DNMT3a mediates developmental plasticity. We used CRISPR/Cas9 to produce single (DNMT3aa^−/−) and double (DNMT3aa^−/−ab^−/−) knock-out lines plus a no-template control (control) (a). In each genotype, we tested the effect of developmental and acclimation temperatures (23 °C blue, 28 °C red) on fecundity, survival, and deformities in F₁ and F₂ fish. We used adult F₂ fish to determine locomotor and metabolic thermal performance curves. The points in the experiment where data in the different figures were collected are marked on top of the panel. At 23 °C developmental temperature (b), survival to the 26-somite stage (early survival) was lowest in the DNMT3aa^−/−ab^x/x (ab mixed genotype) fish (purple circles; p < 0.0001 compared to control), and survival was somewhat reduced in the DNMT3aa^−/− line compared to control at 28 °C (p < 0.04). Deformities (c) increased from control (grey circles) to DNMT3aa^−/− (green circles; p = 0.004) and DNMT3aa^−/−ab^x/x fish (p < 0.0001) at 23 °C developmental temperature and were elevated in DNMT3aa^−/−ab^x/x fish at 28 °C compared to control (p = 0.002). Survival to swim bladder inflation (late survival) (d) at 23 °C was lowest in DNMT3aa^−/−ab^x/x fish (p < 0.0001 compared to control) and intermediate in DNMT3aa^−/− fish (p = 0.02 compared to control). Late survival was reduced in the two knock-out lines at 28 °C (both p < 0.01 compared to control). Means ± s.e. are shown in B-D (black horizontal and vertical lines, respectively), and each symbol represents data from one clutch (n = 19–28 clutches per treatment). Letters indicate significant differences between groups. There were no homozygous knock-out survivors (e) at the ab locus in the F₁ DNMT3aa^−/−ab^x/x line at 23 °C developmental temperature, but 35% of survivors were DNMT3aa^−/−ab^−/− homozygous fish at 28 °C

Parental acclimation rescues offspring survival. Fecundity of F₁ fish (a) reared at 28 °C and acclimated to 23 °C did not differ significantly between genotypes (control = grey circles, DNMT3aa^−/− green circles, and DNMT3aa^−/−ab^−/− purple circles). Genotype or developmental temperature had no effect on F₂ early survival rates (b). There was a small but significant increase in deformity rates in DNMT3aa^−/− larvae at 23 °C (indicated by asterisk) (c). There was no significant effect of genotype or developmental temperature on late survival rates (d). Each circle in a–d represents an individual clutch (n = 17–32 clutches per treatment), and means ± s.e. are shown (black horizontal and vertical lines, respectively). Multiple dimension scaling plot (e) of reduced representation bisulphite sequencing results showed no global differences in methylation pattern between control and DNMT3aa^−/−ab^−/− fish from both developmental temperatures (triangles = 28 °C developmental temperatures, inverted triangles = 23 °C, grey triangles = control, purple triangles = DNMT3aa^−/−ab^−/−)

DNMT3a modulates thermal performance curves. A significant interaction between genotype and developmental temperature manifests as a progressive decline in swimming performance (U_crit) (a) of warm-developed fish (red symbols and lines) from control to DNMT3aa^−/− and to DNMT3aa^−/−ab^−/− where U_crit was lowest. The genotype x developmental temperature interaction determining citrate synthase (CS) activity (b) results from higher activity in warm-developed DNMT3aa^−/− fish compared to cold-developed fish. As for U_crit, activity of warm-developed DNMT3aa^−/−ab^−/− fish was lowest but similar to control. Marginal means (± s.e.) are shown across both acclimation treatments in a and b. Sample sizes were n = 12 for U_crit (24 for marginal means shown) and n = 8–9 for CS activity (16–18 for marginal means). Developmental temperature interacted with acclimation temperature to determine U_crit (c), and CS activity (d); marginal means (± s.e.) across all genotypes and test temperatures are shown