1,2,6-Thiadiazinones as Novel Narrow Spectrum Calcium/Calmodulin-Dependent Protein Kinase Kinase 2 (CaMKK2) Inhibitors

We demonstrate for the first time that 4H-1,2,6-thiadiazin-4-one (TDZ) can function as a chemotype for the design of ATP-competitive kinase inhibitors. Using insights from a co-crystal structure of a 3,5-bis(arylamino)-4H-1,2,6-thiadiazin-4-one bound to calcium/calmodulin-dependent protein kinase kinase 2 (CaMKK2), several analogues were identified with micromolar activity through targeted displacement of bound water molecules in the active site. Since the TDZ analogues showed reduced promiscuity compared to their 2,4-dianilinopyrimidine counter parts, they represent starting points for development of highly selective kinase inhibitors.


Introduction
Protein kinases catalyze phosphate transfer from Adenosine Triphosphate (ATP) to tyrosine, threonine or serine residues in specific target proteins. These phosphorylation events occur in almost every signal transduction pathway and provide regulatory points for therapeutic intervention [1]. Kinases have been successfully utilized as drug targets for the past 30 years, with 38 kinase inhibitors approved by the FDA to date [2]. These drugs are predominantly multi-targeted tyrosine kinase inhibitors for the treatment of cancer [3]. However, approval of kinase inhibitors for the treatment on non-oncological diseases, such as rheumatoid arthritis, psoriasis and lung fibrosis has demonstrated their broader utility in treatment of human disease. There are over 500 kinases in the human genome, suggesting that there remains an untapped potential to treat a wide range of human ailments with new classes of inhibitors. Large scale kinome-wide profiling of ATP-competitive kinase inhibitors has also started to uncover the preferred chemotypes for the inhibition of many of the relatively under-studied kinases or dark kinases [4][5][6]. Despite the success in development of kinase inhibitor drugs, there is a still a need for new heterocycles on which to build ATP-competitive inhibitors [7,8]. One chemotype that has not yet been used in kinase inhibitor design is the 4H-1,2,6-thiadiazin-4-one (TDZ, Figure 1) [9,10] that can be prepared from 2,2-dichloromalononitrile [11].
Dianilinopyrimidines represent a remarkably common chemotype that is found in~10% of the clinically approved kinase inhibitor drugs, including ceritinib and palbociclib ( Figure 1) [12]. Each of these drugs demonstrates potency and efficacy for its primary kinase target but also has cross activity on a broad range of other kinases. As such, these drugs and many other pyrimidine-based inhibitors have limited use as chemical probes to study the biology of specific kinases. As further testament to the broad activity profile of the dianilinopyrimidine chemotype, the 35 examples that are contained in the chemogenomic inhibitor sets PKIS/PKIS2 showed activity on >400 different protein kinases (excluding mutants) in either enzyme inhibition or affinity capture assays [13,14].
Molecules 2018, 23, x 2 of 23 every signal transduction pathway and provide regulatory points for therapeutic intervention [1].
Kinases have been successfully utilized as drug targets for the past 30 years, with 38 kinase inhibitors approved by the FDA to date [2]. These drugs are predominantly multi-targeted tyrosine kinase inhibitors for the treatment of cancer [3]. However, approval of kinase inhibitors for the treatment on non-oncological diseases, such as rheumatoid arthritis, psoriasis and lung fibrosis has demonstrated their broader utility in treatment of human disease. There are over 500 kinases in the human genome, suggesting that there remains an untapped potential to treat a wide range of human ailments with new classes of inhibitors. Large scale kinome-wide profiling of ATP-competitive kinase inhibitors has also started to uncover the preferred chemotypes for the inhibition of many of the relatively understudied kinases or dark kinases [4][5][6]. Despite the success in development of kinase inhibitor drugs, there is a still a need for new heterocycles on which to build ATP-competitive inhibitors [7,8]. One chemotype that has not yet been used in kinase inhibitor design is the 4H-1,2,6-thiadiazin-4-one (TDZ, Figure 1) [9,10] that can be prepared from 2,2-dichloromalononitrile [11]. Dianilinopyrimidines represent a remarkably common chemotype that is found in ~10% of the clinically approved kinase inhibitor drugs, including ceritinib and palbociclib ( Figure 1) [12]. Each of these drugs demonstrates potency and efficacy for its primary kinase target but also has cross activity on a broad range of other kinases. As such, these drugs and many other pyrimidine-based inhibitors have limited use as chemical probes to study the biology of specific kinases. As further testament to the broad activity profile of the dianilinopyrimidine chemotype, the 35 examples that are contained in the chemogenomic inhibitor sets PKIS/PKIS2 showed activity on >400 different protein kinases (excluding mutants) in either enzyme inhibition or affinity capture assays [13,14].

