Evidence that Bacterial ABC-Type Transporter Imports Free EDTA for Metabolism

Abstract

EDTA, a common chelating agent, is becoming a major organic pollutant in the form of metal-EDTA complexes in surface waters, partly due to its recalcitrance to biodegradation. Even an EDTA-degrading bacterium, BNC1, does not degrade stable metal-EDTA complexes. In the present study, an ABC-type transporter was identified for possible uptake of EDTA because the transporter genes and the EDTA monooxygenase gene were expressed from a single operon in BNC1. The ABC-type transporter had a periplasmic-binding protein (EppA) that should confer the substrate specificity for the transporter; therefore, EppA was produced in Escherichia coli, purified, and characterized. EppA was shown to bind free EDTA with a dissociation constant as low as 25 nM by using isothermal titration calorimetry. When unstable metal-EDTA complexes, e.g., (Mg-EDTA)²⁻, were added to the EppA solution, binding was also observed. However, experimental data and theoretical analysis supported EppA binding only of free EDTA. When stable metal-EDTA complexes, e.g., (Cu-EDTA)²⁻, were titrated into the EppA solution, no binding was observed. Since EDTA monooxygenase in the cytoplasm uses some of the stable metal-EDTA complexes as substrates, we suggest that the lack of EppA binding and EDTA uptake are responsible for the failure of BNC1 cells to degrade the stable complexes.

EDTA is a common chelating agent, and it has quietly become a major organic pollutant in surface waters, present at higher concentrations than any other organic pollutants (3, 7). In the environment, EDTA occurs as metal-EDTA complexes (27), and some heavy metal-EDTA complexes can be toxic (10, 30). In addition, EDTA has been disposed of together with nuclear wastes (7, 32), and it can mobilize radionuclides in groundwater (8, 24), spreading contamination. The concerns over EDTA recalcitrance and EDTA's potential mobilization of heavy metals and radionuclides have led the European Union and Australia to ban EDTA in detergent. The high levels of EDTA present in natural waters are due partly to its extensive usage, such as that in industrial cleaning to remove calcium deposits, in the pulp and paper industries to form complexes with metal ions for better bleaching, in detergent as a sequestering agent, in phytoremediation to mobilize heavy metals, and in scientific laboratories as a chelating agent (7, 32). Another major reason is its recalcitrance to biodegradation (8, 24).

Consistent with the relative recalcitrance, only several bacteria that degrade EDTA have been isolated (12, 21, 26, 38). The biochemistry and molecular biology of EDTA degradation have been studied with EDTA-degrading bacterial strains BNC1 and DSM 9103, which are phylogenetically grouped together and are closely related to Mesorhizobium and Agrobacterium species (25), likely forming a new branch within the Phyllobacteriaceae-“Mesorhizobia” family (36). EDTA monooxygenase (EmoA), a reduced flavin mononucleotide (FMNH₂)-utilizing monooxygenase, oxidizes several metal-EDTA complexes, such as (Mg-EDTA)²⁻ and (Cu-EDTA)²⁻, into ethylenediaminetriacetate and then into ethylenediaminediacetate in both BNC1 and DSM 9103 (5, 38). Iminodiacetate oxidase (IdaA) then oxidizes ethylenediaminediacetate into ethylenediamine (22), which is structurally similar to putrescine, a common biological diamine present in bacterial cells. Bacterium BNC1 also degrades nitrilotriacetate (NTA) with the same enzymes. EmoA converts certain metal-NTA complexes into iminodiacetate, and IdaA oxidizes the latter into glycine and glyoxylate (5, 22). The genes coding for these enzymes are organized in a gene cluster in the order of emoA, emoB, emoR (a hypothetical regulatory gene), and idaA (5, 22). The gene emoB codes for the FMN-NADH oxidoreductase that supplies FMNH₂ to EmoA. Immediately upstream of emoA are four genes, eppABCD, coding for a hypothetical ABC (ATP-binding cassette)-type transporter.

Despite the efforts in characterizing EDTA uptake by BNC1 and DSM 9103 cells (17, 37), it is still unclear whether metal-bound EDTA or free EDTA is transported into the cells for metabolism. Witschel et al. (37) reported that free EDTA and unstable metal-EDTA complexes are actively transported into DSM 9103 cells. Although the cells transport free EDTA and (Ca-EDTA)²⁻ at similar rates, the authors favored (Ca-EDTA)²⁻ as the substrate because ⁴⁵Ca²⁺ is also associated with the cells after (⁴⁵Ca-EDTA)²⁻ uptake. Separately, Kluner et al. (17) reported essentially similar findings with detectable metabolism of free EDTA and unstable metal-EDTA complexes by BNC1 cells. However, NCN1 slowly consumes (Zn-EDTA)²⁻, which is stable with a very low dissociation constant (K_d). Consequently, the authors did not conclude whether free EDTA or metal-EDTA complexes were transported into the cells.

