OUP user menu

Abundance and transferability of antibiotic resistance as related to the fate of sulfadiazine in maize rhizosphere and bulk soil

Christoph Kopmann, Sven Jechalke, Ingrid Rosendahl, Joost Groeneweg, Ellen Krögerrecklenfort, Ute Zimmerling, Viola Weichelt, Jan Siemens, Wulf Amelung, Holger Heuer, Kornelia Smalla
DOI: http://dx.doi.org/10.1111/j.1574-6941.2012.01458.x 125-134 First published online: 1 January 2013


Veterinary antibiotics entering agricultural land with manure pose the risk of spreading antibiotic resistance. The fate of sulfadiazine (SDZ) introduced via manure and its effect on resistance gene levels in the rhizosphere were compared with that in bulk soil. Maize plants were grown for 9 weeks in soil fertilized with manure either from SDZ-treated pigs (SDZ treatment) or from untreated pigs (control). CaCl2-extractable concentrations of SDZ dissipated faster in the rhizosphere than in bulk soil, but SDZ remained detectable over the whole time. For bulk soil, the abundance of sul1 and sul2 relative to 16S rRNA gene copies was higher in the SDZ treatment than in the control, as revealed by quantitative PCR on days 14 and 63. In the rhizosphere, sampled on day 63, the relative sul gene abundances were also significantly increased in the SDZ treatment. The accumulated SDZ exposure (until day 63) of the bacteria significantly correlated with the log relative abundance of sul1 and sul2, so that these resistance genes were less abundant in the rhizosphere than in bulk soil. Plasmids conferring SDZ resistance, which were exogenously captured in Escherichia coli, mainly belonged to the LowGC group and carried a heterogeneous load of resistances to different classes of antibiotics.

  • sulfadiazine
  • sul genes
  • manure
  • maize rhizosphere
  • soil
  • mesocosm
  • LowGC-type plasmid


Sulfadiazine (SDZ) belongs to the class of sulfonamide antibiotics that are used for veterinary purposes worldwide, mainly in pig production (Burkhardt et al., 2005; Sarmah et al., 2006). SDZ and its metabolites N-acetyl-SDZ (N-Ac-SDZ) and 4-hydroxy-SDZ (4-OH-SDZ) are almost completely excreted by animals and enter agricultural soils through the use of manure and slurry as fertilizer (Lamshöft et al., 2007, 2010). Here, SDZ can have effects on the functional and structural composition of the soil microbial community and its activity (Hammesfahr et al., 2008, 2011; Kotzerke et al., 2008), and it may increase the abundance and transferability of antibiotic resistance genes (reviewed by Heuer et al.,2011a,b). Bacterial resistance to sulfonamides is mainly mediated by the genes sul1, sul2, and sul3 coding for dihydropteroate synthases, which are insensitive to sulfonamides (Sköld, 2000; Perreten & Boerlin, 2003). Recently, the effect of pig manure spiked with SDZ on the absolute and relative abundance of sulfonamide resistance genes and transfer frequencies of plasmids conferring SDZ resistance in exogenous plasmid isolations were investigated in a microcosm experiment in the absence of plants (Heuer & Smalla, 2007). The SDZ resistance genes sul1 and sul2 were detected in a silt loam and loamy sand, whereas the occurrence of sul3 genes was not observed (Heuer & Smalla, 2007). The study also clearly showed that manure spiked with SDZ significantly increased the abundance of sul1 and sul2 genes in soil bacteria and increased transfer frequencies of captured plasmids conferring SDZ resistance (Heuer & Smalla, 2007).

Bacterial communities of manure and soil were shown to be distinct (Hammesfahr et al., 2008; Heuer et al., 2008). Despite an insufficient adaptation of bacteria from manure to soil conditions, antibiotic resistance genes can be transferred from manure-derived bacteria to soil bacteria through plasmids (Brown, 2003; Sørensen et al., 2005; Binh et al., 2008). A recently described class of self-transferable plasmids with low GC content was identified as a major vector of sul2 in manure and manure-treated soils (Heuer et al., 2009).

