Antibiotics are commonly used in aquaculture and they can change the environmental resistome by increasing antibiotic resistance genes (ARGs). Sediment samples were collected from two fish farms located in the Northern Baltic Sea, Finland, and from a site outside the farms (control). The sediment resistome was assessed by using a highly parallel qPCR array containing 295 primer sets to detect ARGs, mobile genetic elements and the 16S rRNA gene. The fish farm resistomes were enriched in transposon and integron associated genes and in ARGs encoding resistance to antibiotics which had been used to treat fish at the farms. Aminoglycoside resistance genes were also enriched in the farm sediments despite the farms not having used aminoglycosides. In contrast, the total relative abundance values of ARGs were higher in the control sediment resistome and they were mainly genes encoding efflux pumps followed by beta-lactam resistance genes, which are found intrinsically in many bacteria. This suggests that there is a natural Baltic sediment resistome. The resistome associated with fish farms can be from native ARGs enriched by antibiotic use at the farms and/or from ARGs and mobile elements that have been introduced by fish farming.
INTRODUCTION
Aquaculture farms have been suggested to be hotspots for antibiotic resistance gene (ARG) enrichment and transfer due to prophylactic and therapeutic use of antibiotics to treat fish infections at the farms (Cabello et al.2013). Oxytetracycline, a sulfonamide-trimethoprim combination or florfenicol is used to treat fish infections and is also important for human medicine (FAO 2006; Heuer et al.2009). Fish farming may also cause enrichment of mobile genetic elements (MGEs) that carry ARGs and can thus pose a risk of transferring ARGs to other bacteria and in some cases even to clinically relevant bacteria (Cabello et al.2013). It is therefore crucial to investigate the ARGs and associated MGEs in the aquaculture environment to help evaluate the risks to human health (Sapkota et al.2008; Heuer et al.2009; Cabello et al.2013).
Coastal aquaculture farms often use an open cage system, which allows free transfer of water from the farms to the surrounding water and eventually to the sediment (FAO 2007). Therefore the introduced antibiotics may cause the surrounding water and sediments becoming reservoirs of ARGs due to selection pressure (Cabello et al.2013). Reports on the diversity and distribution of ARGs in the sediments below fish farms are typically based on cultivation experiments (Buschmann et al.2012; Yang et al.2013), which might underestimate the resistance potential present in the sediments (Amann, Ludwig and Schleifer 1995; Buschmann et al.2012). ARGs are ubiquitous in the environment (Martinez, Coque and Baquero 2015) and the background of naturally occurring ARGs is diverse in sea sediments (Chen et al.2013).
Highly parallel qPCR arrays currently provide a culture-independent method that permits hundreds of assays to detect and quantify specific ARGs in a single experiment. The method has recently been used to analyze ARG profiles in different environments (Looft et al.2012; Zhu et al.2013; Wang et al.2014), but has not been previously used to study fish farms. We used a qPCR array with 285 primer sets to target ARGs, 9 primer sets targeting transposase alleles and a primer set for the 16S rRNA gene to study gene profiles in sediments below two fish farms and a control site outside the farm influence in the northern Baltic Sea. Fish farming impacts the diversity and distribution of ARGs and transposons in the sediments. By screening a wide range of ARGs, our study also revealed a background ARG resistome in the Baltic Sea sediments.
MATERIALS AND METHODS
Study sites and sediment sampling
The study was carried out at two fish farms (FIN1 and FIN2), which are separated geographically by tens of kilometers, and at a site 1000 m from the FIN1 farm in the Turku archipelago, Finland. The study sites are located in the northern Baltic Sea, which has brackish water (mean salinity: 6.7 parts per thousand). Average water temperature and pH was 13.1°C (+/− SD 1.2°C) and 6.8 (+/− SD 0.5), respectively, during the study times, and average depth was 7.5 m (+/− SD 1.3 m).
