Fig. 1 Experimental space-for-time approach. We characterized DNA methylation profiles (via RRBS) and whole genomes [whole-genome sequencing (WGS)] of fish from three populations of wild-caught three-spined sticklebacks locally adapted to 6 (blue; n = 15), 20 (green; n = 16), and 33 (yellow; n = 15) PSU. We also bred and acclimated sticklebacks from the mid-salinity location (20 PSU) within one (“within-generational”) or over two (“transgenerational”) generations to decreased (6 PSU) or increased (33 PSU) salinity while maintaining a control group at its original salinity (n = 11 to 12 per group; see details in the figure). Differential methylation within and across generations was assessed and compared to natural populations locally adapted to the corresponding salinity, serving as the hypothetical future DNA methylation state to capture long-term adaptation processes.
Fig. 2 Graphical summary of the main results. We used the Baltic Sea salinity gradient to study the role of DNA methylation in local salinity adaptation and the response to salinity change in a space-for-time approach. To assess the potential future acclimatization and adaptation processes of the natural stickleback population from 20 PSU (KIE; green) to the predicted desalination (63), we compared differences in DNA methylation at CpG sites between wild-caught and laboratory-bred sticklebacks. Following the experiment timeline (bottom), we compared methylation levels of the experimental control group from 20 PSU to within- and transgenerational acclimation of 20 PSU sticklebacks to 6 PSU (DNA from left to right). The population locally adapted to 6 PSU serves as the hypothetical future state in which salinities will decrease (blue; DNA on the right). The three main results are written in the circles with schematically and horizontally corresponding DNA methylation changes. (i) Sixty-three percent of the DMS between the populations remained stable under experimental salinity change. (ii) The direction of experimental methylation change was dependent not only on the treatment but also on the degree of genetic differentiation between the populations [see Fig. 4 (A to D) for results]. (iii) Transgenerational salinity acclimation shifted DNA methylation patterns closer to the anticipated adaptive state found in the hypothetical future population [see Fig. 4 (E to H) for results]. For clarity, only one (6 PSU) of the two foreign salinity regimes tested (6 and 33 PSU) is shown. The results for the experimental fish acclimated to 33 PSU were very similar (see Fig. 1 for full experimental design and Fig. 4 for results).
Fig. 3 Gene Ontology terms for biological processes and molecular functions. Gene Ontology (GO) terms for biological processes and molecular functions under salinity increase (20 versus 33 PSU; yellow) and decrease (20 versus 6 PSU; blue) associated with pop-DMS are presented. The graph is split into GO terms associated with pop-DMS from natural stickleback populations across a salinity cline (wild) and their experimental inducibility (inducible and stable) in a two-generation acclimation experiment. The size of the circles refers to the number of genes of this term in the groups (in %), and the transparency refers to the false discovery rate–corrected P value (darker circles refer to a lower adjusted P value). This subset is filtered for GO terms including the following keywords: “channel,” “transport,” “water,” “chloride,” “potassium,” “homeostasis,” “ion-dependent,” “urine,” “ATP” (adenosine 5′-triphosphate), and “metabolic”; see fig. S2 for the full figure. cGMP, guanosine 3′,5′-monophosphate; cAMP, adenosine 3′,5′-monophosphate; G protein, heterotrimeric GTP-binding protein.