For an integrated risk assessment, resistance gene dynamics need to be considered together with the fate of the antibiotic. Soil extractions with mild solvents such as 0.01 M CaCl2 and methanol (Barriuso et al., 2004; Laabs & Amelung, 2005; Förster et al., 2009) yield an easy proxy of antibiotic bio-accessibility in soil, referred to in the following as the easily-extractable fraction (EAS) (Rosendahl et al., 2011). This fraction typically decreases rapidly for SDZ (Förster et al., 2009; Rosendahl et al., 2011). The remobilization of SDZ from a second more strongly bound antibiotic fraction, however, may maintain low concentrations of easily-extractable SDZ over longer periods (Zarfl et al., 2009). Previously, it was shown that SDZ may exert a selective pressure at concentrations down to 150 μg kg−1 bulk soil (Heuer et al., 2008). The level of resistance genes was furthermore maintained at much lower concentrations of easily-extractable SDZ over a period of 150 days (Heuer et al., 2008).

In this context, several findings suggested that the coupling between fate and effects differs between the rhizosphere and bulk soil. On the one hand, it was shown that the addition of artificial root exudates increased the bacterial community tolerance toward SDZ, indicating that the rhizosphere might be a hotspot of resistant bacteria (Brandt et al., 2009). On the other hand, the exposure of bacteria to SDZ is presumably reduced in the rhizosphere, as the dissipation of easily-extractable SDZ was recently shown to be accelerated in rhizosphere soil (Rosendahl et al., 2011). However, little is known about the abundance and dynamics of sulfonamide resistance genes in the rhizosphere. We therefore aimed to compare the fate and effect of SDZ in bulk and rhizosphere soil of maize plants in a mesocosm experiment. In comparison with former laboratory-based microcosm studies (Heuer & Smalla, Heuer et al., 2011a,b), such a mesocosm experiment allows for the control of microclimatic variations for an optimal growth of plants in large containers. In contrast to previous experiments, SDZ was not spiked, but manure from SDZ-treated and untreated pigs was used. Bulk and rhizosphere soil were separately sampled, and a sequential extraction protocol yielded antibiotic fractions of increasing binding strength. The abundance and dynamics of sul1 and sul2 and major plasmid vectors were assessed by quantitative real-time PCR in total community DNA. Transferable SDZ resistance plasmids were captured in GFP-tagged E. coli recipients.

Materials and methods

Mesocosm experiment

Manure was produced by intramuscular administration of the prescribed dose of SDZ (30 mg kg−1 bodyweight) on four consecutive days and collection of manure for 10 days. SDZ injectable solution (200 mg mL−1) for this purpose was kindly supplied by Vetoquinol Biowet (Gorzow Wielkopolski, Poland). Control manure was collected from pigs not treated with antibiotics. The soil used for all trials was the topsoil of a Luvisol from Merzenhausen (Germany) with 1.2% organic carbon, a pH (CaCl2) of 6.3, a cation exchange capacity of 11.4 cmol kg−1 (measured at pH 8.1), 16% clay (< 2 μm), 78% silt (2–63 μm), and 6% sand (63–2000 μm). Manure [90.2 and 87.8% moisture (w/fw) for manure free from antibiotics and manure-containing SDZ] and soil [c. 15% residual soil moisture (w/dw)] were extensively mixed with an electric mixing machine and a backhoe at a ratio of 1 : 25 (w/dw), and the soil was filled into polystyrene tubes (140 × 80 × 40 cm, n = 4) that were planted with maize (Zea mays L.) the next day (see Supporting Information, Fig. S1). No assessment of heterogeneity of the distribution of manure in the tubs was carried out; however, samples were always taken as composite samples to compensate for potential heterogeneity. The temperature was kept constant during incubation (21 °C), and plants were illuminated with two halogen lamps (400 W), each with a day/night cycle of 16:8 h. Soil moisture varied between 8.5 and 15.2% (w/dw), and water loss was replenished regularly. To determine the concentrations of SDZ and its metabolites, soil was sampled on days 0, 1, 7, 14, 28, 42, and 63. Bulk soil (≥ five subsamples) was taken from in between maize rows (Fig. S2) using a soil sampling cylinder (9.2 cm height × 1.2 cm inner diameter). Sampling of rhizosphere was not possible before day 14, because the maize plants were still too small before that sampling date. Rhizosphere soil was sampled by digging out ≥ 4 maize plants and was defined as soil adhering to the roots after shaking (Fig. S3). It has to be noted that the high sensitivity of the molecular methods allowed a more rigorous removal of soil adhering to the root system than the extractions determining the soil concentrations of SDZ, for which a minimum of 10 g of soil sample was required. Therefore, the results of the DNA-based methods for rhizosphere soil refer to soil tightly adhering to the roots.