Here, we provide several lines of evidence indicating that free EDTA, but not metal-EDTA complexes, is transported into BNC1 cells for metabolism. Primarily the periplasmic-binding protein (EppA) of the ABC-type transporter, encoded by eppABCD, bound free EDTA. The periplasmic-binding proteins usually confer the substrate specificity for ABC-type transporters (29). Thus, we suggest that stable metal-EDTA complexes are recalcitrant due to the lack of transport into BNC1 cells for metabolism.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The EDTA-degrading bacterium was kindly provided by Bernd Nörtemann (Technical University of Braunschweig, Braunschweig, Germany). BNC1 was cultured in a mineral salts medium with Na₂-EDTA (0.3 g/liter) (MMEDTA) or NH₃Cl (10 mM) (MMNH₃) as the sole nitrogen source (26). Glycerol was used as the carbon source at 4 g/liter. Escherichia coli strains BL21(DE3) and NovaBlue (Novagen, Madison, WI) containing the expression vector were grown in Luria-Bertani medium with kanamycin at 30 μg/ml.

Degradation of EDTA by BNC1 cell suspensions.

BNC1 cells were grown in the MMEDTA or the MMNH₃ to stationary phase. Cells were harvested, washed twice with 20 mM PIPES [piperazine-N,N′-bis(2-ethanesulfonic acid)] buffer (pH 7.6) or 20 mM MOPS (morpholinepropanesulfonic acid) buffer (pH 7.2), and resuspended in the same buffer to an optical density at 600 nm (OD₆₀₀) of 1. Experiments were started by adding 100 mM metal-EDTA complex or free EDTA stock to the cell suspension to a final concentration of 1 mM. Controls were cell suspensions without the addition of EDTA. The metal/EDTA ratios were 10:1 for Mg²⁺ and Ca²⁺ and 1:1 for Zn²⁺ and Cu²⁺. The EDTA stocks were prepared in the same buffers (PIPES or MOPS) and stored at room temperature in the dark. The level of ammonium produced from EDTA was determined by using an ammonium assay kit (Hach, Loveland, CO). At specific time points, 0.5 ml of the cell suspension was collected for sampling and centrifuged. A 0.4-ml sample of the supernatant solution was diluted to 0.8 ml with distilled water. Added to the sample were 0.l ml of salicylate reagent and 0.1 ml of cyanurate reagent. The mixture was incubated at room temperature for 10 min, and the OD at 655 nm was recorded. Ammonium concentrations were obtained by comparison to a standard curve (5 to 1,000 μM).

RT-PCR.

Total RNA was isolated according to a previously published method (35) from BNC1 cells grown to mid-log phase in MMEDTA or MMNH₃. The RNA was then treated with RNase-free DNase (GIBCO BRL, Gaithersburg, MD) and further purified by using the RNeasy mini kit (QIAGEN, San Diego, CA). RNA samples were screened for DNA contamination by PCR analysis. Samples that contained no target DNA were used for reverse transcription (RT)-PCR analysis. RT-PCRs were carried out by using a one-step RT-PCR kit (GIBCO BRL) in a 100-μl reaction mixture with 2 ng of RNA. Reactions were performed by using various combinations of sequence-specific primers (Table 1). All the primers were checked with genomic DNA and confirmed with the expected PCR products. The products were analyzed on 0.7% agarose gels.

TABLE 1.

PCR primers

Name	Sequence
MS6	5′-GACGACGACAAGATGGACAATTTGGTCACCGGGGAGTTG
MS7	5′-GAGGAGAAGCCCGGTTGATGACGACGAACATGAGAAAGC
RT-1	5′-ATGAATCCGTCCGCCAACTGGTAT
RT-8	5′-GCTCATAGCGATTGTCATGTGTGG
OX-1	5′-CTTCTTCACAGCGGCACACG
RT-2	5′-GCCACATCCAGTATCGGTCGAGAA
EF1	5′-CGACGACCATATGGACAATT
T7R	5′-ATGCTAGTTATTGCTCAGCG

Construction of pEppA.

