**2. Chemical detection using** *S. cerevisiae***-based bioluminescent bioreporters**

Yeast-based bioassays containing human receptors for estrogens and androgens fall into the recombinant receptor/reporter gene assay category. Estrogen or androgen response elements linked to a bioreporter molecule offer a low-cost method for screening samples rapidly for determining the presence of possible endocrine disruptors. Two widely used receptor/reporter assays for detecting estrogenic and androgenic compounds are the Yeast Estrogen Screen (YES) (Routledge & Sumpter, 1996) and the Yeast Androgen Screen (YAS) (Purvis et al., 1991). The *S. cerevisiae* YES and YAS bioreporters are colorimetric *lacZ*-based estrogen and androgen-sensing strains, respectively. The *S. cerevisiae* host strain for YES and YAS, contains the human estrogen receptor (hER-α) and human androgen receptor, respectively (Purvis et al., 1991; Routledge & Sumpter, 1996). Further, each host strain contains a series of either human estrogen response elements (EREs) or human androgen response elements (AREs) fused to the *lacZ* gene. The *lacZ* gene product, β-galactosidase, transforms the chromogenic substrate CPRG to a red product, measured by absorbance at 540 nm. These were the first widely used assays for yeast-based detection of estrogenic compounds.

The YES and YAS assays have been used extensively to measure endocrine responses to specific chemicals including polychlorinated biphenyls (PCBs) and hydroxylated derivatives (Layton et al., 2000; Schultz, 2002; Schultz et al., 1998), polynuclear aromatic hydrocarbons (PAH) (Schultz & Sinks, 2002), pesticides (Sohoni et al., 2001) and other compounds (Schultz et al., 2002). These assays have been adapted to environmental matrices including environmental waterways (Thomas et al., 2002), aquifers (Conroy et al., 2005), wastewater treatment systems (Layton et al., 2000) and dairy manure (Raman et al., 2004). Additional yeast-based bioreporters have been developed using either a colorimetric detection (Bovee et al., 2004; Gaido et al., 1997; Le Guevel & Pakdel, 2001; Rehmann et al., 1999), green

Fig. 1. Schematic representation of *S. cerevisiae* BLYEV (currently known as BLYR). This strain produces light continuously by constitutive expression of the *luxCDABE* genes from

**2. Chemical detection using** *S. cerevisiae***-based bioluminescent bioreporters** Yeast-based bioassays containing human receptors for estrogens and androgens fall into the recombinant receptor/reporter gene assay category. Estrogen or androgen response elements linked to a bioreporter molecule offer a low-cost method for screening samples rapidly for determining the presence of possible endocrine disruptors. Two widely used receptor/reporter assays for detecting estrogenic and androgenic compounds are the Yeast Estrogen Screen (YES) (Routledge & Sumpter, 1996) and the Yeast Androgen Screen (YAS) (Purvis et al., 1991). The *S. cerevisiae* YES and YAS bioreporters are colorimetric *lacZ*-based estrogen and androgen-sensing strains, respectively. The *S. cerevisiae* host strain for YES and YAS, contains the human estrogen receptor (hER-α) and human androgen receptor, respectively (Purvis et al., 1991; Routledge & Sumpter, 1996). Further, each host strain contains a series of either human estrogen response elements (EREs) or human androgen response elements (AREs) fused to the *lacZ* gene. The *lacZ* gene product, β-galactosidase, transforms the chromogenic substrate CPRG to a red product, measured by absorbance at 540 nm. These were

the first widely used assays for yeast-based detection of estrogenic compounds.

The YES and YAS assays have been used extensively to measure endocrine responses to specific chemicals including polychlorinated biphenyls (PCBs) and hydroxylated derivatives (Layton et al., 2000; Schultz, 2002; Schultz et al., 1998), polynuclear aromatic hydrocarbons (PAH) (Schultz & Sinks, 2002), pesticides (Sohoni et al., 2001) and other compounds (Schultz et al., 2002). These assays have been adapted to environmental matrices including environmental waterways (Thomas et al., 2002), aquifers (Conroy et al., 2005), wastewater treatment systems (Layton et al., 2000) and dairy manure (Raman et al., 2004). Additional yeast-based bioreporters have been developed using either a colorimetric detection (Bovee et al., 2004; Gaido et al., 1997; Le Guevel & Pakdel, 2001; Rehmann et al., 1999), green

*Photorhabdus luminescens* and the *frp* gene from *Vibrio harveyi*.

fluorescent protein (Bovee et al., 2007; Bovee et al., 2004) or the firefly luciferase bioreporter (Bovee et al., 2004; Leskinen et al., 2005; Michelini et al., 2005).

While the YES and YAS assays were highly specific for their target compounds, the colorimetric assays have disadvantages including addition of the chromophore for color development and a 3-5 day reaction time. This latter requirement hindered their ability for high-throughput analysis. Further, after 3 -5 days of incubation, it was unknown if any oxidation reactions were occurring that may activate the target compound. Some newer colorimetric assays have dramatically shortened the time required for color development (4- 6 h) through the use of alternative substrates but have the disadvantage of requiring cell lysis steps (Jaio et al., 2008).

