Phosphatidylinositol 5-Phosphate 4-Kinase
Pioneering work in the early 1950s led to the elucidation of the roles of multiple kinases and phosphatases required to generate the diverse array of inositol lipid species in the PI cycle and in the numerous regulatory, signaling, and metabolic pathways that they are involved in (reviewed in Anderson et al. 1999). The first activities, isolated from human erythrocytes, to convert PtdIns4P to PtdIns(4,5)P2 were classified as Type I and Type II enzymes (Bazenet et al. 1990). Subsequent studies revealed that these two enzymes had distinct substrate specificities, and that the Type II enzyme was in fact solely utilizing PtdIns5P, a contaminant of the commercial PtdIns4P substrate in the assay (Rameh et al. 1997). Although the preferred substrate of PI5P4K is PtdIns5P it has also been shown to phosphorylate PtdIns3P in vitro (Fig. 2), producing PtdIns(3,4)P2 (Clarke et al. 2007). Both of these substrates are minor components of the typical cellular phosphoinositide content, PtdIns5P and PtdIns3P constituting 2% and 5% of total PtdInsP, respectively, the majority being PtdIns4P (Toker 2002). As a consequence, the contribution of PI5P4K activity to cellular PtdIns(4,5)P2 synthesis is thought to be restricted to specific isolated compartments, or another possible role of the enzyme is to attenuate the putative signaling function of PtdIns5P itself. Three mammalian PI5P4K isoforms have so far been discovered (α, β, and γ) and each has been shown to have a specific distribution and subcellular localization (Clarke et al. 2010).
PI5P4K Molecular Evolution
Although the PI5P4K isoforms have a significant level of identity at the amino acid level, they share only 30% identity with the PI4P5Ks and constitute a novel family with no significant identity to any other lipid or protein kinases (Anderson et al. 1999).
The N-terminal domain has seven β-sheets and four α-helices, and the crystal structure shows that the protein homodimerizes along the interface of two adjacent β1 strands (Fig. 4) to form a broad, flat face that is able to contact cell membranes by means of electrostatic interaction with lipid head-groups (Rao et al. 1998). Recent data has also shown the potential of the PI5P4K isoforms to heterodimerize, suggesting a possible means of inter-isoform regulation based on protein expression levels in different tissues (Bultsma et al. 2010; Wang et al. 2010; Clarke and Irvine 2013). This domain also contains a putative ATP-stabilizing G-loop motif (Rao et al. 1998).
The C-terminal domain consists of five β-sheets and four α-helices and contains the “activation loop” and “variable insert” (Fig. 4). The activation loop region (amino acids 373–391) confers substrate specificity as demonstrated by domain swapping between PI5P4K and PI4P5K enzymes (Kunz et al. 2000). The variable insert region interrupts the catalytic kinase core and represents the region of least similarity between the PI5P4K isoforms. In PI5P4Kβ the α7 helix in this insert serves as a specific nuclear localization motif (Ciruela et al. 2000), and other inter-isoform differences could be accounted for in this region.
Mammalian PI5P4K Isoforms
Comparison of activities of the PI5P4K isoforms (Clarke and Irvine 2013). Specific activity is defined as nmoles of ATP incorporated into lipid per minute per mg of protein under the same assay conditions. Turnover number (kcat) is defined as the number of ATP (hence PtdIns5P to PtdIns(4,5)P2) reactions per second
KM for ATP (μM)
3.7 × 10−2 (±0.12)
1.1 × 10−2
1.6 × 10−5 (±0.08)
9.6 × 10−5
3.7 × 10−7 (±0.12)
5.7 × 10−6
Yeast two-hybrid screening of a murine cerebellar cDNA library, using a cytoplasmic domain of the human p55 TNF-α receptor as bait, resulted in the cloning of the PI5P4Kβ isoform (Castellino et al. 1997). Although this isoform has 77% amino acid similarity to PI5P4Kα, it has 100-fold less in vitro activity (Table 1). PI5P4Kβ mRNA was detected in all tissues tested and was found to be specifically enriched in skeletal muscle and brain (Fig. 5 and Castellino et al. 1997). At a subcellular level, although PI5P4Kβ has been shown to associate with the TNF-α, EGF, and ErbB2 (HER2) receptors (Castellino et al. 1997; Castellino and Chao 1999), and undoubtedly has a cytoplasmic role, it is also the only isoform to specifically localize to the cell nucleus. The targeting of PI5P4Kβ to the nucleus is affected by a nuclear localization sequence unique to this isoform (Ciruela et al. 2000), and genomic tagging