DNA extraction and denaturing gradient gel electrophoresis

Total community DNA was extracted from day 14 for bulk soil, day 63 for bulk and rhizosphere soil samples, as well as from soil before manure application and manure from treated and untreated pigs. DNA from 0.5 g roots, bulk soil, or manure was extracted using the Fast DNA® Spin Kit for soil (MP Biomedicals, Heidelberg, Germany), followed by a purification step using the Geneclean® Spin Kit (MP Biomedicals), according to the manufacturer.

Amplification of 16S rRNA gene fragments of bacteria from total community DNA of bulk soil, rhizosphere, or manure samples and separation of PCR products in denaturing gradient gel electrophoresis (DGGE) were carried out as previously described (Weinert et al., 2009). After alignment, Pearson correlation was performed with the program GelCompar II® (version 6.5, Applied Maths, Austin, TX). The similarity matrix was used to test for significant treatment effects and to determine the adjusted difference between bacterial communities as described by Kropf et al. (2004).

Exogenous plasmid isolation

Escherichia coli K-12 CV601gfp (Rifr Thr Leu Thi miniTn5::gfp-nptII) was used as recipient in exogenous plasmid isolation, essentially carried out as described previously (Heuer et al., 2002). The recipient strain was incubated in 5 mL of Luria Bertani (LB) broth (Carl Roth GmbH + Co. KG, Karlsruhe, Germany), supplemented with kanamycin and rifampicin (both 50 mg L−1) at 28 °C overnight. Cells were harvested by centrifugation (4880 g, 5 min) and washed twice with 0.85% sodium chloride solution. The pellet was resuspended in 2.5 mL of LB broth. One gram of bulk soil or rhizosphere soil (soil adhering to the roots after gentle shaking) from days 28, 42, and 63 was resuspended in 9 mL of 1/10 tryptic soy broth (Becton Dickinson GmbH, Heidelberg, Germany), supplemented with five sterile glass bowls (4 mm diameter), and incubated on an overhead shaker (GFL 3025, Braunschweiger Labor Bedarf, Germany) for 2 h. Four milliliter of bulk or rhizosphere soil solution was transferred into a new tube, and 1 mL of the recipient strain solution was added. Bulk and rhizosphere soil solutions without recipient solution were used as controls. The tubes were centrifuged at 4880 g for 5 min, and the pellets were resuspended in 100 μL of LB broth and spotted onto sterile filter disks (pore size 0.22 μm, Millipore, Billerica, MA) on plate count agar (PCA, Merck KGaA, Darmstadt, Germany) plates, supplemented with cycloheximide (100 mg L−1). The filters were incubated overnight at 28 °C. The cell lawn was resuspended in 10 mL of 0.85% sodium chloride solution, and transconjugants were obtained after plating serial 10-fold dilutions on Mueller-Hinton Agar (Merck, Darmstadt, Germany), supplemented with kanamycin (50 mg L−1), rifampicin (50 mg L−1), SDZ (50 mg L−1), and cycloheximide (100 mg L−1). The controls were plated undiluted on the same selective media. After 2 days of incubation at 28 °C, transconjugants were selected by GFP fluorescence and transferred to fresh plates of the selective media. Transfer frequencies were obtained by dividing the colony-forming unit (CFU) counts of transconjugants by the CFU counts of recipients. Differences between transfer frequencies obtained on day 63 were tested for significance using the Tukey test (P< 0.05, SAS 9.2, SAS Institute Inc., Cary, NC).

Plasmid extraction

Plasmids from transconjugants were extracted using the Qiagen Plasmid Purification Kit (Qiagen, Hilden, Germany) following a modified protocol for plasmids larger than 50 kpb, suggested by the manufacturer. For desalting, ultrafiltration was applied using Microcon YM-100 filter tubes (Millipore, Bedford, MA) instead of alcohol precipitation.