Synthesis
Following analysis of the kinome-wide profiling of the dianilinopyrimidines in PKIS/PKIS2, we selected five R 1 and R 2 substituent pairs (Table 1 and Figure S1) that showed the broadest range of activity on human kinases. The corresponding dianilino-TDZs (1)(2)(3)(4)(5) were synthesized in two-steps from 3,5-dichloro-4H-1,2,6-thiadiazin-4-one (6) [9]. The reason for this strategy is that the first of the two reactive chlorine atoms of dichlorothiadiazinone 6 can be readily displaced by anilines and alkylamines with stoichiometric amounts of the amine (1 equiv.) and 2,6-lutidine (1 equiv.) as base in EtOH, at ca. 0-20 °C. However, more forcing conditions are typically required for the displacement of the remaining chloride. This is owed to electron release by the 3-amino group into the thiadiazine that decreases the electrophilicity of the 3-amino-5-chloro-1,2,6-4H-thiadiazin-4-one [15]. Nevertheless, we were able to use our recently developed Buchwald-Hartwig coupling conditions to overcome this difficulty [16] and enable the efficient synthesis of unsymmetrical 3,5-diaminothiadiazinones.

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

-
a Average of 4 experiments.

CaMKK2 Crystallography
To investigate the molecular details of interaction of the dianilino-TDZs (1-3) with protein kinases, co-crystallization with the corresponding purified proteins was attempted. Diffracting crystals were obtained with compound 2 in complex with CAMKK2 (see Table S2). The structure was solved by molecular replacement. The CAMKK2 kinase domain adopted an active state conformation in which residues of the regulatory and catalytic spines were aligned ( Figure 2A); residue Glu236 within α-C helix directly contacted Lys194 ("α-C helix in"); and residues Asp330 and Phe331 within the conserved DFG motif pointed towards and away, respectively, from the ATP-binding site ("DFG -in").
The ligand displayed two direct contact points to the hinge region of the ATP-binding pocket: one involving the oxygen atom of the thiadiazinone moiety and the other through the nitrogen atom of the hydroxymethylaniline moiety ( Figure 2B). A water bridge made by the nitrogen atom from the aminobenzamide provided a third contact point to the kinase hinge region. The co-crystal structure revealed that the oxygen atom of the aminobenzamide interacted with the catalytic Lys194 and made a water-bridge with Glu236 of the α-C helix and Asp330 within the conserved DFG motif. Likewise, two water-bridge interactions connected the nitrogen atom from the aminobenzamide moiety and residues Ser175 within P-loop and Asn317 at the bottom of the kinase ATP-binding site. Finally, the ligand aminobenzamide ring made a T-shaped π-π interaction with the Phe267. Compound 2 is bound to CAMKK2 with aniline groups in a twisted conformation relative to the central TDZ ring, as can be seen in the electron density map ( Figure 2C).