To clone eppA into the pET30-LIC vector without the leader peptide-encoding region (nucleotides 1 to 78), primers MS6 and MS7 (Table 1) were designed. PCR yielded a predicted 1,739-bp product, which was treated with T4 DNA polymerase in the presence of dATP and annealed to pET30-LIC to obtain pEppA-N according to the instructions of the supplier (Novagen). pEppA-N was electroporated into E. coli NovaBlue cells for amplification, recovery, and verification by sequencing. The correct pEppA-N carried an N-terminal His tag fusion eppA gene. A plasmid carrying the nonfusion eppA gene was constructed by using primers EF1 and T7R (Table 1) with pEppA-N as the PCR template. pEppA-N contained two NdeI sites: one was part of the start codon for the fusion protein and the other was within the eppA gene (ca. 1.1 kb from the start codon). An NdeI site was introduced with primer EF1, and T7R was on the plasmid, located toward the 3′ end relative to the cloning site. The PCR product (1.9 kb) was cut with NdeI to generate a 1.0-kb fragment, which was used to replace a 1.1-kb NdeI fragment from pEppA-N. The resulting plasmid, pEppA, was confirmed by sequencing and introduced into BL21(DE3) for producing mature EppA with a methione residue in place of the leader peptide (26 amino acid residues).

Overproduction and purification of EppA proteins.

E. coli strain BL21(DE3) with pEppA was grown in 1 liter of Luria-Bertani medium at 37°C to an OD₆₀₀ of 1. Isopropyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 0.2 mM, and the culture was incubated at 30°C for four additional hours. The induced cells were harvested by centrifugation and resuspended in 20 ml of 20 mM potassium phosphate buffer. All the buffers contained the protease inhibitor phenylmethylsulfonyl fluoride at 0.5 mM. The resuspended cells were passed through a French pressure cell (model FA-030; Aminco, Urbana, IL) three times at 260 MPa. The product was centrifuged at 15,000 × g for 20 min to remove unbroken cells. The supernatant was subsequently ultracentrifuged at 183,960 × g (average) for 1 h. Solid ammonium sulfate was added to the supernatant to 70% saturation (room temperature), and the mixture was centrifuged at 10,000 × g for 15 min. The protein precipitates were collected and dissolved in 6 ml of the 20 mM potassium phosphate (pH 7) buffer. The solution was centrifuged at 15,000 × g for 15 min to remove insoluble proteins. The supernatant was then dialyzed in 1 liter of the 20 mM potassium phosphate buffer for 3 h. The dialyzed sample was injected onto an Econo-Pac High Q column (5-ml bed volume; Bio-Rad, Hercules, CA) equilibrated with the 20 mM potassium phosphate buffer. Proteins were eluted with a step and linear gradient of NaCl (percentages of 1 M NaCl in the same buffer, 0%, 10 ml; 20 to 40%, 15-ml gradient; and 100%, 10 ml) by a liquid chromatography system (Bio-Logic; Bio-Rad). EppA was eluted as a major peak around 350 mM NaCl. The protein fractions were pooled and concentrated to about 6 ml by Centriprep-10 (Millipore, Billerica, MA). The purity was checked by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (19). The purified protein was then dialyzed against 1 liter of either 10 mM MOPS (pH 7.2) or Bis-Tris (pH 6.1) buffer with 0.5 mM phenylmethylsulfonyl fluoride for 2 h. The dialysis was repeated three times. The dialyzed protein solution was stored at −80°C, and final dialysis buffer was saved in a refrigerator. The molar extinction coefficient of EppA (62,008 at 280 nm) was determined with the deduced amino acid sequence available online (http://workbench.sdsc.edu/) and used to estimate the EppA concentration.

Measurements of EppA binding by ITC.

EppA binding of EDTA and other substrates was performed with isothermal titration calorimetry by using a microcalorimetry instrument (Microcal LLC, Northampton, MA) with a 1.4-ml reaction cell. Prior to use, the reaction cell and syringe were rinsed three times with double-distilled water and then three times with the final dialysis buffer. Most measurements were done with 10 mM MOPS buffer (pH 7.2), and some were done with 10 mM Bis-Tris buffer (pH 6.1). NaCl was added to the buffer to 150 mM as specified. The dialyzed EppA was diluted with the final dialysis buffer to 10 μM and vacuumed (14 kPa) for 8 min. Briefly, the ITC experiments were performed as follows: the EppA solution was placed in the ITC cell with constant mixing by a propeller at 300 rpm, and 10 μl of the titrant (the solution that is added in a titration, e.g., 300 μM EDTA) was injected 30 times. The duration of each injection was 20 s, the spacing between injections was 400 s, and the temperature was set at 25°C. All the sample solutions were prepared by using the final dialysis buffer. Data were acquired and analyzed by using the Origin software package (provided by Microcal LLC).