Both FIN1 and FIN2 farms use open and floating cage systems that allow free transfer of uneaten fish feeds and fish excrement from the cages to surrounding waters and eventually to sediments. The farms raise European whitefish, Coregonus lavaretus (Linnaeus) and rainbow trout, Oncorhynchus mykiss (Walbaum). Each farm produces approximately 50 tons of fish annually. Detailed records of antibiotics used at the fish farms were not available, however, both farms have used tetracycline, sulfonamide, trimethoprim and florfenicol during past decades. Sediment samples were taken in three biological replicates from the FIN1 farm and from the FIN2 farm below fish cages during summer 2012. In addition, two biological replicates were taken from 1000 m outside the FIN1 farm (Out_12B and Out_12C) during the same sampling campaign, and were used as control sediment samples. Additional control samples collected during the summers of 2008 and 2009 (Out_08 and Out_09), were used to increase the extent of sampling for background levels of ARGs. The three biological replicates of Out_08 and Out_09 were pooled for each year, thus providing a total of four control sediment samples. Sediment samples were collected by sampling the surface (3–10 cm depth) using a Limnos sediment probe (Limnos Ltd, 100 Turku, Finland). Each sample was homogenized manually inside a zipper storage plastic bag and immediately frozen on dry ice. The sediments were stored at –80°C until DNA extraction.
DNA extraction
The total environmental DNA was extracted from 0.5 g wet weight of sediment, using the FastDNA® SPIN kit for soil (MP Biomedicals, Illkrich, France). The standard protocol was modified by adding an extra washing step with 5.5 M of guanidine thiocyanate (Sigma Life Science, Steinheim, Germany), according to the manufacturer's instructions for removing humic acids. The DNA quality and concentration were analyzed with a Nanodrop 1000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA).
qPCR array
qPCR reactions were performed using 295 primer sets in the WaferGen SmartChip Real-time PCR system, as described previously (Wang et al.2014), and a threshold cycle (CT) of 27 was used as the detection limit (Zhu et al.2013). Melting curve analysis was performed on all of the samples for each primer set. Amplicons with unspecific melting curves and multiple peaks based on the slope of melting profile were considered to be false positives and discarded from the analysis. Several primer sets were also designed to target sequence diversity within an ARG to more specifically assess the environmental resistome. Therefore each primer set was analyzed independently to maintain the most specific data possible for all ARGs.
Briefly, the SmartChip has 5184 reaction wells with a volume of 100 nL and filled using the SmartChip Multisample Nanodispenser. PCR cycling conditions and initial data processing was done as previously described (Wang et al.2014). Mean CT of three technical replicates in each qPCR assay was used to calculate the ΔCT values, unless, the genes were detected in only one of the three technical replicates, in which case they were considered false positives and were removed. The 2−ΔCT method (Schmittgen and Livak 2008) where ΔCT = (CT detected gene – CT 16S rRNA gene) was used to calculate the relative abundances of the detected gene in proportion to the 16S rRNA gene in each sediment sample.
Data analysis
Permutational multivariate analysis of variance from the Vegan package in R (Oksanen et al.2014) was performed to determine the differences between ARG profiles in fish farm sediments and the sediments outside the farms. Relative abundance values in proportion to the 16S rRNA gene (Table S1, Supporting Information) were used to calculate the distances between sediment samples using function vegdist from Vegan based on Bray–Curtis method. Data values below detection limits were replaced with zero for the data analysis. Function cmdscale and ordiellipse from Vegan were respectively used to plot the distance matrix and to add 95% confidence region of two groups: the farm sediments and sediments outside the farms. The permutational multivariate analysis of variance was performed using function adonis from Vegan and default parameters with 9999 permutations to test whether the two groups were significantly different. The analysis was considered to be significant at P-values < 0.05. All analyses were performed using RStudio v.0.98.501 (RStudio, Boston, MA).
RESULTS
Gene occurrence in the fish farm and outside sediments
Five of the transposase genes and 66 of the ARGs were detected using qPCR assay with 285 primer sets for the detection and quantification of ARGs and 9 for transposase alleles in the sediment (Fig. 1A; Table S2, Supporting Information). Of the detected 71 genes, 31 were found in all of the sediment samples (Table S1, Supporting Information). A total of 55 ARGs and 4 transposase genes were detected at the FIN1 farm, 46 ARGs and 5 transposase genes at the FIN2 farm, and 45 ARGs and 2 transposase genes outside the farms (Table S2, Supporting Information). The detected ARGs covered resistances to most antibiotics and the three major antibiotic resistance mechanisms: cellular protection, antibiotic deactivation and active efflux (Table S1 and 2, Supporting Information).
