The importance of biological time keeping is apparent across all organisms, from bacteria to humans (Wulund and Reddy, 2015). These time-dependent adaptations can last seconds or minutes and recur throughout the day (ultradian) or endure days or months (infradian). They are an evolutionary trait which enables all living organisms to maximise fitness in anticipation of endogenous (molecular and cellular) and external stimuli or zeitgebers (Andreani et al., 2015; Sánchez-Vázquez et al., 2021). Circadian rhythms are endogenous oscillatory expression of genes with a periodicity of approximately 24-h. These rhythms are the expression of endogenous clocks which synchronises biochemical, physiological, and behavioural responses enabling organisms to respond to diel environmental changes (Li et al., 2015). The transcription of 43% of all protein-coding genes in mice displayed circadian rhythms across numerous organs (Zhang et al., 2014). Clock genes are of particular interest in salmonids due to their association with physiological traits such as reproduction migration and smoltification (Leder et al., 2006; O'Malley et al., 2007, 2010; Paibomesai et al., 2010). Allelic diversity and variation in length polymorphism of the clock PolyQ domain was reported in four Pacific salmon species (chinook, chum, coho, and pink) with overlapping geographical ranges and diversity in spawning times. This implicates clock gene duplicate may be involved in the seasonal and geographical variation in reproduction (O'Malley et al., 2007, 2010). In addition, a copy of the gene clock has been localised to a quantitative trait locus (QTL) responsible for 20–50% of the variation in spawning dates in female rainbow trout (Leder et al., 2006).

The circadian clock consists of intracellular transcriptional-translational feedback loops (TTFL) composed of core clock genes and stabilising accessory loop genes, which drives the rhythmic accumulation of downstream outputs, or clock-controlled genes (Reppert and Weaver, 2001; Partch et al., 2014). Circadian mechanism is highly conserved across animal species (Lowrey and Takahashi, 2011). However, deciphering the circadian clock mechanism in fish is complex. Clock and clock-related genes which are found in single copies in invertebrates such as clock, period and cryptochrome are duplicated in vertebrates (Tauber et al., 2004). In addition, salmonids were subjected to two rounds of whole-genome duplication (WGD) events (Ts3R, teleost-specific third whole-genome duplication, 320 million years ago, and Ss4R, salmonid-specific fourth whole-genome duplication, 80 million years ago) resulting in an abundance of circadian related genes (Lien et al., 2016; Huang, 2018).

The molecular mechanisms underlying circadian rhythmicity have been characterised in several model animal species including the fruit fly (Drosophila melanogaster), mice (Mus musclus), and humans (Homo sapiens), with relatively limited work undertaken in teleosts (Wang, 2008b; Huang, 2018) predominantly centred on zebrafish (Danio rerio) (Cahill, 2002; Vallone et al., 2007). The circadian system comprises all the different components by which light is perceived by the organism and is transformed into a nervous or hormonal signal (Migaud et al., 2010a). Therefore, manipulations of photic inputs and cues impact rhythms, which can be commercially exploited for aquaculture production (Migaud et al., 2010a). Research based on zebrafish has been fundamental in describing and broadly characterising clock genes and circadian rhythmicity as many findings are applicable to numerous vertebrate species. However, fish models have not yet significantly contributed to our understanding of core clock mechanisms and circadian clock control of fish physiology (Frøland Steindal and Whitmore, 2019). Teleosts represent the largest and most diverse group of vertebrates, with over 30,000 species identified to date. Each species possesses distinct characteristics and displays a considerable amount of anatomical and physiological plasticity. This is arguably the direct result of exposure to multiple, variable selection pressures caused by the highly dynamic aquatic environments they have inhabited throughout their evolutionary development (Bone, 2019). Alongside selection pressures, multiple rounds of WGD have had a large impact upon the evolution of lineages to date, with the retention of resultant gene duplicates being biased with regard to gene function (Brunet et al., 2006). Gene duplication through WGD led to redundant genes either los, non-functionalisation, with a different function than the ancestral gene (sub-functionalisation) or which acquired new functions, neo-functionalisation, (Pasquier et al., 2016), thus resulting in genome reshaping (Inoue et al., 2015). An example of this is the salmonid specific WGD (Ss4R) which preceded the origin of anadromy in salmonids, an important milestone in the evolutionary development of salmonid migration (Alexandrou et al., 2013). Whilst WGD events led to gene duplication, these duplicated genes typically resolved over time (Inoue et al., 2015). In rainbow trout, 80–100 million years post Ss4R, 48% of the pre-duplication ancestral genes were retained as duplicates, the remainder of the genes underwent fractionation and the duplicated protein-coding genes were lost (Berthelot et al., 2014; Lien et al., 2016). Analysis of duplicate retention in Atlantic salmon identified that 20% of duplicates from Ts3R and 55% of duplicates from Ss4R were retained as functional copies, with the prominent mechanism for duplicate loss being pseudogenisation (Lien et al., 2016). While there is a general lack of clarity surrounding the duplication and retention of functional genes post WGD, there is an unusually large complement of clock genes in salmonids (West et al., 2020).