CaMKK2 Docking and Water Map Simulation
To further probe the molecular basis of ligand binding to CAMKK2, we compared our co-crystal structure of compound 2 (PDB 5VT1) with the previously published co-crystal structure with STO-609 (PDB:2ZV2) (Figure 3) [19]. We also controlled for hinge contacts in the model by using the 2,4dianilinopyrimidines as a training set (see Figure S2). We noted that STO-609 can displace a bound water molecule from the ATP-binding site that is still present when compound 2 is bound to the enzyme. The carboxylic acid of STO-609 also forms a tighter interaction with the catalytic lysine than the benzamide of compound 2. To improve its activity, we designed analogues of 2 that would be able to displace the bound water molecule and form stronger interactions with the catalytic lysine. In order to use molecular docking to guide the design of new analogues, we used the Maestro suite (2017-3) to create a models of compounds STO-609 and 2 bound to CaMKK2 that accurately reflected what was seen in the crystal structures of both [19]. Water Map using a 2 nS simulation was used to populate the hydration sphere of compounds STO-609 and 2. The resulting model correctly

CaMKK2 Docking and Water Map Simulation
To further probe the molecular basis of ligand binding to CAMKK2, we compared our co-crystal structure of compound 2 (PDB 5VT1) with the previously published co-crystal structure with STO-609 (PDB:2ZV2) (Figure 3) [19]. We also controlled for hinge contacts in the model by using the 2,4-dianilinopyrimidines as a training set (see Figure S2). We noted that STO-609 can displace a bound water molecule from the ATP-binding site that is still present when compound 2 is bound to the enzyme. The carboxylic acid of STO-609 also forms a tighter interaction with the catalytic lysine than the benzamide of compound 2. To improve its activity, we designed analogues of 2 that would be able to displace the bound water molecule and form stronger interactions with the catalytic lysine.

CaMKK2 Docking and Water Map Simulation
To further probe the molecular basis of ligand binding to CAMKK2, we compared our co-crystal structure of compound 2 (PDB 5VT1) with the previously published co-crystal structure with STO-609 (PDB:2ZV2) (Figure 3) [19]. We also controlled for hinge contacts in the model by using the 2,4dianilinopyrimidines as a training set (see Figure S2). We noted that STO-609 can displace a bound water molecule from the ATP-binding site that is still present when compound 2 is bound to the enzyme. The carboxylic acid of STO-609 also forms a tighter interaction with the catalytic lysine than the benzamide of compound 2. To improve its activity, we designed analogues of 2 that would be able to displace the bound water molecule and form stronger interactions with the catalytic lysine. In order to use molecular docking to guide the design of new analogues, we used the Maestro suite (2017-3) to create a models of compounds STO-609 and 2 bound to CaMKK2 that accurately reflected what was seen in the crystal structures of both [19]. Water Map using a 2 nS simulation was used to populate the hydration sphere of compounds STO-609 and 2. The resulting model correctly In order to use molecular docking to guide the design of new analogues, we used the Maestro suite (2017-3) to create a models of compounds STO-609 and 2 bound to CaMKK2 that accurately reflected what was seen in the crystal structures of both [19]. Water Map using a 2 nS simulation was used to populate the hydration sphere of compounds STO-609 and 2. The resulting model correctly identified the crystallographically observed water molecule highlighted in red (Figure 4). In addition, several other water molecules were identified in the 'back' pocket of the CaMKK2 that contained the catalytic lysine. Several analogues of compound 2 were designed to directly displace/interact with the key water molecule. In parallel, we designed a series of analogues to strengthen the interaction with the catalytic lysine.
Molecules 2018, 23, x 6 of 23 identified the crystallographically observed water molecule highlighted in red ( Figure 4). In addition, several other water molecules were identified in the 'back' pocket of the CaMKK2 that contained the catalytic lysine. Several analogues of compound 2 were designed to directly displace/interact with the key water molecule. In parallel, we designed a series of analogues to strengthen the interaction The docking simulations showed that strengthening the interaction with the catalytic lysine while retaining the bound water molecule gave better scores than direct displacement of the water ( Figure 5). We first optimized the core to see if the weak hit (2) could be a tractable starting point. Molecular simulations of compound 2 ( Figure 5A) showed a weaker interaction with the catalytic lysine when the primary amide was switched in compound 10 ( Figure 5B); the interaction with the water produced a more effective docking pose with a strong water mediated interaction with the backbone. We were able to boost the proposed lysine interaction with an imidazole substitution (11) ( Figure 5C). The carboxylic acid in STO-609 appeared to contribute significantly to the binding affinity to CaMKK2. We designed a switch of the primary amide in compound 2 to a carboxylic acid (12) ( Figure 5D) and this gave a 14/15 poses match to where STO-609's carboxylic acid was directed in the co-crystal structure with CaMKK2 [19]. The docking simulations showed that strengthening the interaction with the catalytic lysine while retaining the bound water molecule gave better scores than direct displacement of the water ( Figure 5). We first optimized the core to see if the weak hit (2) could be a tractable starting point. Molecular simulations of compound 2 ( Figure 5A) showed a weaker interaction with the catalytic lysine when the primary amide was switched in compound 10 ( Figure 5B); the interaction with the water produced a more effective docking pose with a strong water mediated interaction with the backbone. We were able to boost the proposed lysine interaction with an imidazole substitution (11) ( Figure 5C). The carboxylic acid in STO-609 appeared to contribute significantly to the binding affinity to CaMKK2. We designed a switch of the primary amide in compound 2 to a carboxylic acid (12) ( Figure 5D) and this gave a 14/15 poses match to where STO-609's carboxylic acid was directed in the co-crystal structure with CaMKK2 [19].

