Endocrine disrupting chemicals (EDCs) interfere with the endocrine system of humans and wildlife, influencing sexual development even at sub-lethal concentrations (in the ng/L range) and could generate alterations at population level including local extinction (Kidd et al., 2007). Within EDCs, estrogenic chemicals or xenoestrogens can generate oxidative stress and feminization in wildlife, and induce proliferation of estradiol dependent carcinomas in human (Ayoola et al., 2011; García-Alonso et al., 2011a; Kabir et al., 2015). Within estrogenic substances we can find compounds with different structures, including natural substances and
a large number of synthetic compounds such as pesticides, surfactants, plasticizers, synthetic hormones and trace metals (De Coster and Van Larebeke, 2012). These chemicals are introduced into the aquatic environment by domestic, agriculture or industrial human activities and might not be removed from the municipal wastewater and their consequences had become a concern (Zhang et al., 2011). The precipitation and accumulation of these compounds in sediments increases the risk of exposure of organisms to cocktails of xenoestrogens (Wang et al., 2011) which may have additive or synergistic effects (Norris and Carr, 2006; Frische et al., 2009). For this reason is common to detect xenoestrogens in sediments of aquatic ecosystems with different land uses (García-Alonso et al., 2011b; Schnell et al., 2013; Wu et al., 2015).
Bioassays based on yeast strains such the Yeast Estrogen Screen (YES) developed by Routledge and Sumpter (1996) are simple, easy handling and low costs non invasive tool for the determination of estrogenic activity (Rehmann et al., 1999). The YES assay is used to evaluate the estrogenic activity in waters and wastewaters (Brix et al., 2010; Li et al., 2014).
Uruguayan water bodies present an actual degradation 94 of their quality parameters, most of them related to eutrophication processess (Bonilla et al., 2015) principally due to agricultural intensification, dairy production and feedlots and low efficiency in sewage treatment plants. In addition, no regulation in production, commercialization and use of already known EDCs exist in Uruguay. Therefore, there is an urgent need to analyze endocrine toxic responses that allow knowing the baseline of the estrogenicity of substances in water systems. The aim of this study was determine the estrogenicity and citotoxicity of sediments of the Santa Lucia River using the in vitro YES assay and analyze the association with land uses and natural and impact markers of the basin. Our hypothesis is that estrogenicity along the Santa Lucia River sediment is associated with different land uses (urban-industrial, agricultural and rangelands), since diffuse and multiple point sources of xenoestrogens occurs from different land uses.
2 Materials and Methods
2.1 Study area and sampling points
The Santa Lucia River basin is located in the South of Uruguay (33o41’S; 54o59’W), comprise an area of 13448 km2 112 with a maximum altitude of 250 meters above sea level (Figure 1). The basin drains into the estuary of the Río de la Plata and is a key estuarine site for conservation, since several estuarine and marine species including commercial and full exploited resources such as sciaenid fishes, spawn in down-stream waters of this river (Jaureguizar et al., 2003; Vizziano et al., 2002). The watershed provides drinking water to more than 60% of Uruguayan besides being used as a water resource for irrigation in 6 administrative regions (Departments) (Achkar et al., 2004). Land uses in the catchment area of the basin include: cattle (71.3%), agriculture (16.2%), forestry (4.2%) and finally the urban-industrial (1.1%) (Achkar et al., 2012; DINAMA, 2011; Inda and Indarte, 2010). The coexistence of these different land uses in the watershed has led to the potential exposition for wildlife, farms and humans of a wide range of toxic substances, with clear effects in fish community (Benejam et al., 2016; Teixeira de Mello, 2007). Fourty two sampling points were selected covering a wide range of land uses and of heterogeneity of landscapes (Figure 1). Collection of samples were carried out in winter of 2014. The drainage area was estimated from a digital elevation model obtained from Nasa Shuttle Radar Topographic Mission (SRTM) data (Jarvis et al., 2008), using GRASS function ‘r.watershed’ in QGIS (Quantum GIS Development Team, 2015). Main land use was estimated defining the area of land use by drawing polygons using Google Earth Pro free package. Land use was defined according four principals categories: Agriculture, rangelands, forestry and urban-industrial. In order to see
spatio-temporal variation and environmental partition of estrogenic compounds, three sampling points were analysed in sediment and water (in triplicate) monthly from December 2014 to February 2015. These sites were located in the Santa Lucia Grande basin (SG1, SG2 and SG4). In all of them, the predominant land use associated was rangelands, but SG1 receives effluents from Minas city while SG2 and SG4 represents sites with lower degree of anthropogenic impact.
