High-throughput quantification of antibiotic resistance genes from an urban wastewater treatment plant

Antibiotic resistance among bacteria is a growing problem worldwide, and wastewater treatment plants have been considered as one of the major contributors to the dissemination of antibiotic resistance to the environment. There is a lack of comprehensive quantitative molecular data on extensive numbers of antibiotic resistance genes (ARGs) in different seasons with a sampling strategy that would cover both incoming and outgoing water together with the excess sludge that is removed from the process. In order to fill that gap we present a highly parallel quantitative analysis of ARGs and horizontal gene transfer potential over four seasons at an urban wastewater treatment plant using a high-throughput qPCR array. All analysed transposases and two-thirds of primer sets targeting ARGs were detected in the wastewater. The relative abundance of most of the genes was highest in influent and lower in effluent water and sludge. The resistance profiles of the samples cluster by sample location with a shift from raw influent through the final effluents and dried sludge to the sediments. Wastewater discharge enriched only a few genes, namely Tn25 type transposase gene and clinical class 1 integrons, in the sediment near the discharge pipe, but those enriched genes may indicate a potential for horizontal gene transfer.

sludge, sewage, qPCR

INTRODUCTION

Antibiotic resistance among bacteria has become a serious threat worldwide. Ordinary infectious diseases currently treatable with antibiotics might soon become untreatable and life threatening even in developed countries. The case will be even worse in Third World developing countries if expensive treatments are needed for common infections. By the year 2050 antimicrobial resistant infections might lead to more deaths than any other major cause (Review on Antimicrobial Resistance, 2014). The World Health Organization (WHO) has stated that a post-antibiotic era is possible if the problem is not taken seriously (World Health Organization, 2014). Of special concern are antibiotic resistance genes (ARGs) located in mobile genetic elements in pathogenic bacteria, which present a serious threat of causing untreatable infections and further dissemination of these genes (Martinez, Coque and Baquero 2014).

Urban wastewater treatment plants (UWTPs) are one of the main sources of both antibiotic resistant bacteria and resistance gene release to the environment (Munir, Wong and Xagoraraki 2011; Rizzo et al.2013; Czekalski, Gascon Diez and Bürgmann 2014). UWTPs are reservoirs and environmental suppliers of antibiotic resistance and considered to be hotspots for horizontal gene transfer (HGT) enabling even broader dissemination of antibiotic resistance genes (Rizzo et al.2013). UWTPs receive sewage from various sources, and bacteria from different environments are able to interact and exchange genes horizontally. The high bacterial densities, biofilms and stress caused by pollutant compounds, such as antibiotics and heavy metals, promote horizontal gene transfer in wastewaters (Aminov 2011). UWTPs are a unique interface between society and the environment because sewage from households and hospitals contain antibiotics and bacteria of human origin, potentially providing a selective pressure for antibiotic resistant bacteria (ARB) and ARGs (Martinez 2009). This unique environment may pose a serious threat of spread of resistance, possibly to pathogenic bacteria.

Antibiotic resistant bacteria, and even resistant pathogens, have been found from UWTPs and their effluents using culture-dependent methods (Figueira et al.2011; Munir, Wong and Xagoraraki 2011; Rizzo et al.2013). Also enrichment of resistant bacteria during the wastewater treatment process has been observed using the same methods (Harris, Cormican and Cummins 2012). Culture-dependent methods work well on known pathogens and human commensals, but molecular methods can be used to measure a high number of resistance-associated determinants in parallel.

Several studies have focused on ARGs with molecular methods, but only a few have analysed several genes at more than one time point and compared the difference between influent and effluent waters, giving only a limited view of the ARG dynamics in UWTPs. Antibiotic resistance genes are widespread in UWTPs, and resistance genes can also be found from effluents (Auerbach, Seyfried and McMahon 2007; Zhang et al.2009; Munir, Wong and Xagoraraki 2011; Rizzo et al.2013; Laht et al.2014). Metagenomic sequencing of environmental DNA from a UWTP in Hong Kong showed seasonal fluctuations of few ARG types and high abundance of several ARGs, but also the reduction of genes in the effluents (Yang et al.2013; Yang et al.2014). In addition to ARGs, genetic elements involved in horizontal gene transfer have been detected from UWTPs (Moura et al.2010; Parsley et al.2010).

