Diel metabolic patterns revealed by in situ transcriptome and proteome in a vertically migratory copepod
Amy E. Maas1, Emma Timmins-Schiffman2, Ann M. Tarrant3, Brook L. Nunn2, Jea Park2, Leocadio Blanco-Bercial1
1 Bermuda Institute of Ocean Sciences, School of Ocean Futures, Arizona State University, St. George’s, Bermuda
2 Department of Genome Sciences, University of Washington, Seattle, WA, USA
3 Biology Department, Woods Hole Oceanographic Institution, Woods Hole, MA, USA
Abstract
Zooplankton undergo a diel vertical migration which exposes them to gradients of light, temperature, oxygen and food availability on a predictable daily schedule. Anticipating and responding to these environmental conditions, which independently are known to influence metabolic rates, likely has an appreciable effect on the delivery of metabolic waste products to the distinctly different daytime (deep) and nighttime (surface) habitats. Disentangling the co-varying and potentially synergistic interactions on metabolic rates has proven difficult, despite the importance of this migration to oceanic biogeochemical cycling. This study examines the transcriptomic and proteomic profiles of the circumglobal migratory copepod,Pleuromamma xiphias, over the diel cycle. The transcriptome showed a large number of up-regulated genes during the middle of the day – the period often considered to be of lowest zooplankton activity. During the day (9:00 and 15:00) there were patterns of increased chitin synthesis and degradation in both the transcriptome and proteome. At 09:00 and 22:00 there were increases in myosin and vitellogenin proteins, which may relate to the stress of migration and/or reproductive investment. There is an inferred switch, based on protein expression, in broad metabolic processes, shifting from electron transport system in the day, to glycolysis and glycogen mobilization in the afternoon and evening. These observations provide evidence of the diel impact of DVM on transcriptomic and proteomic pathways that likely influence metabolic processes and subsequent excretion products, and clarify how this behavior results in the direct rapid transport of active metabolisms that are motored from the surface to the deep ocean.
Introduction
In the ocean, changes in light level are a cue for one of the largest migrations on earth (Forward 1988; Hays 2003; Longhurst 1967). As the sun rises, millions of zooplankton migrate to deeper water, away from the light, reducing the risk of visual predation. This behavior, which varies across regions and among taxa, is thought to be driven by trade-offs between simultaneously increased feeding opportunities and predation risk at higher light levels (Antezana 2009; Benoit‐Bird & Moline 2021; Pinti et al. 2019). During their descent to their daytime habitat, migrators encounter increasingly deoxygenated, colder waters, with lower food availability; phytoplankton are absent from the deep ocean and both particulate organic carbon and total organismal biomass tends to drop exponentially with depth (Buesseler et al.2007; Hernández-León et al. 2020). Each of these factors independently affects zooplankton metabolism, with extensive literature devoted to analyses of critical oxygen partial pressure, temperature coefficients (Q10), and specific dynamic action (increase in metabolic rate after consumption of food; Hochachka & Somero 2002; Ikeda 2014; Kiørboe et al. 1985; Seibel et al. 2021; Svetlichny & Hubareva 2005; Thor 2000). Recent efforts have additionally demonstrated diel rhythms in metabolic enzyme activity and oxygen consumption, both in controlled laboratory conditions where extrinsic environmental cues including food and light were manipulated (Häfker et al. 2017; Maas et al. 2018; Piccolin et al. 2020; Teschke et al. 2011), and in wild-caught organisms captured across diel cycles (De Pitta et al. 2013; Tarrantet al. 2021).
The zooplankton migration supplies an estimated 15-40% of the global flux of carbon into the midwater region (Archibald et al. 2019; Bianchi et al. 2013a; Steinberg et al. 2000), and a similar percentage of nitrogen (Al-Mutairi & Landry 2001; Bianchiet al. 2014; Schnetzer & Steinberg 2002), via the “active flux” pathway of the biological pump. Calculations of these fluxes are based on application of physiological rate measurements from surface caught organisms to an estimated migratory biomass and typically are exclusively corrected for temperature at depth (e.g. Kwong et al.2020; Maas et al. 2021; Steinberg et al. 2008), although some studies are now starting to include the effects of variations in oxygen level (e.g. Kiko & Hauss 2019). Experiments developed to estimate metabolic rates are typically conducted with surface-caught individuals under surface (oxygenated and relatively warm) or deep (cold, sometimes deoxygenated) conditions (see metanalysis of Ikeda 2014); possible daily cycles in these rates associated with migratory effort, food availability, or circadian regulation are usually ignored in biogeochemical modeling (Archibald et al. 2019; Burd et al. 2010). The current assumption, however, is that an underlying circadian clock and the local environmental conditions together influence migratory behavior (Bianchi et al. 2013b; Häfkeret al. 2022; Pinti et al. 2019). In addition, circadian regulation of metabolism, which acts independently of environmental changes in pressure, temperature, oxygen and food availability, appears to moderate the physiology of zooplankton across their vertical migration (Häfker et al. 2017; Maas et al. 2018; Teschkeet al. 2011). In light of all of these co-varying parameters, predicting metabolic patterns of migrators over the course of a day is profoundly difficult.
Despite the difficulties, more precisely estimating the timing of the release of carbon and nitrogen across the diel cycle and across depths has clear biogeochemical relevance. The nighttime surface metabolic rate of zooplankton influences the recycling of nutrients for photosynthesis in the euphotic zone, while metabolic waste production at depth may provide a substantive portion of the labile nutrients available in the midwater (Archibald et al. 2019; Kelly et al. 2019; Steinberg & Landry 2017). The dissolved excreta of migratory copepods has been shown to include labile highly nutritive components, including vitamins (Maas et al. 2020; Shoemaker et al. 2020; Valdéset al. 2017), which would have distinctly different ecological impacts if produced at depth versus being recycled in the euphotic zone. Variations in the types of metabolic waste, or temporal pulses in their production, would thus have profound implications for the local ecosystems across the depth range that migratory zooplankton inhabit. The location of nitrogenous waste production is of particular relevance, as it delivers nitrogen for recycling back into primary production in surface waters, or to support alternate metabolic pathways, such as anammox – the anaerobic oxidation of ammonium, in the midwater (Bianchiet al. 2014; Bronk & Steinberg 2008; Valdés et al. 2018).
Broadly, understanding of the physiological underpinnings and consequences of DVM on zooplankton metabolism still lags far behind similar observations of migration in terrestrial systems, where there are known consequences for blood-oxygen binding, oxidative stress, metabolic substrate use, overall metabolism and broad patterns of gene expression (Doyle et al. 2022; Gerson et al. 2020; Gutiérrez et al. 2019; Olsen et al. 2021). Work in the Sargasso Sea with the copepod Pleuromamma xiphias has shown that this abundant migratory species, which is found at 50-200 m depth during the night and 400-700 m during the day, exhibits true circadian patterns in oxygen consumption rate. Continuous observation over three days in constant dark laboratory conditions revealed a substantial (40%) daily change in oxygen consumption rate, a peak during dawn, and lowest levels during the evening (Maas et al. 2018). Pleuromamma xiphiassampled directly from the field at different times of day showed diel variation in the activity of metabolic enzymes associated with oxygen consumption (citrate synthase and the electron transport system) and ammonium metabolism (glutamate dehydrogenase; Tarrant et al.2021). However, during short duration incubations of these wild-caughtP. xiphias held at common environmental conditions, investigators were unable to detect diel oscillations in oxygen consumption rate. Furthermore, the enzyme patterns detected in the field were inverse in pattern to those observed in dark:dark conditions in the laboratory, with peaks during the late afternoon and early evening. Combined, these studies both demonstrate the difficulty of determining instantaneous changes in metabolic state in organismal studies of small wild-caught organisms, and also suggest potential interactive effects of the various ecological drivers that migrators experience over diel cycles.
To gain further insight into diel transitions in metabolism, we have conducted transcriptomic and proteomic analyses of P. xiphiascaught across the diel cycle, using flash frozen copepods collected at the same time as individuals used in previously published measurements of oxygen consumption, ammonium excretion, fecal pellet production, and enzyme activity (Tarrant et al. 2021). This design thus allows simultaneous analysis of periodicity in both the transcriptomic and proteomic datasets, and an assessment of how these markers relate to metabolic rates.
