ABSTRACT: Ketoprofen and diclofenac are non-steroidal anti-inflammatory drugs (NSAIDs) often used for similar indications, and both are frequently found in surface waters. Diclofenac affects organ histology and gene expression in fish at around 1 M-BM-5g/L. Here, we exposed rainbow trout to ketoprofen (1, 10 and 100 M-BM-5g/L) to investigate if this alternative causes less risk for pharmacological responses in fish. The bioconcentration factor from water to fish blood plasma was <0.05 (4 for diclofenac based on previous studies). Ketoprofen only reached up to 0.6M-bM-^@M-0 of the human therapeutic plasma concentration, thus the probability of target-related effects was estimated to be fairly low. Accordingly, a comprehensive analysis of hepatic gene expression revealed no consistent responses. In some contrast, trout exposed to undiluted, treated sewage effluents bioconcentrated ketoprofen and other NSAIDs much more efficiently, according to a meta-analysis of recent studies. Neither of the setups is however an ideal representation of the field situation. If a controlled exposure system with a single chemical in pure water is a reasonable representation of the environment, then the use of ketoprofen is likely to pose a lower risk for wild fish than diclofenac, but if bioconcentration factors from effluent-exposed fish are applied, the risks may be more similar. Ketoprofen (purity M-bM-^IM-%98%) (Sigma-Aldrich, Steinheim, Germany) was dissolved in water (500 mg/L concentrated solution), stirred vigorously and diluted stepwise to obtain stock concentrations (0.5 mg/L, 5 mg/L and 50 mg/L). Juvenile rainbow trout (Oncorhynchus mykiss) of both sexes (age: approximately six months, weight: 39.4M-BM-16.6g) were obtained from VM-CM-$nneM-CM-%ns fiskodling AB, Sweden, and kept in 500 L holding tanks with sand-filtered, aerated fresh water in a flow-through system for one week prior to the experiment. Eight experimental 48 L glass aquaria, semi-covered in black plastic bags to reduce visual stress from the outside (12:12 h light:dark photocycle), were supplied with filtered, aerated freshwater at a flow rate of 0.25 L/min. At the onset of the experiment 80 trout were randomly divided among the aquaria (ten fish in each). Peristaltic pumps (500 M-BM-5l/min) from each stock solution were used to reach target ketoprofen concentrations of 0 (control), 1 (low), 10 (intermediate) and 100 (high) M-BM-5g/L with two aquaria for each concentration. Water samples were taken from each aquarium at day 1, 7 and 14 and stored at -20M-BM-0C until analysis. To minimize biological variation due to variable food intake, the fish were not fed during the exposure period. It should be stressed that trout readily cope with food deprivation for much longer periods at these temperatures [Kullgren et al., 2010]. To ensure a stable water quality, temperature (12.9M-BM-10.1M-BM-0C), pH (7.4M-BM-10.2) and oxygen saturation (88.3%M-BM-11.3) were monitored every third day. The total organic carbon content of the supply water was 2.5 mg/L. At the experiment termination on day 14, the fish were killed by a rapid blow to the head and blood was collected from the caudal vessels by the use of heparinized syringes. The plasma was instantly separated by centrifugation at 10,000 rpm for 2 min and snap frozen in liquid nitrogen. The length and weight of the fish were measured and their sex determined by macroscopical observation of their gonads. Liver samples for gene expression analyses were collected and snap frozen in liquid nitrogen. Both liver and plasma samples were stored at M-bM-^HM-^R70M-BM-0C until analysis. All animal experiments were approved by the local animal committee in Gothenburg (permission number 216-2010). A 15k rainbow trout gene expression microarray was designed for the RT analyzer platform (febit, Heidelberg, Germany) using The Institute for Genomic Research (TIGR) Rainbow Trout Gene Index (RTGI) database version 7.0 (http://compbio.dfci.harvard.edu/tgi/). Details on the probe design strategy, but for eelpout (Zoarces viviparus), and transcript selection strategy are described elsewhere [Cuklev et al., 2011, Kristiansson et al., 2009, Cuklev et al., 2012]. When available, transcripts at GenBank (http://www.ncbi.nlm.nih.gov/nucleotide) were used. Results from similar microarrays, performed by our research group using the same platform, have shown good correlation with qPCR data [Cuklev et al., Kristiansson et al., 2009, Cuklev et al., 2012, Gunnarsson et al., 2009, Lennquist et al., 2011]. The frozen liver tissue was homogenized using Tissuelyser (Qiagen) and total hepatic RNA was isolated using QIAcube and RNeasyM-BM-. Plus Mini Kit (Qiagen). The RNA quantity was determined with spectophotometric measurements (Nanodrop 1000, NanoDrop Technologies) and the quality was assessed by Experion automated electrophoresis using RNA StdSense chip (Bio-Rad, Sundbyberg, Sweden). Biotinylated aRNA was synthesized using MessageAmpM-bM-^DM-" II-Biotin Enhanced Single Round aRNA Amplification Kit (AmbionM-BM-.). The aRNA samples (20 M-BM-5g) were vacuum dried in a vacuum centrifuge, dissolved in 10 M-BM-5l water and fragmented according to the manufacturerM-bM-^@M-^Ys protocol. The following steps described in this subchapter were all performed by febit. Oligonucleotide arrays were synthesized by photo-controlled in situ synthesis using the Geniom One system (febit). Each biochip consists of eight individually accessible micro-channels, each of which is referred to as a microarray in this manuscript. Four individuals (of both sexes) from each control aquarium and each high exposure concentration aquarium were included in the analysis, i.e. eight fish per exposure concentration. In total, two biochips were analyzed corresponding to 16 microarrays. Pre-hybridization and hybridization were performed based on a customized protocol, described in detail by Cuklev et al. [2011]. The samples were randomly allocated on the biochips. Signals were detected using the internal CCD-camera system of the RT analyzer instrument (febit) and quantified using the Geniom Wizard software. Integration times were 266 and 273 ms, determined automatically by the instrument software. All statistical calculations were done in R-2.12.2 (www.r-project.org) [R Development Core Team, 2010]. The quality of pre- and post-normalized arrays was verified with Box and MA plots. The data analysis was performed in the R-package LIMMA [Smyth, 2004]. Data were normalized using the M-bM-^@M-^XquantileM-bM-^@M-^Y method. Moderated t-statistics and adjusted p-values of differential expression were calculated using the empirical Bayes model. As we found no clear differences between the two control aquaria or between the two aquaria with the highest concentration of ketoprofen (100 M-BM-5g/L), all fish from the same treatment were treated as a single group, i.e. we compared one control group to one group of fish exposed to 100 M-BM-5g/L of ketoprofen.
