High-throughput screening assay for the environmental water samples using cellular response profiles

Ecotoxicology and Environmental Safety 114 (2015) 134–142

Contents lists available at ScienceDirect

Ecotoxicology and Environmental Safety journal homepage: www.elsevier.com/locate/ecoenv

High-throughput screening assay for the environmental water samples using cellular response profiles Tianhong Pan a,b,n, Haoran Li a, Swanand Khare b,f, Biao Huang b,n, Dorothy Yu Huang c, Weiping Zhang d, Stephan Gabos e a

School of Electrical & Information Engineering, Jiangsu University, Zhenjiang, Jiangsu 212013, China Department of Chemical & Materials Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2G6 c Alberta Centre for Toxicology, University of Calgary, Calgary, Alberta, Canada T2N 4N1 d Alberta Health, Edmonton, Alberta, Canada T5J 1S6 e Division of Analytical and Environmental Toxicology, University of Alberta, Edmonton, Alberta, Canada T6G 2G3 f Department of Mathematics, Indian Institute of Technology, Kharagpur, West Bengal 721302, India b

art ic l e i nf o

a b s t r a c t

Article history: Received 8 July 2014 Received in revised form 19 January 2015 Accepted 20 January 2015 Available online 28 January 2015

Chemical and physical analyses are commonly used as screening methods for the environmental water. However, these methods can only look for the targeted substance but may miss unexpected toxicants. Furthermore, the synergistic effects of mixture cannot be detected. In order to set up the assay criteria for determining various biological activities at a cellular level that could potentially lead to toxicity of environmental water samples, a novel test based on cellular response by using Real-Time Cellular Analyzer (RTCA) is proposed in this study. First, the water sample is diluted to a series of strengths (80%, 60%, 40%, 30%, 20% and 10%) to get the multi-concentration cellular response profile. Then, the area under the cellular response profile (AUCRP) is calculated. Comparing to the normal cell growth of negative control, a new biological activity index named Percentage of Effect (PoE) has been presented which reflects the cumulative inhibitory activity of cell growth over the log-phase. Finally, a synthetical index PoE50 is proposed to evaluate the intensity of biological activities in water samples. The biological experiment demonstrates the effectiveness of the proposed method. & 2015 Elsevier Inc. All rights reserved.

Keywords: Cytotoxicity Biological activity Area under the cellular response profile (AUCRP) Concentration–response curve High-throughput screening (HTS)

1. Introduction Around the world, exposure to chemicals, whether acute poisoning through ingestion of large doses or chronic contact with low levels of contaminants, is a public health concern. Poisoning placed second behind automobile accidents as the leading cause of injury-related death in many developed countries. Certain poisonings (e.g., pesticides) occur most frequently in children, 40–60% of those affected being are under the age of six (WHO, 2010). Chronic adverse effects due to exposure to hazardous chemicals are more, subtle but they have been long recognized by occupational health studies conducted on workers. Important occupational toxicities such as neuro-, reproductive-, and immunotoxicity as well as mutagenicity and carcinogenicity are now well documented (Soldán and Badurová, 2013; n Corresponding authors at: Department of Chemical & Materials Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2G6. E-mail addresses: [email protected] (T. Pan), [email protected] (H. Li), [email protected] (S. Khare), [email protected] (B. Huang), [email protected] (D. Yu Huang), [email protected] (W. Zhang), [email protected] (S. Gabos).

http://dx.doi.org/10.1016/j.ecoenv.2015.01.020 0147-6513/& 2015 Elsevier Inc. All rights reserved.

Radić et al., 2013). Evidence is also accumulating that a range of adverse effects and even chronic diseases can occur in the general population at very low chemical concentrations after prolonged periods of time. Most of the population exposures are primarily through food, air, consumer products and water. Conventional toxicity testing for environmental water monitoring has been performed by characterizing the specific substance, quantifying its level and then comparing it to known regulatory guidelines. A wide range of analytical chemistry methods are being used to achieve this goal (Kaza et al., 2007; IAEA, 2009; Comerton et al., 2009). However, there are many substances in the environmental water samples and only well-known toxic substances can be determined using these assay techniques. Thus it is difficult to perform toxicity analysis for substances which are new or not available in the library of the well-known toxic substances. Additionally, the use of only chemical analyses cannot predict the actual human health hazard, because they are limited by the sensitivity and determination of single compound (Lah et al., 2005; Elad et al., 2011). The environmental contaminants are usually present as mixtures and their synergistic effects cannot be obtained by simply summing up the toxicity from individual substances. Currently, little is known about the mixed