Data analysis.

The calorimetric data for EppA binding of EDTA in the presence of Mg²⁺ were fitted with various binding schemes, and the best fit was obtained with the following equations:

(1)

(2)

Data fitting was conducted with the program LETAGROP (2), developed according to the published equations (28). Since the pHs of both the titrant and titrand were well controlled by the dialysis buffer, proton-related reactions were neglected in data fitting.

Aqueous speciation of EDTA.

The aqueous speciation of EDTA was calculated using the chemical equilibrium modeling system MINEQL+, version 3.01 (http://www.mineql.com/). Calculations were based on the experimental concentrations of cations, anions, and EDTA and the pH in a specific assay solution (40). The dominant species discussed for each experiment is the major species in the overall species distribution. The free EDTA mentioned in the text was, more precisely, (H-EDTA)³⁻. When Na₂H₂-EDTA was added to the solution in the absence of additional divalent cations, it was equilibrated mainly into (H-EDTA)³⁻ (∼90%) at pH 7.2, as calculated by the aqueous speciation analysis, consistent with the findings reported in the literature (28).

RESULTS

Sequence analysis of the ABC transporter genes.

The genes (eppABCD) immediately upstream of the EDTA monooxygenase gene, emoA, are predicted to code for a hypothetical ABC-type transporter in bacterium BNC1 (5) because they exhibit strong similarities to some bacterial solute ABC-type transporter genes (15). The first gene in the transport operon, eppA, encodes a 593-amino-acid protein that contains a cleavable signal sequence (1 to 26 amino acid residues), indicative of a periplasmic protein. According to conserved-domain comparison (23), EppA belongs to family 5 of the bacterial periplasmic solute-binding proteins (pfam00496.13) and is most similar to the oligopeptide-binding proteins (COG4166) (1, 29). EppA shows 27% identity and 44% similarity to the oligopeptide-binding protein (OppA) of Mycobacterium tuberculosis (1).

The next gene, eppB, encodes a putative permease protein with 315 amino acid residues. Immediately downstream is eppC, which encodes a 308-amino-acid-residue protein with the conserved EAA loop (residues 201 to 220) between two transmembrane helices (1, 15). Both EppB and EppC contain conserved domains that show the highest homology with the permease components of the ABC-type oligopeptide transport systems, and each has a predicted six-membrane-spanning α-helix. Downstream of eppC is eppD, which encodes an ATP-binding protein that is homologous to many such proteins from different organisms. EppD contains both the Walker A (residues 355 to 363) and Walker B (residues 190 to 197) motifs, which are characteristic of ATP-binding proteins, and an amino acid linker sequence that is a conserved signature of the ATP-binding domains of ABC-type transporters (15).

Further sequence analysis shows that the transporter genes are likely organized in a single operon with the emoA and emoB genes, coding for EDTA monooxygenase and NADH-FMN oxidoreductase, as the stop codon for eppD overlaps with the start codon of emoA. Since emoA and emoB are overexpressed when cells are grown on EDTA and are transcribed on the same mRNA molecules (5), it is likely that the transporter genes, emoA, and emoB are all organized in an operon.

Gene expression analysis.

RT-PCR was used to analyze the coexpression of the transporter genes (eppABCD) and the first gene in the EDTA metabolism system (emoA) in bacterium BNC1 (Fig. 1). Primers were designed to span the entire hypothetical transport gene cluster and emoA. RT-PCR analysis of the total RNA extracted from MMNH₃- and MMEDTA-grown cells showed that the gene cluster was expressed in the MMNH₃-grown cells at a significantly lower level than in the MMEDTA-grown cells (Fig. 1). The expression of the gene cluster was in agreement with results from resting-cell studies. The MMEDTA-grown cells degraded EDTA about 5.2 ± 0.5 times faster (average of results for samples ± standard deviation) than the MMNH₃-grown cells in MOPS buffer with 1 mM EDTA and 10 mM Mg²⁺, consistent with previously reported results (5). A primer located upstream of eppA (OX-1) and primer RT-2 located inside eppA did not produce any detectable products (Fig. 1). Thus, eppA was the first gene of the operon including the transporter and metabolism genes, as emoA and emoB have been shown to be cotranscribed (5). Because the transporter genes and EDTA metabolism genes are organized in a single operon, the ABC-type transporter is possibly involved in EDTA uptake.

FIG. 1.

