This document summarizes a study that evaluated estrogenic disrupting potency in aquatic environments and urban wastewaters by combining chemical and biological analysis. The researchers measured concentrations of natural estrogens and synthetic estrogen in water samples using analytical chemistry and tested the samples using a MELN biological assay. They found that estrone and beta-estradiol were present in all samples at levels from 0.1-15.7 ng/L and 0.1-2.3 ng/L respectively. The biological responses from the MELN test closely matched the results from chemical analysis, indicating that estrogens are good tracers of estrogenic disruption in urban surface waters.
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
Estrogenic Hormones in Aquatic Environments
1. Trends Trends in Analytical Chemistry, Vol. 28, No. 2, 2009
Evaluation of estrogenic disrupting
potency in aquatic environments
and urban wastewaters
by combining chemical
and biological analysis
`
C. Miege, S. Karolak, V. Gabet, M.-L. Jugan, L. Oziol, M. Chevreuil,
Y. Levi, M. Coquery
We studied estrogenic disrupting potency in rivers and wastewaters in the Orge catchment area near Paris, using analytical and
biological approaches simultaneously. The MELN test was applied to surface-water samples, urban storm run-off and wastewater-
treatment plant (WWTP) effluent in parallel with analytical determinations of natural estrogens and synthetic estrogen (ethin-
ylestradiol) using liquid chromatography and tandem mass spectrometry.
We quantified estrone in all samples in the range 0.1–15.7 ng/L. We also quantified b-estradiol in all samples, but at a lower
level (0.1–2.3 ng/L), but we never detected a-estradiol. We quantified ethinylestradiol only in WWTP effluent (0.2 ng/L); we
measured estriol in WWTP effluent (12.1 ng/L) and downstream effluent (4.9 ng/L).
The biological responses using the MELN test closely followed the chemical responses. Analytical quantification of estrogens
appears to be a simple way to trace estrogenic disruption in surface waters of urban areas, as these hormones are mainly
responsible for biological effects.
ª 2008 Elsevier Ltd. All rights reserved.
Keywords: Biological analysis; Chemical analysis; Equivalent Estrogenic Quantity; Estrogenic disrupting potency; Estrogenic hormone; Liquid
chromatography; Mass spectrometry; MELN test; Surface water; Urban wastewater
1. Introduction
`
C. Miege*,
V. Gabet,
M. Coquery Natural estrogens are a group of steroid
Cemagref, UR QELY, 3 bis quai Chauveau, hormones that include the main active
CP 220 - F-69336 Lyon, France hormone, 17ß-estradiol, estrone and
S. Karolak*, estriol. Endocrine disruptors are defined as
M.-L. Jugan, substances that interfere with the endo-
L. Oziol, crine system and disrupt the physiological
Y. Levi
functions of hormones. The presence of
´
Univ. Paris Sud 11, IFR 141, Faculte de Pharmacie,
´
Laboratoire Sante Publique – Environnement,
estrogenic compounds in surface waters
´
5 rue J.B. Clement, 92290 Chatenay-Malabry, has been noted since the early 1980s [1].
France Numerous endocrine-disrupting sub-
M. Chevreuil
stances [e.g., industrial or domestic
Laboratoire Hydrologie et Environnement – EPHE, chemicals (e.g., plasticizers, flame retar-
UMR 7619, 4, place Jussieu 75252 Paris cedex 05, dants and pesticides) and natural or syn-
France thetic hormones excreted by human
bodies] reach the aquatic environment
* daily via sewage systems. Indeed,
Corresponding authors.
E-mail: cecile.miege@cemagref.fr, sara.karolak@u-psud.fr industrial and domestic wastewaters are
186 0165-9936/$ - see front matter ª 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.trac.2008.11.007
2. Trends in Analytical Chemistry, Vol. 28, No. 2, 2009 Trends
recognized as the main sources for these pollutants, Berlin area and detected only E1 (0.16–0.86 ng/L,
which may act with different modes of disruption on n = 5). Ternes [8] did not quantify any of the estrogenic
animal and human endocrine systems. The conse- hormones studied (E1, b-E2, EE2, n = 15). In the dis-
quences of the presence of these substances in the solved phase of German rivers and creeks, Kuch et al. [9]
aquatic environment are still largely unknown, but some quantified E1 in 29 samples out of 31, at an average
negative impacts have been reported (e.g., the femini- concentration of 0.7 ng/L; EE2 and b-E2 were quantified
zation of fish in large rivers and toxicological effects on in about half of the samples, at an average concentration
wildlife) [2,3]. Great scientific efforts are therefore in of 0.8 ng/L and 0.6 ng/L, respectively; a-E2 was quan-
progress to improve evaluation of the presence and the tified in 8 samples out of 31 at an average concentration
effect of these compounds in the environment and to of 0.6 ng/L.
identify their sources and modes of transfer to the Beck et al. [10] measured the concentration of estro-
aquatic ecosystems. genic hormones in the dissolved phase of 10 Baltic Sea
Many papers have reported the presence of estrogenic sites and pointed out that E1 was always quantified
hormones (i.e., estrone [E1], 17b and a-estradiol [17b (0.10–0.53 ng/L), b-E2 and E3 were never detected, and
and a-E2], 17a-ethinylestradiol [EE2] and estriol [E3]) in EE2 was quantified in all the sampling sites except one
wastewaters and surface waters. In a recent paper, (0.45–17.2 ng/L).
