**3. Results**

#### **3.1. Water chemistry**

Mean-measured concentrations of SO<sup>4</sup> 2- and As were similar to nominal concentrations (**Table 1**). Background SO<sup>4</sup> 2- (∼17 mg/L) was present in the pond water; thus, measured concentrations were slightly above nominal concentrations. There was no measureable background As in the pond water.

Water quality measurements were within acceptable ranges for toxicity tests with freshwater mussels. During the study, the pond had a mean measured hardness of 204 mg/L CaCO<sup>3</sup> and a mean measured alkalinity of 196 mg/L CaCO<sup>3</sup> . Mean measured temperatures were similar between treatments, and temperature was generally within 1°C of the 23°C target temperature (**Table 2**). Measured pH was similar across all treatments and varied little over the course of the study. Specific conductivity was similar in the controls and As treatments, and was elevated in the SO<sup>4</sup> 2- treatments as expected. Dissolved oxygen concentrations remained relatively high for the duration of the study (lowest recorded value: 75% saturation, 6.20 mg/L, on Day 28). Dissolved oxygen concentrations were consistent between treatments but did show variation during the study, with a decrease in overall mean saturation from 86% on


a Measurements were pooled for control\_1 and control\_2, since all replicates consisted of 100% unfiltered pond water. b Background concentration of SO<sup>4</sup> 2⁻ in the pond.

c Method detection limit for As.

**Table 1.** Nominal and mean (standard deviation) measured concentrations of SO<sup>4</sup> 2- and As in treatments (*n* = 4 each) and controls (*n* = 8) during the 28-day exposure.


**Table 2.** Mean (standard deviation) measured concentrations of water quality parameters in all treatments and controls (*n* = 4 each) during the 28-day exposure.

Day 7 (range 79.1–88.7%) to 78% on Day 28 (range 75–78.8%). TAN concentrations were also consistent between treatments, but varied over the course of the study. The overall mean TAN concentration was 0.001 mg/L on Day 7 (range 0–0.01 mg/L) and increased to 0.068 mg/L on Day 21 (range 0.05–0.08 mg/L). The maximum measured TAN concentration was 0.080 mg/L in several control and SO<sup>4</sup> 2- units, but this value is far below the 2013 acute and chronic water quality standards for ammonia (0.96 and 0.24 mg/L, respectively, at pH 8.6 and 23°C), derived based on the sensitivity of mussels [32].

#### **3.2. Mussels**

No mussels died during the course of the experiment. The mean length of all exposed mussels was 85.1 mm (range 79.0–99.3 mm). The mean lengths were not significantly different between treatments (GLIMMIX, *p* = 0.22). Mussel growth was not quantified in this study due to the use of adult mussels and the short duration of the exposure.

#### **3.3. Reference assembly**

The analytic program Trinity handled 61,774 transcripts, assembling them into the putative transcripts of 34,019 genes. The average length of these contiguous sequences, or "contigs," was 1012 base pairs.

#### **3.4. Differential expression analysis**

Day 7 (range 79.1–88.7%) to 78% on Day 28 (range 75–78.8%). TAN concentrations were also consistent between treatments, but varied over the course of the study. The overall mean TAN concentration was 0.001 mg/L on Day 7 (range 0–0.01 mg/L) and increased to 0.068 mg/L on Day 21 (range 0.05–0.08 mg/L). The maximum measured TAN concentration was 0.080 mg/L

**Table 2.** Mean (standard deviation) measured concentrations of water quality parameters in all treatments and controls

 **(mg/L) As (µg/L)**

Measurements were pooled for control\_1 and control\_2, since all replicates consisted of 100% unfiltered pond water.

**Treatment Temp. (°C) pH Specific conductivity (mS/**

Arsenic 22.8 (0.9) 8.58 (0.03) 0.436 (0.004) Sulfate 22.6 (0.8) 8.64 (0.05) 2.89 (0.10) Control\_1 22.6 (1.1) 8.51 (0.27) 0.431 (0.005) Control\_2 22.8 (1.2) 8.61 (0.07) 0.434 (0.004)

**Treatment Nominal Mean Nominal Mean** Arsenic - 16.93 (1.44)<sup>b</sup> 1000 1027 (111) Sulfate 1250 1276 (48) 0 <8.1<sup>c</sup> Controla - 16.90 (1.56)<sup>b</sup> 0 <8.1<sup>c</sup>

2⁻ in the pond.

