This led to our following model (in brief): DOG1 prevents AHG1 from promoting AFPs-mediated shut-down of maturation-imposed ABA responses, thereby maintaining seed dormancy. The loss of DOG1's function over time allows AFPs to trigger germination.

We hypothesized that the residual 75% of seeds, might promote germination via AHG3, the second DOG1-controlled PP2C. Introduction of an ahg3 mutant allele into afp dog1 resulted in extreme dormancy, confirming that AFPs are pivotal factors of the AHG1 branch of the DOG1-PP2C module.

Furthermore, we demonstrated that DOG1 requires AFPs in seeds by combining the highly dormant afp mutants with non-dormant dog1 mutant. This resulted in the restoration of some dormancy (25 % of seeds) in the corresponding offspring.

Vice versa the overexpression of AFP2 leads to the absence of dormancy, but can be reversed by the introduction of an ahg1 mutant allele. This shows that AFPs functionally require an active AHG1 to promote germination.

The analysis of higher order afp mutants showed AFP1, 2 and 3 are partially redundant negative regulators of dormancy, whose absence results in extremely strong dormancy in Arabidopsis seeds.

In this context, we identified the ABI FIVE BINDING PROTEINs (AFPs) 1 and 2 as promising targets of DOG1-PP2C as they were less phosphorylated in dog1 seeds. These proteins specifically interacted with the DOG1-controlled PP2C AHG1, which is able to dephosphorylate AFPs in vitro.

We could show that hyperphosphorylated SnRK2s in dog1 seeds are more active compared to WT, despite overall ABA responses appear to be diminished. This suggested that DOG1’s function is not to prevent the deactivation of SnRK2s, but to stabilize ABA responses otherwise.

Interestingly, we detected that class III SnRK2 kinases from the ABA core pathway were hyperphosphorylated at a critical residue in the activation loop in dog1 mutants, which is linked to their activity, despite the PP2CAs AHG1 and AHG3 controlled by DOG1 lack a pivotal upstream regulator.

Using mass spec-based analyses, we found that the absence of DOG1 profoundly impacts the dry and imbibed seed proteome/phosphoproteome. Furthermore, DOG1 appears to be necessary to establish ABA responses in seeds during late maturation/early imbibition at the proteome level.

Both mutants fail to establish dormancy, but while abi2-1 is hyposensitive to ABA during germination, dog1-1 mutants remain fully responsive and retain their ability to inhibit germination. This pointed towards a different mode of action by which DOG1-PP2Cs controls dormancy establishment.