Salmon spawning migration: Metabolic shifts and environmental triggers

Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

Contents lists available at ScienceDirect

Comparative Biochemistry and Physiology, Part D j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / c b p d

Salmon spawning migration: Metabolic shifts and environmental triggers Kristina M. Miller a,⁎, Angela D. Schulze a, Norma Ginther a, Shaorong Li a, David A. Patterson b, Anthony P. Farrell c, Scott G. Hinch d a

Molecular Genetics Section, Pacific Biological Station, 3190 Hammond Bay Road, Fisheries and Oceans Canada, Nanaimo, BC, Canada V9T 6N7 Fisheries and Oceans Canada, Cooperative Resource Management Institute, School of Resource and Environmental Management, Simon Fraser University, Burnaby, BC, Canada V5A 1S6 Department of Zoology and Faculty of Land and Food Systems, University of British Columbia, Vancouver, British Columbia, Canada V6T 1Z4 d Centre for Applied Conservation Research, Department of Forest Sciences, University of British Columbia, 2424 Main Mall, Vancouver, British Columbia, Canada V6T 1Z4 b c

a r t i c l e

i n f o

Article history: Received 18 September 2008 Received in revised form 25 November 2008 Accepted 25 November 2008 Available online 6 December 2008 Keywords: Migration Genomics Microarray Muscle Atrophy Sockeye

a b s t r a c t A large-scale functional genomics study revealed shifting metabolic processes in white muscle during the final 1300 km migration of wild sockeye salmon to their spawning grounds in the Fraser River, British Columbia. In 2006, Lower Adams stock sockeye salmon ceased feeding after passing the Queen Charlotte Islands, 850 km from the Fraser River. Enhanced protein turnover and reduced transcription of actin, muscle contractile and heme-related proteins were early starvation responses in saltwater. Arrival to the estuarine environment triggered massive protein turnover through induction of proteasomal and lysosomal proteolysis and protein biosynthesis, and a shift from anaerobic glycolysis to oxidative phosphorylation. Response to entry into freshwater was modest, with up-regulation of heat shock proteins and nitric oxide biosynthesis. High river temperatures resulted in a strong defense/immune response and high mortalities in 50% of fish. Arrival to the spawning grounds triggered further up-regulation of oxidative phosphorylation and proteolysis, down-regulation of protein biosynthesis and helicase activity, and continued down-regulation of muscle proteins and most glycolytic enzymes. However, sharp up-regulation of PFK-I indicated induction of glycolytic potential at the spawning grounds. The identification of potential environmental cues triggering genome-wide transcriptional shifts in white muscle associated with migration and the strong activation of proteasomal proteolysis were both novel findings. Crown Copyright © 2008 Published by Elsevier Inc. All rights reserved.

1. Introduction Migration is regarded as one of the most demanding and physiologically challenging phases of salmon life history and represents a complex interplay between physiology and behaviour. Anadromous Pacific salmon species spend their first 3-24 months in freshwater (FW), and then migrate to saltwater (SW) where they typically spend 2–4 years before returning to natal rivers to spawn (Groot and Margolis, 1991). Whereas rearing in protective FW streams and lakes before migrating to the productive ocean for rapid growth has benefited salmon for millions of years, the shift between these very different environments is costly. Indeed, significant mortality is experienced during FW/SW transitions (Hinch et al., 2006), with mortality rates up to 95% for juvenile salmon during their first 6 months in the ocean in years when growing conditions are poor due to high temperatures, low food availability or high predation (Parker, 1968; Bax, 1983; Beamish and Mahnken, 2001; Wertheimer and Thrower, 2007). Similarly, adult fish may experience high mortality on their return spawning migration back to FW. En-route

⁎ Corresponding author. Tel.: +1 250 756 7155; fax: +1 250 756 7053. E-mail address: [email protected] (K.M. Miller).

(during FW migration) and pre-spawning (at spawning grounds) mortality typically varies between 10–20%, but over the past decade has averaged N50% for several stocks of sockeye salmon in BC and N90% for Fraser River sockeye (Oncorhynchus nerka Walbaum) stocks in years with unusually high water temperature (Cooke et al., 2004). The large white muscle mass is the energy store for migrating fish (Van den Thillart and van Raaij, 1995). Returning adult salmon cease feeding before entering FW (Burgner, 1991). Consequently, they utilize the substantial lipid and protein stores in white muscle mass as fuel for maintenance, swimming and reproductive development, selectively breaking down the white muscle tissue over time and thereby reducing their ability to burst swim (Mommsen et al., 1980). The physiological shifts in white muscle tissue associated with adult spawning migration involve an integrated response to changes resulting from exercise, fasting, salinity transition and sexual maturation. A previous study by Magnoni et al. (2006) based on plasma analyses showed that lipids and to a lesser extent proteins provided most of the energy for river migration. Additionally, many of the catabolized protein and fat products are funnelled into building blocks for the developing oocytes (hepatic vitellogenin biosynthesis), while the carbohydrates, used intermittently and rebuilt from amino acids, are specifically required for spawning (French et al., 1983; Donaldson et al., 2000; Mommsen, 2004). There exists a balance between energy

1744-117X/$ – see front matter. Crown Copyright © 2008 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.cbd.2008.11.002

76

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

usage to fuel migration and that used for developing oocytes, the nature and control of which is not well understood. It is expected, however, that senescence ensues when there is no fuel left in the white muscle tissue for either process. Salem et al. (2006a) conducted a microarray study to elucidate the processes that occur during atrophy of white muscle tissue specifically associated with the maturation process in salmon. They showed that there were four common trends in the transcription profile of vitellogenesis-induced muscle atrophy in non-starving rainbow trout (Oncorhynchus mykiss, previously: Salmo gairdneri; RBT), including up-regulation of protein turnover (myofibrillar, extracellular and cytoskeleton proteins), suppression of the overall protein biosynthetic process, down-regulation of the glycolytic pathway, and differential regulation of signal transduction cascades that regulate cell cycle and apoptosis. In addition, unlike mammalian muscle atrophy models, the proteasomal proteolytic pathway was not found to be significant in protein turnover in the iteroparous (repeat spawning) RBT. The Salem et al. (2006a) study, while documenting the physiological processes associated with late-stage maturation and providing a good foundation of functional genomics knowledge, highlighted only a single time comparison of mature vs. immature salmon. Conversely, our study sought to identify, over time and space, the series of genomic responses leading ultimately to white muscle atrophy during migration. Our study was based on a single stock of wild-caught sockeye salmon undergoing a final 1330 km migration to the spawning grounds. Transcriptional profiles, conducted over 16,006 gene features and including 11,621 unique genes (identified by the number of gene contigs; http://web.uvic.ca/grasp), represent physiological shifts potentially cued through environmental (salinity and olfactory cues, high river temperature, pheromones at the spawning site), behavioural (prolonged food deprivation), conditional (gross somatic energy, stress or disease), and programmed physiological (maturation) processes. Key goals of this study were two-fold: 1) to provide a detailed analysis of the physiological processes transcriptionally activated or de-activated in white muscle tissue during sockeye salmon adult return migration, and 2) to identify the timing and location where key

physiological processes shift and to use this information to hypothesize potential internal and environmental cues responsible for triggering metabolic shifts in white muscle. To do so, we incorporated information on environmental conditions at each sample site (salinity to determine degree of FW influence in the marine environment, temperature in FW) into the functional genomic analysis. While we expect that energy metabolism was strongly influenced by energy needs to fuel both (swimming) migration and reproductive loading, no direct experimental attempt was made to tease apart these processes. Finally, because our functional genomics study was based on wild-caught fish, we attempted to minimize genetically-based variability by confining analyses to a single stock complex, the late-run Adams/Shuswap complex in the Fraser River, through use of individual-based genetic stock identification using the methods described by Beacham et al. (2005). 2. Materials and methods 2.1. Sampling wild fish Fraser River (FR) sockeye salmon (O. nerka) from the Adams River stock were collected in late July through November, 2006, at seven sites along their migration route to the spawning grounds (Fig. 1). Commercial purse seine boats were used in collections at three marine sampling sites, including the 1) Queen Charlotte Islands (QCI), 850 km from the FR (July 28–Aug 3), 2) Juan de Fuca Strait (JDFS), the outer southern passage route around Vancouver Island approximately 300 km to the FR (Aug 6–10), and 3) Johnstone Strait (JS), the inner northern passage route around Vancouver Island approximately 200 km from the FR (Aug 11–27). A commercial seine boat was also used to capture fish in Strait of Georgia (SOG), 15 km from the FR mouth (Aug 29–31) in the estuarine waters. A commercial gill net boat was used to collect fish from the first of three FW sites, at Whonnock (W), located 69 km upriver from the mouth of the FR (Aug 24–31, Sept 19). A beach seine was used to collect fish within the South Thompson River at Savona (SV), 391 km upriver (Sept 20), and in collections of fish arriving to the Lower Adams spawning grounds (LA), 480 km from the FR entrance (Oct 19). A total of 663 adult migrating sockeye

Fig. 1. Map of British Columbia showing collection sites in the Pacific Ocean and Fraser River. Numbers in parentheses indicate sample sizes of Lower Adams stock fish used in the microarray experiment. The migration tract from the Queen Charlotte Islands to the Lower Adams Spawning grounds spans 1300 km in distance by water.

