We designed a salinity stress treatment that simulated conditions of increased fresh water input (e.g. rainfall runoff) that mussels might encounter in their natural environment. For example, in bay habitats salinity levels can fall rapidly during heavy runoff, but low salinity conditions are often short-lived due to tidal flux (Braby and Somero 2006b). Therefore, we exposed mussels to an abrupt drop in salinity and sampled after a four-hour exposure to measure acute responses in gene expression. To serve as baseline control groups for each species, six mussels of each species from the common-garden aquarium were sacrificed at the start of the experiment and their gill tissues dissected and frozen on liquid nitrogen. Treatment animals were then transferred from the common-garden aquarium to a tank in which they were exposed to a reduced salinity (hypo-osmotic) shock of 850 mOs/kg (29.75 ‰). To serve as a time-control, 6 mussels of each species were simultaneously transferred to a separate tank that remained at a salinity of 1000 mOs/kg (35 ‰). Osmolality was measured at the beginning of the exposure using an Advanced Instruments, Inc. osmometer (Norwood, MA, USA) and was 858 mOs/kg for the hypo-osmotic exposure and 979 mOsm/kg for both the common-garden aquarium (baseline-control) and the time-control aquarium. We inserted corks between the valves of both the treatment and time-control mussels to prevent the valves from closing during the experiment. While the application of corks was not environmentally realistic, we sought to maximize the hypo-osmotic exposure to the tissues and minimize the inter-individual variation in response to salinity stress that might have resulted from behavioral idiosyncrasies in valve (shell) gaping. After four hours, six treatment and six time-control mussels of each species were sacrificed, and their gill tissues were dissected and frozen on liquid nitrogen. We used gill tissues for this study because (1) stress-induced protein expression has been shown to be responsive to environmental stress in Mytilus gill (Hofmann & Somero 1995; Hofmann & Somero 1996), (2) Mytilus californianus has been shown to have highly similar stress-induced transcriptomes between gill, mantle and adductor muscle (Gracey et al. 2008), indicating that gill can be used as a representative tissue; and (3) the microarray was designed from expressed sequence tag (EST) sequences generated from the gill tissue of the out-group species, M. californianus (Gracey et al. 2008; Lockwood et al. 2010). Tissues were stored at -70°C prior to RNA extraction. During the experiment we concurrently exposed a different set of mussels to a more severe salinity stress of 700 mOs/kg (24.5 ‰). Subsequently, a proteomic comparison of these salinity treatments using 2-dimensional gel electrophoresis revealed that the 700 mOs/kg treatment rendered animals unable to elicit broad changes in protein expression (L. Tomanek, personal communication). This indicated that the 700 mOs/kg treatment was too severe for the animals to respond via the cellular stress response (Kültz 2005). Therefore, we focused our transcriptomic analysis on the more moderate stress of 850 mOs/kg in order to investigate immediate and early responses in gene expression to hypo-osmotic shock.
Growth protocol
We collected adult mussels and transported them back to the laboratory of Dr. Lars Tomanek at California Polytechnic State University, San Luis Obispo, as previously described (Lockwood et al. 2010). We chose to conduct our study on adult mussels because the Mytilus transcriptome has only been characterized in adults (Gracey et al. 2008; Lockwood et al. 2010). Mytilus trossulus (Gould 1850) were collected from Newport Harbor, Newport, OR, USA (44°38’25” N, 124°03’10” W), and Mytilus galloprovincialis (Lamarck 1819) were collected from Santa Barbara Harbor, Santa Barbara, CA, USA (34°24’15” N, 119°41’30” W). Specimens were kept in a re-circulating aquarium in which they were common-garden acclimated in seawater (1000 mOs/kg = 35 ‰ salinity) at 13°C for four weeks in order to fully acclimate both species to the same osmotic conditions (Gilles 1972). To ensure that no hybrid individuals were included in our study, we chose both collection sites to be well outside of the California blue mussel hybrid zone (McDonald & Koehn 1988; Braby & Somero 2006b). Molecular species identification was performed on each specimen by published methods (Heath et al. 1995; Rawson et al. 1996).