`
Miege et al. [4] compiled concentrations and removal In the dissolved phase of Italian rivers, Lagana et al.
efficiencies measured in the dissolved phase of influents [11] found the free fraction of E1, b-E2, E3 and EE2
and effluents of wastewater-treatment plants (WWTPs) at 8 ng/L, 4 ng/L, 1 ng/L and 3 ng/L, respectively
with activated sludge processes. Detailed datasets were (7 samples in the Tiber river), while Zuccato et al. [12],
drawn from 117 research papers covering the period who analyzed only EE2, did not quantify it (one sample
from January to June 2006 for international studies and in the Lambro river and 7 samples in the Po river).
to February 2007 for French studies. In Table 1, we Other studies have been realized all over the world.
reported mean, minimum and maximum concentrations Boyd et al. [13] did not quantify E1 or b-E2 in the dis-
in WWTP influent and effluent for the five estrogenic ´
solved phase of American rivers and lakes. Farre et al.
hormones as well as their removals (data from [4]). [14] did not quantify E1, b-E2, E3 or EE2 in the dissolved
Concentrations range from 0.4 ng/L for EE2 to 670 ng/L phase of Spanish rivers (three sites). Morteani et al. [15]
for E1 in influents and from 0.1 ng/L for a-E2 to 275 ng/ studied 19 sites of rivers and creek waters in the Czech
L for E3 in effluents. Removal efficiencies ranged from Republic and found: E1 at 7.4 ng/L and E3 at 1.7 ng/L
68% for EE2 to 92% for E3. Thus, although removal only at one site; b-E2 at 7 sites at a maximum concen-
rates are relatively high, WWTPs nonetheless represent tration of 3.8 ng/L; and, EE2 at 6 sites at a maximum
an evident source of estrogenic hormones contaminating level of 4.6 ng/L.
surface waters. In summary, most studies report concentrations of free
In France, some studies have reported the concentra- estrogens in the dissolved phase and estrogens were not
tion of estrogenic hormones in river water. Cargouet ¨ systematically detected in surface waters. Concentra-
et al. [5] found concentrations of free E1, b-E2, E3 and tions measured were generally in the range of 1 ng/L
EE2 in the dissolved phase in the range 1.0–3.2 ng/L. and rarely over 10 ng/L. Since these molecules are
Labadie et al. [6] analyzed free and conjugated fractions moderately hydrophobic, with log Kow values between
of the same hormones in the dissolved and suspended 2.6 for estriol and 4.1 for 17a-ethinylestradiol, they
particular phase of the Jalles dÕEysine river and detected have also been detected in sediments in a few studies. For
none of them. example, Labadie et al. [16] analyzed 7 samples of river
In the dissolved phase of German rivers, Zuehlke et al. sediments from the River Ouse (UK) supposedly differ-
[7] analyzed the free fraction of E1, b-E2 and EE2 in the ently contaminated (located 0.2–5 km downstream of
Table 1. Mean, minimum, maximum concentrations and removals (with relative standard deviation) for estrogenic hormones in the dissolved
phase of wastewater-treatment plants with activated sludge processes (from [4])
Hormones Influent concentration (ng/L) Effluent concentration (ng/L) Removal (%)
min max mean n min max mean n R RSD n
E1 2.4 670 67 109 0.6 95 21 79 74 39 59
a-E2 1.5 17 7.4 36 0.1 3 0.8 9 79 22 6
b-E2 2.5 125 22 108 0.3 30 2.8 63 88 13 52
EE2 0.4 70 4.2 70 0.2 5 0.9 33 68 33 46
E3 15 660 115 36 0.4 275 13 33 92 20 36
n, Number of individual data.
http://www.elsevier.com/locate/trac 187
3. Trends Trends in Analytical Chemistry, Vol. 28, No. 2, 2009
WWTP effluents). Measured concentrations varied in the The sensitivity of the bioassay is expressed as the EC50
ranges 0.4–3.3 ng/g dry weight for E1 and <0.03– determined with 17b-estradiol. The YES test appears less
1.2 ng/g dry weight for E2, and were below 0.04 ng/g efficient, with an EC50 value of 203 ± 67 pM
dry weight for EE2. (55.3 ± 18.3 ng/L), which is four times larger than that
Chemical analyses of estrogenic hormones have the measured using the E-screen test {i.e., 53.2 ± 7.2 pM
advantages of reaching very low Limits of Detection (14.5 ± 2.0 ng/L) [17]}.
(LODs) (in the sub-ng/L range) and identifying molecules Sonneveld et al. [24] determined an EC50 value of
precisely. But other estrogenic disruptors are known to 16 pM (4.36 ng/L) using the ER-CALUX test. This value
induce estrogenic effects in aquatic environments (e.g., is similar to that calculated for the MELN test by Pillon
bisphenol A, polybrominated diphenyl ethers, phtha- et al. [23] [i.e. 16.6 pM (4.52 ng/L)].
lates, organochlorines, alkylphenols or dioxin). They Using the MELN test, Berckmans et al. [25] reported
would also have to be measured specifically. Hence, in an average value twice that reported by Pillon et al. [23]
order to check if estrogenic hormones are good tracers of (i.e. 33 ± 7 pM or 8.99 ± 1.91 ng/L).
