Microarray data and gene expression statistics for Saccharomyces cerevisiae exposed to simulated asbestos mine drainage

Here we describe microarray expression data (raw and normalized), experimental metadata, and gene-level data with expression statistics from Saccharomyces cerevisiae exposed to simulated asbestos mine drainage from the Vermont Asbestos Group (VAG) Mine on Belvidere Mountain in northern Vermont, USA. For nearly 100 years (between the late 1890s and 1993), chrysotile asbestos fibers were extracted from serpentinized ultramafic rock at the VAG Mine for use in construction and manufacturing industries. Studies have shown that water courses and streambeds nearby have become contaminated with asbestos mine tailings runoff, including elevated levels of magnesium, nickel, chromium, and arsenic, elevated pH, and chrysotile asbestos-laden mine tailings, due to leaching and gradual erosion of massive piles of mine waste covering approximately 9 km2. We exposed yeast to simulated VAG Mine tailings leachate to help gain insight on how eukaryotic cells exposed to VAG Mine drainage may respond in the mine environment. Affymetrix GeneChip® Yeast Genome 2.0 Arrays were utilized to assess gene expression after 24-h exposure to simulated VAG Mine tailings runoff. The chemistry of mine-tailings leachate, mine-tailings leachate plus yeast extract peptone dextrose media, and control yeast extract peptone dextrose media is also reported. To our knowledge this is the first dataset to assess global gene expression patterns in a eukaryotic model system simulating asbestos mine tailings runoff exposure. Raw and normalized gene expression data are accessible through the National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO) Database Series GSE89875 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE89875).


Keywords:
Ultramafic Chrysotile asbestos Mine drainage Gene expression Microarray Saccharomyces cerevisiae Environmental toxicology 100 years (between the late 1890s and 1993), chrysotile asbestos fibers were extracted from serpentinized ultramafic rock at the VAG Mine for use in construction and manufacturing industries. Studies have shown that water courses and streambeds nearby have become contaminated with asbestos mine tailings runoff, including elevated levels of magnesium, nickel, chromium, and arsenic, elevated pH, and chrysotile asbestos-laden mine tailings, due to leaching and gradual erosion of massive piles of mine waste covering approximately 9 km 2 . We exposed yeast to simulated VAG Mine tailings leachate to help gain insight on how eukaryotic cells exposed to VAG Mine drainage may respond in the mine environment. Affymetrix GeneChip s Yeast Genome 2.0 Arrays were utilized to assess gene expression after 24-h exposure to simulated VAG Mine tailings runoff. The chemistry of mine-tailings leachate, mine-tailings leachate plus yeast extract peptone dextrose media, and control yeast extract peptone dextrose media is also reported. To our knowledge this is the first dataset to assess global gene expression patterns in a eukaryotic model system simulating asbestos mine tailings runoff exposure. Raw and normalized gene expression data are accessible through the National Center for Biotechnology Raw and normalized microarray data table  Water and media chemistry metadata table  One-way analysis of variance results gene table  Differentially expressed gene table  Experimental  factors Treatment (asbestos mine tailings-exposed) vs. control

Experimental features
Total RNA was recovered from four Saccharomyces cerevisiae strain NRRLY-12632 (ATCC s 18824) replicates exposed to simulated asbestos mine tailings runoff for a 24-h period and three control replicates.

Data source location
Lowell tailings pile, Vermont Asbestos Group Mine, Lowell, Vermont, USA (near 44°45'51.8"N 72°31'22.2"W) Data accessibility Microarray data and some sample metadata are available at NCBI GEO Series GSE89875: rhttps://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc ¼GSE89875. Water and media chemistry data tables as well as gene statistical data tables are available in this article.

Value of the data
This is the first global gene expression dataset from a eukaryotic model system exposed to asbestos mine tailings runoff.
Water chemistry of the simulated asbestos mine-tailings leachate used in this experiment is geochemically similar to runoff from the Vermont Asbestos Group mine and may well represent physio-chemical properties of other alkaline mine drainage systems.
Our thorough metadata descriptions, including water and growth media chemistry, maximize the utility of these microarray data to others studying the biological effects of asbestos mine drainage.
Both the normalized intensity data and gene expression profile data can be used in functional analysis and systems biology approaches to understanding complex biological processes affected by asbestos mine runoff.