**Table 1.** Nominal and mean (standard deviation) measured concentrations of SO<sup>4</sup>

quality standards for ammonia (0.96 and 0.24 mg/L, respectively, at pH 8.6 and 23°C), derived

No mussels died during the course of the experiment. The mean length of all exposed mussels was 85.1 mm (range 79.0–99.3 mm). The mean lengths were not significantly different between treatments (GLIMMIX, *p* = 0.22). Mussel growth was not quantified in this study due

The analytic program Trinity handled 61,774 transcripts, assembling them into the putative transcripts of 34,019 genes. The average length of these contiguous sequences, or "contigs,"

to the use of adult mussels and the short duration of the exposure.

2- units, but this value is far below the 2013 acute and chronic water

2- and As in treatments (*n* = 4 each) and

**cm)**

in several control and SO<sup>4</sup>

(*n* = 4 each) during the 28-day exposure.

**3.3. Reference assembly**

was 1012 base pairs.

**3.2. Mussels**

a

b

c

based on the sensitivity of mussels [32].

**SO4 2**−

Background concentration of SO<sup>4</sup>

controls (*n* = 8) during the 28-day exposure.

Method detection limit for As.

106 Organismal and Molecular Malacology

Statistically significant (*p* < 0.05) differential expression (DE) of particular genes was observed among control mussels (CT) and those exposed to either arsenate (HA) or sulfate (HS). Chemical stress was as likely to cause underexpression as it was to cause overexpression of genes relative to levels observed in control mussels. Contrasts also were observed among mussels exposed to the respective pollutants, indicating that pollutants induced up- or downexpression of different suites of genes. From 50 to 100 differentially expressed genes were found for each comparison. **Figures 1**–**3** compare levels of expression of particular genes among groups of mussels.

Comparing gene expression among control and arsenate-treated mussels (**Figure 1**), 18,572 transcripts were not affected by treatment, 28 were underexpressed in controls relative to treated mussels, and 30 were overexpressed.

Comparing gene expression among control and sulfate-treated mussels (**Figure 2**), 18,469 transcripts were not affected by treatment, 74 were underexpressed in controls relative to treated mussels, and 86 were overexpressed.

Comparing gene expression among pollutants, 18,444 transcripts were not differentially expressed, 128 were more highly expressed in sulfate-exposed mussels, and 57 more highly in arsenate-exposed mussels (**Figure 3**). Overall, the sulfate treatment was a ∼3× greater stress factor in terms of the number of differentially expressed genes than was the arsenate (HA) treatment. We offer the explanation that in aquatic species, maintaining homeostasis in the face of ionic and osmotic stressors affects many cellular processes, while heavy metal toxicity

**Figure 1.** Summary of differential gene expression (DE) in control (CT) compared to arsenate-treated (HA) mussels. LogCPM is the logarithm of counts per million reads, and logFC is the log-transformed fold change in gene expression. Each dot in the figure represents the expression of one gene. Dots for genes whose expression is not significantly affected fall between the two horizontal lines. Dots for genes whose expression is upregulated relative to controls fall above the upper line. Dots for genes whose expression is downregulated relative to controls fall below the lower line.

**Figure 2.** Summary of differential gene expression (DE) in control (CT) compared to sulfate-treated (HS) mussels. LogCPM is the logarithm of counts per million reads, and logFC is the log-transformed fold change in gene expression. Each dot in the figure represents the expression of one gene. Dots for genes whose expression is not significantly affected fall between the two horizontal lines. Dots for genes whose expression is upregulated relative to controls fall above the upper line. Dots for genes whose expression is downregulated relative to controls fall below the lower line.

**Figure 3.** Summary of differential gene expression (DE) in high arsenate-(HA) compared to high sulfate-(HS) treated mussels. LogCPM is the logarithm of counts per million reads, and logFC is the log-transformed fold change in gene expression. Each dot in the figure represents the expression of one gene. Dots for genes whose expression is not significantly affected fall between the two horizontal lines. Dots for genes whose expression is upregulated relative to controls fall above the upper line. Dots for genes whose expression is downregulated relative to controls fall below the lower line.

affects relatively few. Comparing the effects of HS and HA directly, it was evident that HS causes >2 times more overexpression, again potentially pointing to a greater effect of HS. Interestingly, and as would be expected, the expression of different genes was affected. As discussed below, these genes are candidate markers for indicating exposure of mussels to the respective pollutants.