T. Pan et al. / Ecotoxicology and Environmental Safety 114 (2015) 134–142

cytotoxic effects caused by multiple environmental chemicals and their related human health risks. There is a great need for novel tests that can be used for the screening and monitoring of a wide range of toxic contaminants in the environment; then if desired, more detailed and expensive laboratory analysis can be performed. To achieve a realistic estimation of human risks caused by the environmental water contaminations, the first thing to know is their toxic effects. The biotest is commonly used for the toxicity assessment in the literature (Faria et al., 2007; Mankiewicz-Boczek et al., 2008; Dragun et al., 2009). Lah et al. (2005) implemented the Comet assay and Ames test to monitor the genotoxicity in drinking water. Two human cell lines and protozoa cells were treated with nonconcentrated and 50  concentrated drinking water respectively, and the degree of nuclear DNA damage was proposed to assess the genotoxicity. Kaza et al. (2007) performed a battery of microbiotests to assay the toxicity of water samples from the rivers in Central Poland. As a result, the percentage effect depending on the effect criterion of the respective assay is proposed to rank the water samples into five hazard classes. Soldán and Badurová (2013) used exotoxicological approach to screen the risk of chronic effect of surface water pollution. The risk of toxicity and genotoxicity was investigated to detect the biological impact. Sansom et al. (2013) employed six fish cell lines in 24h/96h viability assays (EC50 and LC50) for rapid fluorometric assessment of cellular integrity and functionality. The physicochemical composition of the tested waters confirmed that the proposed approach can be a biologically relevant tool in the initial assessment of the water toxicity. Although the mentioned biological methods can be used for the evaluation of environmental water contamination, they either focused on the detection of special substances or used the end-point estimation only. Further, the dynamic information during organism (such as Daphnia magna, fish cells) growth was ignored when they were exposed to the environmental water samples. To this end, a Real-Time Cell Analyzer (RTCA) developed by the ACEA Biosciences Inc. (San Diego, USA) presents a real time monitoring platform to record the dynamic process of cell proliferation and changes induced by celltoxicant interaction (Xing et al., 2012). Furthermore, the RTCA MP Station configured with six E-Plates (96 wells on one E-Plate) can improve the assay throughput (Roche, 2011). The basic principle of the RTCA is to monitor the changes in electrode impedance induced by the interactions between testing cells and electrodes. The presence of the cells leads to an increase in the electrode impedance: the more the cells attached to the sensor, the higher the impedance that could be picked up by RTCA. The dynamic data generated by the RTCA reflects cell proliferation. However, measuring the biological activity from this rich dynamic data is a challenge. The traditional indices (such as EC50, IC50, and LC50) can also be used to achieve the target, but these indices are largely dependent on the incubation time. Different assayed time points may lead to different values of traditional indices (Pan et al., 2013a). The influence of incubation time can lead to questions about which time point provides the most scientifically valid results. In this study, we develop a new high throughput screening method based on the RTCA that can be used to monitor environmental water for biological activities at a cellular level that could potentially lead to toxicity. Human cell lines are used to generate data that may be useful for human health risk assessment. The collected water samples are diluted to several concentrations and the human cells are exposed to each concentration of each individual water sample. As a result, the multi-concentration and time-dependent cellular response profile is recorded for each water sample. In order to quantify the cell growth effects, the area under the cellular response profile (AUCRP) is developed to evaluate the extent of exposure to each concentration of each individual sample. By integrating over time rather than looking at multiple endpoint measurements, a more accurate and robust estimate of the overall exposure to the chemical of various concentrations is obtained, which can describe the intensity of cellular level biological

135

activities due to the environmental water. Compared to the negative control (without exposure to the collected water sample), a percentage of effect (PoE) index is proposed for identifying and analyzing the level of water contamination.

