Environ. Sci. Technol. 2007, 41, 5854-5861
Impact of Transgenic Tobacco on Trinitrotoluene (TNT) Contaminated Soil Community E M M A R . T R A V I S , †,‡ NERISSA K. HANNINK,† CHRISTOPHER J. VAN DER GAST,§ IAN P. THOMPSON,§ S U S A N J . R O S S E R , †,‡ A N D N E I L C . B R U C E * ,†,| Institute of Biotechnology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QT, UK, Institute of Biomedical & Life Sciences, University of Glasgow, Glasgow G12 8QQ, UK, NERC Centre for Ecology & Hydrology Oxford, Mansfield Road, Oxford OX1 3SR, UK, and CNAP, Department of Biology, University of York, York YO10 5YW, UK
Environmental contamination with recalcitrant toxic chemicals presents a serious and widespread problem to the functional capacity of soil. Soil bacteria play an essential role in ecosystem processes, such as nutrient cycling and decomposition; thus a decrease in their biomass and community diversity, resulting from exposure to toxic chemicals, negatively affects the functioning of soil. Plants provide the primary energy source to soil microorganisms and affect the size and composition of microbial communities, which in turn have an effect on vegetation dynamics. We have found that transgenic tobacco plants overexpressing a bacterial nitroreductase gene detoxify soil contaminated with the high explosive 2,4,6trinitrotoluene (TNT), with a significantly increased microbial community biomass and metabolic activity in the rhizosphere of transgenic plants compared with wild type plants. This is the first report to demonstrate that transgenic plants engineered for the phytoremediation of organic pollutants can increase the functional and genetic diversity of the rhizosphere bacterial community in acutely polluted soil compared to wild type plants.
Introduction Pollution of the environment with recalcitrant chemicals impacts severely on the microbial community (1-4). Microorganisms are vital for soil fertility and for the degradation of organic matter in soils and sediments. Bacterial communities in soil are likely to respond to pollution by changing their structure to one that favors organisms that are able to thrive under the new selective conditions at the expense of other organisms that are suppressed. The loss of sensitive microorganisms, often performing specific functions, can have serious ecological consequences (5), for instance impacting on nitrogen cycling (6). * Corresponding author address: CNAP, Department of Biology, University of York, PO Box 373, YORK, YO10 5YW, UK; phone: +44 1904 328777; fax: +44 1904 432928; e-mail:
[email protected]. † University of Cambridge. ‡ University of Glasgow. § NERC Centre for Ecology & Hydrology Oxford. | University of York. 5854
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 16, 2007
Explosives are a major cause of organic pollution, with TNT pollution being the most widespread. Historically TNT was the most widely used military explosive (7, 8), with a large number of sites used in World War II still heavily contaminated (9). TNT also contaminates military training ranges and sites of explosives manufacture. TNT, as with most explosive compounds, is toxic, mutagenic, and highly energetic (10), having a serious impact on the environment, and threatening human health (11, 12). TNT is a highly recalcitrant xenobiotic compound and, although soil bacteria have been isolated that can metabolize this explosive (13), the chemical intransigence of TNT at contaminated sites suggests that there is insufficient microbial biomass and activity present in soil to degrade TNT at any appreciable rate. To address this issue we investigated whether plants could be genetically engineered to yield an optimal system for in situ bioremediation of toxic explosives residues in soil. Progress has been made toward this goal by successfully combining the biodegradative capabilities of soil bacteria with the high biomass, stability, and sequestration properties inherent in plants. Increased tolerance to TNT, in sterile liquid cultures, was shown for Nicotiana tabacum (tobacco) engineered to express TNT-transforming bacterial enzymes (14, 15). Two bacterial enzymes were separately expressed in tobacco, pentaerythritol tetranitrate (PETN) reductase (16) and a classical type I nitroreductase (NR) (15), both enzymes deriving from strains of Enterobacter cloacae (17, 18) which have been shown to catalyze a two electron reduction of TNT to produce hydroxylamino- and amino-dinitrotoluene derivatives (13). Vila et al. recently demonstrated that these TNT metabolites are conjugated to glycose in tobacco as part of the detoxification process (19). While the experiments conducted in sterile culture provided valuable information, it is how the plants perform in the more complex soil environment that is of relevance to assess the phytoremediation potential. In soil, both abiotic and biotic factors could influence plant growth, health, and response to TNT. As well as the microbial community potentially influencing the plants, the tobacco plants are predicted to impact on the microbial community of the soil. One important aim was to determine whether NRexpressing tobacco plants were more tolerant to TNTcontaminated soil than wild type plants. One of the overall objectives of phytoremediation is to decrease the toxicity of contaminated soil to allow the indigenous microflora to flourish and contribute to the remediation process; therefore, it was considered imperative to determine the impact of an engineered TNT-tolerant transgenic tobacco line on the ecotoxicology of TNT-contaminated soil. We sought to investigate if transgenic plants can positively influence the functional and genetic diversity of the microbial community of contaminated soil.
