Functional Redundancy of Linuron Degradation in Microbial Communities in Agricultural Soil and Biopurification Systems

ABSTRACT

The abundance of libA, encoding a hydrolase that initiates linuron degradation in the linuron-metabolizing Variovorax sp. strain SRS16, was previously found to correlate well with linuron mineralization, but not in all tested environments. Recently, an alternative linuron hydrolase, HylA, was identified in Variovorax sp. strain WDL1, a strain that initiates linuron degradation in a linuron-mineralizing commensal bacterial consortium. The discovery of alternative linuron hydrolases poses questions about the respective contribution and competitive character of hylA- and libA-carrying bacteria as well as the role of linuron-mineralizing consortia versus single strains in linuron-exposed settings. Therefore, dynamics of hylA as well as dcaQ as a marker for downstream catabolic functions involved in linuron mineralization, in response to linuron treatment in agricultural soil and on-farm biopurification systems (BPS), were compared with previously reported libA dynamics. The results suggest that (i) organisms containing either libA or hylA contribute simultaneously to linuron biodegradation in the same environment, albeit to various extents, (ii) environmental linuron mineralization depends on multispecies bacterial food webs, and (iii) initiation of linuron mineralization can be governed by currently unidentified enzymes.

IMPORTANCE A limited set of different isofunctional catabolic gene functions is known for the bacterial degradation of the phenylurea herbicide linuron, but the role of this redundancy in linuron degradation in environmental settings is not known. In this study, the simultaneous involvement of bacteria carrying one of two isofunctional linuron hydrolysis genes in the degradation of linuron was shown in agricultural soil and on-farm biopurification systems, as was the involvement of other bacterial populations that mineralize the downstream metabolites of linuron hydrolysis. This study illustrates the importance of the synergistic metabolism of pesticides in environmental settings.

INTRODUCTION

Linuron [3-(3,4-dichlorophenyl)-1-methoxy-1-methyl urea] is a phenylurea herbicide that is widely used in various agriculture crops and in orchards (1) but also forms a contaminant in soil, groundwater, and surface water (1, 2). Biodegradation is an important mechanism for removal of linuron in the environment. In agricultural soils that were treated with the compound on a long-term basis, microbial communities respond to linuron application and specific bacterial populations that use linuron as a sole source of carbon, nitrogen, and energy proliferate (2).

The major bacterial pathway for mineralization of linuron in soil is initiated with the hydrolysis of linuron into 3,4-dicholoroaniline (3,4-DCA) and N,O-dimethylhydroxylamine (N,O-DMHA), which are subsequently mineralized (2). Various bacterial isolates and consortia that mineralize linuron were reported. In both cases, members of the genus Variovorax are essential. Most linuron-mineralizing single strains are Variovorax species, and in consortia, the organism performs at least the initial hydrolysis step (3–6). Recently, gene functions linked with linuron mineralization in Variovorax sp. were identified. The libA gene was identified in Variovorax sp. strain SRS16 and encodes the linuron amidase LibA, which hydrolyzes linuron into 3,4-DCA and N,O-DMHA (7). Until recently, libA was the only linuron hydrolysis gene linked with linuron mineralization. In several environments such as agricultural soil and on-farm biopurification systems (BPS) used for the treatment of agricultural pesticide-contaminated wastewater, the abundance of libA increased in parallel with increasing linuron mineralization capacity as a response to linuron application, indicating a prominent role for libA in linuron mineralization in environmental settings (8, 9). However, in addition to LibA, other linuron hydrolases were proposed to contribute to linuron mineralization in Variovorax and in the environment, as some linuron-degrading Variovorax isolates lack a libA homologue and libA did not proliferate in all environments that developed linuron mineralization activity upon exposure to linuron (7). Recently, we identified HylA as a second type of linuron hydrolase in Variovorax sp. strain WDL1 (10). Interestingly, HylA is evolutionarily unrelated to LibA and shows different enzymatic kinetic properties (10). The linuron hydrolysis genes in both WDL1 and SRS16 are combined with highly similar catabolic gene modules encoding the downstream pathway for 3,4-DCA degradation. Apparently, the expansion of a 3,4-DCA catabolic pathway toward linuron degradation in the two strains involved divergent evolution and the independent acquisition of nonrelated isofunctional linuron hydrolytic gene functions by horizontal gene transfer (10). Moreover, in contrast to SRS16, WDL1 is a member of a commensal bacterial consortium in which most of the 3,4-DCA produced from HylA-dependent linuron hydrolysis is channeled to 3,4-DCA catabolic strains (4).

The identification of HylA as an alternative linuron hydrolase in linuron-degrading Variovorax strains poses questions about its contribution to environmental linuron degradation in agricultural environments in addition to that of LibA, i.e., about the functional redundancy of environmental linuron biodegradation. Catabolic enzymes for pesticide degradation are considered to be quite unique, as only a few enzyme types that perform the corresponding activity exist, making their genes excellent molecular markers for assessing the corresponding biodegradation capacity and activity. Several authors (8, 11, 12) reported on the use of catabolic molecular markers to assess herbicide biodegradation in soil. However, isofunctional diversity of enzymes in pesticide biodegradation and the role of their functional redundancy remain largely unexplored and are to date reported only for biodegradation of phenoxy acid herbicides (11, 13–15).