RT-PCR analysis of the cotranscription of eppABCD and emoA. Lane 1, DNA kilobase ladder; lane 2, total RNA extracted from EDTA-grown BNC1 cells with primers RT-1 and RT-8; lane 3, total RNA extracted from MMNH₃-grown cells with primers RT-1 and RT-8; lane 4, total RNA extracted from EDTA-grown cells with primers OX-1 and RT-2.

Overproduction and purification of the periplasmic-binding protein (EppA).

The DNA fragment encoding the mature EppA (amino acid residues 27 to 567) was cloned into expression vector pET30-LIC as a nonfusion protein. When the cells from 500 ml of culture were disrupted, 114 mg of protein was obtained in the extract. Approximately 37 mg of EppA was obtained after processing through the Econo Q column. The pure EppA was dialyzed and stored at −80°C, and it was stable for several months.

EppA binding of EDTA.

The EppA binding of EDTA was studied by ITC. EppA did not bind (Cu-EDTA)²⁻, (Co-EDTA)²⁻, or (Zn-EDTA)²⁻ (data not shown) but bound free EDTA with heat generation (exothermic reaction) (Fig. 2A). Since the EDTA binding of Mg²⁺ absorbed heat (endothermic reaction) (Fig. 2C), the titration behavior depicted in Fig. 2B showed that EDTA, added to the mixture of EppA and Mg²⁺, first bound to EppA and then to Mg²⁺. An earlier study (14) has, by extensive ITC analysis, determined the K_ds of (Mg-EDTA)²⁻ and (Ca-EDTA)²⁻ to be 1,670 and 23 nM, respectively, with the corresponding standard enthalpy changes of reaction, ΔH°, of 4.4 and −6.0 kcal mol⁻¹. When we used EDTA to titrate Mg²⁺ and Ca²⁺ in MOPS buffer (pH 7.2), our experimental data were in agreement with these values. The K_ds of (Mg-EDTA)²⁻ and (Ca-EDTA)²⁻ increased to 10,200 ± 400 nM and 420 ± 30 nM (means ± standard deviations), respectively, in 10 mM Bis-Tris buffer (pH 6.1) with 150 mM NaCl.

FIG. 2.

Isothermal titration of EppA and/or Mg²⁺ by free EDTA. The sample cell contained 1.4 ml of 10 mM MOPS buffer (pH 7.2) with 10 μM EppA (A), 10 μM EppA and 10 μM Mg²⁺ (B), or 10 μM Mg²⁺ (C), titrated with 10 μl of 300 μM EDTA per injection.

More titration experiments were conducted with different combinations of EppA, EDTA, and Mg²⁺. With equations 1 and 2 and the previously determined (Mg-EDTA)²⁻ K_d (1,670 nM) and ΔH° (4.4 kcal mol⁻¹), these data were well fitted and allowed us to obtain the K_d (25 ± 6 nM) and ΔH° (−8.85 ± 0.83 kcal mol⁻¹) for EDTA-EppA. Even though the titrants were used with different Mg/EDTA ratios, the binding parameters were highly similar, indicating that the data follow the same reaction mechanism: free EDTA bound to EppA, and (Mg-EDTA)²⁻ dissociated to provide additional free EDTA. A representative set of raw data and analysis are shown in Fig. 3. The titrant contained EDTA and Mg²⁺ with EDTA in excess so that the Mg²⁺ was primarily in the form of (Mg-EDTA)²⁻. Figure 3A shows that EppA bound free EDTA and that Mg²⁺ was released from (Mg-EDTA)²⁻ during titration. When all EppA became associated with EDTA, further titration resulted in the binding of accessible free EDTA from the titrant with the accumulated Mg²⁺ in the titrand. The data were well fitted with the changes in free EppA, EDTA-EppA, Mg²⁺, and (Mg-EDTA)²⁻ concentrations according to equations 1 and 2 with the determined binding parameters. These results strongly supported one binding scheme, EppA binding of free EDTA.

FIG. 3.

Isothermal titration of EppA by EDTA and Mg²⁺. (A) Raw data; (B) data-fitting results. The sample cell contained 1.4 ml of 10 mM MOPS buffer (pH 7.2) with 10 μM EppA, titrated with 10 μl each of 300 μM EDTA and 150 μM Mg²⁺ per injection. Fitting was conducted with equations 1 and 2 and K_ds of 25 nM and 1,670 nM for EDTA-EppA and (Mg-EDTA)²⁻, respectively. Speciation curves are as follows: dashed line, free EppA; dotted line, EDTA-EppA; dash-dot-dash line, free Mg²⁺; dash-dot-dot-dash line, (Mg-EDTA)²⁻. q_v,p is normalized reaction heat per mole of protein (EppA). Squares indicate experimental data (solid squares are data points expected to be missed before data collection). The solid line shows calculated data.

