished data). No detectable growth was observed in P-deficient liquid media for all strains.

The growth of the DB mutant as a function of the DIPP concentration was determined in broth. The DB mutant grew much better than the wild type at low DIPP concentrations, whereas the growth of the two strains was similar at high DIPP concentrations (Fig. 3A). This finding indicates that the mutation lowered the DIPP concentration at which this OPC is optimally utilized.

FIG. 3.

E. coli growth in broth with different DIPP concentrations as the sole P sources. After 13 h (A) or 20 h (B through D) of growth at 37°C, total cell growth was determined turbidimetrically. Every point with an error bar (standard deviation) represents the average of a minimum of three determinations. (A) Mutant DB mutation enhances growth only at low DIPP concentrations. (B) wild-type transformed with active phnE acquires the DB mutant phenotype. The wild type transformed with the pUC19 vector alone and the wild type transformed with pUC19 carrying the phnE gene on the pFΔRR construct were used. (C and D) Multicopy glpT gene enhances growth at low DIPP concentrations and inhibits growth at high DIPP concentrations, but a low copy number of glpT has not effect. The wild type was transformed with the PCR-amplified glpT gene from the wild-type genome on the multicopy pUC19 plasmid (C) or the low-copy-number pWSK29 plasmid (D) and also with the appropriate control pUC19 and pWSK29 vectors alone. A₆₀₀, absorbance at 600 nm.

Cloning.

In order to understand the nature of the mutation, experiments were conducted to identify the mutated gene. We assumed that the mutation was a dominant gene mutation, and a genomic library was constructed from the DB mutant in a high-copy number-plasmid (pUC19) and was used to transform E. coli wild-type strain JA221. Transformants were screened on DIPP-containing plates for the DB mutant growth (OPU) phenotype (i.e., vigorous growth on the DIPP-containing medium and weak growth on the P_i-deficient medium). Six positive clones were identified.

Subcloning experiments performed with selected restriction endonucleases (PvuII, EcoRI, BamHI, HindIII, PstI, BglI, and EcoRV) revealed the possibility that there were two different genes. Five of the clones (pB1, pB3, pB12, pE2, and pF1) appeared to represent the same contiguous area of genomic DNA, and two of these clones (pB12 and pE2) generated identical restriction fragment (Fig. 1A). The possible second gene was represented by a single isolate, pB9 (Fig. 1B).

pF1, the smallest of the five clones in the first set of genomic clones, was selected for further analysis (Fig. 1A). Two subclones of pF1 (pFΔRL and pFΔRR) were tested, and a 1.9-kb fragment of pFΔRR was found to contain the minimal sequences needed to confer vigorous growth on DIPP-containing plates to wild-type E. coli strain upon transformation. Flanking regions of pFΔRR were sequenced and were found to correspond to a part of the previously reported phnCDEFGHIJKLMNOP operon (6, 33) that is involved in the transport of phospho compounds. The pFΔRR construct contained the complete phnE and partial phnF ORFs. In order to ascertain which of these two genes was the gene of interest, a third construct was made from the pF1 clone; in this construct phnF, phnG, and part of phnH were deleted without affecting the phnE gene. Upon transformation, the resulting pFΔF construct (Fig. 1A) was found to confer the vigorous growth phenotype on DIPP-containing plates to E. coli wild-type strain JA221. Therefore, we concluded that phnE was the gene of interest.

To identify the second gene conferring the OPU phenotype, regions of the pB9 insert were sequenced. We found that this clone contained four previously identified ORFs (glpT, glpA, glpB, and glpC) belonging to two different operons (7, 12, 38). The glpABC operon encodes the anaerobic sn-glycerol-3-phosphate dehydrogenase (12), while the divergently transcribed glpT gene encodes a transport protein for sn-glycerol-3-phosphate (11, 12, 24, 26, 31). To identify which of the glp genes conferred the OPU phenotype, a subclone construct was obtained in which glpBC and all but 259 bp of glpA were deleted from the pB9 clone without affecting the glpT gene (Fig. 1B). The resulting pqTa construct in the high-copy-number pUC19 plasmid was found to confer vigorous growth on DIPP-containing plates to E. coli wild-type strain JA221 upon transformation. Therefore, we concluded that glpT was the second gene of interest.

The phnE gene has been identified in E. coli K-12 as a cryptic gene which can undergo spontaneous activation via an 8-bp deletion in the triple 8-bp tandem repeats present inside its ORF (33). To ascertain whether the DB mutant strain was the result of such a mutation, both the wild-type and mutant genomic DNAs encompassing the triple repeat were sequenced. Sequencing confirmed that the phnE gene in the DB mutant underwent spontaneous activation by excision of the same 8-bp unit.