Based on previously measured diel variation in enzyme activity levels, we hypothesized that we would observe complex regulatory mechanisms at the transcriptomic and proteomic level associated with periods of “rest” during the day, and recovery from migration during dawn and dusk. Informed by previously observed differences in fecal pellet production and enzyme activity (Tarrant et al. 2021), we additionally hypothesized that we would observe diel changes in digestion and excretion pathways as a consequence of different prey fields (high food concentrations at the surface at night and lower concentrations at depth during the day).
Materials and Methods
2.1 Sample Collection
To determine diel patterns of gene expression and protein abundance in a migratory copepod, Pleuromamma xiphias were collected during a cruise aboard the R/V Atlantic Explorer from May 20-22, 2019 in the waters off Bermuda. The objective was to sample individuals from their natural location in the water column at as close to in situ conditions as possible over the course of their diel migration. During the evening, when the animals resided near the surface, organisms were captured using a 1-m diameter Reeve net (Reeve 1981) deployed to 200 m depth, with 150 µm mesh, a 20-L cod end, and a miniSTAR-ODDI temperature and depth sensor. During the daytime, after animals had migrated into their colder, deeper water, tows were conducted from 400-600 m depth using a 1-m2 MOCNESS equipped with 150 µm mesh nets and a custom-built thermally-insulated closing cod end. For both shallow and deep-caught organisms, time in the net was maximally 1.25 hours at a constant temperature. On deck, copepods were rapidly selected from the cod end and placed in filtered seawater. Adult female P. xiphiaswere identified using a stereomicroscope and flash-frozen for transcriptomic and proteomic analyses (<20 minutes processing time). Time points were 22:00 (evening, after the upward migration) on May 21st and 02:00 (midnight), 09:00 (morning, after the downward migration), 15:00 (mid-afternoon), and 22:00 on May 22nd (local time; sunrise during this cruise was ~06:00 and sunset was ~20:00). Individuals from the two separate 22:00 time points were analyzed for the subsequent analyses (total n=8). All transcriptomic and proteomic analyses were done on whole individual copepods to capture biological variation.
2.2 Transcriptomics
Pairwise differential expression analysis, cluster analysis and network analysis were conducted to explicitly examine differences in expression associated with time of day. To accomplish this, RNA was extracted from eight individuals from each time point, then the best six extractions (highest quality on Bioanalyzer) were sequenced on an Illumina NovaSeq6000 (s2-300) as 150 bp paired-end reads (30M reads per sample). Adapter sequences were trimmed, and low-quality reads were removed (detailed methods in SF1).
Multiple transcriptome assemblies were constructed with the Trinity pipeline (v2.3.2; Haas et al. 2013), then compared using BUSCO, and evaluated on the basis of high completeness (BUSCO score; Simãoet al. 2015), higher e90n50, and a lower number of total transcripts (see SF1 for detailed methods and statistics). The assembly chosen for use was built from one individual copepod captured at 02:00 with no transcript clustering. This transcriptome was annotated following the Trinotate (v3.2.0) pipeline using a local NCBInr blastx database (updated Feb 3, 2020). The assembly was filtered to include only sequences that were annotated as metazoan (see SF2 for annotation).
To enable differential expression analysis, reads were mapped against the reference transcriptome using Bowtie2 (v.2.2.9; Langmead & Salzberg 2012), estimates of abundances were made with RSEM (v1.3.0; Li & Dewey 2011), and pairwise comparison of differential expression (DE) among all time points was performed using an edgeR analysis in R v.3.6.0 (Robinsonet al. 2010) following the pipeline packaged with Trinity (v.2.3.2; Haas et al. 2013). Differences were considered significant if the log2-fold change was > 2 (corresponding to a four-fold change in expression), and both the false discovery rate and p value were < 0.05.
All transcripts for which there were no counts in any library were removed from the dataset, leaving 21,863 transcripts for downstream analyses in R v. 4.1.0. Hierarchical clustering was applied to these transcripts. First, read abundances were averaged across replicates (n=6) within timepoints. The average clustering method was used on a Bray-Curtis dissimilarity matrix produced with the vegan package (v 2.5-7; Oksanen et al. 2020). Tree cut-off height was set at 0.5. The data were formatted using the package reshape (Wickham 2007) and plotted in ggplot2 (Wickham 2016) with a loess smoothing curve for each cluster.
Two forms of ordination analyses were performed on the dataset. Nonmetric multidimensional scaling (NMDS) was performed on a Bray-Curtis dissimilarity matrix based on the log(x+1)-transformed transcriptomic data using the vegan package. A second ordination analysis, discriminant analysis of principal components (DAPC), was applied to minimize the within-group/time point variability and better reveal the transcriptomic differences that could be ascribed to time of day in the adegenet package (Jombart 2008). The top 60 transcripts weighted along Linear Discriminant (LD) 1 and 2 of the DAPC analysis were exported for further analysis of the transcripts whose expression patterns were more strongly correlated with time point.
Weighted gene correlation network analysis (WGCNA) was performed on the full transcriptomic dataset following Langfelder and Horvath (2008), and using an interactive GUI (detailed in SF1).
Gene Ontology enrichment (GO enrichment) analyses were run to determine significantly (p-value ≤ 0.01) enriched categories of biological process (BP), molecular function (MF), and cellular component (CC) for all of the differential expression comparisons, as well as WGCNA modules and hierarchical clusters of interest using the GoSeq package in R (Younget al. 2012). The most specific terms were identified with the CompGO visualization tool (https://meta.yeastrc.org/compgo_TS6a_Trinity/pages/goAnalysisForm.jsp ) (Timmins-Schiffman et al. 2017).
Proteomics
Four copepods from each point were selected for proteomics analysis. Copepods were individually homogenized, and proteins were extracted and digested from the homogenates (see SF1). Proteomics mass spectrometry was performed on each copepod peptide digest, in triplicate, on a Q-Exactive Plus mass spectrometer (Thermo Fisher Scientific). Reverse-phase high performance liquid chromatography was performed in-line with the mass spectrometer operated in data dependent acquisition mode (details of mass spectrometry settings available in SF1). Chromatographic and MS consistency were monitored with regular injections of a quality control mix that included BSA and Pierce’s Peptide Retention Time Calibration mix in the Skyline software (MacLeanet al. 2010).
To create a reference proteome, sequences from the reference transcriptome (described above) were translated using Transdecoder v. 2.0.1 (github.com/Transdecoder) and concatenated with standard laboratory contaminant sequences from the cRAPome (Mellacheruvu et al. 2013). Raw files of acquired mass spectra were then searched against the reference proteome using Comet v. 2019.01 rev. 4 (Enget al. 2015; Eng et al. 2013), with the following parameters: concatenated decoy search; peptide mass tolerance = 20 ppm; search enzyme = trypsin; allowed missed cleavages = 2; fragment bin tolerance = 0.02; fragment bin offset = 0. Probability scores for peptide and protein detection were assigned using PeptideProphet and ProteinProphet (Deutsch et al. 2015). Consensus protein inferences for proteins with a false discovery rate < 0.01 or less were determined using Abacus (Fermin et al. 2011). Technical replicates that did not group with the other replicates in a group were excluded from analysis (n=3).
The proteomics dataset was analyzed in parallel to the transcriptomics dataset as described above, using NMDS, DAPC, hierarchical clustering, and WGCNA. The hierarchical clustering tree height was cut at 0.5 for proteomics. Proteins were annotated with the Uniprot trembl database (downloaded May 2019) using BLASTp with an e-value cut-off of 1E-10. QSpec (Choi et al. 2008) was used to detect differentially abundant proteins between grouped day and night time points, as well as pairwise between all time points. Proteins were considered to be significantly differentially abundant if |log fold change| was greater than 0.5 and the |z-statistic| was greater than 2. Enrichment analysis was performed on differentially abundant proteins, on protein clusters from the hierarchical clustering, and on proteins in WGCNA modules using CompGO (Timmins-Schiffman et al. 2017). Gene Ontology terms for proteomics were considered significantly enriched at a p-value of ≤ 0.1. The compGO portal can be accessed at the following url:https://meta.yeastrc.org/compgo_emma_sub_copepod/pages/goAnalysisForm.jsp.
2.4 Integrating Datasets
To explore associations between the ‘omics datasets and the broader metabolic pathways that influence biogeochemical cycling, particularly ammonium production and respiration, we investigated specific pathways in the transcriptome and proteome, plotting expression and abundance patterns that were compared with enzyme and organismal level measurements from Tarrant et al. (2021). Chitin biosynthesis and degradation processes were queried using the inferred crustacean pathways described in Zhang et al. (2021).
