A major goal in biology is to understand how evolution shapes variation in individual life histories. Genome-wide association studies have been successful in uncovering genome regions linked with traits underlying life history variation in a range of species. However, lack of functional studies of the discovered genotype-phenotype associations severely restrains our understanding how alternative life history traits evolved and are mediated at the molecular level. Here, we report a cis-regulatory mechanism whereby expression of alternative isoforms of the transcription co-factor vestigial-like 3 (vgll3) associate with variation in a key life history trait, age at maturity, in Atlantic salmon (Salmo salar). Using a common-garden experiment, we first show that vgll3 genotype associates with puberty timing in one-year-old salmon males. By way of temporal sampling of vgll3 expression in ten tissues across the first year of salmon development, we identify a pubertal transition in vgll3 expression where maturation coincided with a 66% reduction in testicular vgll3 expression. The late maturation allele was not only associated with a tendency to delay puberty, but also with expression of a rare transcript isoform of vgll3 pre-puberty. By comparing absolute vgll3 mRNA copies in heterozygotes we show that the expression difference between the earlyand late maturity alleles is largely cis-regulatory. We propose a model whereby expression of a rare isoform from the late allele shifts the liability of its carriers towards delaying puberty. These results exemplify the potential importance of regulatory differences as a mechanism for the evolution of life history traits.

Crosses used in the study

We controlled-crossed eight 2x2 factorial matings of unrelated individuals that each included one vgll3*EE and one vgll3*LL male and female, thereby resulting in 32 families with reciprocal vgll3 genotypes EE, EL, LE and LL (allele order denotes female and male parents of origin respectively). We used parents from the “Oulujoki” broodstock, which is a mixture of several salmon populations from northern Baltic rivers and is maintained by the Natural Resources Institute Finland (LUKE). All parents were selected for having a homogenous age-at-maturity-associated haplotype encompassing the vgll3 (chromosome 25 paralog) coding sequence, the top non-coding association, and the coding region of the adjacent akap11 gene (collectively referred to as “vgll3genotype”).

Husbandry design and conditions

Following fertilization in October 2017, familial egg batch replicates were raised in two egg incubators until first feeding (~ 2 months post-hatching) with mesh-separators to contain families in individual compartments. Families were assigned to compartments randomly. At first feeding, familial batches were combined and 48 individuals per family were randomly selected and distributed evenly across eight 0.25 m³ tanks using recirculated water i.e. on average, each tank included six individuals per family. Prior to tank allocation, each individual was tagged with visible elastomer (VIE) at the base of the caudal fin to enable identification of the vgll3 genotype of an individual by a fluorescent color (initial n=1300). Light cycle and water temperature (min=6.3, max=17.7 °C) during husbandry corresponded to the local seasonal cycle at 61°N. Fish were fed first with live Artemia for three days, after which we fed commercial aquaculture feed ad libitum (Raisio Baltic Blend) with pellet diameter increasing over time according to fish size. Temperature, oxygen and nitrogen-component levels were monitored on a regular basis. Experimentation was conducted according to the Finnish Government Decree on the Protection of Animals Used for Scientific or Educational Purposes (564/2013), which implements EU directive 2010/63/EU. The experiments in this study were approved by the Project Authorisation Board (ELLA) on behalf of the Regional Administrative Agency for Southern Finland (ESAVI) under experimental license ESAVI/2778/2018.

Individual tagging, family assignment and morphological measurements

At eight months post-hatching and before the onset of the first breeding season we tagged remaining fish using passive integrated transponder (PIT) tags and collected fin-clips. We extracted DNA using standard chelex or salt extraction methods, genotyped all fish for 141 informative SNPs, assigned their genotypic sex based on genotyping coverage of the salmon sex determining locus sdy and assigned individuals to families using the likelihood method implemented in SNPPIT. We recorded the length and mass of all fish monthly after PIT-tagging. We non-lethally assessed the maturity status of all males by testing for the presence of running milt at nine months post-hatching. Morphological measurement and non-lethal maturation assessment were performed in a random order such that the assessor did not have knowledge of vgll3 genotype.

Temporal sampling, RNA extraction and RT-ddPCR

We sampled individuals with vgll3*EE, vgll3*EL, vgll3*LE and vgll3*LL monthly (3-32 in total per genotype per time point) and assessed sex, gonad size and maturity by dissection. We sacrificed the animals by MS-222 overdose and dissected ten tissues (brain (timepoints 12-13 months post hatch); heart, liver, muscle (timepoints 3-13 months post hatch), adipose (timepoints 5-13 months post hatch); immature ovary (timepoints 6-13 months post hatch); immature testes (timepoints 7-13 months post hatch); mature testes (timepoints 8 and 10-12 months post hatch); stripped mature testes (timepoints 12-13 months post hatch) and ejaculate (timepoint 13 months post hatch)). As immature male gonads had no recordable weight on a scale with a detection limit of 10 mg thus preventing the use of gonadal-somatic index, we classified each sacrificed male fish visually as immature (not enlarged, transparent color gonads), pubertal (slightly enlarged, transparent or white color gonads) or mature (enlarged white color gonads). We flash froze the samples on liquid nitrogen and stored the samples in -80 °C. We homogenized the tissues using an OMNI Bead Ruptor Elite machine, in 2 ml or 7 ml (for mature gonads) with 2.4 mm steel beads. We used Macherey-Nagel NucleoSpin RNA kits in single and in 96-well plate formats to extract total RNA and treated the samples with additional DNase (Invitrogen Turbo Dnase) to remove any residual DNA. We verified total RNA quality using Agilent BioAnalyzer for a random set of 24 samples and found all samples to have RIN>9. We measured total RNA yield using ThermoFischer Qubit and Quant-it reagents, diluted the samples to working concentrations and re-measured RNA concentrations. For all gonad samples we repeated the RNA concentration measurement once more and calculated the mean of the duplicate measurements, which we used in data normalization. We used ~20 ng (2-8 months post-hatch) or ~50 ng (8-14 months post-hatch) of total RNA as template in one-step reverse transcription droplet digital PCR (RT-ddPCR) (BioRad) with TaqMan probes specific to vgll3 alleles to quantify vgll3 absolute expression. Measurement of amh and igf3 expression was performed as above, with the exception of using ~2 ng of total RNA template and TaqMan probes that were labeled with alternative fluorophores and assayed in the same RT-ddPCR reactions. We verified the specificity of vgll3-genotype specific TaqMan assays with genomic and synthetic DNA templates and optimized melting temperatures for all assays using a gradient. Genomic DNA could not be used as control for the vgll3 5’ UTR assay because the assay spans an intron-exon boundary. Genomic and synthetic DNA controls and null control were included in each RT-ddPCR vgll3 run, while genomic DNA and null control were included in Amh and Igf3 RT-ddPCR runs. 

Analyses as per the associated article can be reproduced with the R MarkDown scripts provided within this accession.