The Arabidopsis ABA-activated kinase OST1 phosphorylates the bZIP transcription factor ABF3 and creates a 14-3-3 binding site involved in its turnover

Abstract

Background: Genetic evidence in Arabidopsis thaliana indicates that members of the Snf1-Related Kinases 2 family (SnRK2) are essential in mediating various stress-adaptive responses. Recent reports have indeed shown that one particular member, Open Stomata (OST)1, whose kinase activity is stimulated by the stress hormone abscisic acid (ABA), is a direct target of negative regulation by the core ABA co-receptor complex composed of PYR/PYL/RCAR and clade A Protein Phosphatase 2C (PP2C) proteins.

Methodology/principal findings: Here, the substrate preference of OST1 was interrogated at a genome-wide scale. We phosphorylated in vitro a bank of semi-degenerate peptides designed to assess the relative phosphorylation efficiency on a positionally fixed serine or threonine caused by systematic changes in the flanking amino acid sequence. Our results designate the ABA-responsive-element Binding Factor 3 (ABF3), which controls part of the ABA-regulated transcriptome, as a genuine OST1 substrate. Bimolecular Fluorescence Complementation experiments indicate that ABF3 interacts directly with OST1 in the nuclei of living plant cells. In vitro, OST1 phosphorylates ABF3 on multiple LXRXXpS/T preferred motifs including T451 located in the midst of a conserved 14-3-3 binding site. Using an antibody sensitive to the phosphorylated state of the preferred motif, we further show that ABF3 is phosphorylated on at least one such motif in response to ABA in vivo and that phospho-T451 is important for stabilization of ABF3.

Conclusions/significance: All together, our results suggest that OST1 phosphorylates ABF3 in vivo on T451 to create a 14-3-3 binding motif. In a wider physiological context, we propose that the long term responses to ABA that require sustained gene expression is, in part, mediated by the stabilization of ABFs driven by ABA-activated SnRK2s.

Figure 1. OST1 substrate preferences.

(A) The Position Specific Scoring Matrix (PSSM) for the OST1 kinase regroups the calculated weight (W) of each amino acid at each position. The amino acids favoring phosphorylation (W>1) are colored in blue using Excel conditional formatting, while those that negatively affect phosphorylation (W<1) are in red. (B) A series of variant peptides originating from the optimized SnRK2S peptide was used to validate OST1 kinase phosphorylation preferences. The initial phosphorylation speed of the peptides (kinase activity) was calculated with a linear regression using 4 to 5 time-points of the phosphorylation kinetic. The error bars represent 95% confidence interval. The experiment was repeated tree times with similar results.

Figure 2. OST1 interacts with ABF3 in the nucleus of guard cell.

(A) qRT-PCR analysis of ABF genes expression in epidermal peels in response to 50 µM ABA for 3 h. Data were normalized with AtACTIN2 expression and the expression of ABF3 gene without ABA was set to 1 as reference. Error bars indicate standard deviation for two technical replicates. A representative experiment out of three independent biological repetitions is shown (B) qRT-PCR analysis of At2g36640 gene expression in epidermal peels of wild type (Col), ost1 and abf3 plant genotypes in response to 50 µM ABA for 3 h. At2g36640 expression was normalized to the expression of AtACTIN2 in all samples and the value corresponding to non-treated wild type plants (Col) set to 1. A representative experiment out of three independent biological repetitions is shown. (C) Confocal images of guard cells expressing YFP-ABF3 (left) and YFP-OST1 (right). On the bottom panel, the confocal images were merged with the corresponding bright field images. Scale bar corresponds to 8 µm. (D) ABF3 and OST1 interaction is analyzed by Bimolecular Fluorescent Complementation in N. benthamiana tobacco leaves expressing the given constructs. The YFP fluorescent signal is located in the nucleus of epidermal cells. In these experiments, the nuclear localized AKINβ protein is used as negative control. Scale bar corresponds to 20 µm.

Figure 3. OST1 phosphorylates ABF3 on multiple sites in vitro.