Characterization of the captured plasmids

Extracted plasmids were digested with the enzymes BstZ17I and PstI (Fermentas GmbH, St. Leon Rot, Germany), according to the manufacturer. Antibiotic resistance profiles were analyzed by the Robert Koch Institute (Wernigerode, Germany) according to DIN 58940 (German Institute for Standardization, Berlin, Germany) for 10 transconjugants harboring plasmids that were representative for the different plasmid restriction types. Plasmids extracted from transconjugants were characterized by dot-blot and Southern blotting following standard protocols (Sambrook et al., 1989). Dot- and Southern-blotted plasmid DNA was hybridized with digoxigenin-labeled probes according to the manufacturer's instructions (Roche Diagnostics, Mannheim, Germany). Probes for traN and repB (LowGC-type plasmids), trfA (IncP-1ε), and rep (IncN) were generated from PCR products amplified from pHHV216, pKJK5, and RN3, respectively, according to Heuer et al. (2009, 2012) and Götz et al. (1996). Probes for sul1 and sul2 were generated from plasmids R388 and RSF1010 according to Heuer & Smalla (2007). Chemiluminescence was measured with a LAS-3000 luminescent image analyzer (Fujifilm, Düsseldorf, Germany).

Detection and quantification of target genes

The respective gene copies were quantified by real-time PCR 5′-nuclease assays in an CFX96 Real-Time PCR Detection System (Bio-Rad, Hercules, CA) as described previously for sul1, sul2, bacterial 16S rRNA genes (rrn) (Heuer & Smalla, 2007; Heuer et al., 2009) and traN (Heuer et al., 2009). To adjust for differences in bacterial DNA extraction and amplification efficiency between samples, the relative gene abundance was calculated by dividing the copy numbers of the target genes sul1, sul2, or traN by rrn copy numbers (Fig. S4) using the following equation: Embedded Image

Pairwise comparisons were made using the Tukey test (P< 0.05).

Extraction of antibiotics

To obtain antibiotic fractions of different binding strength and bio-accessibility in soil, antibiotics were sequentially extracted as described previously (Förster et al., 2009). Briefly, a shaking extraction (24 h) with 0.01 M CaCl2 was followed by a methanol step (4 h) and a third exhaustive extraction in the microwave using acetonitrile : water (20 : 80, v/v). SDZ and its major metabolites 4-OH-SDZ and N-Ac-SDZ were analyzed in the extracts via LC/MS-MS. Details of the analytical protocol can be taken from Rosendahl et al. (2011). Concentrations in the CaCl2 and the methanol extracts were summed up to give the EAS, which may serve as a proxy for the bio-accessible antibiotic concentrations. The microwave extraction yielded the more strongly bound residual fraction (RES).

The fitting of first-order dissipation models Eqn. was performed using nonlinear regression (Sigma Plot 11.0, Systat Software GmbH, Erkrath, Germany):

Embedded Image

with C(t) defining the concentration (μg kg−1) still present in soil at time t (day), C0 the initial concentration (μg kg−1), and k (d−1) the dissipation rate constant. Dissipation half-lives (DT50) were calculated from Eqn. with DT50 = ln(2) k−1. Differences between concentrations in bulk and rhizosphere soil were tested for statistical significance for each sampling date using a Mann–Whitney U-Test (P< 0.05, Statistica 7, Statsoft, Hamburg, Germany). Differences between DT50 were assessed based on the underlying dissipation rate constant k ± its standard error of parameter estimation. The exposure (E) to SDZ until day 63 was roughly estimated by summing up easily-extractable concentrations (c) of each sampling day (i) starting from day 14 (first sampling of the rhizosphere) multiplied by the days (d) to the previous sampling:

Embedded Image

Correlation was tested by linear regression analysis (P< 0.05, SAS 9.2, SAS Institute Inc., Cary, NC).

Results and discussion

Fate of SDZ and its metabolites

As an estimation of bio-accessibility in soil, the EAS of SDZ was analyzed in the present study. Following the application of manure, concentrations of easily-extractable SDZ, 4-OH-SDZ, and N-Ac-SDZ decreased rapidly in bulk soil (Fig. 1). Fitted dissipation models yielded dissipation half-lives (DT50) of 5.1 days (SDZ), 17.6 days (4-OH-SDZ), and 0.7 day (N-Ac-SDZ). Residual concentrations decreased much slower with DT50 of 85.6 (SDZ) and 161.2 days (4-OH-SDZ) (Fig. 1). Intriguingly, low concentrations (< 5 μg kg−1) of SDZ and 4-OH-SDZ remained easily-extractable over the entire trial period, possibly due to a remobilization from sequestered antibiotic residues, as shown by Zarfl et al. (2009). The risk of a long-term exposure to such subtherapeutic antibiotic concentrations is currently unknown.

