Transcriptomic signatures in Chlamydomonas reinhardtii as Cd biomarkers in metal mixtures

Aquatic Toxicology 100 (2010) 120–127

Contents lists available at ScienceDirect

Aquatic Toxicology journal homepage: www.elsevier.com/locate/aquatox

Transcriptomic signatures in Chlamydomonas reinhardtii as Cd biomarkers in metal mixtures C.M. Hutchins a,∗ , D.F. Simon a , W. Zerges b , K.J. Wilkinson a a b

Department of Chemistry, University of Montreal, P.O. Box 6128, Succ. Centre-Ville, H3C 3J7 Montreal, Canada Biology Department, Concordia University, 7141 Sherbrooke W., H4B 1R6 Montreal, Canada

a r t i c l e

i n f o

Article history: Received 7 April 2010 Received in revised form 12 July 2010 Accepted 15 July 2010 Keywords: Chlamydomonas reinhardtii Gene transcription mRNA expression Cadmium Copper Lead Binary metal mixtures

a b s t r a c t In the natural environment, toxicant effects can be monitored by the signature mRNA expression patterns of genes that they generate in test organisms. The specificity and sensitivity of these transcriptome-based bioassays to a given toxicant can be confounded by temporal changes in biomarker mRNA expression, effects of other toxicants and hardness ions, and non-linear mRNA expression responses of genes. This study provides the foundation for the development of a transcriptomic-based bioassay for bioavailable Cd in the freshwater alga, Chlamydomonas reinhardtii. It characterizes: (1) the Cd regulation of nine genes with respect to their mRNA induction kinetics; (2) the effects of two additional metals common to freshwaters, Cu2+ and Pb2+ , and (3) the relationships between metal bioaccumulation and the transcriptomic responses. Quantitative real time PCR was used to monitor mRNA levels of nine Cd-induced genes following an exposure to 0.01, 0.11 and 1.16 ␮M Cd2+ . Several distinct mRNA expression patterns were observed with time. While the presence of Cu2+ and Pb2+ decreased Cd biouptake, mRNA levels increased for six genes, showing lack of Cd2+ specificity. Nonetheless, the transcriptomic effects of binary metal exposures rarely adhered to a simple additive model based on single metal exposures; rather most exhibited synergistic or antagonistic interactions. While none of these genes could be used as a specific Cd biomarker, the signature mRNA expression profile obtained from a select subset of Cd sensitive genes was a useful biomarker of sublethal effects. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Ecotoxicogenomics uses variations in the transcriptome of a suitable test organism to monitor sublethal responses to a given toxicant. A transcriptomic profile is a snapshot of the relative mRNA expression levels of a gene set, which can range from a few relevant genes to the entire genome (Neumann and Galvez, 2002). In a transcriptomic bioassay, a test organism is exposed to a contaminant solution and a select set of biomarker genes are monitored in order to quantify the effects of a toxicant of interest. The principal advantage of transciptomic bioassays is that they can provide some insight into the test organism’s physiological response(s) to stress. While gene expression depends upon both transcription and translation, mRNA levels of biomarker genes of known function are commonly used to reveal defensive responses or acclimatization to a given toxicant. In any case, measurements of transcription in relationship to variable environmental conditions can serve as a useful biomarker of environmental health. While global profiling of an organism’s entire transcriptome provides the most

∗ Corresponding author. Tel.: +61 754 466 610. E-mail address: [email protected] (C.M. Hutchins). 0166-445X/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.aquatox.2010.07.017

comprehensive information, this requires expensive and laborious microarray analyses or massive-scale “deep” mRNA sequencing. In contrast, the development of a transcriptome-based bioassay for high-throughput analyses in research and water quality testing requires the identification of a minimum number of genes whose relative mRNA levels can provide both sensitive and specific detection of the relevant toxicant(s). It is also essential to characterize the regulation of the biomarker genes by toxicant exposure in order to mitigate the following three problems inherent to transciptomic bioassays. First, toxicant exposure can induce complex temporal patterns of biomarker expression, meaning that very different transcriptomic profiles can be observed throughout a transcriptomic response. For example, phosphate deprivation induces temporal changes in mRNA levels of distinct sets of early and late genes (Moseley et al., 2006). The few studies that have assessed the kinetics of gene regulation by metal exposure suggest a similar time-dependence (e.g. Lemaire et al., 1999). Therefore, it is essential to know the kinetics and temporal complexity of a transcriptomic response on which a bioassay is based. Second, the accuracy of transciptomic bioassays can be adversely affected by non-linearity of biomarker expression as a function of toxicant concentration. To address this issue, rather