The effects of ionic strength, pH, and divalent cations were evaluated (Table 2). The K_d of EDTA-EppA was 25 nM in MOPS buffer at pH 7.2. NaCl at a physiological concentration (150 mM) only slightly increased the K_d to 38 nM. The pH change from 7.2 in MOPS buffer to 6.1 in Bis-Tris further increased the K_d to 141 nM. In addition, ITC titrations of (Ca-EDTA)²⁻ to EppA in MOPS buffer were also attempted, but data fitting with the same fitting scheme used for the Mg-EDTA-EppA system (Fig. 3) failed. As a trial, the data were fitted as if EppA bound (Ca-EDTA)²⁻. The roughly resolved apparent K_ds are listed in Table 2. Interestingly, the apparent K_ds increased with the increase of Ca/EDTA ratios in titrants. This trend indicated that EppA still bound free EDTA. Apparent K_ds for the EDTA binding of (Mg-EDTA)²⁻ showed a similar trend (Table 2). The higher metal ion concentrations shifted equilibrium towards more metal-bound EDTA and less EppA-EDTA. Accordingly, it took more injections of metal-bound EDTA to titrate EppA, giving higher apparent K_ds. Since (Ca-EDTA)²⁻ is more stable than (Mg-EDTA)²⁻, Ca²⁺ had a more significant effect than Mg²⁺ on the apparent K_ds (Table 2), and the values obtained with (Mg-EDTA)²⁻ and (Ca-EDTA)²⁻ titrations were unlikely to represent true values.

TABLE 2.

EDTA titration of EppA

EDTA/metal ratio	Kd (nM) for:
	Mg	Ca
1:0	25 ± 6a	25 ± 6a
1:0	38 ± 2b	38 ± 2b
1:0	141 ± 11c	141 ± 11c
1:0.5	NMd	3,450a
1:0.7	NM	14,100a
1:1	165a	18,500a
1:10	385a	NM

^aEppA at a concentration of 15 μM in 10 mM MOPS buffer, pH 7.2, was titrated with mixture of 0.3 mM EDTA and various Mg²⁺ or Ca²⁺ concentrations.

^bTitration was done with 10 mM MOPS buffer (pH 7.2) and 150 mM NaCl.

^cTitration was done with 10 mM Bis-Tris buffer (pH 6.1) and 150 mM NaCl.

^dNM, not measured.

EDTA degradation by BNC1 cells.

Since the periplasmic-binding protein determines the substrate specificity of the ABC-type transporter (29), our EppA binding data suggest that BNC1 cells can degrade only free EDTA and unstable metal-EDTA complexes, which are able to release free EDTA for uptake and metabolism. To further confirm the binding results, the degradation of free EDTA and metal-EDTA complexes was tested with resting, EDTA-grown BNC1 cells. The cells degraded free EDTA, (Mg-EDTA)²⁻, and (Ca-EDTA)²⁻ but did not metabolize (Zn-EDTA)²⁻ and (Cu-EDTA)²⁻ in MOPS buffer (Fig. 4). BNC1 cells degraded free EDTA, but at a reduced rate in comparison to the rate of (Mg-EDTA)²⁻ or (Ca-EDTA)²⁻ degradation (Fig. 4). Because EDTA is known to disrupt the cell wall (25), the added free EDTA was likely to affect the integrity of BNC1 cells, slowing down the metabolism. The whole-cell results were consistent with the prediction that only free EDTA was transported into BNC1 cells for metabolism.

FIG. 4.

Degradation of EDTA by BNC1 cell suspensions. EDTA-grown cells were suspended in 20 mM MOPS buffer (pH 7.2). Metal-bound EDTA (1 mM) or free EDTA (1 mM) was added to start degradation, monitored as ammonium production. Curve 1, (Mg-EDTA)²⁻ degradation; curve 2, (Ca-EDTA)²⁻ degradation; curve 3, free-EDTA degradation; curve 4, (Cu-EDTA)²⁻ degradation; curve 5, (Zn-EDTA)²⁻ degradation; and curve 6, cell suspension with no added EDTA (control). Aqueous speciation showed the dominant metal-bound EDTA species were approximately 99% (Mg-EDTA)²⁻, 99% (Ca-EDTA)²⁻, 90% (H-EDTA)³⁻, 99.7% (Zn-EDTA)²⁻, and 99.9% (Cu-EDTA)²⁻, respectively. Averages of results for three samples are shown with standard deviations.