urban sources of contamination of endocrine disruptors This last difference underlines the importance of the
in aquatic environments, it appears valuable to combine variability inherent in biological material and operating
chemical analysis of estrogenic hormones with biological conditions (e.g., incubation time or the measurement
tests of estrogenic effects. method for luciferase activity). These factors should be
Estrogenic activity can be measured using different assessed to optimize the sensitivity and the LOD of a
biological tests. Global biological disrupting effects are bioassay.
generally expressed as equivalent estrogenic quantity of The MELN test was adapted in our laboratory and an
17b-estradiol (EEQ). The yeast estrogen screen (YES) EC50 value of 6.2 ± 0.4 pM (1.69 ± 0.11 ng/L) was
test, based on recombinant yeast cultures expressing validated on several assays. It is worth noting that the
human estrogen receptor, has been used for investiga- MELN and ER-CALUX tests have been selected to define a
tions in influents and effluents of WWTPs [17–19] and standardized test for the in vitro evaluation of estrogenic
surface waters [20]. Levels were up to 130 ng/L EEQ in activity [26]. In these studies, biological tests were used
WWTP influents, <20 ng/L EEQ in effluents and around as a global approach to give an indication of the endo-
1 ng/L EEQ in surface waters. crine-disruption risk for aquatic wildlife exposed in the
Korner et al. [21] used the E-screen test involving environments studied.
proliferation of human breast-cancer cells MCF-7 under As mentioned above, several substances are known to
estrogenic control to quantify estrogenic activity in WWTP be estrogenic disruptors. These substances are able to
influent and found concentrations of 58–70 ng/L EEQ. react with an estrogen receptor, and, using this ability,
The ER-CALUX and MELN tests are similar bioassays using they could be detected with the bioassays described
human breast-cancer cells T47D and MCF-7, respectively, previously. However, their affinities to estrogen receptors
stably transfected with luciferase reporter gene. are far weaker than the affinities of natural estrogenic
Murk et al. [22] used the ER-CALUX bioassay and hormones or ethinylestradiol. For example, Pillon et al.
quantified estrogenic agonist activities on estrogen [23] measured EC50 values with the MELN test of
receptors in influents (1.1–119.8 ng/L EEQ), effluents 339 pM (74.70 ng/L) for p-nonylphenol and 11 lM
(0.03–16.1 ng/L EEQ) and surface waters (0.25– (2.42 mg/L) for n-nonylphenol.
1.72 ng/L EEQ). It is interesting to compare biological results of estro-
Using the MELN test, Pillon et al. [23] found 1.4 ng/L genic activity to analytical determinations of estrogenic
EEQ in surface waters. substances, in particular for natural and synthetic
In a previous study around the greater suburban area estrogenic hormones. Such a study was realized by
of Paris, Cargouet et al. [5] applied the MELN test and
¨ Cargouet et al. [5] on four WWTPs located upstream or
¨
found 43–63 ng/L EEQ in influents, 2–24 ng/L EEQ in downstream Paris, France, and in surface waters.
effluents and 1–3.2 ng/L EEQ in surface water. Estrogenic activities assessed by the MELN test were
Estrogenic activities determined in areas located in mainly associated with estrogenic hormone concentra-
different countries therefore appeared relatively homo- tions (E1, E2, E3 and EE2) quantified in WWTP influent
geneous. High levels are reported at the entry of WWTP and effluent samples. Using weighting factors, chemical
influent, then estrogenic activities are significantly re- EEQ was estimated from estrogenic hormone concen-
duced by the WWTPs, in a 60–95% range, leading to trations and compared to biological EEQ; chemical EEQs
residual EEQ values in surface water of nearly 1–4 ng/L. represented half of the biological EEQs in WWTP influent
However, the results may differ depending on the and were equal to the biological EEQs in effluent;
bioassay used. Thus, Nelson et al. [17] compared EEQ chemical EEQs were higher than the biological EEQs in
values obtained with the E-screen and the YES tests and surface-water samples, and this difference was partly
obtained correlation factors of 0.56–0.75, depending on explained by the relatively high EE2 concentrations
the operating conditions. compared to those in WWTP samples. In a similar study
188 http://www.elsevier.com/locate/trac
4. Trends in Analytical Chemistry, Vol. 28, No. 2, 2009 Trends
on wastewaters, Nelson et al. [17] found correlation tion. Two samples were collected at each site, one for
factors (r) of 0.71–0.80 when comparing biological EEQs analytical measurement and one for biological testing.
measured with the E-Screen or YES tests and chemical
EEQs obtained from analytical determinations (n = 10, 2.2. Chemical analysis
effluent or influent from 5 WWTPs). In this study, the We analyzed the dissolved fraction of hormones,
authors obtained chemical EEQs using EEQ factors re- including free and conjugated forms. We have described
ported in the literature for the different bioassays. Simi- the analytical methodology in detail [27]. Below, we
larly, Salste et al. [18], using the YES test, showed that briefly describe the main steps.
the main estrogenic activity observed in the effluents of
one WWTP in Finland was mainly due to E1. 2.2.1. Preparation before extraction. Aqueous samples
The objective of our study was to assess estrogenic were filtered on site, on the same day, through a pyro-
activity using the MELN tests for surface-water samples, lyzed (450°C, 1 h) glass-fiber filter (GF/F, 0.7 lm pore
urban storm run off and WWTP effluent, in parallel with size). The samples were then submitted to enzymatic
analytical determination of natural estrogenic hormones hydrolysis by beta glucuronidase aryl sulfatase from
and a synthetic estrogenic hormone (i.e. ethinylestradiol). Helix pomatia (1/1000, v/v) at pH 5.2 and 52°C for 15 h.
