Assessing the impact of Benzo[a]pyrene on Marine Mussels: Application of a novel targeted low density microarray complementing classical biomarker responses

Abstract

Despite the increasing use of mussels in environmental monitoring and ecotoxicological studies, their genomes and gene functions have not been thoroughly explored. Several cDNA microarrays were recently proposed for Mytilus spp., but putatively identified partial transcripts have rendered the generation of robust transcriptional responses difficult in terms of pathway identification. We developed a new low density oligonucleotide microarray with 465 probes covering the same number of genes. Target genes were selected to cover most of the well-known biological processes in the stress response documented over the last decade in bivalve species at the cellular and tissue levels. Our new 'STressREsponse Microarray' (STREM) platform consists of eight sub-arrays with three replicates for each target in each sub-array. To assess the potential use of the new array, we tested the effect of the ubiquitous environmental pollutant benzo[a]pyrene (B[a]P) at 5, 50, and 100 μg/L on two target tissues, the gills and digestive gland, of Mytilus galloprovincialis exposed invivo for three days. Bioaccumulation of B[a]P was also determined demonstrating exposure in both tissues. In addition to the well-known effects of B[a]P on DNA metabolism and oxidative stress, the new array data provided clues about the implication of other biological processes, such as cytoskeleton, immune response, adhesion to substrate, and mitochondrial activities. Transcriptional data were confirmed using qRT-PCR. We further investigated cellular functions and possible alterations related to biological processes highlighted by the microarray data using oxidative stress biomarkers (Lipofuscin content) and the assessment of genotoxicity. DNA damage, as measured by the alkaline comet assay, increased as a function of dose.DNA adducts measurements using 32P-postlabeling method also showed the presence of bulky DNA adducts (i.e. dG-N2-BPDE). Lipofiscin content increased significantly in B[a]P exposed mussels. Immunohistochemical analysis of tubulin and actin showed changes in cytoskeleton organisation. Our results adopting an integrated approach confirmed that the combination of newly developed transcriptomic approcah, classical biomarkers along with chemical analysis of water and tissue samples should be considered for environmental bioimonitoring and ecotoxicological studies to obtain holistic information to assess the impact of contaminants on the biota.

Fig 1. Mytilus galloprovincialis gene expression profiles of gills tissue in animals exposed to B[a]P (5μg/L, 50μg/L and 100 μg/L).

The heat map (A) (Pearson correlation, complete linkage algorithm) reports the log2 relative expression level with respect to the reference condition. 205 differentially expressed genes were generated in at least one condition. Microarray data were analyzed using the Linear Mode for Microarray Analysis (LIMMA) software. B statistics with adjusted p value, 0.05 and B.0 were used as threshold for rejection of the null hypothesis (no variation). Supporting information to Fig 2 is presented in S2 Table and S3 Table. Venn diagram representation of gene expression patterns (Panel B) clearly depicted that only 14 DEGs are shared between the three experimental conditions. All DEGs are obtained respect to the control condition. Data used to generate the Venn-diagram were obtained from microarray analysis (S3 Table). Four biological replications of were considered.

Fig 2. Mytilus galloprovincialis gene expression profiles of digestive gland tissue in animals exposed to B[a]P (5μg/L, 50μg/L and 100 μg/L).

The heat map (A) (Pearson correlation, complete linkage algorithm) reports the log2 relative expression level with respect to the reference condition. 109 differentially expressed genes were generated in at least one condition. Microarray data were analyzed using the Linear Mode for Microarray Analysis (LIMMA) software. B statistics with adjusted p value, 0.05 and B.0 were used as threshold for rejection of the null hypothesis (no variation). Supporting information to Fig 3 is presented in S2 Table and S4 Table. Venn diagram representation of gene expression patterns (Panel B) clearly depicted that no DEGs are shared between the three experimental conditions. All DEGs are obtained respect to the control condition. Data used to generate the Venn-diagram were obtained from microarray analysis (S4 Table). Four biological replications of were considered.

Fig 3. Over-representation analysis of DEGs in the gills tissue of mussels exposed to B[a]P (5μg/L, 50μg/L and 100 μg/L).

Showed are: experimental condition; biological processes; Number of up-regulated genes; Number of down-regulated genes. The over-represented biological processes in B[a]P-exposed animals versus control. (S3 Table).

Fig 4. Over-representation analysis of DEGs in the digestive gland tissue of mussels exposed to B[a]P (5μg/L, 50μg/L and 100 μg/L).

Showed are: experimental condition; biological processes; Number of up-regulated genes; Number of down-regulated genes. The over-represented biological processes in B[a]P-exposed animals versus control. (S4 Table).

Fig 5

DNA strand breaks, as measured by the comet assay, in mussel (a) gill and (b) digestive gland cells after exposure to 0–100 μg/L BaP (0.02% DMSO; n = 9). Mismatching lowercase letters indicate differences between treatments according to Tukeys HSD applied as post hoc tests following a one-way ANOVA (p< 0.05).

Fig 6. B[a]P-derived-DNA adduct levels, as measured by ³²P-postlabelling, in mussel digestive gland cells, after exposure to 0–1000 μg/L B[a]P for 3 days.

Data are shown with the sexes combined (a) and separated (for those concentrations that resulted in detectable numbers of adducts)(b). All treatments had an n of 9 in total but the distribution of sexes was not even, therefore group n values are indicated in (b). Mismatching lowercase letters indicate differences between groups according to Tukeys HSD applied as post hoc tests following a generalised linear model (p< 0.05; n = 9). Insert: Representative autoradiographic profile of DNA adducts, measured by thin-layer chromatography ³²P-postlabelling, in digestive gland tissue exposed to B[a]P; no B[a]P-derived DNA adducts were detected in control (untreated) tissue (data not shown). The origin (OR), at the bottom left-hand corners, was cut off before exposure. The arrow indicates 10-(deoxyguanosin-N²-yl)-7,8,9-trihydroxy-7,8,9,10-tetrahydro-B[a]P (dG-N²-BPDE).

Fig 7. Effects of B[a]P on lipofuscin lysosomal content in the digestive gland cells of mussels, evaluated in cryostat tissue sections by the Schmorl reaction.

Data, expressed as percent change with respect to control values, represent the mean ± SD of at least 10 replicates. * = p < 0.05 (Mann-Whitney U-test).

Fig 8

(A-D) Anti-tubulin immunohistochemical staining (Chromeo™ 488 conjugated secondary antibody) of digestive gland tissue sections from mussels exposed to different experimental conditions (A = Control; B = B[a]P 5 μg/L; C = B[a]P 50 μg/L; D = B[a]P 100 μg/L). (E) Quantitative fluorescence analysis of anti-tubulin immunoreaction. Data are mean ± SD of at least five replicates; * = p < 0.05 (Mann-Whitney U-test).

Fig 9

Fluorescent staining of the F-actin cytoskeleton by Green Fluorescent Phalloidin Conjugate in digestive gland sections of mussels exposed to different experimental conditions (A = Control; B = B[a]P 5 μg/L; C = B[a]P 50 μg/L; D = B[a]P 100 μg/L).