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

salmon were sampled destructively and 1038 were tagged and sampled non-destructively. The late-run Lower Adams River stock was the most dominant Fraser sockeye run for 2006 and individual fish were confirmed to belong to this stock through use of genetic stock identification (Beacham et al., 2005). The Pacific Salmon Commission utilizes genetic stock identification for in-season management of sockeye salmon, with regular (2–3 times per week) test fisheries at JS, JDFS and W throughout the summer. We used these data to schedule sampling in each area to times when Lower Adams fish were present in high abundance. Muscle biopsy samples were collected non-lethally in JS and JDFS after fish were individually removed from the purse seine by dip net and placed into a trough with flowing seawater for sampling. These fish were subsequently fitted with radio-tags and tracked throughout their migration in the river to identify timing of river entry and subsequent fate in the river (data not shown; sampling methods in English et al., 2005; Cooke et al., 2008). At all other sites, fish were killed quickly by a blow to the head before the biopsy. A muscle biopsy was removed between the lateral line and the dorsal fin using a 3 mm biopsy punch. The horizontal position of this sample was anterior to the adipose fin and basically in line with the mid-point of the anal fin insertion. The sample was primarily white muscle tissue and any small amount of red muscle near the skin was removed later. In all cases, muscle samples were collected within minutes of capture and flash frozen on dry ice or in liquid nitrogen. An adipose fin clip sample was preserved in ethanol for DNA stock identification. 2.2. Total RNA isolation Samples were quickly frozen on dry ice and stored in a charged liquid nitrogen cryoshipper for a few days until they were brought to Pacific Biological Station, at which time they were transferred to −80 °C storage until analyzed. Total RNA was purified from white muscle tissue using Magmax™-96 for Microarrays Kits (Ambion Inc, Austin, TX, USA) with a Biomek NXP (Beckman-Coulter, Mississauga, ON, Canada) automated liquid-handling instrument. Approximately 10 mg of frozen tissue was homogenized with stainless steel beads in TRI-reagent (Ambion) on a MM301 mixer mill (Retsch Inc., Newtown, PA, USA). 100 µL aliquots of homogenate were pipetted into 96-well plates and extractions were carried out according to the manufacturer's instructions using the “No-Spin Procedure” for frozen tissues, on the Biomek FXP. In the final step, RNA was eluted with RNAase-, DNAase-free water and RNA yield was determined by measuring the A260 of the eluate. Purity was assessed by measuring the A260/A280 ratio of the eluate. Solutions of RNA were stored at −80 °C until use for cDNA synthesis or qPCR.

control hybridized onto each microarray slide. All individual (experimental) samples were fluorescently tagged with Alexa 555 and the reference labelled with Alexa 647. Samples and references were cleaned up by adding 150 µL of DNA binding buffer (Zymo) to each Alexa tube and then combining the sample and references for each slide into Zymo25 clean-up columns. The unbound portion of the labelled cDNA was removed by centrifugation at 12,000 g. The labelled cDNA was washed three times with DNA wash buffer (Zymo) and eluted in 9 µL of 1X TE buffer. 2 μL of poly dA was added to the targets, which were then denatured for 10 min at 80 °C, followed by the addition of 125 µL of prewarmed SlideHybe3 buffer (Ambion) before loading into the hybridization chamber in Tecan-HS4800 Pro Hybridization Station (Tecan Trading AG, Switzerland). Hybridizations were performed in batches of 12–24 samples and conducted over a two-week period to minimize technical variance. 2.4. Salmon arrays The 16,006 feature microarray used in this study was produced by the Genome Canada funded Genomics Research on Atlantic Salmon Project (GRASP http://www.uvic.ca/cbr/grasp; B.F. Koop & W. Davidson; von Schalburg et. al., 2005). The experiment included a minimum biological replicate sample size of six to eight fish per sample site, but was increased to 16–18 for JDFS and JS and samples to accommodate intra-site variables of interest such as river entry timing and fate in-river (these data are not presented herein), with a total of 80 experimental microarray slides (Table 1). At W, the first FW site, fish were sampled on three dates, two in August (10 fish) and one in September (6 fish), which allowed us to assess temporal shifts within a site (representing early and late river entry, respectively). A single technical replication at Savona was included in the experiment. In each of the technical steps leading to slide hybridization, samples were either randomized (RNA extraction, labelling, hybridization) or run all at once (amplification); hence, the need for technical replicates was not warranted, especially given the extensive biological replication. Slide processing steps within the Tecan hybridisation station were conducted as follows: one 30 s slide wash step (0.014% SDS) at 23 °C with a 30 s soak time, two 30 s slide wash steps (sdH2O) at 23 °C with 30 s soak times, 3 min slide denaturation at 85 °C with agitation, 1 min wash (5x SSC, 0.01% SDS, 0.2% BSA) at 43 °C with a 30 s soak time, 1 h pre-hybridization at 43 °C with medium agitation, two 1.5 min washes (water) at 43 °C with a 30 s soak time, followed by 1 min wash (8x SSC, 0.1% SDS, 4X Denhardt's) at 43 °C, sample injection at 43 °C with agitation, hybridization for 16 h at 43 °C in high viscosity mode with low frequency agitation, four 30 s wash steps, two at 43 °C (2x SSC, Table 1 Experimental design for white muscle microarrays on Lower Adams sockeye salmon.

2.3. aRNA labelling Because the tissue biopsy samples were small, in order to obtain enough RNA to run on a microarray slide we had to first amplify the total RNA. 500 ng to 5 µg of total RNA was amplified using a MessageAmp™II-96 kit (Ambion), performed manually according to manufacturer's instructions. Amplification steps were performed on a single 96-well plate for all samples at once. 5 µg of aRNA were reverse transcribed into cDNA and labelled with Alexa dyes using the Invitrogen Indirect Labelling Kit, with modifications from the manufacturer's instructions. Briefly, the cDNA was purified using Zymo-25 clean-up columns (Zymo Research, Orange, CA, USA) and eluted using the 2X coupling buffer, supplied by Invitrogen. During dye labelling, samples were processed individually by first resuspending the Alexa dye with DMSO, then adding the cDNA and incubating for 1 h at room temperature. For microarray experiments, a pooled reference comprised of the white muscle RNA from all of the fish used in the experiment was used, with labelled cDNA from an individual fish and the reference

77

Entry-timing Sampling Location

#fish

Queen Charlotte Islands (QCI) Juan de Fuca Strait (JDFS) Johnstone Strait (JS) Strait of Georgia (SOG) Whonnock August (W) Whonnock September (W) Savona (SV) Lower Adams Spawning (LA) TOTAL

8 16 18 7 10 6 6 + 1 tech 8 80

Early

Normal

9 11

7 7

10 0

0 6

Sex Female

Male

4

4

6 9 2 4 8

1 1 4 3 0

On each microarray slide, the amplified cDNA from a single experimental fish (labelled with Alexa 555) was hybridized against a reference control (labelled with Alexa 647) comprised of all fish used in the experiment. Sex could not be identified in nondestructively sampled fish from JDFS or JS. The timing of freshwater (FW) entry (entry timing) was known only for fish radio-tagged in the marine environment (in JS and JDFS) and for fish sampled in W, the first freshwater sampling site. While intra-site variables such as entry timing and sex were included in the experimental design, these are not discussed herein, but rather are part of a larger, multi-stock experiment (Schulze and Miller, unpublished data).

78

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

0.01% SDS) followed by two at 23 °C (0.2x SSC), and slide drying at 23 °C for 2.5 min. The fluorescent images were obtained by scanning the slides using a Perkin Elmer ScanArray Express, adjusting the PMT gain for optimized visualization of each image. The images were quantified using Imagene (BioDiscovery, El Segundo, CA, www.biodiscovery. com) and spots with poor quality or no signal (b2 standard deviations from background) at both wavelengths were flagged. Raw microarray intensity data were normalized in GeneSight (version 4.1, BioDiscovery, Inc., www.biodiscovery.com) using the local intensity-dependent loess normalization in order to remove intensity-dependent dye bias (Yang et al., 2002). Flagged spots were replaced by the experimental means for each gene. All data were log transformed (base 2) and an intensity ratio was computed by taking the differences in log transformed intensities between the sample and reference control. These log-transformed intensity ratios were used in all further analyses. Microarray data were expressed in terms of mean normalized (background corrected) log2 ratios between each fish and the reference control. Microarray data were deposited (according to Microarray Gene Expression Data Society Standard) in the NCBI Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) with the experimental accession number GSE13657 and individual slide series GSM343584–GSM343662. A supplemental Table 1 containing significant gene lists is also included in the GEO submission.

sample sites are and whether there are “outlier” fish within sites that do not match the predominant physiological signatures. To identify genes and processes that were on a trajectory of change, we conducted linear regressions with distance to the spawning grounds for each of the 16,006 genes on the slide. 2.6. Functional analysis The publicly available program Database for Annotation, Visualization and Integrated Discovery (DAVID) was used to identify enriched biological themes, or gene ontology (GO)-terms, within lists of significant genes (http://niaid.abcc.ncifcrf.gov/; Huang et al., 2007). Within DAVID, gene identifiers from Unigene or the non-redundant protein database in GenBank were first converted to DAVID ID's. DAVID analyses compared the functional annotation associated with each group of significantly differentially regulated genes with that present on the slide as a whole to identify over-represented functional categories. While this analysis works well for identifying large, well represented pathways on the slide such as those associated with metabolism, pathways involved in immune and defense response and osmoregulation do not annotate well between genomes for fish and model mammals, and hence are poorly represented within DAVID analysis. Therefore, some of these processes were considered separately. 3. Results

