Introduction

Every year thousands of tons of drugs, after human and veterinary medication, are excreted unmetabolized or as active metabolites (1). Most wastewater biological treatments are unable to effectively remove pharmacologically active ingredients (2, 3). This, along with improper disposal of expired medications, contributes to the pharmaceutical contamination of freshwater aquatic habitats (4), which has been recently confirmed by survey campaigns worldwide. More than 80 pharmaceutically active compounds from various therapeutic classes have been detected up to g/L levels in sewage, surface, and ground waters (1). Such documented evidence raised a novel and intricate toxicological and environmental issue.

Drugs are found in the environment at very low concentrations, far below the therapeutic doses employed for medical practice. Exposure, however, may chronically affect a variety of organisms in different stages of development, and for which no pharmacological action is known. Ad ditionally, pharmaceuticals are ordinarily found in complex mixtures of active ingredients with unrelated biological activities. Toxicological studies in this field, therefore, have to face a number of difficulties, and the potential risk associated with the presence of combinations of pharmaceutical contaminants in surface waters is mostly unknown.

The majority of toxicological studies targeting aquatic organisms have dealt with estrogens and antibiotic type compounds, with sporadic exceptions (5-7). Acute toxicity tests have failed to detect the subtle action elicited by pharmaceuticals at environmental concentrations, and showed that toxicity can be influenced by additive and synergistic effects (5, 7), raising issues concerning the capability of mixtures of inducing genetic alterations/mutations (9).

Here we aimed to investigate the toxicity of 13 therapeutic ingredients, assembled in a mixture that mimics the concentration and combination of drugs detected in northern Italian rivers (3, 10, 11). We approached this issue by in vitro toxicological tests using cell clones to minimize the variation in biological response due to different individuals (12). Water-borne drugs may affect the physiology of aquatic life as well as humans, which impinge upon water resources for supply. We chose the human embryonic kidney cell line HEK293 as the cellular model (13-15), considering embryonic cells as a representative for the most sensitive human archetype. The specific objectives were to evaluate the combined effects of drugs on cells growth and to acquire valuable insights into the mechanisms of toxicity at the molecular level, by studying the expression of genes and proteins.

Experimental Procedures

Active Pharmaceutical Ingredients (APIs). Pharmaceuticals were selected based on occurrence in the environment as well as on long-term use forecast, poor biodegradation, poor metabolization, and high sale levels (3, 4, 10, 11). The consequent list of top pharmaceuticals causing potential environmental risk also comprised the alkylating agent cyclophosphamide (Table 1), a molecule with high activity and potential toxicity. The reference standards of atenolol, bezafibrate, carbamazepine, cyclophosphamide, furosemide, hydrochlorothiazide, ibuprofen, lincomycin, ofloxacin, salbutamol, and sulfamethoxazole were purchased from Sigma-Aldrich (Sigma-Aldrich Co., Dorset, UK); ciprofloxacin was purchased from ICN Biochemicals (Meckenheim, Germany), and ranitidine obtained from GlaxoSmithKline (Philadelphia, PA). Experimental concentrations of APIs were selected based on the concentrations previously measured in the environ ment (3, 4, 10, 11), rounding each environmental value up to the nearest order of magnitude (Table 1). The stock solution of APIs was prepared combining single drugs in methanol at the concentrations reported in Table 1, and stored at -20 C. Aliquots of the stock solution were evaporated to dryness and resuspended in culture media prior to testing at the optimal exposure level. From here on we name Level 1 the mixture of pharmaceuticals at experimental concentrations as reported in Table 1 (this combination of drugs is also referred in the text as "environmental levels"); Level 0.1, 10, and 100 corresponded to one tenth, 10 times, and one hundred times, respectively. Cisplatin stock solution (100 mM) was prepared in DMSO, stored at 4 C protected from light, and diluted into the culture media to obtain the desired concentrations.

Cell Cultures. The immortalized cell line HEK293 was maintained at 37 C with 5% CO₂ in Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 mM glutamine (Euroclone, West Yorks, UK) containing 10% fetal bovine serum (FBS) (Euroclone). Sub-confluent HEK293 cultures were washed with phosphate saline buffer (PBS) (Euroclone) in T-75 flasks, and cells harvested by trypsin (Euroclone) for sub-culturing or the following assays.

