Assimilatory detoxification of herbicides by Delftia acidovorans MC1: induction of two chlorocatechol 1,2-dioxygenases as a response to chemostress

The herbicide 2-(2,4-dichlorophenoxy)propionic acid is used in crop control. The biodegradation of chlorophenoxy acid herbicides by microbial communities was first observed several years ago (Duxbury et al., 1970 ; Evans et al., 1971 ; Pemperton & Fisher, 1977 ; Kilpi, 1980 ). Since then, various species have been isolated which are able to use at least one of these compounds as the sole source of carbon and energy (Pieper et al., 1988 ; Horvath et al., 1990 ). The enzymes and the corresponding genes involved in the biodegradation pathway have been well described in some species (Fukumori & Hausinger, 1993 ; Kaphammer et al., 1990 ). The first step of biodegradation is often catalysed by a 2-oxoglutarate dependent dioxygenase, which is encoded by the gene tfdA. The tfdBCDE genes which, like tfdA, belong to the tfd gene cluster, catalyse ortho-cleavage of the aromatic ring of the first metabolite dichlorophenol and the following dechlorination. However, this knowledge alone seems to be insufficient to explain the low conversion rates of chlorophenoxy acids observed at polluted sites. Therefore, current research is focused on determining the effects of unfavourable environmental factors on the biodegradation rate and on discovering mechanisms that stabilize the biocatalysts. One unfavourable factor could be growth on 2,4-dichlorophenoxyacetic acid (2,4-D) itself, which causes the depletion of ATP (Müller et al., 1997 ). In addition, the presence of 2,4-D induces the synthesis of the heat-shock proteins DnaK and GroEL in Burkholderia sp. YK-2 (Cho et al., 2000 ), indicating the importance of stress proteins for adaptation. These results are consistent with the established fact that detoxifying bacteria respond to chemostress by inducing stress proteins (Lupi et al., 1995 ; Uchiyama et al., 1999 ), as do many other bacteria (Blom et al., 1992 ; van Dyk et al., 1994 ). Previous studies have shown heat-shock proteins to be induced by a variety of mainly hydrophobic compounds and these chaperones may defend cells against the toxic effects of such compounds (Benndorf et al., 1999 ). An example of an active protection mechanism is the induction of the oxidative stress response in Acinetobacter calcoaceticus, which increases the organisms resistance to catechol by detoxifying reactive oxygen species released by redox cycling (Benndorf et al., 2001 ). The induction of further groups of proteins associated with responses to starvation (Blom et al., 1992 ), osmotic shock and pH shock (Vasseur et al., 1999 ) has also been observed.

Our intention here was to study mechanisms of the bacterium Delftia acidovorans MC1, which is able to utilize chlorophenoxy acids (Müller et al., 1999 ), to detoxify 2,4-dichlorophenoxypropionic acid (2,4-DCPP), 2,4-dichlorophenol (2,4-DCP) and 3,5-dichlorocatechol (3,5-DCC). For this purpose we analysed the short-term response by 2D-electrophoresis during growth on the non-toxic growth substrate pyruvate.

Bacterial strain, growth and stress conditions.
Delftia acidovorans MC1 (formerly known as Comamonas acidovorans MC1) was used for this study. It was continuously grown in a fermenter (Braun) with a working volume of 1·5 l. The growth temperature was 30 °C, the dissolved oxygen concentration ranged from 70 to 95% and the pH was maintained at 7·0 by automatic titration with 0·1 M KOH or 0·25 M H₂SO₄, as appropriate. The cultivation started with a batch culture on 3 g sodium pyruvate l^-1 as the sole source of carbon and energy in minimal medium to late exponential growth phase (Müller & Babel, 1986 ). Afterwards, the culture was grown chemostatically on 2 g sodium pyruvate l^-1 (s₀) with a dilution rate (D) of 0·2 h^-1 for at least 20 h. This culture was harvested from the fermenter and 2 g sodium pyruvate l^-1 was added. The culture was then divided into 50 ml portions that were placed in 500 ml shake flasks and exposed for 1·5 h to the following stress factors: 2,4-DCPP (Sigma; at least 95% purity), 2,4-DCP (Sigma; at least 99% purity) and 3,5-DCC (Promochem; 9599% purity) were added at a range of concentrations between 10 and 1000 µM; heat shock at 41 °C; and H₂O₂ (added in nine 1 mM batches at 10 min intervals). All stress experiments were run in duplicate and growth was measured spectrophotometrically by monitoring optical density at 700 nm.

Sample preparation and protein determination.
The bacteria were harvested by centrifugation at 4 °C for 10 min at 8000 g and washed twice with a 50 mM Tris/HCl buffer (pH 7·5) containing 0·1 mg chloramphenicol ml^-1 and 1 mM PMSF. Lysis buffer (50 µl; 50 mM Tris/HCl pH 7·5 containing 2%, w/v, SDS) was added to 50 µl of the washed cell suspensions and the cells were lysed by incubation at 60 °C for 5 min. After adding 1 ml 50 mM Tris/HCl buffer pH 7·5 containing 1 µl Benzonase ml^-1 (Merck), 0·1 mg MgCl₂ ml^-1 and 1 mM PMSF the mixture was incubated at 37 °C for 10 min. Unbroken cells and cell debris were removed by centrifugation for 10 min at 8000 g and 4 °C. The protein content was determined as described by Holtzhauer & Hahn (1988) .

2D-electrophoresis.
Cell-free extracts containing 50 µg protein for analytical separations or 2000 µg for micropreparative separations were precipitated by adding 5 vols ice-cold acetone, incubating for 20 min on ice and centrifuging at 12000 g for 10 min. The precipitated proteins were resuspended in 100 µl resolubilizing solution containing 8 M urea, 2% (v/v) Triton X-100, 2% (w/v) CHAPS, 0·5% (v/v) IPG (immobilized pH gradient) pH 310 NL Buffer (Pharmacia), 65 mM dithioerythritol and 0·01% (w/v) bromophenol blue. After agitating for 10 min, 400 µl of a rehydrating solution containing 8 M urea, 2% (v/v) Triton X-100, 0·5% (v/v) IPG pH 310 NL Buffer (Pharmacia) and 65 mM dithioerythritol was added, and the resulting solution was centrifuged at 12000 g for 30 min to remove precipitates.

Samples of the solution (450 µl) were loaded on 18 cm long Immobiline DryStrip pH 310 NL gels for equilibration overnight. After equilibration the proteins were separated by isoelectric focusing using an IPG Phore electrophoresis unit (Pharmacia). The gels were run for a total of 100 kVh for analytical separations and 240 kVh for preparative separations. After the isoelectric focusing, the gels were equilibrated as described by Benndorf et al. (1999) and the equilibrated gels were stored at -18 °C until use in second dimension electrophoresis.

For the second dimension, the gels were loaded into a layer of 0·5% (w/v) agarose in separation buffer at 80 °C on 918% (w/v) polyacrylamide gradient gels (240x200x1 mm). The gels were run overnight in a Dalt electrophoresis unit (Hoefer) at a voltage of 100 V and a temperature of 12 °C. After completion of SDS-PAGE, the gels were fixed, silver stained as described by Blum et al. (1987) and dried in a stream of unheated air from a GelAir Dryer (Bio-Rad).

Western blotting.
The gels were electroblotted overnight onto PVDF (Bio-Rad) membranes and nitrocellulose membranes for immunochemical staining. The blotting was carried out overnight in the Dalt electrophoresis unit (Hoefer) at a current of 600 mA and a temperature of 12 °C in 10 mM CAPS buffer pH 11 containing 10% (v/v) methanol (Jin & Cerletti, 1992 ). The PVDF membranes were stained with 0·1% (w/v) Coomassie blue in 50% (v/v) methanol and 10 % (v/v) acetic acid for 10 min. After destaining the membranes with several changes of 50% (v/v) methanol and 10% (v/v) acetic acid the membranes were dried and stored at -18 °C. Spots of interest were excised and subjected to N-terminal amino acid sequencing using a model 473A protein sequencer (Applied Biosystems).

The proteins on the nitrocellulose membranes were stained with 0·2% (w/v) Ponceau S in 3% (w/v) TCA and destained with water. For immunochemical stains an Immun-Blot Assay Kit (Bio-Rad) was used with a secondary goat anti-rabbit antibody linked to horseradish peroxidase. The primary antibodies, polyclonal anti-GroEL and anti-HSP 70 from rabbit, were obtained from Sigma and Upstate Biotechnology, respectively.

Comparison of 2D protein patterns.
The dried gels were scanned with a UMAX Power Look 2000 Scanner with an eight bit dynamic range and 200 d.p.i. resolution. The images were analysed and compared by Phoretix 2D 5.01 software (NonLinear Dynamics). For comparing protein patterns we used only spots that were detected in both gels of the duplicated experiments. Protein spots with a twofold, or greater, intensity than the corresponding spots in control gels were considered to be amplified, and spots which were observed following imposition of stress conditions, but were not in the control protein pattern, were considered to represent newly synthesized proteins. Molecular mass and pI were calibrated using internal standards defined by calibration with the 2D SDS-PAGE Standards kit (Bio-Rad).

Influence of 2,4-DCPP and its metabolites on growth rate
The precultivation of D. acidovorans provided us with a reproducible culture to start the stress experiments. After the transfer into batch cultures there was no lag-phase detectable and the observed growth rate of the control was 0·65 h^-1, which corresponds to approximately one doubling in 1·5 h. The presence of 2,4-DCPP, 2,4-DCP and 3,5-DCC caused a reduction in the growth rate of D. acidovorans MC1 to as little as 4% of control rates, depending on the concentration of the compounds (Table 1). 2,4-DCPP (100 µM) and 100 µM 2,4-DCP had similar effects on growth, yielding growth rates of 85% and 94% of control rates, respectively, while at 1000 µM both compounds reduced growth to 12% of control rates. Growth was reduced to half of the control rate in the presence of 100 µM 3,5-DCC and almost totally inhibited in the presence of 1000 µM 3,5-DCC.

Table 1. Influence of chemicals on growth of D. acidovorans MC1 with pyruvate

Influence of DCPP and its metabolites on protein synthesis
Treatment of D. acidovorans with 100 µM 2,4-DCPP, 2,4-DCP and 3,5-DCC during growth on pyruvate resulted in the repression of some proteins (not shown) and in the induction of 58 proteins which were amplified at least twofold or were newly synthesized (Fig. 1). Only two proteins were induced from non-detectable levels by all the compounds, but we detected 50 specific proteins that were amplified by one of these compounds. In addition, a group of four proteins were induced after treatment with 2,4-DCP and 3,5-DCC (but not 2,4-DCPP) and a further two were induced following treatment with 2,4-DCPP and 3,5-DCC. To determine whether the induced proteins included the heat-shock proteins DnaK or GroEL we assessed their levels in Western blots with polyclonal antibodies (Fig. 2). We detected both proteins (identifying two main GroEL spots with molecular masses of 59·1 and 53·0 kDa, and isoelectric points of 4·94 and 5·05, respectively, and one main spot of DnaK with a molecular mass of 70·2 kDa and an isoelectric point of 4·79), and measured induction in heat-shocked cells (up to a threefold level). However, we observed no significant induction of these proteins after treatment with the chemicals (data not shown). Moreover, levels of the oxidative stress protein alkylhydroperoxide reductase C (AhpC), which has been identified as a major component of the response of D. acidovorans MC1 to hydrogen peroxide by electrophoretic analysis and amino-terminal sequencing (Table 2), remained unchanged following treatment with 2,4-DCPP, 2,4-DCP and 3,5-DCC (not shown). Further sequencing of two proteins, one of which was generally induced by treatment with all the chemicals tested and the other of which was only induced by treatment with 2,4-DCP and 3,5-DCC, led to the identification of two chlorocatechol 1,2-dioxygenases (TfdC and TfdCII) in D. acidovorans MC1 (Table 2). Fragments of both enzymes have already been detected in this strain by DNA sequencing of PCR products (Müller et al., 2001 ). The strongest induction of both enzymes was observed in response to 100 µM 2,4-DCP and 100 µM 3,5-DCC (Fig. 3) whereas 1000 µM 2,4-DCPP was necessary to cause a small induction of both enzymes.

(137K): Fig. 1. 2D protein patterns of Delftia acidovorans MC1 after treatment with chemicals. A, control; B, 100 µM 2,4-DCPP; C, 100 µM 2,4-DCP; D, 100 µM 3,5-DCC. Circled spots, more than twofold amplified proteins; boxed spots, new proteins; spots surrounded by dotted lines, proteins induced by at least two chemicals.