To overcome these limitations, bioluminescent version of the YES and YAS reporters were developed by modifying the plasmid constructs of Gupta et al. (2003). Triple repeats of the human ERE were inserted in between the GPD and ADH1 constitutive promoters regulating the *luxA* and *luxB* genes, respectively (Figure 2) generating strain BLYES (Sanseverino et al., 2005). A similar strategy was used for strain BLYAS (Eldridge et al., 2007), which functions in the same way except that it contains the human androgen receptor gene on its genome and *luxAB* are under control of four androgen response elements (AREs), while the constitutive strain (BLYR) has both the *luxAB* and *luxCDEfrp* genes constitutively produced therefore it makes light constantly. The BLYR strain is used to determine whether samples or chemicals are toxic to the yeast, preventing false negatives. If a chemical is highly toxic, killing or inhibiting the cells, no light will be produced and it would be easy to mistake toxicity for no estrogenic response. However, if bioluminescence of the BLYR strain is reduced, since it produces light constitutively, it is obvious that toxicity exists in the sample.

Fig. 2. Schematic representation of *S. cerevisiae* BLYES. Estrogenic compounds cross the cell membrane and bind to the human estrogen receptor (hER). This complex interacts with estrogen response elements (RE) initiating transcription of *luxA* and *luxB*. *S. cerevisiae* BLYES contains the human estrogen receptor in its genome, while *S. cerevisiae* BLYAS has the human androgen receptor in the genome.

Analysis of Environmental Samples with Yeast-Based Bioluminescent Bioreporters 9

(DHT) over concentration ranges of 2.5 x 10-12 thru 1.0 x 10-6 M, respectively. Based on analysis of replicate standard curves, comparison to background controls, and screening a variety of chemicals, a set of quantitative rules was formulated to interpret data and determine if a chemical is potentially hormonally active, toxic, both, or neither (Sanseverino et al., 2009). The results demonstrated that these assays were applicable for Tier I chemical screening in EPA's Endocrine Disruptor Screening and Testing Program as well as for

Additional *S. cerevisiae* bioluminescent bioreporters for estrogens and androgens have been developed using the firefly luciferase as the reporter molecule. The bioreporters of Leskinen et al., (2005) each contain the firefly luciferase gene (*lucFF*) under control of hormoneresponsive promoters. The four strains, designated BMAEREluc/ERα, BMAEREluc/ERβ, BMAEREluc/AR, and BMA64/luc were used to detect estrogens (two versions), androgens, and toxicity, respectively. This bioassay is unique in that it uses two estrogen-sensing bioreporters; one contains the alpha form of the estrogen receptor and one contains the beta form (ERα, ERβ). These bioreporters were used by Svobodova et al. (2009) to test commercially available PCB mixtures and triclosan for estrogenic and androgenic activity but did not detect any activity with these samples (estrogenic or androgenic). This lack of estrogenic response in the bioluminescent assays may be due to the different mode of action of chemicals like triclosan (Stoker et al., 2010). In a study that examined the effects of triclosan exposure on female Wistar rats, triclosan advanced the onset of puberty symptoms. Also, a combination of ethinyl estradiol (EE2) and triclosan increased uterine weight significantly more than EE2 alone while triclosan alone had no effect. Therefore the mode of action of triclosan appears to have a synergistic effect on EE2 activity in Wistar rats. This effect appears to be independent of estrogen receptor binding given that bioluminescent yeast bioassays (Svobodova et al. 2009, Eldridge et al. unpublished data), which measure

In addition to hormone-mimicking chemicals, several other types of contaminants are also detectable with *S. cerevisiae*-based bioluminescent bioreporters. For example, the aryl hydrocarbon-sensing strain of Leskinen et al. (2008) contains genomically integrated human aryl hydrocarbon receptor and human aryl hydrocarbon nuclear translocator genes. In addition, it carries a plasmid-encoded copy of the firefly luciferase gene (*lucFF*) that is regulated by a series of aryl hydrocarbon receptor complex (AHRC) response elements (also called dioxin response elements or xenobiotic response elements, AhREs/DREs/XREs). Aryl hydrocarbon receptor proteins interact with both their AH ligand and the nuclear translocator protein then bind to the AhRE region of the *luc*-containing plasmid, activating transcription of luciferase, similarly to the receptor-response element system present in the

Another *S. cerevisiae*-based bioreporter has been created to measure arsenate and also UV damage (Bakhrat et al., 2011). This strain is based on the BLYES strain of Sanseverino et al. (2005), containing a constitutive *luxCDEfrp* plasmid and a *luxAB* plasmid that has been reengineered to be under control of the UFO1 promoter, which specifically responds to DNA damage by UV light and also arsenate. The strain is able to detect very low concentrations of arsenate (1x10-12 to 1x10-6 M), which makes them useful for environmental monitoring. It was also used to evaluate the level of UV protection in commercial sunscreens. When films of Saran wrap were placed between the cells with SPF100 or SPF15 sunscreen on them, the sunscreen provided 100% and 90% protection, respectively, in comparison to a control in

monitoring endocrine disrupting activity of unknown chemicals in water.

binding to the hER and then EREs, did not respond to triclosan.

BLYES bioassay. Since this bioassay is *luc*-based, D-luciferin must be added.

Comparison of the BLYES and BLYAS strains to their colorimetric counterparts and proofof-concept as to their utility has been established (Eldridge et al., 2007; Sanseverino et al., 2005). The BLYES and BLYAS assays are consistent with previously published yeast-based reporter assays (Sanseverino et al., 2009). The 40 - 50% variability of the EC50 values shown in Figure 3 reaffirms the suggestion that no single assay should be used to determine an absolute EC50 value but rather as a first step in estimating the hormonal activity of a chemical (Beresford et al., 2000).