DISCUSSION

The location of the ABC-type transporter genes next to the EDTA metabolic genes in BNC1 suggests the potential role of the transporter in EDTA uptake. RT-PCR analysis showed that the transporter genes were cotranscribed with emoA, and overexpression was observed when the cells were grown on EDTA (Fig. 1). Since emoA and emoB are cotranscribed (5), all these genes were organized in one operon for EDTA uptake and metabolism. Gene inactivation would provide direct evidence on the function of a specific gene, but a genetic system to inactivate genes in BNC1 has not been worked out. Thus, a biochemical approach is used to show that the transporter is involved in EDTA uptake via binding analysis. Since periplasmic-binding proteins confer substrate specificity for ABC-type solute transporters (29), the purified EppA was used to investigate whether free EDTA or a metal-EDTA complex is transported by the ABC-type transporter into BNC1 cells for metabolism.

Our ITC data showed EppA binding of free EDTA (Fig. 2) but not (Cu-EDTA)²⁻, (Zn-EDTA)²⁻, or (Co-EDTA)²⁻. When (Mg-EDTA)²⁻ or (Ca-EDTA)²⁻ was used in the titration of EppA, ITC revealed binding with heat generation. Data analysis unequivocally showed that (Mg-EDTA)²⁻ dissociated to yield free EDTA, which then bound to EppA. Data fitting of heat changes during titration with the predicted changes in (Mg-EDTA)²⁻ dissociation and EppA binding of free EDTA gave very good fits (Fig. 3B). Although a single binding model could also be forced to fit EppA binding of (Mg-EDTA)²⁻ and (Ca-EDTA)²⁻, the obtained apparent K_ds were likely pseudo values, as they increased significantly at higher Mg²⁺ and Ca²⁺ concentrations (Table 2).

The EppA binding data and analysis suggested that the transporter system could transport only free EDTA into BNC1 cells for metabolism. This finding is in agreement with the results of a previous whole-cell study, i.e., that BCN1 cells can metabolize only free EDTA and unstable metal-EDTA complexes (16). The degradable complexes are those with K_ds of 2.0 × 10⁻³ nM or higher [(Ba-, Mg-, Ca-, and Mn-EDTA)²⁻], and nondegradable complexes have K_ds lower than 2.0 × 10⁻⁵ nM [(Pb-, Ni-, Cu-, and Fe(III)-EDTA)⁻]. The only exception is (Zn-EDTA)²⁻, with a K_d of 6.3 × 10⁻⁶ nM, but this complex is degradable at a much lower rate than unstable metal-EDTA complexes. A different whole-cell study with DSM 9103 for the uptake of metal-EDTA complexes gave essentially the same results, except (Zn-EDTA)²⁻ was not transported into DSM 9103 cells (37). The discrepancy between these studies is whether (Zn-EDTA)²⁻ is transported and metabolized by the cells. Interestingly, both studies used HEPES buffers, but at different pH values. The DSM 9103 cells did not take up (Zn-EDTA)²⁻ at pH 7.0, but the BNC1 cells did at pH 7.5. From a literature search, it is known that the protonated HEPES does not have the ability to form a complex with Zn²⁺ but that the deprotonated HEPES can (4). At pH 7.5, close to the pK_a of deprotonation, 7.55 (Merck Index), more HEPES molecules are in the deprotonated form than the protonated form, forming complexes with Zn²⁺ and releasing EDTA from (Zn-EDTA)²⁻. The released EDTA can then be degraded by the cells. This deduction is also in agreement with our data. We found that BNC1 did not degrade (Zn-EDTA)²⁻ in MOPS buffer (pH 7.2) (Fig. 4) but slowly degraded (Zn-EDTA)²⁻ in PIPES buffer (pH 7.6) (data not shown). In our case, the difference is likely due to the buffers: PIPES has strong Zn²⁺-chelating ability (4), but MOPS does not. PIPES has two amine groups and two sulfonic groups, while MOPS has only one amine group and one sulfonic group. Thus, buffer and pH can affect the dissociation of certain metal-EDTA complexes. The lack of EppA binding of (Cu-EDTA)²⁻ apparently accounts for the failure of (Cu-EDTA)²⁻ uptake and metabolism by BNC1 cells, because EDTA monooxygenase, a cytoplasmic protein, uses (Cu-EDTA)²⁻ as a substrate (14, 16).