We chose to analyze surface-water samples collected in Perdeuterated hormones (E1-D4, 17b-E2-D2, 17a-
different areas on a single catchment basin, with the aim EE2-D4 and E3-D2), used as internal surrogates, were
of following the variation of estrogenic hormone con- spiked before the extraction step, at a concentration of
centration and biological activity along the river flow and 125 ng/L in WWTP influents and 50 ng/L in effluents or
urban and country settings. An urban storm run-off and a river waters.
WWTP were located along the rivers studied, causing
local input of ethinylestradiol and natural estrogenic 2.2.2. Extraction and clean-up protocols. Sample volumes
hormones. were 100 mL for influents and 250 mL for river waters
and effluents. Solid-phase extractions were performed
with an Autotrace workstation (Caliper Life Science)
2. Methods with Oasis HLB cartridges as follows: after washing with
6 mL of methanol and 6 mL of ultrapure water, sample
2.1. Sampling sites and protocols was percolated and elution was achieved with 4 mL of a
´
Sampling was performed on the Predecelle and Orge mixture ethyl acetate/methanol (70/30, v/v). The ex-
rivers, the latter being a tributary of the River Seine. tract was evaporated to dryness and reconstituted in a
They merge in an urban area upstream of Paris (Fig. 1). mixture of 1 mL of methylene chloride/heptane (50/50,
The Orge watershed has a surface of 952 km2; it is v/v). The extract was then purified on Florisil as follows:
covered by agricultural land upstream and is entirely after percolation of the extract, 5 mL of a mixture of
urbanized downstream. The Orge river, in contrast with acetone/heptane (75/25, v/v) were used for elution,
its two main tributaries (the Yvette and Remarde rivers) then evaporation to dryness was performed and the ex-
does not receive any WWTP input. However, some tract was reconstituted in 200 lL of a mixture of water/
diffuse domestic wastewater discharges can reach the acetonitrile (60/40, v/v). Finally, b-estradiol acetate,
Orge river due to poor connections on the stormwater- used as internal standard, was spiked at 40 lg/L just
sewer system. Sampling sites were chosen upstream or before injection into the chromatographic system.
downstream of specific points (i.e. urban sites, WWTP
effluent, stormwater output, tributary junctures and a 2.2.3. Liquid chromatography and tandem mass
marsh area). Samples were collected on the 24 Sep- spectrometry (LC-MS2). Chromatographic analyses
´
tember 2007 on the Predecelle River and on the 25 were performed on an Xbridge Waters C18 end-capped
September on the Orge River downstream the Remarde column (150 mm · 2.1 mm · 3.5 lm) and guard
tributary (Fig. 1). As we wanted to characterize the column with an Agilent 1110 coupled with an API
longitudinal gradient of contamination, we choose to 4000 with triple-quadrupole mass spectrometer (Applied
sample during low flow; flow rates were <0.9 m3/s for Biosystems-MDS Sciex). The injected volume was 10 lL.
the Orge river. A rain event occurred on the 24 A gradient with LC-grade water and acetonitrile (flow
September and allowed us to collect a stormwater sam- rate, 0.2 mL/min) was applied for the separation of the 5
ple (i.e. mixture of rain-water run-off and domestic hormones: 40% acetonitrile 0–2 min, up to 80% aceto-
wastewater from a combined sewer system). nitrile at 4.5 min and until 15 min. The column tem-
For each sampling site, 1 L surface water or WWTP perature was set at 35°C. Ionization was performed with
effluent was collected in amber-glass bottles with Teflon an electrospray source in a negative mode and acquisi-
caps, previously washed and rinsed with methanol and tion was achieved in multiple reaction monitoring
ultrapure water. Special care was taken to rinse the (MRM) mode. As recommended in EU Commission
bottle at least twice with sampling water before collec- Decision 2002/657/EC [28], the MS2 conditions
http://www.elsevier.com/locate/trac 189
5. Trends Trends in Analytical Chemistry, Vol. 28, No. 2, 2009
Figure 1. Sampling sites in the Orge catchment, Paris area (n and )) and sites for river-flow measurements (=).
included the use of 2 ionization transitions for each 2.2.4. Performances of the analytical method and quality
compound (except for the perdeuterated surrogates), one controls. We have reported the performances of the
for the quantification (QT) and one for the identity method in detail [27]. The method was validated
confirmation (CT). Final concentrations were calculated according to the French standard NF XPT 90-210 [29].
using recoveries obtained for the internal perdeuterated Acceptable linear responses were obtained for all 5
surrogates (17a-E2-D2 is corrected by 17b-E2-D2). hormones using standard mixtures containing
190 http://www.elsevier.com/locate/trac
6. Trends in Analytical Chemistry, Vol. 28, No. 2, 2009 Trends
0.5–80 lg/L of hormones in vials before injection, which of methanol at a flow rate of 1 mL/min. The extract
correspond to concentration ranges of 1.0–200 ng/L for was then evaporated to dryness under nitrogen at 40°C
influents and 0.4–80 ng/L for effluents and river waters. and dissolved in 350 lL DMSO; it was then stored at
During validation of the method, limits of quantification À20°C before analysis. Just before biological testing, the
(LOQs) were estimated from 0.4 ng/L for E1 and a-E2 to extract has to be diluted 1000-fold to avoid cellular
1.0 ng/L for EE2 in surface and effluent waters, and from toxicity.