3. Results
3.1 Transcriptomics
The transcriptomic assembly used for gene expression analysis was generated from a single individual with no transcript clustering, and initially consisted of 123,272 genes and 174,733 transcripts (see SF2 for transcript counts and annotation). The assembly had high N50 (1275 bp), e90n50 (1764), and BUSCO completeness (C:96% [S:51.7%, D:44.3%], F:1.7%, M:2.3) scores. Annotation of this “raw” transcriptome indicated contamination, likely due to the gut contents of these wild caught organisms. For gene expression analysis, the assembly was filtered to only retain transcripts that were annotated as metazoan in origin (21,872 genes and 38,517 transcripts) while those that were non-metazoan (3,043) and unannotated (133,173) were removed. On average, 49% of all sequences from the samples mapped back uniquely onto the filtered reference transcriptome, while only 2.5% mapped back to multiple contigs (details about trimming, assembly, annotation, and mapping are all available in SF1).
When total gene expression was statistically investigated, the time points were significantly different based on an ANOSIM analysis (R = 0.2498; p-value = 0.003, NMDS SF3), and day versus night clustering was significant (R = 0.1654 and p=0.011). Hierarchical clustering of total transcript expression among all samples resulted in 48 statistically distinct clusters of gene expression patterns over the diel cycle (SF2; SF3). The two predominant expression patterns were a peak during the middle of the day (15:00) and reduced expression in the middle of the day (15:00). Likewise, the WGCNA module-trait correlation analysis similarly suggested that the most abundant gene expression patterns were associated with a peak at 15:00 (SF2; SF3). Clustering (DAPC analysis; Figure 1A) was most significantly attributable to a distinct gene expression pattern at 15:00 (LD1 axis, 86.5% of the variation), with no clear progression between sequential time points.
In pairwise comparisons, a total of 7,956 differentially expressed genes (DEG) were identified (Table1). This constitutes 39% of the transcripts analyzed. Many of the genes were differentially expressed in multiple pairwise comparisons, resulting in 15,566 significant relationships; 96% of these represented the upregulation of a gene at the 15:00 time point, and 3% were downregulation of a gene at the 15:00 time point. Of the 327 genes that were downregulated in pairwise comparisons at 15:00, there were no statistically significant enriched GO terms.
During the mid-day peak in transcription (15:00) numerous transcripts associated with glycolysis, gluconeogenesis, protein and fatty acid catabolism based on their GO annotation were upregulated (SF3). Oxidative stress response genes, including peroxidases and superoxide dismutase, were additionally upregulated in this period, along with a subset of the heat shock proteins. Various forms of vitellogenin were significantly upregulated at 15:00. These observations were statistically supported by the GO enrichment analysis of the pairwise comparisons to 15:00, which demonstrated that the upregulated genes were enriched for GO terms associated with protein metabolic process (especially proteolysis), lipid transport, response to oxidative stress and motor activity (SF4). The GO enrichment analysis also suggested cellular detoxification, immune response, ionic transport and a suite of biological regulation processes as transcriptionally upregulated during the middle of the day. Finally, a number of core circadian genes (period, PER; timeless, TIM; and cryptochrome 1, CRY1) peaked in the middle of the day with statistically significant patterns in gene expression (Fig. 2)
Beyond the overarching pattern of a mid-day peak in transcription at 15:00, there were a set of transcripts that were upregulated in the 09:00 time point, particularly in contrast to the 22:00 time point. These transcripts were frequently observed as also upregulated in the 15:00 pairwise comparisons suggesting a “daytime” set of upregulated pathways. Specifically, during the day at depth (9:00 and 15:00 time points) 92 genes were upregulated compared to one of the night time points. Of these upregulated DE genes, almost half (44) were associated with muscle tissue and contractile function (i.e. actin, myosin, smoothelin, tropomodulin, paladin, and troponin). There was also upregulation of numerous (15) transcripts associated with smooth muscle, collagen, cuticle proteins, cuticlin, and peritrophin – proteins that are thought to be associated with the production of fecal pellets (which involves packaging of material in a chitinous package within the gut). Motor activity was overrepresented in the GO enrichment analysis of genes with higher expression during the day (SF4). These patterns were even more pronounced when assessing genes that were upregulated during the 15:00 time point. Almost all sequences annotated as associated with muscle tissue and contractile function were upregulated in the middle of the day (15:00), relative to at least one other time point. Most enzymes in the inferred crustacean chitin metabolism and degradation pathways (Zhang et al. 2021) were also significantly upregulated in the middle of the day (SF2), as well as the sequences identified as being potentially associated with the cuticle and the peritrophic membrane.
3.2 Proteomics
Across all time points, 2214 proteins were identified with high confidence (see SF5 for protein counts and annotation). Hierarchical clustering resulted in 27 statistically distinct clusters of protein abundance (SF2). Proteomes differed significantly by time point and globally between night and day (ANOSIM by time point: R = 0.4611 with p-value = 0.001, night vs. day: R = 0.2484 with p-value = 0.013; NMDS SF3). In the proteome DAPC analysis, much of the variation appeared to be associated with time point (which correlated with LD1, 74.1% of the variation). DAPC analysis also revealed differences within the quantitative proteome between night and day, which also represents the depth of sampling (correlated with LD2; Fig. 1B; 24% of the variation in protein abundance). In pairwise differential abundance comparisons between time points there were 133 unique proteins that were significantly differentially abundant (SF5). When time points were clustered by day and night (i.e., 09:00 and 15:00 vs 22:00 and 02:00) there were 36 differentially abundant proteins.
Across analyses (e.g., clustering, WGCNA, QSpec), specific markers of the nighttime proteome and daytime proteome of P. xiphias were revealed. At night (i.e., 22:00 and 02:00), when P. xiphias was found nearer the surface and was presumably actively feeding, the proteome shifted towards functions of glycolysis followed by changes in amino acid biosynthesis. Specifically, during the evening (22:00), when the copepods come to the surface, there was a peak in abundance of glucose-6-phosphate 1-dehydrogenase, glyceraldehyde-3 phosphate dehydrogenase, pyruvate kinase, and phosphoglycerate mutase. While the latter three enzymes are involved in the glycolysis pathway, glucose-6-phosphate 1-dehydrogenase catalyzes the first step in the pentose phosphate pathway, an alternate pathway for the breakdown of glucose-6-phosphate, which finishes with the glycolytic intermediate glyceraldehyde 3-phosphate. Many protein clusters show peaks at the 22:00 time point (clusters 1, 6, 8, 9, 14, 16, 26; SF3Fig2) and the diverse functions of proteins included in these clusters may point to a time of high metabolic activity. In the later part of the night (02:00) the protein betaine-homocysteine S-methyltransferase (BHMT) peaked (compared to all other time points). BHMT is involved in amino acid biosynthesis, specifically of the amino acid L-methionine, and is part of the larger cysteine and methionine metabolism pathway. In humans, supplements of betaine increase power (Cholewa et al. 2013), and similarly, the increased BHMT may be indicative of high levels of reactant to produce increased betaine prior to the downward migration. Also at 02:00, the abundance of the enzyme responsible for the rate-limiting step of the methionine cycle, S-adenosylmethionine synthetase, was decreased in comparison with all other time points. The cluster analysis reveals some groups of proteins that peak in abundance at 02:00, including molecular chaperones (cluster 7) and a variety of ATPases (cluster 15).
When the copepods were at depth (~400-600 m) during the daylight hours (09:00 and 15:00 sampling time points), their proteomes shifted towards chitin metabolism, alternate pathways of carbohydrate metabolism, amino acid metabolism, and mitochondrial respiration.  Multiple observations suggest an increase in chitin metabolism in the daytime proteome. In the enrichment of cluster 12 (SF3Fig2; proteins elevated in the daytime), “chitin metabolism” proteins included two isoforms of protein obstructor-E (a cuticular protein). Cuticle protein 6 was increased in abundance at 15:00 (compared to 22:00) and chitinase 4 was suppressed at 15:00 compared to all other timepoints. Levels of carbonic anhydrase, which has a variety of roles including pH regulation and physiological homeostasis, increased in the morning (09:00 vs. 02:00) and then rose again as the day progressed (15:00 vs. 09:00 and 02:00). Potential evidence of gluconeogenesis was also present during the daytime, with the inclusion of pyruvate carboxylase in daytime-peaking cluster 21 and glyceraldehyde 3-phosphate dehydrogenase, which is active in the gluconeogenesis and glycolysis pathways, in clusters 11 and 13 (SF3Fig2). Proteins in the WGCNA blue module (SF5) trended towards higher abundance during the day and were most strongly correlated with night:day (correlation coefficient of -0.552). This module was enriched for GO terms tricarboxylic acid cycle (e.g., oxoglutarate dehydrogenase, malate dehydrogenase) and electron transport chain (e.g., cytochrome c oxidase, electron transfer flavoprotein). Cytochrome c, a mitochondrial protein involved in the electron transport chain, was detected at significantly elevated levels during the day (09:00 vs. 02:00 and 22:00; 15:00 vs. 02:00 and 22:00).
Some proteins were statistically more abundant in pairwise differential analyses at both the 09:00 and 22:00 time points relative to 02:00 and 15:00. Both periods, directly after the migrations, had higher levels of multiple isoforms of vitellogenin, a major egg yolk protein precursor, the increased abundance of which was responsible for the enrichment of the GO term lipid transport (biological process) in the nighttime proteome and lipid transporter activity (molecular function) in the WGCNA brown module (SF5). The brown module, which was moderately correlated with hour (correlation coefficient of 0.453) and night/day (0.492), contained many proteins with elevated abundance at 22:00 and at 09:00, and was also enriched with proteins that matched to GO terms proteasome complex, Golgi-associated vesicle membrane, isomerase activity, and threonine-type endopeptidase activity. Proteins involved in muscle contraction, including myosin, troponin and tropomyosin were frequently upregulated in the 09:00 and 22:00 time points. At 09:00 vs. 02:00 other muscle proteins detected at increased abundance included actin, smoothelin, a calponin-homology domain-containing protein, and a protein unc-45 homolog (also elevated at 09:00 vs 22:00). Hemerythrin, a protein that binds oxygen, among other functions, was also elevated at 09:00 and at 22:00 vs.15:00. Finally, the protein peroxiredoxin, which is associated with oxidative stress, was upregulated at 22:00.
3.3 Integrated Transcriptomics and Proteomics
In a comparison of differentially abundant markers between ‘omics datasets, 31 unique transcripts were also differentially abundant in the protein dataset. The temporal pattern, or direction of regulation, was not always conserved between transcription and translation. Often (>70%), the ‘omics markers followed opposite patterns, which suggests a temporal shift in the regulation of transcripts and proteins; transcripts did not always peak before proteins for the same marker (SF3).
Most of the transcripts and proteins that represented specific enzymes that are frequently used as proxies for organismal level processes, such as glutamate dehydrogenase for ammonium excretion or citrate synthase and the electron transport system enzymes for oxidative respiration, were not significantly different between time points in the pairwise statistical comparisons. This was inconsistent with the observed statistical differences in enzyme activity that were made during excretion experiments conducted with organisms flash frozen at the same time as our samples (Fig. 3).
An additionally interesting pattern was the presence of transcripts with low coefficients of variation (e.g., stable expression across samples) for which the corresponding proteins demonstrated large changes in abundance. These may suggest proteins that are post-transcriptionally regulated. These proteins are described in SF3 and include, for example, electron transfer flavoprotein, signal peptidase complex catalytic subunit, and ADP ribosylation factor.
Discussion
Diel vertical migration of metazoan animals is a common behavior observed across phyla in marine and limnological systems. With 70% of the earth covered in water, this means that a diverse set of ecosystems are biogeochemically and trophically impacted by this phenomenon. In this study, we leverage a suite of thousands of molecular markers (transcripts and proteins) to better understand how the physiology of the abundant circumglobal copepod, Pleuromamma xiphias , varies over the course of the day. Our findings suggest that migratory species are responding to the variety of changing environmental factors over their diel cycle by partitioning components of their metabolic and repair processes to different periods of the day, with important implications for their biogeochemical and ecological roles (Fig. 4). There are distinct metabolic processes upregulated by these copepods while feeding at the surface at night and during the “non-feeding” period at depth during the day, as well as responses to the periods of migration. The proteomic and transcriptomic signaling associated with these cycles is often temporally uncoupled. As this species is a dominant migrator in the region, responsible for an average of 23% of the migratory population biomass (Steinberg et al. 2000), more explicitly detailing how the physiology of P. xiphias responds to the suite of co-varying environmental forces over the diel cycle has implications for our understanding of the local active flux of carbon and nitrogen in the Sargasso Sea. Echoing the complexity demonstrated in terrestrial and model organisms, these results suggest that estimating biological rates from ‘omics in metazoans will never be straightforward. These findings do, however, position P. xiphias as a model for increasing our understanding of how diel changes in migrator physiology contribute to the spatial and temporal complexity of midwater ecology and global biogeochemical cycles.
One limitation of our design was that it integrated signals across multiple tissue types, potentially masking distinct diel patterns of stimulation in different tissue types. It was, however, reflective of the “whole organism” response that is relevant to the broader zooplankton physiology and biogeochemical literature (Ikeda 2014; Steinberg & Landry 2017), as well as the co-captured organismal level metrics (Tarrant et al. 2021). The integration of these synchronously collected datasets additionally emphasizes that sampling at the transcript, protein, enzyme, and organismal (i.e. oxygen consumption, ammonium excretion) level provides very different perspectives of organismal physiology, making mechanistic descriptions or biogeochemical predictions from any single level of organization difficult. Oceanographers are seeking ways to implement ‘omics tools to more comprehensively understand the physiology of marine zooplankton (Lenz et al. 2021; Matos et al. 2020; Tarrant et al. 2019), which are classically difficult to capture, culture, and mechanistically understand. The fact that the transcriptomics patterns were not closely correlated with patterns at the other levels (proteomics, enzymatic or organismal) raises concerns the present application of transcriptomic approaches in oceanographic studies to infer physiological activity, and to derive biogeochemical significance. In the light of these results, and as has been previously demonstrated in terrestrial and model organisms (Kassahn et al. 2009; Vogel & Marcotte 2012), it appears that combined approaches likely provide the best understanding of migrator physiology and the associated consequences for pelagic ecology.
Circadian Signals
The transcriptomic and proteomic responses documented here provide a detailed look at how P. xiphias gene expression and protein abundance shifts throughout the diel cycle. While we did not specifically study the circadian rhythm of this copepod, we detected transcripts of most of the core circadian genes previously observed in copepods. Many known circadian genes were differentially expressed in our pairwise comparisons and peaked in the middle of the day, including period (PER), timeless (TIM), cryptochrome-1 (CRY1) and aryl hydrocarbon receptor nuclear translocator-like protein 1 (ARNTL, also called BMAL1 or cycle; Fig. 2). Two other core circadian genes, CLOCK and cryptochrome-2 (CRY2), showed antiphase expression with a nadir mid-afternoon, although these changes were not statistically significant. The trend of CLOCK being in antiphase with PER, is consistent with prior analyses in sub-polar and temperate populations of the copepod Calanus finmarchicus (Häfker et al. 2017; Hüppe et al. 2020). In studies of C. finmarchicus,variations in day length, light field, season, and depth of capture result in differences in the timing of circadian gene expression, indicating a complex interaction between the environment and circadian signaling. Further studies of copepods from a variety of environments are called for to complement previous studies conducted in C. finmarchicus , which have shown that photoperiod, and potentially food moderate the expression of circadian regulatory genes (Häfker et al. 2017; Häfker et al. 2018). Genes identified as having rhythmicity in polar copepods, including those associated with neurotransmitters, oxido-reduction, carbohydrate metabolism, lipid metabolism, and proteolysis processes (Payton et al. 2021), were also upregulated at 15:00 in our dataset. This suggests that these are robust indicators of diel rhythmicity across copepods, despite substantial differences in light level and species ecology.
