Differential proteomic response of Sydney rock oysters (Saccostrea glomerata) to prolonged environmental stress

Aquatic Toxicology 173 (2016) 53–62

Contents lists available at ScienceDirect

Aquatic Toxicology journal homepage: www.elsevier.com/locate/aquatox

Differential proteomic response of Sydney rock oysters (Saccostrea glomerata) to prolonged environmental stress A.R. Melwani a,b,∗ , E.L. Thompson a,b , D.A. Raftos a,b a b

Department of Biological Sciences, Macquarie University, NSW 2109, Australia Sydney Institute of Marine Science, NSW 2088, Australia

a r t i c l e

i n f o

Article history: Received 2 November 2015 Received in revised form 4 January 2016 Accepted 8 January 2016 Available online 23 January 2016 Keywords: Ecotoxicology Molluscs Proteomics Stress Contamination

a b s t r a c t Exposure to prolonged environmental stress can have impacts on the cellular homeostasis of aquatic organisms. The current study employed two-dimensional electrophoresis (2-DE) to test whether exposure to impaired water quality conditions in the Sydney Harbour estuary has significantly altered the proteomes of the resident Sydney rock oyster (Saccostrea glomerata). Adult S. glomerata were sampled from four bays in the estuary. Each bay consisted of a “high-impact” site adjacent to point sources of chemical contamination (e.g., storm drains/canals or legacy hotspots) and a “low-impact” site located ∼5 km away from point sources. A mixture of environmental stressors differed significantly between high- and low-impact sites. Specifically, PAHs, PCBs, tributyltin, Pb, and Zn were significantly elevated in oyster tissues from high-impact sites, together with depleted dissolved oxygen and low pH in the water column. A 2-DE proteomics analysis subsequently identified 238 protein spots across 24 2-DE gels, of which 27–50 spots differed significantly in relative intensity between high- and low-impact sites per bay. Twenty-five percent of the differential spots were identified in more than one bay. The identities of 80 protein spots were determined by mass spectrometry. HSP 70, PPIB, and radixin were the three most highly expressed differential proteins. Despite the largely unique proteomes evident in each bay, functional annotations revealed that half of the differentially expressed proteins fell into just two subcellular functional categories—energy metabolism and the cytoskeleton. These findings provide a framework to further investigate adaptation of cellular mechanisms to prolonged stress in S. glomerata. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Anthropogenic stressors have significantly altered the condition of estuarine environments worldwide (Birch and Taylor, 2000; Brown et al., 2003; Hedge et al., 2009). Pollution, habitat alteration, freshwater inputs, and invasive species are among the many stressors that persist in heavily modified estuaries. Exposure to stressful conditions for prolonged periods (months/years) can impact on a variety of genetic and physiological traits of organisms. Such changes may explain particular tolerance or sensitivities to environmental change (Brown et al., 2003; Butt and Raftos, 2007; Thompson et al., 2007, 2011). Understanding the adaptive response of organisms to prolonged stress is essential for biomonitoring and maintaining the biodiversity of native species. Proteins play an integral role in how organisms can respond and adapt to environmental change. Advancements in the field of pro-

∗ Corresponding author. E-mail address: [email protected] (A.R. Melwani). http://dx.doi.org/10.1016/j.aquatox.2016.01.003 0166-445X/© 2016 Elsevier B.V. All rights reserved.

teomics to discover subtle changes in protein concentrations have provided novel methods to address the functional mechanisms of biological effects. Recent studies have identified a suite of proteins responding to rapid environmental changes in invertebrates. These proteomic profiles reflect substantial changes in energy metabolism, cytoskeleton function and cellular stress responses (Apraiz et al., 2006; Muralidharan et al., 2012; Neave et al., 2012) that are consistent across a range of stressors (e.g., metals, temperature, CO2; Thompson et al., 2015, 2012; Tomanek and Zuzow, 2010). However, few studies have examined the proteomic response of organisms after long-term exposure to stress in the field, or within the context of natural environmental variability. Evaluating such responses is a necessary next step in furthering our understanding of the subcellular responses to stress and its effects on aquatic health. The Sydney estuary in New South Wales is by far the most heavily impacted estuary in Australia (Birch, 2007). Sydney was the first major maritime port (ca. 1811) and historically supported the needs of major industries in southeastern Australia. For over a century, the estuary catchment (480 km2 ) has been subjected

54

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

Fig. 1. Map of study locations in Sydney Harbour, Australia. Locations marked in black (“-1”) comprise the high-impact sites, exhibiting elevated contaminants in oyster tissue and altered hydrologic factors, relative to the low-impact sites in white (“-2”).

to intense urban and industrial development (∼86%; Birch and Rochford, 2010) along the shoreline of Sydney Harbour and adjoining bays. As a result, Sydney is the most populous city in Australia, with a large proportion of its inhabitants residing within a few kilometers of the estuary and its waterways. Not surprisingly, stormwater and inputs of fluvial sediments are the most attributable current sources of altered water quality conditions in the Sydney estuary. An overabundance of nutrients, low dissolved oxygen, and significantly elevated contaminants (e.g., heavy metals, chlorinated organic compounds, and hydrocarbons) are common characteristics of the estuary’s embayments (Birch and Taylor, 1999, 2000; Hatje et al., 2003). Consequently, waterways surrounding industrial and urban developments in Sydney are known to contain some of the highest concentrations of toxic contaminants in sediment reported in the world (Birch and Taylor, 1999; Birch et al., 2007). However, these impacts are not ubiquitous. Gradients of decreasing sediment contamination toward the axis of the estuary are documented features of bays in the system (Birch, 2011). As its name suggests, the Sydney rock oyster (Saccostrea glomerata) was first described in Sydney Harbour. It is an iconic bivalve that is well suited for testing the effects of environmental stress. The species is endemic to Australia and has been a commonly used to assess water quality conditions and test responses to various stressors (Gall et al., 2012; Hardiman and Pearson, 1995; Scanes, 1996; Vorkamp et al., 2010). In the Sydney estuary, S. glomerata are abundant along the shoreline of many sheltered embayments and are often found fringing storm water pipes and canals. Oysters tend to closely reflect local water quality conditions and rely on suitable conditions for growth and survival. Changes to hydrologic regimes (e.g., rainfall, water temperature, salinity, and sediment loads) have been associated with impacts on oyster growth, development and survival (Bergquist et al., 2006). The current study tests whether prolonged exposure to stress in the Sydney Harbour estuary has significantly altered the proteomes of resident S. glomerata. It fills a crucial knowledge gap by

identifying the proteins and subcellular systems associated with responses to stress in native aquatic organisms from a heavily impacted urban/industrial environment.