0.8 ng/L for a-E2 to 3.0 ng/L for EE2 in influent waters.
However, LOQs greatly depend on the sample matrix and 2.3.3. Biological tests. One bioluminescent cellular
on the sensitivity of the instrument, which can vary model was used to test estrogen-receptor agonist potential
from day to day. For this study, results were considered of the samples extracts. The MELN cells were seeded into
higher than the LOQ when: 96-wells, white opaque, culture plates at a density of
(i) the 2 ionization transitions (for QT and identity CT) 2 · 104 cells per well and left to develop 24 h before use.
were confirmed, as explained in the EU Commission DMSO extracts of sample or calibration standards of b-E2
Decision 2002/657/EC [28]; (10À13–10À8 mol/L) were left 16 h for incubation at
(ii) the concentration value was within the range of 37°C. The cells were then washed twice with PBS buffer,
the calibration curve. and luciferase activity was measured on lysed cells. Each
Within-day recoveries obtained for 5 replicate samples analysis was repeated 5 times using 5 replicate culture
of surface water, WWTP influent and effluent were wells. The mean of the 5 luminescence activities was used
generally in the range 82–115% with relative standard for calculation and results were expressed as relative
deviations <22%. The specificity of the method was luminescence unit (RLU) that corresponded to the mean
verified for the 5 estrogenic hormones, which meant that luminescence value related to the one of DMSO control.
matrix effects were not significant (i.e. the use of per- In parallel to the MELN cells, the MTT test was used to
deuterated hormones as internal surrogates appears to verify cellular viability, as described by Mosmann [30].
be an efficient method to correct for matrix effects).
During sample analysis, we obtained satisfactory qual- 2.3.4. Performances of the biological tests. LOD, esti-
ity controls: none of the 5 estrogenic hormones was de- mated as the concentration of hormone leading to
tected in blank samples and, by using standard solutions, luciferase activity significantly different (p = 0.05) from
we verified that instrumental sensitivity did not vary. the DMSO control, was 0.1 pM (0.03 ng/L of b-E2).
Repeatability was around 12% for each test (n = 5).
2.3. Biological analysis Sigmoıdal dose-response curves were estimated from
¨
2.3.1. Materials and chemicals. 17ß-estradiol (b-E2) was calibration standards allowing LOQs in a range 10À12–
from Sigma-Aldrich (St-Quentin-Fallavier, France). 10À9 M (0.3–272 ng/L of b-E2).
Standard solutions were made in dimethylsulfoxide In order to determine the relative biological activity of
(DMSO, HPLC grade, Sigma-Aldrich). For sample prep- the chemically-analyzed estrogens, dose-response curves
aration, glass-fiber filters (1 lM) were from Whatman were drawn using calibration standards of hormone for
and Oasis HLB-500 mg cartridges were purchased from each compound leading to EC50 values of 193.3 pM
Waters (Guyancourt, France). Methanol HPLC grade, (52.3 ng/L) for E1, 6.3 pM (1.69 ng/L) for b-E2, 69.4 pM
acetone Normapur and hexane Pestinorm were from (20.0 ng/L) for E3 and 3.9 pM (1.16 ng/L) for EE2. The
VWR (Strasbourg, France). The material for cell culture relative potencies to b-E2 (EC50 ratio), estimated from 3
was supplied by Life Technologies (Cergy-Pontoise, repeated curves in a same run, were 0.04 ± 0.01,
France). The luciferase reporter-gene assay kit was 0.11 ± 0.04 and 1.79 ± 0.45 for E1, E3 and EE2,
supplied by Roche Applied Science (Meylan, France) and respectively.
a Centro LB 960 microplate luminometer (Berthold,
Thoiry, France) was used to measure luminescence.
3. Results and discussion
2.3.2. Liquid-sample preparation. Sample preparation
procedures were similar to that for chemical analysis, 3.1. Chemical analysis
but special care was taken to avoid contamination from Fig. 2 shows the concentrations of the 5 estrogenic
extraction material that could lead to false-positive re- hormones.
sults. Sample preparation was progressed within 24 h E1 was quantified in all samples: from 0.1 ng/L
after collection. Filtered liquid phase was extracted on downstream Limours (Station 2) to 15.7 ng/L in Briis
Oasis HLB cartridges previously washed with 10 mL WWTP effluents (Station 3). A relatively high concen-
methanol and 10 mL purified water. Then, 1 L of water tration (13.7 ng/L) was also measured in the urban
sample was passed through the cartridge at a flow rate of storm run-off (upstream of Limours, Station 1).