Concentrations of SDZ, 4-OH-SDZ, and N-Ac-SDZ in the easily extractable (EAS) and residual (RES) fraction in bulk soil (means of four replicates ± SE); lines represent modeled dissipation curves.

In rhizosphere soil, easily-extractable and residual concentrations of SDZ and 4-OH-SDZ were mostly lower than in bulk soil (Table S1), and this difference was significant for easily-extractable SDZ on days 28 and 63, that is, on two of four sampling days. After 14 days, the maize plants were still very small so that the rhizosphere effect of the 14-day-old plants may have impacted the dissipation of SDZ less strongly than the rhizosphere of well-developed plants on day 63. The fact that differences in the SDZ concentrations between bulk and rhizosphere soil could not be detected on day 14 after manure application is thus in line with earlier statements that rhizosphere effects take time to develop (e.g. Hsu & Bartha, 1979; Chaineau et al., 2000; Yu et al., 2003).

When applying a nonlinear regression analysis for the calculation of dissipation half-lives (see Eqn. ), the DT50 values of the EAS of SDZ were, in the range of the analytical precision, smaller in rhizosphere soil (DT50 = ln(2) k−1 ± SE k = 4.0–4.1 d; Table S2) than in bulk soil (4.6–5.8 days). Though with overlapping standard errors, the same applies for 4-OH-SDZ (range of DT50 = 13.1–17.8 days in the rhizosphere and 13.9–24.7 days in bulk soil; Table S2). There was no detectable difference for the residual fraction (DT50: 93.6 days for SDZ and 133.3 days for 4-OH-SDZ). Significantly different soil concentrations of easily-extractable SDZ and different dissipation rate constants together with earlier findings of accelerated dissipation of CaCl2-extractable SDZ in a field trial (Rosendahl et al., 2011) all together suggest a slightly faster dissipation of bio-accessible SDZ in rhizosphere soil compared with bulk soil.

Taking into consideration that plant uptake is generally too small to explain the loss of SDZ from rhizosphere soil (Rosendahl et al., 2011) and that an increased sequestration of SDZ in the residual fraction in the rhizosphere soil was not observed, we suggest in line with Rosendahl et al. (2011) that the degradation of easily-extractable SDZ may be accelerated by rhizosphere microorganisms. These effects might even be more pronounced in the soil tightly adhering to the plant roots, which was used for the DNA-based analysis.

Effect of SDZ manure treatment on bacterial communities in bulk and rhizosphere soil

Based on the previous results, day 14 and day 63 were selected for in-depth analysis of the abundance of SDZ resistance and plasmid replicon genes in the bulk and rhizosphere soil of maize (Heuer & Smalla, 2007; Heuer et al., 2008, 2009). Thereby, the effect of manure from SDZ-treated or untreated pigs on the abundance of sul genes was investigated as well as on the abundance of traN genes of LowGC-type plasmids identified as a major vector of sul2 in manure and manure-treated soils (Heuer et al., 2009). The effect of SDZ manure in the rhizosphere of maize compared with bulk soil was determined on day 63, which was the longest time of unrestricted growth of the maize plants in the mesocosms. The usage of manure from SDZ-treated or untreated pigs in contrast to spiked manures represents a more realistic scenario concerning concentrations of SDZ and its metabolites as well as of resistance gene abundance. Correspondingly, the relative abundance of sul1 and sul2 as well as of traN was about one half (sul1, sul2) to two (traN) orders of magnitude higher in the manure from SDZ-treated pigs than in the manure from untreated pigs (−1.6, −1.8, and −1.6 compared with −2.1, −2.3, and −3.6, respectively; one composite sample analyzed). This correlation between antibiotic use and the abundance of resistance genes in field-scale manures was already described in previous studies (Binh et al., 2008, 2010; Heuer et al., 2011a,b). Additionally, DGGE analysis of 16S rRNA gene fragments indicated differences in the bacterial community composition between manure originating from SDZ-treated pigs and manure from untreated pigs (Fig. S5).

