**2. Methods**

#### **2.1. Mussel collection and holding**

Pheasantshell mussels (*A. pectorosa*) were collected from the Clinch River, Hancock County, Tennessee, on September 23, 2014. A total of 143 individuals ranging in size from 70 to 90 mm were collected by hand using snorkel and mask. Mussels were held for 14 days of acclimation at the Freshwater Mollusk Conservation Center (FMCC), Virginia Tech, Blacksburg, Virginia, in flow-through systems continuously replenished with unfiltered water from a man-made, on-site pond.

#### **2.2. Chemicals**

Sodium arsenate heptahydrate (Na<sup>2</sup> HAsO<sup>4</sup> \*7H<sup>2</sup> O, American Chemical Society (ACS)-certified reagent grade) was purchased from J.T. Baker Co. (Avantor Performance Materials, Center Valley, PA). A 100-mM (7.492 g/L) As (V) stock solution in ultrapure water was prepared at the beginning of the study. ACS-certified reagent-grade sodium sulfate (Na<sup>2</sup> SO<sup>4</sup> ) was purchased from Fisher Scientific (Fair Lawn, NJ). Stock solutions were prepared weekly in ultrapure water.

#### **2.3. Mussel exposure system**

Mussels were exposed in 18-L downweller-bucket systems [22] modified to accommodate adult mussels [23]. Buckets were placed into a 757-L container filled with well water to serve as a temperature control bath. The target exposure temperature of 23°C was maintained in the water baths using aquarium heaters. Each bath held up to five buckets. In the bucket systems, unfiltered water from the FMCC pond was used for acclimation, control water, and makeup water for each treatment. Mussels were fed daily with a 1:1:1 algal cell ratio from three premixed commercial microalgae diets (Nanno 3600, Shellfish Diet 1800, and TP 1800; Reed Mariculture, Campbell, CA). A concentrated stock solution was prepared in deionized water and 1 mL of stock was added to each bucket to achieve a final concentration of 24,000 cells/mL.

#### **2.4. A 7‐day acute exposure**

Thirty mussels were allocated to 10 buckets (three mussels each, arbitrarily chosen) on October 7, 2014 for a preliminary 7-day test of acute toxicity. This test was conducted to determine the potential for mussel survival in solutions of As(V) and SO<sup>4</sup> 2- over 28 days; treatments were not replicated. Five buckets (*n* = 1) were used to test acute toxicity of five SO<sup>4</sup> 2- concentrations: 1500, 1000, 500, 250, and 0 mg/L (control). Treatments were prepared via direct addition of appropriate volumes of a 607-mM (58.3-g/L) stock solution of SO<sup>4</sup> 2-. Five buckets were used to test acute toxicity of five As(V) concentrations: 2000, 985, 515, 170, and 0 µg/L (control). Treatments were prepared via direct addition of the 100-mM stock solution of As(V). On Day 7, all 30 mussels were alive; there was no mortality in any of the As(V) or SO<sup>4</sup> 2- treatments or controls.

#### **2.5. A 28‐day chronic exposure**

chronic toxicity studies conducted with other bivalves (*S. simile* and *C. fluminea;* [13, 17, 18, 21]). The majority of toxicity tests with As have been conducted with arsenite (As(III)), the more toxic form of As. However, due to the oxygenated environment in our exposure system, mussels were exposed to As as arsenate (As(V)). Hence, the purpose of this study was to determine

these changes, and the relationships between biochemical, histological, and genetic markers.

Pheasantshell mussels (*A. pectorosa*) were collected from the Clinch River, Hancock County, Tennessee, on September 23, 2014. A total of 143 individuals ranging in size from 70 to 90 mm were collected by hand using snorkel and mask. Mussels were held for 14 days of acclimation at the Freshwater Mollusk Conservation Center (FMCC), Virginia Tech, Blacksburg, Virginia, in flow-through systems continuously replenished with unfiltered water from a man-made,

HAsO<sup>4</sup>

beginning of the study. ACS-certified reagent-grade sodium sulfate (Na<sup>2</sup>

\*7H<sup>2</sup>

reagent grade) was purchased from J.T. Baker Co. (Avantor Performance Materials, Center Valley, PA). A 100-mM (7.492 g/L) As (V) stock solution in ultrapure water was prepared at the

from Fisher Scientific (Fair Lawn, NJ). Stock solutions were prepared weekly in ultrapure water.

Mussels were exposed in 18-L downweller-bucket systems [22] modified to accommodate adult mussels [23]. Buckets were placed into a 757-L container filled with well water to serve as a temperature control bath. The target exposure temperature of 23°C was maintained in the water baths using aquarium heaters. Each bath held up to five buckets. In the bucket systems, unfiltered water from the FMCC pond was used for acclimation, control water, and makeup water for each treatment. Mussels were fed daily with a 1:1:1 algal cell ratio from three premixed commercial microalgae diets (Nanno 3600, Shellfish Diet 1800, and TP 1800; Reed Mariculture, Campbell, CA). A concentrated stock solution was prepared in deionized water and 1 mL of stock was added to each bucket to achieve a final concentration of 24,000 cells/mL.

Thirty mussels were allocated to 10 buckets (three mussels each, arbitrarily chosen) on October 7, 2014 for a preliminary 7-day test of acute toxicity. This test was conducted to determine the

2- caused sublethal changes in freshwater mussels, the nature of

O, American Chemical Society (ACS)-certified

SO<sup>4</sup>

2- over 28 days; treatments were not

2- concentrations: 1500,

) was purchased

whether As (as As(V)) and SO<sup>4</sup>

102 Organismal and Molecular Malacology

**2.1. Mussel collection and holding**

Sodium arsenate heptahydrate (Na<sup>2</sup>

**2.3. Mussel exposure system**

**2.4. A 7‐day acute exposure**

potential for mussel survival in solutions of As(V) and SO<sup>4</sup>

replicated. Five buckets (*n* = 1) were used to test acute toxicity of five SO<sup>4</sup>

**2. Methods**

on-site pond.

**2.2. Chemicals**

Sixty-four mussels were randomly selected on October 16, 2014. Four mussels were allocated to each of 16 buckets, and the length of each mussel was recorded using dial calipers. Mussels were allowed to acclimate in the buckets for 11 days prior to the start of the chronic exposure. During the acclimation period, dissolved oxygen and ammonia levels were monitored using the methods described below to ensure that the feeding rate was appropriate to maintain acceptable water quality.

On Day 0 of the exposure (October 27, 2014), each bucket received a 100% water exchange and was randomly assigned to one of four treatments/controls (*n* = 4 for each treatment/control). Because of the availability of experimental units and mussels, two separate controls were used (control\_1 and control\_2); both consisted of 100% unfiltered pond water. The As(V) treatment (hereafter, HA) concentration of 1000 µg/L was achieved by adding appropriate volumes of 100 mM stock solution to unfiltered pond water. The SO<sup>4</sup> 2- treatment (hereafter HS) of 1250 mg/L was achieved by adding appropriate concentrations of a 569-mM (54.65-g/L) stock solution to unfiltered pond water. The final water volume in all treatment and control buckets was 17.5 L.

For each bucket, mussel mortality was checked daily and recorded, and a 100% water exchange was conducted weekly. Prior to water exchanges, temperature (°C), specific conductance (µS/cm), dissolved oxygen (DO; mg/L and % saturation), and pH were measured using a calibrated YSI 556 Multi-Probe Sensor (YSI Inc., Yellow Springs, OH), and water samples were collected. Total ammonia-nitrogen (TAN) was measured in unfiltered samples using a Hach DR/2400 meter following the manufacturer's methods. Concentrations of elemental As and S were measured in filtered (0.45-µm) water samples. Elemental quantification was performed by the Virginia Tech Soil Testing Laboratory using an inductively coupled plasma atomic emission spectrometer (ICP-AES, Spectro Analytical Instrumentation, Kleve, Germany) following the laboratory's standard operating procedures and quality assurance/quality control methods [24]. Sulfate concentrations were calculated from measured total S; all S were assumed to be present as SO<sup>4</sup> 2- due to the buckets' oxygenated environment. Although only total As was measured, all As are assumed to have been present as As (V) due to the oxygenated environment in the buckets. Alkalinity and hardness of the pond water were measured weekly as part of a concurrent study [24]. The 28-day exposure concluded on November 24, 2014.

#### **2.6. Mussel dissection**

At the end of the study on Day 28, all mussels were removed from the buckets. Each mussel was measured using dial calipers and opened using reverse pliers. Each mussel was evaluated for gravidity, that is, the presence of glochidia in the outer marsupial gills. Mussels were removed from their shells by severing the adductor muscles. The mantle tissue was removed and samples of gill, kidney, and gonad were excised and fixed in neutral buffered formalin for histological evaluation. The digestive gland was removed from the remainder of the tissue and divided into three portions, with one portion each preserved separately for histological evaluation and RNA extraction. The remaining portion was further divided into thirds for separate biochemical analyses. Samples for biochemical work were immediately frozen. Samples for screening of gene expression were preserved in RNALater (Qiagen). All dissection tools were sterilized in a concentrated bleach solution between dissections of each mussel.

#### **2.7. Transcriptome analysis**

Only parasite-free, female mussels (determined via histological analysis) were used for transcriptomic analysis. RNA isolated from digestive gland tissue samples was transcribed in vitro to double-stranded cDNA, yielding a library of transcripts for each individual mussel. Individual-specific adapters were ligated onto transcripts for each individual so that we could subsequently identify RNA sourced from the respective individuals. Barcoded cDNA samples three from control and six from contaminant-exposed individuals (three per treatment)—were sequenced using the Illumina HiSeq platform. Adapters were removed from the raw sequencing reads. Duplicated and low-quality reads were discarded using FastqMcf [25] with default parameters. To exclude possible contamination, all reads were aligned to a bacterial database downloaded from NCBI (http://www.ncbi.nlm.nih.gov/), and only unmapped reads were used for assembly. Processed reads from all nine samples were merged together, assembled with Trinity [26] with parameter—trimmomatic, after duplications were removed. TransDecoder (Trinity package) was used to identify candidate-coding regions within assembled transcripts, and transcripts with open-reading frame (ORF) lengths less than 300 (100 amino acids) were removed from the assembly. The final transcript assembly was used as a reference for gene annotation and expression calculation. Transcripts were mapped to the nonredundant protein database [(NR database) from NCBI] using BLAST (v. 2.2.28). Alignments with threshold *e*-values greater than 1*e*-10 or identity less than 50% were discarded. The clean reads were mapped to the reference assembly using Bowtie v. 1.0.0 [27] with parameters set to '-l 25 -I 1 -X 1000 -a -m 200.' RSEM [28] was used to calculate the gene expression. Differentially expressed genes were calculated using the edgeR [29] package in R software (http://www.r-project.org/), and Benjamini-Hochberg adjusted *p*-values less than 0.05 were considered to be significant.

#### **2.8. Histological analysis**

Sections from gonads, digestive glands, and gills were stained with hematoxylin and eosin for routine histological evaluations, and sections from kidneys were stained with Long Ziehl-Neelsen stain for elaboration of lipofuscin [30]. Histological evaluations of tissues were conducted by light microscopy using point counting [31]. Evaluations determined fractions of reproductive acini containing mature and/or developing gametes, acini containing atretic and/ or resorbing gametes, digestive gland diverticula cells with abnormal cytoplasm, gill filament termini without cilia, and kidney diverticula cells containing lipofuscin inclusions. Genders (female, male, hermaphrodite, and indeterminate) of mussels and incidences of parasitic infestation were recorded. Histological data were analyzed using generalized linear-mixed models for binomial data with SAS GLIMMIX.