6 mL/min. After drying the cartridge for 5 min under b-E2 was also quantified in all samples, but at a lower
vacuum aspiration, elution was carried out using 10 mL level: from 2.3 ng/L in the urban storm run-off (up-
http://www.elsevier.com/locate/trac 191
7. Trends Trends in Analytical Chemistry, Vol. 28, No. 2, 2009
1 Limours
WWTP
18 Urban storm effluent 2
Pr
run off
Yv
16
éd
3
4
ett
ec
e
el
14
le
12 5 8
6 7 9 10
-1
10 Orge
ng.L
8
6
4
2
0
Downstream
Downstream
Downstream
Downstream
Effluent Briis
Athis-Mons
Upstream
Villemoisson
Vaugrigneuse
Yvette - Viry
les Arpajons
Briis WWTP
St Germain
Epinay
Downstream
Limours
Upstream
Chatillon
Limours
Yvette-
WWTP
pond
Prédecelle river Orge river Yvette Orge river
river
E1 alpha E2 bêta E2 E3 EE2
Figure 2. Concentration (ng/L) of the 5 estrogenic hormones in the dissolved phase of surface waters, storm run-off and wastewater-treatment
plant (WWTP) effluent from selected sites in the Orge catchment.
stream of Limours, Station 1) to 0.1 ng/L downstream of the 5 hormones concentrations · river flow). This
St. Germain les Arpajons (Station 6). estimated hormones flow in the Orge river at Morsang
a-E2 was never detected. downstream the Yvette river (Station 9, 4.07 lg/s) is
EE2 was only quantified in Briis WWTP effluent similar to that in the Yvette river (Station 8, 1.76 lg/s)
(0.2 ng/L, Station 3). plus the one in the Orge river upstream the Yvette river
E3 was measured in Briis WWTP effluent (12.1 ng/L, (Station 7, 2.00 lg/s). These results allow us to validate
Station 3) and downstream Briis effluent (4.9 ng/L, our concentration measurements.
Station 4). For information, from [31], the mean annual river flow
When compared with the levels of other hormones, is 1.33 m3/s for the Yvette river at Villebon (evaluated in
the higher concentrations of E1 can partly be explained the period 1968–2008), 2.23 m3/s for the Orge river up-
by it being a degradation product of b-E2 and EE2. stream the junction with the Yvette river (evaluated in the
The decreasing concentrations of E1 and b-E2 from period 1982–2008) and 3.89 m3/s for the Orge river at
the urban storm run-off upstream Limours to down- Morsang, downstream of the junction with the Yvette
stream of Limours (Stations 1 and 2) can be explained by river (evaluated in the period 1967–2008).
dilution in the river flow, and degradation and adsorp-
tion on river sediment.
If we consider the concentration of E1 or the sum of 3.2. Biological analysis
concentration of the five hormones, we observe a strongly Estrogenic potential, reported as RLUs in Fig. 3, was
´
decreasing gradient all along the Predecelle river, from the observed for all samples. A high RLU value of 22.6 was
WWTP Briis effluent input (Station 3), identified as a observed in WWTP effluent that decreased downstream
source of contamination, to the junction with the Orge as a function of the distance from the river input. In the
river at St. Germain les Arpajons (Station 6). same way, an RLU value of 14.1 was observed in
As mentioned in Fig. 1, river flows measured on the the Yvette river, with a constant decrease observed after
25 September were equal to 0.65 m3/s for the Yvette the junction with the Orge river.
river at Villebon (i.e. 10 km upstream the junction with Meanwhile, estrogenic activities were generally low
the Orge river), 1.16 m3/s for the Orge river upstream of and Quantification was only possible for 5 out of 10
the junction with the Yvette river and 1.94 m3/s for the samples: 2.8 ng/L EEQ was measured in the WWTP
Orge river at Morsang, downstream the junction with effluent and values near 1 ng/L EEQ were measured for
the Yvette river. the four river samples (Fig. 4). These values are similar
We can estimate the mean daily flow of the sum of the to those reported in surface waters of the River Seine in
5 hormones from the measured concentrations (i.e. sum our previous study [5].
192 http://www.elsevier.com/locate/trac
8. Trends in Analytical Chemistry, Vol. 28, No. 2, 2009 Trends
1 Limours
30
2
Pr
WWTP effluent
Yv
éd
25 3
4
ett
ec
e
Urban storm
el
le
run off
20 5 8
6 7 9 10
Luciferase activity (RLU)
Orge
15
10
5
0
Athis-Mons
Effluent Briis
Villemoisson
Upstream
Downstream
Downstream
Downstream
Downstream
Vaugrigneuse
les Arpajons
Yvette - Viry
Briis WWTP
Epinay
St Germain
Downstream
Limours
Upstream
Chatillon
Limours
Yvette-
WWTP
pond
Prédecelle river Orge river Yvette Orge river
river
Figure 3. Estrogenic disruption measured by MELN tests (expressed in relative luminescence units, RLUs) in the dissolved phase of surface
waters, storm run-off and wastewater-treatment plant (WWTP) effluent from selected sites in the Orge catchment.
3.3. Chemical vs. biological analysis (i.e. a weighting factor equal to 1.79 for EE2, 1.00 for
Chemical analysis and RLU in Figs. 2 and 3, respectively, b-E2, 0.11 for E3, and 0.04 for E1). In Table 2, these
showed similar profiles, especially for the decreasing chemical EEQ values are compared with the RLU values
concentrations downstream of the WWTP effluent input obtained from the MELN test. The correlation was con-
and for the mixing of Yvette and Orge rivers beyond their firmed as good with a Spearman Rank test coefficient of
junction. 0.87 (p < 1%).