2. Methods 2.1. Oysters Adult Sydney rock oysters (Saccostrea glomerata; shell length 40–85 mm) were collected from four bays in the Sydney Harbour estuary (Rozelle-Blackwattle Bay, Iron Cove, Five Dock Bay, Gore Cove). Oysters were sampled from two treatment sites in each bay. They comprised one site adjacent to stormwater canals and/or legacy sources of contamination, and a second site near the mouth of each bay ∼5 km from point sources (Fig. 1). Our initial identification of treatment sites was based on an extensive dataset of sediment contamination previously collected over several years in the estuary (Birch and Taylor, 1999, 2000; Birch et al., 2008). This evaluation clearly indicated that shoreline areas were heavily degraded relative to the axis of the estuary. However, due to the lack of conventional reference sites in the estuary, treatment sites were nested within each bay. This design helped to control for potential differences in hydrology and genetic variability between oysters from the two treatment sites. Three replicate areas were targeted at each site in each bay. Ten individual oysters were collected at each replicate area for proteomics and 5–10 oysters were composited for tissue chemistry. Oysters were measured for shell length and then shucked. Gills (for proteomics) or whole-body tissues (for chemistry) were removed and immediately snap-frozen in a portable liquid nitrogen freezer. Basic hydrologic parameters (pH, temperature, salinity, and dissolved oxygen) were measured in surface waters using a YSI Professional Series (Ohio, USA) handheld data logger. Three measurements were taken per replicate area at the same point in the tidal cycle, immediately prior to oyster collection.

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

2.2. Chemical analyses Composited oysters were kept frozen in their original bags until analyzed. All tissue chemistry analyses were performed at the National Measurement Institute (Sydney, Australia). Four metals (cadmium, mercury, nickel, zinc) were analyzed using inductively coupled plasma mass spectrometry. Polychlorinated biphenyls (PCBs), polyaromatic hydrocarbons (PAHs), and tributyltin were assessed by gas chromatography and mass spectrometry. Clean techniques were followed during preparation of samples, blanks, and standards, using ASTM Type II water and analytical grade chemicals. Blanks, duplicates, and a pair of spiked samples were analyzed with each set of samples. The matrix spike recoveries were all within the acceptable range of recovery (generally 70–120%). Relative Percent Differences (RPDs) for spiked samples and lab duplicates were within the acceptable range (<40%), and all method blanks were below the detection limit. Metal concentrations are reported in ␮g/g or parts-per-million, wet weight. PCBs, PAHs, and tributyltin are reported in ng/g or parts-per-billion, wet weight.

55

gel electrophoresis (SDS-PAGE) with 12% Mini-PROTEAN TGX precast gels (Bio-Rad). Gels were stained with Lava Purple (Gel Company) and then imaged using a Bio-Rad Pharos scanner (GE Healthcare). 2.6. Protein spot quantification Protein spots (n = 238) were manually selected and matched across the 24 2-DE gels using PDQuest proteomic analysis software (Bio-Rad). Normalized spot abundances were generated automatically using the software, based on the spot intensity, total spot volume, and background of the gel. Mean normalized spot volumes (n = 3 replicate areas per site) were then used to calculate fold differences between each pair of sites (high-and low-impact) within a bay. Fold difference was calculated as the ratio of mean normalized spot volume between treatments. Positive values represented spots with a higher relative abundance in the high-impact treatment and negative values represented spots with lower relative abundance in the high-impact treatment.

2.3. Protein extraction

2.7. Trypsin in-gel digestion

Protein extraction procedures were modified from the methods of Thompson et al. (2011, 2012). Proteins were extracted by homogenizing gill tissues in 1 ml of Tri-reagent (Sigma–Aldrich). RNA and DNA were removed using bromochloropropane and 100% ethanol, respectively. Proteins were then precipitated by adding 3× sample volume of ice-cold acetone followed by centrifugation for 10 min at 12,000 × g (4 ◦ C). The protein pellets were washed three times in 2 ml of 0.3 M guanidine hydrochloride in 95% ethanol and 2.5% glycerol, and once in 2 ml of 95% ethanol and 2.5% glycerol. Each wash was followed by centrifugation at 8000 × g for 5 min (4 ◦ C) and removal of the supernatant. The resulting protein pellets were airdried before re-suspension in 50 ␮l rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS; 50 mM dithiothreitol).

Eighty differential protein spots were manually excised from a fresh set of coomassie-stained 2-DE gels and digested using trypsin. Each gel plug was washed for 10 min with 100 mM NH4 HCO3 , de-stained for 10 min in 50% acetonitrile (ACN)/50 mM NH4 HCO3, dehydrated for 5 min in 100% ACN, and subsequently air dried. Gel plugs were then reduced with 100 mM DTT in 100 mM NH4 HCO3 at 56 ◦ C for 1 h and alkylated with 55 mM iodoacetamide in 100 mM NH4 HCO3 for 45 min in the dark at room temperature, before being washed and dehydrated again. Peptide digestion was initiated by adding 30 ␮l of trypsin solution to gel plugs (12.5 ng/␮l in 50 mM NH4HCO3, Promega) followed by incubation for 30 min at 4 ◦ C. Samples were then stored overnight at 37 ◦ C. The following day, gel plugs were washed twice in 50% ACN/2% formic acid for 30 min. The resulting supernatants containing peptides were concentrated to 10 ␮l in a vacuum centrifuge. Peptides were then purified using C18 zip-tips by washing with Buffer A (2% ACN, 0.1% formic acid) and elution with Buffer B (99.9% ACN, 0.1% formic acid). Peptides were finally dried in a vacuum centrifuge, reconstituted in 20 ␮l of 2% formic acid, and stored at −80 ◦ C.