2.5. Statistical analysis of array data 3.1. Intensity and patterns of physiological change associated with migration In defining significance, we focused initially on the intensity of physiological change, i.e. how many genes statistically differentiate expression profiles from one site to another. However, for complex comparisons such as these, we also looked closely at the pattern of physiological change (i.e., the physiological relationships among individuals and sites). Therefore, we asked questions such as: Do salmon sampled at sites along the migration path carry unique expression profiles indicative of physiological change? Are there regional clusters of sites that carry similar patterns? Are there specific points along the migration path where major shifts in gene expression occur? Are there outlier samples that appear physiologically distinct from the majority of samples within a site? A simple ANOVA conducted over all sample sites was used initially to identify, through unsupervised hierarchical clustering, the most differentiated sites. This ANOVA did not address nested intra-site variables such as sex or river entry timing, which are a topic of a companion manuscript in prep. Thereafter, pairwise t-tests conducted between sample sites were used to determine whether fish sampled from adjacent sites were statistically differentiated on the basis of gene expression data from the 16,006 genes on the array. Additional t-tests were performed between the two most distal sites, QCI and LA, between the two marine sites reflecting different approaches to the Fraser River, JS and JDFS, and between these sites and the first FW entry site, W. A conservative p-value cut-off of 0.001 was used to identify significant gene lists for functional analysis. At this p-value, we expect approximately 16 false positives by chance. Randomized simulations of the data yielded an estimate of 12–22 false positives. We chose to use p b 0.001 as opposed to a more stringent false discovery rate that accounts for multiple testing to maximize our ability to resolve functional shifts among sites as opposed to individual biomarkers. These initial statistical analyses identified genes that were significantly up- and down-regulated between sites. Significant gene lists were combined and clustered using unsupervised two-way hierarchical clustering based on Pearson correlation coefficients to reveal the relationships among genes, individuals and sites. Hierarchical clusters resolve physiological relationships among individuals (x-axis) and genes (y-axis) simultaneously, and can be used to identify where major shifts in physiology take place. They can also reveal how distinctive

The microarray experiment was designed to document the physiological changes that occur in white muscle during the course of migration and to ascertain where and when major physiological shifts occurred. The most pronounced transcriptional shifts occurred between the most distal sampled marine site (QCI) and the spawning grounds (LA) (1601 gene features; hereafter referred to as genes, but note that all are not unique) (Fig. 2). Between adjacent sites, maximum divergence was observed between the outer marine route (JDFS) and the estuary (SOG) (427 genes) and QCI and the inner marine route (JS) (414 genes). Conversely, only minimal change was noted between the SOG estuary and first entry into the FR (W) (57 genes) despite the fact that this movement involved a shift from SW to FW and initiated the upstream swimming. Similar numbers of genes were differentially regulated between the four saltwater (SW) sites (union of site to site t-tests was 1181 genes) and between the three FW sites (1177 genes; with SV treated as a single group, see below). However, if we discount the estuarine SOG sample (it was included

Fig. 2. Spatial patterns of significant transcriptional shifts, as shown by the numbers of differentially regulated genes in pairwise t-tests between sites. Comparisons between adjacent sites are shown in black, and non-adjacent sites in grey. Site abbreviations as in Table 1. Graph starts with QCI, the most distal site from the spawning grounds (1300 km to LA).

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

as a SW site because of the environment, but clearly carries greater resemblance physiologically to FW samples), then only 578 genes were significant in SW (between QCI, JS and JDFS combined). Unsupervised hierarchical plots were used to visually resolve the relationships among samples and sites and to identify where major physiological switch points in gene expression occurred. Distinct profiles were observed between fish sampled in marine and FW environments, with fish from the estuarine SOG clustering closely with those from FW (Fig. 3a). The technical replicates at SV clustered tightly together, carrying a correlation coefficient over all significant genes of 0.73 compared to an average correlation coefficient among

79

pairwise samples within sites of 0.38. JS fish showed a high degree of individual divergence, some clustering with SW and others intermediate between FW and SW. In 2006, the FW influence of the FR extended all the way to JS at the time of fish sampling (Rick Thompson, personal communication); hence the variability within JS may have been influenced by exposure to lower salinity water within the inner passage. Alternately, given the broad temporal sampling of JS fish (Aug 11–27), temporal variation could also be a factor; three of the four fish collected the last week in August clustered with FW samples. Overall, it is clear that exposure to the estuarine environment signals the most profound shifts in physiology occurring in white muscle.

Fig. 3. Hierarchical clustering plots of a union of all significant genes over the course of migration (a) and a union of significant genes in FW, treating the two highly differentiated profiles in SV as separate groups (b). The clustering is “unsupervised”, with genes clustering on the y-axis and samples on the x-axis. Plot B also shows the over-representation of biological processes within each cluster, as resolved through DAVID analysis. Colored branches on the x-axis in indicate the location of samples collected, with (a) including red—QCI, yellow—JDFS, green—JS, dark blue—SOG, pink—W, orange—SV, and light blue—LA, and (b) the same except SV is shown in yellow and gold as opposed to orange to reflect the two distinct profiles at this site. Site abbreviations as in Fig. 1.

80

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

Table 2 Potential environmental and developmental cues triggering shifts in physiology at sites along the migration path shown with functional analysis of genes significantly differentially regulated between adjacent sites along the migration tract.

Abbreviations for locations are as follows: QCI—Queen Charlotte Islands, JDFS–—Juan de Fuca Strait, JS—Johnstone Strait, SOG—Strait of Georgia, W—Whonnock, SV—Savona, and LA—Lower Adams spawning grounds. Functionally annotated processes deemed significant based on the heat map results but not identified in DAVID are italicized. An indication of which cue was more likely the cause of each physiological shift was indicated in brackets (i.e. strong or weak) where triggers were potentially overlapping. The processes which occurred more than once were colour coded.

A second major spatial shift in physiology occurred at the spawning grounds, which was apparent both from the statistical analyses (over 700 genes differentially regulated between W and LA) and from the hierarchical clustering plot restricted to FW samples

(Fig. 3b). Salmon sampled in SV, a high water temperature region of the river approximately 80% of the distance from W and LA spawning grounds and 1.5 weeks later in the migration, displayed a bimodal pattern in their expression profile, with 50% of the fish clustering

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

intermediate between W and the LA spawning grounds (SV-A), and 50% clustering as an outlier group to all other FW samples (with the exception of one LA fish; SV-B). 3.2. Shifting biological processes 3.2.1. Identifying potential cues that trigger shifts in energy metabolism during migration Biological processes affected during migration between QCI and JDFS/JS included the down-regulation of actin filaments, muscle contraction and glycolysis, and up-regulation of proteolysis (particularly proteasomal) and protein biosynthesis as well as variability in the regulation of DNA-dependent transcription (Table 2 and Fig. 4). Given the common effects of these processes in both JDFS and JS, they were likely triggered by the early stages of starvation or cessation of growth. Salinity cues (FW influence in SOG, and to a lesser extent JS) may also account for many of these physiological responses, as most were more highly affected in JS, and their directional trajectories of expression continued through the SOG. The down-regulation of muscle contraction occurred at similar magnitudes in both JS and JDFS, indicative of a stronger influence of starvation than salinity. In contrast, although the up-regulation of protein biosynthesis, proteolysis and the ubiquitin cycle occurred in all marine sites south of QCI, these responses were more prevalent in JS and SOG indicative of a potentially stronger influence of salinity. Four up-regulated physiological processes, including oxidative phosphorylation, lysosomal proteolysis, glycolytic regulation (PFK) and protein folding, and the down-regulation of oxygen transport were

81

likely triggered by salinity or olfactory cues stemming from the FR, as these processes were most significantly affected when fish reached the SOG, although there was a slight response also noted in JS. Alternately, the down-regulation of the mitochondrial electron transport chain and up-regulation of nitric oxide biosynthesis may have been additionally cued by chemical differences in SW and FW as these pathways did not shift until fish reached the FW W site, the latter continuing on a downward trajectory throughout FW migration. Shifting biological processes associated with migration from W or SV to the spawning grounds (LA) included protein biosynthesis and muscle contraction, both down-regulated, and proteolysis (lysosomal and proteasomal) and oxidative phosphorylation, up-regulated (Table 2 and Fig. 4). Each of these processes was differentially regulated in a trajectory of change (Table 3). Additional up-regulated pathways exclusive to the spawning grounds included glycolytic regulation (PFK), calcium ion binding (particularly muscle relaxation– through parvalbumin), zinc binding (LIM), symporter activity and secretion (Golgi apparatus). ATPase activity was exclusively downregulated at the spawning site, where variability in the regulation of DNA-dependent transcription was also noted. These changes could have been influenced by maturation, advanced muscle degradation and/or senescence. There were two distinct transcriptional profiles at SV, one whereby fish clustered somewhat intermediate to fish sampled in W and the LA spawning grounds (SV-A), and the other reflecting a divergent and possibly disadvantageous physiological state (SV-B) (Fig. 3b). Over 1430 genes in total were differentially regulated in FW, as revealed by an ANOVA between W, SV, and LA, with SV-A and SV-B groups

Fig. 4. Heatmaps showing differentially regulated biological processes on a trajectory of change from QCI to the spawning grounds. A star is placed over the SOG site, in which the largest transcriptional shift occurred. Site abbreviations as in Fig. 1.

82

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

Table 3 Ranking of significant biological processes resolved using DAVID based on correlation analysis with distance to the spawning site and pairwise t-tests between adjacent sites along the migration tract taken by Lower Adams Fraser River sockeye salmon.

Abbreviations for locations (sites) as shown in Table 1.

