Validation of COR28 effects using knockout lines and co-segregation analysis
11 mutant lines representing eight gene candidates were genotyped to confirm homozygous mutations before being bulked for seed. Confirmed mutants were then assayed for their circadian rhythms using DF under the same conditions as for the 191 accession screen. Mutant details and genotyping results can be viewed in Supplementary File 2. Cor27mutants and the double mutant cor27-1/28-2 were also assayed ascor27 is known to be partially redundant with cor28 . Analysis of the DF rhythms confirmed that periods in cor28mutants (cor28-1 =25.3h, SD=1.73; cor28-2 =25.3h, SD=1.78) were significantly longer than their respective controls (23.9h, SD=1.62 and 24.0h SD=2.02) (Welch Two Sample t-test p< 0.05) as shown in Table 2. This confirms results previously reported using leaf movement, qPCR and ProCCR2:LUC bioluminescence rhythms (X. Li et al., 2016; Wang et al., 2017). Cor27 mutants showed no significant difference in period length, however the double knock-outcor27-1/28-2 had an exaggerated long period phenotype (26.4h, SD=2.37). The peak phases of cor27 , cor28 and especiallycor27-1/28-2 were also earlier than for Col-0, withcor27-1/28-2 having a mean phase peak 2.6h earlier than the control. Other mutants characterized were not found to have significantly altered periods or phases compared to their WT controls (Supplementary Tables 12-14).
[Table 2]
Finally, we carried out a co-segregation analysis for the COR2858S allele to demonstrate that this SNP was causal for the long period phenotype. Crosses were made between the line T800 (Watkins 6133) and St-0 (Watkins 8387), with T800 carrying the 58S minor allele and St-0 carrying the 58W major allele. Parent individuals were genotyped using a CAPs assay prior to crossing and were found to be homozygous for their respective SNPs. We selfed three F1 individuals from this cross and analyzed the genotypes and circadian period phenotypes in the F2 offspring using DF on dissected leaves (Table 3.) Restriction digest gels from F2 genotyping can be viewed in Supplementary File 6. F2 individuals carrying the 58S SNP had a mean period length of 23.7h (SD=1.95) compared to a mean of 21.3h (SD=1.65) for those carrying the 58W SNP (a difference of 2.4h) which was shown to be highly significant (F( 2,62)=12.27, p<0.0001, One-way ANOVA. TukeyHSD for SNP 58S and 58W: p<0.001). Heterozygous individuals had period lengths which were on average closer to the long period phenotypes of the 58S homozygote individuals (23.1h (SD=1.95)) suggesting a semi-dominant effect for the 58S allele.
[Table 3]