2.4. Protein quantification Proteins were quantified using the Bio-Rad Protein Assay based on the Bradford method. Two microlitres of the sample were added in triplicate to a 96-well micro-titration plate. Forty microlitres of Bradford reagent and 160 ␮l of deionised water were then added to each well, and the plate was left to incubate at room temperature for 5 min. Subsequently, absorbance was measured at 595 nm on an X-Mark spectrophotometer (Bio-Rad). Final protein concentrations were interpolated from a standard curve generated with bovine serum albumin. Based on the results of protein quantification, proteins from five oysters per replicate area per site were pooled to obtain 150 ␮g of total protein per replicate for subsequent 2-DE gel analysis. Overall, three biological replicates were employed per treatment (site) in each bay. This resulted in 24 gels in total across the four sampling areas. 2.5. 2-DE analysis Immobilized pH linear gradient gel strips (7 cm, pH 4–7; GE Healthcare) were rehydrated overnight in 125 ␮l of rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS, 50 mM DTT, 1% ampholytes) containing 150 ␮g of protein. First-dimension separation was then undertaken by isoelectrofocusing (IEF) on an Ettan IPGphor system (Amersham Biosciences). IEF of gel strips was performed in a four step procedure: 1) 100 V for 2 h; 2) 500 V for 20 min; 3) along a gradient from 100–5000 V for 2 h; and 4) 5000 V for 2 h. Gel strips were then reduced in 1% DTT for 20 min and alkylated in 2.5% iodoacetamide for 20 min. Second-dimension separation was performed by sodium dodecyl sulfate-polyacrylamide

2.8. Protein identification by mass spectrometry Peptides were analyzed by liquid chromatography-tandem mass spectrometry (LC–MS/MS) using a Velos Pro (Thermo Fisher, California, USA) linear ion-trap mass spectrometer. Chromatography was performed in reversed phase columns packed with Magic C18 AQ resin (200 Å, 5 ␮m, Michrom Bioresources, California) in a fused silica capillary with an integrated electrospray tip. An aliquot of 10 ␮L of each sample was injected into the mass spectrometer and scanned in the mass range 400–1500 amu. Spectra were acquired over 50 min for each sample using Xcalibur software (v. 2.06). Automated peak recognition, MS/MS of the top six most intense precursor ions at 35% normalization collision energy, and dynamic exclusion duration of 90 s was performed. The acquired spectra were compared against a database of 65,133 peptide sequences (as of May 19, 2014), consisting of all bivalve genomes on the National Centre for Biotechnology Information website (NCBI; www.ncbi.nlm.nih.gov), in addition to common human and trypsin peptide contaminants. A reverse sequence comparison was included in database runs to eliminate falsediscoveries. Database searches were performed using the Global Proteome Machine (v. 2.2.1) desktop software employing the X! Tandem Sledgehammer algorithm (v. 2013.09.01.2). Only peptides with log (e) values of ≤−10 and ≥8 spectral counts were retained

56

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

Fig. 2. PAH, PCB, tributyltin, cadmium, nickel, and zinc concentrations measured in oyster samples collected in Sydney Harbour. Gray bars are the high-impact sites and white bars are the low-impact sites. Centerlines on the boxplots are the medians; box limits indicate the 25th and 75th percentiles; and whiskers extend 1.5 times the interquartile range. Each boxplot is based on n = 3 samples.

from the GPM. Using these criteria, no reverse database peptide identifications were detected. Acceptable identifications were further scrutinized for molecular weight and isoelectric point before being finalized. The final list of putative identifications for each spot were assigned biological functions based on an assembled database of annotations from NCBI. Five broad biological function categories (energy metabolism, cytoskeleton, stress response, cell communication, protein synthesis) were used to characterize the proteins based on the consensus model of stress responses in oysters from Anderson et al. (2015). Chi-squared goodness-of-fit tests were used to test if the number of differentially expressed proteins differed in each biological function category. 2.9. Statistical analysis Two-sample t-tests were used to compare differences in contaminant concentrations of oyster tissues and hydrologic data between treatment sites in each bay. Summing of PAHs (12 PAH metabolites) and PCBs (14 PCB congeners) followed standard protocols. All values below the detection limit were set to zero (this influenced less than 10% of the entire dataset). Prior to statistical analysis, data were log-normalized to meet assumptions of normal distribution and equal variances. Significance in p-values was adjusted using the Bonferroni correction for multiple comparisons. The Significance Analysis of Microarrays test (Meunier et al., 2005) employing a two-class unpaired design (samr package; R Statistical Software) was used to examine differences in spot abundances between treatment sites in each bay. The non-parametric SAM test is similar to a t-test, except that the False Discovery Rate (FDR) can be accounted for through permutation. Significance is determined by the value ∂ (analogous to the t-statistic), which can be adjusted to reduce the influence of false positives on the test statistic. In our analyses, ∂ value was set to 0.3, which provided for an FDR of 10%. There was no significant effect on the FDR by selecting a smaller ∂ value. Only protein spots with ≤ −2 or ≥+2 fold change were retained in the analysis. A two-way nested non-parametric multivariate analysis of variance (PERMANOVA) was conducted to assess overall patterns in oyster proteomes between sites using all differential spots. The analysis included factors for treatment site (two levels) and bays (nested within site; four levels). Based on the outcome of this anal-

Table 1 Mean water temperature, salinity, pH, and dissolved oxygen at eight sites where oysters were collected in Sydney Harbour. Sites denoted by ‘-1 are the high-impact treatment and sites denoted by ‘-2 are the low-impact treatment. Values are based on averages of three biological replicates per site. Site

Temperature (◦ C)

Salinity (ppt)

pH

DO (%)