C.M. Hutchins et al. / Aquatic Toxicology 100 (2010) 120–127

than using the absolute mRNA levels of biomarker genes in a transcriptomic profile, the bioassay can also simply consider if the biomarker gene is induced and, thereby, determine whether toxicant concentration is above or below its “no transcriptional effect level” (NOTEL). The NOTEL is the maximal toxicant concentration that has no detectable effect on the mRNA expression of a biomarker gene (Ankley et al., 2006; Lobenhofer et al., 2004). To date, NOTELs have not been determined for metals and many other toxicants. Third, external factors in the environment can alter a transcriptomic response and, thereby, decrease the accuracy of a bioassay based upon it. For example, natural waters contain multiple metals including hardness ions (Ca2+ , Mg2+ ) in addition to numerous ligands. These can alter the biouptake of the toxicant and, hence, the transcriptomic response of the organism. Therefore, it is necessary to characterize the effects of extraneous physicochemical factors on the transcriptomic response. Unfortunately, most toxicogenomic studies on metals have measured single metal responses without addressing the role of multiple metal contaminants (Dardenne et al., 2008). Compounding these issues is the rarity of biomarker genes that are specifically induced by a given toxicant. For this reason, the field is starting to turn to toxicant-specific signature patterns of differential mRNA expression of genes within a transciptomic profile. The present study contributes to the development of a transcriptomic bioassay for bioavailable Cd by characterizing the transcriptional regulation of nine genes that were previously identified as potential Cd biomarkers using global gene expression profiling in the freshwater eukaryotic green alga, Chlamydomonas reinhardtii (Simon et al., 2008). Transcriptomic response to Cd exposure was assessed for kinetic variability and in the presence of additional metals Cu and Pb commonly found in situ in contaminated waters. C. reinhardtii was selected because it is endogenous to both soils and freshwaters and, therefore, it is expected to be a robust test organism for in situ measurements. Furthermore, it is also an established model organism for research into metalinduced stress and metal homeostasis. Researchers benefit from a battery of experimental tools, extensive literature on its physiology, cell biology and genetics, and the sequenced and annotated genome. Our results reveal that the responses of the Cd biomarkers were not well predicted by either free Cd in solution or bioaccumulated Cd. Rather, the use of a small subset of genes may be more useful for a specific and sensitive bioassay of bioavailable Cd. 2. Theory Mixed toxicant models can provide an initial understanding of how toxicants interact (additively, synergistically or antagonistically) relative to a single compound. Two models are predominantly used to predict mixture toxicities: concentration addition (CA) and independent action (IA). The CA model sums the concentrations of the components after adjusting for the differences in potencies and assumes the same toxic mode of action:



ci

(1)

ECxi

where ci represents the concentration of the compound i and ECxi represents the x% effect concentration for compound i. Unfortunately, when employing mRNA expression as an end point, the determination of comparable ECx levels can prove problematic since the maximum response level is often unknown, making a classic ECx approach impossible (Dardenne et al., 2008). Alternatively, the IA model assumes differing modes of action: E(cmix ) = 1 −



sa

(2)

121

where E(cmix ) refers to the total effect of the mixture and E(ci ) is the effect of compound i. The IA model appears better suited to mRNA expression (transcriptional) analysis. However, both models are necessarily oversimplifications of the complex interactions in metal mixtures and thus results may not fit either model (Chu and Chow, 2002; Preston et al., 2000).

3. Methods 3.1. Culture and exposure media C. reinhardtii (C137) was cultured to mid-log phase in fourfold diluted Tris-acetate-phosphate (dTAP) medium (Gorman and Levine, 1965), under a 12:12 h light:dark cycle of fluorescent white light and orbital shaking. Cells were pelleted by centrifugation (3300 × g for 7 min), rinsed by resuspension in 200 mL of dTAP or HEPES (4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid, 0.01 M, pH 7.0) for 2 min and pelleted again. The cell pellet was resuspended in 50 mL (to ca. 6.4 × 105 cells mL−1 ) and aliquots were diluted to a cell density of ca. 3–4 × 105 cells mL−1 , corresponding to 1 cm2 mL−1 in the exposure solutions (450 mL). As indicated below, the dTAP medium was employed to investigate the kinetics of the transcriptomic response (1–8 h), while the metal mixture experiments (2 h) were performed in 0.01 M HEPES in order to minimize complexation of the metals by the buffer components.

3.2. Time series exposures Total Cd concentrations of 0.089, 0.90 and 8.8 ␮M were buffered with 500 ␮M citrate in dTAP medium in order to give 0.01, 0.11 and 1.16 ␮M of Cd2+ , as determined from thermodynamic calculations (MINTEQA2, version 1.50). Experiments were performed in triplicate on three different days with independent cultures. Cells were sampled at 1, 2, 4, 6 and 8 h for the determination of internalised metal and the analysis of mRNA expression. Replicate 25 mL samples were first centrifuged at 3300 × g for 10 min. The pellet was subsequently washed in dTAP medium containing 10−3 M EDTA for 1 min in order to remove metal bound to the algal cell wall (Hassler et al., 2004). For the genomic analysis, the cell suspension was then centrifuged at 12,280 × g for 5 min. Cell pellets were frozen on dry ice and then stored at −80 ◦ C until RNA extraction. For the analysis of internalised metal, cells were filtered on a 5 ␮m nitrocellulose membrane (Durapore, Millipore) then washed twice with dTAP.

3.3. Mixed metal exposures Based upon the results of the time series experiments, all further metal exposures occurred over 2 h. Mixed metal exposure experiments were based upon a design described by Stratton (1988). In order to assess expression levels for the single metals, cells were first exposed to 0.05, 0.1, 0.5, 1.0 and 5.0 ␮M of Pb or Cu. In a second set of experiments designed to evaluate expression following exposure to two metals, a fixed concentration of Cd2+ (0.5 ␮M) was added to all conditions. Finally, cells were exposed to 0.5 ␮M Cd2+ alone. All experiments were performed in 0.01 M HEPES at pH 7 using three independent biological replicates. Free metal concentrations corresponded to 99.9%, 83% and 78–80% of the total metal concentrations for Cd, Pb and Cu, respectively (Table S1). Following metal exposure, 5 mL of 1 mM EDTA was added to 45 mL of the exposure solution (corresponding to ca. 1.5–1.6 × 107 cells). Cells were prepared for bioaccumulation and genomic analysis as described above.