One major concern is the release of ARGs from UWTPs to the environment. Wastewater release has been shown to increase the abundance of resistance genes in river sediments downstream from UWTPs (Pruden, Arabi and Storteboom 2012; Marti, Jofre and Balcazar 2013). UWTPs were the main explanation of ARG pollution in the sediments of the River Thames (Amos et al.2015), and in Switzerland sulfonamide and tetracycline resistance genes accumulated in freshwater lake sediments near the UWTP effluent discharge pipe with the abundance decreasing exponentially with distance (Czekalski, Gascon Diez and Bürgmann 2014).

Regardless of work conducted on antibiotic resistance in UWTPs, there is still a lack of comprehensive quantitative molecular data on an extensive number of ARGs in different seasons with sampling strategy that would cover both incoming and outgoing water together with the excess sludge that is removed from the process. In order to fill that gap we present here a highly parallel quantitative analysis of transposase and antibiotic resistance gene abundances from a UWTP in Helsinki, Finland over four seasons during one year. The impact of wastewater release on the environment was quantified by sampling sediment near the discharge pipe using a parallel qPCR assay that has been previously used to analyse these genes in feces from pigs fed growth-promoting antibiotics (Looft et al.2012) and from commercial swine farms (Zhu et al.2013) as well as in studying the effect of reclaimed water irrigation on park soils (Wang et al.2014). Our results show the higher abundance of antibiotic resistance and transposase genes in the raw sewage compared with effluents and dried sludge leaving the treatment plant. This UWTP with tertiary treatment system substantially decreased the gene abundance and richness. The wastewater discharge was shown to have a limited impact on the antibiotic resistance abundance in the sediment.

MATERIALS AND METHODS

Sampling and DNA extraction

The Helsinki UWTP uses a combined mechanical, chemical and biological process with a tertiary treatment (biofilters) for enhanced nitrogen removal, and has been described previously (Laht et al.2014). The UWTP uses anaerobic digestion for the activated sludge before drying. The dried sludge is composted and land applied. The UWTP does not have a disinfection step. Wastewater sampling was done as previously described for Helsinki UWTP (Laht et al.2014). Briefly, 24 h composite samples were taken from raw influent and final effluents during four seasons from summer 2010 to spring 2011. Each season was sampled on three days at 1- to 3-day intervals. Wastewater temperature was monitored automatically at the UWTP. Water samples were stored at 4°C and the samples collected on filters within 2 h after sampling. Filtration was done through 0.22 μm pore size polycarbonate filters (GE Water & Process Technologies). Three technical replicates were filtered from all samples. Samples from dried sludge were collected on the same days by the UWTP personnel. Surface sediment samples near the discharge site were collected using a Limnos sediment probe (Limnos, Ltd, Turku, Finland) at a single time point in summer 2011. Control sediments were collected >200 km away from the discharge site using the same device. The control sites did not have any clear impact from human activities (e.g. wastewater discharge or agricultural run-offs). There were three biological replicates for the discharge site and five for the control site. All sediment samples and filters were stored at –80°C before DNA extraction.

DNA was extracted from water samples using MoBio PowerWater DNA isolation kit (MoBio Laboratories Inc., CA, USA). From sediment and dried sludge samples three technical replicate DNA extractions were done with FastDNA spin kit for soil (MP Biomedicals, Illkrich, France). Extracted DNA was stored at –20°C prior to analysis. All biological and technical replicates were pooled before the qPCR analysis. Replicates were pooled to reduce variations in extraction efficiencies and sampling. Biological and extraction replicates were not barcoded, and thus could not be differentiated on the qPCR array. Altogether 20 samples were analysed. There were four samples (seasons) from each raw influent, final effluent and dried sludge. From release site sediments there were three and from controls sediments five samples.

Quantitative PCR

All quantitative PCR reactions were performed using the WaferGen SmartChip Real-time PCR system, as described previously (Wang et al.2014), except that a threshold cycle (T_C) of 27 was used as the detection limit as in other studies (Looft et al.2012; Zhu et al.2013). Briefly, 5184 parallel 100 nl reactions were dispensed into the SmartChip using the SmartChip Multisample Nanodispenser. Each sample–primer set combination was tested with three technical replicates on the array. PCR cycling conditions and initial data processing were according to the procedure in Wang et al. (2014). Average C_T, standard deviations and fold change were all calculated using Microsoft Excel 2008. Genes detected in only one of the three technical replicates were considered false positives and were removed. ΔC_t calculations were done as previously described (Livak and Schmittgen 2001; Zhu et al.2013).