The 66 ARGs and 5 transposase genes detected in the sediments at the FIN1 farm (dark green), FIN2 farm (green) and outside the farms (black). The ARGs were grouped based on the mechanism of resistance and the classification of the antibiotics that they confer resistance to: (A) percentage of positive qPCR assays designed to target each group of ARGs and transposons in proportion to the total assays; (B) total relative abundance of genes in proportion to the 16S rRNA gene in each sediment sample (as log values). The box plot presents the biological replicates at FIN1 farm (n = 3), FIN2 farm (n = 3) and outside the farms (n = 4). ND indicates the genes not detected in the sediments.
The gene profile detected in the sediments at the two farms were different from the outside sediments (Fig. 2) and significantly different based on permutational multivariate analysis of variance (R2 = 0.62, P-value < 0.01).
Principal Coordinate Analysis (PCoA) of genes’ relative abundance detected in the sediments below two fish farms, FIN1 (square-dark green) and FIN2 (square-green), and outside the farms (circle-black) based on Bray–Curtis distance. Red circle line shows 95% confidence regions of the two farms and the outside sediments. The distribution of the genes’ relative abundance detected in the farm sediments is significantly different from the sediments outside the farms (permutational multivariate analysis of variance R2 = 0.62; P-value < 0.01).
ARG and transposase abundance in the fish farm and outside sediments
ARGs conferring resistance to tetracycline, sulfonamide and trimethoprim were enriched in the farm sediments compared to the outside sediments (Fig. 1B; Table S3, Supporting Information). The tetracycline resistance genes enriched at the farms belonged to two resistance classes: ribosomal protection proteins (tet(32), tetM, tetO, tetS and tetW) and efflux pumps (tetA, tetE, tetG and tetH). Resistance genes sul2 of sulfonamide resistance gene and dfrA1 of trimethoprim resistance gene were also enriched in fish farm sediment (Fig. 3).
Gene profiles of ARGs and transposase genes detected in the northern Baltic Sea sediments. Y-axis presents the assayed ARGs grouped by antibiotics and transposons. X-axis presents the sampling locations organized by the sediment sample type. The sediments outside the farms are represented in black and the sediments below the two fish farms, FIN1 in dark green and FIN2 in green. The color scale indicates the relative abundance of the detected genes in proportion to the 16S rRNA gene. White indicates that the respective gene was not detected or below the detection limit of each assay (CT cut-off was at cycle 27) in the qPCR array.
ARGs encoding aminoglycoside resistance, aadA, aadA1, aadA2 and strB were the only other ARGs enriched at the farms, despite aminoglycosides not being used at the fish farms (Fig. 3). MGE associated genes, tnpA genes encoding transposon associated genes and qacEΔ1, a marker gene of clinical class 1 integrons also were enriched at the fish farms compared to the outside sediments (Fig. 3; Table S1, Supporting Information).
Genes, mexF, oprD, oprJ, pncA, yecLblaOXY, blaCTX, blaSHV, acrA, vanC and aacC were ubiquitous in the sediment samples with relative abundance ca. 10−4–10−2 of the 16S rRNA gene (Fig. 3). The beta-lactam, macrolide, vancomycin and quinolone resistance genes’ relative abundance in proportion to the 16S rRNA gene was higher in the control sediments (outside the farms) (Figs 1B and 3). Also, the total overall relative abundance of all the ARGs in proportion to 16S rRNA gene was higher in the sediments outside the farms compared sediments at the fish farms (Fig. 1B; Table S3, Supporting Information).
DISCUSSION
A qPCR array consisting of 295 primer sets detected a wide range of ARGs occurring in the sediment resistome associated with fish farming in the Northern Baltic Sea, Finland. Altogether 71 different ARGs and transposases were detected and quantified. Previous studies using standard qPCR have investigated only 15 or fewer ARGs for the antibiotics used at aquaculture farms (Tamminen et al.2011b; Gao et al.2012; Muziasari et al.2014; Xiong et al.2015). However, the diversity of ARGs found at fish farms is much more complex, since other ARGs can also occur in these environments (Yang et al.2013). Sediment bacterial communities are known to contain a multitude of different resistance genes even without anthropogenic antibiotic input (Chen et al.2013). The previous qPCR studies have yet to investigate the ARGs for antibiotics that have not been used in fish farming and thus have not estimated the natural resistome that is also present in the farm sediments.
The ARG profiles of the sediments collected outside the fish farms were significantly different from the ARG profiles of both FIN1 and FIN2 farms based on multivariate analysis. The farms’ resistance gene profiles were similar, although the farms are geographically separated by tens of kilometers. The observed shift in the resistance gene profiles between the farms and outside the farms is therefore likely due to fish farming, possibly partly because the bacterial community changes in response to the farming. Actinobacteria, Chloroflexi, and Firmicutes were prominent in the farm sediments whereas in the outside sediments the prominent phyla included Alphaproteobacteria, Cyanobacteria, Deltaproteobacteria and Verrucomicrobia (Tamminen et al. 2011a). The impact of fish farming on the sediment bacterial community's resistance profiles compared to control sediments has also been studied previously using culturing and antibiotic sensitivity testing (Chelossi et al.2003; Buschmann et al.2012). In this study, the culture independent method used gives a more comprehensive and quantitative assessment of the change in the ARG profiles in the sediments caused by fish farming and hence compliments the previous studies.
We observed a high abundance and diversity of ARGs even in sediments that were collected outside the fish farms and are not considerably affected by municipal or agricultural activities. Total relative abundance of resistance genes outside the farms was in fact relatively higher compared to that of the two studied fish farms, FIN1 and FIN2. It seems that in the Baltic Sea sediments, there is a background resistome consisting mainly of genes encoding multidrug or efflux pumps conferring resistance to antibiotics, which has not been reported previously in brackish water sediment. Efflux pump resistance genes are known to be dominant in the background sea sediment resistome (Chen et al.2013). ARGs are commonly found in environmental bacteria without any record of antibiotic exposure (Bhullar et al.2012; Segawa et al.2013) and have other functions besides antibiotic resistance, such as increasing fitness (Martinez et al.2009). For example, mexF, which encodes an efflux protein conferring resistance to chloramphenicol, was the most abundant gene found in every Baltic Sea sediment sample studied with relative abundance ca. 10−3–10−2 of the 16S rRNA gene. mexF confers the capability not only to transport out antibiotics, but also other compounds, such as humic acids, which are commonly present in the sediment (Groh et al.2007).
The fish farming process alters the bacterial community in the farm sediments possibly due to the selection for bacteria harboring a suite of ARGs relevant to antibiotics used at the farms. Indeed, we observe that aquaculture enriches ARGs associated with antibiotics used in fish farming (tetracycline, sulfonamide and trimethoprim) in a reproducible manner in both FIN1 and FIN2 farms. Both farms are still using antibiotics occasionally and have been operating with antibiotic use for over two decades (Tamminen et al.2011b). The enrichment of ARGs for antibiotics used has been described previously (Tamminen et al.2011b; Muziasari et al.2014). However, the sulfonamide, tetracycline and trimethoprim resistance genes were mostly not detected in the outside sediments despite otherwise high abundance and diversity of potential ARGs, and thus it is likely that these resistance genes were introduced to the sediment due to fish farming using antibiotics. This study and our previous studies show a similar and local impact of fish farming on the diversity and distribution of ARGs in the northern Baltic Sea sediments.
Beside the ARGs that encode the resistance to used antibiotics, aadA, aadA1, aadA2 and strB genes encoding aminoglycoside resistance were also ARGs enriched in the fish farm sediments. Aminoglycosides are not used in fish farming but are commonly used in human medicine. The aadA gene is found commonly in gene cassettes of integrons (Partridge et al.2009) also the strB gene (Sundin 2002) which can be transferred to fish farm environments from clinical environments (Gaze et al.2011). Thus, the enrichment of the aadA and strB might be due to enrichment of clinical integrons such as class 1 and class 2 integrons which are known to carry these genes (Roe, Veda and Pillai 2003). The presence of strB could also be explained by other MGEs such as IncQ or Tn5393 (Sundin and Bender 1996). Quaternary ammonium compound resistance gene, qacEΔ1, was also enriched in the farms although the farms have no record of quaternary ammonium compound use. qacEΔ1 and sul1 genes are often found in the 3’ conserved region of clinical class 1 integrons (Partridge et al.2009; Gaze et al.2011). The enrichment of the sul1 and intI1 integrase gene of class 1 integrons in the two Baltic farm sediments has been reported previously (Muziasari et al.2014). The primers used in this study target at different allele of the sul1 gene and therefore did not amplify sul1 gene allele found in the farm sediments. It is likely that fish farming has enriched clinical class 1 integrons carrying aadA, qacEΔ1 and sul1 genes because of co-selection caused by antibiotics such as sulfonamides (Gaze et al.2011). Therefore, the possibility of co-selecting resistance determinants and MGEs encoding resistance for antibiotics not used in fish farming should also be considered as a potential risk.
In addition to the enrichment of integron associated genes, also tnpA (transposase genes) associated with transposons were 10-fold more abundant in the fish farm sediments compared to the outside sediments. This is particularly notable since integron and transposon associated genes were mostly undetected in the sediments outside the farms. Our study supports the finding that anthropogenic activity causes the enrichment of MGEs in the environment (Gaze et al.2011). Estimating the likelihood of MGEs mobilizing ARGs found in the natural resistome is not simple. However, insight of the resistance potential present in the fish farming sediments is valuable, since there might be a risk of the ARGs spreading to other environments. Certain ARGs have in fact first been detected in water or aquaculture related environments from which they have spread to clinical settings when the genes have been captured in MGEs and horizontally transferred to human pathogens (Sorum 1998; Sapkota et al.2008).
The specific cause of enrichment of ARGs and integron and transposon associated genes in the fish farm sediment has yet to be elucidated. There are many causes of the enrichment of ARGs in the environment, e.g. selection pressure caused by antibiotic use. However, the concentration of oxytetracycline, tetracycline, sulfamethoxazole, sulfadiazine and trimethoprim is very low (1–100 ng g−1) in these farm sediments (Tamminen et al.2011b; Muziasari et al.2014). However, the presence of sub-inhibitory concentrations of antibiotics can also play a role in the enrichment of ARGs by causing selection pressure (Gulberg et al.2011). Sub-inhibitory concentrations of antibiotics increase horizontal gene transfer by activating recombinases such as integrases and transposases (Andersson and Hughes 2014) and mediate the spread of ARGs (Knapp et al.2008). ARG selection can also occur in the fish intestine during antibiotic treatment (Giraud et al.2006). Alternatively, selection can be caused independent of antibiotics by co-selection with other chemicals such as heavy metals (Liebert, Hall and Summers 1999). Mercury is often enriched in fish farms due to mercury found in fish feed (Mente et al.2006) and could also possibly contribute to the enrichment of some resistance genes caused by fish farming. Mercury resistance genes are known to be enriched in the studied fish farm sediments (Pitkanen et al.2011) and thus may partly explain the enrichment of some ARGs.
CONCLUSION
This study shows that fish farms impact the diversity and distribution of ARGs in the sediments and the enrichment at the farms is mainly limited to the ARGs associated with the antibiotics which are used at the farms. In addition, the presence and enrichment of transposons and integrons in the farm sediments may potentially lead to co-selection other ARGs and facilitate the transfer of ARGs to other bacteria, including human pathogens. This study also establishes that there is a natural environmental resistome in the northern Baltic Sea sediments. It is important to determine the sources of ARGs and to use practices that minimize their introduction into the environment, especially the ARGs in MGEs.
SUPPLEMENTARY DATA
The authors thank Johanna Muurinen and Harri Urponen for technical support on sampling at the fish farms and Dr Hanna Sinkko for assisting with statistical analysis.
FUNDING
The research was funded by and grants and with support from the and the .
Conflict of interest. None declared.
REFERENCES