(originally identified through hierarchical clustering) treated as separate samples. A large percentage of this FW signal emanated from genes differentiating SV-B from all other samples. Functional genomic analysis revealed a strong signal associated with extracellular matrix (important in wound healing), protein amino acid ADPribosylation (important in DNA repair and telomere maintenance, as well as bacterial toxin responses), defense/inflammatory/stress response, cell adhesion, and heme binding/iron ion binding/cytochrome P450 up-regulated in SV-B salmon relative to all other fish sampled in FW. This profile may reflect strong response to high water temperature stress. The remaining processes were down-regulated in SV-B fish, but none exclusively. Maturation associated processes including oxidative phosphorylation, proteasomal proteolysis, carbohydrate metabolism and cell death/apoptosis were up-regulated in LA and SV-A fish relative to SV-B and W, while cellular development process was down-regulated in all SV and LA fish relative to W. Alternately, LA and SV-B salmon down-regulated the processes of protein biosynthesis, protein folding/response to stress, TGF-beta signalling and response to DNA damage/MAPK signalling. The defense response emanating from SV-B salmon included the up-regulation of interferon interacting proteins – interferon-inducible protein Gig2, toll-like receptors 9, cytochrome P450 3A45, and serum

amyloid P (Fig. 5). Additional up-regulated defense genes included MHC class Ia and Ib, small inducible cytokine SCYA105, Ig mu chain C region membrane-bound form, macrophage colony-stimulating factor receptor, IgH-A, barrier-to-autointegration factor, TNF-alpha 2, perforin, T cell receptor beta, CD9 antigen, and alpha-2-macroglobulin. Up-regulated in the stress-response of SV-B fish were two general stress response genes, diamine acetyltransferase 1 and aldose reductase, and two genes involved in oxidative stress, sensor protein dcuS and hemopexin. Serine/threonine–protein kinase/endoribonuclease IRE1 precursor, which responds to unfolded proteins by initiating cell cycle arrest and apoptosis, was up-regulated in all SV fish. Importantly, heat shock proteins responding to unfolded proteins, including HSP90 alpha and beta, HSP-30, HSc70, and activator of 90 kDa heat shock protein ATPase homolog 1, while initially up-regulated upon entry into FW, were down-regulated in the highly responsive SV-B fish as well as at the spawning grounds. One HSP binding protein, phospholipid hydroperoxide glutathione peroxidase, was exclusively down-regulated in SV-B salmon, while another, FK506-binding protein 3, was exclusively up-regulated. Stress response genes up-regulated in SV-A salmon and the spawning grounds included those involved in oxidative stress (proteasome subunit beta type 5 precursor), osmotic stress (26 S proteasome non-ATPase regulatory subunit 12, also up in

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

Fig. 5. Heatmap showing differentially regulated genes at Savona involved in stress and defense. Site abbreviations as in Fig. 1.

83

84

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

W), and hypoxia (cytochrome c oxidase subunit 4 isoform 1, mitochondrial precursor). Importantly, F-actin capping protein subunit alpha-1 and 2, which is required for assembly or stability at high temperatures of the F1 sector of mitochondrial F1F0 ATP synthase species, was also strongly up-regulated in SV-A and the spawning grounds, but down-regulated in SV-B. Two genes associated with oxidative stress were exclusively up-regulated at the spawning grounds, glutathione peroxidase 1 and ADP/ATP translocase 2. Carbonic anhydrase, an important enzyme involved in blood and tissue pH homeostasis, was solely up-regulated in the SV-B group. 3.2.2. Trajectories of change Genes and pathways on a (directional) trajectory of change through migration may provide insight into the energetic needs and processes that salmon use to fuel migration and maturation processes. Linear regression analysis was used to identify the genes and pathways transcriptionally associated with distance to the spawning grounds, and elucidated 1366 genes correlated at a P b 10− 5, and 74 genes at P b 10− 25. Fifteen biological processes were on a strong directional trajectory of change correlating with distance to the spawning grounds, and these were ranked by statistical strength through the DAVID gene functional classification tool enrichment scores (Table 3). Most notably, all of the genes correlating with distance were also on the combined gene list from site to site pairwise t-tests. The five most strongly correlated pathways included glycolysis, muscle protein and daughter groups actin filament and calcium binding EF-hand, all down-regulated with decreasing distance (positively correlated), and proteolysis, up-regulated as fish neared spawning grounds. 4. Discussion This study is the first to define the transcriptional changes in white muscle tissue of wild salmon over long distance migration to their spawning grounds. During migration, salmon experience extreme changes in their environment (e.g. salinity, temperature, water chemistries, olfactory cues, flow) which when overlaid on programmed physiological and developmental shifts associated with starvation, maturation, and senescence can create a complex, interacting matrix of processes. In our analyses, we reasoned that the elucidation of locations along the migratory route in which biological processes begin to shift may point to potential environmental cues triggering specific physiological events. For instance, at the time of collection, sockeye salmon sampled in 2006 had full stomachs at QCI while those sampled at JDFS and JS had empty stomachs. Therefore, fish ceased feeding after passing through QCI. Oceanographic data indicated that JS also contained an unusually high level of FW influence in 2006, and was, in essence, a weak estuarine environment. Therefore, shifts occurring between QCI or JDFS and JS could equally be cued by the slightly reduced salinities in JS as well as from the early stages of starvation. Indeed, transcriptional data from gill tissue revealed a stronger FW osmotic signature in fish sampled from JS relative to JDFS (Miller, unpublished data). These processes should have been even more strongly affected, however, upon arrival to the SOG estuary, in which lower salinities and higher concentrations of olfactory cues stemming from the FR are expected. As W was the first fully FW site sampled, shifts exclusive to or starting in W may have been cued by the difference in water chemistry (ionic, pH, chemical contaminants) between SW and FW, as well as high concentrations of olfactory cues. SV was located in a high temperature region of the river. Radio-tracking and ibutton data revealed that during the time of sampling, salmon in SV would have been exposed to elevated water temperatures (N18 °C) for 5–7 days prior, but shortly after leaving SV, they would enter Kamloops Lake, which contains cooler temperatures at depth and is utilized by many fish as a thermal refuge during migration (D. Patterson; unpublished data). Hence, signals exclusive to SV may have been in response to chronic temperature stress. Finally, the spawning grounds generally contain high pheromone levels

produced by mature and spawning fish, which could trigger the final stages of maturation of fish upon arrival to this site. In addition, fish at the spawning grounds may be affected by limited energy reserves and senescence. 4.1. Starvation An early trigger in the coastal return migration that affected energy metabolism of white muscle was likely behaviourally induced when fish ceased feeding south of the QCI. Energy metabolism would have continued to be affected by prolonged starvation as fish became more anorexic along the migration path. If atrophy of white muscle during migration were driven simply by a response to starvation, the mobilization sequence of different energy sources should follow that previously identified in fish during the course of starvation. In teleosts, white muscle is the first tissue to respond to food deprivation, and does so through an immediate reduction in protein synthesis, followed approximately 1 week later by an increase in lysosomal proteolytic enzymes (Haschemeyer, 1983; Loughna and Goldspink, 1984). Lipid mobilization should occur during initial fasting, simultaneous to or following carbohydrate mobilization but always prior to protein degradation (Navarro and Guitierrez, 1995). Amino acids (breakdown products of muscle protein) are the major source of energy for fish during extended periods of starvation (Love, 1980; Walton and Cowey, 1982) and their oxidization in white muscle provides the bulk of energy used by red muscle for sustained swimming (Jurss and Bastrop, 1995). There were several deviations from this starvation model observed in the transcriptional profiles of white muscle from migrating sockeye salmon. First, protein biosynthesis was transcriptionally upregulated simultaneous to the up-regulation of proteolysis in the SOG, but followed a down-ward trajectory during the FW phase of migration to the spawning grounds. The initial up-regulation of protein biosynthesis may have been influenced by the need to maintain homeostasis under changing salinity regimes or by the extensive biosynthetic demands of the proteasomal machinery. The early use in starvation of carbohydrate and lipid stores should have been reflected by up-regulation of key pathways involved in glycogenolysis, glycolysis, lactic acid fermentation, TAG breakdown as well as evidence of mobilization of fatty acids to other tissues (e.g. red muscle and liver) to be utilized as fuel. However, although genes involved in glycogenolysis (glycogen catabolism) and lactic acid fermentation were indeed progressively down-regulated between JDFS/JS and the spawning grounds, they were not transiently upregulated between QCI and JS/JDFS in response to cessation of feeding. The observed glycolytic profile was similar to the continuous decline in the activity of white muscle metabolic enzymes (including HK, GAPDH, PK, LDH) observed in Mommsen et al. (1980) over the spawning migration of Stuart sockeye salmon, however their study did not contain a migration sample point prior to the cessation of feeding. Similarly, the lack of initial up-regulation early in starvation of lipid metabolism and mobilization genes [triacylglycerol catabolism, beta oxidation, fatty acid transport proteins and some apolipoproteins (A-IV and B)] was inconsistent with expectations of the early mobilization of lipid reserves, although many of these genes were progressively down-regulated during migration. Perhaps these results indicate that these processes were initiated prior to the QCI. In another earlier study of migrating, spawning Stuart sockeye salmon, Kiessling et al. (2004) reported an 85–90% reduction in the intraand extra-muscular adipose tissue fat depots from JDFS to the spawning grounds and found that the various lipid depots were utilized during different stages of the salmon migration. As in the study by Mommsen et al. (1980), they found that the extra-muscular depots (intestinal) were already quite depleted by JDFS, indicating that the fish may have already begun utilizing these lipid depots prior to starvation.

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

The early transcriptional down-regulation of muscle structural and contractile proteins occurring immediately after fish ceased feeding in the marine environment was most likely triggered as a means to conserve energy. White muscle accounts for more than half of the total fish mass and therefore suspension of its growth would eliminate the use of substantial amounts of energy and amino acids during starvation (Mommsen, 2004). Lowery and Somero (1990) observed a 90% decrease in actin synthesis following 23 days of starvation in sand bass (Paralabrax nebulifer). However, as these same proteins were transcriptionally down-regulated in fed maturing RBT (Salem et al., 2006a), the extra metabolic demands of spawning may also play a role this process. Lastly, although the down-regulation of oxygen transport was likely triggered by changes in salinity, there is some evidence that starvation may also trigger this process. Salem et al. (2007) found this same down-regulation in liver tissue of starved RBT. While our experiment was not designed to specifically tease out the effects of starvation versus those associated with reproductive maturation, the inconsistencies with expectations under a “starvation model” and the high degree of consistency between our findings and those of Salem et al. (2006a) on maturing but continuously fed RBT would suggest a strong maturation-based response. 4.2. Energy shifts: anaerobic to aerobic metabolism Glycolytic enzyme levels, as all proteins, can be affected by transcription, mRNA processing and export, translation and post-translational modification. The glycolytic enzymes in white muscle tissue should be expressed at a level that is adequate to meet the infrequent demands of high intensity activity. Our finding that glycolytic enzymes were transcriptionally down-regulated during muscle atrophy is well supported by previous studies on both mammals and fish (Batt et al., 2006; Salem et al., 2006a; Lecker et al., 2004), and it is possible that their down-regulation is a means to conserve energy by reducing transcription of proteins with long half-lives (Kuehl and Sumsion, 1970) to counterbalance the increased biosynthetic demand for other products. Given their long half lives, it is possible that the transcriptional levels of these genes do not appreciably reflect levels of circulating enzymes or their activity. However, Mommsen et al. (1980) also reported a steady decline in LDH, PK and HK enzyme levels in return migrating Stuart sockeye salmon over a range similar to that from the SOG to the spawning grounds in our study. Alternately, enzyme levels of GAPDH were unchanged in the Mommsen study, while changing profoundly at the transcriptional level in our study. Hence, it appears that at least some transcriptional changes in glycolytic enzymes do affect protein levels, while others may not. We surmise, however, that the massive down-regulation of most glycolytic enzymes does affect the activity of the glycolytic pathway. Salem et al. (2006a) proposed that decreased expression of glycolytic enzymes in atrophying white muscle of non-starving RBT may be due to insufficient levels of glycogen as a substrate for anaerobic respiration. On the surface, our data would seem to support this hypothesis. However, we suggest an alternate hypothesis, namely that the glycolytic pathway is conserved as energetic reserves diminish through migration to ensure adequate glycogen levels required to support spawning are maintained. Three lines of evidence lend support to this hypothesis. First, Idler and Clemens (1959) found that salmon maintained their carbohydrate stores throughout their spawning migration and French et al. (1983) found that the liver and white muscle glycogen levels of migrating sockeye salmon did not significantly change over migration except for an increase just prior to spawning. Second, Mommsen et al. (1980) found that carbohydrates were necessary for spawning in sockeye salmon and rebuilt from amino acids. Third, we observed a strong up-regulation at the spawning grounds of two key enzymes, PFK-I, the most influential regulator of the glycolytic pathway, and UDP-glucose pyrophosphorylase 1, involved in glycogen synthesis. Assuming that sufficient

85

protein levels of the transcriptionally down-regulated glycolytic enzymes are still available at the spawning grounds, PFK-I has the capacity to turn on the glycolytic pathway. Furthermore, we expect that lactate will accumulate in white muscle tissue and will require conversion to glycogen through the glycogen synthesis pathway. The sharp increase in the expression of the glycolytic regulatory enzyme (PFK-1) observed at the spawning site in our study was not observed in Salem's RBT study. Perhaps this contradiction points to a difference in fuel source and/or reserves at the final stage of spawning between semelparous (sockeye) and iteroparous (rainbow trout) salmon species. While there was a general down-regulation of glycolysis throughout migration, some copies of enzymes that link glycolysis with the TCA cycle, required to drive oxidative phosphorylation, were synchronously up-regulated with enzymes involved in oxidative phosphorylation. Pyruvate kinase is an important regulatory point that pulls the glycolysis pathway to completion, and a single copy of the M2 isoform was up-regulated, although a shorter, more 3' EST representing the same gene (contig) was down-regulated similar to other glycolytic enzymes. Whether these divergent results reflect alternate splicing or differences in specificity of the two EST fragments is not known. Pyruvate dehydrogenase links glycolysis directly to the TCA cycle, and both pyruvate dehydrogenase kinase isozyme 1 and pyruvate dehydrogenase E1 component beta subunit were upregulated during the FW phase of migration. Similarly, two enzymes involved in the TCA cycle, malate dehydrogenase and isocitrate dehydrogenase, were up-regulated in the FW phase of migration, although both were sharply down-regulated in some fish at the high temperature SV site. Oxidative phosphorylation is the terminal step in the aerobic respiration pathway and the main producer of energy (ATP). The expression profiles of the proteins involved in the electron transport chain and the ATP synthase followed the same trends as those shown for the TCA cycle except that there was a sharp up regulation at the spawning grounds. Taken together, these patterns of expression indicate a shift towards the more efficient aerobic pathway to generate energy that starts by the time fish reach the Fraser estuary. 4.3. Proteolytic pathways Three separate proteolytic systems which target different substrates are thought to cooperate in the controlled and selective breakdown of skeletal muscle in vertebrates: the ATP-dependent ubiquitin–proteasome complex (Lecker et al., 1999), the calpains, and the lysosomes (containing cathepsins and other acidic hydrolases; Cuervo and Dice, 1998) (Hershko et al., 2000). In mammals, the proteasomal system is considered the most prevalent in the normal turnover of muscle proteins as well as for degradation of proteins in muscle wasting (Lecker et al., 2004). Under a mammalian model, proteasomal proteolysis is up-regulated after 1–2 weeks of starvation (Haschemeyer, 1983; Loughna and Goldspink, 1984). However, in fish the proteasomal proteolytic pathway has been considered of relatively minor importance compared to lysosomal proteolysis through cathepsin enzymes (Mommsen, 2004). In support of this assertion, the proteasomal pathway was actually down-regulated in white muscle tissue in independent studies on starvation (Salem et al., 2007) and vitellogenesis (Salem et al., 2006a,b), while lysosomal proteolysis was up-regulated (ibid and in Mommsen et al., 1980; Yamashita and Konagaya, 1990; Toyohara et al., 1991). Our study is the first to highlight the potential functional importance of the proteasomal proteolytic pathway in salmon. From the SOG to the spawning grounds, a majority of the 163 significant proteasomal features (including ubiquitin, proteasome maturation factor, ubiquitin conjugating enzymes and ligases, ubiquitin carboxylterminal hydrolases, 20S proteasome alpha and beta subunits, 26S proteasome non-ATPase regulatory subunits and 26S proteasome

86

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

regulatory subunits) were up-regulated in white muscle tissue of migrating sockeye salmon. While the fold change averages were relatively small (1.78 for 26S and 2.12 for 20S from QCI to LA), the strength of the response in terms of numbers of differentially regulated genes was large and undeniable. Alternately, the average fold change of lysosomal pathway enzymes was on par to slightly higher (1.70 for cathepsin L; 3.23 for cathepsin D). The seemingly contradictory up-regulation of the proteasomal pathway in our study may be a response specific to anadromous salmon and the stresses related to this lifestyle, which would not be the case in RBT (both Salem studies). Moreover, this process may only be initiated in semelparous fish such as sockeye that will ultimately die post spawning (i.e. senescence related processes). Indeed, in humans, ubiquitination is one of many processes up-regulated in the process of aging and senescence (Welle et al., 2003). The findings of Toyohara et al. (1991) may lend credence to these theories as they also showed that the multi-catalytic proteinase (a proxy for proteasomal activity) was up-regulated 2-fold in ayu (Plecoglossus alivelis) males 2 months prior to and during the spawning event (recalculated data reported in Mommsen, 2004). Although these spawning osmeriformes are not starved as in sockeye salmon, they do exhibit reduced feeding behaviours prior to this event (Nagasaka et al., 2006). Importantly, they also exhibit anadromous and semelparous life history strategies. The decrease in proteasomal enzyme levels of starving rainbow trout observed in Martin et al. (2002) could be due to the proteasome itself becoming degraded by the lysosomal pathway (in late stage starvation) (Cuervo et al., 1995) rather than transcriptional down regulation, which, if true, would be consistent with our findings. Alternately, perhaps the transcriptional changes observed in the proteasome do not amount to translational shifts in abundance of proteasomal proteins. Salem et al. (2007) observed a slight but significant enhancement of 20S proteasome activity, but not mRNA levels, in the liver of starved RBT. Third, it is possible that the differences between Martin et al. and our study are due to the expanded scope provided by the use of microarrays. In Martin et al., two proteasomal genes (proteasome subunit N3 and polyubiquitin) were measured. In our study, the expression of the ubiquitin genes was quite variable and not typical of the majority of genes within the proteasomal pathway. In addition, 17% of the 20S proteasomal beta subunit genes showed expression patterns that would better match those seen by Martin and colleagues. Regardless, it is apparent that the up-regulation of this energetically expensive process must serve some important function in these fish. Interestingly, the other potentially aberrant finding in our study was the up-regulation of protein biosynthesis upon arrival to the estuary, which may have been required for the synthesis of the extensive array of proteasomal proteins required to activate this pathway. While further directed study will be required to tease these hypotheses apart, it is clear that up-regulation of the lysosomal proteolytic pathway is required for fuelling maturation and spawning, while proteasomal proteolysis may be required for energy in migrating, anadromous and semelparous salmon. The intracellular, pH neutral, calcium-dependent cysteine proteases (calpains) are known to play a role in the degradation of myofibrillar proteins for protein turnover associated with animal growth (Goll et al., 1992; Huang and Forsberg, 1998) as well as in the breakdown of vertebrate skeletal muscle due to exercise and/or altered nutritional states (Belcastro et al., 1996). These enzymes are involved in a more regulatory rather than digestive role due to their targeting of selected proteins [tropomyosin, troponins, desmin and titin, but not alpha-actin, alphaactinin or myosin heavy chains (Huang and Forsberg,1998)] during early stages of protein turnover (Goll et al., 1992). Mommsen (2004) speculated that fasting could initiate the calpains, as in mammals (Ilian and Forsberg, 1992) and found evidence of massive and early disappearance of Z-disks in white muscle tissue early in the migration (300 km from the ocean) of spawning sockeye salmon. In addition, Salem et al. (2005, 2006a, 2006b, 2007) found calpain genes were up-regulated in