2. Materials and methods 2.1. Materials 2.1.1. Cell lines Four cell lines (i.e. A549, ACHN, HepG2, and SK-N-SH) were purchased from the American Type Culture Collection (ATCC) and maintained in 37 °C incubators containing 5% CO2. The cells were amplified and frozen in aliquots to ensure that the same source of cells was used for the investigation. 2.1.2. Controls Controls are subjects closely resembling the experimental subjects but not receiving the treatment, thereby serving as a comparison group when treatment results are evaluated. Positive control: Arsenic in Alberta's ground waters has been a concern since the early 1990s (Surveillance, 2000). The ground waters were monitored and arsenic was found to be at relatively high levels in some of the ground water samples. Therefore, arsenic III and a mixture of the trace elements were chosen as positive controls for the cytotoxicity assay, in which the affected result can be predicted. Negative control: A negative control is a group that has not been administered the drug of interest. In this experiment, negative control contains the target cells, the culture medium, and the maximum concentration of the solvent used in dissolving chemicals, where no phenomena are expected. Here, H2 O was included as the negative control for environmental water analysis. 2.1.3. Water samples As examples for testing the proposed method, three types of samples were collected from a specific well, lake or storm pond, which does not imply any similarity of these samples as other wells, lakes or storm ponds of the region (Pan et al., 2013a). Well samples: Private domestic wells are the drinking and household water sources for rural families. The samples were analyzed for routine chemistry, trace elements, as well as cytotoxicity. Storm pond samples: Storm water ponds are frequently built into urban areas in North America to provide storm water flow control and improve water quality. The suspended sediments are also collected in storm water, which are often found in high concentrations in storm water due to upstream construction and sand applications to roadways. Storm water ponds could be chemical soups of pesticides, fertilizers, pet wastes, oil, grease and other contaminants. The samples were analyzed for routine chemistry, trace elements, pesticides, VOCs, as well as cytotoxicity. Lake samples: Water samples were collected from lakes across Alberta. The samples were analyzed for routine water chemistry, trace elements, total microcystins, as well as cytotoxicity. Dilution rule: Each water sample was diluted to a series of strength (80%, 60%, 40%, 30% 20% and 10%) to get the multi-concentration cellular response profile. The advantage of this method is that it can achieve the concentration–response curve which is similar to the chemical compound exposure. 2.2. Methods 2.2.1. RTCA HT system and ecotoxicological testing The xCELLigence RTCA HT system was used as the platform to facilitate this study. The system has been developed by the ACEA Biosciences Inc. (San Diego, USA) in the 96  well plate format. It

136

T. Pan et al. / Ecotoxicology and Environmental Safety 114 (2015) 134–142

measures the electronic impedance as the results of cell contacting biocompatible microelectrode fabricated at the bottom of well of a tissue culture plate (E-Plate). Using ACEA's proprietary algorithm, the impedance value is converted to Cell Index (CI), which closely reflects not only cell growth and cell death, but also cell morphology (attachment and spreading/ shrinking) and adhesion. As the measurement is non-invasive and label free, the system can continuously monitor the cells from the time when cells are seeded. Quality control of the cells (e.g. doubling time, attachment) becomes a built-in feature of the system. More importantly, unlike end-point assays, cell responses from minutes to days after substance addition are recorded, which ensure that no meaningful time points are missed for analysis. In this study, we consider xCELLigence Multi-plates (MP) system with six 96  E-plate format for the experiment. Two water samples with six dilutions are arranged in the two sides of each 96  E-plate, and the positive control and the negative control are arranged in the middle of E-plate. Each dilution of each water sample has been repeated four times. As a result, 12 water samples are monitored by the RTCA HT system in each experiment. First, 100 μL of the culture medium was added to each well of the 96-well E-Plate. The background check was run without the cells and a baseline cell index was set for the assay. In the interim, the target cells were detached from T75 tissue culture flask, one cell line at a time, and counted by a hemacytometer. After the baseline was measured, the E-plate was disconnected from workstation and 100 μL of the cells was seeded in each well of the 96-well E-plate at the desired density using a 12-well multichannel pipetteman. The E-plate was incubated at room temperature for 20–30 min for the cells to settle, and then returned to the workstation in the incubator. The plated cells were monitored every hour for 20–24 h. At the end of the incubation, cells were exposed to either control chemicals or environmental samples. The cell index was monitored every hour for 96 h. The schematic of RTCA and testing procedure was shown in Fig. 1.

cellular response profiles upon treatment with biologically active compounds. Based on measured impedance, a dimensionless parameter termed Cell Index (CI) is derived to provide quantitative information about the physiological and pathological responses of the living cells to a given substance:

2.2.2. Cell index Continuously monitored cell substrate impedance in real-time has been considered to produce very specific time-dependent

2.2.3. Area under the cellular response profile (AUCRP) It is well known that the cells in culture media alone (negative controls), without exposure to test substances, demonstrate

⎡ R cell (f ) ⎤ l CI = max ⎢ − 1⎥ l = 1, … , L ⎢ ⎣ R b (fl ) ⎦⎥

(1)

where Rb (fl ) and Rcell (fl ) are the frequency-dependent electrode impedance (resistance) without and with cells present in the wells respectively, and L is the number of the frequency points at which the impedance is measured. In our analysis, normalized CI (NCI) is considered. NCI is the ratio of cell index at a particular point of time to the cell index at the time of exposure, as shown in the following equation (Pan et al., 2013b):

NCI(k) =

CI(k) , CI(0)

k = 1, 2, …, K

(2)

where k is the sampling sequence number and K is the total number of sampling points. Example is shown in Appendix A. The real-time concentrationspecific growth reflects the growth pattern change in comparison with the negative control. Some of the curves are close to the negative control (S4 in Appendix A), indicating no clear biological activities of the water sample. Others depart from the negative control, which demonstrate the existence of biological activities at a given concentration level (S11 in Appendix A). The observed TCRPs reflect the dynamic evolution of cell proliferation status. The patterns show the following properties:

 Time kinetic: NCI(k) changes with exposure time t(k).  Concentration-dependent: The TCRPs of test substances vary at different concentration levels.

Fig. 1. Schematic of real-time cell analyzer (RTCA) and testing procedure.

T. Pan et al. / Ecotoxicology and Environmental Safety 114 (2015) 134–142

137

typical cell growth curve with four phases: lag-phase, log-phase, plateau-phase and decline-phase (Davis, 2011). The number of viable cells declines due to the natural cycle exhibited by cells and a shortage of nutrient supplements at the decline-phase. If the decline-phase is included into the assay, the results will not be deemed credible, since there would be an uncertainty in the cause of death. Fig. 2 shows the TCRPs of negative control in six E-Plate. It can be seen that the time of decline-phase in those TCRPs is greater than 79 h. To keep the toxicity assay within the cell proliferation boundary and to have the same evaluation baseline, the exposed time used for the assessment is selected as 72 h. As shown in Fig. 2, the shape of TCRP demonstrates the cell population changes over the exposure time. Therefore, the area under the TCRP reflects the level of biological activities of the water sample. In order to evaluate the overall activities level, integral of NCI is proposed as follows. AUCRPi, j , the area under the jth TCRP of ith water sample, can be expressed as 72

Fig. 2. HepG2 TCRPs of negative control of six E-Plates in one experiment. The black circle denotes the maximum NCI of negative control in each E-Plate.

a

AUCRPi, j =

∑

NCIi, j (k)(t [k] − t [k − 1]) 2

k=2

(3)

where NCIi, j (k) is the cell index of jth concentration of ith water sample, i = 1, 2, … , 12; j = 1, 2, … , 6, and t is the sampling time. Here, the trapezoid rule is used to numerically approximate the integration. The AUCRPi, j evaluates the extent of cell growth inhibition or cell killing with treatment under this concentration. Similarity, the area under the TCRP of negative control can also be calculated as 72

AUCRPm, NC =

∑

NCIm, NC (k)(t [k] − t [k − 1]) 2

k=2

(4)

where NCIm, NC (k) is the cell index of negative control in mth E-plate, m = 1, 2, … , 6.

b

2.2.4. Biological activity index Although the AUCRPi, j evaluates the biological activity level when cells are exposed to a water sample with j different dilutions, it does not compare to the negative control in each E-Plate. In order to get the intensity of biological activity of each water sample, a new index termed as Percentage of Effect (PoE) is presented in this study as in the following equation:

PoEi, j =

AUCRPi, j AUCRPm, NC

× 100%

(5)

where PoEi, j is the activity index of ith water sample with jth concentration. AUCRPm, NC is the AUCRP of negative control in mth E-Plate, each E-Plate including two water samples. The value of PoE directly measures the biological activity of water sample with the selected concentration. It reflects the accumulative biological effect within the definitive time range. In order to better comprehend exposure-related effects, a concentration–response curve based on PoE is introduced. The curve measures the magnitude of an effect (for example, cell viability) across a range of concentrations, and is typically represented graphically with log-linear plots (Sichani et al.). Based on the curve shape, the power-based model is selected to describe the relationship between the PoE and water sample concentrations. The main reason behind the choice of the power function is that it has convex downward shape and produces straight lines in log–log space: Fig. 3. Toxicity assessment for positive control (HepG2). (a) TCRPs of arsenic III with 0.33 mM, 0.11 mM, 12.35 μM, 0.46 μM . (b) The PoE model of S11 is PoE(x) = 15.34 × x1.96 + 93.83 and PoE50 = 33.11 μM.

PoE(x) = p1 × x p2 + p3

(6)

where x denotes the concentration, p1, p2 and p3 are parameters independent of water sample concentrations and PoE(x) denotes

138

T. Pan et al. / Ecotoxicology and Environmental Safety 114 (2015) 134–142

a

b

c

d

Fig. 4. Water toxicity assessment (HepG2). (a) Histogram of activity index PoE for 12 water samples with 6 dilutions. (b) Cool colormap of activity index PoE for 12 water samples with 6 dilutions. (c) The PoE model of S4 is PoE(x) = 1878.51 × x−0.001 − 1783.33 and PoE50 = 7.02 × 109 . (d) The PoE model of S11 is PoE(x) = 1.65 × x−1.43 + 5.49 and PoE50 = 0.1.

Table 1 PoE50 values and Microtox comparison. Samples

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12

Description

Process water Tap water Tap water Single distilled water Double distilled water Ground water Ground water Ground water Surface water Surface water Surface water Surface water

Microtox (IC50)

PoE50 A549

ACHN

HepG2

SK-N-SH

– 1.2  10210 1.7  105 3.7  105 – – 6.9  105 1.3  1095 0.04 – 10.7 1.4  109

64.3 1.8  102 – – 1.7  107 4.3  104 1.5  102 6.5  102 0.11 16.5 0.27 –

41.6 1.9  106 – 7.0  109 1.2  1023 4.4  102 – – 0.12 3.3  10103 0.10 6.2  106

– – – – – 2.6  1029 9.3  1014 4.3  1011 0.11 – 0.22 1.2  1014

71 >91 >91 >91 >91 >91 >91 >91 1.1 >91 56 >91

– means that there is no real solution for PoE50, which reflects no toxicity in this water sample.

the PoE value when the cells are treated with x concentration of a water sample. The regression analysis to estimate the parameters was done using “nlinfit” function provided by the software MATLAB (Mathworks, Natick, MA, USA).

The concentration–response curve describes the relationship between the level of exposure and water sample concentration. If the water sample has biological activity, then the curve regressed as in Eq. (6) will decline faster with higher values of

T. Pan et al. / Ecotoxicology and Environmental Safety 114 (2015) 134–142

139

The concept of PoE50 can be generalized to PoEn, i.e. n% of the PoE value in reference to the untreated cell response over the logphase, where n is any value between 0 and 100. The procedures for screening the water samples are summarized as follows: Step 1: Screen the unreasonable cellular responses using the criteria with m ± 2δ . Step 2: Get the median value of each quadruplicate cellular responses. Step 3: Calculate the AUCRPi, j , AUCRPm, NC and PoEi, j using the trapezoid algorithm. Step 4: Fit the concentration–response curve and determine the parameters p1, p2, and p3. Step 5: Calculate the activity index PoE50 by using the determined p1, p2, and p3.

a

3. Results and discussions 3.1. Results

b

Fig. 5. The reproducibility and repeatability assessment (HepG2). (a) Inter/Intraplate reproducibility assessment for negative control. The boxes at P1 to P6 are the CV% of six E-Plates in this experiment. The box at the all-plates is the CV% calculated by using the six average of four TCRPs in each E-Plate (shown in Fig. 5). (b) The repeatability analysis for S11 in three experiments. The average of the six coefficients of variation (CV%) is 2.04% which is acceptable. Besides the average CV%, theCV% of PoE50 is 10.16% which is very small.

concentrations. On the other hand, if the water sample has no biological activity, then the curve in Eq. (6) remains flat at a certain value. In order to quantify biological activity assessment, a new activity index, PoE50, is defined as the concentration of the water sample that results in a PoE value at 50% of the PoE value found from the untreated cells over the log-phase. From Eq. (6), PoE50 can be calculated as follows:

⎛ 50 − p ⎞1/ p2 3 ⎟⎟ PoE50 = ⎜⎜ ⎝ p1 ⎠

(7)

If the value of PoE50 is between 0 and 1, we have a reason to believe that the water sample exhibits biological activity. The smaller the PoE50, the higher the biological activity of water sample. As a result, PoE50 can be used as an index to screen the water sample. This analysis can then be followed by a more detailed and expensive laboratory analysis if desired.

Data from 6 batches of experiments and 12 water samples were selected to validate the proposed method. In those experiments, the HepG2 cell line was repeated three times to confirm the reproducibility of the bio-experiment, and A549/ACHN/SK-N-SH cell lines were tested only one time. The 12 water samples included one sample of process water, two samples of tap water, two samples of distilled water, three samples of ground water and four samples of surface water. We first compared the quadruplicate cellular responses and excluded the unreasonable responses. The screening criteria was that the TCRP should fall inside of a band with m ± 2δ (m and δ are the mean and standard deviation of quadruplicate TCRPs respectively). We used statistical analysis associated with the remaining responses in the subsequent analysis. Before assessing the biological activities of water sample, the positive control was used to assess test validity. Here, arsenic III were taken as positive control, whose TCRPs were shown in Fig. 3 (a). It could be seen that some of the TCRPs were close to the negative control, and some of them departed from the negative control. In order to evaluate the degree of cellular level biological activity, the proposed activity index PoE was calculated as shown in Fig. 3(b). Furthermore, the concentration response curves were fitted and the PoE50 was calculated based on the fitted parameters. The PoE50 of arsenic III was 33.11 μM and had the same order of magnitude of IC50 at 72 h (IC50 = 12.1 μM) mentioned in the literature (Rangwala et al., 2012). Similarly, the activity indexes PoE of 12 water samples were calculated as shown in Fig. 4(a) and (b). Fig. 4(a) demonstrates the PoE of each water sample at each concentration (from high to low). If the bar is high enough and around the 100% reading, it means that the water sample does not affect the cell growth over the log-phase. The corresponding color map of PoE is shown in Fig. 4(b). The brighter the color, the more biological activity the water sample has. As a result, S9 and S11 showed growth inhibitory activity in HepG2 cells, which coincided with the original TCRP shown in Appendix A. To quantitatively determine the biological activities of water sample, the concentration response curves were fitted and the PoE50 was calculated based on the fitted parameters. Two examples are shown in Fig. 4(c) and (d). The PoE50 of S4 and S11 are 7.02  109 and 0.1 respectively. Killing half of untreated cells over the log-phase, the water sample from S4 should be concentrated as 7.02  109 time, while the water sample from S11 only need 0.1 dilution. From this analysis, it is concluded that S11 exhibits high biological activity whereas S4 does not show any biological

140

T. Pan et al. / Ecotoxicology and Environmental Safety 114 (2015) 134–142

activity. The PoE50 values of 12 water samples in 4 cell lines are shown in Table 1. There are three cases in this experiment.

 Case 1: PoE50 is between 0 and 1, which reflects that the water





sample has the cellular level biological activity. The smaller the value of PoE50, the higher the biological activity of the water sample under investigation. Case 2: PoE50 is greater than 1. It means that the water sample should be concentrated to achieve the killing of half of untreated cells over the log-phase. The case reflects that the water sample has little or no biological activity. Case 3: There is no real solution of PoE50 (denoted as ‘–’ in Table 1). It means that the water sample does not affect the cells.

It can be seen that the water samples from S9 and S11 exhibit high biological activities, which is consistent with Fig. 4. In order to confirm the results and validate the proposed method, the Microtox test system was taken for comparison. As mentioned in the literature, the Microtox is a standardized toxicity test system which is rapid, sensitive, reproducible, ecologically relevant and cost effective (Lukawska-Matuszewska et al., 2009). Here, Vibrio fischeri was taken as the indicatory bacteria to conduct the ecotoxicity evaluation. The decrease in luminescence of Vibrio fischeri showed the intensity of biological effect after exposure to water samples. The exposed time was set as 30 min and the Microtox Omni software was used to analyze the response data. The toxicity index, i.e. IC50, at the endpoint was derived to demonstrate the level of toxicity. As shown in Table 1, the IC50 values of S9 and S11 showed lower values because of their higher biological activity. A comparison between both tests demonstrates the potential usefulness of the proposed approach as a screening tool. If PoE50 is less than 1, the water sample is regarded as having biological activities. 3.2. Discussion Although numerous toxicity assessment methods have been proposed for the environmental water monitoring, the design of in vitro assay, data collection and treatment remain a huge problem. The primary goal of this study is to explore a cell-based high throughput screening in which TCRPs can be used to assay the biological activity of environmental water sample that could potentially lead to toxicity effect. The success of the proposed method is based on the concept that the Percentage of Effect reflects the cumulative biological activity effect compared to the negative control over the log-phase. Compared to the synthetic index proposed by Pan et al. (2013a), the method proposed in this work does not need to predefine the evaluated threshold. Furthermore, the concept of concentration–response curve is presented in this study which has not been considered in the previous work. The concentration–response curve is inspired by the cytotoxicity test of a chemical, which not only demonstrates the level of biological activity for each concentration, but also gives the varying activity trend with the concentration variation. The power model of three parameters is also used to describe the concentration–response curve. To the best of our knowledge, the design of cell-based dynamic test experiment for water cellular level biological activity assessment has not appeared in the literature before. A challenge exists when maintaining low assay variance involving inter/intra-plate. Here, the coefficient of variation (CV%) is used to provide a quantitative indicator of the repeatability of the tests (Xing et al., 2012). Fig. 5(a) shows the negative control's CV% of six E-Plates in one experiment. The CV% in this figure are all less than 20%, which can be accepted. Furthermore, the robustness of