Data analysis

All further data analysis was done using R version 3.1.1 (R Core Team 2014). Gene quantification was relative to the 16S rRNA gene abundance in the samples. Relative gene abundances (R) were calculated from the ΔC_t values with R = 2^−ΔCt and log10 transformed for the statistical analyses. Fold changes were calculated from the ΔΔC_t values as described previously (Livak & Schmittgen 2001). Genes under detection limit were given a ΔC_t value of 15, which was higher than any observed ΔC_t (14.9) for detected genes in the samples. The resistance genes were grouped into nine classes based on the antibiotic to which they confer resistance, and the mean relative abundance and proportion of positive assays were calculated for each class. To study the statistical significance of the differences in the samples, the fold changes were modeled using linear mixed effect models (LMM). Linear mixed effect models were fitted to the data using function lmer from the lme4 package version 1.1 (Bates et al.2015). Overall relative gene abundance was modeled using the sampling site as fixed effect and the seasons and the genes as random effects. Single genes were modeled using the sampling site as fixed effect and the season as random effect. We used the likelihood ratio test to test the significance of the model compared with an intercept only model (see Supplementary Methods available at FEMSEC online). Models were checked with residual plots and normal probability plots using the raw residuals. P-values were corrected for multiple testing with the false discovery rate method using the R function p. adjust. Multidimensional scaling was done using function cmdscale from the vegan package version 2.2 (Oksanen et al.2015) with Bray–Curtis dissimilarity index (see Supplementary Methods).

RESULTS

A highly parallel qPCR array of 285 primer sets targeting antibiotic resistance genes (ARGs) and nine transposase genes related to horizontal gene transfer (HGT) was used with samples collected from an urban wastewater treatment plant (UWTP). Genes were quantified over four seasons during one year to examine possible seasonal variations in gene abundances. In addition, the long-term effect of wastewater release on the sediments near the treated wastewater release pipe was analysed and the results compared with control sediments (about 200 km away) and without direct impact of wastewater release. All nine primer sets targeting transposase genes and 175 of the 285 primer sets targeting ARGs were detected in the UWTP (Supplementary Table S1). The raw influent had higher abundance of total ARGs compared with both final effluent and sludge (P < 0.001 for both, LMM, see Material and methods).

The resistance profiles of the samples cluster by sample location and there is a shift from raw influent through the final effluents and dried sludge to the sediments (Fig. 1). The shift detected is in line with the expectation that raw influent has higher ARG abundance than final effluents and dried sludge. The different seasons show only minor differences in the ARG abundance, which indicates that the ARG load in the wastewater and the UWTP process are quite stable through the year (Supplementary Fig. S1). The wastewater temperature at the UWTP remained relatively stable during the study period even though the outside temperature varied at the sampling times from –19°C in the winter to 21°C in the summer. The daily average temperatures of the wastewater varied from 12°C in the spring to 18°C in the autumn season. The wastewater temperatures in the summer and winter remained stable and were 14–15°C.

Figure 1.

Principle coordinates analysis (PCoA) ordination plot from the antibiotic resistance gene (ARG) abundance profiles showing the clustering of different sample types and the change from raw influent to the final effluents and dried sludge. The release site sediments did not differ substantially from the control sediments based on the ARG profile. ARG results were normalized to relative abundances with 16S rRNA gene copy numbers. Bray–Curtis dissimilarity index was used in the multidimensional scaling plot. Ellipses were drawn with 95 % confidence. Colors indicate UWTP sampling seasons; summer green, autumn brown, winter blue, spring yellow. Sediment samples (3 discharge and 5 control site) are shown in black.

To get more a detailed view on the ARG dynamics in the wastewater, the ARGs were grouped into nine classes based on the mechanism by which they confer resistance. In almost all classes the raw influent showed higher abundance and gene richness than final effluent or sludge. The MLSB (macrolide, lincosamide and streptogramin B resistance) class genes were more abundant in the sludge whereas in the final effluent resistance to the sulfa drug class was relatively more abundant. The number of genes detected (richness) in both classes was higher, however, in the raw influent (Fig. 2).

Figure 2.

The mean ARG abundance (A) and proportion of positive assays (B) over all seasons in each ARG class in the UWTP. ARG results were normalized to relative abundances with 16S rRNA gene copy numbers. The positive assays are presented as proportions due to the different number of genes in different classes. Error bars show the standard error. Results are calculated from four raw influent, four final effluent and four dried sludge samples.