In this study, we examined whether hylA-hosting organisms can successfully compete with libA-hosting organisms in linuron-exposed environments and whether hylA gene copy numbers explain increases in linuron mineralization capacities in linuron-exposed environments where libA failed to do so. Moreover, we examined whether commensalism in linuron metabolism as observed for hylA-carrying WDL1 and hence multispecies food webs are involved in linuron biodegradation in the environment. For those purposes, gene copy numbers of hylA as well as dcaQ were determined in available DNA extracts from two different ecosystems for which the responses of the resident Variovorax community, of the libA gene copy number, and of the intrinsic capacity to mineralize linuron to the application of linuron were studied previously (9, 16, 17). The gene dcaQ encodes the glutamine-aminotransferase-like component of the multicomponent enzyme 3,4-dioxygenase converting 3,4-DCA into chlorocatechol and functions as a marker for linuron biodegradation beyond 3,4-DCA (18, 19). The first ecosystem is an agricultural soil with a history of linuron treatment and was studied as a field experiment followed by a lab-scale soil microcosm (SM) experiment that was initiated to verify results of the field experiment in a controlled way (9). The second ecosystem mimicked the environment of an on-farm BPS. Microcosms (BMs) containing BPS material were treated with linuron (7, 16), and different stress events were imposed to study their effects on linuron mineralization (20, 21).

MATERIALS AND METHODS

qPCR.

Primer3web and Primer-BLAST were used to design primers targeting hylA, encoding the linuron hydrolase HylA, and dcaQ, encoding the glutamine-aminotransferase-like component of the multicomponent enzyme 3,4-dioxygenase. Primers for hylA were designed based on the sequences of the corresponding gene in the genomes of the linuron-degrading Variovorax sp. strains WDL1 and PBS-H4. Primers for dcaQ were designed based on the corresponding gene sequences identified in Variovorax strains SRS16, WDL1, and PBS-H4 and in a range of chloroaniline-degrading bacteria, including the 3,4-DCA-degrading Comamonas testosteroni strain WDL7, which is part of the linuron-mineralizing commensal bacterial consortium that includes WDL1 (Table 1). Phylogenetic analysis of dcaQ genes in chloroaniline-degrading bacteria showed the existence of two main groups that show around 80% identity at the nucleotide level and 79% at the amino acid level. The groups were designated dcaQ_I and dcaQ_II, and primer sets that allowed discrimination between the two groups were designed. Specificities of the primer sets were assessed in silico with Primer-BLAST and a BLAST search and performance of PCR and quantitative real-time PCR (qPCR) on genomic DNA from several linuron-/3,4-DCA-degrading and non-linuron-/3,4-DCA-degrading Variovorax strains and bacteria of related and nonrelated genera. qPCR was performed in a Rotor Gene real-time centrifugal DNA amplification apparatus (Corbett Research, Australia). The real-time PCR mixtures contained 7.5 μl of Absolute QPCR SYBR Green mix (Thermo Fisher Scientific, United Kingdom), 0.30 μl of forward primer (200 nM), 0.30 μl of reverse primer (200 nM), 3.90 μl of nuclease-free water, and 3 μl of 10-fold-diluted template DNA. The exception was the hylA PCR mixture, which used a 100 nM solution of reverse primer HylA-RT-F instead of a 200 nM solution. Reaction conditions were 15 min at 95°C, followed by 40 cycles of 15 s at 94°C, 15 s at 60°C, and 15 s at 72°C. Standard curves for qPCR were compiled using 10-fold serial dilutions of amplicons (ranging from 1 copy/μl up to 10⁸ copies/μl) of appropriate gene fragments generated by conventional PCR from genomic DNA of strains SRS16 (dcaQ_I) and WDL1 (for hylA and dcaQ_II) as reported below. The fragments were purified from agarose gels using the QIAquick gel extraction kit (Qiagen). DNA concentrations of the purified DNA fragments were determined with the NanoDrop 1000 spectrophotometer (Thermo Scientific). The limit of detection for all genes by qPCR was 1.2 × 10³ copies g (dry weight) of soil/BM material⁻¹. For each DNA extract, qPCRs were performed in duplicate. Gene abundances are expressed either as the gene copy number of hylA/dcaQ_I/dcaQ_II per copy of bacterial 16S rRNA gene or as the percentage of the bacterial 16S rRNA gene copy number. The values of bacterial 16S rRNA gene copy numbers used originated from the work of Bers et al. (9, 16, 17). hylA, dcaQ_I, and dcaQ_II gene copy numbers were determined on the same DNA extracts that were previously used to determine libA and 16S rRNA gene copy numbers (9, 16, 17).

TABLE 1.

Primer pairs used for either regular PCR or real-time PCR targeting linuron-specific catabolic genes

Target	Primera	Primer sequence	Amplicon size (bp)	GenBank sequence accession no.
hylA	HylA-F	AGGTCATGTCCACTCGCGTCT	1,905	KC146403; KC146406
	HylA-R	GCCGATGCATAGGGCCATATTTGCT
	HylA-RT-F	GCATGGGTCTGTTGCTGATAC	90
	HylA-RT-R	CTGCGTGGAACTTCACTGTTAG
dcaQI	dca I-F	CTCTCATGGCCGGATCAATA	272	JN104632.1
	dca I-R	TACAGATCGGCCAGCATCCA
	dca I-RT F	AAGGGATTGAACACGAAGGC	137
	dca I-RT R	TGGCCGGATCAATATGGTCTG
dcaQII	dca II-F	CGCCCACTGGTCATGTAAAG	377	KC146405.1
	dca II-R	GAAAAGCACGGCATCTGGTC
	dca II-RT F	GCCAAGACAACCGAACCATC	80
	dca II-RT R	GGATACCCAGAAAGCCGCA

^aPrimers including RT in their designation were used for qPCR. The others were used for conventional PCR. F, forward; R, reverse.