BW-1 BW-2 IRC-1 IRC-2 FD-1 FD-2 GC-1 GC-2

20.1 19.2 18.9 19.2 24.0 23.6 23.1 21.4

36.7 37.1 36.2 36.8 34.8 36.0 36.4 37.1

7.99 8.03 7.82 8.02 7.76 7.96 7.83 8.06

84 95 53 87 78 95 84 118

ysis, separate two-way ANOVAs of the individual spot abundances were conducted using the same design as PERMANOVA. Significance was again assessed at p < 0.05 with Bonferroni correction. Finally, an ordination for each bay was created to visualize the overall patterns between sites (e.g., IRC-1 vs. IRC-2), using a non-metric multidimensional scaling (nMDS) method. Both PERMANOVA and nMDS were based on Bray–Curtis dissimilarity matrices, calculated using the normalized abundances of all differential protein spots in oysters from each bay. 3. Results 3.1. Water quality Study sites adjacent to point sources (BW-1, IRC-1, FD-1, and GC-1) exhibited adverse water quality conditions compared to the sites further away from the point sources (BW-2, IRC-2, FD-2, and GC-2) (Table 1 and Fig. 2). Dissolved oxygen (DO) at BW-1, IRC-1, FD-1, and GC-1 ranged from 53-84%, which was significantly lower (p < 0.05) than the DO measured at the sites further away from the point sources (87–118%). In addition, water at FD-1, IRC-1, and GC1 was significantly more acidic (lower pH) than at FD-2, IRC-2, and GC-2, respectively. Salinity and temperature did not differ between sites in any of the sampled areas. This was consistent with sampling during dry-season conditions. Seven chemicals (polychlorinated biphenyls, polyaromatic hydrocarbons, tributyltin, cadmium, mercury, nickel, zinc), rep-

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

57

Fig. 3. 2-DE gel image showing the positions of all 132 differential protein spots identified among the four bays. Average abundance (+S.E.) for two spots (2805, 8407) that exhibited a statistically significant difference between high-impact (black bars) and low-impact sites (white bars) are highlighted as examples.

resenting both legacy and contemporary contaminants, were measured in oyster tissues. Concentrations were quite variable between replicates (e.g., 71–242 ng/g PAHs). However, numerous contaminants still diverged significantly between treatment sites (−1 vs. −2) in each bay (Fig. 2). Blackwattle Bay and Iron Cove exhibited substantial variation in several ‘persistent organic pollutants’ between treatment sites. Elevated (p < 0.05) concentrations of tributyltin (TBT) and polyaromatic hydrocarbons (PAHs) were found at both BW-1 and IRC-1 compared to BW-2 and IRC-2, respectively. Polychlorinated biphenyls (PCBs) were also significantly elevated (p < 0.05) at BW-1. Samples from Five Dock Bay and Gore Cove did not vary in any organic contaminants, but had higher concentrations of zinc (GC-1 and FD-1) and nickel (GC-1). Cadmium concentrations were indistinguishable, and mercury was not detectable (data not shown), between sites in any of the bays. The water quality data presented above corroborates our a priori selection of treatment sites based on proximity to point sources within each bay. In the remainder of this paper, the term “high-impact” and “low-impact” will be used to refer to the two treatments. 3.2. Protein spot analysis A 2-DE proteomic analysis was used to quantify the concentrations of 238 protein spots on 24 gels (4 bays × 2 sites × 3 replicates) (Fig. 3). The relative intensities (spot volume) of 170 spots differed significantly between the high and low-impact sites of the four bays (Table 2). Of these, 132 spots differed significantly between treatment sites in only one bay, whilst the remainder were identified as significantly different in more than one bay. Blackwattle Bay contributed 16% (28 of 170) of the significantly differential spots, while the other three bays contributed greater proportions: Gore Cove (44 of 170, 26%), Iron Cove (49 of 170, 29%), and Five Dock Bay (50 of 170, 29%). These differences in individual protein spots meant that the total proteomes of oysters from each of the high-impact sites were clearly differentiated from the proteomes of oysters from

Table 2 Total number of differential protein spots identified between high and low-impact sites in each bay. Relative intensity was calculated for high-impact sites relative to low-impact sites (i.e., “Decreased relative intensity” = spots with lower spot intensities at the high-impact site). Bay

Total number of differential spots

Increased relative intensity

Decreased relative intensity

BW IRC FD GC Total

27 49 50 44 170

0 37 9 32 78

27 12 41 12 92

the corresponding low-impact sites (Fig. 4). PERMANOVA revealed a significant difference in oyster proteomes between treatment sites (F1,2 = 2.72, p = 0.016) and among bays within treatment (F6,16 = 2.67, p = 0.008). Two-way nested ANOVAs of the individual spot abundances showed that 12 spots were primarily responsible for the treatment difference. These analyses also identified 10 spots that varied significantly between bays but not between high- and low-impact sites. Twenty-five percent of the differential spots were identified as significantly different between high- and low-impact sites in more than one of the bays. Thirty spots were significantly different in two bays and four spots differed in three bays. However, none of the spots differed significantly between high and low sites in all four bays. Forty seven percent of the differential spots that were identified in multiple bays showed the same directionality of response (up or down regulation) in each bay. The majority of differential spots identified in oysters from Blackwattle Bay and Five Dock Bay had lower relative intensities at the high-impact site compared to the low-impact site (100% and 82% of differential spots, respectively). In contrast, oysters from Iron Cove and Gore Cove predominately had differential spots of higher relative intensities at the high-impact site relative to the low-impact site (75% and 72%, respectively). Overall, fewer differential spots increased in abundance than decreased in abundance at the high-impact sites. However, the magnitude of change was far greater in spots of higher abundance. Of the 170 differential

58

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

Fig. 4. Non-metric multidimensional scaling ordination of oyster proteomes from high-impact (black) and low-impact (white) sites for each sampling area. Ovals enclose data points that were not statistically different (at p < 0.05). Stress values ranged from 0.06 (BW) to 0.1 (IRC).

spots identified, 46% (78 of 170) exhibited higher relative intensity at high-impact sites. These spots corresponded to a range in fold change from +2 to +672, with an average fold increase of +36 ± 13 and median fold increase of +3. Of the other 92 spots that exhibited a decreased relative intensities at the high-impact sites, the range in fold change was from −2 to −55, with an average fold decrease of −4 ± 1 and a median fold decrease of −3.