Fig. 4 The duration of acclimation (within-generational versus transgenerational) and level of genomic differentiation between populations influence DNA methylation at inducible sites. (A and B) Mean FST values for inducible pop-DMS (with a ± 5-kb window) under experimental salinity decrease (top; blue) and increase (bottom; yellow) that shifted methylation levels toward the values observed in either the field (expected) or the opposite direction (opposite). A randomization test (with 10,000 bootstraps) was performed for the difference between expected and opposite mean FST value (δ.mean.FST = expected mean FST – opposite mean FST) (C and D). Under the one-tailed hypothesis of increased genetic differentiation at opposite sites and an α of 0.05, the P value was calculated as values smaller than the true difference divided by 10,000 bootstraps. In (E to H), the y axis shows the percentage match between the within- and transgenerational acclimation groups in relation to the methylation differentiation level found in natural populations at inducible pop-DMS. This value was obtained by calculating the difference between the methylation change in the experiment (meth.diff.exp in %; control versus within-generational or control versus transgenerational) and the difference in methylation between natural populations (meth.diff.wild in %) as δ.meth.diff = 100 − (meth.diff.wild − meth.diff.exp). Mean values ± 95% confidence interval are shown for within- and transgenerational acclimation to decreased and increased salinity at expected and opposite inducible sites. Colors refer to the direction of DNA methylation change (hypomethylation or hypermethylation). Values closer to 100 indicate a shift in methylation pattern toward adaptive methylation levels found in natural populations, and asterisks indicate the significance level (***P ≤ 0.001 and **P ≤ 0.01) for the comparison between within- and transgenerational acclimation. “Main effect” refers to an effect of acclimation (within- or transgenerational), and “interaction effect” refers to an interaction of acclimation and methylation direction (hypo- or hypermethylation).
Fig. 5 Effects of salinity acclimation on fitness-correlated factors. For all five acclimation groups [control group (20 PSU), within-generational, and transgenerational acclimation to 6 or 33 PSU], survival rates in percent (A), standard length in centimeters (B), hepatosomatic index (C), and total weight in grams (D) are displayed. Letters indicate significant differences resulting from Tukey post hoc tests (table S3). HSI, hepatosomatic index.
- Table 1 Differentially methylated genes across natural populations along a salinity cline.
Genes derived from DNA methylation comparisons between natural populations associated with ≥10 pop-DMS [decreased salinity: KIE (20 PSU) versus NYN (6 PSU); increased salinity: KIE (20 PSU) versus SYL (33 PSU)]. Ensembl gene ID and name as well as the position on the chromosome are listed. The numbers refer to the numbers of DMS in the population comparison (wild). These DMS were classified into inducible, inconclusive, and stable sites according to their behavior in a two-generation salinity acclimation experiment with laboratory-bred sticklebacks from the mid-salinity population (20 PSU) exposed to experimental salinity increase or decrease (33 and 6 PSU, respectively). Furthermore, inducible sites were distinguished whether they matched methylation levels of the locally adapted population (expected) or not (opposite). Genes written in bold vary in both population comparisons. We used a Fisher’s exact test to assess whether pop-DMS associated to the same gene are correlated in their response to experimental salinity change (nonrandom distribution among the categories stable, inducible, and inconclusive) and reported corresponding P values. For a full table on all genes associated with one or more pop-DMS, see table S2 (A and B).