2.2 Water and sediment quality parameters
In all site physico-chemical variables (pH, dissolved oxygen, temperature and conductivity) were measured using field sensors and four replicates samples to measure total phosphorous (TP), ammonium (NH4+), total suspended solids (TSS) and suspended organic matter (MOS) according to the standards of Valderrama (1981) and APHA (1985). At each sampling point, four sediment samples were also taken to measure organic matter percentage (%MO) and grain size (Phi). All fractions > 2 mm were separated by sieving in successive intervals, while fractions < 2 mm were determined with a laser diffraction particle analyzer (Shimadzu model SALD-3101). Three different certified samples were used to calibrate the method (JISS 11, Licopodium and glass beads). The granulometric parameter mean grain size (Phi) was compiled in SYSGRAM 3.0 using the equation proposed by Folk and Ward (1957), using laser method in the Laboratory of Environmental Sciences (UENF, Rio de Janeiro-Brazil).
2.2 Sediment and water extractions for YES
Before collection, all glass materials were previously rinsed with alcohol and acetone analytical grade (HPLC grade, Tedia). Superficial sediment and water samples were taken in amber glass bottles for the YES assay. Samples were transported on ice, and stored at -20°C. Ten grams of dried sediment samples were extracted by sonication with methanol (10 mL; 5 min). Subsequently, the liquid phase was separated by centrifugation (2500 g; 5 minutes) three times collecting supernatant and added ultra161
pure water up 250 mL.
Water samples were filtered through 1.2 μm glass fiber filters (Merck) and 0.45 μm cellulose acetate filters (Merck), purified using columns of solid phase extraction (SPE, Strata-X, Phenomenex®) 500 mg / 6 mL. Columns were preconditioning with 6 mL of hexane, 2 mL of acetone, 6 mL of methyl alcohol and 10 mL of ultrapure water with pH 3. For sediment extract and water 250 and 1000 mL respectively were extracted in the columns (pH adjusted to 2 with HCl). After that, samples were eluted with 4 mL of acetone, dried in nitrogen flow and resuspended in 2 mL of ethanol for YES assay.
2.3 Yeast Estrogen Screen (YES) bioassay
Yeast strain was kindly provided by Prof. Marcia Dezotti (UFRJ, Brazil). The assay procedure was according to the original protocol (Routledge and Sumpter, 1996) following the adaptations of Bila et al. (2007). Briefly, the yeast stock stored at -20°C in a cryogenic tube (2mL) with growth medium and glycerol (40 %) was added to 10 mL of the growth medium and grew on an orbital shaker 48 hs. Hundred microliters of culture were added for a new growth medium (10 mL) and grew on an orbital shaker for another 24 hs to obtain an enough cell density. The assay medium was prepared by mixing 25 μL of the above solution, 25 mL of growth medium, and 250 μL of the chromogenic substrate chlorophenol red-β-D- galactopyranoside (CPRG, Sigma-Aldrich®180 ). The 17 β-estradiol (E2) standard solution and the samples extracts were serially diluted in ethanol and 10 μL of each dilution were transferred (in duplicate) into a 96-well optically flat microtiter plate and allowed to
evaporate until dryness. Then two hundreds microliters were seeded into 96-well test plates (Kasvi®183 ) and each time dilution series of E2 (> 98%, Sigma-Aldrich®) were used as a calibration curve. The plates were sealed with masking tape and vigorously shaken on a plate shaker for 2 min. Then were incubated in darkness at 30 °C during 72 h and absorbance was read at 540 nm for colour development (estrogenicity) with a plate reader (Softmax Pro 5 SpectraMax M3). Limit of quantification (LQ) were determined according to Fávaro (2010). Their value was 0.035 ng/g of sediment. Turbidity correction was applied in all sample extracts and standards according to Coleman et al. (2004). The estrogenic activity was calculated as E2 equivalents (EQ-E2) by interpolation from the E2 standard curves (ng/L).
Citotoxicity of sediment samples were obtained by measurements of inhibition of yeast cell growths by reduction of absorbance at 620 nm, compared to reference wells (Frische et al., 2009)
2.4 Statistical analysis
In order to investigate whether estrogenicity was affected by environmental variables a Generalized Linear Model GLM (with response variable log transformed) was carried out. It includes linear regressions between estrogenicity and environmental variables, retaining those variables with significant effects. All values of parameters were transformed using exponential function for correct interpretation of values. All analyses were performed using R-statistical free package.
3 Results
3.1 Land uses and environmental quality parameters
The drainage area of each sampling point denotes heterogeneous sizes ranging from 1 to
9122 Km2 (Table 1). Land uses varied considerably between sub-basins and sampling points. The predominant land use represented in the sampling points was agriculture with a mean of 74.1% (including 3.5% of forestation with Pine spp. and Eucaliptus spp.), followed by rangelands (21.3%) and urban-industrial (4.9%). Environmental variables showed a wide range of values (Table 1). For example, Total phosphorus (TP) and ammonium ranges were from 4.2 to 1530 and 4.1 to 10596 μg/L respectively (Figure 2).
3.2 Estrogenicity and citotoxicity
Estrogenicity was observed in 14 of the 42 sites analyzed (Figure 3). All sub-basins showed estrogenicity except Canelón Chico (CC). The maximum value (EQ-E2 8.49 ng/g of sediment) was found in Colorado sub-basin (Figure 3). Inhibition of cell proliferation (cytotoxicity) was observed in 9 sampling points, covering all land uses and sub-basins. The greatest inhibition of yeast growth (92%) occurred downstream from the city of Progreso (Colorado stream,), which represent an important urbanized center whose water quality problems have been reported (Teixeira de Mello, 2007) (Figure 3).
3.3 Estrogenicity and environmental variables
The environmental variables that showed significant relationships with estrogenicity were total phosphorous, organic matter in sediment, urban-industrial land use area and total suspend solids (slope = 0.02, 0.61, 1.00 and 0.98 respectively) (Table 2).
3.4 Spatio-temporal dinamic of estrogenicity and citotoxicity
When compared estrogenicity in water and sediment samples in three sampling points (SG1, SG2 and SG4), no estrogenicity was detected in sediments during this period of study. We only found activity above the LQ in water samples of SG1 in February, in which no estrogenicity was found previously (winter). Relative low levels of citotoxicity were found in sediments of SG4 (2%) 239 and in water of SG2 with 3% of inhibition (Table 4).
4 Discussion
Water quality parameters observed in this study confirm the eutrophication process of the Santa Lucía River, which was mainly reflected by high levels of TP and ammonium in sites associated with intensive agriculture and urban land use, and is generally in agreement with other studies (Chalar et al., 2013; DINAMA, 2011; Goyenola et al., 2015; Inda and Indarte, 2010). The main site with serious problems of water quality was SJ5 inside San José city (both nutrient overload and low values of OD were found). Other points with higher TP and ammonium levels where sites of the Colorado stream sub-basin particularly from Las Piedras stream (C1), Paso Severino damming (SC7) located in Santa Lucía chico river sub-basin and sites located along the Canelón Grande stream sub-basin due to high charge of TP (CG1 and CG2). While most sites showed acceptable levels of OD (> 5 mg/L) according to the standard values of water quality (D.253/79), the sampling site that presented the lowest values (2.75 mg/L) is located on a stream inside San Jose city (SJ5), this site have very low water movement (i.e. pool) and heavily loaded organic waste. Only 4 sites showed values of TP below the maximum concentration (25 μg/L) established for drink water source according to national regulations (D.253/79). Many of these sites are established within watersheds with intense agricultural activity, besides the presence of major cities nearby. Therefore could receive diffuse inputs of TP probably of agrochemical origin, and punctual contributions from domestic or industrial sources. In fact a marginal association between TP and agricultural 263 (r = 0.29, p = 0.06) and urban (r = 0.26, p = 0.09) land use was found. Ammonium values were also high. The maximum value (10595 μg/L) was found on SJ5 exceeding standards established by national legislation (D.253/79). These and others high concentrations are probably linked to the discharge of untreated domestic sewage. Most of the cities and villages in Uruguay do not have sewage primary treatment plant.
Estrogenic and citotoxic activities were found in sediments of Santa Lucia Basin, in all sub- basins and associated with different land uses, reinforcing the idea that environmental estrogenicity is a multifactorial response, depending on multiple human activities, and diffuse and punctual sources (Gorga et al., 2015). However, a significant relationship between estrogenicity and urban-industrial land use area was found (Table 2) which could indicate that point source discharges are contributing to the release of substances with estrogenic potential. This is particulary important taking into account that sewage effluents are the major source of estrogenic compounds in the aquatic environment (Ying et al., 2009).
The presence of xenoestrogens was determined mainly from hydrophobic compounds
present in the sediments, becoming our observations limited to these kind of compounds, therefore false positives are discarded while false negative could potentially exist if presence of hydrophilic estrogenic chemicals are present. Within 14 estrogenic sites, values were generally higher when compared to other studies. In most of these works samples analyzed come from watersheds that receive raw and treated discharges from urban and industrial regions, but also two of them explore areas with intensive agriculture activity and livestock operation. Therefore, could be compared with sites from Santa Lucía basin. However, it is emphasized that the technique used is a fairly conservative approach that is to say values were obtained with high LQ (0.035 ng/g of sediment). Highest estrogenicity values in
sediment and water within this literature correspond with Yundang Lagoon (China), which receive the discharge of municipal sewage (Table 4).
Most of estrogenic sites were located in Colorado sub-basin (Figure 3), which are associated with areas of high urbanization and agriculture (Teixeira de Mello, 2007). However, estrogenicity was found in sites that were considered of relatively low human impact, such as up-stream sampling points of the Santa Lucia Grande and Chico rivers. This could be associated with multiple sources that release substances which may be acting as xenoestrogens in addition to urban and agricultural activities. At Canelón Grande (CG) and Chico (CC) high values of estrogenicity were found. It is important to note that these streams fed the Santa Lucía river 1 km upstream of a dam for extracting water for potabilization (Aguas Corrientes dam, OSE). These results, like occur with San Francisco stream near city of Minas, reflect the inefficiency to remove EDCs compounds with only primary treatment plants of wastewater (Auriol et al., 2006; Xu et al., 2012).
Besides 14 estrogenic samples, 9 sites presented cytotoxicity. One of them was in Paso Severino dam (where estrogenicity also was found). It would be interesting to assess the estrogenicity and cytotoxicity at this point in water samples, since these waters are used for human consume and therefore is important to consider the effectiveness of the type of treatment carried out here. Significant association between estrogenicity and environmental variables were observed. Negatively associated variables were TP, %MO and TSS, while urban land use area was positively associated with estrogenicity 313 (Table 2). The temporal and space estrogenicity variation observed indicate how dinamic is this aquatic system and estrogenicity in water and sediment dependent on point and diffuse sources discharges that may have occurred at certain points and times. The dinamic of estrogencity observed during temporal study and the absence of association with indicators of water quality parameters or area drainage reported at each sampling point, are consistent with the existence of multiple sources of pollutants (Gorga et al., 2015; Lopes et al., 2010). In addition, the analysis of land use was carried out on a large scale, but there may be multiple small-scale activities that are affecting environmental estrogenicity and therefore should be used as indicators to consider in future studies. Finally, there may be multiple chemical compounds that can act as EDCs, and even more depending on the presence of other organic and inorganic compounds, their bioavailability and toxicity may vary. That is why we consider the analysis of response (i.e. estrogenicity) and not the analytical quantification of a given compound as the best tool to determine the degree of environmental quality of a body of water.
5 Conclusions
Results obtained in this study show that contamination by estrogenic substances in sediments occurs in various locations of the Santa Lucia River basin, associated with multiple sources of contamination. This work describes for first time in Uruguay the presence of estrogenic substances in the environment using a direct assay method (YES), indicating the relevance of employment of these measurements in human sensitive watersheds.