Structural insight into the mechanism of synergistic autoinhibition of SAD kinases

Abstract

The SAD/BRSK kinases participate in various important life processes, including neural development, cell cycle and energy metabolism. Like other members of the AMPK family, SAD contains an N-terminal kinase domain followed by the characteristic UBA and KA1 domains. Here we identify a unique autoinhibitory sequence (AIS) in SAD kinases, which exerts autoregulation in cooperation with UBA. Structural studies of mouse SAD-A revealed that UBA binds to the kinase domain in a distinct mode and, more importantly, AIS nestles specifically into the KD-UBA junction. The cooperative action of AIS and UBA results in an 'αC-out' inactive kinase, which is conserved across species and essential for presynaptic vesicle clustering in C. elegans. In addition, the AIS, along with the KA1 domain, is indispensable for phospholipid binding. Taken together, these data suggest a model for synergistic autoinhibition and membrane activation of SAD kinases.

Figure 1. Two elements within the non-catalytic region autoinhibit SAD-A activity.

(a) Schematic diagram of mouse SAD-A. The structural elements are coloured as follows: the kinase domain (N-lobe, pink; C-lobe, yellow; activation segment, red), UBA (slate blue), AIS (dark green) and KA1 (green). The sequence of AIS is provided, with the key residues highlighted in red. (b) Phosphorylation of SAD-A Thr175 by LKB1 analyzed using an anti-AMPK-pT172 antibody. (c) Representative time course of PP2Cα-catalyzed dephosphorylation of activated SAD. The reaction mixture contains 1 μM SAD-A full-length protein and 250 nM PP2Cα. (d) Time courses of SAD-catalyzed phosphorylation of Cdc25C. Reactions were initiated by adding 5 nM indicated SAD-A protein to the reaction mixture containing 7.5 μM Cdc25C peptide. (e) Comparison of the initial rates of 5 nM various SAD-A fragments towards 7.5 μM Cdc25C peptide (mean±s.e.m., n=3). (f) Plots of initial rate of SAD-catalyzed Cdc25C phosphorylation versus the Cdc25C concentration. The solid lines represent the best fitting results to the Michaelis–Menten equation with k_cat and K_m values listed at the top.

Figure 2. Distinct binding modes between SAD-A kinase and UBA domains.

(a) Overall structure of SAD-A KD-UBA. The color scheme follows that in Fig. 1a. (b) Comparison of the KD-UBA structures from SAD-A, AMPK (PDB code: 3H4J), MARK1 (2HAK) and MELK (4IXP) upon superposition of kinase C-lobes. For clarity, only the UBA domains from AMPK, MARK1 and MELK are displayed. (c,d) Close-up views of the KD-UBA interface. The interacting residues from UBA are highlighted as cyan sticks, and those from the kinase N- and C-lobes are shown as magenta and orange sticks, respectively. Blue dashed lines represent the polar interactions. (e,f) Structure-based sequence alignments of the UBA domains and the αC regions from SAD-A, BRSK2, AMPK, MARK1 and MELK. Residues involved in respective KD–UBA interactions are highlighted in blue. Residues at the KD-UBA interface of SAD-A are also indicated by asterisks, and the AIS-interacting residues are indicated by triangles. The code following each protein name is the corresponding UniProt ID. m, mouse; h, human; sp, Schizosaccharomyces pombe.

Figure 3. Binding of UBA immobilizes the αC-out kinase conformation.

(a) Comparison of the autoinhibited SAD-A with the active conformations of PKA (PDB code: 1ATP) and Snf1 (2FH9). For clarity, only strands β3, helices αC and activation segments of PKA (marine blue) and Snf1 (green) are displayed. The conserved Lys and Glu side chains and two hydrophobic residues in the activation segment of SAD-A are highlighted as sticks. (b) Effects of the UBA mutations on the catalytic efficiencies of KD-UBA. The k_cat/K_m values are determined as that in Supplementary Table 1.

Figure 4. AIS binds at the KD-UBA junction and inhibits SAD-A activity.

(a) Trans-inhibition of different C-terminal fragments on the activity of KD-UBA (10 nM). The continuous curves were the best-fit to the non-competitive model using equation v_i/v₀=(K_i+a[I])/(K_i+[I]), where K_i and a are the apparent inhibition constant and residual activity, respectively. (b) Overall structure of the KD-UBA and AIS-KA1 complex. The color scheme for the complex follows that in Fig. 1a. (c) Comparison of the KD-UBA conformations in complex with AIS-KA1 and in isolation. The KD-UBA alone is shown in dark grey. (d,e) Close-up views of AIS binding to the KD-UBA junction. The interacting residues in the AIS sequence are highlighted as green sticks, and those from the kinase and UBA domains are shown as magenta and cyan sticks, respectively.

Figure 5. AIS and UBA synergistically regulates SAD activity.

(a) Effects of the AIS and KA1 mutations on the trans-inhibition of AIS-KA1 on the KD-UBA activity. The assays were performed with 10 nM wild-type KD-UBA and 50 μM AIS-KA1 mutants. (b) Comparison of the catalytic activities of various KD-UBA mutants in the absence or presence of AIS-KA1. The assays were performed with 5 nM KD-UBA mutants and 50 μM wild-type AIS-KA1. (c) Activities of full-length SAD-A bearing various UBA and/or AIS mutations. Reactions were initiated by adding 5 nM indicated SAD-A proteins. (mean±s.e.m., n=3)

Figure 6. C. elegans SAD-1 mutations mislocate presynaptic vesicle clusters to dendrite of DA9 neuron.

(a) Sequence alignment of UBA-α3 and AIS from mouse SAD-A and C. elegans SAD-1. Residues involved in intramolecular interactions are shown in bold, and those mutated in following assays are highlighted in blue or green. (b) Localization of synaptic vesicle-associated GFP::RAB-3 in wild-type animals. (c–g) Distribution of GFP::RAB-3 in animals expressing single-copy insertion of wild-type and mutant sad-1. Scale bar, 10 μm. (h) Quantification of ectopic dendritic puncta in wild-type and mutant animals. The columns represent the percentage of animals with 0, 1–3, 4–6, or >6 extra presynaptic puncta in the dendrite.

Figure 7. Model for the autoinhibition and activation of SAD-A.

(a) Sequence alignment for the AIS-KA1 regions of SAD and MARK kinases. The positively charged residues required for phospholipid binding are highlighted in blue. h, human; m, mouse; sp, Schizosaccharomyces pombe; sc, Saccharomyces cerevisiae. (b) Three basic clusters in the AIS-KA1 fragment. The basic residues are shown as blue sticks, and the surface representation is coloured according to electrostatic potential (positive, blue; negative, red). (c) Lipid-binding assays for two representative C-terminal fragments. PG, phosphatidylglycerol; PE, phosphatidylethanolamine; Chl, cholesterol; SPM, sphingomyelin; PS, phosphatidylserine; PIP₂, phosphatidylinositol-4,5-bisphosphate; PA, phosphatidic acid; PC, phosphatidylcholine. (d) Conserved (i–iv) and specific (SAD) hydrophobic pockets surrounding helix αC. Three AIS-binding pockets are indicated by green circles, and that for UBA binding is coloured in blue. For clarity, only helices α1 and α3 of UBA are displayed. (e) Regulation model for SAD-A. The extensive intramolecular interactions keep SAD-A in an autoinhibited conformation, where the kinase activity is synergistically inhibited by UBA and AIS. Membrane association of the AIS-KA1 fragment, probably triggered by regulatory protein(s), might release the AIS (and UBA) inhibition.