To determine whether the OPU DB mutant contained an additional mutation in glpT, the complete glpT ORF of this strain (including the upstream 233-bp regulatory region) was sequenced. The sequence obtained was found to be identical to the previously reported glpT gene sequence (7, 12). This demonstrated that the OPU phenotype associated with glpT was not due to a spontaneous mutation in the gene. Furthermore, to test whether the OPU phenotype conferred by the glpT gene was due to an enhanced copy number of the gene, glpT genes (including upstream 176 bp of the regulatory region) were obtained from wild-type E. coli JA221 and OPU mutants and were subcloned into both the high-copy-number pUC19 vector and the low-copy-number pWSK29 vector. The resulting constructs were used to transform wild-type E. coli JA221 and were subsequently tested for the presence of the OPU phenotype. All transformants harboring the glpT gene on the high-copy-number pUC19 plasmid exhibited robust growth on DIPP-containing plates (Fig. 2C). In comparison, all transformants harboring the glpT gene on the low-copy-number pWSK29 plasmid (six to eight copies per cell [48]) exhibited weak growth on DIPP-containing plates (unpublished data). These results suggest that the OPU phenotype conferred by glpT on the pUC19 plasmid was not the result of a mutation but was due to the high level of expression of the gene carried on the multicopy vector.

Utilization of organophosphorus sources.

To ascertain the effects of the glpT and active phnE genes on the metabolism of various organophosphates, the growth of the mutant and the growth of wild-type strain JA221 transformed with these two genes were compared with the growth of the wild-type strain transformed with the appropriate vector alone on media containing different organophosphates. The cells containing an active phnE gene (the DB mutant and the pFΔRR E. coli transformant) metabolically broke down the chromogenic phosphorus source X-P_i, producing a characteristic blue metabolite (13). The wild-type strain transformed with the active phnE gene (the pFΔRR construct) was phenotypically identical to the E. coli DB mutant on X-P_i-containing media. In comparison, the strains lacking the active phnE gene (the wild-type strain and alone the wild-type strain transformed with the multicopy glpT gene) showed no evidence of X-P_i metabolism. Similar results were observed on the X-P_i-containing plates and when the X-P_i-containing liquid medium was supplemented with a growth-limiting concentration (0.1 mM) of P_i (13). Thus, the mutated phnE gene facilitates utilization of X-P_i, whereas the multicopy glpT gene does not.

In comparison to the all-or-nothing effects of the active phnE and glpT genes on the metabolism of X-P_i, the influence of these genes on utilization of DFP and DIPP as sole phosphorus sources was more complex. The wild-type cells containing inactive phnE did exhibit some utilization of DFP and DIPP as sole phosphorus sources, albeit at a much reduced rate compared with active phnE cells. In liquid media containing low DFP (unpublished data) and DIPP concentrations, cells with an active phnE gene exhibited stronger growth than wild-type phnE cells, whereas at high DIPP concentrations the gene appeared not to have a significant effect (Fig. 3A and B). In the case of the multicopy glpT gene, growth of E. coli JA221 was enhanced at low DIPP concentrations in liquid medium. However, the multicopy glpT gene inhibited growth at high DIPP concentrations (Fig. 3C). The glpT gene carried on the low-copy-number pWSK29 plasmid appeared not to affect growth significantly at all DIPP concentrations (Fig. 3D).

Crude cell extracts of E. coli wild-type strain JA221 transformed with the active phnE gene or the vector alone were compared to determine their phosphatase and phosphodiesterase activities. We observed no significant differences between these extracts in their abilities to break down X-P_i, p-nitrophenyl phosphate, and bis(p-nitrophenyl) phosphate (unpublished data).

To gain further insight into the effects of the glpT and active phnE genes on E. coli growth in DIPP-containing broth, a comparative time course study was performed. In this experiment, the effects of the glpT and active phnE genes on the doubling and lag times were tested at different concentrations of DIPP as sole phosphorus sources and in the P_i-rich medium (Table 1). This study revealed that growth enhancement in the presence of low concentrations of DIPP by both active phnE (mutant DB and phnE in pUC19) and glpT (in pUC19) genes was achieved through a significant reduction in the cell doubling time. As an increase in the DIPP concentration led to a corresponding decrease in the doubling time of the wild type, the growth-enhancing effect of the active phnE gene became less pronounced (Table 1). In contrast, the inhibitory effect of the multicopy glpT gene on cell growth reflected a concentration-dependent increase in the lag time. In addition, the doubling time was significantly increased at the highest DIPP concentration tested (25 mM). No significant effect was observed on either doubling or lag times with the glpT gene carried on the low-copy-number pWSK29 plasmid at all DIPP concentrations (Table 1). No measurable growth was observed in P-deficient medium, indicating that no significant carryover of P_i from the preincubation media occurred (unpublished data). Together, these studies demonstrated that the active phnE and glpT genes have distinct effects on utilization of X-P_i and DIPP by E. coli.