The chemical EEQ values were calculated from E1, The biological EEQ values were determined for the five
b-E2, E3 and EE2 concentrations weighted by a factor samples with RLU values above the LOQ. From Fig. 4, we
obtained from relative estrogenic potential on MELN cells observe that the chemical EEQ is higher than the bio-
4 Urban storm WWTP
run off 1 Limours
effluent
2
3
Pr
Y
Yv
ré
3
d
de
4
e
et
ce
tte
e
-1
lle
e
ng.L
2 5 8
6 7 9 10
Orge
1
< LOQ < LOQ < LOQ < LOQ < LOQ
0
Athis-Mons
Upstream
Vaugrigneuse
Downstream
Downstream
Downstream
Downstream
Villemoisson
Effluent Briis
les Arpajons
Yvette - Viry
Briis WWTP
Epinay
St Germain
Downstream
Limours
Upstream
Chatillon
Limours
Yvette-
WWTP
pond
Prédecelle river Orge river Yvette Orge river
river
Chemical EEQ Biological EEQ
Figure 4. Comparison of equivalent estrogenic quantity (EEQ) obtained from chemical or biological measurements. Biological EEQ was not
quantified for samples with relative luminescence unit (RLU) values under the limit of quantification (LOQ).
http://www.elsevier.com/locate/trac 193
9. Trends Trends in Analytical Chemistry, Vol. 28, No. 2, 2009
Table 2. Comparison study between relative luminescence units (RLU) values and chemical equivalent estrogenic quantity (EEQ)
Sample sites RLU Chemical EEQ (ng/L)
Urban storm run-off, upstream of Limours 14.77 2.82
´
Predecelle river, downstream of Limours 3.22 0.25
Effluent from Briis WWTP 22.66 3.13
´
Predecelle river, downstream of Briis WWTP 16.34 1.28
´
Predecelle river, downstream of Vaugrigneuse pond 13.97 0.53
Orge river, downstream of St. Germain les Arpajons 3.46 0.10
Orge river, upstream of Yvette Villemoisson 4.97 0.43
Yvette river, Epinay 14.09 0.51
Orge river, downstream of Yvette Viry-Chatillon 10.84 0.40
Orge river, Athis-Mons 9.07 0.84
logical EEQ in the urban storm run-off. Chemical and effluent and an urban storm run-off. This study showed
biological EEQs are comparable for WWTP effluent and that the biological responses using the MELN test closely
downstream of the WWTP, in which estrogenic hor- followed the chemical responses.
mones seem to be responsible for more than 90% of the The total (including the conjugated fraction) dissolved
biological effect. concentrations of the 5 hormones seemed to be a good
For five samples, chemical EEQs can be quantified (i.e. tracer of urban sources of contamination of estrogenic
´
in the Predecelle river downstream of Limours and in the disruptors in wastewaters and surface waters.
four sites of the Orge river), unlike biological EEQs Chemical analysis had the following advantages:
(estrogenic disruption was detected but biological activ- (i) reaching lower LOQs than MELN tests, as veri-
ity could not be quantified). fied on river samples collected at 5 sites (estro-
´
For two river samples (in the Predecelle river down- genic disruption was detected but biological
stream of the Vaugrigneuse pond and in the Yvette river activity could not be quantified).
at Epinay), the chemical EEQs are lower than the bio- (ii) being specific (i.e. not affected by matrix interfer-
logical EEQs. For these biological EEQs, a contribution ents), thanks to the use of perdeuterated hor-
from other estrogenic disruptors has to be taken into mones;
account; this may explain the lower contribution (about (iii) being selective (i.e. quantifying each of the 5
50%) of estrogenic hormones to the biological effect. estrogenic hormones individually).
The result obtained in the urban storm run-off up- Bioassays (e.g., MELN tests) have the advantage of
stream of Limours clearly differs from others with a measuring the estrogenic effect related to hormones and
chemical EEQ twice as large as the biological EEQ. other estrogenic disruptors present in the samples, so
In our previous study carried out near WWTPs in the they can be better adapted to screen estrogenic disrup-
river Seine [5], chemical EEQ values varied in the range tion in aquatic environments exposed to urban and
4.1–7.3 ng/L for surface water, whereas biological EEQs industrial sources of contamination. However, the pos-
remained around 1 ng/L and the contribution of EE2 sible inhibition effect from a mixture of pollutants needs
was estimated to be 35–48%. In the present study, EE2 to be taken into account by performing chromatographic
was not quantified in surface water. fractionation of samples and biological testing of the
The low biological activity observed upstream of Li- isolated fractions individually.
mours could partly be explained by an inhibition effect In conclusion, analytical quantification of estrogens
due to a mixture of organic pollutants being present in appears to be a simple way to trace estrogenic disruption
the sample. This was clearly observed by Salste et al. [18], in surface waters of urban areas, as these hormones are
who studied some chromatographic fractions from mainly responsible for biological effects.
WWTP-effluent samples and showed the inhibition of
b-E2 activity measured with YES tests. The compounds
responsible for this inhibition effect were said to interfere Acknowledgments
with the estrogen receptor. This study was supported by the Piren Seine program
(CNRS) and a Ph.D. grant from Cemagref for V. Gabet.
We thank Ph. Bados (Cemagref) for help for sample
4. Conclusion ´
analysis, M. Bimbot and V. Huteau (Faculte de Phar-
´
macie – Laboratoire Sante Publique Environnement,
Combining chemical and biological analyses of estro- Univ. Paris Sud 11) for technical support for biological
genic disruptors allowed us to confirm a tendency for tests, M. Deschamps (Syndicat intercommunal dÕassainiss-
contamination to decrease along the rivers studied ement des communes de Limours) for access to river sites,
´
(Predecelle, Yvette and Orge) downstream of a WWTP and M. Hollander (Lyonnaise des eaux) for access to the
194 http://www.elsevier.com/locate/trac
10. Trends in Analytical Chemistry, Vol. 28, No. 2, 2009 Trends
WWTP. MELN cells were kindly provided by P. Balaguer [18] L. Salste, P. Leskinen, M. Virta, L. Kronberg, Sci. Total Environ.