Salmonids are amongst the most widely studied groups of fish species both scientifically and commercially, as many species of salmonid are of significant economic, societal, and environmental importance (Thorgaard et al., 2002). Many of the species within the 11 genera of the Salmonidae (Nelson et al., 2016) are of great commercial value and contribute significantly to both local and global economies through aquaculture, wild stock fisheries and recreational sport (Reppert and Weaver, 2001; Davidson et al., 2010; Frøland Steindal and Whitmore, 2019). Unlike zebrafish, salmonid species are highly seasonal in their physiology, including migration, smoltification and reproduction. Lighting and temperature manipulations are routinely used by industry to manipulate commercial broodstock ovulatory rhythm and smoltification, enabling year-round production (Migaud et al., 2013). However, understanding the intricate interactions between zeitgebers, circadian rhythmicity, seasonality, and the control of biochemical, physiological, and behavioural rhythms is complex (Migaud et al., 2010b). The completion and publication of salmonid genomes (Salmo salar, Salvelinus alpinus, Oncorhynchus mykiss, Oncorhynchus kisutch, and Oncorhynchus tshawytscha to date) as part of the Functional Annotation of All Salmonid Genomes (FAASG) project (Macqueen et al., 2017), alongside RNAseq have provided great tools to study clock genes of salmonids.

The aim of this study was to identify the full complement of clock genes in Atlantic salmon (Salmo salar) in comparison to other commercially important salmonid species and evaluate the expression patterns of the identified genes. To do so, phylogenetic analysis of clock gene ohnologs [functional product of WGD event (Ohno, 1970)] has been performed to classify and name salmonid clock genes based on published zebrafish references and nomenclature. This was confirmed by a transcriptomic approach looking into gene expression over 24-h in freshwater salmon kept under a controlled lighting regime. This study provides a new nomenclature for salmon clock genes that will serve as a tool for further circadian research in salmon.

In the present study, we have identified and characterised 48 core clock and accessory loop genes in silico in several salmonid species with published genomes using the latest common ancestors as reference points. For each of the salmonid species with published genomes, the core clock and accessory loop genes were identified and characterised in silico. There is a differential retention of genes originating in the latest common ancestors, E. lucius and D. rerio. In addition to this, there are numerous ohnologs (Ohno, 1970) which are consistent with the salmonid specific whole genome duplication event Ss4R (Lien et al., 2016). There is also evidence of differential gene loss in salmonids based on the current gene annotations (Table 1). Findings from this study further highlight the rich complexity of core clock genes previously outlined in S. salar (West et al., 2020), and is displayed across a wider complement of salmonid species. The gene families identified differ slightly to those explored by West et al. (2020) as the focus of this study was to identify all members of gene families clock, arntl, period, cryptochrome, ror, nr1d, and csnk1. Identified genes common to both studies coincide with one another and paralog pairs identified using ML phylogenetic alignments are also identified as paralogs (West et al., 2020).