starvation-induced atrophy in muscle and liver tissue but not in late stage spawning-induced muscle atrophy. Based on these studies, we anticipated up-regulation of the calpain system prior to the other two proteolytic pathways; however, the 5 significant calpain features were not induced until the SOG-FW transition, with a fold change from QCI to LA of only 0.56 for calpain subunit 1. 4.4. Salinity/olfactory cues White muscle underwent dramatic metabolic reconfigurations coincident with arrival of salmon to the estuarine environment in the SOG, whereby proteolysis, oxidative phosphorylation, protein biosynthesis and folding were up-regulated and cytoskeleton organization and biogenesis, muscle development, muscle contraction and relaxation pathways and oxygen transport were down-regulated. We hypothesize that the low salinity and enhanced olfactory cues present in the estuarine environment triggered the switch to white muscle tissue as a key source of energy for migration and reproductive loading, with pheromones at the spawning grounds triggering a second shift to further enhance reproductive loading. The estuarine shift in white muscle expression contrasted sharply with observations of other tissues (gill, whole brain, and hypothalamus) in which the SOG estuary appeared to be more of an admixture of physiologically distinct fish migrating from JS and JDFS (Miller, unpublished data). Adding to this is the fact that, in recent years, some late-run FR fish enter the river upon arrival while others hold in the SOG estuary for 1–2 weeks before entry (Lapointe et al., 2004; Cooke et al., 2004). Hence, the finding that individual variation in white muscle tissue metabolism actually diminished in the SOG was further evidence that the observed metabolic shifts were likely triggered environmentally as opposed to being part of a slow progression due to enhanced starvation. 4.5. Temperature Exposure to high water temperatures can have profound negative effects on the survival of sockeye salmon in the river (Crossin et al., 2008), and has been attributed as the cause of en route losses in the spawning migration of FR sockeye salmon in numerous reviews (Pearse and Larkin, 1992; Macdonald, 2000; Williams, 2005). Salmon exposed to water temperature increases above their optimal level (N16 °C) experience enhanced stress and parasite loads (Brett, 1965; Gilhousen, 1980; Macdonald, 2000; Wagner et al., 2005), which can affect disease progression (Sanders et al., 1978; Crossin et al., 2008), reproduction (Pankhurst et al., 1996) and swimming performance (Lee et al., 2003). During the 20th century, BC experienced an atmospheric warming of 0.8 °C (Mote et al., 2003), and in the past 50 years alone, the FR experienced a 2 °C rise in peak summer water temperature. As a result, late-run stocks of FR sockeye salmon that enter the river by early September experience river temperatures 5– 6 °C higher than historically due both to their earlier than normal entry timing and higher peak summer temperatures (Cooke et al., 2004; Lapointe et al., 2004). In our study, fish sampled at Savona on September 20th would have travelled through water temperatures N18 °C for at least 6 days based on radio-tagging data, hence any transcriptional response observed in our study would be chronic, not acute. The transcriptional activation of HSPs in poikilotherms is one of the main cellular biochemical responses upon exposure to an acute temperature elevation (Sonna et al., 2002). The increase of mRNA levels of both HSP 30 and 70 has been observed after acute temperature shifts in goldfish, where HSP 70 was induced after a 10–15 °C temperature rise and HSP 30 was induced in shifts greater than 15 °C (Kondo et al., 2004). In our study, HSPs c70, 30, and 90 alpha and beta were up-regulated as fish entered FW, where fish would have experienced elevated temperatures as well as changes in salinity and

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

water chemistry. However, these chaperones were subsequently down-regulated under chronic high water temperature exposure at SV, most strongly in SV-B fish, and even further down-regulated at the spawning grounds. The SV-B data contradicts the pattern observed in killifish (Astrofundulus limnaeus) whereby HSPc70 and HSP90 continued to be highly expressed under chronic exposure (Podrabsky and Somero, 2004). The bimodal pattern of expression of fish sampled at SV suggests the potential for differences in adaptive capacity of fish from the Lower Adams stock complex. Moreover, the strong stress and defense signalling in the high responder group, SV-B, may have resulted from a greater influence of disease in these fish, or may have simply been a maladaptive response to high water temperature stress. Whereas 50% of fish sampled in SV carried the high-responder SV-B signature, only 12.5% of the fish at the spawning grounds carried a similar signature, potentially suggesting a higher mortality rate in the high-responder fish. 4.6. Maturation We hypothesize that olfactory cues and/or low salinity in the SOG estuary trigger enhanced reproductive maturation of FR salmon by turning on processes that provide both the raw materials (amino acids, lipids) and energy to fuel the maturation process. We term this the “energetic maturation hypothesis”. Further support to this hypothesis could be gained by physiological assessments of brain, liver and gonadal tissues to determine whether shifts more directly associated with maturation (e.g. hormonal, ovarian developmental) occur coincidentally in the SOG in these tissues, data our lab is in the process of obtaining. Furthermore, correlation analyses with physiological measures of maturation (e.g. hormone levels, egg mass size) could potentially provide greater resolution between the energetic processes associated with starvation versus reproductive loading. Physiological shifts associated with arrival to the spawning grounds may have been influenced by maturation, advanced muscle degradation and/or senescence. GO processes exclusive to the spawning grounds included the up-regulation of calcium ion binding, muscle relaxation (parvalbumin), zinc binding (LIM), symporter activity and secretion and the down-regulation of ATPase activity. The up-regulation of the secretion pathway in the Golgi apparatus may signal the release of the final post-translationally modified proteins necessary for reproduction, and the up- and down-regulation of the symport activity of various solute carrier family genes may allow the final transport of goods necessary to fuel reproduction. Alternately, the fact that none of the up-regulated LIM, zinc binding proteins involved in transcription, cytoskeleton organization and development was identified in the Salem et al. (2006a) study indicates a potentially greater role of these processes in muscle senescence than in the maturation process. Indeed, there is some precedence for the association between LIM proteins and senescence in humans (Shibanuma et al., 1997). Three of the LIM proteins, four and a half LIM domains protein 1, LIM domain-binding protein 3, and PDZ and LIM domain protein 1, were highly up-regulated at the spawning site, the latter of which was also associated with morbidity in gill tissue (Jeffries and Miller, unpublished data). 4.7. Exploring mammalian models of senescence and aging In the animal kingdom, semelparous salmon offer an extreme example of rapid senescence, whereby a large reproductive investment is followed by senescence within days to weeks after spawning (Mcphee and Quinn, 1998). At the spawning grounds, we collected fish that had arrived but not yet spawned; hence these fish were, for the most part, not in late-stage senescence but rather were presenescent. When we compare the expression profile of these presenescent fish with that expected under an aging model in mammals,

87

there are a number of notable contrasts. In mammals, the accumulation of oxidative damage causing mitochondrial dysfunction is associated with the aging process in many tissues (including muscle), limiting ATP synthesis and mitochondrial respiration (Shigenaga et al., 1994; Welle et al., 2003). In salmon, ATP synthesis through oxidative phosphorylation was up-regulated at the spawning site, indicating high functionality of the mitochondria at a pre-spawning state. A decline of antioxidant activity is also an important component of the aging process in mammals (Finkel and Holbrook, 2000). However, two active antioxidants, peroxiredoxin and thioredoxin-dependent peroxide reductase, were transcriptionally up-regulated at the spawning grounds, whereas selenoprotein Pa precursor was down-regulated. Finally, up-regulation of creatine kinase, a critical target for ROSinduced inactivation, is associated with the aging process in mammals (Lee et al., 1999), but was down-regulated with migration in white muscle tissue in salmon. Whether these processes diminish after spawning as fish approach a more senescent state is unknown at this time. A more specific comparison between the aging of mammalian and salmon muscle tissue suggests greater consistency among species, despite the above discrepancies. In human muscle tissue from the vastus lateralis, there was a slight reduction in the transcription of myosin heavy chain mRNA in older men, and a strong up-regulation in metallothionein, high-mobility-group (HMG), and heterogeneous nuclear ribonucleoprotein mRNAs (Welle et al., 2003). These results were consistent with the gradual down-regulation of myosin heavy chain and up-regulation of HMG (B1, B3, 2 and 14), and strong upregulation at the spawning grounds of metallothionein (A, B, and I) and hnRNP A0 observed in salmon. As well, there is some evidence that migrating salmon may use behavioural and molecular mechanisms to retard senescence until after spawning. In mammals, caloric restriction is the only behavioural intervention known to slow the intrinsic rate of aging (Weindruch and Walford, 1988). Given that salmon have physiologically evolved to sustain enhanced periods of starvation through the development of large energy stores in the white muscle tissue, perhaps the cessation of feeding serves a similar purpose as that of caloric restriction in mammals, at least for the period until energy reserves are sufficiently diminished. In aging mammals, the mechanistic link between caloric restriction and aging in muscle is hypothesized to be the metabolic shift toward increased protein turnover and decreased macromolecular damage (Lee et al., 1999). Both of these processes were affected in migrating salmon after the cessation of feeding, and coincident with the arrival to the FR. Over expression of the oxygen radical metabolizing enzyme Cu/Zn superoxide dismutase is an additional molecular mechanism that can retard age-related senescence (Orr and Sohal, 1994); this gene was mildly upregulated in the white muscle of salmon entering FW. Finally, stress resistance is associated with longevity; the 50% of fish responding only weakly to the high water temperatures at SV experienced a higher rate of survival to the spawning grounds than those reacting strongly. 5. Summary In summary, this research has provided novel insight into the environmental cues triggering shifts in energy metabolism in migrating adult salmon, and has identified a more substantial role of the proteasome than previously documented in fish. We identified the reduced salinity and enhanced olfactory stimulation in the estuarine environment as a major trigger for enhanced reproductive loading in anadromous migrating sockeye salmon. We identified a highly divergent response to high water temperature stress among fish within a single stock, suggesting that at least some fish within the stock carry the adaptive capacity to withstand temperatures 3–5 °C higher than their historic norms. Finally, we identified the final stages of energy fueling for the maturation process at the spawning grounds, which