There was no clear enrichment of ARGs or HGT genes (Fig. 3), indicating that the Helsinki UWTP largely removes antibiotic resistances when individual genes are considered. From the 180 genes found in the raw influent, 77 and 111 had decreased to below the detection limit in the final effluents and dried sludge, respectively. Sixty genes were persistent in the UWTP process and could be found from all three locations. There were four genes that were not detected in the raw influent, but were found either in the final effluents or in the dried sludge (Supplementary Fig. S2). To confirm the decrease in the amount of ARGs, we calculated abundance fold changes including also the less abundant genes from raw influent to final effluent and sludge (Supplementary Fig. S3). The fold changes were small and in both cases most of the genes decreased from raw influent to final effluents or dried sludge. The highest enrichment was a 25-fold increase of the erythromycin resistance gene, ermF, in the sludge (P < 0.001). The ermF enrichment explains the MLSB class level enrichment in the sludge. In the sludge we also saw 20-fold (P < 0.001) and 8-fold (P < 0.05) enrichment of tetracycline resistance genes tetPB and tetPA, respectively, and almost 4-fold enrichment of related tetracycline resistance regulator gene tetR (P < 0.001) (Supplementary Fig. S4).

Figure 3.

Relative abundance of the most abundant resistance genes in the UWTP. Samples were clustered with hierarchical clustering based on Bray–Curtis dissimilarities. ARG results in the heatmap were normalized to relative abundances with 16S rRNA gene copy numbers. Genes with maximum abundance >0.05 were selected.

There was high abundance of almost all analysed transposases in the raw influent (Fig. 3). The high abundance of transposases in the raw influent might indicate an HGT potential in the wastewater. Tn25 type of transposase gene (tnpA-04) was not abundant in the raw influent, but was enriched in the final effluents and sludge by almost 3-fold (P < 0.01) and 2.5-fold (P < 0.01), respectively.

There are only minor differences between the wastewater release site and control sediments on the ARG class level (Fig. 4), indicating that the release of purified wastewater has only a limited effect on the abundance and richness of resistance genes in the sediment. A heatmap of the most abundant genes shows clearly that three genes, tnpA-04, ermB and qacEΔ1, were enriched in the sediments (Fig. 5). Tn25 type transposase gene (tnpA-04) was also enriched in the final effluents, which might explain its enrichment in the sediments.

Figure 4.

The mean ARG abundance (A) and proportion of positive assays (B) in each ARG class in the sediments. ARG results were normalized to relative abundances with 16S rRNA gene copy numbers. The positive assays are presented as proportions due to the different number of genes in different classes. Error bars show the standard error. Results are calculated from three discharge site and five control site sediment samples.

Figure 5.

Relative abundance of the most abundant resistance genes in the sediments. Samples were clustered with hierarchical clustering based on Bray–Curtis dissimilarities. ARG results in the heatmap were normalized to relative abundances with 16S rRNA gene copy numbers. Genes with maximum abundance >0.005 were selected.

DISCUSSION

We analysed almost 300 primer sets with a highly parallel qPCR array and sampled four seasons covering one year from the urban wastewater treatment plant with tertiary treatment in Helsinki, Finland. We found high abundance and diversity of antibiotic resistance and horizontal gene transfer genes in the raw sewage. All quantified transposases and two-thirds of the ARGs assayed for were detected showing a high level of resistance in the sewage that can pose a risk of resistance gene dissemination.

When analysing the risk of antibiotic resistance gene spread to the environment, a reduction in abundance and richness of resistance genes was seen from raw influent to effluent waters and dried sludge. Helsinki UWTP reduces the amount of bacteria by two to three orders of magnitude from raw influent to final effluents (Laht et al.2014), so when relative abundance of resistance genes does not increase, there is a more than 99% reduction of ARGs in the effluents released to the environment. Yang et al. (2014) observed that most of the antibiotic resistance genes were removed from UWTP influent with more than 98% reduction in the effluent by analysing antibiotic resistance genes with metagenomic sequencing.

