Differential AMP-activated Protein Kinase (AMPK) Recognition Mechanism of Ca2+/Calmodulin-dependent Protein Kinase Kinase Isoforms*

Ca2+/calmodulin-dependent protein kinase kinase β (CaMKKβ) is a known activating kinase for AMP-activated protein kinase (AMPK). In vitro, CaMKKβ phosphorylates Thr172 in the AMPKα subunit more efficiently than CaMKKα, with a lower Km (∼2 μm) for AMPK, whereas the CaMKIα phosphorylation efficiencies by both CaMKKs are indistinguishable. Here we found that subdomain VIII of CaMKK is involved in the discrimination of AMPK as a native substrate by measuring the activities of various CaMKKα/CaMKKβ chimera mutants. Site-directed mutagenesis analysis revealed that Leu358 in CaMKKβ/Ile322 in CaMKKα confer, at least in part, a distinct recognition of AMPK but not of CaMKIα.

ulation of appetite and glucose homeostasis (16), stimulation of mitochondrial fatty acid oxidation by thyroid hormone T3 (17), and regulation of autophagy by amino acid starvation (18). According to studies using either RNA interference or pharmacological inhibition of CaMKK in HeLa cells, which do not express LKB1, an alternative AMPK kinase, CaMKK␤ has been shown to be responsible for Ca 2ϩ -dependent activation of AMPK in vivo, whereas both CaMKK isoforms are capable of phosphorylating the AMPK␣ subunit at Thr 172 in vitro (13)(14)(15). This was confirmed by the fact that STO-609 (19), a CaMKK inhibitor, suppressed ionomycin-induced AMPK phosphorylation in A549 cells (a human lung adenocarcinoma epithelial cell line) expressing STO-609-resistant CaMKK␣ but not STO-609-resisitant CaMKK␤ (20). These results indicate that the CaMKK␤ isoform, rather than CaMKK␣, preferably recognizes AMPK, suggesting that the recognition mechanism of AMPK by CaMKK isoforms as a substrate may differ from that of CaMKI. A previous report showed that CaMKK␤, but not CaMKK␣, forms a stable complex with AMPK, which could explain why CaMKK␤ is an AMPK kinase and CaMKK␣ is not (21). However, Fogarty et al. (22) reported that CaMKK␤ activates AMPK without forming a stable complex with AMPK. Therefore, despite the well characterized regulatory mechanisms of CaMKK, including an autoinhibitory mechanism (23), and the role of Ca 2ϩ /CaM-binding in the expression of its kinase activity (24), little is known about the molecular mechanism of substrate recognition of CaMKKs. To clarify the differential substrate specificity of CaMKK isoforms, especially for AMPK, we investigated enzymatic characterization of CaMKKs using various chimeras and site-directed mutants of CaMKK isoforms and identified a single residue in subdomain VIII that may be essential for the discrimination of AMPK as a substrate.
In Vitro CaMKK Activity Assay-Purified recombinant CaMKK isoforms, including site-directed mutants and GST-CaMKK chimera mutants (ϳ1 g/ml unless otherwise indicated), were incubated individually with GST-CaMKI␣ 1-293, KE (0.5 mg/ml) or AMPK K45R (0.5 mg/ml) at 30°C for the indicated time periods in a solution containing 50 mM HEPES (pH 7.5), 10 mM Mg(CH 3 COO) 2 , 1 mM DTT, and 200 M [␥-32 P]ATP (200 -700 cpm/pmol) in the presence of either 4 mM CaCl 2 /10.0 -16.5 M CaM or 2 mM EGTA. Each reaction was initiated by addition of [␥-32 P]ATP and terminated by addition of 2ϫ SDS-PAGE sample buffer. Samples were then subjected to 10% SDS-PAGE followed by autoradiography. 32 P incorporation into each substrate was estimated by Cerenkov counting of the excised gels.