Ensembl gene ID Chromosome Start
positionEnd
positionGene
nameWild Inducible Expected
inducibleOpposite
inducibleStable Inconclusive Fisher’s
exact (P)Salinity decrease: ENSGACG00000008328 Chr10 12860144 12863850 si:dkey-166 k12.1 24 0 0 0 9 15 0.005 ENSGACG00000019416 Chr7 4451892 4453656 HMX1 ortholog 17 0 0 0 9 8 0.033 ENSGACG00000013229 Chr18 15327717 15352321 15 0 0 0 3 12 0.011 ENSGACG00000017287 Chr3 13454527 13465167 mmp16b 12 0 0 0 12 0 0.001 ENSGACG00000017584 Chr3 14690814 14694448 CCNY 12 12 12 0 0 0 0.001 ENSGACG00000018249 Chr4 12141625 12143011 si:ch211-153b23.5 12 1 1 0 3 8 0.188 ENSGACG00000008034 Chr6 9368187 9380941 11 10 10 0 0 1 0.014 ENSGACG00000009469 Chr1 9166576 9173856 egln2 11 0 0 0 11 0 0.001 ENSGACG00000004433 Chr17 2127457 2211376 igsf21a 10 10 10 0 0 0 0.003 ENSGACG00000007343 Chr10 10666995 10679875 col9a2 10 0 0 0 6 4 0.227 ENSGACG00000018407 Chr4 13828336 13837518 Sncb 10 2 2 0 5 3 0.848 Salinity increase: ENSGACG00000020323 Chr7 17010160 17011176 23 0 0 0 22 1 <0.001 ENSGACG00000013229 Chr18 15327717 15352321 15 10 10 0 1 4 0.125 ENSGACG00000013359 Chr11 12960883 12968110 sec14l1 15 0 0 0 12 3 0.011 ENSGACG00000019416 Chr7 4451892 4453656 HMX1 ortholog 15 3 3 0 5 7 0.745 ENSGACG00000002948 Chr8 218240 221355 ddx10 14 0 0 0 6 8 0.077 ENSGACG00000016350 Chr14 3603545 3604923 14 1 0 1 7 6 0.277 ENSGACG00000006636 Chr18 4780893 4786820 ZC3H12D 13 0 0 0 3 10 0.034 ENSGACG00000004667 Chr12 4273498 4286193 tti1 12 0 0 0 12 0 0.001 ENSGACG00000015566 Chr2 9043062 9051779 casc4 10 0 0 0 10 0 0.003
Supplementary Materials
Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/6/12/eaaz1138/DC1
Fig. S1. Significant DMS throughout the genome for comparison between KIE versus NYN (20 versus 6 PSU; blue fish) and KIE versus SYL (20 versus 33 PSU; yellow fish).
Fig. S2. GO terms for biological processes, cellular components, and molecular functions under salinity increase (20 versus 33 PSU; yellow) and decrease (20 versus 6 PSU; blue) associated with pop-DMS.
Table S1. Relative distribution of DMS among genomic features.
Table S2A. Differentially methylated genes between populations from KIE (20 PSU) and NYN (6 PSU).
Table S2B. Differentially methylated genes between populations from KIE (20 PSU) and SYL (33 PSU).
Table S3A. Tukey post hoc test results for survival rate.
Table S3B. Tukey post hoc test results for SDL.
Table S3C. Tukey post hoc test results for HSI.
Table S3D. Tukey post hoc test results for total weight.
Table S4. Summary statistics for whole-genome resequencing of wild-caught sticklebacks.
Table S5A. Summary statistics for the RRBS of experimental fish.
Table S5B. Summary statistics for the RRBS of wild-caught fish.
Table S6. The number of DMS for each of the two pairwise population comparisons (pop-DMS).
Additional Files
Supplementary Materials
This PDF file includes:
- Fig. S1. Significant DMS throughout the genome for comparison between KIE versus NYN (20 versus 6 PSU; blue fish) and KIE versus SYL (20 versus 33 PSU; yellow fish).
- Fig. S2. GO terms for biological processes, cellular components, and molecular functions under salinity increase (20 versus 33 PSU; yellow) and decrease (20 versus 6 PSU; blue) associated with pop-DMS.
- Table S1. Relative distribution of DMS among genomic features.
- Table S2A. Differentially methylated genes between populations from KIE (20 PSU) and NYN (6 PSU).
- Table S2B. Differentially methylated genes between populations from KIE (20 PSU) and SYL (33 PSU).
- Table S3A. Tukey post hoc test results for survival rate.
- Table S3B. Tukey post hoc test results for SDL.
- Table S3C. Tukey post hoc test results for HSI.
- Table S3D. Tukey post hoc test results for total weight.
- Table S4. Summary statistics for whole-genome resequencing of wild-caught sticklebacks.
- Table S5A. Summary statistics for the RRBS of experimental fish.
- Table S5B. Summary statistics for the RRBS of wild-caught fish.
- Table S6. The number of DMS for each of the two pairwise population comparisons (pop-DMS).
Files in this Data Supplement:
