Inhibitor Cocrystal Structures 2.1

Structures of PKA with Isoquinoline Sulphonamide Derivatives

PKA was among the first kinases cocrystallized with protein kinase inhibitors (Schulze Gahmen et al. 1995; De Azevedo et al. 1996; Engh et al. 1996; Xu et al. 1996). The isoquinoline sulphonamide derivatives H7 (1-(5-isoquinolinesulfonyl)-2-methylpiperazine), H8 (N-[2-(methylamino)-ethyl]-

5-isoquinolinesulphonamide), and H89 (N-[2-(p-Bromocinnamylamino)eth-yl]-5-isoquinolinesulphonamide) were chosen, because these inhibitors were at that time in wide use, and all of them inhibit PKA (PDB-codes 1YDR, 1YDS, 1YDT). H7 has a broad specificity within the AGC kinases and inhibits PKC as well as PKA in the micromolar range. Its ability to inhibit PKC (at 6 mM) was the reason for the enormous popularity of H7, used by many laboratories to dissect the role of PKC in signal transduction. Rho kinase, actually, is a 20-fold better target for H7 (Uehata et al. 1997) (300 nM), but this enzyme was not described before 1995 (Leung et al. 1995). H8, a micromolar inhibitor of PKA and PKC, inhibits PKG also. Still, H89 (48 nM for PKA) is one of the more selective PKA inhibitors, although when tested against a wider panel of protein kinases, some other AGC kinases, such as Rho kinase and PKB/Akt from the AGC group and S6K1 and MSK1 from other branches of the kinase family, are found to be inhibited by H89 too at comparable concentrations.

The H-inhibitors are characterized by an isoquinoline sulphonamide group, and a sidechain, where a two-carbon spacer separates two amide groups.

The binding of H7 to PKA is representative for some typical aspects of PKA inhibitor binding. The conserved isoquinoline moiety of the inhibitor, a planar double ring with one nitrogen atom acting as proton acceptor, occupies the position of the adenine purine of ATP and mimics the ATP N1 H-bond to the hinge region Val123 amide (Fig. 1). The planar isoquinoline group is embedded between the residues Val57, Ala70 and Leu49 from the small lobe, and Leu173 from the large lobe; these interactions contribute significantly to the number of van der Waals (VDW) contacts to the enzyme. Usually protein kinase inhibitors bind alike, with a proton accepting interaction from a planar structure. Few structures are known where the inhibitor does not follow this pattern. TBB (tetrabromo-2-benzotriazole) (PDB-code 1J91) is such an example, which makes no hinge atom contact in CK2 (Battistutta et al. 2001); however, this appears to be a specific feature of CK2,


Fig. 1 Superimposition of H8 and MnAMP-PNP, showing spatial congruence with adenosine group and of function groups. (Engh et al. 1996)