88

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89

included enhanced stimulation of ATP-generating processes (oxidative phosphorylation and glycolytic potential) and provision of amino acids potentially to support biosynthetic processes in other tissues (proteolysis). The down-regulation of the protein biosynthetic machinery and helicase activity at the spawning grounds were likely energy saving mechanisms, but could also relate to enhanced senescence signatures. A detailed description of senescence related processes in mammals and salmon, and the potential for delayed senescene through caloric restriction, was also presented. Many specific findings of our study are well corroborated by smaller-scale studies and synthetic reviews by Mommsen et al. (1980), Mommsen (2004) and Salem et al. (2006a, b). In fact, the general strength and validity of the microarray approach is exemplified by the striking similarity between the profiles associated with maturation in wild sockeye salmon here and those elucidated earlier for captive RBT. Both studies employed the same array, with Salem et al. (2008) subsequently using an oligonucleotide array to corroborate most of their findings. Furthermore, when we compared our results with Mommsen's (2004) proteomic analyses of migrating sockeye salmon from the Stuart system, the results are again quite similar. In all, these data imply that large-scale functional genomic studies on wild caught organisms can be used to gain considerable insight and depth into the influences of environment variation on their physiology. Acknowledgements This study was part of a large collaborative multidisciplinary research program aimed at identifying the causes and consequences of early FW entry timing of late-run FR sockeye salmon funded through contributions from the Pacific Salmon Commission Southern Endowment Fund, the Canadian Department of Fisheries and Oceans Genomic Research and Development Fund and Environmental Watch Program, and the Natural Sciences and Engineering Research Council (NSERC) of Canada. We thank Karl English, Dave Robichaud, Jayme Hills, Ivan Olsson, Kyle Hanson, Lucas Pon, and Glenn Crossin for field and lab assistance, Liane Stenhouse and Kathryn Horst for laboratory assistance, and Colin Wallace for assistance with data analysis. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.cbd.2008.11.002. References Batt, J., Bain, J., Goncalves, J., Michalski, B., Plant, P., Fahnestock, M., Woodgett, J., 2006. Differential gene expression profiling of short and long term denervated muscle. FASEB J. 20, 115–117. Bax, N.J., 1983. Early marine mortality of marked juvenile chum salmon released into Hood Canal, Puget Sound, Washington in 1980. Can. J. Fish. Aquat. Sci. 40, 426–435. Beacham, T.D., Candy, J.R., McIntosh, B., MacConnachie, C., Tabata, A., Kaukinen, K., Deng, L., Miller, K.M., Withler, R.E., Varnavskaya, N.V., 2005. Estimation of stock composition and individual identification of sockeye salmon on a Pacific Rim basis using microsatellite and major histocompatibility complex variation. Trans. Am. Fish. Soc. 134, 1124–1146. Beamish, R.J., Mahnken, C., 2001. A critical size and period hypothesis to explain natural regulation of salmon abundance and the linkage to climate and climate change. Prog. Oceanogr. 49, 423–437. Belcastro, A.N., Albisser, T.A., Littlejohn, B., 1996. Role of calcium-activated neutral protease (calpain) with diet and exercise. Can. J. Appl. Biol. 21, 328–346. Brett, J.R., 1965. The swimming energetics of salmon. Sci. Am. 213, 80–85. Burgner, R.L., 1991. Life history of sockeye salmon (Oncorhynchus nerka). In: Groot, C., Margolis, L. (Eds.), Pacific Salmon Life Histories. University of British Columbia Press, Vancouver, B.C., pp. 3–117. Cooke, S.J., Hinch, S.G., Farrell, A.P., Lapointe, M., Healey, M.C., Patterson, D., Macdonald, S., Jones, S.R.M., Van Der Kraak, G., 2004. Early-migration and abnormal mortality of laterun sockeye salmon in the Fraser River, British Columbia. Fisheries 29, 22–33. Cooke, S.J., Hinch, S.G., Crossin, G.T., Patterson, D.A., English, K.K., Healey, M.C., Macdonald, J.S., Shrimpton, J.M., Young, J.L., Lister, A., Van Der Kraak, G., Farrell, A.P., 2008. Physiological correlates of coastal arrival and river entry timing in Late Summer Fraser River sockeye salmon (Oncorhynchus nerka). Behav. Ecol. 19, 747–758.

Crossin, G.T., Hinch, S.G., Cooke, S.J., Welch, D.W., Lotto, A.G., Patterson, D.A., Jones, S.R.M., Leggatt, R.A., Mathes, M.T., Shrimpton, J.M., Van Der Kraak, G., Farrell, A.P., 2008. Exposure to high temperature influences the behaviour, physiology, and survival of sockeye salmon during spawning migration. Can. J. Zool. 86, 127–140. Cuervo, A.M., Dice, J.F., 1998. Lysosomes, a meeting point of proteins, chaperones, and proteases. J. Mol. Med. 76, 6–12. Cuervo, A.M., Palmer, A., Rivett, A.J., Knecht, E., 1995. Degradation of proteosomes by lysosomes in rat liver. Eur. J. Biochem. 227, 792–800. Donaldson, E.M., Smith, J., Barnes, D., Clarke, W.C., Gordon, R., Martens, D., 2000. Physiological and endocrine changes during the anadromous migration of early Stuart sockeye salmon (Oncorhynchus nerka) in 1997. In: Macdonald (Ed.), Canadian Technical Report of Fisheries and Aquatic Sciences, vol. 2315, pp. 67–73. English, K.K., Koski, W.R., Sliwinski, C., Blakley, A., Cass, A., Woodey, J.C., 2005. Migration timing and river survival of late-run Fraser River sockeye salmon estimated using radiotelemetry techniques. Trans. Am. Fish Soc. 134, 1342–1365. Finkel, T., Holbrook, N.J., 2000. Oxidants, oxidative stress and the biology of ageing. Nature 408, 239–247. French, C.J., Hochachka, P.W., Mommsen, T.P., 1983. Metabolic organization of liver during spawning migrations of sockeye salmon. Am. J. Physiol. 245, R827–R830. Gilhousen, P., 1980. Energy sources and expenditures in Fraser River sockeye salmon during their spawning migration. Int. Pac.Salmon Fish. Comm. Bull. No. 22. Goll, D.E., Thompson, V.F., Taylor, R.G., Christiansen, J.A., 1992. Role of the calpain system in muscle growth. Biochimie 74, 225–237. Groot, C., Margolis, L., 1991. Pacific Salmon Life Histories. University of British Columbia Press, Vancouver. British Columbia. Haschemeyer, A.E.V., 1983. A comparative study of protein synthesis in Nototheniids and icefish at Palmer station, Antarctica. Comp. Biochem. Physiol. 76B, 541–543. Hershko, A., Ciechanover, A., Varshavsky, A., 2000. The ubiquitin system. Nat. Med. 6, 1073–1081. Hinch, S.G., Cooke, S.J., Healey, M.C., Farrell, A.P., 2006. Behaviour and physiology of fish. In: Sloman, K., Balshine, S., Wilson, R. (Eds.), Fish Physiology, vol. 24. Elsevier, pp. 239–295. Huang, J., Forsberg, N.E., 1998. Role of the calpain in skeletal-muscle protein degradation. Proc. Natl. Acad. Sci. U.S.A. 95, 12100–12105. Huang, D.W., Sherman, B.T., Tan, Q., Collins, J.R., Alvord, W.G., Roayaei, J., Stephens, R., Baseler, M.W., Lane, H.C., Lempicki, R.A., 2007. The DAVID gene functional classification tool: a novel biological module-centric algorithm to functionally analyze large gene lists. Genome Biol. 8, R183. Idler, D.R., Clemens, W.A., 1959. The energy expenditures of Fraser River sockeye salmon during the spawning migration to Chilko and Stuart Lakes. Int. Pacific Salmon Fish. Comm. Prog. Rept. No. 6 (80 pp.). Ilian, M.A., Forsberg, N.E., 1992. Gene expression of calpains and their specific endogenous inhibitor, calpastatin, in skeletal muscle of fed and fasted rabbits. Biochem. J. 287 (Pt 1), 163–171. Jurss, K., Bastrop, R., 1995. Amino acid metabolism in fish. In: Hochachka, P.W., Mommsen, T.P. (Eds.), Biochemistry and Molecular Biology of Fishes, vol. 4. Elsevier, Amsterdam, pp. 159–189. Kiessling, A., Lindahl-Kiessling, K., Kiessling, K.H., 2004. Energy utilization and metabolism in spawning migrating Early Stuart sockeye salmon (Oncorhynchus nerka): the migratory paradox. Can. J. Fish. Aquat. Sci. 61, 452–465. Kondo, H., Harano, R., Nakaya, M., Watabe, S., 2004. Characterization of goldfish heat shock protein-30 induced upon severe heat shock in cultured cells. Cell Stress Chaperones 9 (4), 350–368. Kuehl, L., Sumsion, E.N., 1970. Turnover of several glycolytic enzymes in rat liver. J. Biol. Chem. 245 (24), 6616–6623. Lapointe, M.F., Cooke, S.J., Hinch, S.G., Farrell, A.P., Jones, S., Macdonald, S., Patterson, D., Healey, M.C., Van Der Kraak, G., 2004. Late-run sockeye salmon in the Fraser River, British Columbia are experiencing early upstream migration and unusually high rates of mortality: what is going on? In: Vancouver, B.C., Droscher, T.W., Fraser, D.A. (Eds.), Proceedings of the 2003 Georgia Basin/Puget Sound Research Conference, 31 March – 3 April 2003. Puget Sound Action Team, Olympia, Wash, pp. 1–14. Lee, C.-K., Klopp, R.G., Weindruch, R., Prolla, T.A., 1999. Gene expression profile of aging and its retardation by caloric restriction. Science 285, 1390–1393. Lee, C.G., Farrell, A.P., Lotto, A., MacNutt, M.J., Hinch, S.G., Healey, M.C., 2003. The effect of temperature on swimming performance and oxygen consumption in adult sockeye (Oncorhynchus nerka) and coho (O. kisutch) salmon stocks. J. Exp. Biol. 206, 3239–3251. Lecker, S.H., Solomon, V., Mitch, W.E., Goldberg, A.L., 1999. Muscle protein breakdown and the critical role of the ubiquitin–proteosome pathway in normal and disease states. J. Nutr. 129, S227–237. Lecker, S.H., Jagoe, R.T., Gilbert, A., Gomes, M., Baracos, V., Bailey, J., Price, S.R., Mitch, W.E., Goldberg, A.L., 2004. Multiple types of skeletal muscle atrophy involve a common program of changes in gene expression. FASEB J. 18, 39–51. Loughna, P.T., Goldspink, G., 1984. The effects of starvation upon protein turnover in red and white myotomal muscle of rainbow trout, Salmo gairdneri. J. Fish. Biol. 25, 223–230. Love, R.M., 1980. The Chemical Biology of Fishes, vol. 2. Academic Press, New York. Lowery, M.S., Somero, G.N., 1990. Starvation effects on protein synthesis in red and white muscle of the barred sand bass, Paralabrax nebulifer. Physiol. Zool. 63, 630–648. Macdonald, J.S., 2000. Mortality during the migration of Fraser River sockeye salmon (Oncorhynchus nerka): a study of the effect of ocean and river environmental conditions in 1997. Can. Tech Rep. Fish. Aquat. Sci. 2315, 39–57. Magnoni, L.J., Patterson, D.A., Farrell, A.P., Weber, J.-M., 2006. Effects of long-distance migration on the circulating lipids of sockeye salmon (Oncorhynchus nerka). Can. J. Fish. Aquat. Sci. 63, 1822–1829.