Cell Titer Proliferation Assay. For cytotoxicity tests, HEK293 cells were resuspended in DMEM 0.5% FBS, with 5 × 10³ cells per well seeded in 96-well flat-bottom microtiter plates and incubated for 4 h at 37 C. Culture media were then changed with 100 L of DMEM 10% FBS added with increasing levels of the drug mixture. Experimental controls consisted of untreated cells (negative), and cells exposed to cisplatin as positive reference for proliferation assays (with DMSO as its own negative control). Microtiter plate tests were performed using the CellTiter 96 non-radioactive cell proliferation assay kit (Promega Corporation, Madison, WI). Assays were performed as suggested in the standard protocol provided with the kit. The dye solution of tetrazolium salt (MTT) was added at different times, and the stop solution was added 4 h after incubation at 37 C with MTT. Plates were incubated for 2 h at room temperature and absorbance recorded at 600 nm with a microplate reader Metertech 960 (Metertech Inc., Taipei, Taiwan).

Antibodies and Molecular Probes. Commercial antibod ies were obtained from Cell Signaling Technology, Inc. (Beverly, MA): Pathscan Multiplex Western Cocktail II, Apoptosis Sampler, and Phospho-MAPK Family Antibody Sampler. For immunoblotting, concentrations of primary and secondary antibodies were set as suggested by the manufacturer. TaqMan probes (FAM-dye) for real-time PCR were purchased from Applied Biosystems (Foster City, CA): -actin (Hs99999905_m1) (internal reference gene), glutathione-S-transferase P1 (GSTP1) (Hs00168310_m1), p53 (Hs00153349_m1), p16 (CDKN2A; Hs00233365_m1), p21^Cip1 (Hs00355782_m1), and cyclin-dependent kinase 2 (CDK2) (Hs00608082_m1).

Protein Activation Analysis. To study protein kinases phosphorylation and apoptosis, 125 × 10³ HEK293 cells were seeded in 6-well plates and prepared by overnight incubation at 37 C in DMEM 0.5% FBS. Cells were assayed for 2h in 100 L of fresh DMEM 0.5% FBS containing pharmaceuticals at the desired concentrations, washed on ice with 4 C PBS, and collected from 6-well plates with a sterile scraper in 40 L of extraction buffer. To investigate the activation of caspase-3, caspase-9, and poly(ADP-ribose) polymerase (PARP), total proteins were extracted in Chaps Cell Extract Buffer (Cell Signaling Technology) from cells treated for 48 h with the tested compounds. For protein kinases analysis, cells were extracted in Immunoassay Cell Extraction Buffer (BioSource International, Inc., Camarillo, CA) after 2 h of exposure. Extraction buffers were added with 5L/mL of both Protease Inhibitors Cocktail (Sigma) and phenyl-methyl-sulfonyl-fluoride (200 mM in ethanol, Sigma). Total protein content was measured by Bradford method (16) and 60 g of protein extract was loaded for each sample on a 12% acrylamide gel. Proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a Hybond nitrocellulose membrane (Amersham, Little Chalfont, UK). Membranes were then washed, blocked, and incubated with primary and secondary antibodies according to the standard protocols suggested by the antibody manufacturer (Cell Signaling Technology). Membranes were developed using Lumiglo (Cell Signaling Technology), and chemiluminescence detected by exposure to CL-XPosure films (Pierce Biotechnology, Rockford, IL).

Gene Expression Studies. To investigate gene expression, 125 × 10³ HEK293 cells were plated in 6-well plates, treated for different periods of time (0, 1, 4, and 24 h) with therapeutic drugs, washed with 4 C PBS on ice, and collected from 6-well plates with a sterile scraper in 500 L of TRI Reagent LS (Sigma). Total RNA was extracted with chloroform and precipitated with 2-propanol as suggested by TRI Reagent user's protocol (Sigma). Pellets were then resuspended in 30 L of RNase-free water pooling experimental duplicates. For each sample, approximately five micrograms of total RNA were retro-transcribed using SuperScript II RNase H^- Reverse Transcriptase (Invitrogen, Carlsbad, CA) primed by random nonamers (Sigma). First strand cDNA synthesis was achieved following the protocol suggested by Invitrogen. Real-time PCR reactions (25 L volume) were performed using 100-times diluted cDNA samples, TaqMan Universal PCR Master Mix (Applied Biosystems), and TaqMan probes (Applied Biosystems) as suggested by the provider. Reactions were cycled using the standard two-step real-time PCR protocol in an Applied Biosystems ABI PRISM 7000 Sequence Detec tion Systems, and analyzed by ABI PRISM 7000 SDS software (version 1.1).

