The overall shape of p iSH
The overall shape of p110α/iSH2 is dramatically different from DGKB or SK1, instead of forming a homodimer, p110α/iSH2 heterodimer has a triangular shape with iSH2 laying on the top and ABD, RBD, helical and kinase domain of p110α sitting in the bottom (Fig. 2B) (Gabelli et al., 2010; Huang et al., 2007). The C2 domain of p110α is in close proximity to iSH2 and both of them are thought to directly contact the membrane (Gabelli et al., 2010; Huang et al., 2007). While the N terminus of p110α is mainly involved in interacting with the regulatory subunit, the C-terminal catalytic domain of p110α maintains a conserved two lobes structures of αβ fold common to general kinase. The ATP binding site locates in the cleft of the two lobes as illustrated in Fig. 2B (Gabelli et al., 2010).
With respect to its catalytic mechanism, PI3Ks have an interesting history. While this enzyme is known for its lipid kinase activity, it also has protein kinase activity (Dhand et al., 1994; Foukas and Shepherd, 2004). This has led to the recognition that it is has dual kinase specificity: a lipid kinase activity that catalyzes the ATP-dependent phosphorylation of the 3′-hydroxyl of phosphoinositides and a protein-kinase activity that includes the catalysis of autophosphorylation (Dhand et al., 1994; Foukas and Shepherd, 2004). Recently, critical catalytic residues have been identified as well as and a potential catalytic mechanism involving a ternary complex with ATP and PI(4,5)P2 has been proposed (Maheshwari et al., 2017) The data indicate that Lys776, within the P-loop region, is involved in both lipid and ATP substrate recognition and is critical for autophosphorylation. One interesting observation involved the roles of His936 and His917. While mutation of His917 abolished both lipid and autophosphorylation, mutation of His936 abolished lipid kinase activity only. This is similar to studies of another lipid kinase, a PtdIns-5 kinase (PIKfyve), that also displays protein kinases activity. A double mutation in a putative lipid substrate binding region of this enzyme (K1999E/K2000E) ablates lipid kinase activity but not protein kinase activity (Ikonomov et al., 2002).
Lessons from known structures and catalytic mechanisms The catalytic mechanism and molecular mechanisms underlying the regulation of mammalian DGKs is of major interest. This is because this knowledge will not only fill an important gap in our understanding of these enzymes, such information is also critical to the generation of specific inhibitors. In terms of catalytic mechanism, it is noteworthy that most kinases utilize a catalytic N6-Methyl-ATP that abstracts a proton from the substrate hydroxyl and orients the substrate for in-line attack (associative) on the ATP γ-phosphate (Mildvan, 1997). For DGKB, a glutamate (E273) was proposed to serve this role (Miller et al., 2008). This is similar to the proposed role for glutamate-69 (E69) in DGKA (Li et al., 2015). In sphingosine kinase (SK1), aspartate-61 (D61) is likely the catalytic base for this enzyme (Wang et al., 2013). These data suggest mammalian DGKs will employ an associative mechanism involving a catalytic base. It is noteworthy, however, that there are data to suggest some tyrosine kinases use a dissociative mechanism involving attack on a metaphosphate intermediate (Wang and Cole, 2014). Hopefully, future studies will clarify the catalytic mechanism of the mammalian DGKs. While a complete review of the molecular mechanisms regulating mammalian DGK activities is beyond the scope of this review, other reviews have covered much of this aspect (Shulga et al., 2011a; Topham and Epand, 2009; Tu-Sekine and Raben, 2011). In general, aside from enzyme expression levels, this regulation may involve one of more of the following: interactions with specific ions, protein-protein or protein-lipid interactions, post-translational modifications, membrane localization/substrate availability. By way of illustration, some isoforms (Class I DGKs) are sensitive to calcium levels, and responses to specific phospholipids varies among the various isoforms (Shulga et al., 2011b; Topham and Prescott, 2002). Additionally, phosphorylation may modulate activities of some DGK isoforms. For example, membrane association of DGK-δ (Imai et al., 2002) and DGK-θ (van et al., 2005), the nuclear localization of DGK-ζ (Topham et al., 1998), and intrinsic activity of DGK-α (Baldanzi et al., 2008) may involve phosphorylations. DGK-δ may also be regulated by oligomerization (Knight et al., 2010). In contrast, our understanding of protein modulators of DGK activities is less clear. DGK-ζ is activated by the hypo-phosphorylated Rb protein (pRb) as well as two related pocket proteins p107 and p130 (Los et al., 2006), while DGK-θ is inhibited by RhoA (Houssa et al., 1999), and is activated by proteins containing polybasic rich regions, termed polybasic activators (PBAs) (Tu-Sekine et al., 2013; Tu-Sekine and Raben, 2012). Importantly, how these mechanisms alter DGK conformations to affect activity remains unresolved. It is tempting to speculate that a hinge region similar to that seen in DGKB may be involved in modulating DGK structures but this appears to be unclear for SK1 (Wang et al., 2013). Clearly, there is a critical need for information regarding the three-dimensional structure of mammalian DGKs which will help elucidate the catalytic mechanisms. Solving these structures by rather conventional approaches, such as x-ray crystallography or NMR, have been hampered largely by the inability to obtain sufficient quantities of highly purified, monodispersed enzymes. The rapid development of single-particle cryo-electron microscopy (cryo-EM) has made it the most compelling technique for solving these challenging protein structures. Indeed, cryo-EM has recently been used to successfully solve the structure of a complex containing a lipid kinase, PtdIns4KIIIα, and two regulatory subunits designated TT7 and FAM126 (Lees et al., 2017). Importantly, besides solving large symmetric proteins, cryo-EM has been used to solve the structure of proteins size around 100 kD. Such small proteins were once thought to be beyond the resolution limit of cryo-EM but recent have advances have overcome the original limitations. For example, cryo-EM has been used to determine the structures of isocitrate dehydrogenase (IDH, 93 kD) and lactate dehydrogenase (LDH, 145 kD) at near atomic resolution (Merk et al., 2016). It is hoped that new approaches such as cryo-EM will finally allow us to obtain high resolution structural data of mammalian DGKs.