(INSERM U 439, Montpellier, France). 348 (2007) 343.
[19] B.V. Rutishauser, M. Pesonen, B.I. Escher, G.E. Ackermann,
H.R. Aerni, M.J. Suter, R.I. Eggen, Environ. Toxicol. Chem. 23
(2004) 857.
References [20] I.C. Beck, R. Bruhn, J. Gandrass, Chemosphere 63 (2006) 1870.
[1] M.L. Richardson, J.M. Bowron, J. Pharm. Pharmacol. 37 (1985) 1. [21] W. Korner, U. Bolz, W. Sussmuth, G. Hiller, W. Schuller, V. Hanf,
[2] B. Halling-Sorensen, S.N. Nielsen, P.F. Lanzky, F. Ingerslev, H. Hagenmaier, Chemosphere 40 (2000) 1131.
H.C.H. Lutzhoft, S.E. Jorgensen, Chemosphere 36 (1998) 357. [22] A.J. Murk, J. Legler, M.M. van Lipzig, J.H. Meerman, A.C. Belfroid,
[3] M.Y. Gross-Sorokin, S.D. Roast, G.C. Brighty, Environ. Health A. Spenkelink, B. van der Burg, G.B. Rijs, D. Vethaak, Environ.
Perspect. 114 (2006) 147. Toxicol. Chem. 21 (2002) 16.
` `
[4] C. Miege, J.M. Choubert, L. Ribeiro, M. Eusebe, M. Coquery, [23] A. Pillon, A.M. Boussioux, A. Escande, S. Ait-Aissa, E. Gomez,
Environ. Poll. (in press). H. Fenet, M. Ruff, D. Moras, F. Vignon, M.J. Duchesne, C. Casellas,
[5] M. Cargouet, D. Perdiz, A.M. Souali, S. Tamisier-Karolak, Y. Levi, J.C. Nicolas, P. Balaguer, Environ. Health Perspect. 113 (2005)
Sci. Total Environ. 324 (2004) 55. 278.
[6] P. Labadie, H. Budzinski, Environ. Sci. Technol. 39 (2005) [24] E. Sonneveld, J.A. Riteco, H.J. Jansen, B. Pieterse, A. Brouwer,
5113. W.G. Schoonen, B. van der Burg, Toxicol. Sci. 89 (2006) 173.
[7] S. Zuehlke, U. Duennbier, T. Heberer, J. Sep. Sci. 28 (2005) 52. [25] P. Berckmans, H. Leppens, C. Vangenechten, H. Witters, Toxicol.
[8] T.A. Ternes, Trends Anal. Chem. 20 (2001) 419. In Vitro 21 (2007) 1262.
[9] H.M. Kuch, K. Ballschmiter, Environ. Sci. Technol. 35 (2001) [26] S. Bremer, Oral communication in ECVAM (European Center for
3201. the Validation of Alternative Methods) meeting, Refinements of In
[10] I.C. Beck, R. Bruhn, J. Gandrass, W. Ruck, J. Chromatogr., A 1090 Vivo Tests Assessing the Reproductive/Developmental Hazards of
(2005) 98. Chemicals, 7 September 2006, Ispra, Italy.
[11] A. Lagana, A. Bacaloni, I. De Leva, A. Faberi, G. Fago, A. Marino, `
[27] C. Miege, P. Bados, C. Brosse, M. Coquery, Trends Anal. Chem. 28
Anal. Chim. Acta 501 (2004) 79. (2009) 238–245.
[12] E. Zuccato, S. Castiglioni, R. Fanelli, J. Hazard. Mater. 122 (2005) [28] European Commission, Commission Decision of 12 August 2002
205. implementing Council Directive 96/23/EC concerning the perfor-
[13] G.R. Boyd, H. Reemtsma, D.A. Grimm, S. Mitra, Sci. Total mance of analytical methods and the interpretation of results
Environ. 311 (2003) 135. (2002/657/EC), EC, Brussels, Belgium, 2002, 29 pp.
[14] M. Farre, R. Brix, M. Kuster, F. Rubio, Y. Goda, M. Lopez De Alda, ´
[29] AFNOR, Norme NF XPT 90-210, Protocole dÕevaluation dÕune
D. Barcelo, Anal. Bioanal. Chem. 285 (2006) 1001. ´
methode alternative dÕanalyse physico-chimique par rapport a `
[15] G. Morteani, P. Moller, A. Fuganti, T. Paces, Environ. Geochem. ´ ´´
une methode de reference, AFNOR, Paris, France, 1999, 58 pp.
Health 28 (2006) 257. [30] T. Mosmann, J. Immunol. Methods 65/1–2 (1983) 55.
[16] P. Labadie, E.M. Hill, J. Chromatogr., A 1141 (2007) 174. [31] Hydrological database (http://www.hydro.eaufrance.fr/index.
[17] J. Nelson, F. Bishay, A. van Roodselaar, M. Ikonomou, F.C. Law, php).
Sci. Total Environ. 374 (2007) 80.
http://www.elsevier.com/locate/trac 195