3.3. Peptide identifications Eighty spots were analyzed by LC–MS/MS, of which 61 were assigned putative identities. These 61 spots incorporated 80 distinct proteins. Table 3 lists the identified proteins with the largest fold changes in relative concentration between high and lowimpact sites across the four bays. Supplementary Table A lists all the identities obtained for differential protein spots. The largest fold changes occurred in proteins that increased in abundance, particularly HSP70, peptidyl-prolyl cis-trans isomerase B (PPIB), and radixin. HSP70 had significantly different concentrations between sites in three bays, where fold changes ranged from −9 at Blackwattle Bay to ‘on’ at Five Dock Bay (i.e. spot was only identified at the high-impact site). Lactoylglutathione lyase was identified as the protein with the largest decrease in relative concentration. Many of the proteins identified differed significantly in abundance between high and low-impact sites in multiple bays. Actin differed significantly in abundance in all four bays (Table 4), although not in the same 2-DE spot. ATP synthase, HSP 70, protein disulphide isomerase, and tektin were found at differential concentrations in oysters from three bays, and 29 other proteins were differential in oysters from two bays.

3.4. Functional categories The 80 identified proteins were assigned to five broad functional categories (energy metabolism, cytoskeleton, protein synthesis, cell communication, stress response) (Fig. 5). Over half were associated with energy metabolism (25%) and cytoskeletal activity (28%). These two categories were also the two most frequent functional categories among differential proteins within each bay, except at Five Dock where cell communication was the predominant functional category of differential proteins. A goodness-of-fit test confirmed that the frequencies of energy metabolism and cytoskeletal proteins were significantly higher than proteins in the other functional categories for each of the bays assessed.

4. Discussion Prolonged exposure to environmental stress can significantly affect cellular function in aquatic species, such as oysters. Meistertzheim et al. (2007) found that 24 days of laboratory exposure to thermal stress (25 ◦ C) induced differential gene expression in Crassostrea gigas, primarily by inactivating protein synthesis and metabolic activity. Similarly, four weeks of laboratory exposure to elevated pCO2 (856 ␮atm) severely decreased the activity of major functional processes in the proteome of selectively bred S. glomerata, particularly protein synthesis and cytoskeletal activity (Thompson et al., 2015). However, few field studies have been conducted that focus on the long-term response to multiple stressors (Liu and Wang, 2012). This led us to conduct a baseline study to determine whether the proteomes of Sydney rock oysters (S. glomerata) are substantially affected by prolonged exposure to degraded water quality in the Sydney Harbour estuary. We identified a mixture of persistent organic pollutants, metals, and hydrologic

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

59

Table 3 Proteins with largest fold change in relative concentration between treatment sites. Where the same protein was identified as differential in oysters from multiple bays (e.g., HSP 70), the range in expression is shown (min to max). ‘On’ refers to proteins that were present at the high-impact site, but entirely absent at the low-impact site. Accession #

Protein name

Fold change

Biological function

gi|4838561 gi|405963687 gi|405962295 gi|405954824 gi|405974790 gi|405951006 gi|405967058 gi|405964146 gi|405972836 gi|390979785 gi|405971692 gi|405974628 gi|405961849 gi|405963427 gi|405952287 gi|405968678 gi|405970994 gi|405956520 gi|405965637 gi|405963909

HSP 70 Peptidyl-prolyl cis-trans isomerase B Radixin Severin ATP synthase EF-hand domain-containing protein Glutathione S-transferase P1 Protein disulfide-isomerase Retinal dehydrogenase 1 Procollagen-proline dioxygenase beta Tumor protein D54 Calcium-dependent protein kinase isoform 2 Ras-related protein Rab-7a Malate dehydrogenase Glycerol-3-phosphate dehydrogenase Transmembrane emp24 domain-containing protein 7 Calcyphosin-like protein Putative aminopeptidase W07G4.4 Tubulin alpha-1C chain Lactoylglutathione lyase

−9 to On On −3 to On 3 to On 2 to 5 3 to 5 2 to 5 −4 to 5 4 2 to 4 −7 −7 −7 to 3 −7 −7 −7 to −2 −7 −7 to −2 −10 −13 to −3

Stress response Protein synthesis Cytoskeleton Cytoskeleton Energy metabolism Protein synthesis Stress response Energy metabolism Energy metabolism Energy metabolism Cell communication Cell communication Cell communication Energy metabolism Energy metabolism Cell communication Cell communication Protein synthesis Cytoskeleton Energy metabolism

Fig. 5. Biological functions associated with 80 differential proteins identified by mass spectrometry. The number of proteins in each category is shown for individual bays.

variables that may be associated with changes in gene expression or protein concentration (Liu and Wang, 2012; Muralidharan et al., 2012; Rodriguez-Ortega et al., 2003; Schmidt et al., 2014). Subsequently, 2-DE gel analysis provided strong evidence that oyster proteomes were significantly altered at high-impact sites. This corresponded to sets of differential proteins that were unique to each bay. However, these unique sets of differential proteins contributed to functional pathways that were affected at high-impact sites across all of the bays, suggesting a commonality of subcellular responses. The proportion of protein spots that differed significantly in intensity between treatment sites ranged from 11–21% per bay (n = 238). This was higher than the proteome response revealed in previous laboratory exposures of S. glomerata to four heavy metal contaminants. Thompson et al. (2012) found that the relative intensities of 13% of protein spots (n = 129) were significantly altered after 10-days of exposure to elevated concentrations of cadmium, copper, lead, or zinc. In a field study of S. glomerata, Amaral et al. (2012) found that 7 of 132 (5%) protein spots in oysters collected from acid sulfide soil impacted sites had significantly different

intensities compared to non-impacted sites. In a separate study, S. glomerata transplanted for four days into Lake Macquarie, NSW, where elevated concentrations of metals and hydrologic variation were evident, 46 out of 514 (9%) protein spots differed between high and low-impact sites. These data highlight the substantial proportion of the oyster proteome (5–20%) that can be affected by environmental stress. The differential proteins identified in the current study fell into five broad categories of subcellular function. More than half of these proteins were attributed to energy metabolism and cytoskeletal activity. Thompson et al. (2015) used similar functional categories to assess the response of S. glomerata to elevated pCO2 exposure. They found that cytoskeletal functions accounted for approximately one-third of the differential response, similar to our study. However, protein synthesis weighed heavily in the response to elevated pCO2 , but did not correspond to a significant proportion of the differential proteomes observed in the current study. Similar responses to metals have been found in C. gigas. For example, short term exposure to cadmium resulted in elevated protein synthesis and increased expression of stress proteins (Ivanina et al., 2008).