Conventional PCR.

Conventional PCR targeting hylA, dcaQ_I, and dcaQ_II was performed using the primer sets reported in Table 1. PCR mixtures contained 5 μl DreamTaq Green buffer (10 ×) (Life Technologies), 5 μl 1% bovine serum albumin (BSA), 4 μl 2.5 mM deoxynucleoside triphosphates (dNTPs), 0.25 μl 0.1 mM forward and reverse primers, and 0.25 μl DreamTaq polymerase (5 units/μl) (Life Technologies) adjusted to a total volume of 50 μl with nuclease-free water. PCRs were performed in a Biometra Thermocycler (AnalytikJena), and reaction conditions were 15 min at 95°C, followed by 30 cycles of 1 min at 94°C, 1 min at 60°C, and 1 min at 72°C and a final elongation step at 72°C for 10 min. Amplicons were visualized by agarose gel electrophoresis (1% agarose, 75 min, 90 V) using GelRed (Biotium) as a nucleic acid stain.

Agricultural soil DNA extracts.

The agricultural soil DNA extracts used originated from a field experiment and a concomitant SM experiment which assessed the responses of the Variovorax community, libA gene copy number, and linuron mineralization potential to linuron application. A detailed description of those experiments is found in the work of Bers et al. (9). Briefly, two adjacent potato field plots either were not treated with any herbicide (plot P0) or, at day 0, were treated with 450 g linuron ha⁻¹ (plot PF). Topsoil samples were taken from each plot at three different positions (marked as A, B, and C in plot PF and D, E, and F in plot P0) at day 0, before the pesticide application, and at days 20 and 34 after the treatment. At each position, three soil samples were taken in a radius of 0.2 m. Soil samples were homogenized and used for DNA extraction for molecular analysis, dry weight measurement, and [¹⁴C]linuron mineralization assays in triplicate (9). After 176 days of pesticide treatment, soil samples, taken from all six positions in plots P0 and PF, were mixed and six SMs consisting of glass columns (height, 10 cm; diameter, 4 cm) filled with the soil mixture were set up. All SMs were incubated at 25°C. Three SMs were irrigated with tap water with linuron, and three were irrigated with linuron-free tap water (9) according to the scheme shown in Fig. 1 (9). For some SM replicates that originally received linuron, linuron application was intermittently stopped as pictured in Fig. 1. At selected time points, soil samples were taken for DNA extraction as reported previously (9). Other DNA extracts used were those recovered previously from linuron-mineralizing liquid enrichment cultures that were started from samples of linuron-fed soil microcosms (L SM A, L SM B, and L SM C) as described previously (9).

FIG 1.

Treatment schemes used in the microcosm experiments. In the agricultural soil microcosm (SM) experiment (top), as described by Bers et al. (9), SMs A, B, and C were treated with linuron with an intermittent period of no linuron application for SM A and SM B. SMs D, E, and F were never treated with linuron. A drought period without linuron application was imposed as indicated. In the BPS microcosm (BM) experiment (bottom) as described by Bers et al. (21) and Sniegowski et al. (20), the BPS matrix contained either linuron-primed soil L or nonprimed soil C. The BMs were treated either with linuron (L+/C+) or with water (L−/C−). Linuron treatment was stopped between week 12 and week 22 for BMs L+ and C+, and an intermittent drought and cold period without linuron application was imposed for all BMs.

DNAs from biopurification systems.

The BPS DNA extracts used originated from a BM experiment previously described by Sniegowski et al. (16, 20). A detailed description of the experiment is found in the work of Sniegowski et al. (16, 20). Briefly, the experiment made use of BMs set up in glass columns (height, 10 cm; diameter, 4 cm) filled with a mixture of 25% (vol/vol) cut straw, 25% (vol/vol) peat, and 50% (vol/vol) soil. The soil was either a non-linuron-primed soil, C (BM type C), or a linuron-primed soil, L (BM type L). Soil L originated from the agricultural field studied in the field experiment described above but was taken at another location within the field and 2 years before the field experiment. Soil C was a subsurface soil obtained from a construction site. All setups were performed in triplicate. The setups were subjected to different treatments and periods of stress (drought and freezing) as shown in Fig. 1 (16, 20). Samples were taken as outlined in Fig. 1.

Data analysis.

libA gene copy numbers, 16S rRNA gene copy numbers, and linuron mineralization capacity data used in this study were taken from previous reports (9, 16, 20, 21). The lag times determined in the linuron mineralization kinetics recorded in linuron mineralization assays using samples taken from the field or SM and BM experiments as inoculum were used as a measure for the linuron mineralization capacity, i.e., the shorter the lag phase, the higher the mineralization capacity (9, 16). The results were subjected to Student's t test on a significance level of 0.05 to assess differences between gene copy numbers and to assess effects of linuron application and perturbations.

RESULTS

hylA and dcaQ dynamics in agricultural soil.