Before manure application, the log relative abundances of sul1, sul2, and traN genes in Merzenhausen soil were −5.0, −5.8, and −3.8, respectively (measured in one composite sample). Fourteen days after manure application from treated or untreated pigs, the relative gene abundances in soil were increased by about one (traN) to three (sul1, sul2) orders of magnitude (Fig. 2) as previously observed in the same trend by Heuer & Smalla (2007) and Heuer et al. (2009) for the same soil. This increase in relative abundance was significantly higher after the application of manure from SDZ-treated pigs (Fig. 2), which might have been caused by the selective effect of SDZ but also by the increased abundance of sul and traN genes in the applied manure and the differences in the manure bacterial communities.

Effects of manure without and with SDZ (gray and dark gray, respectively) on the abundances of sulfonamide resistance genes sul1 and sul2 and plasmid replicon gene traN (affiliated with pHHv216) relative to rrn gene copy numbers in (a) bulk soil on day 14 and (b) bulk and rhizosphere soil of maize on day 63, quantified by quantitative real-time PCR. Error bars indicate the standard deviations (n= 4). For quantification of traN genes on day 63 (SDZ treatment, rhizosphere), only three replicates were analyzed. Different letters indicate significant effects of SDZ for each gene (Tukey test, P< 0.05).

The effect of SDZ on the abundance of sul and traN genes is likely as in previous microcosm experiments SDZ was spiked to manure just before its application to Merzenhausen soil and thus manure contained comparable bacterial communities and just differed in the presence of SDZ (Heuer & Smalla, 2007; Heuer et al., 2009). In these studies, significant effects of SDZ at much higher concentrations as in the present study were detected on the abundance of sul and traN genes even 61 days after manure application.

In the present study, concentrations of easily-extractable SDZ in the bulk soil ranged from 3 to 40 μg kg−1 until day 14. Between day 14 and day 63, the concentration of easily-extractable SDZ ranged from 1.8 to 0.2 μg kg−1 in the rhizosphere and from 3 to 1 μg kg−1 in the bulk soil (Table S1) which is far below easily-extractable concentrations of SDZ of 0.15 mg kg−1 reported to influence the sul2 gene frequency in soil (Heuer et al., 2008). Considering that the protocol for chemical extraction required a relatively big amount of soil material (10 g) and therefore the spatial distribution of SDZ is averaged over a relatively big distance to the roots, the SDZ concentrations affecting the bacterial community in direct vicinity to the roots might have been even lower. However, it is known that subinhibitory concentrations of antibiotics can have multiple effects on bacterial cells (Davies et al., 2006). Recently, it was shown that even concentrations of antibiotics of up to several hundred-fold below the minimal inhibitory concentration could select for resistant bacteria (Gullberg et al., 2011). Concomitantly, although the relative sul2 abundance in bulk soil was increased on day 63 compared with day 14 in both the SDZ treatment and the control (Fig. 2), the difference in relative abundance of sul1 between SDZ treatment and control was significantly higher on day 63 than on day 14 (t-test, P< 0.05), and the same trend was observed for the relative abundance of sul2. While Heuer & Smalla (2007) observed that SDZ promotes antibiotic resistance for over 2 months at concentrations in the mg kg−1 range, this increased difference in relative abundance of sul1 on day 63 may further suggest a selection for resistant bacteria and an influence of SDZ on the soil bacterial community at more realistic, very low easily-extractable concentrations. It should be noted, however, that despite low concentrations of SDZ detected in bulk and rhizosphere soils, there still might be sites with higher concentrations of SDZ because of a heterogeneous distribution of humic substances in soil that rapidly integrate SDZ (Junge et al., 2011).

On day 63, the absolute copy numbers of sul1 and sul2 as well as the 16S rRNA gene copy numbers of bulk soil were about one half to two orders of magnitude higher than on day 14 (see Fig. S4). The increase in 16S rRNA gene copies of about 0.5 log units might be either because of an increased relative abundance of populations with multiple 16S rRNA gene operons or because of an increased number of bacteria grown on maize root exudates and manure-derived nutrients within the containers. DGGE analysis of 16S rRNA gene fragments revealed that the bacterial community composition of the soil treated with SDZ containing manure differed significantly from the control 14 and 63 days after manure application (permutation test according to Kropf et al. (2004), P< 0.05). This difference between treatments was probably due to resistant bacterial populations either applied to soil through manure or enriched after the application of manure containing SDZ. The difference between SDZ treatment and control in bulk soil was higher on day 63 than on day 14 (differences between treatments of 14 and 9%, respectively), which may further support the concept of selection for resistant bacteria and an influence of SDZ on the soil bacterial community at very low easily-extractable concentrations.