AMPK Phosphorylation in A549 Cells-Human lung adenocarcinoma epithelial cell line A549 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% FBS, 100 units/ml penicillin, and 100 units/ml streptomycin at 37°C in 5% CO 2 . A549 cells in 6-well dishes were infected with FLAG-CaMKK␣ mutants expressing retroviruses that were generated by transfection of the pVSV-G vector and retroviral transfer vectors (pMSCV-MCS-IRES-EGFP) harboring either the FLAG-CaMKK␣ double mutant (A292T/L233F) as described previously (20) or the FLAG-CaMKK␣ triple mutant (A292T/L233F/I322L) into the GP2-293 packaging cell line. After 18 h of culture, the cells were cultured in the absence of FBS for 6 h and then treated with 1 M ionomycin for 5 min. The cells were extracted with 1ϫ SDS-PAGE sample buffer (100 l), followed by immunoblot analysis using indicated antibodies.
Statistics-Student's t tests were used to evaluate statistical significance when two groups were compared. p Ͻ 0.05 was considered to be statistically significant.
Other Methods-The CaM overlay method was performed using 0.5 g/ml biotinylated CaM in the presence of 1 mM CaCl 2 , followed by detection of the CaM-binding signal using a chemiluminescence reagent (PerkinElmer Life Sciences) as described previously (28). Protein concentration was estimated by staining the samples with Coomassie Brilliant Blue (Bio-Rad) using bovine serum albumin as a standard.

Results and Discussion
CaMKK␤, but Not CaMKK␣, Preferentially Phosphorylates AMPK in Vitro-It has been reported that the purified CaMKK␤ from rat brain activated AMPK 7-fold more rapidly than purified CaMKK␣ in vitro (14), whereas both CaMKK isoforms equally phosphorylated CaMKI. To confirm a distinct substrate preference of CaMKK isoforms, we attempted to measure the direct phosphorylation of the activation-loop Thr (Thr 172 ) of the ␣ subunit of the AMPK heterotrimeric complex by purified recombinant CaMKK isoforms and compared this with the phosphorylation of the catalytic domain of rat CaMKI␣ (GST-CaMKI␣ 1-293, K49E) at Thr 177 . Recombinant CaMKK isoforms were expressed in the E. coli BL21 Star (DE3) strain and purified by CaM-coupled Sepharose column chromatography. The amounts of the enzymes for measuring sub-strate phosphorylation were comparable, as judged by the CaM overlay method (Fig. 1A), because both CaMKK isoforms contain similar CaM-binding sequences (24). Both CaMKK substrates were mutated at a residue in the ATP-binding site (K45R in AMPK␣ and K49E in CaMKI␣) to generate kinase-dead enzymes to avoid the feedback phosphorylation of CaMKKs by activated CaMKK target kinases (29). Time course experiments of phosphorylation of the catalytic domain of GST-CaMKI␣ 1-293, K49E (Fig. 1B), and AMPK (Fig. 1C) were performed with the same concentrations (1 g/ml) of CaMKK isoforms in the presence of Ca 2ϩ /CaM. Although the phosphorylation profiles of CaMKI by both CaMKK isoforms were indistinguishable (Fig. 1B), CaMKK␤ was shown to phosphorylate the Thr 172 of AMPK 14-fold more rapidly than CaMKK␣ (Fig. 1C) under this experimental condition. When we compared the kinetic constant (K m ) of CaMKK isoforms for AMPK based on the double reciprocal plots of phosphorylation data using various concentrations of AMPK ( Fig. 2A), it was clear that the main reason for the distinct activities of CaMKK isoforms for AMPK phosphorylation was the significantly higher K m (13.1 M) of CaMKK␣ for AMPK compared with that of CaMKK␤ (1.5 M). These results are in good agreement with a previous report using purified CaMKKs from rat brain (14), suggesting that CaMKK␤ is the dominant isoform with respect to the regulation of AMPK.