developed activity index PoE is also tested. As mentioned before, the HepG2 cell line has been repeated three times. Fig. 5 (b) demonstrates the PoE boxplot for S11 with six concentrations. The average of the six CV%'s is 2.04% and the CV of PoE50 is 10.16%. The result shows a more robust repeatability performance of the proposed activity index. The reason is that PoE evaluates the cumulative activity effect but not at a single point of time. Cell cultures under laboratory conditions are free of biological contaminants such as bacteria, mold, yeast, virus, protozoa, and mycoplasma. These biological contaminants can achieve high densities altering the growth and characteristics of the culture, potentially leading to inaccurate and erroneous results in the cell based assay. Therefore, it is very important to have sterile cell culture. When applying the cytotoxicity assay to environmental samples, the biggest challenge was to avoid the biological contamination from the environmental samples. Since pre-treatment steps are labor intensive with longer turn around time and higher assay cost, it will be optimal if no pre-treatment is involved. Twelve samples were analyzed by the cytotoxicity assay using the SK-N-SH cells. No bacteria or mycoplasma contaminations were observed under current assay conditions (AlbertaHealth, 2012). Using liquid handling devices, RTCA and its control software, and the proposed data processing algorithm, 12 water samples can be assessed at the same time. Through this process, one can rapidly identify active water samples and high-throughput screening can be achieved. The results of these experiments provide starting points for evaluating the impact of water contamination.

4. Conclusions In order to assess human health risk to contaminated environmental water, the use of human cell based assay is a relevant approach. The real-time cell electronic sensing system not only provides the high-throughput assay platform, but also collects multi-dimensional data that allows complex analysis, clustering and profiling. In this study, a novel method is proposed for the cellular level biological activity assessment of environmental water. Twelve samples from ground water, storm ponds and lake water are analyzed by the proposed method. Data analysis demonstrates the sensitivity of the assay and the range of biological activities that may be expected from environmental samples. It should be noted that the composition of contamination in water will change with respect to time. In this regard, only one time bio-test may not be sufficient. The logical next step is to perform a longer term assessment by evaluating the impact of water contamination over a longer sampling period. It should be pointed that existence of biological activities at the cellular level as detected by the proposed method does not necessarily mean that the water sample already has the toxicity. The proposed methods should only be treated as a potential high throughput screening tool for preliminary environmental tests. Although our results are consistent with traditional Microtox test and cellular responses, other and more complex environmental aqueous samples need to be evaluated before conclusions about making a more general applicability. The future works will focus on

 testing pure individual substances and substance mixtures to better understand the mixture effects;

 monitoring a vast variety of environmental aqueous to improve the assay criteria;

 optimizing the Standard Operating Protocol (SOP) to increase the laboratory analytical capacity;

 working with different toxicity testing systems.

T. Pan et al. / Ecotoxicology and Environmental Safety 114 (2015) 134–142

Acknowledgments

Appendix A

The work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC), Alberta Health, the National Nature Science Foundation of China [Grant number 61273142], the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD) and Foundation for Six Talents by Jiangsu Province. We would like to thank David Kinniburgh from Alberta Centre for Toxicology for scientific advice and technical support, anonymous reviewers for their helpful comments.

See Fig. A1.

Appendix B See Fig. B1.

Fig. A1. TCRPs of 12 water samples (HepG2).

141

142

T. Pan et al. / Ecotoxicology and Environmental Safety 114 (2015) 134–142

Fig. B1. Examples of area under the cellular response profile (HepG2).