Flow-Cytometry. Analysis of cell-cycle phase populations was performed using either 4 × 10⁵, 5 × 10⁵, or 8 × 10⁵ HEK293 cells, harvested and plated in T-25 flasks as described above. Cells were collected 48 h after exposure without PBS washing, fixed in 80% ethanol 20% PBS, and stored at -20 C. Prior to reading on a BD-FACSCalibur system (Becton-Dickinson Biosciences, San Jose, CA), cells were treated with RNaseA and stained with propidium iodide (PI).

Results

Effects on Embryonic Cells Growth. Cytotoxicity tests revealed that the proliferation of HEK293 cells was inhibited by the pharmaceutical mixture at environmental levels or higher, with a maximum result of ca. 30-40% growth reduction compared to negative control values after 48 h (Figure 1A,D). Effects of the cytotoxic chemical cisplatin (positive control) on HEK293 are shown for comparison (Figure 1A). Cells were also treated for 72 h time. When culture media added with pharmaceuticals were maintained unchanged over the experimental time period, cells reached maximum growth inhibition after 48 h of exposure (Figure 1B). On the contrary, changing culture media supplemented with fresh drug mixture every 24 h, resulted in a continuous decrease in cells proliferation over time for all treatments except the lowest dose (Figure 1C). This suggested that therapeutics could be subjected to physical or biological degradation in our experimental conditions, or otherwise that drugs may accumulate in cells. To rule out the possibility that cyclophosphamide, the strongest cytotoxic agent included in the drug mixture, was entirely responsible for the observed inhibition of cell proliferation, we compared the effects of cyclophosphamide alone to the complete pharmaceutical mixture (including cyclophosphamide itself). Instead of the growth inhibition shown by the entire mixture, cyclophosphamide appeared to stimulate cell proliferation at all concentrations tested and only inhibited growth at the highest concentration (Figure 1D).

Figure 1 Effects of pharmaceuticals on HEK293 cells proliferation. Values are reported as % proliferation compared to untreated controls (100%) and expressed as average ± SE (N = 3). Level 1 corresponds to experimental concentrations as reported in Table 1. (A) 48 h dose-response plot of serial dilutions of the therapeutic drug mixture () compared to different concentrations of cisplatin as reference for cytotoxicity (M, ); DMSO control for cisplatin was not shown. (B) Cytotoxicity of pharmaceutical dilutions over time. (C) Cytotoxicity of pharmaceuticals over time, changing culture media added with fresh drug mixture every 24 h. Exposure: is Level 0.1; is Level 1; is Level 10; is Level 100. (D) Comparison between the complete drug mixture () and corresponding concentrations of cyclophosphamide () after 48 h exposure without replacing the drug mixture.

Induction of Cellular Stress Responses. We hypothesized that growth reduction by APIs in HEK293 cells could have been induced via the phosphorylation of stress/mitogen-activated proteins (SAP/MAP) kinase. Exposure to rising levels of the drug mixture had no effect on p38 kinase (Phospho-p38, Figure 2A) and cJun N-terminal kinase (JNK) (data not shown). Conversely, phosphorylation of the extracellular signal regulated kinases 1/2 (Phospho-ERK1/2) increased in a dose-dependent manner as a result of 2 h treatment with pharmaceuticals (Figure 2A). The 90 kDa ribosomal S6 kinases 1-3 (Phospho-p90RSK), showed stronger phosphorylation after exposure to environmental levels of the drug mixture (Figure 2A). We also studied the influence of the SAPK/MAP kinase inhibitor SB203580 on cells growth and SAP/MAP activation (see Supporting Information), and verified that ERK1/2 signaling pathway was crucial in determining the observed reduction in cell proliferation (Figure S2). In relation to the activation of cellular stress responses, we investigated variations in GSTP1 gene expression after 24 h of exposure. Treatment of HEK293 cells with environmental levels of the pharmaceutical mixture resulted in a 2-fold increase in cellular accumulation of GSTP1 mRNA, with a 3-fold rise in gene expression for the highest dose tested, compared to the controls and relative to internal reference levels of -actin (Figure 2B). The other drug levels evaluated here gave no statistically significant results (Figure 2B).

Figure 2 (A) Effects of pharmaceuticals on stress kinases phosphorylation; elongation factor 4E (elF4E) was used as internal control. (B) Expression levels of GSTP1 in consequence to drug stress. Values are expressed as mean ± SE (N = 3). * P < 0.05 (Student t test). Level 1 corresponds to experimental concentrations as reported in Table 1, 0 = untreated controls.