Outline of Compounds
We proposed a series of modifications of compound 2 relating to the crystal structure and modelling ( Figure 6). These included a switch of the para-amide to the ortho-position (10) to better interact with the water and a direct substitution of the para-amide with a carboxylic acid (12) to form

Outline of Compounds
We proposed a series of modifications of compound 2 relating to the crystal structure and modelling ( Figure 6). These included a switch of the para-amide to the ortho-position (10) to better interact with the water and a direct substitution of the para-amide with a carboxylic acid (12) to form a stronger interaction with the water network. A series of mono-substituted ortho-, metaand paracyano analogues (13-15, respectively) probed the space available in this pocket and checked conformation constraints. A substitution on the adjacent anilino-nitrogen to the hinge binder to directly replace the water with a 4-methyloxazole (16) was encouraging. A methanol substitution at the ortho position (17) also looked promising and to increase the π-stacking potential of this analogue we added a meta-trifluoromethyl group (18). The use of an imidazole as a hydrogen bond donor/acceptor and the model ( Figure 4C) suggested that, though out of plane, it could align with the catalytic lysine. The para-substituted imidazole (11) is about 30-40 • out of plane and looked favorable to form a networked interaction between Lys197 and the wider water network. The final analogue was an arylthiadiazinone with a substitution of 2-cyclopentylbenzoic acid (19). This tactic for interaction of the wider water network with the para-carboxylic acid while having an adjacent meta-cyclopentyl to form a π-stacking/lipophilic interaction was previously used to successfully target SGK1, a regulator of epithelial sodium channels (eNaCs) [20].
Interestingly, a different route was used to access imidazole derivative 11 as the Buchwald-Hartwig coupling of thiadiazinone 7 with 4-(1H-imidazol-2-yl)aniline led to a complex mixture of products. This was resolved by switching the reaction sequence and performing first the nucleophilic displacement of the 3-chloride of dichlorothiadiazinone 6 with 4-(1H-imidazol-2-yl)aniline to afford anilinothiadiazine 21 and subsequently performing the Buchwald-Hartwig coupling with 5-amino-2-methylphenol (Scheme 2).