122

C.M. Hutchins et al. / Aquatic Toxicology 100 (2010) 120–127

3.4. Bioaccumulation Following their exposure to the metal(s), cells were filtered as above (5 ␮m nitrocellulose). Cells and filters were digested by the addition of 300 ␮L ultrapure HNO3 (JT Baker), incubated overnight at 80 ◦ C then diluted to 5 mL with deionised water (R > 18 M cm, organic carbon 85%. To generate the standard curve, qPCR reactions were performed using cDNA serial dilutions of 1/5, 1/25, 1/125 and 1/625. qPCR reactions were performed using the SYBR Green FASTA ABI Universal PCR Master Mix (AmpliTaq Gold DNA Polymerase) with a 1/5 dilution of the reverse transcription products and a primer concentration of 200 nM in a final volume of 10 ␮L. PCR conditions were: 2 min at 50 ◦ C, 3 min at 95 ◦ C followed by 50 cycles of dual temperature (5 s at 95 ◦ C and 30 s at 60 ◦ C). The data were analysed using the SDS 2.2 sequence detection software (Applied Biosystems). Relative mRNA levels were analysed using the 2−CT method (Livak and Schmittgen, 2001) with the threshold cycle (CT), i.e. the cycle at which an increase in the fluorescence is statistically significant from the background, in the exponential phase of amplification. Non-induced control genes 18S rRNA and N1 201 (a gene with no observable changes in intensity over the range of metals tested, Simon et al., 2008) were used for normalisation of the qPCR. 3.7. Data analysis For each gene and each condition, mRNA expression levels were divided by those obtained for an identical experiment containing no metals in order to obtain the fold change (FC). The following IA model was used to predict effects in the presence of the two metals: FCM = 1 − ((1 − FCA ) + (1 − FCB ))

(3)

where FCM is the fold change following exposure to the metal mixture (M) or to the individual metals (A or B). When measured levels of induction for the mixture fit this model, the response is considered as additive relative to the responses for single metals. When observed levels of induction are greater than predicted, metal interactions are assumed to be synergistic while less induction than predicted suggests that interactions are antagonistic.

Fig. 1. Gene transcription normalised to control (no metal) exposures, represented as fold change induction (FC). Cells were exposed to 1.0 ␮M Cd2+ over 1–8 h and mRNA expression was measured for: (a) AOT4 (), PBD1 (♦) and CHLH1 ( ); (b) SDC1 (), DRP1 () and METE (ⵦ); (c) SIR1 (), NUO11 (䊉) and LCI11 (). Error bars represent the standard errors obtained from three biological replicates.

4. Results 4.1. Biomarker genes exhibit temporal patterns of Cd regulation The dynamics of the regulation of the nine biomarker genes were characterized by quantifying their mRNA levels during exposure to 1.16 ␮M Cd2+ over 8 h (Fig. 1). Diverse temporal patterns of mRNA expression were observed. The earliest induction was observed for AOT4, DRP1, SDC1; these genes were transcriptionally maximally induced by 2 h (FC of 8.0, 3.1 and 4.7, respectively) with their mRNA levels declining thereafter. LCI11 mRNA expression increased over the first 4 h (8.2 FC) and then gradually declined to the basal level by 8 h. PBD1 showed a broad temporal pattern of induction, reaching maximal mRNA levels between 4 and 6 h of exposure. SIR1 showed a bimodal temporal pattern of weak transcriptional induction, with maxima at 1–2 and 6–8 h. Finally, CHLH1, METE and NUO11 were not transcriptionally induced by Cd2+ in this study, but rather their mRNA expression was either unchanged (METE) or slightly repressed (65% and 50% repression for CHLH1 and NUO11, respectively). Since the highest levels of induction provide the best analytical signal for biomarkers, an

C.M. Hutchins et al. / Aquatic Toxicology 100 (2010) 120–127

123

Fig. 2. Intracellular metal concentrations (mol/cm2 ) for: (a) Cu or (b) Pb in single metal exposures (open symbols) or in the presence of 0.5 ␮M Cd2+ (closed symbols). Intracellular Cd concentrations () are presented for the mixtures containing (c) Cu and (d) Pb. Intracellular Cd concentrations for the control exposure (0.5 ␮M Cd, no additional metal) are represented by the arrow. Error bars represent the standard error of three biological replicates.

exposure time of 2 h was deemed appropriate for the subsequent metal mixture experiments. In order to estimate NOTEL values for Cd2+ in C. reinhardtii, the same time course experiment was carried out at lower Cd2+ concentrations (0.01 or 0.11 ␮M, Fig. S1). At 0.11 ␮M Cd2+ , transcriptional induction was observed only for AOT4 (max. 2.3 FC) and LCI11 (max. 3.4 FC). At 0.01 ␮M Cd2+ , no significant effects were observed for any of the genes, suggesting that NOTEL values using the genes investigated were between 0.01 and 0.11 ␮M Cd2+ .