Experimental Section Amendment of Soil with TNT. A well-characterized Rowland Series soil was chosen because of its low organic and clay content (20), as TNT is known to bind to clay (21) and organic matter (22). Soil was sieved to 2 mm size particles and water content was balanced at 10% (v/w). TNT was solubilized in acetone and added dropwise to soil to reach final concentrations of 0, 25, 50, 150, 500, 1500, and 4000 mg TNT/ kg soil. Soil Characterization. Soil pH was assessed directly by dissolving 5 g of soil in 5 mL of distilled water for 30 min prior to measurement using an EDT Instruments (Kent, UK) 10.1021/es070507a CCC: $37.00
2007 American Chemical Society Published on Web 07/12/2007
pH reader. The partition coefficient (Kd) for TNT in soil was estimated at 4.0 mL/g, based on Pennington (23). Growth of Plants in TNT-Amended Soil. Wild type and transgenic tobacco seeds were planted in a mixture of three parts Rowland Series soil to one part of sand and germinated at 23 °C with an 18-hour photoperiod. After five weeks of growth, uniform-sized plantlets were selected and transferred to 88 mm diameter, free-draining soil lysimeter columns containing TNT-amended soil. Unplanted soil columns served as controls. All conditions were performed in triplicate, with the soil for each plant analyzed individually. All columns were watered with identical volumes. Trays of plants were frequently rotated to ensure even growth of all plants. After 12 weeks growth, the rhizosphere soil (defined as the soil that remained attached to the roots after shaking) was separated from the bulk soil (the remaining soil). The lengths of the longest root and shoot were determined. Colony-Forming Unit Count and Biolog Eco Plates. Soil was diluted to 100 mg/mL in 50 mM phosphate buffered saline pH 8.0 (PBS) and homogenized by vortexing on maximum speed for 1 min. Serial dilutions were performed (from 10-4 to 10-7) and spread onto tryptic soya agar plates (Oxoid, Basingstoke, Hampshire, UK) which were incubated at 25 °C, with colonies counted every 48 h for 6 days. Univariate analysis of variance (ANOVA) was employed to test for significant associations. Aliquots of 150 µL of a 10-3 dilution of washed soil suspension were inoculated into each well of a Biolog EcoPlate (Biolog. Inc., Hayward, CA) prior to incubation at 25 °C. Absorbance at 595 nm was assessed by an EL800 Universal Microplate Reader (Bio-Tek Instruments, Inc. Winooski, VT) every 24 h for 6 days. Well color density was corrected compared to the control water-containing well, with negative values discounted for further analysis. The area under the absorbance versus time curve (AAT) was calculated using a trapezoidal approximation (24), with the end time point (96 h) determined when the majority of substrates had reached Vmax. The average well color development (AWCD) was calculated as the average area under the curve for all substrates for each sample to provide a measure of overall metabolic activity. ANOVA was employed to test for significant associations. The relative AAT compared to AWCD was calculated for each substrate to minimize bias in the data set due to varying inoculum densities between samples (25). The multivariate analytical procedure, principal component analysis (PCA), was then applied to the relative AATs (25) to further investigate the relationships between different samples’ carbon substrate utilization. Each principal component (PC) extracts a portion of the variance within the original data set, with the greatest amount of variance reflected in PC1. Plotting samples in two-dimensional space based on their first and second component allows relationships among samples to become apparent. Variables with low overall variance and therefore low individual measure of sampling adequacy (L-serine, Tween 80) were excluded from the PCA to increase the overall sampling adequacy of the analysis. Denaturing Gradient Gel Electrophoresis (DGGE). Genomic DNA (gDNA) was extracted directly from soil using the Ultra Clean Soil DNA Isolation Kit (Mo Bio Laboratories, Inc., Solana Beach, CA) and used as the template in polymerase chain reaction (PCR) to amplify a fragment of the 16S rDNA as described by Whiteley and Bailey (26). The primers used for PCR (GC338F (5′ - CGC CCG CCG CGC CCC CGC CCC GGC CCG CCG CCC CCG CCC ACT CCT ACG GGA GGC AGC -3′) and 530R (5′ - GTA TTA CCG CGG CTG CTG - 3′) yielded an amplicon corresponding to bases 338 to 530 (Escherichia coli numbering) of the bacterial 16S rDNA with a GC clamp.