Optimization Results on CaMKK2
To more accurately determine the relative changes in potency of the TDZ analogues and to enable measurement of accurate IC50 ' s, we developed a Time-Resolved Fluorescence Resonance Energy Transfer (TR-FRET) assay. The TR-FRET tracer displacement assays were generated using a protocol derived from the Lanthascreen binding assays (ThermoFisher Scientific, Waltham, MA, USA) [21]. In this assay, we measured the effect of ATP competing compounds that are able to displace a fluorophore-labeled pan-kinase inhibitor (tracer 236) from the ATP binding site. We used ponatinib and staurosporine as internal controls to calibrate the FRET assay. The results of the CAMKK2 FRET assay are shown in Figure 7 and Table 2 (see Figures S3 and S4, SI). Surprisingly, TDZ analogues 1 and 2 showed no measurable activity at a concentration up to 50 μM but analogue 3 gave weak activity with an IC50 34 μM. Nevertheless, several of the structure-optimized analogues showed improvements in potency. TDZ's 10-12 had the highest affinity for CAMKK2 with IC50 7.8, 3.2 and 10.5 μM, respectively.

Optimization Results on CaMKK2
To more accurately determine the relative changes in potency of the TDZ analogues and to enable measurement of accurate IC 50 ' s, we developed a Time-Resolved Fluorescence Resonance Energy Transfer (TR-FRET) assay. The TR-FRET tracer displacement assays were generated using a protocol derived from the Lanthascreen binding assays (ThermoFisher Scientific, Waltham, MA, USA) [21]. In this assay, we measured the effect of ATP competing compounds that are able to displace a fluorophore-labeled pan-kinase inhibitor (tracer 236) from the ATP binding site. We used ponatinib and staurosporine as internal controls to calibrate the FRET assay. The results of the CAMKK2 FRET assay are shown in Figure 7 and Table 2 (see Figures S3 and S4, SI). Surprisingly, TDZ analogues 1 and 2 showed no measurable activity at a concentration up to 50 µM but analogue 3 gave weak activity with an IC 50 34 µM. Nevertheless, several of the structure-optimized analogues showed improvements in potency. TDZ's 10-12 had the highest affinity for CAMKK2 with IC 50 7.8, 3.2 and 10.5 µM, respectively.
ponatinib and staurosporine as internal controls to calibrate the FRET assay. The results of the CAMKK2 FRET assay are shown in Figure 7 and Table 2 (see Figures S3 and S4, SI). Surprisingly, TDZ analogues 1 and 2 showed no measurable activity at a concentration up to 50 μM but analogue 3 gave weak activity with an IC50 34 μM. Nevertheless, several of the structure-optimized analogues showed improvements in potency. TDZ's 10-12 had the highest affinity for CAMKK2 with IC50 7.8, 3.2 and 10.5 μM, respectively.    The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17)(18) were also effectively inactive, likely due in part to the lack of ability to reach the water interaction.  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17)(18) were also effectively inactive, likely due in part to the lack of ability to reach the water interaction.  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17)(18) were also effectively inactive, likely due in part to the lack of ability to reach the water interaction.  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17)(18) were also effectively inactive, likely due in part to the lack of ability to reach the water interaction.  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17-  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17-  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15)   The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17-  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM

18
Molecules 2018, 23, x 10 of 23 The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17)(18) were also effectively inactive, likely due in part to the lack of ability to reach the water interaction.  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM

19
Molecules 2018, 23, x 10 of 23 The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 μM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17)(18) were also effectively inactive, likely due in part to the lack of ability to reach the water interaction.  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the paraamide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the monocyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 μM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15)  The switch of the amide from the para-position (2) to the ortho-position (10) provided a >8-fold boost likely related to the new more favorable water mediated interaction. The exchange of the para-amide (2) to the para-carboxylic acid (12) led to a >5-fold increase in potency. However, the mono-cyano group substitutions were relatively in-effective. The ortho-cyano (13) showed some activity (43 µM) likely do to the water network interaction but meta-cyano (14) and para-cyano (15) were >50 µM against CaMKK2. This was the same result for the 4-methyloxazole (16), which was surprising but could be related to an inability to be accommodated in the active site. The methanol analogues (17)(18) were also effectively inactive, likely due in part to the lack of ability to reach the water interaction. The imidazole (11) preformed well as we expected from our model ( Figure 4C) and appeared to form the water network interaction in the back pocket of CaMKK2. The direct carbon-carbon bonded para-carboxylic acid with adjacent meta-cyclopentyl compound (19) was only weakly active (38 µM) and was likely out of position on this scaffold to form the optimal interaction as in STO-609.