Water and media chemistry data
Simulated asbestos mine drainage water chemistry shows similarities to the geochemistry of surface water samples collected previously from the VAG Mine by the Vermont Agency of Natural Resources (VT-ANR), particularly pH and conductivity for simulated asbestos mine drainage and impaired site Burgess Branch 5.0 [1]. Table 1 includes select chemical properties from simulated asbestos mine drainage from this research as well as four localities from the VT-ANR report [1], including Hutchins Brook 3.1 and Burgess Branch Tributary 8, two control sites that do not receive asbestos mine drainage runoff and two asbestos mine drainage-impaired sites downstream of the mine, Hutchins Brook 2.1 and Burgess Branch 5.0. These results establish the mine-tailings leachate prepared in the lab for this study as a reasonable substitute for VAG Mine drainage. Complete water chemistry for this study are available in Supplementary Table 1.
Control group media chemistry is notably different than simulated asbestos mine drainage media ( Table 2). Total magnesium concentration is considerably higher in simulated asbestos mine drainage than control media (100 mg/L and 2.6 mg/L, respectively). Additionally, the pH of the experimental group media suggests that simulated asbestos mine drainage water (pH 9.2) was buffered by the yeast extract peptone dextrose (YPD) media (pH 7.23). Table 1 Select parameters from simulated asbestos mine drainage water chemistry from this study compared to previously analyzed environmental samples from VT-ANR water quality reports/surface water analyses [1].

Parameter
Units Values in bold indicate large differences between impaired and control sites in the VT-ANR report.

Microarray data
Raw and normalized microarray data for all Affymetrix GeneChip s Yeast Genome 2.0 Array probesets are available through NCBI GEO Series GSE89875: https://www.ncbi.nlm.nih.gov/geo/query/ acc.cgi?acc ¼GSE89875. Microarray data exploration at the sample-level using principal components analysis (PCA) identifies distinct molecular phenotypes in simulated asbestos mine drainage versus control samples and shows no significant outliers in the study set (Fig. 1). The first principal component (PC1) correlates with our single biological factor, simulated asbestos mine drainage exposure. Gene-level ANOVA with S. cerevisiae-only probesets found 735 genes that are differentially expressed between treatment and control groups (unadjusted p-value o0.05; fold change 4|1.5|) (Supplementary Tables 2-3). Of these 735 genes, 404 genes are down-regulated and 331 genes are upregulated (Fig. 2). Of the 761 genes that passed a more conservative false discovery rate (FDR) filtering threshold of 0.05, 405 exhibited a fold change 4|1.5| ("step-up" p-value, Supplementary Tables 2-3, column K). A hierarchical clustering map includes the highest and lowest expressers (FDR o0.05; fold change 4|1.9|) for treatment and control groups (Fig. 3).

Mine tailings collection and processing
A composite sample of the mine tailings surface was collected on June 19, 2013 from the Lowell Quarry at the VAG mine, Lowell, Vermont, USA (near 44°45 0 51.8″N 72°31 0 22.2″W) shown in Fig. 4. The tailings were collected in sterile Whirl-pak s bags (Nasco, Fort Atkinson, Wisconsin, USA) and stored at 4°C until use. Tailings were processed and the leachate was prepared in the Dolci Laboratory at Johnson State College (Johnson, Vermont, USA) in February 2014. A 250 g sample of tailings was mixed with an equal volume (250 ml) of deionized water followed by exposure to three freeze-thaw cycles ( À80°C) over ten days. The thaw cycle was performed at room temperature with gentle rocking. Water was retrieved from settled tailings followed by filtration through a 0.22 mm polycarbonate filter. Modified YPD growth media was prepared with either simulated asbestos mine drainage or deionized water as follows: 12.5 g YPD powder (Fisher Scientific catalog #BP2469), 10 g glucose, 500 ml water and autoclaved [2]. Note, mineralogical characterization of VAG mine tailings was not performed for this study, however, it has been shown previously that magnesium, silicon, iron, and aluminum are major constituents and nickel and chromium are minor constituents of the tailings [3,4].