indicates that the majority of endogenous PI5P4Kβ is transported across the nuclear membrane in DT40 cells (Richardson et al. 2007). Activation of PI5P4Kβ has been shown to be mediated by TNF-α in specific cytokine-responsive cells (Castellino et al. 1997). Within the nucleus, PI5P4Kβ activity is inhibited by p38 mitogen-activated protein kinase (MAPK) phosphorylation of serine 326 after ultraviolet irradiation, leading to increases in nuclear PtdIns5P, which is sensed by the cell stress–responsive nuclear adapter protein ING2 (Jones et al. 2006); note that this residue is also present in PI5P4Kα, so it is possible that nuclear PI5P4Kα (see above) is similarly regulated. PI5P4Kβ that is co-localized to nuclear speckles (pre-mRNA processing sites) with the Cul3-SPOP ubiquitin ligase complex is itself targeted for degradation by ubiquitylation (Bunce et al. 2008). Intriguingly, PtdIns5P activation of p38 MAPK stimulates the activity of the Cul3-SPOP complex, suggesting that a feed-forward mechanism of regulation exists, which could in part be explained by the interaction between PI5P4Kβ and PI5P4Kα in the nucleus (Bultsma et al. 2010). There is also evidence that PI5P4Kβ is able to regulate insulin-induced signaling in cells by an indirect mechanism. Carricaburu et al. showed that PI5P4Kβ-regulated PtdIns5P levels were responsible for activation of a PtdIns(3,4,5)P3-specific 5-phosphatase, which removed this PI 3-kinase (PI3K) signaling product, resulting in reduced Akt/PKB protein kinase phosphorylation (Carricaburu et al. 2003).
The PIP4Kγ isoform is the most recent to be discovered and was isolated from a rat brain cDNA library by Itoh et al. in 1998. PI5P4Kγ shares approximately 63% similarity with PI5P4Kα and PI5P4Kβ at the protein level, but has a 2,000-fold lower in vitro catalytic turnover of PtdIns5P than the PI5P4Kα isoform (Table 1), and no detectable activity against any other phosphoinositide substrates. Highest levels of PI5P4Kγ expression are observed in the kidney and brain (Fig. 5), with significant levels also in the heart, ovary, and testis (Itoh et al. 1998; Clarke et al. 2008, 2009). Within the kidney, the expression is limited to the distal nephron and restricted to the epithelial cells of the thick ascending limb of the loop of Henle and to intercalated cells of the collecting duct (Clarke et al. 2008). Expression of PI5P4Kγ is also seen at early developmental stages in the zebra fish embryo, localized to the pronephric duct and brain. In the mouse brain PI5P4Kγ has a restricted expression, found only in specific neuronal subpopulations, such as cerebellar Purkinje cells, hippocampal pyramidal cells, and mitral cells in the olfactory bulb, and is excluded from granule cells (Clarke et al. 2009). At a subcellular level, endogenous PI5P4Kγ is partially associated with the cis-Golgi matrix and predominantly with an unidentified vesicular compartment, which in kidney epithelial cells is concentrated at the apical secretory membrane (Clarke et al. 2008). PI5P4Kγ undergoes protein phosphorylation on serine residues in vivo, and at least two phosphorylated forms are differentially expressed in mouse brain regions (Itoh et al. 1998; Clarke et al. 2009). In a β-pancreatic cell line, the mTORC1 complex is responsible for phosphorylation of PI5P4Kγ at two specific sites and contributes to the feedback regulation of mTORC1 activation during starvation (Mackey et al. 2014), potentially by a localization mechanism rather than in vivo activation of the enzyme. Phosphorylation may also be regulated by mitogenic cell stimulation with EGF or PDGF and to a lesser extent with lysophosphatidic acid or bradykinin (Itoh et al. 1998). Enzyme expression may also be regulated in thyrocytes by thyroid-stimulating hormone (Park et al. 2001).
PI5P4Ks and Disease
Aberrant regulation of cellular phosphoinositides has been implicated in a number of diseases including cancer and diabetes. Interference with PtdIns(3,4,5)P3 regulation of Akt signaling via enzymes such as PI3K and the phosphatases PTEN and SHIP1 are linked to many forms of cancer (McCrea and De Camilli 2009). PtdIns(4,5)P2 can also regulate ion channels, the dysfunction of which can lead to epilepsy and rare forms of congenital channelopathies (Halstead et al. 2005). In these cases any involvement of the PI5P4Ks would be indirect, via the downstream production of lipid substrates and second messengers themselves. However, some direct effects of the PI5P4Ks in disease have been reported.