Advanced Enzyme Assay Results on CaMKK2 Demonstrating Functional Inhibition
To further characterize the activity of 10-12 as inhibitors of CAMKK2, the compounds were subjected to an enzyme inhibition assay. CaMKK2 activity was measured by determining the rate of transfer of radiolabeled phosphate from [γ-32 P]-ATP to a synthetic peptide substrate [22]. Compounds 10-12, when initially screened at a concentration of 1 µM, showed statistically significant inhibition of CAMKK2 kinase activity. The compounds were then screened at 7 concentrations (see Table S3, SI) to produce moderately potent IC 50 's. TDZs 10-12 were demonstrated to be competent inhibitors of the CaMKK2 enzyme with enzymatic IC 50 's of 11.9, 6.5 and 4.1 µM, respectively (Table 3).

Discussion
We demonstrate, for the first time, that the 4H-1,2,6-thiadiazin-4-one (TDZ) chemotype can function as an ATP-competitive kinase inhibitor. TDZ represents a novel hinge binder with the potential to be further optimized into a high quality chemical probe for kinases such as CaMKK2. Furthermore, we report the first protein co-crystallization with this rare heterocycle. The electronics of the TDZ core allows for participation of the sulfur atom to be part of extended conjugated electronic exchanges through the core units to transfer charge [23]. This electronic property, exploited in solar cell applications, can partly explain the general lack of kinome promiscuity compared to the dianilinopyrimidine. The modular synthesis and relative narrow kinome spectrum make the TDZ an attractive chemotype for further development.
CaMKK2 is predominantly expressed in the brain, with trace expression in peripheral tissues such as the testis, spleen and lung [24,25]. In addition to recently being linked to appetite in vivo [26], CAMKK2 is over-expressed in multiple cancers [27,28]. The knockout of CaMKK2 can reduce cell proliferation and tumorigenicity in vivo, making CAMKK2 an attractive drug target. The only reported potent small molecule inhibitor of CaMKK2 is STO-609, which has several liabilities limiting its use as a probe of CaMKK2 activity. These include poor solubility and unfavorable off-target kinome profile with kinases that would cloud the interpretation of a phenotype including ERK8, MNK1 and PIM3 [29][30][31].
In addition, STO-609 is an agonist of the arylhydrocarbon receptor (AhR) [32]. The complicating factors highlight the need for the design and development of high quality inhibitors targeting CAMKK2.
Our results add further credence to the importance of water networks in optimization of kinase inhibitors. The advent of powerful modelling tools such as Water Map and the Schrodinger Maestro platform have made manipulating the water network more accessible [33][34][35]. There are two distinct water network regions that the TDZ core can exploit in binding to CaMKK2. We have shown an ability to exploiting the water network and lysine interactions, we improved on the activity of compound 2 and produced compound 11 that is >15-fold more potent. Our discovery of the TDZ core as a useful chemotype for kinase inhibitor design adds a new a hinge binding heterocycle to the medicinal chemistry tool box and provides another example of the application of sulfur in drug design.
Starting from 100 µM protein stocks, our kinase panel enzymes were diluted to 1 µM in buffer 100 mM K 2 HPO 4 pH 7.5 containing 150 mM NaCl, 10% glycerol and 5X dye (Applied Biosystems catalogue 4461806). The protein/dye mixture was transferred to a 384-well PCR microplate having 20 µM per well. Compounds in DMSO at 10 mM concentration were added in 20 nL volume, using a liquid handling device setup with a pin head, to make 10 µM compound concentration in the assay plate.
Protein thermal shift data was measured in a qPCR instrument (Applied Biosystems QuantStudio 6) programmed to equilibrate the plate at 25 • C for 5 min followed by ramping the temperature to 95 • C at a rate of 0.05 • C/s. Data was processed using Protein Thermal shift software (Applied Biosystems) by fitting experimental curves to a Boltzmann function to calculate differential thermal shifts (dT m ) referenced to protein/dye in 0.2% DMSO.