Extracted molecule
total RNA
Extraction protocol
Total RNA was extracted from gill tissues with a TissueLyser homogenizer (Qiagen; Valencia, CA, USA) and RNeasy kit (Qiagen).
Label
Alexa Fluor 647
Label protocol
Total RNA was reverse-transcribed and amplified with the AminoAllyl MessageAmp II kit (Ambion; Foster City, CA, USA). To reduce problems in transcript binding due to sequence differences between species, we employed a reference design that used separate pools of reference RNA for each species. Reference amplified RNA (aRNA) was created for each species by pooling RNA before and after amplification. The reference pool was made up of RNA from six different samples: two base-line control samples from the beginning of the experiment, two treatment samples from the end of the four-hour hypo-osmotic exposure, and two time-control samples from the end of the four-hour exposure time. aRNAs from reference and experimental samples were separately labeled with Alexa Fluor fluorescent dyes (Invitrogen, Carlsbad, CA, USA), 555 for reference and 647 for control or treatment.
Channel 2
Source name
M. trossulus pooled: Time-point zero control, 4 hour control, 850-mOsm_kg_salinity_stress
We designed a salinity stress treatment that simulated conditions of increased fresh water input (e.g. rainfall runoff) that mussels might encounter in their natural environment. For example, in bay habitats salinity levels can fall rapidly during heavy runoff, but low salinity conditions are often short-lived due to tidal flux (Braby and Somero 2006b). Therefore, we exposed mussels to an abrupt drop in salinity and sampled after a four-hour exposure to measure acute responses in gene expression. To serve as baseline control groups for each species, six mussels of each species from the common-garden aquarium were sacrificed at the start of the experiment and their gill tissues dissected and frozen on liquid nitrogen. Treatment animals were then transferred from the common-garden aquarium to a tank in which they were exposed to a reduced salinity (hypo-osmotic) shock of 850 mOs/kg (29.75 ‰). To serve as a time-control, 6 mussels of each species were simultaneously transferred to a separate tank that remained at a salinity of 1000 mOs/kg (35 ‰). Osmolality was measured at the beginning of the exposure using an Advanced Instruments, Inc. osmometer (Norwood, MA, USA) and was 858 mOs/kg for the hypo-osmotic exposure and 979 mOsm/kg for both the common-garden aquarium (baseline-control) and the time-control aquarium. We inserted corks between the valves of both the treatment and time-control mussels to prevent the valves from closing during the experiment. While the application of corks was not environmentally realistic, we sought to maximize the hypo-osmotic exposure to the tissues and minimize the inter-individual variation in response to salinity stress that might have resulted from behavioral idiosyncrasies in valve (shell) gaping. After four hours, six treatment and six time-control mussels of each species were sacrificed, and their gill tissues were dissected and frozen on liquid nitrogen. We used gill tissues for this study because (1) stress-induced protein expression has been shown to be responsive to environmental stress in Mytilus gill (Hofmann & Somero 1995; Hofmann & Somero 1996), (2) Mytilus californianus has been shown to have highly similar stress-induced transcriptomes between gill, mantle and adductor muscle (Gracey et al. 2008), indicating that gill can be used as a representative tissue; and (3) the microarray was designed from expressed sequence tag (EST) sequences generated from the gill tissue of the out-group species, M. californianus (Gracey et al. 2008; Lockwood et al. 2010). Tissues were stored at -70°C prior to RNA extraction. During the experiment we concurrently exposed a different set of mussels to a more severe salinity stress of 700 mOs/kg (24.5 ‰). Subsequently, a proteomic comparison of these salinity treatments using 2-dimensional gel electrophoresis revealed that the 700 mOs/kg treatment rendered animals unable to elicit broad changes in protein expression (L. Tomanek, personal communication). This indicated that the 700 mOs/kg treatment was too severe for the animals to respond via the cellular stress response (Kültz 2005). Therefore, we focused our transcriptomic analysis on the more moderate stress of 850 mOs/kg in order to investigate immediate and early responses in gene expression to hypo-osmotic shock.