60

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

Table 4 Proteins identified with differential concentrations between high and low-impact sites in oysters from four different bays in Sydney Harbour. Pie segments: blue = Blackwattle Bay; red = Iron Cove; green = Five Dock Bay; and orange = Gore Cove. For interpretation of the references to color in this table legend, the reader is referred to the web version of this article. Protein name

Differential expression

Actin ATP synthase HSP 70 Protein disulfide-isomerase Tektin Arginine kinase Coactosin-like protein Cytosolic non-specific dipeptidase EF-hand domain-containing protein D2 Ferritin Glutathione S-transferase P 1 Glyceraldehyde-3-phosphate dehydrogenase Glycolipid transfer protein Growth factor receptor-bound protein 2 HSP 60 Lactoylglutathione lyase Major vault protein Mitotic checkpoint protein BUB3 Neurocalcin-like protein Procollagen-proline dioxygenase beta subunit Proteasome subunit alpha type-2 Protein dodo Putative aminopeptidase W07G4.4 Radixin Ras-related protein Rab-11A S-crystallin SL11 Severin Stress-70 protein T-complex protein 1 subunit theta Transaldolase Transgelin-2 Tropomyosin Tryptophanyl-tRNA synthetase, cytoplasmic Tubulin alpha-1C chain

Proteins involved in cell communication are also commonly associated with differential proteomes, confirming active responses to stress (Thompson et al., 2015). However, in the current study, intracellular signaling only represented a significant proportion of the differential proteome at Five Dock Bay. Despite these differences, the five major categories of subcellular function (cytoskeleton, energy metabolism, protein synthesis, cell communication, and stress response) identified in this study corresponded well with the types of functional changes evident in other proteomic analyses of environmental stress in oysters (Tomanek, 2011). These commonalities in the results of proteomic analyses point toward universal cellular response to stress in oysters (Kültz, 2003, 2005), particularly with respect to the cytoskeleton. Impacts on cytoskeletal activity have previously been shown in bivalves exposed to a range of stressors, including heavy metals, polyaromatic hydrocarbons, and thermal stress (Boutet et al., 2004; Farcy

et al., 2007; Meistertzheim et al., 2007; Muralidharan et al., 2012). In the current study, a number of proteins related to structural components of the cytoskeleton (e.g., actin, tubulin, tropomyosin and severin) were differentially expressed in S. glomerata from high-impact sites. Severin, for example, was highly upregulated in oysters from high-impact sites at Iron Cove and Gore Cove. Severin, an actin-binding protein, is known to be activated in response to severe environmental stress, such as oxygen depletion (Mary et al., 2010). It has been proposed that altered concentrations of cytoskeletal proteins indicates an overabundance of reactive oxygen species (ROS) within cells (Anderson et al., 2015; Kültz, 2003). Oxidative stress resulting from excess ROS production has been associated with cytoskeletal damage in oysters and mussels, requiring increased protein turnover (Boutet et al., 2004; McDonagh et al., 2005; Tomanek et al., 2011). The need to protect cells from excess ROS may explain why proteins associated with cellular stress responses, particularly molecular chaperones (e.g., HSPs) and oxidative stress (e.g., glutathione S-transferase) were among the most differential proteins identified in the current study. HSPs have well documented roles in maintaining protein homeostasis (Fabbri et al., 2008), particularly when changes to the actin cytoskeleton are involved (Dineshram et al., 2012; Rodriguez-Ortega et al., 2003). The significant upregulation of stress response proteins may be an indicator of disruption to the cytoskeleton. Hence, the proteomic changes identified in the current study may reflect an adaptive response to excess ROS production through the activation of intracellular stress responses and modified production of cytoskeletal proteins (Tomanek, 2014). Other aspects of the differential proteomes identified in the current study point to a source of excess ROS production in oysters from high-impact sites. Changes in metabolic activity were a common feature of responses at high-impact sites. ATP synthase, protein disulfide-isomerase, and retinal dehydrogenase were among the 10 most highly differential proteins across multiple bays. Significant changes in mitochondrial energy production have been reported in numerous studies of environmental stress in oysters (Anderson et al., 2015; Ivanina and Sokolova, 2013; Tomanek, 2014). These changes appear to be necessary to maintain energy balance in organisms under stress (Kültz, 2005). ROS production is an inescapable by-product of metabolic activity. Thus increasing energy production can lead to oxidative stress within cells. This fits with the overall proteomic responses of oysters from high-impact sites in the current study. A plausible scenario is that exposure to high levels of environmental disturbance leads oysters to increase subcellular energy production with the concurrent production of excess ROS. This may initiate compensatory changes in the cellular stress response and production of cytoskeletal proteins. Even though the majority of differential proteins fell into a common group of just a few functional categories, the majority of differential protein spots were unique to each bay. Only 25% of differential spots occurred in multiple bays, and none of the differential protein spots changed in response in all four bays. Furthermore, differential spots that were in common across bays did not always change in the same direction (up or down regulation). Two factors may have contributed to these differences in the proteomes of oysters between bays. Even though 2-DE can resolve large numbers of protein spots, it may lack the statistical sensitivity to identify all of the differential proteins in comparisons between experimental treatments. Hence, some proteins may have been affected in all of the bays, but those changes may not have been statistically significant in one or two bays. This is an inherent limitation of 2-DE. However, it does not affect the veracity of the broad conclusions of our study, which focus on total proteomes and the broad functional categories of differential proteins. Another explanation for the differences in proteomes between bays is that each bay had a relatively unique combination of contaminants