K.M. Miller et al. / Comparative Biochemistry and Physiology, Part D 4 (2009) 75–89 Martin, S.A., Blancy, S., Bowman, A.S., Houlihan, D.F., 2002. Ubiquitin–proteasome-dependent proteolysis in rainbow trout (Oncohorhynchus mykiss): effect of food deprivation. Plungers Arch. 445, 257–266. McPhee, M.V., Quinn, T.P., 1998. Factors affecting the duration of nest defense and reproductive lifespan of female sockeye salmon, Oncorhynchus nerka. Environ. Biol. Fishes 9, 211–268. Mommsen, T.P., 2004. Salmon spawning migration and muscle protein metabolism: the August Krogh principle at work. Rev. Comp. Biochem. Physiol. B Biochem. Molec. Biol. 39, 383–400. Mommsen, T.P., French, C.J., Hochachka, P.W., 1980. Sites and patterns of protein and amino acid utilization during the spawning migration of salmon. Can. J. Zool. 785–1799. Mote, P.W., Parson, E.A., Hamlet, A.F., et al., 2003. Preparing for climatic change: the water, salmon, and forests of the Pacific Northwest. Clim. Change 61, 45–88. Nagasaka, R., Okamoto, N., Ushio, H., 2006. Increased leptin may be involved in the short life span of ayu (Plecoglossus altivelis). J. Exp. Zool. 305A, 507–512. Navarro, I., Guitierrez, J., 1995. Fasting and starvation. In: Hochachka, P.W., Mommsen, T.P. (Eds.), Biochemistry and Molecular Biology of Fishes, vol. 4. Elsevier, Amsterdam, pp. 393–434. Orr, W.C., Sohal, R.S., 1994. Extension of life-span by overexpression of superoxide dismutase and catalase in Drosophila melanogaster. Science 263, 1128–1130. Pankhurst, N., Purser, G., Van der Kraak, G., Thomas, P., Forteath, G., 1996. Effect of holding temperature on ovulation, egg fertility, plasma levels of reproductive hormones and in vitro ovarian steroidogenesis in the rainbow trout, Oncorhynchus mykiss. Aquaculture 146, 277–290. Parker, R.R., 1968. Marine mortality schedules of pink salmon of the Bella Coola River, Central British Columbia. J. Fish Res. Board Can. 25, 757–794. Pearse, P.H., Larkin, P.A., 1992. Managing salmon in the Fraser. Vancouver, B.C., Department of Fisheries and Oceans. Podrabsky, J.E., Somero, G.N., 2004. Changes in gene expression associated with acclimation to constant temperatures and fluctuating daily temperatures in an annual killifish Austrofundulus limnaeus. J. Exp. Biol. 207, 2237–2254. Salem, M., Nath, J., Rexroad, C.E., Killefer, J., Yao, J., 2005. Identification and molecular characterization of the rainbow trout calpains (Calpn1 and Capn2): their expression in muscle wasting during starvation. Comp. Biochem. Physiol. B 140, 63–71. Salem, M., Kenney, P.B., Rexroad, C.E., Yao, J., 2006a. Microarray gene expression analysis in atrophying rainbow trout muscle: a unique nonmammalian muscle degradation model. Physiol. Genomics 28, 33–45. Salem, M., Kenney, P.B., Rexroad, C.E., Yao, J., 2006b. Molecular characterization of muscle atrophy and proteolysis associated with spawning in rainbow trout. Comp. Biochem. Physiol. D 1, 227–237. Salem, M., Silverstein, J., Rexroad, C.E., Yao, J., 2007. Effect of starvation on global gene expression and proteolysis in rainbow trout (Oncorhynchus mykiss). BMC Genomics 8, 328. Salem, M., Kenney, P.B., Rexroad, C.E., Yao, J., 2008. Development of a 37 k high-density oligonucleotide microarray: a new tool for functional genome research in rainbow trout. J. Fish Biol. 72, 2187–2206.

89

Sanders, J.E., Pilcher, K.S., Fryer, J.L., 1978. Relation of water temperature to bacterial kidney disease in coho salmon (Oncorhynchus kisutch), sockeye salmon (Oncorhynchus nerka), and steelhead trout (Salmo gairdneri). J. Fish. Res. Board Can. 35, 8–11. Shibanuma, M., Mochizuki, E., Maniwa, R., Mashimo, J., Nishiya, N., Imai, S., Takano, T., Oshimura, M., Nose, K., 1997. Induction of senescence-like phenotypes by forced expression of hic-5, which encodes a novel LIM motif protein, in immortalized human fibroblasts. Mol. Cell. Biol. 17, 1224–1235. Shigenaga, M.K., Hagen, T.M., B.N., 1994. Oxidative damage and mitochondrial decay in aging. Proc. Natl. Acad. Sci USA 91, 10771–10778. Sonna, L.A., Gaffin, S.L., Pratt, R.E., Cullivan, M.L., Angel, K.C., Lilly, C.M., 2002. Effect of acute heat shock on gene expression by human peripheral blood mononuclear cells. J. Appl. Physiol. 92, 2208–2220. Toyohara, H., Ito, K., Ando, M., Kinoshita, M., Shimizu, Y., Sakaguchi, M., 1991. Effect of maturation on activities of various proteases and protease inhibitors in the muscle of ayu (Plecoglossus altivelis). Comp. Biochem. B 99, 419–424. Van den Thillart, G., van Raaij, M., 1995. Endogenous fuels: noninvasive versus invasive approaches. In: Hochachka, P.W., Mommsen, T.P. (Eds.), Biochemistry and Molecular Biology of Fishes, vol. 4. Elsevier, Amsterdam, pp. 33–63. von Schalburg, K.R., Rise, M.L., Cooper, G.A., Brown, G.D., Gibbs, R.A., Nelson, C.C., Davidson, W.S., Koop, B.F., 2005. Fish and chips: various methodologies demonstrate utility of a 16,006-gene salmonid microarray. BMC Genomics 6 (1), 126. Walton, M.J., Cowey, C.B., 1982. Aspects of intermediary metabolism in salmonid fish. Comp. Biochem. Physiol. B 73, 59–79. Wagner, G.N., Hinch, S.G., Kuchel, L.J., Lotto, A., Jones, S.R.M., Patterson, D.A., Macdonald, J.S., Van Der Kraak, G., Shrimpton, M., English, K.K., Larsson, S., Cooke, S.J., Healey, M.C., Farrell, A.P., 2005. Metabolic rates and swimming performance of adult Fraser River sockeye salmon (Oncorhynchus nerka) after a controlled infection with Parvicapsula minibicornis. Can. J. Fish. Aquat. Sci. 62, 2124–2133. Weindruch, R., Walford, R.L., 1988. The Retardation of Aging and Disease by Dietary Restriction. Thomas, Springfield, IL. 1988. Welle, S., Brooks, A.I., Delehanty, J.M., Needler, N., Thornton, C.A., 2003. Gene expression profile of aging in human muscle. Physiol. Genomics 13, 149–159. Wertheimer, A.C., Thrower, F.P., 2007. Mortality rates of chum salmon during their initial marine residency. In: Grimes, C.B., Brodeur, R.D., Haldorson, L.J., McKinnell, S.M. (Eds.), Ecology of Juvenile Salmon in the Northeast Pacific Ocean: Regional Comparisons. Am. Fish. Soc. Symp., vol. 57. Bethesda, Maryland, pp. 233–247. Williams, B., 2005. 2004 Southern salmon fishery post-season review. Part one: Fraser River sockeye report. Report for Canadian Minister of Fisheries and Oceans, pp. 1–91 (http://wwwcomm.pac.dfompo.gc.ca/publications/2004psr/williams_e.pdf). Yamashita, M., Konagaya, S., 1990. High activities of cathepsins B, D, H and L in the white muscle of chum salmon in spawning migration. Comp. Biochem. Physiol. B 95, 149–152. Yang, Y.H., Dudoit, S., Luu, P., Lin, D.M., Peng, V., Ngai, J., Speed, T.P., 2002. Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res. 30, e15-e15.