#### **3.5. Annotation of expressed genes**

**Figure 3.** Summary of differential gene expression (DE) in high arsenate-(HA) compared to high sulfate-(HS) treated mussels. LogCPM is the logarithm of counts per million reads, and logFC is the log-transformed fold change in gene expression. Each dot in the figure represents the expression of one gene. Dots for genes whose expression is not significantly affected fall between the two horizontal lines. Dots for genes whose expression is upregulated relative to controls fall above the upper line. Dots for genes whose expression is downregulated relative to controls fall below the

**Figure 2.** Summary of differential gene expression (DE) in control (CT) compared to sulfate-treated (HS) mussels. LogCPM is the logarithm of counts per million reads, and logFC is the log-transformed fold change in gene expression. Each dot in the figure represents the expression of one gene. Dots for genes whose expression is not significantly affected fall between the two horizontal lines. Dots for genes whose expression is upregulated relative to controls fall above the upper line. Dots for genes whose expression is downregulated relative to controls fall below

lower line.

the lower line.

108 Organismal and Molecular Malacology

The proportion of genes that we were able to annotate by reference to GenBank was relatively low (63%), as mollusks are nonmodel organisms and hugely underrepresented in genomic databases. We note that the proportions of genes that we annotated are very similar to those reported by Bai et al. [33, 34] for the freshwater pearl mussel *Hyriopsis cumingii.*

Among annotated underexpressed genes in the arsenate treatment relative to controls were guanylate binding protein 1 (*GBP1*) and poly [ADP-ribose] polymerase (BRAFLDRAFT\_74879). In mammalian cells, *GBP1* has been shown to respond to exposure to multiple stress agents including paclitaxel and doxorubicin [35]. The enzymes poly(ADP-ribose) (*PAR*) and polymerase-1 (*PARP‐1*) are central for cellular stress responses and for directing cells to specific fates (e.g., DNA repair vs. cell death) [36]. Among annotated overexpressed genes in arsenate relative to control were poly [ADP-ribose] polymerase 15-like (LOC101731886), and sodiumand chloride-dependent glycine transporter 2-like (LOC100899820).

Among annotated underexpressed genes in the sulfate treatment relative to controls were serine protease inhibitor dipetalogastin precursor and zinc-binding Sp-Hypp\_8991. Interestingly, the expression of serine protease inhibitor seems to offer stress tolerance via delayed senescence [37]. Among annotated overexpressed genes in sulfate relative to control were NADPH-dependent alpha-keto amide reductase (YDL124W), and poly [ADP-ribose] polymerase 15-like (LOC101731886). Multiple aldo-keto reductases fulfill functionally redundant roles in stress response in yeast [38].

#### **3.6. Correlations with histological, biochemical, and gene expression markers**

Significant correlations (*r*, *p* < 0.05) were observed among genetics and histological data from the nine sampled females. Fractions of reproductive acini containing mature or developing oocytes were significantly negatively correlated with gene identification code (gene code) c140332\_g1 (gene symbol = LOC101731886) (*r* = -0.78, *p* = 0.01). Fractions of digestive diverticula cells with lesions were significantly correlated with gene code c145102\_g1 (none) (*r* = 0.74, *p* = 0.04) and negatively correlated with gene codes c103784\_g1 (CGI\_10002926) (*r* = -0.76, *p* = 0.03), c149150\_g3 (none) (*r* = -0.78, *p* = 0.02), and c152424\_g6 (none) (*r* = -0.73, *p* = 0.04). Fractions of kidney cells containing lipofuscin inclusions were significantly correlated with c138105\_g1 (none) (*r* = 0.76, *p* = 0.02).

There were strong correlations (*r* > 0.85, *p* < 0.01) between the expression of five genes and biochemical measurements. Numerous weaker correlations (0.05 > *p* > 0.01) are not presented or discussed. The expression of one annotated gene, c150857\_g2 (*GBP1*), was positively correlated with the activity of glutathione-S-transferase, but there was little separation between treatments and control\_2 for both gene expression and enzyme activity. The expression of one annotated gene (c152797\_g1; BRAFLDRAFT\_74879) and one unannotated gene (c126914\_g1) was positively correlated with the concentration of reduced glutathione, and there was good separation between treatments and control\_2. The expression of two unannotated genes (c155139\_g1 and c155139\_g2) was negatively correlated with the concentration of reduced glutathione and there was good separation between the treatments and control\_2. Correlation between the expression of a sixth gene (unannotated; c154720\_g4) did not meet the α = 0.01 criterion for strong correlation, but the plot demonstrated a similar pattern as the other two genes negatively correlated with reduced glutathione concentration, with good separation between treatments and control\_2. For the eight mussels included in these correlations, concentrations of reduced glutathione were higher in the control and lower in the two treatments, whether these genes are directly related to reduced glutathione production or utilization is unknown. The small sample of mussels utilized for genetic analysis limits the interpretation of the relationships but suggests that these genes warrant further investigation, particularly as related to As(V) exposure.