The dynamics of hylA gene abundances were similar to those previously reported for libA, but throughout the experiment, hylA gene copy numbers were at least 2 log₁₀ units higher than those of libA in both the linuron-treated and nontreated plots (Fig. 2). At day 20, as was the case for libA, hylA abundances had significantly increased (40- to 80-fold) at two of the three sampling positions in the linuron-treated plot PF compared to day 0 and were significantly higher than those in the nontreated plot P0 (Fig. 2). Similar to libA gene copy numbers at day 34, hylA gene copy numbers had decreased again to an abundance similar to this at day 0. Overall, gene copy numbers of dcaQ_I and dcaQ_II followed the same dynamics as libA and hylA, with the highest number at day 20 in the linuron-treated plots. However, dcaQ_I and dcaQ_II abundances were never in the same order as those of libA and hylA. In particular, dcaQ_II showed high copy numbers that were 10 times higher than those of hylA and up to 1,000 times higher than those of libA at day 20 in the linuron-treated plots. We conclude that, as for libA and hylA, dcaQ_I and dcaQ_II abundances were highest when the mineralization capacity in the linuron-treated plots was highest compared to the nontreated plots, i.e., at day 20. At that time point, the lag phase of linuron mineralization as a measure for the mineralization capacity was reduced to approximately 5 days in the treated plot compared to 8 days in the nontreated plot. Apparently, to obtain this reduction, the gene copy numbers of libA, hylA, dcaQ_I, and dcaQ_II have to reach (1.7 ± 0.4) × 10⁻⁴%, (8.3 ± 31) × 10⁻²%, (1.8 ± 40) × 10⁻³%, and (4.7 ± 17.3) × 10⁻¹% of the total number of 16S rRNA gene copies, respectively.

FIG 2.

Abundances of libA, hylA, dcaQ_I, and dcaQ_II in samples taken at days 0, 20, and 34 from positions A, B, and C in the linuron-treated field plot PF and positions D, E, and F in the nontreated field plot P0. Reported values are the log₁₀ values of the average hylA gene copy numbers expressed as a percentage of the bacterial 16S rRNA gene copy number with standard deviation (n = 6; 3 soil samples per position, 2 qPCRs per sample) (approximately 10⁹ bacterial 16S rRNA gene copies/g soil). Student's t test was used to determine significant differences in treatment and time points (P < 0.05). An asterisk above a bar marks a significant difference in gene abundance at a specific position (n = 6) in a linuron-treated plot compared to the average abundance of that gene determined at positions D, E, and F of the nontreated field plot P0 at the same time point. To indicate whether gene abundance is significantly different between different time points at a specific position, “a,” “b,” and “c” are used as markers above the bars of each position. “a” marks a significant increase between day 0 and day 20, “b” marks a significant decrease between day 20 and day 34, and “c” marks a significant increase between day 0 and day 34. Lag time (bottom) as a measure for the linuron mineralization capacity is shown for each position of plots PF and P0 on days 0, 20, and 34. Values are averages (n = 3, 3 soil samples per position) with standard deviations shown as error bars. Values for libA abundances and mineralization capacity (lag time) were taken from the work of Bers et al. (9).

In the controlled SM experiment, hylA gene copy numbers were initially (before feeding with linuron) relatively high and, as in the field, exceeded libA gene copy numbers by a factor of 100 to 1,000 despite the low linuron mineralization capacity (Table 2). These initial high hylA abundances were maintained only in microcosms treated with linuron (Table 2), implying that the maintenance of the hylA-containing population(s) depended on linuron application. At day 149, hylA gene copy numbers, however, also increased in the control microcosms, which can be attributed to the drought period between day 78 and day 110. A similar observation was previously done for libA (Table 2) (9). In the case of linuron application, hylA abundance increased further to 2.82% ± 0.72% of the bacterial 16S rRNA gene copies at day 491. The dependency of hylA gene copy numbers on linuron was further apparent from the drops in hylA abundance in replicate L SM C and replicate L SM B after stopping linuron application from day 159 until day 285 and from day 285 until day 491, respectively (Table 2). Compared to hylA, both dcaQ_I and dcaQ_II abundances were a factor of 1,000 and 10 less at the start of the SM experiment, respectively (Fig. 3 and Table 2). dcaQ gene copy numbers and especially those of dcaQ_II responded positively on linuron application but never coincided with hylA or libA abundances. As with libA and hylA, both dcaQ_I and dcaQ_II increased in abundance independently of linuron application after the drought period between days 78 and 110. Stopping linuron application in microcosm L SM B affected dcaQ_II abundance (Fig. 3 and Table 2). Overall, it can be concluded that gene copy numbers of most catabolic markers (except dcaQ_I) were highest when the mineralization capacity was highest. Lag times as low as 0.7 days were reached in linuron-fed SMs which corresponded with libA, hylA, dcaQ_I, and dcaQ_II abundances of, respectively, (5.4 ± 1.4) × 10⁻¹%, (1.5 ± 0.6) × 10⁻¹%, (8.2 ± 3.4) × 10⁻³%, and (5.1 ± 2.2) × 10⁰% of the total number of 16S rRNA gene copies. In addition, as previously found for libA, a correlation was found between dcaQ_II gene copy numbers and lag time (representing the mineralization capacity) (Fig. 4). This correlation was less obvious for hylA, particularly due to the gene copy numbers recorded at day 0 and for microcosm L SM B at day 491, where hylA gene copy numbers were fairly high but mineralization capacity was relatively low (Fig. 4).