Physio-chemistry methods for water and media
Samples of the simulated asbestos mine drainage water (150 ml) as well as prepared simulated asbestos mine drainage media and control media (500 ml each) were sent to Endyne Inc. (Williston, Vermont, USA) for analysis. The methodology for the water chemical analyses reported in Table 1

Strain and growth conditions
A 48-h culture of S. cerevisiae strain NRRL Y-12632 (ATCC s 18824) grown in the control YDP media (described above) was used as inoculum for seven independent sample cultures, four experimental and three control. Growth was monitored by measuring the absorbance, A, at 600 nm (A 600 ) and samples were inoculated into either fresh YPD control media or YPD simulated asbestos mine drainage media (A 600 0.05). The cultures were incubated aerobically (225 rpm) at 30°C for 24 h, at which point cells from each sample were harvested (1.5 ml of culture media) and pelleted by centrifugation at 4000 Â g (6800 rpm) for RNA isolation.

RNA isolation and target preparation
Total RNA isolation for all samples was performed as described in [2]. Briefly, to form spheroplasts, all samples were treated with lyticase (10U/μl). Guanidinium isothiocyanate contained in the RLT buffer from the RNeasy s Plus Mini kit (Qiagen, Valencia, California, USA) was used to rupture cell membranes. Total RNA was collected using silica gel spin columns and DNase I was added to degrade genomic DNA. A Nanodrop 1000 Spectrophotometer (Thermo Scientific, Madison, Wisconsin, USA) was used to measure the quantity and quality of the resulting total RNA. Additionally, RNA quality was assayed using the Agilent BioAnalyzer (Agilent Technologies, Santa Clara, California, USA). No samples were excluded due to low RNA quality.
Total RNA input of 50 ng was used to generate cDNA with the Ovation s Pico WTA System V2 kit (NuGEN Technologies Inc., San Carlos, California, USA). cDNA samples were then purified using an Agencourt s RNAClean XP magnetic bead protocol (Beckman Coulter Inc., Brea, California, USA). Following purification, samples were amplified using SPIA reagents from the Ovation s Pico kit. A final cDNA purification was performed using an Agencourt s RNAClean XP magnetic bead protocol. Sample concentrations were determined using a 33 μg/ml/A260 constant on a Nanodrop 1000 Spectrophotometer. Approximately 4 μg of cDNA was fragmented and labeled with the NuGEN Encore s Biotin Module. Efficiency of the biotin labeling reaction was verified using NeutrAvidin (10 mg/ml) with a gel-shift assay.
A hybridization mixture with controls, including control B2 oligo, 20 Â eukaryotic control mix, herring sperm, acetylated BSA, 2 Â hybridization buffer and water, was added to all seven samples. 90 μl of this mixture was injected into a GeneChip s Yeast Genome 2.0 Array (Affymetrix, Santa Clara, California, USA) and placed in the Affymetrix GeneChip s Hybridization Oven 640 at 45°C and 60 rpm for 18 h overnight. A staining mixture containing streptavidin-phycoerytherin dye (SAPE) and biotinylated anti-phycoerytherin antibody (BAP) was added to amplify signal intensities. Arrays were stained and washed in the Affymetrix GeneChip s Fluidics Station 450 using the Mini_euk2V3_450 fluidics script. All arrays were scanned with the Affymetrix GeneChip s Scanner 3000 and raw analysis performed with Affymetrix Expression Console TM software. The raw data images produced from the scanner were processed into CEL files, which contained measure intensities for each probe on the array.

Gene expression statistical methods
Seven CEL files (four experimental and three control) were imported into the Gene Expression Workflow in Partek s Genomics Suite (PGS) version 6.6 (Partek Inc., St. Louis, MO, USA, www.partek. com). Background correction, quantile normalization, log2 transformation, and probeset summarization were performed using default settings for the Robust Multichip Average (RMA) procedure [5][6][7]. After Schizosaccharomyces pombe probesets, also present on the GeneChip s Yeast Genome 2.0 Array, were removed, a PCA was performed using a covariance dispersion matrix as part of the quality control of the data.
Differential expression (DE) between yeast exposed to simulated asbestos mine drainage (experimental) and unexposed yeast (control) was predicted at the gene-level (probesets summarized into transcript clusters/genes). One-way analysis of variance (ANOVA) was used to compare the individual gene expression data with respect to mine-tailings exposure, i.e. simulated asbestos mine drainage versus control. DE genes were defined based on an absolute fold change equal to or greater than 1.5 and an unadjusted p-value o 0.05. Multiple comparison correction was performed using a FDR of 0.05 [8]. Input for the heatmap was the RMA-normalized expression data standardized to a mean of zero and a standard deviation of one for the subset of DE genes with an absolute fold change equal to or greater than 1.9 and a FDR o 0.05.