(A) OST1 kinase activity towards ABF3 derived peptides corresponding to the putative OST1 phosphorylation sites. The OST1 activity was calculated with a linear regression using 4 to 5 time-points of the phosphorylation kinetic. The error bars represent a 95% confidence interval. The experiment was repeated tree times with similar results. (B) The C-terminally truncated ABF3 protein (ABF31–351), and the corresponding mutant forms, in which target Ser were either individually or together (3xSA) mutated to Ala, were phosphorylated by OST1 in vitro in presence of ATP^γ32 (upper panel). ABF3 phosphorylation was quantified by phosphorimaging and normalized to protein amount using Coomassie brilliant blue (CBB) gel staining (lower panel). ABF3 phosphorylation was set to 100% for the wild type protein. (C) ABF31–351 phosphorylation was analyzed by LC-MS/MS. Phosphorylation of S32, S126, S55 and T169 is revealed in MS2 spectra by the neutral loss of phosphoric acid group (H3PO4, 98 Da) from the corresponding precursor ion: the doubly charged TRQNpS32VFSLTFDEFQNSW ion at m/z 1144.02 was produced by chymotrypsin hydrolysis with miscleavages, the triply charged TGGSLQRQGpS126LTLPRTI ion at m/z 622.86 was produced by chymotrypsin hydrolysis, the doubly charged DFGpS55MNDELLK ion at m/z 740.59 and the doubly charged QQpT169LGEMTLEEFLVR ion at m/z 929.07 were produced by trypsin hydrolysis. (D) OST1 phosphorylation sites identified on ABF3 (At4g34000) were aligned with corresponding sequences in Arabidopsis ABF1 (At1g49720), ABF4 (At3g19290), ABF2 (At1g45249), ABI5 (At2g36270) and rice TRAB1 (Os08g0472000). Amino acids predicted to favor OST1 phosphorylation are underlined.

Figure 4. Phosphorylated C4 domain binds 14-3-3 protein.

The binding of phosphorylated ABF3T451 (pABF3T451) and ABF3S126 (pABF3S126) peptides to the Arabidopsis 14-3-3 protein GF14phi was analyzed by in vitro pull-down assay and competition experiments. This experiment was repeated twice and error bar represent the standard error of the mean.

Figure 5. ABF3 is phosphorylated on LXRXX(S/T) motif(s) in vivo in response to ABA.

Phosphorylation of immunoprecipitated YFP-ABF3 (A) and mutants YFP-abf3^S1261A and YFP-abf3^T451A (B) was analyzed by immunoblotting using the PKD substrate antibody that specifically recognized the phosphorylated LXRXXp(S/T) motif. In these experiments plants were treated with MG115 and MG132 for 24 h before treatment with 100 µM ABA for 1 h. The amount of immunoprecipitated proteins was controlled with anti-GFP antibody.

Figure 6. ABF3 is stabilized in response to ABA.

(A) YFP-ABF3 stability in response to ABA was analyzed by immunoblotting with anti-GFP antibody using protein extract from transgenic plants expressing YFP-ABF3 under the control of the constitutive 35S promoter (upper panel). Uniformity of protein loading was verified by CBB gel staining. This result is representative of three independent experiments. When indicated (+Inh), MG115 and MG132 proteasome inhibitors were added 1 h before ABA stimulation. The time course activation of OST1 by ABA was analyzed by kinase assay using 10xHis-ABF3_1-351 as substrate (lower panel). The ost1 knockout mutant is used as negative control in this experiment. (B) The effect of ABA on YFP-abf3^S126A stability was analyzed using the same experimental procedure. (C) The effect of ABA and proteasome inhibitors (24 h treatment) on YFP-abf3^T451A stability was analyzed by confocal microscopy (upper panel, merged fluorescent and bright field pictures, scale bar corresponds to 8 µm) using transgenic plants expressing YFP-abf3^T451A under the control of the constitutive 35S promoter. This experiment was repeated independently twice with the same results. The stabilization of YFP-abf3^T451A by proteasome inhibitors was confirmed by immunoblotting (lower panel).