High throughput nano-liter RT-qPCR to classify soil contamination using a soil arthropod

Abstract

Abstract

Background: To incorporate genomics data into environmental assessments a mechanistic perspective of interactions between chemicals and induced biological processes needs to be developed. Since chemical compounds with structural similarity often induce comparable biological responses in exposed animals, gene expression signatures can serve as a starting point for the assessment of chemicals and their toxicity, but only when relevant and stable gene panels are available. To design such a panel, we isolated differentially expressed gene fragments from the soil arthropod Folsomia candida , a species often used for ecotoxicological testing. Animals were exposed to two chemically distinct compounds, being a metal (cadmium) and a polycyclic aromatic hydrocarbon (phenanthrene). We investigated the affected molecular responses resulting from either treatment and developed and validated 44 qPCR assays for their responses using a high throughput nano-liter RT-qPCR platform for the analysis of the samples.

Results: Suppressive subtractive hybridization (SSH) was used to retrieve stress-related gene fragments. SSH libraries revealed pathways involved in mitochondrial dysfunction and protein degradation for cadmium and biotransformation for phenanthrene to be overrepresented. Amongst a small cluster of SSH-derived cadmium responsive markers were an inflammatory response protein and an endo-glucanase. Conversely, cytochrome P450 family 6 or 9 was specifically induced by phenanthrene. Differential expressions of these candidate biomarkers were also highly significant in the independently generated test sample set. Toxicity levels in different training samples were not reflected by any of the markers' intensity of expressions. Though, a model based on partial least squares differential analysis (PLS-DA) (with RMSEPs between 9 and 22% and R² s between 0.82 and 0.97) using gene expressions of 25 important qPCR assays correctly predicted the nature of exposures of test samples.

Conclusions: For the application of molecular bio-indication in environmental assessments, multivariate analyses obviously have an added value over univariate methods. Our results suggest that compound discrimination can be achieved by PLS-DA, based on a hard classification of the within-class rankings of samples from a test set. This study clearly shows that the use of high throughput RT-qPCR could be a valuable tool in ecotoxicology combining high throughput with analytical sensitivity.