Relative sul gene abundance related to SDZ exposure

Based on the observation of Brandt et al. (2009) that the addition of artificial root exudates increased the bacterial community tolerance toward SDZ, an increased proliferation of SDZ resistance in the rhizosphere of maize could be expected. The rhizosphere is well known as hot spot of horizontal gene transfer (Heuer & Smalla, 2012) as root exudates increase the relative abundance and metabolic activity of some bacteria, and often the cell densities are higher in the vicinity of the roots. Intriguingly, the differences between SDZ treatment and control for the relative abundance of sul1 and sul2 on day 63 were significantly higher for the bulk than for the rhizosphere soil (t-test, P< 0.05). This smaller difference observed in the rhizosphere between SDZ treatment and control fits well with the observed faster dissipation of easily-extractable SDZ in rhizosphere soil. Linear regression analysis showed a correlation between relative sul gene abundances on day 63 and the estimated accumulated exposure to SDZ in bulk and rhizosphere soil until day 63 (Fig. 3). Although we cannot exclude that differences in the bacterial community composition of rhizosphere and bulk soil, as stated previously, may have contributed to the different effect on relative sul abundance, the results of the regression analysis suggest that smaller concentrations of SDZ in the rhizosphere compared with bulk soil may have led to a reduced selective pressure in the rhizosphere. This reduced selective pressure may have had a bigger impact on relative sul abundance than enhanced microbial activity and rate of horizontal gene transfer related to nutrients provided by root exudation.

Relationship between relative abundance of sul1 (a) or sul2 (b) and exposure to SDZ, calculated for day 63 of bulk and rhizosphere soil. The exposure of bacteria to SDZ over time was roughly estimated by summing up easily-extractable concentrations of each sampling multiplied by the days to the previous sampling. Linear regression analyses showed a significant relation for both genes (P< 0.05). Gray areas represent the 95% confidence limits (n = 8).

Capturing and characterization of antibiotic resistance plasmids

Exogenous isolation of transferable elements conferring SDZ resistance was used to evaluate the effect of antibiotics in pig manure on the transferability of mobile genetic elements conferring SDZ resistance. For this purpose, filter matings with soil bacteria as plasmid donors and gfp-tagged and rifampicin-resistant E. coli recipient strains were performed. Escherichia coli was chosen as recipient because of its importance in human medicine as well as its sensitivity to SDZ. The determination of transfer frequencies (TF) on day 63 revealed no significant increase in the SDZ treatment compared with the control for bulk soil (log(TF) ± SD of −6.3 ± 0.2 and −6.7 ± 0.3, respectively) and for the rhizosphere (log(TF) ± SD of −5.9 ± 0.2 and −5.8 ± 0.5, respectively; Tukey test, P< 0.05). Hence, an SDZ effect on the TF as previously observed by Heuer & Smalla (2007) at concentrations of 10–100 mg SDZ kg−1 soil could not be confirmed in the present study. However, significantly higher TF were obtained for both rhizosphere treatments compared with bulk soil control confirming previous studies identifying the rhizosphere as a ‘hot spot’ of plasmid transfer that appears to be controlled by exudation and root growth affecting the cell density and distribution (for a review see Mølbak et al., 2007). A total of 293 antibiotic resistance plasmids were captured by exogenous plasmid isolation in gfp-tagged rifampicin-resistant E. coli, and 119 of them were further analyzed by hybridization. The plasmids were predominantly affiliated to the LowGC-type (66%), and all carried sul2 genes, indicating a high abundance and involvement of this previously described plasmid type (Heuer et al., 2009) in SDZ resistance. On day 63, 69% and 75% of the transconjugants from bulk and rhizosphere soil, respectively, captured LowGC-type plasmids. The higher percentage of captured LowGC-type plasmids in the rhizosphere compared with bulk soil is in agreement with the significant increase in the relative abundance of the traN gene in the rhizosphere of maize grown in SDZ manure-treated soil compared with the control, while in bulk soil, a significant increase in the SDZ manure-treated soil compared with the control could be observed on day 14, but not on day 63 (Fig. 2). Our data therefore suggest an enhanced propagation of populations carrying this plasmid type in the rhizosphere or a higher competitiveness in the rhizosphere bacterial community even at the low-level concentrations of easily-extractable SDZ present in rhizosphere soil on day 63. Hence, the LowGC-type plasmids seem to be highly present in piggery manure-treated soils confirming the findings from soil microcosm experiments (Heuer et al., 2009), although it has to be considered that only plasmids able to transfer to and replicate in E. coli were captured. Plasmids belonging to the novel group of LowGC-type were also observed in manure samples from manure storage tanks of 15 farms in Germany (Binh et al., 2008).