References AlbertaHealth, 2012. Water Toxicity Testing. Technical Report, Edmonton, Alberta, Canada. Comerton, A., Andrews, R., Bagley, D., 2009. Practical overview of analytical methods for endocrine-disrupting compounds, pharmaceuticals and personal care products in water and wastewater. Philos. Trans. 367 (1904), 3923–3939. Davis, J., 2011. Animal Cell Culture. Wiley-BlackWell, West Sussex UK. Dragun, Z., Podrug, M., Raspor, B., 2009. Combined use of bioindicators and passive samplers for the assessment of river water contamination with metals. Arch. Environ. Contam. Toxicol. 57 (2), 211–220. Elad, T., Almog, R., Yagur-Kroll, S., Levkov, K., Melamed, S., Shacham-Diamand, Y., Belkin, S., 2011. Online monitoring of water toxicity by use of bioluminescent reporter bacterial biochips. Environ. Sci. Technol. 45 (19), 8536–8544. Faria, M., Lopes, R., Nogueira, A., Soares, A., 2007. In situ and laboratory bioassays with Chironomus riparius larvae to assess toxicity of metal contamination in rivers: the relative toxic effect of sediment versus water contamination. Environ. Toxicol. Chem. 26 (9), 1968–1977. IAEA, 2009. Analytical Methods for Environmental Water Quality. Technical Report, United Nations Environment Programme 2009. Kaza, M., Mankiewicz-Boczek, J., Izydorczyk, K., Sawicki, J., 2007. Toxicity assessment of water samples from rivers in central poland using a battery of microbiotestsc a pilot study. Pol. J. Environ. Stud. 16 (1), 81–89. Lah, B., Žinko, B., Narat, M., Marinšek-Logar, R., 2005. Monitoring of genotoxicity in drinking water using in vitro comet assay and ames test. Food Technol. Biotechnol. 43 (2), 139–146. Lukawska-Matuszewska, K., Burska, D., Niemirycz, E., 2009. Toxicity assessment by microtox (r) in sediments, pore waters and sediment saline elutriates in the gulf of gdansk (Baltic sea). Clean-Soil Air Water 37 (7), 592–598. Mankiewicz-Boczek, J., Nalecz-Jawecki, G., Drobniewska, A., Kaza, M., Sumorok, B., Izydorczyk, K., Zalewski, M., Sawickib, J., 2008. Application of a microbiotests battery for complete toxicity assessment of rivers. Ecotoxicol. Environ. Saf. 71 (3), 830–836.

Pan, T., Huang, B., Khare, S., Zhang, W., Gabos, S., Huang, D., 2013a. Development of toxicity index to evaluate the level of water contamination by using TCRCs. In: IFAC Symposium on Computer Application in Biotechnology, IFAC, December 16–18, 2013, Mumbai, India, pp. 18-23. Pan, T., Huang, B., Zhang, W., Gabos, S., Huang, D., Devendran, V., 2013b. Cytotoxicity assessment based on the AUC50 using multi-concentration time-dependent cellular response curves. Anal. Chim. Acta 764 (1), 44–52. Radić, S., Gregorović, G., Stipaničev, D., Cvjetkoa, P., Šuta, M., Vujčić, V., Oreščanin, V., Klobučar, G.I.V., 2013. Assessment of surface water in the vicinity of fertilizer factory using fish and plants. Ecotoxicol. Environ. Saf. 96 (1), 32–40. Rangwala, F., Williams, K., Smith, G., Thomas, Z., Allensworth, J., Lyerly, H., Diehl, A., Morse, M., Devi, G., 2012. Differential effects of arsenic trioxide on chemosensitization in human hepatic tumor and stellate cell lines. BMC Cancer 12 (2), 1–12. Roche, 2011. xCELLigence System Application Note No.16: RTCA HT Instrument High-Throughput GPCR Screening. Technical Report, Mannheim, Germany. Sansom, B., Vo, N.T.K., Kavanagh, R., Hanner, R., MacKinnon, M., Dixon, D.G., Lee, L.E. J., 2013. Rapid assessment of the toxicity of oil sands process-affected waters using fish cell lines. In Vitro Cell. Dev. Anim. 49 (1), 52–65. Sichani, M.F., Honarnejad, S., Heiser, L., Gray, J.W., Sorger, P. Soldán, P., Badurová, J., 2013. A method for screening for the risk of chronic effects of surface water pollution. Environ. Monit. Assess. 185 (1), 21–30. Surveillance, H., 2000. Arsenic in Groundwater from Domestic Wells in Three Areas of Northern Alberta. Technical Report. WHO, 2010. WHO Human Health Risk Assessment Toolkit: Chemical Hazards. World Health Organization. Xing, J., Gabos, S., Huang, B., Pan, T., Huang, M., Chen, J., 2012. High-throughput quantitative analysis with cell growth kinetic curves for low copy number mutant cells. Anal. Bioanal. Chem. 404 (6–7), 2033–2042.