TABLE 2.

Percentages of libA and hylA gene copies within the bacterial soil community of water-treated SMs and linuron-fed SMs as determined by hylA-specific, libA-specific, and dcaQ-specific qPCR and recorded linuron mineralization lag times as determined by [¹⁴C]linuron mineralization assays,^fg

Sampling day	% copies of libA or hylA gene/bacteria for SMb				Linuron mineralization lag time (days)c		% copies of dcaQ gene/bacteria for SMb
	libA		hylA				dcaQI		dcaQII
	W SM	L SMd	W SM	L SMd	W SM	L SM	W SM	L SMd	W SM	L SMd
0	(3.0 ± 1.5) × 10−4	(3.2 ± 1.8) × 10−4a	(1.4 ± 0.2) × 10−1a	(9.3 ± 6.5) × 10−2a	6.9 ± 0.3	7.0 ± 0.4	(8.2 ± 3.4) × 10−4a	(3.2 ± 2.4) × 10−4a	(6.1 ± 3.2) × 10−2a	(1.2 ± 0.6) × 10−2a
28	(3.0 ± 2.0) × 10−4a	(7.9 ± 2.3) × 10−3ae	(6.9 ± 2.8) × 10−4ae	(1.9 ± 0.3) × 10−1ae	6.9 ± 0.3	3.1 ± 0.1	(1.8 ± 1.0) × 10−4a	(1.7 ± 0.6) × 10−2a	(2.6 ± 1.3) × 10−3a	(1.1 ± 0.4) × 100a
110	(2.8 ± 1.1) × 10−4	(8.7 ± 4.1) × 10−3a	(3.5 ± 1.8) × 10−4e	(2.1 ± 1.0) × 10−1a	ND	ND	(4.1 ± 3.3) × 10−4a	(1.4 ± 0.6) × 10−2a	(4.4 ± 1.4) × 10−3a	(3.0 ± 1.6) × 100a
149	(1.9 ± 0.9) × 10−2ae	(1.6 ± 4.4) × 10−1e	(3.2 ± 1.6) × 10−2ae	(3.6 ± 1.1) × 10−1	ND	ND	(6.3 ± 3.3) × 10−3ae	(1.7 ± 1.0) × 10−2a	(1.2 ± 0.7) × 10−1a	(1.9 ± 0.8) × 100a
245
A	(6.8 ± 3.8) × 10−3ae	(5.4 ± 1.4) × 10−1e	(2.5 ± 0.4) × 10−3ae	(1.5 ± 0.6) × 10−1ae	3.3 ± 0.8	0.7 ± 0.1	(8.2 ± 3.4) × 10−4a	(4.2 ± 2.0) × 10−3a	(5.1 ± 2.2) × 10−2a	(3.2 ± 1.8) × 100a
B		(8.0 ± 0.4) × 10−1ae		(2.0 ± 0.6) × 10−1a		0.7 ± 0.1		(5.7 ± 2.0) × 10−3a		(3.8 ± 1.6) × 100a
C		(4.4 ± 1.0) × 10−2a		(2.3 ± 0.3) × 10−2ae		1.9 ± 0.5		(1.1 ± 1.2) × 10−3a		(3.3 ± 3.4) × 100a
491
A	BDL	(1.6 ± 0.1) × 10−1ae	(9.5 ± 2.9) × 10−3e	(2.8 ± 0.7) × 10−0ae	7.9 ± 0.1	1.1 ± 0.2	(7.3 ± 1.6) × 10−4a	(8.5 ± 4.7) × 10−4a	(1.1 ± 0.4) × 10−2a	(5.9 ± 2.0) × 100a
B		(8.3 ± 0.3) × 10−3ae		(2.3 ± 0.1) × 10−1a		6.2 ± 0.3		(7.3 ± 1.7) × 10−4a		(3.9 ± 0.1) × 10−1a
C		(2.1 ± 0.1) × 10−1e		ND		0.5 ± 0.2		ND		ND

^aSignificant (P < 0.05) differences between the abundance of libA and hylA and that of dcaQ_I and dcaQ_II in a setup at a specific sampling day.

^bAverage values (Avg) with standard deviation (SD) determined for triplicate SMs, all three measured in duplicate (n = 6). Notation is as follows: Avg ± SD × 10^Y with an exponent Y referring to both the average value and the standard deviation.

^cAverage values with standard deviation determined for duplicate samples of each SM.

^dAt day 245 and day 491, the results for microcosms L SM A, L SM B, and L SM C are shown separately, with SM replicates L SM A, L SM B, and L SM C marked as A, B, and C, respectively, in the first column together with the sampling day. Linuron supply was stopped after day 159 and resumed at day 285 for L SM C. After day 245, linuron supply was stopped for L SM B.

^eSignificant (P < 0.05) differences in the abundance of genes on a sampling day compared to the preceding one within a specific setup.

^fAbbreviations: W SMs, water-treated SMs; L SMs, SMs fed with linuron until day 245, when feeding conditions were changed as described elsewhere in the footnotes; ND, not determined; BDL, below detection limit.

^glibA gene copy numbers and values of linuron mineralization lag times originated from the work of Bers et al. (9).

FIG 3.