Cloning, Protein Expression and Purification
The crystallization of CAMKK2 was performed with a construct of CAMKK2 isoform 7 residues 161-449 (NCBI NP_001257415.1 -SGC construct CAMKK2B-cb002) containing the wild-type kinase domain in vector pNIC28-Bsa4. The construct was transformed into BL21(DE3) Escherichia coli cells that co-express λ-phosphatase and three rare tRNAs (plasmid pACYC-LIC+). Cells were cultured in TB medium containing 50 µg/mL kanamycin and 35 µg/mL chloramphenicol at 37 • C with shaking until the OD600 reached~3 and then cooled to 18 • C for 1 h. Isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.1 mM and the cultures were left overnight at 18 • C. The cells were collected by centrifugation then resuspended in 2× lysis buffer [1× lysis buffer is 50 mM HEPES buffer, pH 7.5, 0.5 M KOAc, 10% (v/v) glycerol, 50 mM each arginine/glutamate, 10 mM imidazole, 1.0 mM tris(2-carboxyethyl)phosphine (TCEP), Protease Inhibitor Cocktail Set VII (Calbiochem, 1/500 dilution)] and flash-frozen in liquid nitrogen. Cells were lysed by sonication on ice. The resulting proteins were purified using Ni-Sepharose resin (GE Healthcare) and eluted stepwise in 1× lysis buffer with 300 mM imidazole. Removal of the hexahistidine tag was performed at 4 • C overnight using recombinant TEV protease. The protein was further purified using reverse affinity chromatography on Ni-Sepharose followed by gel filtration (Superdex 200 16/60, GE Healthcare). The protein in gel filtration buffer (10 mM HEPES, 500 mM KOAc, 1.0 mM TCEP, 5% (v/v) glycerol, 50 mM each arginine/glutamate) was concentrated to 8.5 mg/mL (measured by UV absorbance in a NanoDrop spectrophotometer (Thermo Scientific, Waltham, MA, USA) using the calculated molecular weight and estimated extinction coefficient) using 30 kDa molecular weight cut-off centrifugal concentrators (Sigma-Aldrich Corp., St. Louis, MO, USA) at 4 • C. The concentrated protein was flash-frozen in a liquid nitrogen bath and stored at −80 • C until use.

Protein Crystallization
Kinase inhibitor (dissolved in 100% DMSO) was added to the protein in 3-fold molar excess and incubated on ice for approximately 30 min. The mixture was centrifuged at 15,000 rpm for 10 min at 4 • C before setting up 150 nL volume sitting drops at three ratios (2:1, 1:1, or 1:2 protein-inhibitor complex to reservoir solution). Crystallization experiments were performed at 20 • C. Crystals were cryoprotected in mother liquor supplemented with 25-30% glycerol before flash-freezing in liquid nitrogen for data collection. Diffraction data were collected at 100 K at Diamond Light Source beamline I03. Crystal optimization used Newman's buffer system [36].

Structure Solution and Refinement
Diffraction data were integrated using XDS [37] and scaled using AIMLESS from the CCP4 software suite (version 7.0.057, London, UK) [38]. Molecular replacement (MR) was performed with Phaser [19] using the CAMKK2 bound to STO-609 co-structure (PDB ID 2ZV2) [19]. Automated refinement was performed with Refmac [39,40]. Coot [41] was used for manual model building and refinement. Structure validation was performed using MolProbity [42]. Structure factors and coordinates have been deposited in the PDB (see Table S2, SI).