4.2. Intracellular metal concentrations Since toxicity is presumed to be a product of bioaccumulated metal, metal biouptake was monitored in order to assess its relationship to the elicited transcriptomic response. The addition of a second metal to the experimental medium is predicted either to have no effect (i.e. no interaction or binding to an independent uptake site) or to decrease the biouptake of the first metal (i.e. competition for a similar uptake site) (Slaveykova and Wilkinson, 2005). In contrast, toxicity assessments of metal mixtures can show additive, synergistic or antagonistic effects when compared to the exposure of a single metal (Norwood et al., 2003). In the absence of Cd, Cu and Pb biouptake increased as a function of the external metal concentration (Cu: 0.046–0.43 nmol/cm2 ; Pb: 0.01–0.33 nmol/cm2 ) (open symbols in Fig. 2a and b). When 0.5 ␮M Cd2+ was added to the experimental media, Cu and Pb uptake were not significantly altered except at the highest Pb2+ concentration (5.0 ␮M) (closed symbols in Fig. 2a and b). Under this condition, the addition of 0.5 ␮M Cd2+ doubled Pb biouptake (from 0.33 ± 0.02 to 0.72 ± 0.06 nmol/cm2 ) (Fig. 2b). While an enhancement of biouptake was unexpected, similar increases have been observed previously for Cu uptake in the presence of Zn, Pb and Ni in Chlorella kesslerii (Hassler et al., 2004), for Ni uptake in C. reinhardtii in the presence of Pb (Worms and Wilkinson, 2007) and for Pb uptake in the presence of Cu for C. reinhardtii (Chen et al., 2010). In contrast, in the presence of the highest Cu and Pb concentrations, Cd biouptake decreased, consistent with competitive interactions at the uptake sites (Fig. 2c, d). Most notably, Cd biouptake decreased by 53% and 77% (0.026 and 0.012 ± 0.005 nmol/cm2 ) when 1.0 or 5.0 ␮M Pb2+ was added to the exposure media (Fig. 2d). Similarly, in the presence of 5.0 ␮M Cu2+ , Cd biouptake was also inhibited by 25% (Fig. 2c). If the Cd-regulated genes selected for this study are indeed Cd specific, the observed decrease in Cd biouptake implies that a similar decrease in the transcriptomic response would be expected.

4.3. Transcriptomic effects of Cu and Pb on the Cd-regulated genes (single metal exposures) In order to assess the specificity of the biomarker genes to Cd and to determine the extent to which transcriptional regulation was reflected by metal biouptake, biomarker gene mRNA levels were quantified in the same experimental systems as above. If the mRNA expression of the biomarker genes is specific for Cd, then both in the absence and presence of Cd, mRNA levels should be constant as Cu2+ or Pb2+ concentrations are increased from 0.05 to 5.0 ␮M. In fact, for the single metal exposures, mRNA expression changed significantly with an increasing concentration of Cu or Pb (open symbols in Fig. 3: Cu and Fig. 4: Pb). The qPCR results suggested that the transcriptomic responses were non-metal specific, with the exception of DRP1, NUO11 and LCI11, which showed no change in mRNA expression levels. In general, biomarker mRNA levels were directly correlated with Cu2+ or Pb2+ concentrations, although maximum mRNA expression was often observed at 1.0 ␮M Cu2+ or Pb2+ (Cu: SDC1, SIR1; Pb: AOT4, PBD1, SIR1) rather than at 5 ␮M (Cu: AOT4, METE, CHLH1; Pb: METE, SDC1). There were a couple of exceptions: PBD1 expression inversely correlated with [Cu2+ ] while expression levels of CHLH1 inversely correlated with [Pb2+ ]. Finally, Pb and Cu had different effects on the biomarker mRNA levels. For example, PBD1 mRNA levels increased from 2.6 to 4.3 FC as Pb2+ concentrations increased from 0.05 to 1.0 ␮M whereas the same variation in Cu2+ caused a decrease from 2.7 to 0.7 FC (Figs. 3b and 4b). Similarly, marked differences in AOT4 mRNA levels were observed for Cu and Pb. Concentration-dependent regulation of mRNA levels of these genes in cells exposed to Cu or Pb alone suggests that the interpretation of gene response in multiple metal solutions will be even more complex. 4.4. Transcriptomic effects of binary metal mixtures In order to assess transcriptomic responses in mixed metal solutions and to determine if mRNA levels were predictably based on single metal results, cells were exposed to similar levels of Cu2+ or Pb2+ (0.05–5.0 ␮M) but in the presence of 0.5 ␮M Cd2+ . For the majority of the tested genes, the addition of 0.5 ␮M Cd2+ (solid points in Figs. 3 and 4) increased mRNA levels relative to cells exposed only to the Cu2+ or Pb2+ (open points in Figs. 3 and 4). For example, the addition of 0.5 ␮M Cd2+ generally enhanced AOT4 induction with respect to exposures Cu or Pb alone (Fig. 3a and b). Furthermore, the highest concentrations of Pb appeared to have a greater negating effect on mRNA expression than did Cu, consistent with the greater observed effect of Pb on Cd biouptake. On the

124

C.M. Hutchins et al. / Aquatic Toxicology 100 (2010) 120–127

Fig. 3. mRNA levels relative to no metal controls (FC) are shown for AOT4 (a), PBD1 (b), METE (c), SDC1 (d), CHLH1 (e) and SIR1 (f) exposed to single Cu (, 0.05–5.0 ␮M) or Cd, Cu mixtures (䊉, 0.05–5.0 ␮M Cu; 0.5 ␮M Cd). The arrow indicates the mRNA levels in 0.5 ␮M Cd exposed cells. The horizontal line at FC = 1 indicates the level obtained for identical conditions in the absence of metal. Error bars represent the standard error of three biological replicates.

other hand, PBD1 mRNA levels were not significantly different in the presence or absence of Cd (Fig. 4b) and the induction of CHLH1 was similar in all cases, i.e. when alga was exposed to binary (Cd/Cu; Cd/Pb) or single (Cu, Pb) metal solutions (Figs. 3e and 4e). The extent of interactive effects in the metal mixtures was not uniform, differing for each gene. For example, mRNA expression levels for CHLH1 and METE were well predicted by the simple additive IA model (Fig. 5) as was to be expected due to minimal induction in 0.5 ␮M Cd2+ only exposures (arrow Figs. 3 and 4). On the other hand, despite no differential mRNA expression in 0.5 ␮M Cd2+ only exposures (arrow Figs. 3 and 4), SDC1 mRNA expression levels in the mixtures (Fig. 6) were higher than predicted from single metal exposures, suggesting a synergistic interaction between the metals. In contrast, for PBD1, metal interactions were antagonistic for both binary mixtures, at all concentrations. No systematic trend in the data was observed for AOT4 or SIR1, with varying synergistic or antagonistic interactions that depended on the precise metal mixture and metal concentrations. Improved relationships were not obtained when biomarker mRNA levels were plotted as a function of internalised metal concentrations (Figs. S1 and S2). For single metal exposures, expression levels generally increased

as a function of internalised metal, until a maxima (open symbols Figs. S1 and S2). On the other hand, when single and binary metal exposures were combined, it became clear that no single concentration of intracellular metal (Cu, Pb or Cd) was a good predictor of the observed transcriptomic effects (S1 (Cu), S2 (Pb) and S3 (Cd)). Such an observation clearly complicates the ability to use single genes as biomarkers for metals in solutions containing multiple metals. 5. Discussion 5.1. Kinetic response following Cd exposure Differential mRNA levels, as a biomarker for the presence of a toxicant, only provides a snapshot of the temporally complex transcriptional response. For the majority of the examined Cd biomarkers, short exposures of 2–4 h were sufficient to produce maximal or near-maximal changes in the mRNA levels of the biomarker genes. Dynamic patterns of mRNA expression after Cd treatment have been reported for a range of genes and species (Ren et al., 2003; Soetaert et al., 2007). In C. reinhardtii, transient transcriptional induction of genes encoding two thioredoxin isoforms

Fig. 4. mRNA levels relative to no metal controls (FC) are shown for AOT4 (a), PBD1 (b), METE (c), SDC1 (d), CHLH1 (e) and SIR1 (f) exposed to Pb (ⵦ, 0.05–5.0 ␮M) or Cd, Pb mixtures (, 0.05–5.0 ␮M Pb; 0.5 ␮M Cd). The arrow indicates the mRNA levels in 0.5 ␮M Cd exposed cells. The horizontal line at FC = 1 indicates the level obtained for identical conditions in the absence of metal. Error bars represent the standard error of three biological replicates.

C.M. Hutchins et al. / Aquatic Toxicology 100 (2010) 120–127

125

Fig. 5. Fits of observed gene transcription in Cd,Cu () and Cd,Pb (䊉) mixtures with predicted values for genes under the assumption of a simple additive IA model: (a) AOT4, (b) PBD1, (c) METE, (d) SDC1, (e) CHLH1 and (f) SIR1. The diagonal line bisecting the graph represents the values predicted for additive effects, i.e. in the absence of synergy or inhibition.

has been reported following exposure to 100 ␮M Cd, with mRNA expression maxima (4 FC) at 2 and 3 h, respectively (Lemaire et al., 1999). These data have clear implications for the appropriate exposure of C. reinhardtii as in situ biomarkers. Obviously further investigation is required to confirm corresponding changes to the gene product and to validate this observation for a greater range of metals and metal-induced genes. 5.2. Single metal exposures All genes examined in this study were initially screened for their specificity for Cd using microarray technology (Simon et al., 2008). Regulation of the six Cd2+ responsive genes: AOT4, DRP1, SDC1, LCI11, PBD1, SIR1, was generally consistent with reported previous results (Simon et al., 2008), with a few key differences. After a 2 h exposure to 1.2 ␮M Cd2+ , transcriptional AOT4 induction was lower in this study (8 FC as opposed to 43 FC) while that of LCI11 and SIR1 was greater (6.6 vs. 2.5 FC and 1.7 vs. 0.4 FC). Furthermore, in contrast to the previous results, many of the genes were not Cd specific in their regulation, although admittedly, they had not been previously rigorously tested for the effects of Cu and Pb. Even in the single metal exposures (Cd2+ , Cu2+ or Pb2+ ), biomarker levels were not linearly related to toxicant concentration. Such concentration-dependent response curves, in which the genes are transcriptionally up-regulated at low concentrations,

Fig. 6. Gene transcription signatures of exposure to 1.0 ␮M Cd, Pb, Cu, or two metal mixtures CdPb and CdCu. Error bars represent the standard error of three biological replicates.

with less induction at higher concentrations, are not uncommon in genotoxicity experiments. Recent studies investigating gene responses following toxicant exposure highlight the importance of considering a dose-dependence when interpreting expression data (Jamers et al., 2009). Observations for which differential mRNA expression increased with concentration to a threshold prior to stabilizing or decreasing clearly shows that two vastly different metal concentrations can induce the same level of mRNA expression of a gene. Unfortunately, this limits the practical concentration range for these genes as biomarkers of metal toxicity and complicates the identification of ECx values for individual toxicants (Dardenne et al., 2008). In addition, as the metal concentrations increase, the number of differentially expressed genes related to general stress responses also increase, overshadowing the metal specific modes of toxicity (Poynton et al., 2008a). Indeed, of the biomarker genes encoding proteins of known function, none have previously been shown to be directly involved in Cd tolerance. For example, SDC1 encodes serine decarboxylase, a catalyst in the production of glycinebetaine, known to be part of a general abiotic stress response (Sakamoto and Murata, 2000). Genes encoding for serine decarboxylase have been shown to be induced by Cu in the aquatic plant Lemna gibba (Akhtar et al., 2005) and by Ni and Mn in Arabidopsis (Fujimori and Ohta, 2003). Similarly, CHLH1 encodes the H subunit of Mg chelatase, an enzyme in chlorophyll biosynthesis, which is also the receptor component of a binding protein (ABAR) for abscisic acid (ABAR) (Shen et al., 2006), the biosynthesis of which is regulated by general abiotic stressors (Gong et al., 2008). The products of both SIR1 and METE function in the biosynthesis of methionine, which is subsequently required for the synthesis of cysteine, glutathione and phytochelatines. Both glutathione and phytochelatines are non-metal specific and can be up-regulated following exposure to Cu (Wu et al., 2007; Guo et al., 2008; Helbig et al., 2008), Pb (Figueroa et al., 2008) or Cd (Maier et al., 2003). AOT4 encodes a predicted amino acid transporter (Simon et al., 2008). Its induction in the presence of increasing concentrations of Cu and Pb suggests that its role is not limited to Cd (Fig. 2). In this study, the absence of significant differential mRNA expression below 0.01 ␮M Cd2+ suggested that the NOTEL is in this concentration range for C. reinhardtii. This value is in similar range (0.5 nM–0.16 ␮M Cd) to that identified by Poynton et al. (2008a) for Daphnia magna. With regard to Cu and Pb, the observed significant mRNA expression of a number of genes (PBD1,

126

C.M. Hutchins et al. / Aquatic Toxicology 100 (2010) 120–127

SIR1, SDC1, LCI11) at the lowest exposure concentration (0.05 ␮M) suggested that for C. reinhardtii, NOTEL values would be at concentrations below 0.05 ␮M. This observation is again quite similar to the results for Cu of Poynton et al. (2008a,b), who observed 25 differentially transcribed genes in D. magna exposed to 0.03 ␮M Cu while only 1 differentially transcribed gene was observed at an exposure concentration of 0.015 ␮M Cu. In order to ensure the accurate determination of NOTEL concentrations for metal exposure in C. reinhardtii, it is essential that substantially larger numbers of genes be investigated and that future studies assessing genomic metal biomarkers incorporate appropriately low (environmentally relevant) exposure concentrations to best represent in situ contamination.

The lack of metal specificity of the nine Cd biomarkers, the non-linearity of concentration-response curves and the observed interactive effects in mixed metal solutions limit the practical employment of these genes as unique biomarkers of contaminated systems. An alternative to the use of single gene responses as a biomarker in contaminated systems is to cluster the transcriptomic responses for a number of genes in order to provide a distinct signature for single and multiple metal solutions. For example, distinct signature patterns from the nine biomarker genes examined here could feasibly be employed to identify metal contamination in situ (Fig. 5). While the analysis of mRNA expression clusters is considerably more complex than the single gene responses, it may provide the most useful means of assessing the toxicological responses of complex multiple metal solutions.

5.3. Binary metal exposures 6. Conclusion Biological effects of exposure to a trace metal are most often related to the external concentrations of free metal (free ion activity model (FIAM); Campbell, 1995; Kola and Wilkinson, 2005) or to the concentration of metal bound to sensitive biological sites (biotic ligand model (BLM); Slaveykova and Wilkinson, 2005). The FIAM and BLM models will always predict lower (or unchanged) bioaccumulation and toxicity in the presence of a second competing ion. While decreased bioaccumulation is indeed often observed due to competition effects (Kola and Wilkinson, 2005; Worms et al., 2007; Borgmann et al., 2008), additive, synergistic or antagonistic effects are observed when applying toxicological endpoints to metal mixtures (Borgmann et al., 2008; Vandenbrouck et al., 2009). For binary metal mixtures of Cd in the presence of increasing Cu or Pb, the potential toxicant interaction was evaluated by a comparison of mRNA levels with the levels predicted by a simple additive IA model based on single metal exposures. Deviations from the predictive model for PBD1, AOT4, SDC1 and SIR1 suggest that binary metal exposures can cause interactive effects which differ at the individual gene level. The variability of interactive effects across the genes suggests they are the product of intracellular feedback or regulatory processes (i.e. translational efficiency, mRNA stability, protein stability, etc.) that may be specific to each gene and each binary metal combination. Given that mRNA expression experiments performed on a given sample showed additive, synergistic and antagonistic interactions, depending on the gene examined, competition effects on biouptake cannot be invoked to explain the results, i.e. internal rather than external processes are responsible for the differences in mRNA expression. In equitoxic binary metal mixtures of Ni with Cu and Pb, Vandenbrouck et al. (2009) also showed that mRNA expression patterns were not merely the simple sum of their individual compounds. Instead, solutions containing several metals triggered multiple additional response pathways and interactive molecular responses (Vandenbrouck et al., 2009). The interactive effects in multiple metal solutions also depended on the particular metals in binary exposures and the concentration. Complex concentration ratios and concentration-dependent synergistic/antagonistic interactions between toxicants, as observed for AOT4 and SIR1, are common in assessments of mixture toxicity (Jonker et al., 2009) and have been reported for multiple metal solutions including cadmium and lead (Bae et al., 2001) and binary solutions of Cu and diazinon (Van der Geest et al., 2000; Banks et al., 2003). Clearly, the modelling of toxicant effects, based on single contaminant exposures, will provide only a preliminary understanding of how contaminants interact to affect gene expression. Metal interactions and concentration-dependent effects, in addition to the gene dependent nature of the metal interactions (i.e. additive, synergistic or antagonistic), complicate the practical application of using mRNA expression of a gene as a biomarker of metal contamination.