Fig. 3. **A.** *S. cerevisiae* BLYES standard curve (n = 13) using 17β-estradiol. **B.** *S. cerevisiae* BLYAS standard curve (n = 13) using dihydrotestosterone as a standard. Open circle: calculated EC50 values with error bars. A 50% effective concentration (EC50) value was determined from the midpoint of the linear portion of the sigmoidal dose response curve. The mean and standard deviation values were calculated from replicate EC50 values for each standard to determine the variability between assays. **C.** Summary of EC50 values for BLYES and BLYAS strain with upper and lower limits of detection.

*S. cerevisiae* BLYES, *S. cerevisiae* BLYAS, *S. cerevisiae* BLYR, were used to assess their reproducibility and utility in screening 69, 68, and 71 chemicals for estrogenic, androgenic, and toxic effects, respectively (Sanseverino et al., 2009). This screening was part of an assessment of the United States Environmental Protection Agency's Tiered screening of chemicals for endocrine-disrupting ability. The 3-tier system includes (i) priority setting, (ii) Tier 1 screening, and (iii) Tier 2 screening. Priority setting focuses on identifying chemicals that require further testing; i.e., excluding chemicals with little or no known hormonal activity and that are generally regarded as safe. The intent of Tier I screening is to rapidly identify chemicals that interact with the estrogen, androgen, and thyroid systems while Tier 2 screenings provide a more in-depth study of how each chemical interacts with each endocrine system. In this study, EC50 values were 6.3 ± 2.4 x 10-10 M (n = 18) and 1.1 ± 0.5 x 10-8 M (n = 13) for BLYES and BLYAS, using 17β-estradiol and 5α-dihydrotestosterone

Comparison of the BLYES and BLYAS strains to their colorimetric counterparts and proofof-concept as to their utility has been established (Eldridge et al., 2007; Sanseverino et al., 2005). The BLYES and BLYAS assays are consistent with previously published yeast-based reporter assays (Sanseverino et al., 2009). The 40 - 50% variability of the EC50 values shown in Figure 3 reaffirms the suggestion that no single assay should be used to determine an absolute EC50 value but rather as a first step in estimating the hormonal activity of a

(M)

BLYES 17β-Estradiol 6.3 ± 2.4 x 10-10 5.0 x 10-9 2.5 x 10-11 BLYAS 5α-Dihydrotestosterone 1.1 ± 4.6 x 10-8 5.0 x 10-8 1.0 x 10-10 Fig. 3. **A.** *S. cerevisiae* BLYES standard curve (n = 13) using 17β-estradiol. **B.** *S. cerevisiae* BLYAS standard curve (n = 13) using dihydrotestosterone as a standard. Open circle: calculated EC50 values with error bars. A 50% effective concentration (EC50) value was determined from the midpoint of the linear portion of the sigmoidal dose response curve. The mean and standard deviation values were calculated from replicate EC50 values for each standard to determine the variability between assays. **C.** Summary of EC50 values for BLYES

*S. cerevisiae* BLYES, *S. cerevisiae* BLYAS, *S. cerevisiae* BLYR, were used to assess their reproducibility and utility in screening 69, 68, and 71 chemicals for estrogenic, androgenic, and toxic effects, respectively (Sanseverino et al., 2009). This screening was part of an assessment of the United States Environmental Protection Agency's Tiered screening of chemicals for endocrine-disrupting ability. The 3-tier system includes (i) priority setting, (ii) Tier 1 screening, and (iii) Tier 2 screening. Priority setting focuses on identifying chemicals that require further testing; i.e., excluding chemicals with little or no known hormonal activity and that are generally regarded as safe. The intent of Tier I screening is to rapidly identify chemicals that interact with the estrogen, androgen, and thyroid systems while Tier 2 screenings provide a more in-depth study of how each chemical interacts with each endocrine system. In this study, EC50 values were 6.3 ± 2.4 x 10-10 M (n = 18) and 1.1 ± 0.5 x 10-8 M (n = 13) for BLYES and BLYAS, using 17β-estradiol and 5α-dihydrotestosterone

Upper Limit of Detection (M)

Lower Limit of Detection (M)

chemical (Beresford et al., 2000).

**C**

Assay Chemical Standard EC50

**A B** 

and BLYAS strain with upper and lower limits of detection.

(DHT) over concentration ranges of 2.5 x 10-12 thru 1.0 x 10-6 M, respectively. Based on analysis of replicate standard curves, comparison to background controls, and screening a variety of chemicals, a set of quantitative rules was formulated to interpret data and determine if a chemical is potentially hormonally active, toxic, both, or neither (Sanseverino et al., 2009). The results demonstrated that these assays were applicable for Tier I chemical screening in EPA's Endocrine Disruptor Screening and Testing Program as well as for monitoring endocrine disrupting activity of unknown chemicals in water.

Additional *S. cerevisiae* bioluminescent bioreporters for estrogens and androgens have been developed using the firefly luciferase as the reporter molecule. The bioreporters of Leskinen et al., (2005) each contain the firefly luciferase gene (*lucFF*) under control of hormoneresponsive promoters. The four strains, designated BMAEREluc/ERα, BMAEREluc/ERβ, BMAEREluc/AR, and BMA64/luc were used to detect estrogens (two versions), androgens, and toxicity, respectively. This bioassay is unique in that it uses two estrogen-sensing bioreporters; one contains the alpha form of the estrogen receptor and one contains the beta form (ERα, ERβ). These bioreporters were used by Svobodova et al. (2009) to test commercially available PCB mixtures and triclosan for estrogenic and androgenic activity but did not detect any activity with these samples (estrogenic or androgenic). This lack of estrogenic response in the bioluminescent assays may be due to the different mode of action of chemicals like triclosan (Stoker et al., 2010). In a study that examined the effects of triclosan exposure on female Wistar rats, triclosan advanced the onset of puberty symptoms. Also, a combination of ethinyl estradiol (EE2) and triclosan increased uterine weight significantly more than EE2 alone while triclosan alone had no effect. Therefore the mode of action of triclosan appears to have a synergistic effect on EE2 activity in Wistar rats. This effect appears to be independent of estrogen receptor binding given that bioluminescent yeast bioassays (Svobodova et al. 2009, Eldridge et al. unpublished data), which measure binding to the hER and then EREs, did not respond to triclosan.