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

and hydrological conditions. This means that the fine responses to various combinations of stressors are likely to have differed in oysters from the different bays, in addition to a common stress response. There are numerous studies that clearly identify a core set of genes and proteins that contribute to a common stress response pathway in oyster proteomes (Anderson et al., 2015; Tomanek, 2011). However, the manner in which combinations of stressors impact other components of the proteome appears to differ greatly. Thompson et al. (2012) found unique proteome signatures in S. glomerata exposed to different combinations and concentrations of heavy metals and hydrologic variables during three separate field experiments. These data suggest that oyster proteomes reflect an integrated response depending on the specific conditions affecting each population. Regardless of fine differences in oyster proteomes, the current study provides a baseline dataset to investigate adaptation of cellular mechanisms to prolonged stress in S. glomerata. In that context, our data provide an unequivocal outcome. That is, oysters do not acclimate to prolonged, intergenerational exposure to environmental disturbance. Instead, they retain proteomes that are clearly distinct from oysters living at less impacted sites, and those proteomes reflect ongoing cellular stress. Different scales of response may explain this prolonged change. Firstly, S. glomerata may simply induce short-term changes in gene activity when cellular homeostasis is disrupted. Thus, the data could represent a transient expression of subcellular responses. Alternatively, our observations may indicate an adaptive genetic change in S. glomerata. In this case, proteomic responses could be due to the modulation of small sets of genes, or the result of broader changes across the genome. The next phase of this research will investigate these different scales of response to better understand the capacity for local adaptation in S. glomerata. 5. Conclusions Oysters are an important keystone species, yet few studies have investigated the molecular mechanisms behind how they respond to prolonged stress in the natural environment. In this study, we identified a suite of differentially expressed proteins in S. glomerata that appear to reflect an integrated, long-term response to stress. The key finding was that, despite the largely unique changes in the concentrations of individual proteins at different geographic locations, functional annotations identified a common intracellular response to prolonged environmental disturbance. The proteomic response was predominantly associated with impacts on only two broad categories of subcellular function—energy metabolism and the cytoskeleton. Importantly, changes in these important cellular activities are persistent and are not ameliorated by prolonged, intergenerational exposure to stress. Acknowledgements A Melwani is supported by an International Postgraduate Research Scholarship at Macquarie University and a Thyne-Reid Doctoral Fellowship from the Sydney Institute of Marine Science. This work was also partially funded by the Australian Research Council’s Discovery Projects funding scheme (DP120101946). Research was conducted in the Molecular Biology Facility at the Sydney Institute of Marine Science (Publication # 157). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.aquatox.2016.01. 003.

61

References Amaral, V., Thompson, E.L., Bishop, M.J., Raftos, D.A., 2012. The proteomes of Sydney rock oysters vary spatially according to exposure to acid sulfate runoff. Mar. Freshw. Res. 63, 361–369. Anderson, K., Taylor, D.A., Thompson, E.L., Melwani, A.R., Nair, S.V., Raftos, D.A., 2015. Meta-analysis of studies using suppression subtractive hybridization and microarrays to investigate the effects of environmental stress on gene transcription in oysters. PLoS One 10, e0118839. Apraiz, I., Mi, J., Cristobal, S., 2006. Identification of proteomic signatures of exposure to marine pollutants in mussels (Mytilus edulis). Mol. Cell. Proteomics 5, 1274–1285. Bergquist, D.C., Hale, J.A., Baker, P., Baker, S.M., 2006. Development of ecosystem indicators for the Suwannee River estuary: oyster reef habitat quality along a salinity gradient. Estuaries Coasts 29, 353–360. Birch, G., 2007. A short geological and environmental history of the Sydney estuary, Australia. In: Birch, G. (Ed.), Water Wind Art and Debate—How Environmental Concerns Impact on Disciplinary Research. Sydney University Press, Sydney, Australia, pp. 214–242. Birch, G., 2011. Contaminated soil and sediments in a highly developed catchment-estuary system (Sydney estuary, Australia): an innovative stormwater remediation strategy. J. Soils Sediments 11, 194–208. Birch, G., Taylor, S., 1999. Source of heavy metals in sediments of the Port Jackson estuary, Australia. Sci. Total Environ. 227, 123–138. Birch, G.F., Harrington, C., Symons, R.K., Hunt, J.W., 2007. The source and distribution of polychlorinated dibenzo-p-dioxin and polychlorinated dibenzofurans in sediments of Port Jackson, Australia. Mar. Pollut. Bull. 54, 295–308. Birch, G.F., McCready, S., Long, E.R., Taylor, S.S., Spyrakis, G., 2008. Contaminant chemistry and toxicity of sediments in Sydney Harbour, Australia: spatial extent and chemistry-toxicity relationships. Mar. Ecol. Prog. Ser. 363, 71–87. Birch, G.F., Rochford, L., 2010. Stormwater metal loading to a well-mixed/stratified estuary (Sydney Estuary, Australia) and management implications. Environ. Monit. Assess. 169, 531–551. Birch, G.F., Taylor, S.E., 2000. Distribution and possible sources of organochlorine residues in sediments of a large urban estuary, Port Jackson, Sydney. Aust. J. Earth Sci. 47, 749–756. Boutet, I., Tanguy, A., Moraga, D., 2004. Response of the Pacific oyster Crassostrea gigas to hydrocarbon contamination under experimental conditions. Gene 329, 147–157. Brown, C.L., Parchaso, F., Thompson, J.K., Luoma, S.N., 2003. Assessing toxicant effects in a complex estuary: a case study of effects of silver on reproduction in the bivalve, Potamocorbula amurensis, in San Francisco Bay. Hum. Ecol. Risk Assess. Int. J. 9, 95–119. Butt, D., Raftos, D., 2007. Immunosuppression in Sydney rock oysters (Saccostrea glomerata) and QX disease in the Hawkesbury River, Sydney. Mar. Freshw. Res. 58, 213–221. Dineshram, R., Wong, K.W., Xiao, S., Yu, Z., Qian, P.Y., Thiyagarajan, V., 2012. Analysis of Pacific oyster larval proteome and its response to high-CO2. Mar. Pollut. Bull. 64, 2160–2167. Fabbri, E., Valbonesi, P., Franzellitti, S., 2008. HSP expression in bivalves. Invertebr. Surviv. J. 5, 135–161. Farcy, E., Voiseux, C., Lebel, J.-M., Fievet, B., 2007. Seasonal changes in mRNA encoding for cell stress markers in the oyster Crassostrea gigas exposed to radioactive discharges in their natural environment. Sci. Total Environ. 374, 328–341. Gall, M.L., Poore, A.G.B., Johnston, E.L., 2012. A biomonitor as a measure of an ecologically-significant fraction of metals in an industrialized harbour. J. Environ. Monit. 14, 830–838. Hardiman, S., Pearson, B., 1995. Heavy metals, TBT and DDT in the Sydney rock oyster (Saccostrea commercialis) sampled from the Hawkesbury River estuaryNSW, Australia. Mar. Pollut. Bull. 30, 563–567. Hatje, V., Apte, S.C., Hales, L.T., Birch, G.F., 2003. Dissolved trace metal distributions in Port Jackson estuary (Sydney Harbour), Australia. Mar. Pollut. Bull. 46, 719–730. Hedge, L.H., Knott, N.A., Johnston, E.L., 2009. Dredging related metal bioaccumulation in oysters. Mar. Pollut. Bull. 58, 832–840. Ivanina, A.V., Cherkasov, A.S., Sokolova, I.M., 2008. Effects of cadmium on cellular protein and glutathione synthesis and expression of stress proteins in eastern oysters, Crassostrea virginica. J. Exp. Biol. 211, 577–586. Ivanina, A.V., Sokolova, I.M., 2013. Interactive effects of pH and metals on mitochondrial functions of intertidal bivalves Crassostrea virginica and Mercenaria mercenaria. Aquat. Toxicol. 144–145, 303–309. Kültz, D., 2003. Evolution of the cellular stress proteome: from monophyletic origin to ubiquitous function. J. Exp. Biol. 206, 3119–3124. Kültz, D., 2005. Molecular and evolutionary basis of the cellular stress response. Annu. Rev. Physiol. 67, 225–257. Liu, F., Wang, W.-X., 2012. Proteome pattern in oysters as a diagnostic tool for metal pollution. J. Hazard. Mater. 239–240, 241–248. Mary, J., Rogniaux, H., Rees, J.-F., Zal, F., 2010. Response of Alvinella pompejana to variable oxygen stress: a proteomic approach. Proteomics 10, 2250–2258. McDonagh, B., Tyther, R., Sheehan, D., 2005. Carbonylation and glutathionylation of proteins in the blue mussel Mytilus edulis detected by proteomic analysis and Western blotting: Actin as a target for oxidative stress. Aquat. Toxicol. 73, 315–326.