Molecular Modelling
Molecular modelling was performed using Schrödinger Maestro software package (version 2018-1, Schrödinger, Mannheim, Germany) [43]. Structures of small molecules were prepared using and the LigPrep module of Schrodinger suite employing OPLS3 force for all computations. X-ray crystal structure for the CaMMKK2 (PDB:5VT1/2ZV2) was pre-processed using the protein preparation wizard of Schrödinger suite to optimize the hydrogen bonding network [43].
Prior to Glide docking, the grid box was centered using corresponding x-ray ligand as template. The ligand docking was performed using default SP settings of Schrodinger Glide with softened vdw's potential (0.6) and additional hydrogen bond constraints to NH of V270 (hinge residue). Graphical illustrations were generated using Schrödinger Maestro software (version 2018-1, Schrödinger, Mannheim, Germany).

Hydration Site Analysis
Hydration site analysis calculated with Water Map (Schrödinger Release 2017-3: Water Map, Schrödinger, LLC, New York, NY, 2017. The 5VT1 structure was prepared with Protein Preparation Wizard (as above). Waters were analyzed within 6 Å of the co-crystallized ligand and the 2 ns simulation was conducted with OPLS3 force field.

CaMKK2B TR-FRET Assay
CAMKK2 kinase domain (132-470) was cloned in a pNIC-Bio2 vector in fusion with N-terminal 10xHis tag followed by a TEV protease cleavage site and a C-terminal biotin ligase recognition sequence. This construct was used in the expression of CAMKK2 in E. coli BL21(DE3)-R3-BirA [44]. Protein was purified in a Ni-NTA column (Thermo Scientific, Waltham, MA, USA) followed TEV digestion overnight, dialysis to remove imidazole and re-purification in Ni-NTA to remove undigested samples and TEV protease (made in house with an N-terminal 6xHis tag). As a last step, this sample was loaded to a HiLoad Superdex 200 16/600 column (GE Healthcare, Chicago, IL, USA) for final polishing and buffer exchange.

General Methods and Materials
All chemicals were commercially available except those whose synthesis is described. Anhydrous Na 2 SO 4 was used for drying organic extracts and all volatiles were removed under reduced pressure. 1,4-Dioxane was dried by refluxing over CaH 2 . All reaction mixtures and column eluents were monitored by TLC using commercial glass backed thin layer chromatography (TLC) plates (Merck Kieselgel 60 F 254 ) [46]. The plates were observed under UV light at 254 and 365 nm. The technique of dry flash chromatography was used throughout for all prep scale chromatographic separations using Merck Silica Gel 60 (less than 0.063 mm). Melting points were determined using a PolyTherm-A, Wagner & Munz, Koefler-Hotstage Microscope apparatus or were determined using a TA Instruments DSC Q1000 with samples hermetically sealed in aluminum pans under an argon atmosphere; using heating rates of 5 • C/min (DSC m.p. listed by onset and peak values). Solvents used for recrystallization are indicated after the melting point. UV spectra were obtained using a Perkin-Elmer Lambda-25 UV/vis spectrophotometer and inflections are identified by the abbreviation "inf." IR spectra were recorded on a Shimadzu FTIR-NIR Prestige-21 spectrometer with Pike Miracle Ge ATR accessory and strong, medium and weak peaks are represented by s, m and w, respectively. 1 H and 13 C-NMR spectra were recorded on a Bruker Avance 300 (at 300 and 75 MHz, respectively), or a 500 machine (at 500 and 125 MHz, respectively). Deuterated solvents were used for homonuclear lock and the signals are referenced to the deuterated solvent peaks. APT NMR studies identified quaternary and tertiary carbons, which are indicated by (s) and (d) notations, respectively. MALDI-TOF mass spectra were recorded on a Bruker Autoflex III Smartbeam instrument.