PI5P4Kβ, as discussed above, has been implicated in the regulation of Akt/PKB activation during insulin signaling by reducing PtdIns5P levels that inhibit the enzymatic removal of PtdIns(3,4,5)P3 (Carricaburu et al. 2003). In agreement, PI5P4Kβ-/- knockout mice are viable but hypersensitive to insulin, suggesting that the presence of the PI5P4Kβ substrate, PtdIns5P, is directly required for maintenance of PtdIns(3,4,5)P3 levels for enhanced Akt/PKB phosphorylation. Removal of PtdIns5P by PI5P4Kβ reduces the effectiveness of insulin signaling, which may have consequences in type 2 diabetes (Lamia et al. 2004).
Early reports suggest that PIPK activity was upregulated in malignant tumors (Singhal et al. 1994). Recent studies have shown that as well as PI5P4Kβ interacting with the HER2b receptor (see above), the PIP4K2B gene is located close to the HER2 locus on chromosome 17. Gene amplification of HER2 occurs in 25% of breast cancer cases, and increased levels of PI5P4Kβ have also been detected in these tumors (Luoh et al. 2004). Recently depletion of PI5P4Kβ in HER2-positive breast cancer cell lines (p53 deficient) was shown to markedly reduce tumorigenesis (Emerling et al. 2013). PIP4K2B has also been linked to neuroblastoma development, as this gene is one of three to be disrupted by a translocation breakpoint characteristic of this tumor type (Schleiermacher et al. 2005).
Treatment of bipolar disorder with lithium is thought to be involved with phosphoinositide metabolism. Subsequent linkage studies have implicated PI5P4Kα as a candidate gene for susceptibility to schizophrenia, either through a common single nucleotide polymorphism or transcriptional repression (Stopkova et al. 2003).
Recent observations that inhibition of PI5P4Ks, particularly PI5P4Kγ, can positively regulate autophagy in cells (Vicinanza et al. 2015, Droubi et al. 2016), concomitant with reports identifying isoform-specific inhibitors (Voss et al. 2014, Clarke et al. 2015), suggest that these enzymes may also have a potential role in combatting neurodegenerative disease.
PI5P4K enzymes phosphorylate their preferred phospholipid substrate, PtdIns5P, to produce PtdIns(4,5)P2. The relative scarcity of PtdIns5P in cells suggests that the PI5P4Ks function to remove a lipid signal provided by this phosphoinositide, or to produce a small pool of PtdIns(4,5)P2 that is required in an isolated cellular compartment. PI5P4K exists as three distinct isoforms (α, β, and γ) with PI5P4Kα being significantly more active than PI5P4Kβ or PI5P4Kγ. Each isoform has a different tissue distribution and subcellular localization, and this is mediated in part by differences in highly variable regions within the more conserved common catalytic core domain. The isoforms form dimers by interaction between β1-sheets, and the formation of heterodimers suggests a mechanism by which isoforms are targeted to different cellular locations in a tissue and possibly cell-type-dependent manner. PI5P4Ks may have direct links to diabetes and neurological disorders, and PI5P4Kβ has recently been shown to be upregulated in human cancers, suggesting that these enzymes may be valid targets for disease treatments. The recent discovery of a number of inhibitors for PI5P4Ks also suggests that these targets are potentially druggable.
- Carricaburu V, Lamia KA, Lo E, Favereaux L, Payrastre B, Cantley LC, et al. The phosphatidylinositol (PI)-5-phosphate 4-kinase type II enzyme controls insulin signaling by regulating PI-3,4,5-trisphosphate degradation. Proc Natl Acad Sci USA. 2003;100(17):9867–72.PubMedPubMedCentralCrossRefGoogle Scholar
- Emerling BM, Hurov JB, Poulogiannis G, Tsukazawa KS, Choo-Wing R, Wulf GM, Bell EL, Shim H-S, Lamia KA, Rameh LE, Bellinger G, Sasaki AT, Asara JM, Yuan X, Bullock A, DeNicola GM, Song J, Brown V, Signoretti S, Cantley LC. Depletion of a putatively druggable class of phosphatidylinositol kinases inhibits growth of p53-null tumors. Cell. 2013;155:844–57.PubMedPubMedCentralCrossRefGoogle Scholar
- Voss MD, Czechtizky W, Li Z, Rudolph C, Petry S, Brummerhop H, Langer T, Schiffer A, Schaefer H-L. Discovery and pharmacological characterization of a novel small molecule inhibitor of phosphatidylinositol-5-phosphate 4-kinase, type II, beta. Biochem Biophys Res Commun. 2014;449(3):327–31.PubMedCrossRefGoogle Scholar
- Wang M, Bond NJ, Letcher AJ, Richardson JP, Lilley KS, Irvine RF, et al. Genomic tagging reveals a random association of endogenous PtdIns5P 4-kinases IIalpha and IIbeta and a partial nuclear localization of the IIalpha isoform. Biochem J. 2010;430(2):215–21.PubMedPubMedCentralCrossRefGoogle Scholar