The identification of biomarkers for use in metal ecotoxicogenomics requires a thorough understanding of both the kinetics and specificity of the targeted genes. Extending our knowledge of nine Cd-regulated genes, the present study identified distinct temporal dependence of their mRNA expression in C. reinhardtii following Cd exposure. For the investigated genes, exposures of 2–4 h maximised the level of transcriptional induction or repression. Nonetheless, the observed lack of Cd specificity for six of the genes, with varying levels of transcriptional induction/repression observed in the presence of Cu2+ or Pb2+ , greatly reduced the practicality of using these genes as individual Cd specific biomarkers. In multiple metal solutions, the presence of more than one contaminant complicates the transcriptomic response via processes in the external exposure solution (i.e. competitive reduction of Cu uptake) and gene specific interactive (synergistic/antagonistic) metal effects. The observed lack of specificity, the absence of a strictly concentration-dependent induction and the observation of synergistic, antagonistic and additive metal interactions limits the potential of a using a single gene as a Cd specific biomarker. In the future, the development of a signature transcription profile using a cluster of genes will likely provide the best potential for a biomarker which can integrate both temporal and multiple contaminant effects. Acknowledgments The authors gratefully acknowledge the support of the NSERC MITHE Research Network and the Fonds Quebecois de la recherche sur la nature et les technologies (FQRNT team grant KJW, WZ). A complete list of MITHE sponsors is available at www.mithe-morg. Technical assistance from P. Chagnon is greatly appreciated. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.aquatox.2010.07.017. References Akhtar, T.A., Lampi, M.A., Greenberg, B.M., 2005. Identification of six differentially expressed genes in response to copper exposure in the aquatic plant Lemna gibba (duckweed). Environ. Toxicol. Chem. 24, 1705–1715. Ankley, G.T., Daston, G.P., Degitz, S.J., Denslow, N.D., Hoke, R.A., Kennedy, S.W., Miracle, A.L., Perkins, E.J., Snape, J., Tillitt, D.E., Tyler, C.R., Versteeg, D., 2006. Toxicogenomics in regulatory ecotoxicology. Environ. Sci. Technol. 40, 4055–4065. Bae, D.-S., Gennings, C., Carter Jr, W.H., Yang, R.S.H., Campain, J.A., 2001. Toxicological interactions among arsenic, cadmium, chromium, and lead in human keratinocytes. Toxicol. Sci. 63, 132–142. Banks, K.E., Wood, S.H., Matthews, C., Thuesen, K.A., 2003. Joint acute toxicity of diazinon and copper to Ceriodaphnia dubia. Environ. Toxicol. Chem. 22, 1562–1567.

C.M. Hutchins et al. / Aquatic Toxicology 100 (2010) 120–127 Borgmann, U., Norwood, W.P., Dixon, D.G., 2008. Modelling bioaccumulation and toxicity of metal mixtures. Hum. Ecol. Risk Assess. 14, 266–289. Campbell, P.G.C., 1995. Interactions between trace metals and aquatic organisms: a critique of the free-ion activity model. In: Tessier, A., Turner, D.R. (Eds.), Metal Speciation and Bioavailability in Aquatic systems. John Wiley, Chichester, UK. Chen, Z., Zhu, L., Wilkinson, K.J., 2010. Validation of the biotic ligand model (BLM) in metal mixtures: bioaccumulation of lead and copper. Environ. Sci. Technol. 44, 3580–3586. Chu, K.W., Chow, K.L., 2002. Synergistic toxicity of multiple heavy metals is revealed by a biological assay using a nematode and its transgenic derivative. Aquat. Toxicol. 61, 53–64. Dardenne, F., Nobels, I., De Coen, W., Blust, R., 2008. Mixture toxicity and gene inductions: can we predict the outcome. Environ. Toxicol. Chem. 27, 509–518. Figueroa, J.A.L., Wrobel, K., Afton, S., Caruso, J.A., Gutierrez Corona, J.F., Wrobel, K., 2008. Effect of some heavy metals and soil humic substances on the phytochelatin production in wild plants from silver mine areas of Guanajuato, Mexico. Chemosphere 70, 2084–2091. Fujimori, K., Ohta, D., 2003. Heavy metal induction of Arabidopsis serine decarboxylase gene expression. Biosci. Biotechnol. Biochem. 67, 896–898. Gong, Z., Chinnusamy, V., Zhu, J.-K., 2008. The molecular networks of abiotic stress signaling. Annu. Plant Rev., 33. Gorman, D.S., Levine, R.P., 1965. Cytochrome f and plastocyanin: their sequence in the photosynthetic electron transport chain of Chlamydomonas reinhardi. Proc. Natl. Acad. Sci. U.S.A. 54, 1665–1669. Guo, W.-J., Meetam, M., Goldsbrough, P.B., 2008. Examining the specific contributions of individual Arabidopsis metallothioneins to copper distribution and metal tolerance. Plant Physiol. 146, 1697–1706. Hassler, C.S., Slaveykova, V.I., Wilkinson, K.J., 2004. Discriminating between intraand extracellular metals using chemical extractions. Limnol. Oceanogr. Methods 2, 237–246. Helbig, K., Bleuel, C., Krauss, G.J., Nies, D.H., 2008. Glutathione and transition-metal homeostasis in Escherichia coli. J. Bacteriol. 190, 5431–5438. Jamers, A., Blust, R., de Coen, W., 2009. Omics in algae: paving the way for a systems biological understanding of algal stress phenomena. Aquat. Toxicol. 92, 114–121. Jonker, M.J., Svendsen, C., Bedaux, J.J.M., Bongers, M., Kammenga, J.E., 2009. Significance testing of synergistic/antagonistic dose level-dependent, or dose ratio-dependent effects in mixture dose–response analysis. Environ. Toxicol. Chem. 24, 2701–2713. Kola, H., Wilkinson, K.J., 2005. Cadmium uptake by a green alga can be predicted by equilibrium modelling. Environ. Sci. Technol. 39, 3040–3047. Lemaire, S., Keryer, E., Stein, M., Schepens, I., Issakidis-Bourguet, E., Gerard-Hirne, C., Miginiac-Maslow, M., Jacquot, J.-P., 1999. Heavy-metal regulation of thioredoxin gene expression in Chlamydomonas reinhardtii. Plant Physiol. 120, 773– 778. Livak, K., Schmittgen, T.D., 2001. Analysis of relative gene expression data using real-time quantitative PCR and the 2T −C method. Methods 25, 402–408. Lobenhofer, E.K., Cui, X., Bennett, L., Cable, P.L., Merrick, B.A., Churchill, G.A., Afshari, C.A., 2004. Exploration of low-dose estrogen effects: identification of no observed transcriptional effect level (NOTEL). Toxicol. Pathol. 32, 482–492. Maier, T., Yu, C., Kullertz, G., Clemens, S., 2003. Localization and functional characterization of metal-binding sites in phytochelatin synthases. Planta 218, 300– 308. Moseley, J.L., Chang, C.-W., Grossman, A.R., 2006. Genome-based approaches to understanding phosphorus deprivation responses and PSR1 control in Chlamydomonas reinhardtii. Eukaroytic Cell 5, 26–44.