CaMKK␤ Leu 358 Plays Important Roles in Efficiently Phosphorylating AMPK-To identify the crucial amino acid(s) in CaMKK␤ required for efficiently phosphorylating AMPK␣, we  produced a full-length CaMKK␣ mutant in which Ala 321 -Ile 322 was replaced by the corresponding amino acid residues in CaMKK␤ (Ser 357 -Leu 358 ) (Fig. 5A) and measured the CaMKK activity toward CaMKI␣ and AMPK␣ as substrates. The amounts of the enzymes for measuring substrate phosphorylation were comparable, as judged by the CaM overlay method (Fig. 5B). The CaMKK␣ A321S mutant was shown to phosphor-ylate AMPK␣ with a lower efficiency, in a similar manner as the CaMKK␣ wild type. In contrast, the CaMKK␣ I322L mutant exhibited significantly enhanced kinase activity toward AMPK␣ in a similar manner as the CaMKK␤ wild type. We observed the AMPK phosphorylating activity of the CaMKK␣ I322L mutant in a complete Ca 2ϩ /CaM-dependent manner  (supplemental Fig. 1), indicating that the enhanced kinase activity toward AMPK␣ as a substrate (Fig. 5C) was not likely due to disruption of the autoinhibitory mechanism of CaMKK␣ (23) by this mutation. Therefore, we measured the kinetic constant (K m ) of the CaMKK␣ I322L mutant for AMPK based on the double reciprocal plots of phosphorylation data using various concentrations of AMPK (Fig. 2B) and obtained a K m value (4.9 M) that was significantly lower than that of CaMKK␣ (13.1 M, Fig. 2A). Finally, to confirm the data obtained with in vitro experiments as described above by using living cells, we attempted to test the ionomycin-induced AMPK phosphorylation in A549 cells (a human lung adenocarcinoma epithelial cell line) in which FLAG-tagged CaMKK␣ mutants were exogenously expressed with a retrovirus expression system (Fig. 6). When we expressed a FLAG-tagged CaMKK␣ double mutant (A292T/L233F, AT/LF) in A549 cells, which was a CaMKK inhibitor (STO-609)-resistant mutant (20), 10 g/ml STO-609 (19) treatment completely inhibited ionomycin-induced AMPK phosphorylation because CaMKK␤ was thought to be responsible for AMPK phosphorylation in A549 cells but not CaMKK␣. In contrast, we could observe significant ionomycininduced AMPK phosphorylation in cells expressing a CaMKK␣ triple mutant (A292T/L233F/I322L, AT/LF/IL) even in the presence of STO-609, indicating that the CaMKK␣ triple mutant (AT/LF/IL) acquired an ability for phosphorylating AMPK in living cells. These results suggest that a single amino acid difference (Leu 358 in CaMKK␤/Ile 322 in CaMKK␣) in the catalytic domain of CaMKK isoforms dictates the efficiency of the kinase to phosphorylate AMPK␣ in vivo as well as in vitro.
In conclusion, CaMKK␤, but not CaMKK␣, was shown to be an upstream activating kinase for AMPK (13)(14)(15) because of its efficient phosphorylating activity of CaMKK␤ with an ϳ9-fold higher affinity for AMPK than CaMKK␣ had. It is noteworthy that the kinase activities of both CaMKK isoforms toward CaMKI␣ were indistinguishable. Our mutagenesis studies clearly indicated that a single amino acid (Leu 358 in CaMKK␤/ Ile 322 in CaMKK␣) in subdomain VIII plays a role, at least in part, in the discrimination of AMPK as a native substrate but not in the discrimination of CaMKI␣. This finding is in good agreement with a previous report demonstrating that subdomain VIII in MAPK/ERK kinase kinase 1 (MEKK1) was a contact site for its substrate, MKK4, thereby discriminating the substrate (30). In addition, Leu 358 in CaMKK␤ is also conserved in mammalian CaMKK␤s and in its counterpart (ckk-1) in the roundworm Caenorhabditis elegans (31) (Fig. 5A) but not in mammalian CaMKK␣ isoforms, suggesting that the CKK-1/ AMPK pathway may be functional in nematodes.
Author Contributions-Y. F., Y. K., and T. F. performed the experiments. N. K. and M. M. supervised the experiments and helped to edit the manuscript. H. T. designed the study and wrote the manuscript.