In addition to hormone-mimicking chemicals, several other types of contaminants are also detectable with *S. cerevisiae*-based bioluminescent bioreporters. For example, the aryl hydrocarbon-sensing strain of Leskinen et al. (2008) contains genomically integrated human aryl hydrocarbon receptor and human aryl hydrocarbon nuclear translocator genes. In addition, it carries a plasmid-encoded copy of the firefly luciferase gene (*lucFF*) that is regulated by a series of aryl hydrocarbon receptor complex (AHRC) response elements (also called dioxin response elements or xenobiotic response elements, AhREs/DREs/XREs). Aryl hydrocarbon receptor proteins interact with both their AH ligand and the nuclear translocator protein then bind to the AhRE region of the *luc*-containing plasmid, activating transcription of luciferase, similarly to the receptor-response element system present in the BLYES bioassay. Since this bioassay is *luc*-based, D-luciferin must be added.

Another *S. cerevisiae*-based bioreporter has been created to measure arsenate and also UV damage (Bakhrat et al., 2011). This strain is based on the BLYES strain of Sanseverino et al. (2005), containing a constitutive *luxCDEfrp* plasmid and a *luxAB* plasmid that has been reengineered to be under control of the UFO1 promoter, which specifically responds to DNA damage by UV light and also arsenate. The strain is able to detect very low concentrations of arsenate (1x10-12 to 1x10-6 M), which makes them useful for environmental monitoring. It was also used to evaluate the level of UV protection in commercial sunscreens. When films of Saran wrap were placed between the cells with SPF100 or SPF15 sunscreen on them, the sunscreen provided 100% and 90% protection, respectively, in comparison to a control in

Analysis of Environmental Samples with Yeast-Based Bioluminescent Bioreporters 11

Numerous methods for solid phase extraction exist but commonly a modification of United States EPA 1694 (2007) is used. Briefly, Oasis filters (Waters, Inc.) or cartridges are conditioned with methanol and water, then the sample (typically ~1 L) is passed though the membrane slowly under a small amount of pressure. Chemicals are eluted in a solvent, either singly or in combinations, such as methanol or a methanol:acetone mixture. The solvents are evaporated to dryness and may be used immediately or stored at -20oC for future use. For the BLYES/BLYAS/BLYR assays, samples are resuspended in methanol (or DMSO) such that they are 1000x concentrated compared to the original sample, e.g. 1 L of sample is concentrated, dried, and resuspended in 1 mL of solvent, yielding an effective concentration

In the bioassays, samples are serially diluted in methanol to achieve a range of concentrations (1000x-2.5x). In addition, standard chemicals (17β-estradiol (E2) for BLYES/BLYR and 5α-dihydrotestosterone (DHT) for BLYAS) are suspended in methanol at 0.01 M and then serially diluted 18 times to generate a concentration range of 4x10-7 M to 1x10-12 M for E2 and 4x10-6 M to 1x10-11 M for DHT. Samples and standards (50 µL) are then spotted into the wells of 96-well plates (Figure 4). Triplicate plates are made (one for each of

For preparation of the bioassay, each yeast strain is grown overnight at 28oC with shaking (150 rpm) in Yeast Minimal Media (YMM) without leucine or uracil (Routledge & Sumpter, 1996) to an OD600 of 1.0. Yeast strains (200 µL) are spotted into the wells of 96-well plates containing dry samples and standards, beginning the exposure. This generates a concentration range of 250x-0.625x for environmental samples, 1x10-7 M to 2.5x10-13 M for E2, and 1x10-6 M to 2.5x10-12 M for DHT. Negative controls included wells with (i) medium + cells and (ii) medium + cells + evaporated methanol, to monitor whether estrogenic or androgenic substances are present in the solvent. Plates are then placed into a plate reader (such as Perkin-Elmer Victor2 Multilabel Counter) with an integration time of 1 s/well. Bioluminescence is measured every 30-60 min for four hours. Relative light unit data (as counts per second) is plotted versus the log of concentration in SigmaPlot (or similar

For each chemical, the log of bioluminescence (counts per second) versus the log of chemical concentration (M) is plotted, generating a sigmoidal curve for hormonally active compounds. A 50% effective concentration (EC50) value is determined from the midpoint of the linear portion of the sigmoidal curve. The mean and standard deviation values are calculated from replicate EC50 values for standards to determine the variability between assays. Detection limits are determined by calculating the concentration of chemical at background bioluminescence plus three standards deviations. Toxicity is calculated as the concentration of sample that reduces the signal from the constitutively bioluminescent strain (BLYR) by 20% (IC20). For environmental samples, the concentration factor that yields 50% maximal response is considered the EC50 and when this value is divided by the EC50 for that assay's standard, estrogenic or androgenic equivalents are calculated (in terms of E2 or DHT, respectively); this determines the amount of potentially estrogenic substances that are

For samples in which DMSO is the preferred solvent, a 4% solution of DMSO is used for the serial dilutions of environmental samples and standards (by incubating the sample in a small volume of 100% DMSO for 15 minutes then adding ultra-pure water to achieve a final DMSO concentration of 4%). Next, 100 µL of sample or standard are spotted into 96-well plates along with 100 µL of yeast cells (without drying the samples/standards), yielding a

factor of 1000x. This may then be split for chemical analysis and bioassays.

three strains) and then methanol is evaporated at room temperature.

statistical software) (Figure 5).

present in a sample relative to the standard.

which samples were shielded with only Saran wrap. Studies of this type demonstrate this bioassay's usefulness on complex samples.