Abundances of catabolic genes encoding linuron mineralization in the water-treated (left) and linuron-treated (right) SM setups of the agricultural soil experiment. Linuron hydrolase genes libA (●) and hylA (○) and genes dcaQ_I (▲) and dcaQ_II (△) are expressed as percentages of the bacterial 16S rRNA gene copy number (approximately 10⁹ bacterial 16S rRNA gene copies/g soil). Reported values are average values with the standard deviation (lag time, n = 3 [3 replicates]; qPCR, n = 6 [3 replicate SMs, 2 qPCRs]) indicated by the error bars. Values for libA abundances were taken from the work of Bers et al. (9).

FIG 4.

Correlation between lag time of linuron mineralization and abundance of the linuron hydrolase (left) (libA and hylA) and dcaQ (right) (dcaQ_I and dcaQ_II) genes as log₁₀ value of the percentage of the bacterial 16S rRNA gene copy number for the water-fed (open symbols) and linuron-fed (solid symbols) SMs containing agricultural soil. Reported values are average values with the standard deviation (lag time, n = 3 [3 replicates]; qPCR, n = 6 [3 replicate SMs, 2 qPCRs]) shown by the error bars. Linear regressions are shown as dashed lines. R² values are shown. Linuron mineralization lag times and libA gene copy numbers were taken from the work of Bers et al. (9).

In the work of Bers et al. (9), liquid enrichment cultures using linuron as the sole source of carbon and energy were obtained from soil samples taken from plot PF 20 days after application of linuron and from soil samples taken from the linuron-fed SMs on day 134 of incubation. In these enrichment cultures, libA abundance fell under the detection limit after a few transfers, although Variovorax was still present (9). Conventional PCR targeting hylA showed that hylA instead of libA became the dominant linuron hydrolysis gene in those cultures upon prolonged enrichment (see Fig. S1 in the supplemental material).

hylA and dcaQ dynamics in biopurification systems.

In BMs containing primed soil L, overall hylA behaved similarly to libA (Fig. 5, top), although hylA abundances were generally higher than those of libA (2 to 10 times higher). Similarly to libA, hylA clearly responded to the linuron feed with increasing gene copy numbers in the treated microcosms and a decrease of gene copy numbers when linuron feeding stopped. A cold period without linuron application did not affect hylA gene copy numbers. While the drought period without linuron application affected the abundance of neither libA nor hylA, consequent rewetting and resuming the feed with linuron after the drought period resulted in a dramatic increase of hylA gene copy numbers to (2.3 ± 0.3) × 10⁻⁴% of the 16S rRNA gene copy number. In the non-linuron-treated BMs containing primed soil, drought-wetting events affected both libA and hylA gene copy numbers positively. The abundance of dcaQ_I followed an increasing trend similar to that of libA in the linuron-treated BMs, although the initial increase between week 0 and week 2 was 10-fold higher than for libA. dcaQ_II gene copy numbers remained similar to those of hylA and followed a similar trend but diverged toward the end to a 10-fold difference (dcaQ_II > hylA) where dcaQ_II became the dominant dcaQ variant. In the BMs containing primed soil, the highest linuron mineralization capacity (lowest lag phase, 0.8 ± 0.1 days) was reached in the linuron-treated BMs at week 55, corresponding to the highest percentage of dcaQ_II ([2.9 ± 0.3] × 10⁻²%), dcaQ_I ([1.4 ± 0.3] × 10⁻³%), hylA ([2.3 ± 0.2] × 10⁻³%), and libA ([5.0 ± 2.7] × 10⁻⁵%) of the 16S rRNA gene copy number.

FIG 5.

Dynamics of libA, hylA, dcaQ_I, and dcaQ_II abundances and of the linuron mineralization lag phase (as a measure for linuron mineralization capacity) in BMs inoculated with linuron-primed soil (top) (approximately 10⁹ to 10¹⁰ bacterial 16S rRNA gene copies/g soil) and non-linuron-primed soil (bottom) (approximately 10⁸ to 10⁹ bacterial 16S rRNA gene copies/g soil) treated either with water (left) or with linuron (right). Abundances of the hydrolase genes (○) (libA [solid line]; hylA [dashed line]) and dcaQ genes (△) (dcaQ_I [solid line]; dcaQ_II [dashed line]) are expressed as log₁₀ values of the average ratio of the respective gene copy number to the bacterial 16S rRNA gene copy number (n = 6, 3 replicates; 2 qPCRs for each replicate) with the standard deviations shown by the error bars. Values of libA for the linuron-treated BM containing nonprimed soil are from only one replicate BM (BM3) since libA was not detected for the other two BMs. Values of libA abundance were taken from the work of Bers et al. (21). Values below the x axis are below the detection limit. Values of the lag time (days) recorded in linuron mineralization assays as a measure for the linuron degradation capacity were taken from the work of Bers et al. (21). Gray bars are average values with standard deviations (n = 3; 3 replicate microcosms) for the water- and the linuron-treated BM setups.