TABLE 1.

Effects of phnE and glpT on E. coli growth

P source	Concn (mM)	Doubling time (min)						Lag time (min)
		pUC19 inserts			pWSK29 inserts		Mutant DB	pUC19 inserts			pWSK29 inserts		Mutant DB
		None	phnE	glpT	None	glpT		None	phnE	glpT	None	glpT
Pi	2	76 ± 3a	90 ± 4	73 ± 1	66 ± 1	66 ± 1	70 ± 3	65 ± 6	41 ± 7	71 ± 3	2 ± 2	4 ± 3	10 ± 4
DIPP	0.3	397 ± 12	116 ± 10	93 ± 4	304 ± 17	307 ± 23	89 ± 3	127 ± 50	33 ± 10	35 ± 25	209 ± 34	158 ± 17	10 ± 10
	1	204 ± 3	115 ± 4	81 ± 8	178 ± 7	161 ± 6	91 ± 1	57 ± 2	66 ± 8	147 ± 14	17 ± 3	43 ± 12	10 ± 5
	5	113 ± 5	116 ± 3	93 ± 3	96 ± 3	91 ± 2	78 ± 6	54 ± 5	12 ± 10	216 ± 4	10 ± 7	5 ± 5	10 ± 10
	25	98 ± 2	104 ± 4	148 ± 4	71 ± 1	69 ± 1	72 ± 2	40 ± 5	57 ± 5	384 ± 16	2 ± 2	3 ± 2	10 ± 10

^aMean ± standard error. 

DISCUSSION

This study demonstrated that utilization of low concentrations of DIPP can be enhanced through activation of phnE or overexpression of glpT. phnE has been shown to be a cryptic gene in E. coli K-12 whose spontaneous activation is the result of a mutation (33, 47, 49) that deletes a single 8-bp unit from a triple 8-bp repeat in the coding region (33). The resulting frameshift not only alters downstream amino acids but also annuls the early stop codon of the wild-type gene (33), resulting in elongation of the PhnE protein from 189 to 259 amino acids. Sequencing confirmed that the E. coli JA221 mutation isolated in this study also represents this 8-bp unit deletion from the wild-type genome. It is noteworthy that the phnE gene is functional in the wild-type E. coli B strain (6) due to the absence of the same 8-bp insert (33). Slipped-strand mispairing during DNA synthesis in a recA E. coli strain (29) is thought to be the mechanism by which this 8-bp unit insertion or deletion might arise (33).

The metabolic breakdown of X-P_i and the enhanced utilization of DIPP in strains with active phnE do not reflect increased activity of X-P_i- and DIPP-hydrolyzing enzymes in these organisms. The transformations with the active phnE gene did not affect the expression of phosphodiesterase and phosphomonoesterase in E. coli wild-type strain JA221, as judged from the enzymatic activities in the crude extracts. It should be noted that the E. coli strain used in this study, JA221, like strains used in previous studies (33, 36), appears to be defective in the periplasmic alkaline phosphatase (phoA) gene since it failed to metabolize X-P_i in the absence of active phnE. X-P_i is normally broken down in other strains of E. coli by the phoA gene product (36, 46). Thus, consistent with previous reports (6, 36), phnE operates at the level of phosphorus compound transport and not phosphorus compound enzymatic breakdown.

The likelihood that phnE codes for an integral membrane protein was demonstrated by a Kyte-Doolittle hydropathy plot (23), which showed that the PhnE protein is quite hydrophobic (6) and has six putative transmembrane segments (unpublished data). This transmembrane arrangement is the pattern that is typical of the integral membrane proteins of the binding protein-dependent transport systems, where they have been found to form dimers (15). Therefore, since the PhnE protein is the single integral membrane component, it presumably functions as a homodimer.

All of the phospho compounds previously shown to be affected by the PhnCDE system (36, 37) have at most only a single substituent at the phosphorus atom (48a). Our studies indicate that active phnE is also involved in enhancing the utilization of DIPP, which has two substituents at the phosphorus atom. This suggests that DIPP may be converted to isopropyl phosphate (IPP) (monosubstituted at the phosphorus atom) by a periplasmic phosphodiesterase prior to its transport into the cytoplasm, where it is converted to P_i by a cytoplasmic phosphatase(s). Alternatively, it is equally plausible that the PhnCDE system directly transports DIPP. If this is the case, the entire degradation of DIPP to yield P_i is a cellular process that involves a cytoplasmic or cytoplasmic membrane phosphodiesterase(s) and a cytoplasmic phosphatase(s).