#### **3. Analysis of aqueous environmental samples**

For use on environmental samples, the BLYES/BLYAS/BLYR bioreporter suite is particularly well-suited. They require no substrate addition or illumination source, are inexpensive to use, and are optimized for 96-well plate formats. For water samples where OWC are typically found in the ppb range, a concentration step is necessary. Figure 4 outlines the procedures for analysis of aqueous samples. For wastewater effluents and source drinking water samples, solid phase extraction is performed to isolate and concentrate any chemical contaminants.

Fig. 4. Schematic of sample preparation and analysis. Typically, 1 L samples are collected aseptically and passed through a solid phase extraction unit. After elution by an appropriate solvent, concentrating the sample 1,000-fold, the sample is analyzed by the bioassays and/or with chemical analysis such as GC/MS or LC/MS. Typically, eight samples are analyzed on a 96-well plate, including standards and control wells (both solvent control and no treatment controls). By combining multiple plates in one assay run, numerous samples are processed at one time. Bioluminescence is monitored and recorded over time using a photon-counting system.

which samples were shielded with only Saran wrap. Studies of this type demonstrate this

For use on environmental samples, the BLYES/BLYAS/BLYR bioreporter suite is particularly well-suited. They require no substrate addition or illumination source, are inexpensive to use, and are optimized for 96-well plate formats. For water samples where OWC are typically found in the ppb range, a concentration step is necessary. Figure 4 outlines the procedures for analysis of aqueous samples. For wastewater effluents and source drinking water samples, solid phase extraction is performed to isolate and concentrate any chemical contaminants.

Fig. 4. Schematic of sample preparation and analysis. Typically, 1 L samples are collected aseptically and passed through a solid phase extraction unit. After elution by an appropriate solvent, concentrating the sample 1,000-fold, the sample is analyzed by the bioassays and/or with chemical analysis such as GC/MS or LC/MS. Typically, eight samples are analyzed on

treatment controls). By combining multiple plates in one assay run, numerous samples are processed at one time. Bioluminescence is monitored and recorded over time using a

a 96-well plate, including standards and control wells (both solvent control and no

photon-counting system.

bioassay's usefulness on complex samples.

**3. Analysis of aqueous environmental samples** 

Numerous methods for solid phase extraction exist but commonly a modification of United States EPA 1694 (2007) is used. Briefly, Oasis filters (Waters, Inc.) or cartridges are conditioned with methanol and water, then the sample (typically ~1 L) is passed though the membrane slowly under a small amount of pressure. Chemicals are eluted in a solvent, either singly or in combinations, such as methanol or a methanol:acetone mixture. The solvents are evaporated to dryness and may be used immediately or stored at -20oC for future use. For the BLYES/BLYAS/BLYR assays, samples are resuspended in methanol (or DMSO) such that they are 1000x concentrated compared to the original sample, e.g. 1 L of sample is concentrated, dried, and resuspended in 1 mL of solvent, yielding an effective concentration factor of 1000x. This may then be split for chemical analysis and bioassays.

In the bioassays, samples are serially diluted in methanol to achieve a range of concentrations (1000x-2.5x). In addition, standard chemicals (17β-estradiol (E2) for BLYES/BLYR and 5α-dihydrotestosterone (DHT) for BLYAS) are suspended in methanol at 0.01 M and then serially diluted 18 times to generate a concentration range of 4x10-7 M to 1x10-12 M for E2 and 4x10-6 M to 1x10-11 M for DHT. Samples and standards (50 µL) are then spotted into the wells of 96-well plates (Figure 4). Triplicate plates are made (one for each of three strains) and then methanol is evaporated at room temperature.

For preparation of the bioassay, each yeast strain is grown overnight at 28oC with shaking (150 rpm) in Yeast Minimal Media (YMM) without leucine or uracil (Routledge & Sumpter, 1996) to an OD600 of 1.0. Yeast strains (200 µL) are spotted into the wells of 96-well plates containing dry samples and standards, beginning the exposure. This generates a concentration range of 250x-0.625x for environmental samples, 1x10-7 M to 2.5x10-13 M for E2, and 1x10-6 M to 2.5x10-12 M for DHT. Negative controls included wells with (i) medium + cells and (ii) medium + cells + evaporated methanol, to monitor whether estrogenic or androgenic substances are present in the solvent. Plates are then placed into a plate reader (such as Perkin-Elmer Victor2 Multilabel Counter) with an integration time of 1 s/well. Bioluminescence is measured every 30-60 min for four hours. Relative light unit data (as counts per second) is plotted versus the log of concentration in SigmaPlot (or similar statistical software) (Figure 5).

For each chemical, the log of bioluminescence (counts per second) versus the log of chemical concentration (M) is plotted, generating a sigmoidal curve for hormonally active compounds. A 50% effective concentration (EC50) value is determined from the midpoint of the linear portion of the sigmoidal curve. The mean and standard deviation values are calculated from replicate EC50 values for standards to determine the variability between assays. Detection limits are determined by calculating the concentration of chemical at background bioluminescence plus three standards deviations. Toxicity is calculated as the concentration of sample that reduces the signal from the constitutively bioluminescent strain (BLYR) by 20% (IC20). For environmental samples, the concentration factor that yields 50% maximal response is considered the EC50 and when this value is divided by the EC50 for that assay's standard, estrogenic or androgenic equivalents are calculated (in terms of E2 or DHT, respectively); this determines the amount of potentially estrogenic substances that are present in a sample relative to the standard.

For samples in which DMSO is the preferred solvent, a 4% solution of DMSO is used for the serial dilutions of environmental samples and standards (by incubating the sample in a small volume of 100% DMSO for 15 minutes then adding ultra-pure water to achieve a final DMSO concentration of 4%). Next, 100 µL of sample or standard are spotted into 96-well plates along with 100 µL of yeast cells (without drying the samples/standards), yielding a

Analysis of Environmental Samples with Yeast-Based Bioluminescent Bioreporters 13

samples in which targeted estrogens were not detected by chemical analysis, the yeast bioassay also did not detect any estrogenic activity. However, in some samples from different surface water intake points, often yeast bioassays detected estrogenic activity at levels that chemical analysis data did not predict. For example, BPA detected at 3.53 ng/L (according to chemical analysis) is not a sufficient concentration to elicit an estrogenic response. However, the same sample elicited a response equivalent to 0.7 ng/L of E2 equivalents (according to the yeast bioassay). This suggests that *S. cerevisiae* BLYES was responding to a) something that was not recognized by chemical analysis, b) by-products of the targeted chemicals, or c) that a mixture effect is causing a synergistic estrogen response. The reader is cautioned that these assays should be a first determination of estrogenic activity. *S. cerevisiae* does not have an endocrine system and cannot explicitly identify endocrine disruptors. Advanced testing with alternate assays (i.e. mammalian-based assays)

Fig. 6. Surface water samples from Brazil assessed with the BLYES bioassay. Surface water samples were solid phase extracted and then processed with the *S. cerevisiae* BLYR and

Alvarez et al. (2009) used BLYES for the analysis of Potomac River water samples in a study on the reproductive health of bass. The authors criticized the collection of single grab samples, in favor of using passive samplers to concentrate contaminants. They examined extracts from passive samplers that had been deployed for 31 days in the Potomac River and its tributaries, which receive significant amounts of flow from WWTP effluents. Samples

BLYES.

should be used for confirmation of endocrine-disrupting activity.

final DMSO concentration of 2% in all wells. Negative controls should consist of wells with (i) medium + cells and (ii) medium + cells + DMSO to monitor whether DMSO is toxic to yeast cells and whether the solvent contains potentially estrogenic substances.

Fig. 5. Yeast assay data using environmental samples. The graphs show the responses of the yeast strains *S. cerevisiae* BLYR and BLYES in response to the 17β-estradiol standard and serial dilutions of a solid phase extracted sample. The sample consisted of surface water from the Cotia River in Brazil, which has a high concentration of estrogenic substances present (1.2 ng E2 equivalents/L) and exhibits marked toxicity. Analysis of surface and groundwater are of particular interest to regulatory agencies (and the public) because they are source waters for drinking water treatment plants.

Using this bioassay, surface water samples were surveyed from the U.S. and Brazil (Eldridge-Umbuzeiro, unpublished data, Figures 5 and 6), with both studies determining that the estrogen-sensing strain detects more estrogen-like activity than predicted through chemical analysis alone. This is expected however, given that chemical analysis targets certain contaminants and cannot be expected to screen samples for all known estrogens. In addition, it is relatively unknown if/how chemicals act synergistically to promote estrogenor androgen-like activity. The assay can provide a clear evaluation on the levels of potential estrogenicity in monitoring studies of surface water samples as can be seen from Figure 6 (unpublished data). The levels varied from 0.01 to 19.3 ng/L of E2 equivalents per liter of water. In this particular case, the river water that was monitored was from Brazil, with the highest levels of pollution expected to occur in the dry season (corresponding to June-October).

In Jardim et al. (2011), surface water samples from Brazil were also examined. The samples were collected from sites classified by the São Paulo State Environmental Agency (CETESB) as excellent, good, medium, fair, and poor. Both bioassays and chemical analysis were performed on samples following solid phase extraction. The authors targeted estrone (E1), 17β-estradiol (E2), ethinyl estradiol (EE2), estriol (E3), bisphenol A (BPA), 4-n-octophenol (OP), and 4-n-nonylphenol (NP) in their chemical analysis and used the estrogen-sensing BLYES as the bioassay. From this data, the authors determined that the bioassay data is not fully explained by the amount and strength of the detected estrogens. For example, the highest estrogenic response determined by the yeast bioassay (BLYES) was also determined to have the highest concentrations of estrogen by chemical analysis. Also, in drinking water

final DMSO concentration of 2% in all wells. Negative controls should consist of wells with (i) medium + cells and (ii) medium + cells + DMSO to monitor whether DMSO is toxic to

Fig. 5. Yeast assay data using environmental samples. The graphs show the responses of the yeast strains *S. cerevisiae* BLYR and BLYES in response to the 17β-estradiol standard and serial dilutions of a solid phase extracted sample. The sample consisted of surface water from the Cotia River in Brazil, which has a high concentration of estrogenic substances present (1.2 ng E2 equivalents/L) and exhibits marked toxicity. Analysis of surface and groundwater are of particular interest to regulatory agencies (and the public) because they