There is not only a reduction in the absolute abundance due to reduction in bacterial numbers, but also in relative abundances of all resistance gene classes, and this phenomenon is constant through all four season. Nine resistance genes (Supplementary Fig. S4) were enriched in the effluent community, showing that some genes can enrich in the process. A reduction of resistance genes during wastewater purification has been shown (Zhang et al.2009; Munir, Wong and Xagoraraki 2011; Laht et al.2014), but studies have also shown that there is no change in the relative numbers or the numbers have increased (Auerbach, Seyfried and McMahon 2007; Laht et al.2014). In a meta-analysis of antibiotic resistance in UWTPs, Harris, Cormican and Cummins (2012) concluded that the proportion of antibiotic resistant bacteria increases during the purification process. The study by Laht et al. (2014) shows that there can also be differences between different UWTPs, possibly due to the differences in the purification process or operation. Reasons for enrichment of certain genes might be selection pressure favoring the resistance gene, conditions favoring the bacteria carrying the gene or horizontal spread of the gene in the community. Results from this work and others call for extended studies on several UWTPs with differential processes to better understand the ARG dynamics and process efficiency in removing ARGs.

UWTP produces effluent waters and sludge from settled biosolids that are land applied after composting. The most enriched genes in the dried sludge were ermF from the MLSB resistance gene class and two tetracycline resistance genes, tetPA and tetPB (Supplementary Fig. S4). Yang et al. (2014) found enrichment of genes conferring resistance to polymyxin, tetracycline, vancomycin and MLSB class antibiotics in UWTP sludge. Munir and colleagues (2011) concluded that UWTP sludge was the main source of tetracycline and sulfonamide resistant bacteria and genes released into the environment. These results combined show the importance of also considering UWTP sludge in assessing the threat of antibiotic resistance gene release into the environment. Often only the raw influent and final effluents are screened for genes or resistant bacteria.

In the sediments near the release pipe, three genes were enriched compared with the control sediments, but there were no differences in resistance classes. The natural background resistance is even higher for some ARG classes (Fig. 4A). The genes enriched in the release site sediment compared with the control sea sediments were erythromycin resistance gene ermB, Tn25 type transposase gene (tnpA-04) and quartenary ammonium resistance gene qacEΔ1 (Fig. 5). Two different primer pairs targeting qacEΔ1 both showed its enrichment. A study using the same qPCR array found tnpA-04 gene to be the most enriched transposase gene in park soils irrigated with reclaimed water. The gene was enriched together with ermB gene (Wang et al.2014). These genes can be found together from two different E. coli plasmids (accession numbers: CP009860 and FQ482074). This might indicate that these genes are co-selected in these samples. Another study using normal qPCR found erythromycin resistance gene ermB and tetracycline resistance genes tetO and tetW to be enriched in river sediment downstream from a UWTP (Marti, Jofre and Balcazar 2013). Clinical class 1 integrons have been proposed as a marker for anthropogenic pollution (Gillings et al. 2015) and qacEΔ1 gene belongs to the backbone of clinical class 1 integrons (Gillings et al.2008; Gaze et al.2011; Wellington et al.2013). The detection of qacEΔ1 gene is an indication of clinical class 1 integron presence in the release site sediments and supports the use of clinical class 1 integrons as a marker gene for anthropogenic pollution. However, class 1 integrons should not be used as universal markers for ARG enrichment due to the different genetic context of different AR genes. Based on single time point comparison of discharge site sediments to control site sediments without a strong anthropogenic impact, the results show that wastewater only slightly increases the overall incidence of antibiotic resistance in the sediments.

We showed that raw sewage entering a UWTP is rich in antibiotic resistance and transposase genes possibly posing a threat of spread and release of resistance genes to the environment. But this study also shows that an urban wastewater treatment plant with tertiary treatment can work efficiently in reducing ARG contamination from sewage. The final effluents contain far fewer ARGs than influents and do not seem to be an important source of ARG contamination, since only three genes were enriched in sediments near the discharge pipe. The sludge can play a significant part in environmental ARG contamination and should not be overlooked in UWTP studies. Our results give broader insight to the ARG dynamics in wastewaters and this kind of array will be used in future studies to examine more extensively and quantitatively the different UWTP designs and their operation for maximizing the effectiveness in antibiotic resistance gene removal.

SUPPLEMENTARY DATA

Supplementary Data.

The authors want to acknowledge the staff from HSY at Helsinki UWTP for providing the samples. This work has been financed through the project funded by Academy of Finland, in the frame of the collaborative international consortium StARE of the Water Challenges for a Changing World Joint Programming Initiative (Water JPI) Pilot Call. All authors declare they have no actual or potential competing financial interest.

Conflict of interest. None declared.