Migration
As per our hypothesis, there appear to be proteomic responses that distinctly correlate to specific periods of migration. Interestingly, some of these were additionally present in the transcriptome throughout the middle of the day resulting in patterns that distinguish the middle of the night from all other time points. For example, myosin and other motor activity proteins showed complex patterns of expression that appear to be linked to migration. Transcripts associated with motor and contractile function were upregulated throughout the daytime period. This is supported by the transcriptomic upregulation of myosin, actin and various other transcripts associated with the motor function GO term at the 15:00 time point. In the proteome, many isoforms of myosin peaked after the migrations (9:00 and 22:00), while smoothelin, tropomyosin and actin were more abundant during the daytime points (9:00 and 15:00). Repair of the muscular machinery after the effort of diel vertical migration seems a potential cause of this signal during the periods after upward and downward migration. The peaks in the middle of the day, when swimming activity is thought to be at its lowest as organisms avoid detection by visual predators, may suggest that during this period the musculature continues to be under repair. Circadian patterns in muscle repair and regeneration during the rest period are, in fact, well documented in mammalian literature (reviewed in: Chatterjee & Ma 2016). Consequently, the observed daily cycles in investment in muscle tissue are perhaps unsurprising for organisms that undergo such extensive migrations, but are highly relevant for analysis of prior work in migrators since actin, in particular, was historically used as a housekeeping gene in RNA studies (reviewed by Tarrant et al.2019).
Similar to the patterns observed in myosin and motor activity, vitellogenin transcripts were statistically more abundant at 15:00, and vitellogenin protein peaks occurred at both migratory time points (09:00 and 22:00) in pairwise differential abundance analysis. Vitellogenin is ubiquitous in copepods, which possess multiple homologs of the gene in their genomes (Hwang et al. 2010; Hwang et al. 2009; Leeet al. 2008; Lee et al. 2016; Semmouri et al.2020). Vitellogenin RNA shows variation in expression level across developmental stages (Lee et al. 2008) and molt cycles (Tarrantet al. 2014), and is expressed in both mature male and female copepods but at much higher levels in females than males (Hwang et al. 2010; Hwang et al. 2009; Lee et al. 2008; Leeet al. 2016). Pleuromamma xiphias have asynchronous oogenesis (egg production) meaning all stages of unfertilized oocytes are present within all mature females. All individuals used in this study were mature adult females, and consequently were all 1) engaged in vitellogenesis (production of yolk bodies) in their eggs (Eckelbarger & Blades-Eckelbarger 2005; Niehoff 2007), and 2) not molting anymore. The variation in vitellogenin thus could be related to patterns in allocation of energy to reproductive investment over the daily cycle. Alternatively, vitellogenin has additionally been shown to act as an antioxidant, as well as to play a role in the immune system in other copepod taxa (Zhang et al. 2011). In the copepod Tigriopus kingsejongensis, previous researchers have suggested that homologs of vitellogenin had other physiological roles, including potential immune-related activity (Hwang et al. 2010; Hwang et al.2009; Lee et al. 2016). Consequently, it is possible that some of the vitellogenin patterns observed in our protein dataset are associated with antioxidant or immune defense activity in P. xiphias , which may be differentially regulated during DVM.
Finally, in both the transcriptome and the proteome there were signals of oxidative stress response, a common consequence of acute exercise (Simioni et al. 2018), after the periods of migration and during the day. In the transcriptome this was characterized by upregulation of heat shock proteins, ubiquitins, superoxide dismutase, and hemicentin. In the proteome there were increased abundances in ubiquitin-like modifier-activating enzyme, hemerythrin, multiple forms of peroxiredoxin, a glutaredoxin domain-containing protein, and glucose-6-phosphate dehydrogenase. The former proteins are well known components of the oxidative stress response across taxa, and the latter may indicate a need to control oxidative stress via the production of NADPH (Préville et al. 1999). There was stronger signal of oxidative stress response in the proteome after the upward migration (22:00), which may be due to the increased temperatures experienced after arriving in the evening surface habitat in conjunction with migratory effort, or a distinction between the effort of upward versus downward swimming. Redox state has, however, been suggested to provide non-transcriptional control over diel rhythmicity especially via peroxiredoxin proteins (Reddy & Rey 2014). The prevalence of peaks in the peroxiredoxin proteins at 22:00, along with genes associated with ubiquitination, response to oxidative stress and oxidation-reduction processes, have been observed in our studies and others (Maas et al. 2018; Payton et al. 2021). This suggests that, beyond being a consequence of migratory activity, there may be a role of oxidation-reduction cycles in the diel rhythmic signaling in zooplankton.
Midwater Daytime Activity
The daytime has classically been considered a period of quiescence for copepods that are waiting out the day avoiding predators, with lower feeding activity, lower temperatures, and reduced oxygen thought to contribute to various types of behavioral and metabolic suppression (Lampert 1989; Ohman 1988; Seibel 2011; Torgersen 2003). Our dataset suggests the daytime is, however, a period of complex and distinct cellular activity. The highly upregulated transcriptomic response at 15:00 has not been previously reported for zooplankton and implies that during the daytime midwater animals are not simply “resting” in a semi-dormant state, but are actively engaged in transcription, repair and biological regulation. Prior enzymatic analyses indicate that the mid-day (15:00) is a peak for the electron transport system (Tarrantet al. 2021), while the current data reveals an upregulation in transcripts associated with glycolysis, gluconeogenesis, protein and fatty acid catabolism, the electron transport chain, lipase activity, cell growth/development, lipid metabolism or transport, and protein maturation/processing. Processes elevated in the proteome during the day include carbohydrate metabolism, amino acid metabolism, and electron transport chain.
Additionally, ‘omics markers suggest that the copepods are actively engaged in processes that involve chitin metabolism and binding during the daytime. Specifically, analysis of the transcriptome reveals that the mid-afternoon time point (15:00) is the peak in the differential expression of a suite of chitin biosynthesis transcripts including trehalase (TRE), chitin synthase (CHS), glutamine-fructose-6-phosphate aminotransferase (GFAT), glycogen phosphorylase and phosphoglucomutase. Chitin metabolism proteins (e.g. protein obstructor E, cuticle protein 6) were also elevated during daytime time points. These genes and their associated proteins are known to be involved the production of the intestinal peritrophic matrix (Zhang et al. 2021), which produces the chitinous covering of fecal pellets. In measurements made during the same cruise (Tarrant et al. 2021) fecal pellet production was highest during the middle of the night (02:00), but remained at ~50% at depth during the morning (09:00). Thus, the increased chitin synthesis and metabolism during the daytime period may reflect the recovery of this membrane after fecal pellet expulsion. Alternative roles for chitin metabolic processes include repair or growth of the exoskeletal structures. Once they reach an adult stagePleuromamma sp. have been documented to experience cuticular thickening and modification in the postmolt and intermolt stage, although they no longer molt (Park 1995). Shifts in chitin metabolism proteins, which were concurrent with increases in some muscle proteins, could thus be indicative of an investment in tissue and exoskeleton repair during this time. Additionally, some of the chitin degradation transcripts, like beta-N-acetylglucosaminidase and chitinase, were also upregulated in the mid-afternoon and could be associated with exoskeletal repair, or could similarly be indicative of the digestion of crustacean prey items. Krill that undergo DVM similarly show changes in gene expression suggesting an up-regulation of chitin synthesis and catabolism during the day (Biscontin et al. 2019), further suggesting a general pattern in migratory Crustacea.
Daily Metabolic Partitioning
As predicted, there was partitioning of metabolic processes that was obvious in the proteomic dataset. During the day proteins associated with the electron transport system (ETS; cytochrome c and NADH dehydrogenase) were increased in abundance. This is consistent with the previous findings of increase electron transport enzyme assay activity in the mid-day (Tarrant et al. 2021), and peak in oxygen consumption during daytime hours (6:00-12:00) in constant dark laboratory conditions (Maas et al. 2018). Fructose bisphosphate aldolase, a glycolysis enzyme, is more abundant in the middle of the day (15:00), and right after the migration (22:00) a suite of proteins associated with glycolysis including glyceraldehyde-3-phosphate dehydrogenase and pyruvate kinase, as well as glycogen phosphorylase, which is the first enzyme of glycogenolysis and provides glucose-6-phosphate for glycolysis, are more abundant. These proteins suggest an increase in carbohydrate metabolism in the evening, transitioning to the mobilization of glycogen in the later evening, additionally supported by the increased abundance of a protein involved in the pentose phosphate pathway at 22:00. This pattern of metabolic partitioning matches well with the observed transcriptomic patterning in polar krill Euphausia superba (Biscontin et al. 2019), where ETS peaked in the later afternoon, and was followed by glycogen mobilization and glycolysis in the early evening.