Onset of Programmed Cells Death. No evidence of apoptosis was found in HEK293 cells exposed to rising levels of pharmaceuticals, either by DNA fragmentation analysis (see Supporting Information, Figure S1A) or by activation of caspase-3 (Figure S1B), caspase-9, and PARP (data not shown).

Induced Morphological Changes in Embryonic Cells. By microscopic examination, HEK293 cultures treated with drugs were characterized by smaller proliferative colonies compared to control samples (Figure 3 A,B). No evident difference between controls and treatments was found in the levels of cell lysis or in the total number of dead cells, as revealed by trypan blue staining (Figure 3). We observed that, morphologically, HEK293 cells exposed to pharmaceuticals displayed preferentially an enlarged round shape, losing cell-cell contacts, in contrast with control cultures dominated by elongated star-shaped cell morphology, as depicted in Figure 3 (highlighted areas). No histochemical evidence was found of pH 6.0-dependent -Gal activity as an indication of cellular senescence induction, either in control HEK293 cultures or serial treatments (see Supporting Information).

Figure 3 Light microscopy (20X magnification) of HEK293 cells untreated (A) and exposed to Level 100 of the pharmaceutical mixture (B). Pictures were taken in 96-well plates during standard 48 h cytotoxicity tests. Note the differences in colony dimensions between untreated (A) and treated samples (B). Highlighted areas (circles) in (B) show examples of cells morphology as modified by the tested drug mixture.

Effects on Cell-Cycle Progression. We speculated that growth inhibition associated with an enlargement in cell morphology could have been linked to an arrest in the eukaryotic cell-cycle, as induced by stress response activation. Rising exposure levels of the drug mixture showed no evident effect on both phosphorylation and expression of the DNA damage responsive protein p53 (Figure 4A,B). On the other hand, pharmaceuticals induced an increased expression of genes encoding for the CDK inhibitor proteins p16 and p21, and of CDK2 (Figure 4B), compared to control levels and relative to the internal reference gene (-actin). These observed effects for p21 and CDK2 genes, a 2-fold and 3-fold increase in expression, respectively, were significant only for the highest level tested. On the contrary, the expression of p16 increased in a dose-dependent manner (Figure 4B). Expression patterns of the same genes were analyzed over time under environmental levels of the drug mixture (Figure 5), with accumulation of p53 mRNA evident 24 h after exposure. Expression of CDK inhibitors, instead, raised constantly and significantly over time, and reached a maximum increase of 2-fold after 4 h and 3-fold after 24 h for p16 and p21, respectively (Figure 5). No significant change was observed for CDK2 mRNA levels in such conditions. Given these changes in expression of cell-cycle related genes, flow-cytometry analysis revealed subtle differences between treatments and controls, with respect to cell populations in the different phases of the cell-cycle (Supporting Information, Figure S3). Increased percentage of cells in the G2/M phase was, however, constant for all cultures treated 48 h with the pharmaceutical mix, regardless of the amount of cells (4, 5, or 8 × 10⁵) being seeded for the test (Figure 6).

	Figure 4 (A) Effects of pharmaceuticals on p53 phosphorylation (2 h exposure); elF4E was used as internal control. (B) Expression levels of genes involved in cell cycle progression under drug treatments (24 h exposure). Values are expressed as mean ± SE (N = 3). * P < 0.05 (Student t test). Level 1 corresponds to experimental concentrations as reported in Table 1, 0 = untreated controls.
	Figure 5 Expression of p53, p16, p21, and CDK2 genes over a 24 h time exposure to the drug mixture at Level 1 (experimental concentrations, Table 1). Values are expressed as mean ± SE (N = 6). * P < 0.05 (Student t test).
	Figure 6 Percentage of HEK293 cells in each phase of the cell cycle. Quantitation refers to the observed differences (fluorescence automated analysis of 103 cells) between cultures exposed to Level 100 of the pharmaceutical mixture for 48h and controls. Values are expressed as mean ± SE (N = 5). *** P = 0.00156 (Student t test).