In BMs containing nonprimed soil C, libA was previously recorded in gene copy numbers above the detection limit in only one replicate BM, i.e., BM3 of setup C⁺, although similar high linuron mineralization capacities developed in the other two replicates (9). In contrast to libA, hylA was detected at most time points except when the mineralization capacity was very low (for instance, week 17). However, even when the mineralization capacity was relatively high (short lag time), for instance, at week 55, hylA abundances were extremely low (down to [1.4 ± 1.6] × 10⁻⁷%), even in replicates where libA gene copy numbers were below the detection limit (Fig. 5, bottom). Despite the low abundances of libA or hylA gene copies and the low linuron mineralization capacity, both dcaQ genes increased in the first 17 weeks of linuron treatment to reach and remain at a high abundance until the end of the linuron treatment. Nevertheless, the linuron mineralization lag time was at a minimum when the copy numbers of dcaQ genes were at a maximum, i.e., at week 55. No linuron mineralization occurred in the samples taken from water-treated BMs with nonprimed soil. libA was not detected in any of these samples, while hylA and both dcaQ genes were detected at low copy numbers (approximately 10⁻⁷% of the 16S rRNA gene copy number). We conclude that in the BMs containing nonprimed soil, the highest linuron mineralization capacity (lowest lag time, 1.3 ± 0.3 days) was reached in the linuron-treated BMs at day 55 corresponding to the highest percentage of dcaQ_II ([1.06 ± 1.06] × 10⁻⁴%), dcaQ_I ([5.6 ± 1.2] × 10⁻⁴%) of the 16S rRNA gene copy number but not of hylA ([1.4 ± 1.6] × 10⁻⁷%), and libA (3.3 × 10⁻⁵% in only one replicate, BM3).

DISCUSSION

libA, hylA, and dcaQ dynamics in agricultural soil.

Our results show that hylA gene copy numbers, like libA gene copy numbers, clearly depended on and responded to the application of linuron in agricultural soil with a history of linuron application. Both hylA- and libA-containing bacteria were previously isolated from that soil and, hence, known to be endogenous to the studied field (22). We conclude that as a response to linuron application, both hylA- and libA-carrying bacteria grow in the soil, indicating that bacteria carrying hylA compete successfully with those carrying libA in the same soil. Neither of the two is really outcompeted, indicating that the two benefit simultaneously from the applied linuron to grow and/or maintain their population size. It also indicates that hylA-carrying hosts, in addition to libA-carrying hosts, contribute to linuron degradation in the soil. This is in contrast to studies that follow the dynamics of alternative tfdA genes that encode the enzyme that initiates the metabolism of phenoxy acid herbicides in soils as a response to treatment with different types of phenoxy acid herbicides. It was found that depending on the phenoxy acid substrate, specific tfdA gene groups became dominant, despite the initial presence of multiple groups (11, 13, 23), implying that in the case of phenoxy acid herbicides the initial assessment of functional genes in soils does not necessarily reflect the organisms or genes that proliferate and perform the degradation of the compounds in question, which is the case in our study.

The high initial abundance of hylA in the agricultural soil at the start of the field experiment might be a result of growth of hylA-containing bacteria on residual linuron from previous applications in the field and a higher persistence of those populations than of the libA-carrying populations under field conditions. Surprisingly, in contrast to libA, relatively high gene copy numbers of hylA in the agricultural soil did not always correspond to a high linuron mineralization capacity (9) (Fig. 4 and Table 2, day 0 and day 491 for microcosm L SM B after linuron application was stopped), indicating that hylA gene copy numbers do not always contribute to the measured mineralization capacity. This is explained by the abundances of dcaQ involved in the downstream metabolism of linuron and hence actual ¹⁴C-labeled CO₂ production. In contrast to other time points, for which a high mineralization capacity was recorded, dcaQ, and more precisely dcaQ_II, showed relatively low gene copy numbers at day 0 and, for microcosm L SM B, at day 491. The high gene copy numbers of hylA-containing bacteria and low gene copy numbers of dcaQ-containing bacteria indicate that at those time points, linuron was converted to 3,4-DCA but only slowly mineralized beyond 3,4-DCA and, as such, explain the apparent incongruence between high hylA abundances and low linuron mineralization capacity. These data have other important implications since they show that hylA-containing bacteria do not always contain the downstream pathway and as such must compose a part of consortia, including 3,4-DCA-mineralizing organisms that do not convert linuron into 3,4-DCA. At other time points, though, dcaQ gene copy numbers often exceeded hylA gene copy numbers, which implies that (i) each dcaQ-containing cell might contain an hylA copy but also that (ii) populations that contained only the 3,4-DCA catabolic pathway profited from organisms that perform HylA/LibA activity. The complex dynamics of the examined catabolic genes suggest a high plasticity of the different catabolic gene modules involved in linuron degradation in the agricultural soil as a response to linuron application potentially involving horizontal gene transfer, as previously suggested by Dunon et al. (24). Interestingly, hylA was identified in Variovorax strains that are part of consortia in which efficient linuron conversion to 3,4-DCA and the further mineralization of 3,4-DCA depend on synergistic metabolic interactions between Variovorax and other bacteria. In the consortium reported by Dejonghe et al. (4), Variovorax sp. strain WDL1, which carries hylA, converts linuron to 3,4-DCA. Strain WDL1 can grow on 3,4-DCA, but the conversion of 3,4-DCA is not that efficient, leading to a release of 3,4-DCA that is used for growth by a second consortium member, Comamonas testosteroni WDL7. Concomitant removal of 3,4-DCA in the culture in turn results in an improved rate of conversion of linuron by strain WDL1. In another similarly composed consortium that depends on hylA for initial linuron hydrolysis, a similar cooperation exists (3). IncP-1 plasmids are part of these consortia and, for instance, carry the genes for 3,4-DCA biodegradation in strain WDL7 (19). In contrast to hylA, libA was primarily found in Variovorax strains, such as strain SRS16 (6), which show efficient conversion of 3,4-DCA and hence growth on linuron as single strains. libA gene copy numbers were always below dcaQ gene copy numbers, and hence, each libA-containing cell might contain a dcaQ gene copy. Interestingly, in liquid cultures that were enriched for organisms that use linuron as sole carbon source and were initiated from the studied agricultural soil, the coexistence of libA- and hylA-containing bacteria is lost and hylA-containing consortia start dominating. Apparently, libA-containing bacteria are less competitive than hylA-containing strains under those culture conditions.