It is noteworthy that the phnE gene is functional in most E. coli strains and in bacterial species belonging to other genera of the family Enterobacteriaceae, including the genera Citrobacter, Enterobacter, Hafnia, Klebsiella, and Serratia (49). Therefore, the PhnCDE transport system is likely to facilitate the mineralization of DFP, DIPP, and other OPCs by a number of environmental bacteria.

We also observed that E. coli utilization of low concentrations of DIPP could be enhanced by the multicopy glpT gene encoding sn-glycerol-3-phosphate permease (2, 11, 12, 24, 26, 30, 31). The observed vigorous growth on DIPP conferred by glpT to E. coli was not due to a glpT gene mutation but was due to the high level of expression of the gene carried on the high-copy-number pUC19 plasmid. In E. coli, glpT is normally repressed by the glycerol-3-phosphate-inducible GlpR repressor (25, 27). In addition, the gene is under catabolite repression (25, 27, 30); i.e., it is negatively regulated by glucose. Therefore, in a glucose-containing and glycerol-deficient medium, such as the medium used in our study, the level of expression of glpT is expected to be low, resulting in a low basal level of transport. This basal level of transport may in fact account for the slow but detectable growth of the wild-type E. coli strain (lacking active phnE) at low DIPP concentrations and its more robust growth at high DIPP concentrations (Fig. 3 and Table 1).

The presence of glpT on a low-copy-number plasmid (six to eight copies per cell [48]) did not enhance utilization of DIPP (Fig. 3D and Table 1). It appears that a very high number of copies of glpT is needed to titrate the repressor molecules to an extent that allows sufficient expression of glpT to affect its function significantly.

At low DIPP concentrations, both the active phnE gene and the multicopy glpT gene enhanced E. coli growth. However, at high DIPP concentrations, the multicopy glpT gene inhibited E. coli growth, while the active phnE gene had no effect. It appears that while the glpT product is capable of transporting elevated toxic levels of its substrate, the phnE product transports only low concentrations of the phosphorus compound. This difference could be the result of differential regulation of the respective transport systems. PhnE is part of the multicomponent transport system that is the product of phnCDE, the expression of which is under control of the phosphate (pho) regulon (33, 35, 37, 50). During P_i limitation, the pho regulon is derepressed, resulting in increased expression of the Phn transport proteins and uptake of the substrate (50). After the phospho compound substrate is transported into the bacterial cytoplasm, it is metabolized, and P_i is released. The released P_i then interacts with the pho box of the phn operon, downregulating expression of the phnCD chromosomal genes and resulting in a decrease in further substrate uptake. Such feedback control by the internally released P_i has been observed for another pho-regulated, Ugp (uptake of glycerol phosphate) transport system (4, 53). In contrast, the GlpT transport system consists of a single protein oligomer (26, 31) that is independent of pho. Therefore, the P_i released after phospho compound substrate uptake by the glpT system cannot downregulate glpT gene expression, which results in continued uncontrolled accumulation of higher levels of the substrate and consequent growth inhibition.

In another scenario, the observed effect may be due to differences in the substrate specificities of the Phn and GlpT transport systems. One system may preferentially transport DIPP, whereas the other preferentially transports IPP, the periplasmic diesterase metabolite of DIPP. However, it is thought that both systems are capable of transporting IPP (and perhaps DIPP also), which is a phosphomonoester, since the natural substrates for both Phn and GlpT are phosphomonoesters (24, 26, 30, 31, 36).

The glpT and phnE gene products differs with regard to known substrate specificities in E. coli. In addition to sn-glycerol-3-phosphate, the glpT gene product can also transport 3,4-dihydroxybutyl 1-phosphonate, fosfomycin, arsenate, and P_i (30), whereas the phnE gene product is thought to be the integral membrane component of the binding protein-dependent phosphonate transporter, which can also transport phosphites, P_i esters, and P_i (6, 33, 35, 37, 49, 51). We also observed functional differences between glpT and phnE when X-P_i was used as a substrate. The glpT gene product appeared to be incapable of X-P_i transport, as demonstrated by its inability to facilitate the metabolic breakdown of X-P_i, whereas the active phnE plays an essential role in this process (13). Transport studies with labeled IPP and DIPP are needed to establish the functions of phnE and glpT gene products with regard to these two substrates.

Nevertheless, our studies suggest that glpT and phnE represent overlapping, but not redundant, systems for the transport of OPCs in E. coli.