DGGE analysis was performed as described by Whiteley and Bailey (26). Band profiling was performed using Scion Imaging for Windows Software (Scion Corporation, Frederick, MD) and dendograms were constructed for comparison of DGGE samples by the Pearson Correlation method. Individual bands were excised, extracted, PCR amplified using primers 338F (5′- ACT CCT ACG GGA GGC AGC -3′) and 530R, and sequenced at the PNAC facility, University of Cambridge, UK. TNT Extraction from Soil. Nitroaromatics were extracted from soil using an adaptation of the EPA 8330 method (27), prior to analysis by reverse-phase high-performance liquid chromatography (HPLC) using a Waters 4 µm C8 Nova Pak column on a Waters 2690 separations module with a Waters photodiode array detector (Waters Corporation, Milford, MA) (28). Statistical Analyses. Statistical analyses were performed using SPSS for Windows (Rel. 11.0.1. 2001. SPSS Inc., Chicago, IL).
Results Tobacco Characterization. NR-expressing tobacco grew better on soil contaminated with high concentrations of TNT than wild type tobacco, suggesting that the introduction of the bacterial nitroreductase gene increases the tolerance of tobacco to TNT (Figure 1a). At high levels of TNT (150 mg/kg TNT and above) the wild type tobacco exhibited greater signs of phytotoxicity than the transgenic NR-expressing tobacco, including stunting of roots and chlorosis of leaves. As the concentration of TNT present in the soil increased, the differences observed between the wild type and NR-expressing tobacco plants became more pronounced, with the wild type plants experiencing wilting and necrosis at 4000 mg/kg TNT, with extreme stunting of the roots (Figure 1b), whereas the NR-expressing plants still achieved a small amount of growth and sustained a low level of root development. Nitroaromatic Fate. The fate of TNT and its transformation products in soil was explored by extraction and HPLC (Figure 2). In general, the concentrations of TNT and its monoamino derivative 2ADNT were highest in unplanted soil, with the lowest levels observed in NR-expressing tobacco-planted soil. The ratio of 4ADNT to 2ADNT was higher in NR-expressing tobacco compared with wild type planted soil, or unplanted soil. In unplanted and planted soil 2,4-diaminonitrotoluene (2,4-DANT) appeared at low levels in soil contaminated with high concentrations of TNT. 2,4-DANT was assumed to be the product of anaerobic bacterial transformation. Planting with either wild type or NR-expressing tobacco had no detectable effect on the bulk soil microbial community, compared with the unplanted soil. Differences in the soil microbial community were accounted for by the concentration of TNT contamination. Therefore, for ease of interpretation, all figures illustrate the results for bulk soil from the lysimeters containing wild type plants, omitting the unplanted soil and the bulk soil from the lysimeters containing the NR-expressing transgenic plant line. Bacterial Population Enumeration. Culturable, aerobic, heterotrophic bacterial numbers varied both with level of TNT contamination and soil treatment (plant line and rhizosphere or bulk soil) as shown in Figure 3a. Univariate analysis of variance (ANOVA), using log transformed data to reduce heterogeneity of variance, showed that there was a main effect due to the factor of soil treatment: F(5,89) ) 54.5; p < 0.01. There was also a main effect due to the factor of TNT concentration: F(7,89) ) 11.2; p < 0.01 and an interaction effect between the factors soil type and TNT concentration: F(33,89) ) 2.0; p < 0.01 was also seen. The CFU counts for unplanted and the two bulk soils at low concentrations of TNT (25, 50 mg/kg) were not signifiVOL. 41, NO. 16, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5855
FIGURE 1. Comparison of tobacco plants grown in a range of TNTamended soil for 12 weeks. Photographs of (a) wild type (WT) tobacco plants, and (b) transgenic NR-expressing (NR) tobacco plants. TNT concentration increased from left to right and concentrations were as follows: 0, 50, 150, 500, 1500, and 4000 mg/kg TNT. (c) Comparison of root length for wild type (dotted bar) and transgenic NR-expressing (solid bar) tobacco. It should be noted that at TNT concentrations below 150 mg/kg, roots all grew to the maximum allowed by the experimental design (12 cm) and are therefore not illustrated. Bars represent the mean of triplicate samples and error bars show + 1 SE of the mean. cantly different from the control (acetone-amended) soil. At concentrations of 150 mg/kg and above there was a marked decrease in CFU counts for all three soils. The highest CFU count was observed in the rhizosphere soil samples. Both wild type and NR-expressing transgenic tobacco rhizosphere samples resulted in high CFU counts at 0, 25, and 50 mg TNT/ kg soil. There was a marked difference between wild type and transgenic rhizosphere samples at higher concentrations of TNT. For wild type rhizosphere samples the CFU count dropped dramatically to levels seen in bulk or unplanted soil at equivalent concentrations of TNT (150, 500 mg/kg). No roots were present for wild type tobacco when grown in highly TNT-contaminated soil (1500 and 4000 mg/ kg) and therefore there were no rhizosphere samples. In sharp contrast, for the rhizosphere samples from NR-expressing 5856
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 16, 2007