Growth protocol
We collected adult mussels and transported them back to the laboratory of Dr. Lars Tomanek at California Polytechnic State University, San Luis Obispo, as previously described (Lockwood et al. 2010). We chose to conduct our study on adult mussels because the Mytilus transcriptome has only been characterized in adults (Gracey et al. 2008; Lockwood et al. 2010). Mytilus trossulus (Gould 1850) were collected from Newport Harbor, Newport, OR, USA (44°38’25” N, 124°03’10” W), and Mytilus galloprovincialis (Lamarck 1819) were collected from Santa Barbara Harbor, Santa Barbara, CA, USA (34°24’15” N, 119°41’30” W). Specimens were kept in a re-circulating aquarium in which they were common-garden acclimated in seawater (1000 mOs/kg = 35 ‰ salinity) at 13°C for four weeks in order to fully acclimate both species to the same osmotic conditions (Gilles 1972). To ensure that no hybrid individuals were included in our study, we chose both collection sites to be well outside of the California blue mussel hybrid zone (McDonald & Koehn 1988; Braby & Somero 2006b). Molecular species identification was performed on each specimen by published methods (Heath et al. 1995; Rawson et al. 1996).
Extracted molecule
total RNA
Extraction protocol
Total RNA was extracted from gill tissues with a TissueLyser homogenizer (Qiagen; Valencia, CA, USA) and RNeasy kit (Qiagen).
Label
Alexa Fluor 555
Label protocol
Total RNA was reverse-transcribed and amplified with the AminoAllyl MessageAmp II kit (Ambion; Foster City, CA, USA). To reduce problems in transcript binding due to sequence differences between species, we employed a reference design that used separate pools of reference RNA for each species. Reference amplified RNA (aRNA) was created for each species by pooling RNA before and after amplification. The reference pool was made up of RNA from six different samples: two base-line control samples from the beginning of the experiment, two treatment samples from the end of the four-hour hypo-osmotic exposure, and two time-control samples from the end of the four-hour exposure time. aRNAs from reference and experimental samples were separately labeled with Alexa Fluor fluorescent dyes (Invitrogen, Carlsbad, CA, USA), 555 for reference and 647 for control or treatment.
Hybridization protocol
Competitive hybridization (experimental - AlexaFluor 647 versus reference - AlexaFluor 555). We followed the recommended protocols from Agilent Technologies, Inc. for all hybridization and washing steps.
Scan protocol
Microarrays were scanned with an AXON GenePix 4000B scanner (Axon Instruments, Molecular Devices, Sunnyvale, CA, USA).
Description
SC0_tr_1 Biological replicate 1 of 6. M. trossulus control, 1000-mOsm_kg, beginning of experiment.
Data processing
Fluorescent signal intensities for each spot on the microarray were extracted and Lowess normalized within each microarray using Feature Extraction Software (version 9.5.3.1; Agilent Technologies, Inc.). Given the heterologous hybridization design, we performed stringent quality control to eliminate data from probes that had low signal intensities in any one of the 36 samples. We followed the recommendations of Agilent Technologies and considered a probe to have poor hybridization if (1) the signal intensity was less than 150% of the local background and (2) the hybridized spot diameter was less than 30% that of the nominal spot diameter. Since multiple probes represented each gene on the microarray, data from good probes were summarized by computing the geometric mean of signal intensities for each probe set on a single microarray. We then computed the log2-ratios of the experimental channel divided by the reference channel. In order to minimize the effects of time-of-day on gene expression, we normalized the expression values of the treatment individuals to the median log2-ratio of the time-control samples of the same species. Log2-ratios of the baseline-control samples from the beginning of the experiment were normalized to the median log2-ratio within that group, separately for each species. The normalized log2-ratio data from the treatment samples were compared to those of the baseline-control samples to assess changes in gene expression.