In comparison to D. rerio with 29 reference clock genes (Wang, 2008b; Huang et al., 2010; Toloza-Villalobos et al., 2015), most clock genes identified in salmonids appear to have been duplicated. In addition to these lineage-specific duplications there is also evidence of differential ohnolog loss in some species of salmonids based on the current genome annotations, per1b, cry2, and cry5 appear to have been retained across all species from the latest common ancestors, per3 appears to only be retained in S. salar and S. alpinus and cry4 appears to only be retained in S. salar. Whereas, clock1b, cry1b, cry3a, csnk1da all appear to have been lost in salmonids after the latest common ancestor E. lucius. Some species of salmonids have better genome annotations than others, and currently it is not possible to identify the effects of the WGD for every core clock gene family in O. tshawytscha at present. As expected across most of the gene families except where there is limited information, the Oncorhynchus spp. tend to appear together on the same clade with S. salar and S. alpinus commonly grouped together in a sister clade. A typical example of the clock gene families identified is the cryptochrome family. cry1a is one of the major clades in the cryptochrome family, as a result of duplication there are three main paralogous subfamilies cry1a, cry1b and cry1c. cry1a and cry1b are phylogenetically related to tetrapods, with cry1a being the most conserved of the two subfamilies (Mei et al., 2015). Zebrafish possess both cry1a and cry1b however, all the salmonid species only appear to have inherited cry1a which has been duplicated as a result of the WGD, resulting in two paralogs cry1a.1 and cry1a.2. Except for O. tshawytscha (for which there is not currently enough information in the genome annotation) cry3b is duplicated in all salmonid species with an apparent gene loss in both O. kisutch and O. mykiss. This is suggestive of the non-functualisation and subsequent loss of cry3b.2 in the Oncorhynchus spp. As a result of the Ss4R WGD clock most genes appear to be duplicated in salmonids with some exceptions that appear to have been lost in several of the salmonids. As annotations improve, it is thought that additional gene losses will become more apparent. This study gives insight into the post duplication effects on clock gene family members and provides a fundamental tool for further circadian work in salmonids.

Clock genes interact with each other, generating oscillations in gene expression. Their underlying principle is to create successive gene activation in the form of a cycle, forming an autoregulatory feedback loop which perpetually cycles approximately every 24 h (Ripperger and Albrecht, 2009). This in turn influences downstream targets, whose time-of-day specific expression is determined by the central circadian mechanism (Ripperger and Albrecht, 2009). This study confirmed the expression of the identified clock genes in the brain of S. salar smolts over a 24-h period and showed that 11 out of the 48 core clock genes were highly significantly (p < 0.001) rhythmically expressed. We used a two complementary approach, RAIN which allowed us to detect accurately rhythms in time series irrespectively of the shape of the expression pattern, and JTK with a focus on circadian expression patterns. The vast majority of the JTK findings are included in the RAIN results (Supplementary Table 3). This allowed us to distinguish between rhythmic expressed genes and circadian pattern expressed genes.

The period family is particularly interesting, as except for one gene (per3) all the genes were significantly rhythmically expressed. A difference in acrophase between paralogs was also identified. There was a clear difference in expression pattern observed between per1a paralogs (per1a.1, per1a.2) and per1b, with per1a peaking at 06:00 displaying a positive phase shift in comparison to per1b which peaks at 08:30. This positive phase shift coincides with results observed in zebrafish, it was reported that per1a and per1b paralogs displayed a shifted phase of gene expression in zebrafish with per1a peaking 4 h prior to per1b under a 12:12 LD lighting regime (Amaral and Johnston, 2012). In zebrafish, per1 paralogs were shown to display a distinct difference in spatial and temporal expression in the brain, providing strong evidence for the duplicate pair to have undergone sub- or neo-functionalisation (Wang, 2008a). The duplicated Atlantic salmon per1a paralogs (per1a.1 and per1a.2) share the same periodicity and acrophase, thus indicating that both genes may share a similar functionality. This indicates that the paralogs may have undergone sub-functionalisation—in which the ancestral functions have become partitioned and each paralog potentially has particular adaptations for different tissues, developmental stages or environmental conditions (Innan, 2009). In mammalian literature, it is widely reported that Cry1/2 form heterodimers with Per1/2/3 (Rosensweig et al., 2018). However, in this study, the only highly significantly rhythmically entrained cryptochrome genes was the cry3b.2 paralog which peaked two h before the closest period family member (per2b) at 20:30, in agreement with the pattern of cry2 expression identified in zebrafish 12-h after light onset (Hirayama et al., 2019). In mammals, Per2 is said to repress Nr1d2 transcription thus upregulating arntl1 transcription (Chiou et al., 2016), in keeping with these findings in this study nr1d2b.2 peaks at 08:30, around 10 h before peak expression of arntl1a.1/1a.2/1b.1 (18:00) and per2b peaks at 01:00. This suggests that the relationship between clock gene paralogs in salmonids may be similar to that of those in mammalian species.