Discussion

We found that environmental levels of pharmaceuticals inhibited human embryonic cells growth by 10 to 30% compared to controls, depending on whether cultures were exposed daily to fresh doses of pharmaceuticals or not (Figure 1B,C). Similar values of growth reduction were confirmed on different cell lines (mouse fibroblasts, human tumor cells, zebrafish liver cells) (F. Pomati, unpublished). No clear dose-response model fitted with our cytotoxicity data (Figure 1A). Compared to the effect of cisplatin, the drug mixture appeared to inhibit growth to a similar extent regardless of the increasing levels of exposure. Additionally, our results promote the possibility that the experimental effect was due to the association of drugs rather than to a single toxic compound such as cyclophosphamide (Figure 1D). Both the indicia observed above have been documented in the literature (7, 8, 17), and point to the hypothesis that interactions may take place among different APIs. The nature of such interactions may change with varying levels of exposure, with certain chemicals being dominant at low doses, others at high doses. This issue represents an additional serious toxicological concern with regards to pharmaceutical mixtures, and it is also largely unresolved in medicine (18).

Here, we aimed to investigate the mechanisms by which our mixture of therapeutic drugs was affecting cell growth. The possibilities of apoptosis and senescence have been discounted in the present study, and the end product of pharmaceutical treatments suggested the onset of quiescence (19-22), both morphologically (Figure 3) and with regards to the expression genes such as p16 and p21 (Figures 4 and 5). Taken together, our data support the observations made by Tang and co-workers (23) on several human cell lines, in which ERK activation has been found to induce p21 expres sion independently of p53 stabilization, resulting in G2/M arrest (23). Results obtained in the present study (Figures 2A, 4, 5, and 6) indicate that treatment with the therapeutic drug mixture could have triggered the same model of cellular response in HEK293 cells, which are deficient in active p53 (24).

Tang et al. (23) also demonstrated that ERK may function in the response to DNA damage, as induced in their work by etoposide treatment, which interferes with the function of DNA topoisomerase II. In the mixture of drugs selected for this study, two quinolone antibiotics (ciprofloxacin and ofloxacin) are present and known to exert the same cellular effects of etoposide, although with less potency (25). We speculate that ciprofloxacin and ofloxacin may have interfered with DNA synthesis in HEK293 cells prompting the activation of the ERK signaling pathway and retardation of the cell-cycle. Our results on HEK293 are also consistent with previous data observed for kidney cells, which survive moderate oxidative stress by concomitant activation of ERK1/2 and G2/M arrest (15). Activation of p90RSK (Figure 2A) and a general increase in GSTP1 expression (Figure 5B) suggested here that HEK293 cultures exposed to the drug mixture were facing a moderate oxidative stress (26, 27).

Compared to real-world scenarios, the exposure assays performed in vitro in the present study can be considered artificial, particularly regarding two different aspects. In primis, environmental drug levels tested in this study were slightly higher that those detected in most aquatic habitats (Table 1). Our cytotoxicity data, however, may still underestimate the real effect of environmental mixtures. Organisms living in contaminated surface waters are, in fact, chronically exposed to combinations of water-borne drugs together with other chemical pollutants, such as heavy metals and endocrine disruptors (9). Second, the use of laboratory cell lines may overestimate the effects of drugs at the level of organisms, since long-term consequences on adult animal survival can be balanced by homeostasis processes. Nevertheless, the observed results on cellular physiology suggest that the investigated mixture of drugs could interfere with mitogenic and antimitogenic signals, such as those involved in dif ferentiation and development. This indicates that organisms in larval and early stages of growth may be sensitive to pharmaceutical contamination in aquatic environments. A reduction in reproductive functions also cannot be excluded for adult aquatic animals subjected to drug exposure (9, 18).

The major finding of this study is that a combination of diverse therapeutic drugs at ng/L levels (Table 1), as found in some typologies of surface waters, can significantly inhibit embryonic cells growth in vitro. We observed that reduced cell proliferation was associated with activation of functional pathways linked to cellular stress responses and with control of the cell-cycle progression. Due to the general lack of information regarding this relatively recent subject of concern, results presented here disclose a new topic for future discussion and pose a challenge for toxicologists to explore the effects of low doses of mixed therapeutic ingredients. Additionally, our data indicate that there is a feasible health issue associated with the presence of pharmaceuticals in the environment. Further investigations are required to specif ically identify long-term exposure risks and evaluate the nature of possible interactions taking place among different active ingredients, such as those contaminating aquatic environments. Results will be useful in the future to draw a potential activity profile of water-borne drug mixtures at the level of organisms in vivo.

Acknowledgment

F.P. is grateful to M. Molteni and to M. Gariboldi for assistance and advice in cells culturing and flow cytometry, respectively. The authors acknowledge the "Centro Grandi Attrezzature per la Ricerca Biomedica" Università degli Studi dell'Insubria, for instruments availability, and thank B. Burns for reviewing the manuscript. This research was supported in part by the Italian Ministry of Scientific Research (COFIN 2004034992).