62

A.R. Melwani et al. / Aquatic Toxicology 173 (2016) 53–62

Meistertzheim, A.-L., Tanguy, A., Moraga, D., Thébault, M.-T., Meistertzheim, A.-L., Thébault, M.-T., 2007. Identification of differentially expressed genes of the Pacific oyster Crassostrea gigas exposed to prolonged thermal stress. FEBS J. 274, 6392. Meunier, B., Bouley, J., Piec, I., Bernard, C., Picard, B., Hocquette, J.F., 2005. Data analysis methods for detection of differential protein expression in two-dimensional gel electrophoresis. Anal. Biochem. 340, 226–230. Muralidharan, S., Thompson, E.L., Raftos, D., Birch, G., Haynes, P.A., 2012. Quantitative proteomics of heavy metal stress responses in Sydney rock oysters. Proteomics 12, 906–921. Neave, M.J., Streten-Joyce, C., Nouwens, A.S., Glasby, C.J., McGuinness, K.A., Parry, D.L., Gibb, K.S., 2012. The transcriptome and proteome are altered in marine polychaetes (Annelida) exposed to elevated metal levels. J. Proteomics 75, 2721–2735. Rodriguez-Ortega, M.J., Grosvik, B.E., Rodriguez-Ariza, A., Goksoyr, A., Lopez-Barea, J., 2003. Changes in protein expression profiles in bivalve molluscs (Chamaelea gallina) exposed to four model environmental pollutants. Proteomics 3, 1535–1543. Scanes, P.R., 1996. ‘Oyster watch’: Monitoring trace metal and organochlorine concentrations in Sydney’s coastal waters. Mar. Pollut. Bull. 33, 226–238. Schmidt, W., Rainville, L.C., McEneff, G., Sheehan, D., Quinn, B., 2014. A proteomic evaluation of the effects of the pharmaceuticals diclofenac and gemfibrozil on marine mussels (Mytilus spp.): evidence for chronic sublethal effects on stress-response proteins. Drug Test. Anal. 6, 210–219. Thompson, B., Adelsbach, T., Brown, C., Hunt, J., Kuwabara, J., Neale, J., Ohlendorf, H., Schwarzbach, S., Spies, R., Taberski, K., 2007. Biological effects of anthropogenic contaminants in the San Francisco Estuary. Environ. Res. 105, 156–174.

Thompson, E.L., O’Connor, W., Parker, L., Ross, P., Raftos, D.A., 2015. Differential proteomic responses of selectively bred and wild-type Sydney rock oyster populations exposed to elevated CO2. Mol. Ecol. 24, 1248–1262. Thompson, E.L., Taylor, D.A., Nair, S.V., Birch, G., Haynes, P.A., Raftos, D.A., 2011. A proteomic analysis of the effects of metal contamination on Sydney Rock Oyster (Saccostrea glomerata) haemolymph. Aquat. Toxicol. 103, 241–249. Thompson, E.L., Taylor, D.A., Nair, S.V., Birch, G., Haynes, P.A., Raftos, D.A., 2012. Proteomic discovery of biomarkers of metal contamination in Sydney Rock oysters (Saccostrea glomerata). Aquat. Toxicol. 109, 202–212. Tomanek, L., 2011. Environmental proteomics: changes in the proteome of marine organisms in response to environmental stress, pollutants, infection, symbiosis, and development. Annu. Rev. Mar. Sci. 3, 1–27. Tomanek, L., 2014. Proteomics to study adaptations in marine organisms to environmental stress. J. Proteomics 105, 92–106. Tomanek, L., Zuzow, M.J., 2010. The proteomic response of the mussel congeners Mytilus galloprovincialis and M. trossulus to acute heat stress: implications for thermal tolerance limits and metabolic costs of thermal stress. J. Exp. Biol. 213, 3559–3574. Tomanek, L., Zuzow, M.J., Ivanina, A.V., Beniash, E., Sokolova, I.M., 2011. Proteomic response to elevated P(CO2) level in eastern oysters, Crassostrea virginica: evidence for oxidative stress. J. Exp. Biol. 214, 1836–1844. Vorkamp, K., Strand, J., Christensen, J.H., Svendsen, T.C., Lassen, P., Hansen, A.B., Larsen, M.M., Andersen, O., 2010. Polychlorinated biphenyls, organochlorine pesticides and polycyclic aromatic hydrocarbons in a one-off global survey of bivalves. J. Environ. Monit. 12, 1141–1152.