The finding that free EDTA was transported into the BNC1 cell was somewhat surprising, as previous research on the biodegradation of chelating agents has implied that a possible metal-EDTA complex may be transported. First, EDTA monooxygenases of bacterial strains BNC1 and DSM 9103 and NTA monooxygenase of Aminobacter aminovorans (formerly Chelatobacter heintzii) metabolize only certain metal-EDTA and metal-NTA complexes but not free EDTA and NTA (5, 33, 38, 39). Second, whole-cell studies suggest that (Ca-EDTA)²⁻ is transported into DSM 9103 cells (37) and that (Ca-NTA)⁻ is transported into A. aminovorans cells (34) for metabolism. Third, citrate, a natural chelating agent, has been shown to be transported into Bacillus subtilis cells by two homologous citrate transporters specific for different metals: CitM transports metal-citrate complexes with Mg²⁺, Ni²⁺, Co²⁺, Mn²⁺, and Zn²⁺, and CitH takes up Ca²⁺-, Sr²⁺-, and Ba²⁺-citrate complexes (18). Because of these circumstantial reasons, we had expected EppA binding of some metal-EDTA complexes, but our data supported EppA binding only of free EDTA. Given that EppA is highly homologous to OppA (20), an oligopeptide-binding protein of an ABC-type transporter, we can speculate that EppA has evolved from an oligopeptide-binding protein. Oligopeptides also contain free carboxylic groups and amine groups, as EDTA does. EppA bound EDTA with a K_d of 25 nM (Table 2), and OppA is known to bind some oligopeptides with K_ds in this range (31). Most periplasmic-binding proteins are substrate specific and confer the specificity for the ABC-type transporters (29). However, OppA binds oligopeptides without sequence specificity. Further, some tightly bound oligopeptides are transported into the cells at lower rates than less tightly bound substrates (11). This observation has been attributed to the slow dissociation of the tightly bound substrate. As regards EppA, the tight binding may also lower the rate of EDTA pumping by the transporter. However, EppA has to compete with metal ions for free EDTA. Since BNC1 grows very slowly on EDTA (13), it may be more important for EppA to bind free EDTA than for the transporter to achieve a maximal uptake rate.

On the basis of the information presented, we propose a model of EDTA uptake via EppA and the corresponding transporter in bacterium BNC1 (Fig. 5). The hydrophilic metal-EDTA complexes, smaller than 500 Da, should diffuse freely through porins in the outer membranes of gram-negative bacteria (9). In the periplasm, unstable metal-EDTA complexes dissociate. Then EppA binds free EDTA and delivers it to the transporter proteins (EppBCD) on the cytoplasmic membrane for uptake. Inside the cell, EDTA binds Mg²⁺ to form (Mg-EDTA)²⁻, which is the substrate of EDTA monooxygenase (5). In the model, the description of metal-EDTA complexes as unstable is relative, depending on the presence of other complex-forming agents and the pH. As discussed above, weak complex-forming agents such as HEPES, PIPES, and also phosphate, when present at high concentrations at relatively high pHs, can bind Zn²⁺ and release free EDTA from the (Zn-EDTA)²⁻ complex. In the absence of other competing complex-forming agents, the transporter depends on EppA to compete with metal ions for EDTA binding on the basis of affinity. The reported stabilities of metal-EDTA complexes vary in the literature due to the conditions used in assays. Numbers indicating stability can be used only as references and not as absolute values. For example, the K_d of (Mg-EDTA)²⁻ ranges from 0.4 to 1.7 μM (14, 17). Further, the periplasm is usually more acidic than the cytoplasm. In E. coli, the periplasmic pH is about 1.7 units below the cytosolic pH (16), and the latter is generally between 7.5 and 7.7 (6). At pH 6.1 in Bis-Tris buffer, the K_ds of (Mg-EDTA)²⁻ and EDTA-EppA increased to 10,200 and 141 nM (Table 2), respectively. These values still allow us to draw the same conclusion drawn from the data obtained at neutral pH. Thus, our model (Fig. 5) in which only free EDTA is transported into the bacterium for metabolism is in agreement with the findings of whole-cell studies that highly stable complexes [e.g., (Cu-EDTA)²⁻] are recalcitrant (17, 37). Although the evidence presented here points to the likelihood that the ABC-type transporter is solely responsible for EDTA uptake in BNC1, gene inactivation studies are required to unequivocally resolve any uncertainties.

FIG. 5.

Proposed model of EDTA uptake by bacterium BNC1. Me²⁺, divalent metal cation.