127

Norwood, W.P., Borgmann, U., Dixon, D.G., Wallace, A., 2003. Effects of metal mixtures on aquatic biota: a review of observations and methods. Hum. Ecol. Risk Assess. 9, 795–811. Neumann, N.F., Galvez, F., 2002. DNA microarrays and toxicogenomics: applications for ecotoxicology? Biotechnol. Adv. 20, 391–419. Poynton, H.C., Loguinov, A.V., Varshavsky, J.R., Chan, S., Perkins, E.J., Vulpe, C.D., 2008a. Gene expression profiling in Daphnia magna. Part 1: concentrationdependent profiles provide support for the no observed transcriptional effect level. Environ. Sci. Technol. 42, 6250–6256. Poynton, H.C., Zuzow, R., Loguinov, A.V., Perkins, E.J., Vulpe, C.D., 2008b. Gene expression profiling in Daphnia magna. Part 1: validation of a copper specific gene expression signature with effluent from two copper mines in California. Environ. Sci. Technol. 42, 6257–6263. Preston, S., Coad, N., Townend, J., Killham, K., Paton, G.I., 2000. Biosensing the acute toxicity of metal interactions: are the additive, synergistic, or antagonistic? Environ. Toxicol. Chem. 19, 775–780. Ren, X.-Y., Zhou, Y., Zhang, J.-P., Feng, W.-H., Jiao, B.-H., 2003. Metallothionein gene expression under different time in testicular sertoli and spermatogenic cells of rats treated with cadmium. Reprod. Toxicol. 17, 219– 227. Sakamoto, A., Murata, N., 2000. Genetic engineering of glycinebetaine synthesis in plants: current status and implication for enhancement of stress tolerance. J. Exp. Bot. 51, 81–88. Shen, Y.-Y., Wang, X.-F., Wu, F.-Q., Du, S.-Y., Cao, Z., Shang, Y., Wang, X.-L., Peng, C.-C., Yu, X.-C., Zhu, S.-Y., Fan, R.-C., Xu, Y.-H., Zhang, D.-P., 2006. The Mg-chelatase H subunit is an abscisic acid receptor. Nature 443, 823– 826. Simon, D.F., Descombes, P., Zerges, W., Wilkinson, K.J., 2008. Global expression profiling of Chlamydomonas reinhardtii exposed to trace levels of free cadmium. Environ. Toxicol. Chem. 27, 1668–1675. Slaveykova, V.I., Wilkinson, K.J., 2005. Predicting the bioavailability of metals and metal complexes: critical review of the biotic ligand model. Environ. Chem. 2, 9–24. Soetaert, A., Vandenbrouck, T., van der Ven, K., Maras, M., van Remortel, P., Blust, R., de Coen, W.M., 2007. Molecular responses during cadmium-induced stress in Daphnia magna: integration of differential gene expression with higher-level effects. Aquat. Toxicol. 83, 212–222. Stratton, G.W., 1988. Method for determining toxicant interaction effects toward microorganisms. Toxicol. Assess. 3, 345–353. Vandenbrouck, T., Soetaert, A., van der Ven, A., Blust, R., de Coen, W., 2009. Nickel and binary metal mixture responses in Daphnia magna: molecular fingerprints and (sub)organismal effects. Aquat. Toxicol. 92, 18–29. Van der Geest, H.G., Greve, G.D., Boivin, M.E., Kraak, M.H.S., Van Gestel, C.A.M., 2000. Mixture toxicity of copper and diazinon to larvae of the mayfly (Ephoron virgo) judging additivity at different effect levels. Environ. Toxicol. Chem. 19, 2900–2905. Worms, I.A.M., Parthasarathy, N., Wilkinson, K.J., 2007. Ni uptake by a green alga. 1. Validation of equilibrium models for complexation effects. Environ. Sci. Technol. 41, 4258–4263. Worms, I.A.M., Wilkinson, K.J., 2007. Ni uptake by a green alga. 2. Validation of the equilibrium models for competition effects. Environ. Sci. Technol. 41, 4264–4270. Wu, L., Cheng, S., Yang, L., Wu, Z., 2007. Stress responses and resistance mechanism of Canna indica linn to cadmium and copper. Nongye Huanjing Kexue Xuebao 26, 1365–1369.