tobacco, the CFU count was high across the range of TNT concentrations. NR-expressing tobacco allowed for higher bacterial densities than wild type plants in the rhizosphere of highly TNT-contaminated soil. Carbon Substrate Utilization. The metabolic functioning of the soil microbial community was investigated through analysis of carbon substrate utilization patterns using Biolog EcoPlates. These plates have been designed to provide 31 carbon sources of ecological relevance for microbial growth. An increase in the concentration of TNT in the bulk soil also led to a shift in the carbon utilization pattern of the culturable microbial community, reflecting the change in its functional diversity. The AWCD (a measure of the metabolic activity of the culturable portion of the soil microbial community) varied with TNT concentration and soil treatment (Figure 3b). ANOVA allowed statistical determination of associations between type of soil, TNT concentration, and AWCD. It showed that there was a main effect due to the factor of soil treatment: F(5,79) ) 14; p < 0.01. There was also a main effect due to the factor of TNT concentration: F(7,79) ) 18; p < 0.01. An interaction effect between the factors soil treatment and TNT concentration: F(33,79) ) 6; p < 0.01 was also seen. The CFU per gram of soil was considered as a covariate but had no significant main effect on AWCD. For unplanted and bulk soil the AWCD was highest at intermediate concentrations of TNT (50, 150, 500 mg/kg). At the highest levels of TNT (1500 and 4000 mg/kg), however, the metabolic activity declined, as did the number of carbon sources assimilated by the microbial community. The rhizosphere samples for both wild type and transgenic tobacco displayed a different pattern compared with the bulk or unplanted soil. For the rhizosphere samples, the low and high concentrations exhibited a higher AWCD than the intermediate TNT concentrations (50, 150 mg/kg). It was in the transgenic rhizosphere samples at high TNT concentrations (500, 1500, and 4000 mg/kg TNT) that the highest AWCD from any soil treatment or concentration of TNT was observed. Principal component analysis was performed to obtain a more detailed insight into the patterns of substrate utilization. PCA analysis of the Biolog EcoPlate data indicated that the patterns of carbon utilization varied across the TNT concentration in a fashion dependent on the treatment of soil. Figure 4 shows PC1 plotted against PC2 for bulk and rhizosphere soils across a range of concentrations of TNT. The variation in PC1 was most strongly associated with the type of soil tested. In general, rhizosphere soil samples exhibited a positive PC1 (Figure 4b and c), while soil samples from non-rhizosphere origins tended to be negative (Figure 4a). A trend also existed for PC1 across the range of concentrations of TNT-contaminated soils, with the more highly contaminated soil showing a more positive PC1 than soils with low levels of contamination. The variation in PC2 could mainly be attributed to changes in TNT concentration. For the unplanted and bulk soils uncontaminated soil, or soil that was only contaminated with low concentrations of TNT (25, 50 mg/kg), exhibited a negative PC2, whereas more highly contaminated soil showed a positive PC2. For rhizosphere samples from both tobacco genotypes, the PC2 value became more positive as TNT concentration increased. This trend was less pronounced than for bulk soil. It was of interest that five of the eight substrates that were strongly loaded in a positive fashion on PC1 were carbohydrates, while none of the six substrates negatively loaded on PC1 were carbohydrates. No clear pattern of substrate type emerged as being strongly loaded either negatively or positively on the second principal component. Genetic Diversity. Genetic diversity was investigated by extraction of the genomic DNA from the soil, PCR amplification of 16S rDNA fragments, followed by separation by DGGE.
FIGURE 2. Concentrations of nitroaromatics extracted from the soil compared to the initial concentrations of TNT added to the soil: (a) TNT, (b) 2ADNT, (c) 4ADNT, and (d) 4HADNT. Bars represent the mean of triplicate samples and error bars show ( 1 SD of the mean. Genomic DNA of between 5 and 12 kb was extracted from all samples, as confirmed by gel electrophoresis, with typical yields of 4 ( 1 µg/g soil. DGGE revealed that the genetic diversity of the soil altered in a manner dependent on both the treatment of the soil and the concentration of TNT at which the soil had been amended. Typical DGGE images of samples across the range of TNT concentrations and derived from each soil treatment are shown in Figure 5a. Triplicate samples of soil for each condition yielded similar banding patterns. For unplanted and bulk soils, as the concentration of TNT increased, additional bands became prominent. Figure 6 shows a phylogenetic tree of the excised and sequenced bands. Sequencing and alignment of the emergent bands revealed the similarity to pseudomonads and a Burkholderia sp. The most dominant band arose from a 16 S rDNA sequence that was 100% identical to a Pseudomonas putida. The banding patterns of the rhizosphere soil samples exhibited a greater diversity of bands, with diversity increasing as TNT concentration increased. Pearson correlation (Figure 5b) of the digitized gel lanes showed that samples contaminated with low concentrations of TNT (