because the same inhibitor binds in CDK2 according to the common pattern (De Moliner et al. 2003). Also the inhibitor BIRB 796 (PDB-code 1KV1) binds to a 'DFG-out' conformation of P38 in an atypical way without a hinge region contact (Pargellis et al. 2002). In many AGC kinases, such as PKA, or PKB, a residue from outside the catalytic core, Phe327, contributes to the completeness of the adenosine binding site. Phe327 or its homologue is positioned on a C-terminal polypeptide stretch that expands from the large lobe via the catalytic cleft to the N-terminal lobe and ends in the hydrophobic motif beside helix C. This aromatic residue is not conserved outside the group of AGC kinases, nor has a comparable contribution been observed in other kinases so far. The H-inhibitors mimic not only the purine of ATP, but also aspects of the ribose. The 2' and 3' OH groups of ATP in PKA make contacts to the backbone carbonyl of Glu170, and in addition to the side-chain of Glu127, the only charged residue in the adenosine pocket. The sul-phonamide group together with the sidechain mimics a part of the ribose spatially, and the distal amine of the aminoethyl sidechain of most H-in-hibitors forms a similar contact to the Glu170 carbonyl. Interestingly, one of the sulphonyl oxygens is in the same spatial position as the ring oxygen of the ribose. This is not only observed with H7, but is a general feature of H-inhibitor binding in PKA, and is observed also with the indolocarbazole staurosporine. In none of these cases is a typical H-bond contact formed from the oxygen, but the Ca-hydrogen of Gly50 is close enough for a weak CH-O interaction, as was postulated for staurosporine (Zhu et al. 1999). The triphosphoryl subsite, however, is not occupied by most of the PKA inhibitors, with the exception of balanol (see below), which makes electrophil-ic contacts to Lys72 (in its normal function involved in binding the phos-phoryl groups) and the backbone amides of the glycine loop. The consequence is that, in the case of H7 and H8, the glycine-rich flap is not in contact with the inhibitors and has an increased mobility. This is different in the case of H89, which possesses a large bromocinnamyl sidechain, binding underneath and stabilizing the structure of the glycine flap (Fig. 2). The specificity of the H-inhibitors is defined by differences in their sulphona-mide substituents, as all (with the exception of H-1152P) have identical iso-quinoline sulphonamide head groups. Accordingly, the number of VDW contacts to the isoquinoline is almost identical for all three inhibitors, only H89 is packed slightly better. Differences are observed for the contacts of the amino group of the sidechains, although this group is present in all three inhibitors. Only H8 makes a contact from here to the sidechain of Asp184. Asp184 usually has a different rotamer and is in contact with the basic charge of Lys72, or, in the ATP-bound structure, with one of the metal ions. Apparently, the distal amine in the H8 sidechain attracts this residue more than the comparable nitrogens in H7 or H89, which even is structurally identical to H8 in this region. The bromocinnamoyl group and the pipera-zine ring weaken the partial-positive charge of the secondary amine, which

Fig. 2 H89 in the binding pocket of PKA. The bromocinnamoyl group interacts with the glycine-rich motif. (Engh et al. 1996)

then is less attractive for the Asp residue and does not induce a conforma-tional change. The overall number of polar and VDW interactions of these inhibitors varies; H89 has three H-bond contacts to the enzyme and the highest number of VDW interactions. A correlation of structural features to the inhibitory activity was found in the total area of the buried surface of the inhibitor (Engh and Bossemeyer 2002). Inhibitor selectivity in closely related kinases is determined mostly by sidechain differences in the ATP pocket. One PKA related kinase, hardly inhibited by H89, is PDK1. Few residues differ in the ATP binding pocket between PDK1 and PKA: Val123 is Ala in PDK1, Met120 is Leu in PDK1; both residues, however, interact with the iso-quinoline head group, and not with the H89-specific sidechain. A significant difference in a sidechain-critical region is Gly55, the first residue following the turn in the glycine-rich beta sheet. This residue is Ser in PDK1. Interestingly, Phkg, although also quite similar to PKA in the binding pocket, is also much more weakly inhibited by H89 and also contains a serine residue in the Gly55 position. Another general difference between PKA and PDK1 is the absence of the C-terminal stretch with the phenylalanine residue which, as Phe327, makes contacts to the isoquinoline of all H-inhibitors. Perhaps the binding pocket for comparably small molecules such as H-inhibitors often requires some completeness, provided for most AGC kinases by this C-terminal phenylalanine residue. It cannot be excluded that this residue is one reason for the general preference of several H-inhibitors, such as H89, or HA1077 for AGC kinase, such as MSK1 or S6kinase (Davies et al. 2000). Several derivatives of the isoquinoline sulphonamides have been made which are active against protein kinases. A cognate of H7, HA-1077 or fa-sudil, was especially successful as an inhibitor of Rho kinase (see the chapter on fasudil by Hidaka et al., this volume). Fasudil and two other Rho kinase-selective inhibitors, H-1152P and Y-27632, were cocrystallized with PKA (Breitenlechner et al. 2003).

Was this article helpful?

0 0

Post a comment