#### **2.9. Biochemical analysis**

of gill, kidney, and gonad were excised and fixed in neutral buffered formalin for histological evaluation. The digestive gland was removed from the remainder of the tissue and divided into three portions, with one portion each preserved separately for histological evaluation and RNA extraction. The remaining portion was further divided into thirds for separate biochemical analyses. Samples for biochemical work were immediately frozen. Samples for screening of gene expression were preserved in RNALater (Qiagen). All dissection tools were sterilized in a

Only parasite-free, female mussels (determined via histological analysis) were used for transcriptomic analysis. RNA isolated from digestive gland tissue samples was transcribed in vitro to double-stranded cDNA, yielding a library of transcripts for each individual mussel. Individual-specific adapters were ligated onto transcripts for each individual so that we could subsequently identify RNA sourced from the respective individuals. Barcoded cDNA samples three from control and six from contaminant-exposed individuals (three per treatment)—were sequenced using the Illumina HiSeq platform. Adapters were removed from the raw sequencing reads. Duplicated and low-quality reads were discarded using FastqMcf [25] with default parameters. To exclude possible contamination, all reads were aligned to a bacterial database downloaded from NCBI (http://www.ncbi.nlm.nih.gov/), and only unmapped reads were used for assembly. Processed reads from all nine samples were merged together, assembled with Trinity [26] with parameter—trimmomatic, after duplications were removed. TransDecoder (Trinity package) was used to identify candidate-coding regions within assembled transcripts, and transcripts with open-reading frame (ORF) lengths less than 300 (100 amino acids) were removed from the assembly. The final transcript assembly was used as a reference for gene annotation and expression calculation. Transcripts were mapped to the nonredundant protein database [(NR database) from NCBI] using BLAST (v. 2.2.28). Alignments with threshold *e*-values greater than 1*e*-10 or identity less than 50% were discarded. The clean reads were mapped to the reference assembly using Bowtie v. 1.0.0 [27] with parameters set to '-l 25 -I 1 -X 1000 -a -m 200.' RSEM [28] was used to calculate the gene expression. Differentially expressed genes were calculated using the edgeR [29] package in R software (http://www.r-project.org/), and

Benjamini-Hochberg adjusted *p*-values less than 0.05 were considered to be significant.

Sections from gonads, digestive glands, and gills were stained with hematoxylin and eosin for routine histological evaluations, and sections from kidneys were stained with Long Ziehl-Neelsen stain for elaboration of lipofuscin [30]. Histological evaluations of tissues were conducted by light microscopy using point counting [31]. Evaluations determined fractions of reproductive acini containing mature and/or developing gametes, acini containing atretic and/ or resorbing gametes, digestive gland diverticula cells with abnormal cytoplasm, gill filament termini without cilia, and kidney diverticula cells containing lipofuscin inclusions. Genders (female, male, hermaphrodite, and indeterminate) of mussels and incidences of parasitic infestation were recorded. Histological data were analyzed using generalized linear-mixed

concentrated bleach solution between dissections of each mussel.

**2.7. Transcriptome analysis**

104 Organismal and Molecular Malacology

**2.8. Histological analysis**

models for binomial data with SAS GLIMMIX.

Digestive gland samples for biochemical analysis were placed in 1.5-mL tubes, immediately frozen, and held at -80°C. Analyses of biochemical parameters in mussel digestive glands were conducted in duplicate and expressed per mg of protein determined using a bicinchoninic acid method (Pierce BCA Protein Assay Kit, Thermo Scientific, Rockford, IL). Activities of catalase (CAT), glutathione S-transferase (GST), glutathione peroxidase (GPx), glutathione reductase (GR), malondialdehyde (MDA), total- and Na+, K+−ATPase, and lactate dehydrogenase (LDH), and the concentration of reduced glutathione (GSH) were determined.

#### **2.10. Correlations with histological and biochemical markers**

Correlations between biochemical measurements, histological variables, and gene expression were evaluated using rank-transformed data (Spearman's rho) to minimize effects of the different data types and scales for each class of variable. Due to the large number of variables examined, the criterion for statistical significance was reduced (α = 0.01) and plots of all significant relationships were examined to ensure linearity. To select genes for assessment of correlations, genes with the lowest false-detection rate (<9.67 × 10-5) and *p*-values (<9.5 × 10-8) were identified among genes with differential expression between control, HS, and HA. All annotated genes with differential expression between control, HS, and HA females were also selected. For these 48 selected genes, gene expression was correlated with biochemical measurements and histological variables. Digestive glands from nine mussels were included in the analysis of gene expression, but only eight mussels were included in correlations with biochemical measurements of the glutathione antioxidant system; that is, one extract lost during the set of biochemical analyses was one of the mussels randomly selected for analyses of gene expression.