How salmonids fit into the widely accepted mechanism for circadian rhythmicity is difficult to evaluate due to the limited understanding surrounding circadian mechanisms in fish in general and the complexity are yet to be exploited to the full potential (Frøland Steindal et al., 2018), there are still many unknowns surrounding the mechanism in salmonids as it has been far less explored than other species of teleost. In comparison to species such as D. melanogaster, M. musclus, H. sapiens, and even D. rerio there has been limited work undertaken in salmonids to further identify clock genes and their respective roles within the circadian system (Frøland Steindal et al., 2018). So far, the expression of individual clock genes has been previously investigated in salmonids (Davie et al., 2009; Huang et al., 2010; West et al., 2020). Clock gene member identification has previously been hampered by paralogs with high sequence similarity, which does not easily allow for individual identification by qPCR. Individual gene identification using RNA sequencing has allowed for paralogs to be better classified and individual gene expression patterns ascertained. Although, single gene duplication events in mammalian species have enabled the evolution of specialisation and regulatory sophistication in the temporal regulation of local physiology (Looby and Loudon, 2005); the core clock genes associated with circadian rhythmicity remains largely conserved across a diverse range of organisms spanning vast evolutionary time periods (Bell-Pedersen et al., 2005; Cox and Takahashi, 2019). This study supports that whilst the compliment of clock genes is far richer in salmonids, the function of core clock genes remains conserved and are therefore likely function similarly to other more studied organisms. Some salmonid species display reciprocal gene losses post Ss4R, but many of clock genes identified appear to be duplicated in the salmonid species investigated. The existence of such paralogs whilst increasing genetic complexity may enable gene family members to specialise and extend their ancestral role, which can lead to a shift in the identity of components of the molecular clock (Looby and Loudon, 2005; West et al., 2020). Ambient temperature affects gene expression and physiology in ectotherms; in zebrafish, the clock seems to be temperature-compensated, changing the amplitude of some critical clock genes (Lahiri et al., 2005). However, the influence of temperature on the clock system in salmonids remains poorly studied and should be investigated further.

Whilst this study has furthered the identification of clock gene family members in central salmonids, additional work is required to further elucidate the complexity of the circadian mechanism and how the complement of clock genes identified individually function as components of this mechanism. It is important to note that alongside individual variability, the whole brain being analysed will have influenced gene expression levels and rhythmicity due to the highly decentralised organisation of clock genes in teleosts and localisation of clock gene expression in specific regions of the brain (Moore and Whitmore, 2014). This study provides a fundamental tool to explore the role of the enriched number of clock genes related to seasonal driven life history transition in salmonids.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ebi.ac.uk/ena, PRJEB41327.

The animal study was reviewed and approved by UK Animals (Scientific Procedures) Act 1986 Amended Regulations (SI 2012/3039) Animal Welfare and Ethical Review Body of the University of Stirling (AWERB/19 20/097/).

CB, JH, MB, and HM: study conception and design. CB and HM: sample and data acquisition. CB, MB, and ME: data analysis. HM and JH: funding acquisition. CB, HM, ME, and MB: drafting and editing of the manuscript. All authors contributed to the article and approved the submitted version.

The study was funded by the University of Stirling PhD match funding scheme, the UKRI project ROBUSTSMOLT (BB/S004432/1) and the Research Council of Norway (The effect of narrow banded LED light on development and growth performance, grant no. 254894). For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.