(27K): Fig. 2. Detection of the heat-shock proteins DnaK and GroEL of D. acidovorans MC1 by immunostaining. A, silver stained protein pattern; B, Ponceau S stained protein pattern; C and D, immunochemical stains of protein patterns.

Table 2. N-terminal sequences of identified proteins of D. acidovorans MC1

(14K): Fig. 3. Induction ratios of the chlorocatechol 1,2-dioxygenases TfdC (a) and TfdCII (b) of D. acidovorans MC1. Black bars, 2,4-DCPP; grey bars, 2,4-DCP; white bars, 3,5-DCC.

D. acidovorans responds to 2,4-DCPP and its first metabolites at the protein synthesis level. In this paper we focused on the induced proteins and not on the repressed proteins since we expected that they are predominantly involved in adaptive responses. Two generally induced proteins may indicate a common mechanism of resistance against compounds like 2,4-DCPP, whereas the more specifically induced proteins may reflect the differences between the test compounds. The concentration of 2,4-DCPP and its metabolites which cause the strongest induction of stress proteins in D. acidovorans is 3- to 30-fold lower than the concentration of 2,4-D that causes their strongest induction in Burkholderia sp. YK-2 (Cho et al., 2000 ). The weaker induction of the chlorocatechol 1,2-dioxygenases by 2,4-DCP and 3,5-DCC at 1000 µM than at 100 µM correlates with the pronounced reduction of growth rate that the higher concentration caused. But 2,4-DCPP seems to be a less effective inducer of both enzymes although 1000 µM 2,4-DCPP inhibits the growth to the same extent. In contrast to results obtained with Burkholderia sp. YK-2 (Cho et al., 2000 ) and with an Escherichia coli strain harbouring a fusion of grpE and dnaK heat-shock promoters with lux genes (van Dyk et al., 1994 ), we did not observe an increase of heat-shock proteins when we measured the proteins GroEL and DnaK after treatment with the lipophilic 2,4-DCPP and its metabolites. This was surprising and interesting insofar as a high degree of lipophilicity has proven to be a common property of chemicals that induce heat-shock proteins, e.g. primary alcohols in A. calcoaceticus 69-V (Benndorf et al., 1999 , 2001 ). A further important property of potential stress chemicals, which may account for the induction of other groups of stress proteins, is their reactivity. For example, A. calcoaceticus 69-V synthesizes oxidative stress proteins in response to catechol (Benndorf et al., 2001 ), which detoxifies reactive oxygen species generated by redox cycling (Schweigert et al., 2001 ). It is also noteworthy that the reactive compound 3,5-DCC induced no change in the level of alkylhydroperoxidereductase subunit C, which is strongly induced by hydrogen peroxide. This protein is a gene product of the oxyR regulon of many Gram-negative bacteria, so the failure of 3,5-DCC to induce it excludes the possibility that OxyR-dependent oxidative stress proteins are involved in the response to 2,4-DCPP and its metabolites. Instead of the synthesis of well-known stress proteins, two chlorocatechol 1,2-dioxygenases (TfdC and TfdCII), both of which are involved in the metabolism of the inducing compounds, were strongly increased in response to 100 µM 2,4-DCP and 3,5-DCC but only weakly induced in response to 1000 µM 2,4-DCPP. The induction of both enzymes, within 1·5 h of treatment with these compounds during growth on pyruvate, indicates that degradation is a defence mechanism in D. acidovorans. The presence of a readily utilizable carbon and energy source may support the induction. The identification of further proteins involved in the degradation and assimilation of 2,4-DCPP would confirm our conclusion. Two further catabolic enzymes were assigned by isoelectric point and molecular mass (Fig. 1C) as TfdB (chlorophenol hydroxylase) and TfdD (chloromuconate cycloisomerase), but so far, amino terminal sequencing has been unsuccessful due to the low amounts of sample.

D. acidovorans MC1 defends against potential toxic organic compounds by inducing proteins capable of degrading these compounds. This seems to be rather the rule than an exception in micro-organisms which are resistant to hazardous organic compounds.

We would like to thank Dr P. Mak and Dr A. Dubin (BioCenter Krakow, Institute of Molecular Biology, Jagiellonian University, Krakow, Poland) for the determination of N-terminal amino acid sequences. This work has been supported by a grant from the European Commission within its Fifth Framework Programme (Project designation: HERBICBIOREM, contract no. QLK3-CT-1999-00041).