libA, hylA, and dcaQ dynamics in BPS.

The results show that, as in the agricultural soil, in the BPS matrix containing primed soil, hylA-containing bacteria compete successfully with libA-containing bacteria for linuron as a carbon source and grow alongside. hylA gene copy numbers did not respond on linuron application in BMs containing nonprimed soil, indicating that hylA-carrying microorganisms likely originate from soil L, which was included in the BM matrix, indicating a successful invasion of hylA linuron-degrading bacteria from that primed soil in the overall system. The same was concluded for libA (21). As with libA, hylA abundance appears to correlate with the observed dynamics of the linuron mineralization capacity in the system containing primed soil, i.e., hylA gene copy numbers were relatively high and low when the linuron mineralization capacity was high and low, respectively (Fig. 6). Interestingly, between week 51 and week 55, libA gene copy numbers remained the same or even tended to decrease in linuron-amended BMs while the linuron mineralization capacity significantly increased. In contrast, hylA gene copy numbers were extremely high and had increased 8-fold from week 51 onward. This occurred after the drought-rewetting period implemented between weeks 42 and 51. A similar observation was done for hylA in the nontreated BMs containing primed soil, indicating that this stress situation benefitted hylA-carrying populations in BPS. In contrast, libA-carrying bacteria seem to benefit first from the drought stress in the non-linuron-treated microcosms. In contrast to the agricultural soil, dcaQ gene copy numbers always exceeded hylA/libA gene copy numbers, implying that each hylA/libA-containing cell might also contain a dcaQ gene copy. However, populations that contained only dcaQ clearly benefited from linuron application, suggesting that as in the agricultural soil, 3,4-DCA-degrading organisms that contain only the pathway beyond 3,4-DCA profited from LibA/HylA activity performed by other organisms.

FIG 6.

Correlation between the lag time of linuron mineralization and the abundance of libA, hylA, dcaQ_I, and dcaQ_II as the log₁₀ value of the percentage of the bacterial 16S rRNA gene copy number in BM setups with linuron-primed soil treated with linuron (black) and water (white) of the BPS experiment. Values are average values (lag time [n = 3]; qPCR, n = 6). Linear regressions for libA and dcaQ_II and exponential regressions for hylA and dcaQ_II are shown as a dotted line. R² values and equation parameters are shown on the right. Symbols related to day 0 and day 491 are indicated.

In the linuron-fed BM setup that contained non-linuron-primed soil, some response of hylA gene copy numbers was recorded upon addition of linuron, but this population seems to deteriorate over time despite the maintenance of a high linuron mineralization capacity. Previously, libA was recorded to respond to linuron application in only one of the replicates of that SM setup, and these data also did not always explain the observed high mineralization capacity. As such, we conclude that still other linuron hydrolases or biochemical systems to initiate linuron mineralization exist in that environment, likely originating from the nonprimed soil C. PuhA (25) and PuhB (26) are other enzymes isofunctional to LibA and HylA, involved in linuron hydrolysis by Gram-positive bacteria, but they have never been linked with mineralization of linuron. They might form alternatives for initiating linuron mineralization in the environment carrying that capacity, potentially in accordance with dcaQ-containing populations that clearly proliferated upon linuron addition in the linuron-amended SMs containing nonprimed soil.

Conclusions.

Our results provide further insight into the microbial ecology of linuron biodegradation in agricultural environments. Both organisms containing libA and hylA coexist and contribute to linuron biodegradation. However, situations occur in which one is privileged over the other as observed under liquid enrichment conditions in minimal medium and in the BM experiment after the drought-wetting period. This might be related to the particular bacterial host which carries the genes or otherwise to the enzymatic kinetic parameters of the corresponding enzymes. LibA is known to have a 2-fold-higher affinity than HylA for linuron. Due to a lack of data on the maximal specific conversion rates for LibA, it is difficult, however, to draw conclusions from the role of the linuron hydrolysis kinetic parameters for usage of linuron as a growth substrate. Our study further shows that both hydrolysis genes and downstream catabolic genes should be used for assessing linuron mineralization capacity in environmental samples. However, other (unknown) enzymes initiating linuron mineralization apparently exist in the environment. Furthermore, our data strongly suggest that consortia involving both linuron-hydrolyzing organisms and organisms that further degrade the produced 3,4-DCA cooperate in complete linuron mineralization in the examined ecosystems. Moreover, complex dynamics and interactions of hydrolysis genes and downstream catabolic genes exist in the examined ecosystems, potentially involving horizontal gene transfer. This study proves the presence and activity of pesticide degraders functioning as synergistic consortia in natural environments.