4.5 Data Integration
The expansive upregulation of transcripts during the middle of the day, which was the overwhelming signal in the transcriptomic dataset, points to mid-afternoon being a period of regulation and repair of the cellular machinery. The profound differential expression of transcripts did not, however, typically translate to changes at the protein level, suggesting that the transcripts were produced in response to or expectation of a need to replace proteins to maintain a relatively stable level. Estimations of transcript expression do not provide an accurate picture of the true biological state, as mRNA profiles do not capture regulatory processes or post transcriptional modification that directly influence the amount of active protein (reviewed in: de Sousa Abreu et al.2009; Maier et al. 2009; Vogel & Marcotte 2012). Unfortunately, transcriptomics tend to yield extreme highs and lows in transcript abundances- resulting in exaggerated log fold changes. Additionally, turnover time for proteins is generally substantially longer than that of RNA, so subtle shifts in their abundance could signal the production of transcripts that allowed for replacement, maintaining the observed steady state.
To add to the complexity of data interpretation, the assembled transcriptome often had multiple “genes” that had the same annotation, and the redundant annotations propagated to the protein level where various protein sequences were observed with similar names. These likely represent splice variants, as is typical in de novo assembled copepod transcriptomes (Tarrant et al. 2019), and is supported by the relatively high proportion of duplication within the assembly (BUSCO score; 44.3%). We chose not to cluster these sequences by a similarity threshold to allow observation of variation in the expression patterns. It is likely that these sequences perform distinct roles that have not yet been elucidated. These variants would be the result of post-transcriptional modification that produces distinct mature mRNA and subsequent proteins that are tuned to different environmental conditions or different cell lines. Alternative splicing (one of the forms of post-transcriptional modification) has been observed in response to thermal or salinity stress in metazoan animal species including fish and rats (Huang et al. 2022; Tan et al. 2019; Tian et al. 2020). It is possible that the changes in environment that are part of the diel migration result in the preferential production of different isoforms at surface and depth. Exploring this possibility, which would require a P. xiphias genome assembly, would be instructive for our understanding of the structure and function of transcriptomes in planktonic species.
The observed variations in pattern from transcript to protein to enzyme assay emphasize the complexity of metabolic regulation (Figure 4). Although enzyme assays typically provide unlimited substrate, they are influenced by the presence of phosphorylation, inhibitor and promotor molecules within the copepod cells, demonstrating a further level of biological regulation of diel patterning (Reddy & Rey 2014; Thurleyet al. 2017), which has been observed in marine primary producers (Tan et al. 2020; Welkie et al. 2019). Finally, the time lag for protein production and the variable turnover rate for protein degradation can exacerbate discontinuities between protein levels and enzyme activity levels. Combined, our findings suggest that although total oxygen consumption and ammonium excretion do not change detectably over the course of the day (Tarrant et al. 2021), the cellular machinery of the copepods are undergoing complex and distinct processes throughout the day, with potential profound responses for the production of waste products at the surface and at depth. While diel variation in fecal pellet production is already accounted for in biogeochemical studies (Schnetzer & Steinberg 2002; Steinberg et al. 2023), diel variation in excretion of dissolved metabolites is typically ignored and merits further investigation.
As the field of oceanography increasingly explores the use of transcriptomic, proteomic, and enzyme assays as proxies for ecological function, it is imperative to document and acknowledge the diel variability and asynchronous patterning of these markers of metabolism. Our results suggest that in copepods, post-transcriptional and post-translational modifications may play a large role in metabolic regulation. Consequently ‘omics datasets could be difficult to use to directly predict biogeochemically relevant rates of carbon or nitrogen consumption and waste production. Thurley et al. (2017) found that metabolic pathways with circadian rhythms are often propagated by rhythmic but asynchronous patterning across multiple enzymes. Consequently, phenotype (what matters functionally in biogeochemistry/ecology) is controlled by a complex set of temporally asynchronous enzymatic processes. However, the strength of the ‘omics analyses, which are high-throughput and more temporally precise, is that they can be used to prioritize enzymes for rate-measurements as we move to more intensive molecular analyses in the ocean environment. The cautionary tale is that fully integrated datasets are required to understand diel physiology. If the “products” (i.e. simple functions like oxygen consumption) are measured independently, the diel controls may not be obvious. In contrast, if analyses are only made at the transcriptome level, it may appear that there would be profound physiological activity in the middle of the day. Exploring the phase lag between various proteins, enzymes and organismal function, especially waste production, would require higher temporal resolution sampling, but will be important for the application of ‘omics and enzymes as biogeochemical proxies in the future.
Acknowledgements
We would like to thank Captain George Gunther and the crew of theR/V Atlantic Explorer , as well as cruise participants Hannah Gossner, Andrea Miccoli, Lindsey Cunningham, Susanne Neuer, and Nora McNamara-Bordewick for the assistance with diel copepod sampling. We are grateful for the efforts of Jason Kapit and his team of engineers who designed and fabricated our closing cod ends. Alex Federation provided assistance with the mass spectrometry. This work is supported in part by the University of Washington’s Proteomics Resource (UWPR95794) and the UW BioStatistics department provided guidance on the combined ’omics analysis. Funding this work was supported by the National Science Foundation Grants OCE-1829318 (to AEM, LBB, ETS, BLN) and OCE-1829378 (to AMT).
References
Al-Mutairi H, Landry MR (2001) Active export of carbon and nitrogen at Station ALOHA by diel migrant zooplankton. Deep Sea Research Part II: Topical Studies in Oceanography 48 , 2083-2103.
Antezana T (2009) Species-specific patterns of diel migration into the Oxygen Minimum Zone by euphausiids in the Humboldt Current Ecosystem.Progress In Oceanography 83 , 228-236.
Archibald KM, Siegel DA, Doney SC (2019) Modeling the impact of zooplankton diel vertical migration on the carbon export flux of the biological pump. Global Biogeochemical Cycles 33 , 181-199.
Benoit‐Bird KJ, Moline MA (2021) Vertical migration timing illuminates the importance of visual and nonvisual predation pressure in the mesopelagic zone. Limnology and Oceanography .
Bianchi D, Babbin AR, Galbraith ED (2014) Enhancement of anammox by the excretion of diel vertical migrators. Proceedings of the National Academy of Sciences 111 , 15653-15658.
Bianchi D, Galbraith ED, Carozza DA, Mislan K, Stock CA (2013a) Intensification of open-ocean oxygen depletion by vertically migrating animals. Nature Geoscience 6 545–548.
Bianchi D, Stock C, Galbraith ED, Sarmiento JL (2013b) Diel vertical migration: Ecological controls and impacts on the biological pump in a one‐dimensional ocean model. Global Biogeochemical Cycles27 , 478-491.
Biscontin A, Martini P, Costa R, et al. (2019) Analysis of the circadian transcriptome of the Antarctic krill Euphausia superba.Scientific Reports 9 , 1-11.
Bronk D, Steinberg D (2008) Nitrogen regeneration. Nitrogen in the marine environment 2 .
Buesseler KO, Lamborg CH, Boyd PW, et al. (2007) Revisiting carbon flux through the ocean’s twilight zone. Science316 , 567-570.
Burd AB, Hansell DA, Steinberg DK, et al. (2010) Assessing the apparent imbalance between geochemical and biochemical indicators of meso-and bathypelagic biological activity: What the@ $♯! is wrong with present calculations of carbon budgets? Deep Sea Research Part II: Topical Studies in Oceanography 57 , 1557-1571.
Chatterjee S, Ma K (2016) Circadian clock regulation of skeletal muscle growth and repair. F1000Research 5 .
Choi H, Fermin D, Nesvizhskii AI (2008) Significance analysis of spectral count data in label-free shotgun proteomics. Molecular & Cellular Proteomics 7 , 2373-2385.
Cholewa JM, Wyszczelska-Rokiel M, Glowacki R, et al. (2013) Effects of betaine on body composition, performance, and homocysteine thiolactone. Journal of the International Society of Sports Nutrition 10 , 39.
De Pitta C, Biscontin A, Albiero A, et al. (2013) The Antarctic krill Euphausia superba shows diurnal cycles of transcription under natural conditions. PloS One 8 , e68652.
de Sousa Abreu R, Penalva LO, Marcotte EM, Vogel C (2009) Global signatures of protein and mRNA expression levels. Molecular Biosystems 5 , 1512-1526.
Deutsch EW, Mendoza L, Shteynberg D, et al. (2015) Trans‐proteomic pipeline, a standardized data processing pipeline for large‐scale reproducible proteomics informatics.PROTEOMICS–Clinical Applications 9 , 745-754.
Doyle T, Jimenez‐Guri E, Hawkes WL, et al. (2022) Genome‐wide transcriptomic changes reveal the genetic pathways involved in insect migration. Molecular Ecology 31 , 4332-4350.