Analysis of the complete sequence of three LowGC-type plasmids exogenously captured from manure-treated soils (Heuer et al., 2009) revealed a conserved 30-kbp backbone with only 36% G+C content, composed of transfer and maintenance modules with moderate homology to plasmid pIPO2 and a replication module (rep and oriV) of other descent. But the plasmids differed in composition of the accessory region (27.0–28.3 kbp) that carried ISCR2 and several resistance genes.

However, linear regression analysis showed no correlation between relative traN gene abundance on day 63 and the estimated exposure to SDZ in bulk and rhizosphere soil until day 63 (data not shown). This indicates that besides LowGC-type plasmids also other incompatibility groups may have played an important role in the proliferation of SDZ resistance. The antibiotic resistance patterns of 10 analyzed transconjugants, which were affiliated by PCR replicon typing to the LowGC-type, IncP-1ε, and IncN plasmids, revealed a heterogeneous load of resistances to different classes of antibiotics (Table 1). Six different antibiotic resistance profiles were obtained including plasmids conferring resistance against multiple antibiotics of different substance classes like for example sulfonamides, tetracycline, and beta-lactam antibiotics. This may suggest a possible public health risk emerging from the propagation of LowGC-type plasmids, for which Acinetobacter spp., E. coli, and other Gram-negative genera were identified as potential hosts or recipients (Heuer et al., 2009), that is, a transfer of different antibiotic resistance genes to human pathogens appears possible.

View this table:

Antibiotic resistance profiles determined for 10 transconjugants harboring plasmids that were representative for restriction types

GroupAntibiotic resistance profilesIncompatibility group (number of transconjugants)
1SulfamerazineLowGC (1)
2Oxytetracycline, sulfamerazineLowGC (2)
3Chloramphenicol, oxytetracycline, sulfamerazineLowGC (3)
4Chloramphenicol, oxytetracycline, sulfamerazine, streptomycinLowGC (1), IncP-1ε (1)
5Ampicillin, oxytetracycline, sulfamerazine, streptomycinIncN (1)
6Chloramphenicol, oxytetracycline, sulfamerazine, streptomycin, trimethoprimIncP-1ε (1)
  • In addition to SDZ resistance.


The results of this study demonstrated that the application of manure from pigs treated with SDZ increased the relative abundance of the SDZ resistance genes sul1 and sul2 in bulk and rhizosphere soil of maize compared with the control, confirming results of previous soil microcosm experiments that were performed with higher concentrations of SDZ spiked to manure. Intriguingly, the increase in relative sul gene abundance was less pronounced in the rhizosphere, which coincided with an accelerated dissipation of SDZ in this soil compartment. Moreover, our results support the recently suggested important role of LowGC-type plasmids for bacterial adaptation to SDZ selective pressure in the agro-ecosystem (Heuer et al., 2009). The complete sequence recently determined for three LowGC-type plasmids pointed to the importance of co-selection as various different resistance determinants were localized on the ‘antibiotic resistance gene island’ carried by these plasmids.


C.K. and S.J. are shared first authors. C.K., H.H., I.R., and S.J. were funded by the DFG Deutsche Forschungsgemeinschaft (FOR566, SM59/5-3; SM59/12-1, and AM134/6-3). We would like to thank Ilse-Marie Jungkurth for proofreading this manuscript, the Forschungszentrum Jülich and especially Herbert Rützel for practical support in experimental setup, the Robert Koch Institute (Prof. Dr. W. Witte), Wernigerode, for performing the antibiograms as well as the colleagues of FOR566 for inspiring discussions.


  • Editor: Angela Sessitsch


View Abstract