Using this bioassay, surface water samples were surveyed from the U.S. and Brazil (Eldridge-Umbuzeiro, unpublished data, Figures 5 and 6), with both studies determining that the estrogen-sensing strain detects more estrogen-like activity than predicted through chemical analysis alone. This is expected however, given that chemical analysis targets certain contaminants and cannot be expected to screen samples for all known estrogens. In addition, it is relatively unknown if/how chemicals act synergistically to promote estrogenor androgen-like activity. The assay can provide a clear evaluation on the levels of potential estrogenicity in monitoring studies of surface water samples as can be seen from Figure 6 (unpublished data). The levels varied from 0.01 to 19.3 ng/L of E2 equivalents per liter of water. In this particular case, the river water that was monitored was from Brazil, with the highest levels of pollution expected to occur in the dry season (corresponding to June-

In Jardim et al. (2011), surface water samples from Brazil were also examined. The samples were collected from sites classified by the São Paulo State Environmental Agency (CETESB) as excellent, good, medium, fair, and poor. Both bioassays and chemical analysis were performed on samples following solid phase extraction. The authors targeted estrone (E1), 17β-estradiol (E2), ethinyl estradiol (EE2), estriol (E3), bisphenol A (BPA), 4-n-octophenol (OP), and 4-n-nonylphenol (NP) in their chemical analysis and used the estrogen-sensing BLYES as the bioassay. From this data, the authors determined that the bioassay data is not fully explained by the amount and strength of the detected estrogens. For example, the highest estrogenic response determined by the yeast bioassay (BLYES) was also determined to have the highest concentrations of estrogen by chemical analysis. Also, in drinking water

are source waters for drinking water treatment plants.

October).

yeast cells and whether the solvent contains potentially estrogenic substances.

samples in which targeted estrogens were not detected by chemical analysis, the yeast bioassay also did not detect any estrogenic activity. However, in some samples from different surface water intake points, often yeast bioassays detected estrogenic activity at levels that chemical analysis data did not predict. For example, BPA detected at 3.53 ng/L (according to chemical analysis) is not a sufficient concentration to elicit an estrogenic response. However, the same sample elicited a response equivalent to 0.7 ng/L of E2 equivalents (according to the yeast bioassay). This suggests that *S. cerevisiae* BLYES was responding to a) something that was not recognized by chemical analysis, b) by-products of the targeted chemicals, or c) that a mixture effect is causing a synergistic estrogen response. The reader is cautioned that these assays should be a first determination of estrogenic activity. *S. cerevisiae* does not have an endocrine system and cannot explicitly identify endocrine disruptors. Advanced testing with alternate assays (i.e. mammalian-based assays) should be used for confirmation of endocrine-disrupting activity.

Fig. 6. Surface water samples from Brazil assessed with the BLYES bioassay. Surface water samples were solid phase extracted and then processed with the *S. cerevisiae* BLYR and BLYES.

Alvarez et al. (2009) used BLYES for the analysis of Potomac River water samples in a study on the reproductive health of bass. The authors criticized the collection of single grab samples, in favor of using passive samplers to concentrate contaminants. They examined extracts from passive samplers that had been deployed for 31 days in the Potomac River and its tributaries, which receive significant amounts of flow from WWTP effluents. Samples

Analysis of Environmental Samples with Yeast-Based Bioluminescent Bioreporters 15

chlorophenol caused a response. This is in contrast to the results obtained with the Yeast Estrogen Screen (YES) colorimetric bioreporter, which detected estrogenic activity with all the tested chemicals except chlorophenols. One reason for this difference may be the different length of exposure time between the two bioassays. The YES was incubated with chemicals for three days whereas the Leskinen strains were only incubated with the chemicals for 2.5 h prior to sample processing. It is possible that over three days' time, the PCBs may have oxidized (yeast are incubated in aerobic conditions) to forms that are more likely estrogenic. Indeed, hydroxylated PCBs have been demonstrated to harbor estrogenic activity (Korach et al., 1988; Schultz, 2002; Schultz et al., 1998). Interestingly, using the bioluminescent androgen-sensing bioreporter (BMAEREluc/AR), androgenic activity was detected with the commercial PCB mixture, but not with chlorobenzoic acids, chlorophenols, or triclosan. Triclosan has been demonstrated to have no activity with the

*Saccharomyces cerevisiae*-based bioluminescent bioreporters offer excellent opportunities beyond bacterial bioreporters for rapid analysis of chemicals with human and environmental significance. Expression of the bacterial *lux* cassette in a lower eukaryote offers many opportunities not only for high-throughput screening systems but also bioprocess monitoring, diagnostic applications, fungal gene expression analysis, and *in vivo* sensing of fungal infections (Gupta et al., 2003). Expression in *S. cerevisiae* has led to advances in transferring this system to mammalian cell lines (Close et al., 2010; Patterson et

The advantages for detection of endocrine-disrupting chemicals in water by *S. cerevisiae lux*based bioreporters are numerous including accuracy, ease of use, not expensive, and amenable to automation in performing and collection of data. In addition to screening aqueous samples, BLYES, BLYAS, and BLYR, and other variants described in the literature are useful for Tier I screening as proposed by the EPA, analysis of wastewater influent and effluent, chemical leaching from manufactured products, for example. In fact, the State Environmental Agency of São Paulo (CETESB) in Brazil is considering using the *S. cerevisiae* BLYES bioassay for routine monitoring of surface and ground water samples for the presence of potentially estrogenic substances. Two of the authors (M.E. and G.S.) have begun routine monitoring of wastewater treatment plant effluents from a treatment facility in TN as well as screening 250 water samples across the state of Tennessee in a broad