Eckelbarger KJ, Blades-Eckelbarger PI (2005) Oogenesis in calanoid copepods. Invertebrate Reproduction & Development 47 , 167-181.
Eng JK, Hoopmann MR, Jahan TA, et al. (2015) A deeper look into Comet—implementation and features. Journal of the American Society for Mass Spectrometry 26 , 1865-1874.
Eng JK, Jahan TA, Hoopmann MR (2013) Comet: an open‐source MS/MS sequence database search tool. Proteomics 13 , 22-24.
Fermin D, Basrur V, Yocum AK, Nesvizhskii AI (2011) Abacus: a computational tool for extracting and pre‐processing spectral count data for label‐free quantitative proteomic analysis. Proteomics11 , 1340-1345.
Forward R (1988) Diel vertical migration: zooplankton photobiology and behaviour. Oceanogr. Mar. Biol. Annu. Rev 26 , 1-393.
Gerson AR, DeSimone JG, Black EC, Dick MF, Groom DJ (2020) Metabolic reduction after long-duration flight is not related to fat-free mass loss or flight duration in a migratory passerine. Journal of Experimental Biology 223 , jeb215384.
Gutiérrez JS, Sabat P, Castañeda LE, et al. (2019) Oxidative status and metabolic profile in a long-lived bird preparing for extreme endurance migration. Scientific Reports 9 , 17616.
Haas BJ, Papanicolaou A, Yassour M, et al. (2013) De novo transcript sequence reconstruction from RNA-seq using the Trinity platform for reference generation and analysis. Nature Protocols8 , 1494-1512.
Häfker NS, Connan-McGinty S, Hobbs L, et al. (2022) Animal behavior is central in shaping the realized diel light niche.Communications Biology 5 , 562.
Häfker NS, Meyer B, Last KS, et al. (2017) Circadian clock involvement in zooplankton diel vertical migration. Current Biology 27 , 2194-2201. e2193.
Häfker NS, Teschke M, Last KS, et al. (2018) Calanus finmarchicus seasonal cycle and diapause in relation to gene expression, physiology, and endogenous clocks. Limnology and Oceanography 63 , 2815-2838.
Hays GC (2003) A review of the adaptive significance and ecosystem consequences of zooplankton diel vertical migrations.Hydrobiologia 503 , 163-170.
Hernández-León S, Koppelmann R, Fraile-Nuez E, et al. (2020) Large deep-sea zooplankton biomass mirrors primary production in the global ocean. Nature Communications 11 , 1-8.
Hochachka PW, Somero GN (2002) Biochemical Adaptation: Mechanism and Process in Physiological Evolution Oxford University Press, New York.
Huang S, Dou J, Li Z, et al. (2022) Analysis of Genomic Alternative Splicing Patterns in Rat under Heat Stress Based on RNA-Seq Data. Genes 13 , 358.
Hüppe L, Payton L, Last K, et al. (2020) Evidence for oscillating circadian clock genes in the copepod Calanus finmarchicus during the summer solstice in the high Arctic. Biology Letters 16 , 20200257.
Hwang D-S, Lee K-W, Han J, et al. (2010) Molecular characterization and expression of vitellogenin (Vg) genes from the cyclopoid copepod, Paracyclopina nana exposed to heavy metals.Comparative Biochemistry and Physiology Part C: Toxicology & Pharmacology 151 , 360-368.
Hwang DS, Lee KW, Lee JS (2009) Cloning and expression of vitellogenin 2 gene from the intertidal copepod Tigriopus japonicus .Annals of the New York Academy of Sciences 1163 , 417-420.
Ikeda T (2014) Respiration and ammonia excretion by marine metazooplankton taxa: synthesis toward a global-bathymetric model.Marine Biology 161 , 2753-2766.
Jombart T (2008) adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics 24 , 1403-1405.
Kassahn KS, Crozier RH, Pörtner HO, Caley MJ (2009) Animal performance and stress: responses and tolerance limits at different levels of biological organisation. Biological Reviews 84 , 277-292.
Kelly TB, Davison PC, Goericke R, et al. (2019) The importance of mesozooplankton diel vertical migration for sustaining a mesopelagic food web. Frontiers in Marine Science 6 , 508.
Kiko R, Hauss H (2019) On the estimation of zooplankton-mediated active fluxes in Oxygen Minimum Zone regions. Frontiers in Marine Science 6 , 741.
Kiørboe T, Møhlenberg F, Hamburger K (1985) Bioenergetics of the planktonic copepod Acartia tonsa : relation between feeding, egg production and respiration, and composition of specific dynamic action.Marine Ecology Progress Series , 85-97.
Kwong LE, Henschke N, Pakhomov EA, Everett JD, Suthers IM (2020) Mesozooplankton and Micronekton Active Carbon Transport in Contrasting Eddies. Frontiers in Marine Science 6 , 825.
Lampert W (1989) The adaptive significance of diel vertical migration of zooplankton. Functional Ecology 3 , 21-27.
Langfelder P, Horvath S (2008) WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9 , 559.
Langmead B, Salzberg SL (2012) Fast gapped-read alignment with Bowtie 2.Nature Methods 9 , 357-359.
Lee K-W, Hwang D-S, Rhee J-S, et al. (2008) Molecular cloning, phylogenetic analysis and developmental expression of a vitellogenin (Vg) gene from the intertidal copepod Tigriopus japonicus.Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular Biology 150 , 395-402.
Lee SR, Lee J-H, Kim AR, et al. (2016) Three cDNAs encoding vitellogenin homologs from Antarctic copepod, Tigriopus kingsejongensis: Cloning and transcriptional analysis in different maturation stages, temperatures, and putative reproductive hormones. Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular Biology192 , 38-48.
Lenz PH, Lieberman B, Cieslak MC, Roncalli V, Hartline DK (2021) Transcriptomics and metatranscriptomics in zooplankton: wave of the future? Journal of Plankton Research 43 , 3-9.
Li B, Dewey CN (2011) RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics 12 , 323.
Longhurst AR (1967) Vertical distribution of zooplankton in relation to the eastern Pacific oxygen minimum. Deep Sea Research14 , 51-63.
Maas AE, Blanco-Bercial L, Lo A, AM T, Timmins-Schiffman E (2018) Variations in copepod proteome and respiration rate in association with diel vertical migration and circadian cycle. The Biological Bulletin 235 , 30-42.
Maas AE, Gossner H, Smith MJ, Blanco‐Bercial L (2021) Use of Optical Imaging Datasets to Assess Biogeochemical Contributions of the Mesozooplankton. Journal of Plankton Research 43 , 475–491.
Maas AE, Liu S, Bolanos LM, et al. (2020) Migratory Zooplankton Excreta and its Influence on Prokaryotic Communities. Frontiers in Marine Science 7 , 573268.
MacLean B, Tomazela DM, Shulman N, et al. (2010) Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics 26 , 966-968.
Maier T, Güell M, Serrano L (2009) Correlation of mRNA and protein in complex biological samples. FEBS Letters 583 , 3966-3973.
Matos A, Ledoux J-B, Domínguez-Pérez D, Almeida D, Antunes A (2020) Omics advances in the study of zooplankton: Big data for small drifting organisms. Zooplankton Ecology , 264-277.
Mellacheruvu D, Wright Z, Couzens AL, et al. (2013) The CRAPome: a contaminant repository for affinity purification–mass spectrometry data. Nature Methods 10 , 730-736.
Niehoff B (2007) Life history strategies in zooplankton communities: The significance of female gonad morphology and maturation types for the reproductive biology of marine calanoid copepods. Progress in Oceanography 74 , 1-47.
Ohman M (1988) Behavioral responses of zooplankton to predation.Bulletin of Marine Science 43 , 530-550.
Oksanen J, Blanchet FG, Friendly M, et al. (2020) vegan: Community Ecology Package. R package version 2.5-7.
Olsen L, Thum E, Rohner N (2021) Lipid metabolism in adaptation to extreme nutritional challenges. Developmental Cell 56 , 1417-1429.
Park JS (1995) Biology of deep-sea calanoid copepod genus Pleuromamma with particular references to phylogeny, pore signatures, moulting and life history.
Payton L, Hüppe L, Noirot C, et al. (2021) Widely rhythmic transcriptome in Calanus finmarchicus during the high Arctic summer solstice period. Iscience 24 , 101927.
Piccolin F, Pitzschler L, Biscontin A, Kawaguchi S, Meyer B (2020) Circadian regulation of diel vertical migration (DVM) and metabolism in Antarctic krill Euphausia superba. Scientific Reports10 , 1-11.
Pinti J, Kiørboe T, Thygesen UH, Visser AW (2019) Trophic interactions drive the emergence of diel vertical migration patterns: a game-theoretic model of copepod communities. Proceedings of the Royal Society B 286 , 20191645.
Préville X, Salvemini F, Giraud S, et al. (1999) Mammalian small stress proteins protect against oxidative stress through their ability to increase glucose-6-phosphate dehydrogenase activity and by maintaining optimal cellular detoxifying machinery. Experimental Cell Research 247 , 61-78.
Reddy AB, Rey G (2014) Metabolic and nontranscriptional circadian clocks: eukaryotes. Annual review of biochemistry 83 , 165-189.
Reeve M (1981) Large cod‐end reservoirs as an aid to the live collection of delicate zooplankton. Limnology and Oceanography 26 , 577-580.
Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: a Bioconductor package for differential expression analysis of digital gene expression data.Bioinformatics 26 , 139-140.
Schnetzer A, Steinberg DK (2002) Active transport of particulate organic carbon and nitrogen by vertically migrating zooplankton in the Sargasso Sea. Marine Ecology Progress Series 234 , 71-84.
Seibel BA (2011) Critical oxygen levels and metabolic suppression in oceanic oxygen minimum zones. Journal of Experimental Biology214 , 326-336.
Seibel BA, Andres A, Birk MA, et al. (2021) Oxygen supply capacity breathes new life into critical oxygen partial pressure (Pcrit). Journal of Experimental Biology 224 .
Semmouri I, De Schamphelaere KA, Van Nieuwerburgh F, et al.(2020) Spatio-temporal patterns in the gene expression of the calanoid copepod Temora longicornis in the Belgian part of the North Sea.Marine Environmental Research 160 , 105037.
Shoemaker KM, McCliment EA, Moisander PH (2020) Copepod-associated gammaproteobacterial alkaline phosphatases in the North Atlantic Subtropical Gyre. Frontiers in Microbiology 11 , 1033.
Simão FA, Waterhouse RM, Ioannidis P, Kriventseva EV, Zdobnov EM (2015) BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics 31 , 3210-3212.
Simioni C, Zauli G, Martelli AM, et al. (2018) Oxidative stress: role of physical exercise and antioxidant nutraceuticals in adulthood and aging. Oncotarget 9 , 17181.
Steinberg DK, Carlson CA, Bates NR, et al. (2000) Zooplankton vertical migration and the active transport of dissolved organic and inorganic carbon in the Sargasso Sea. Deep Sea Research Part I47 , 137-158.
Steinberg DK, Landry MR (2017) Zooplankton and the Ocean Carbon Cycle.Annual Review of Marine Science 9 , 413-444.
Steinberg DK, Stamieszkin K, Maas AE, et al. (2023) The outsized role of salps in carbon export in the subarctic Northeast Pacific Ocean.Global Biogeochemical Cycles , e2022GB007523.
Steinberg DK, Van Mooy BAS, Buesseler KO, et al. (2008) Bacterial vs. zooplankton control of sinking particle flux in the ocean’s twilight zone. Limnology and Oceanography 53 , 1327-1338.
Svetlichny LS, Hubareva ES (2005) The energetics of Calanus euxinus : locomotion, filtration of food and specific dynamic action.Journal of Plankton Research 27 , 671-682.
Tan MH, Smith SR, Hixson KK, et al. (2020) The Importance of Protein Phosphorylation for Signaling and Metabolism in Response to Diel Light Cycling and Nutrient Availability in a Marine Diatom.Biology 9 , 155.
Tan S, Wang W, Tian C, et al. (2019) Heat stress induced alternative splicing in catfish as determined by transcriptome analysis.Comparative Biochemistry and Physiology Part D: Genomics and Proteomics 29 , 166-172.
Tarrant AM, Baumgartner MF, Hansen BH, et al. (2014) Transcriptional profiling of reproductive development, lipid storage and molting throughout the last juvenile stage of the marine copepodCalanus finmarchicus . Frontiers in zoology 11 , 1.
Tarrant AM, McNamara-Bordewick N, Blanco‐Bercial L, Miccoli A, Maas AE (2021) Diel metabolic patterns in a migratory oceanic copepod.Journal of Experimental Marine Biology and Ecology 545 .
Tarrant AM, Nilsson B, Hansen BW (2019) Molecular physiology of copepods-from biomarkers to transcriptomes and back again.Comparative Biochemistry and Physiology Part D: Genomics and Proteomics 30 , 230-247.
Teschke M, Wendt S, Kawaguchi S, Kramer A, Meyer B (2011) A circadian clock in antarctic krill: an endogenous timing system governs metabolic output rhythms in the euphausid species Euphausia superba .PloS One 6 , e26090.
Thor P (2000) Relationship between specific dynamic action and protein deposition in calanoid copepods. Journal of experimental marine biology and ecology 245 , 171-182.
Thurley K, Herbst C, Wesener F, et al. (2017) Principles for circadian orchestration of metabolic pathways. Proceedings of the National Academy of Sciences 114 , 1572-1577.
Tian Y, Wen H, Qi X, et al. (2020) Alternative splicing (AS) mechanism plays important roles in response to different salinity environments in spotted sea bass. International Journal of Biological Macromolecules 155 , 50-60.
Timmins-Schiffman EB, Crandall GA, Vadopalas B, et al. (2017) Integrating discovery-driven proteomics and selected reaction monitoring to develop a noninvasive assay for geoduck reproductive maturation.Journal of Proteome Research 16 , 3298-3309.
Torgersen T (2003) Proximate causes for anti-predatory feeding suppression by zooplankton during the day: reduction of contrast or motion—ingestion or clearance? Journal of Plankton Research25 , 565-571.
Valdés V, Carlotti F, Escribano R, et al. (2018) Nitrogen and phosphorus recycling mediated by copepods and response of bacterioplankton community from three contrasting areas in the western tropical South Pacific (20° S). Biogeosciences 15 , 6019-6032.
Valdés VP, Fernandez C, Molina V, Escribano R, Joux F (2017) Dissolved compounds excreted by copepods reshape the active marine bacterioplankton community composition. Frontiers in Marine Science 4 , 343.
Vogel C, Marcotte EM (2012) Insights into the regulation of protein abundance from proteomic and transcriptomic analyses. Nature reviews genetics 13 , 227-232.
Welkie DG, Rubin BE, Diamond S, et al. (2019) A hard day’s night: cyanobacteria in diel cycles. Trends in Microbiology 27 , 231-242.
Wickham H (2007) Reshaping data with the reshape package. Journal of statistical software 21 , 1-20.
Wickham H (2016) ggplot2: elegant graphics for data analysisspringer.
Young MD, Wakefield MJ, Smyth GK, Oshlack A (2012) goseq: Gene Ontology testing for RNA-seq datasets.
Zhang S, Wang S, Li H, Li L (2011) Vitellogenin, a multivalent sensor and an antimicrobial effector. The international journal of biochemistry & cell biology 43 , 303-305.
Zhang X, Yuan J, Li F, Xiang J (2021) Chitin Synthesis and Degradation in Crustaceans: A Genomic View and Application. Marine Drugs19 , 153.
Data Accessibility
Cruise data and metadata including net tow data and environmental conditions are available at BCO-DMO project number 764114. Transcriptome and raw gene expression data are available at NCBI PRJNA766852. This Transcriptome Shotgun Assembly version described in this paper is the first version, GJVP01000000. Raw mass spectrometry and supporting files for analysis were deposited to the ProteomeXchange Consortium via PRIDE with the dataset identifier PXD021156 (reviewer log-in credentials:reviewer83975@ebi.ac.uk, password = 7hySyEDg )
Author Contributions
A.E.M, L.B.-B., A.T. and E.T.-S. conceived and planned the experiments. A.E.M, L.B.-B., A.T., B.N. and E.T.-S. participated in the cruise execution and animal collection. A.T. and E.T.-S. carried out laboratory work. A.E.M, J.P. and E.T.-S. were responsible for bioinformatics analysis. A.E.M and E.T.-S. took the lead in writing the manuscript. All authors provided critical feedback and helped shape the research, analysis and manuscript.