Ideally, detection of potential endocrine disruptors (or any other chemical of interest) by bioluminescent bioreporter strains would be coupled to remote detection systems for continuous real-time monitoring. Bioluminescent bioreporter integrated circuits fuse reporter cells to an integrated circuit containing a photodetector (e.g. Sayler et al., 2001; Nivens et al., 2004; Sayler et al., 2004). These devices could be distributed in networks and coupled with wireless communications would send signals indicating the presence/absence of chemical contaminants. Roda et al. (2011) have developed a device that couples estrogenor androgen-sensing *S. cerevisiae* expressing firefly bioluminescence to fiber optics with detection by a CCD sensor, yielding a fully functional biosensor. While this device resulted in strains whose detection limit was approximately 10-fold higher than bioassays performed in the lab and was larger than previously reported remote detection systems, it does

BLYES and BLYAS bioassays as well (data not shown).

**4. Future applications** 

al., 2005).

survey.

were collected once yearly, for two years (2005-2006), both upstream and downstream of known WWTP discharge sites. They also performed both chemical analysis (targeting E1, E2, EE2, and E3) and bioassays (using BLYES and BLYR). They were able to detect potentially estrogenic compounds at levels statistically different than the field blanks. Levels of E2 equivalents were detected in the nanomolar range. The authors were able to measure estrogenic responses with BLYES but they were not able to detect a seasonal difference in estrogenicity (some chemicals were detected seasonally via chemical analysis but others were not) though it is unclear whether there was no seasonal effect or whether the estrogens were detected at such a low concentration that a conclusion cannot be drawn.

In both studies (Alvarez et al., 2009; Jardim et al., 2011), the expected response with bioassays was lower than the actual response determined with the bioassay. Expected responses are calculated by multiplying a chemical's concentration (determined through chemical analysis) by its potency relative to a reference estrogen, such as E2. It is expected that if all contaminating estrogenic molecules are detected by chemical analysis then the expected responses should match actual bioassay responses. However, it is difficult to anticipate (and therefore target) all possible endocrine-active contaminants that are present in environmental samples. In addition, prediction of the effects of mixtures of chemicals, especially at low concentrations, has proven to be problematic. Moreover, bioassays are likely to detect metabolites of estrogenic chemicals, as long these molecules continue to interact with the human hormone receptor/response element-sensing systems. Given these reasons, it is natural to expect that chemical analysis is unlikely to ever fully predict actual bioassay responses.

The androgen-sensing strain of Leskinen et al. (2005) has been used to monitor wastewater before and after treatment in wastewater treatment plants in several cities in Italy (Michelini et al. 2005). It was determined that both samples (pre- and post-treatment) contained chemicals with androgenic activity, however treatment decreased this activity. They determined that approximately 30% of androgenic activity was typically removed but occasionally activity was reduced by 90%. They attributed the decreased activity to the presence of carbon-based filters, which should bind chemicals, thereby removing them from wastewater. This study illustrates the effectiveness of yeast-based bioreporters for the rapid analysis of samples before and after water treatment. It also demonstrates that wastewater treatment does not necessarily remove chemicals associated with potential endocrine disrupting activity.

In addition, the strains of Leskinen et al. (2005) (BMAEREluc/ERα, BMAEREluc/ERβ, BMAEREluc/AR, and BMA64/luc) were used to test several lotion samples, as a simulation of using the strains on complex sample matrices. Five of the seven lotion samples demonstrated estrogenic activity, even at dilutions as low as 1:175. The authors attributed this activity to parabens present in the lotions, given that samples with no parabens were not estrogenic but samples with mixtures of parabens were. The authors state that parabens are present in many cosmetic products and are generally considered safe (Soni et al., 2002), despite having been demonstrated to produce an estrogenic response (Routledge et al., 1998) and being present in breast cancer tumors (Darbre et al., 2004). No androgenic activity was found for any of the samples.

More recently, Svobodova et al. (2009) examined the endocrine disrupting potential of a commercial PCB mixture (Delor 103) and a series of potential PCB degradation metabolites (chlorobenzoic acids and cholorophenols). The authors did not detect any estrogenic activity with any of the chemicals or mixtures tested using bioluminescent yeast, except that 5 mg/L chlorophenol caused a response. This is in contrast to the results obtained with the Yeast Estrogen Screen (YES) colorimetric bioreporter, which detected estrogenic activity with all the tested chemicals except chlorophenols. One reason for this difference may be the different length of exposure time between the two bioassays. The YES was incubated with chemicals for three days whereas the Leskinen strains were only incubated with the chemicals for 2.5 h prior to sample processing. It is possible that over three days' time, the PCBs may have oxidized (yeast are incubated in aerobic conditions) to forms that are more likely estrogenic. Indeed, hydroxylated PCBs have been demonstrated to harbor estrogenic activity (Korach et al., 1988; Schultz, 2002; Schultz et al., 1998). Interestingly, using the bioluminescent androgen-sensing bioreporter (BMAEREluc/AR), androgenic activity was detected with the commercial PCB mixture, but not with chlorobenzoic acids, chlorophenols, or triclosan. Triclosan has been demonstrated to have no activity with the BLYES and BLYAS bioassays as well (data not shown).
