The Role of Phosphoinositide 3-Kinase Lipid Products in Cell Function*

Phosphoinositide 3-kinases (PI 3-Ks) are a subfamily of lipid kinases that catalyze the addition of a phosphate molecule specifically to the 3-position of the inositol ring of phosphoinositides. Phosphatidylinositol (PtdIns), the precursor of all phosphoinositides (PI), constitutes less than 10% of the total lipid in eukaryotic cell membranes (Fig. 1). Approximately 5% of cellular PI is phosphorylated at the 4-position (PtdIns-4-P), and another 5% is phosphorylated at both the 4and 5-positions (PtdIns-4,5-P2). However, less than 0.25% of the total inositol-containing lipids are phosphorylated at the 3-position, consistent with the idea that these lipids exert specific regulatory functions inside the cell, as opposed to a structural function. To date, nine members of the PI 3-K family have been isolated from mammalian cells. They are grouped, as suggested by Domin and Waterfield (1), into three classes according to the molecules that they preferentially utilize as substrates. Four different lipid products can be generated by the different PI 3-K members: the singly phosphorylated form PtdIns-3-P; the doubly phosphorylated forms PtdIns-3,4-P2 and PtdIns-3,5-P2; and finally the triply phosphorylated form PtdIns-3,4,5-P3 (Fig. 1). PI 3-K was first described as a PI kinase activity associated with the viral oncoproteins, v-Src, v-Ros, and polyomavirus middle T. Mutational studies of these oncoproteins more than 10 years ago indicated a critical role for the associated PI kinase in cell transformation (reviewed by Ref. 2). Recent advances in the field have been achieved by the development of new techniques to probe for the direct targets of PI 3-K lipid products. The chemical synthesis of short chain fatty acid versions of these lipids (3–5) has been a crucial step in determining the specificity of lipid-binding proteins. Additionally, new cloning strategies have been developed to isolate new lipid-binding proteins (6). Here we will review the most recent advances in our understanding of the role of PI 3-K in cell function by dissecting the contribution of each of its lipid products.

Consistent with this, integrin cross-linking causes PtdIns-3,4-P 2 production without PtdIns-3,4,5-P 3 production and results in Akt activation (27). In vivo, full activation of Akt also depends on the phosphorylation of a threonine (Thr-308) and a serine (Ser-473) (33). Phosphorylation of these residues was shown to be dependent on PI 3-K. The Thr-308 kinase PDK1 (for phosphoinositide-dependent kinase; also called PKB kinase) was recently purified and cloned based on its ability to catalyze the phosphorylation of Thr-308 of Akt in a PtdIns-3,4,5-P 3 -dependent manner (34 -37). Like Akt, PDK1 contains a PH domain and binds with high affinity to PtdIns-3,4-P 2 and PtdIns-3,4,5-P 3 (37). To understand the mechanism by which these lipids activate Akt in vivo, a model was proposed in which the PI 3-K lipid products are involved in recruiting Akt and its upstream kinases to the membrane and also in promoting conformational changes in Akt that expose Thr-308 and Ser-473 to be phosphorylated by PDK1 and other kinases (reviewed in Refs. 29 and 38).
PDK1 can also phosphorylate the activation loop of p70 S6-K (36,39) and PKC family members (40,41) in vitro. In most cases, phosphorylation at the activation loop of these various kinases is necessary for activation, but phosphorylation of other residues is also required for full activity.
PKC⑀ expressed in baculovirus and PKC purified from brain are significantly activated by PtdIns-3,4-P 2 and PtdIns-3,4,5-P 3 (42,43). It is likely that the activity of these isoforms is affected by PDK1 phosphorylation (⑀ and ) as well as by a direct interaction with these phosphoinositides (⑀). Consistent with these in vitro studies, inhibition of PI 3-K blocks PDGF and insulin-dependent activation of p70 S6-K in vivo (44), insulin-dependent activation of PKC in vivo (45,46), and PDGFdependent membrane recruitment and activation of PKC⑀ in vivo (47). However, the enzymatic activity of PDK1 is not dependent on phosphoinositides (48). Thus, the requirement of PI 3-K for PDK1dependent phosphorylation of various enzymes probably reflects a role for PtdIns-3,4-P 2 and PtdIns-3,4,5-P 3 in co-localizing PDK1 and its substrates at specific membranes.
Cellular Functions-Activation of Akt mediates the transduction of cell survival signals. Activated Akt can phosphorylate and inactivate Bad, a protein involved in promoting cell death (reviewed by Ref. 50). In addition to Bad, Akt must have other targets that mediate cell survival because it promotes survival of cells that lack Bad (49). Glycogen synthase kinase 3 and phosphofructokinase are also in vitro substrates for Akt, implicating PI 3-K lipid products in gluconeogenesis and glycolysis (51,52).

PI-3,5-P 2
Regulation-In mouse fibroblasts, the major route for PtdIns-3,5-P 2 synthesis is wortmannin-sensitive and involves the consecutive phosphorylation of PtdIns by a PI 3-K (presumably the Class III enzyme) and of PtdIns-3-P by a PtdIns-3-P 5-kinase (53). In vitro, this second reaction can be catalyzed by the PtdIns-P 5-kinases ␣ and ␤ (also known as type I PtdIns-P kinases) (54). Likewise, PtdIns-3,5-P 2 synthesis in yeast involves phosphorylation of PtdIns-3-P by a 5-kinase and requires Vps34p PI 3-K (55). Fab1, a gene that is highly homologous to the mammalian PtdIns-P 5-kinases, was recently identified as the yeast PtdIns-3-P 5-kinase (56). Dramatic increases in PtdIns-3,5-P 2 levels were observed in response to hyperosmotic shock of yeast cells. In mammalian cells, the levels of PtdIns-3,5-P 2 decrease moderately with hyperosmotic shock and increase with hypo-osmotic shock. In vitro, PtdIns-3,5-P 2 can also be generated through phosphorylation of the novel lipid PtdIns-5-P by the Class I A PI 3-K (28).
Targets-PtdIns-3,5-P 2 is a newly identified molecule, and no direct target for this lipid has been found. PH domain-containing proteins are likely candidates for PtdIns-3,5-P 2 downstream effectors. Previous studies of lipid binding specificity of PH domains did not investigate this lipid.
Cellular Functions-Because mutations in the yeast Fab1 cause enlargement of the vacuole (57), PtdIns-3,5-P 2 may be involved in vesicle trafficking (reviewed by Emr and colleagues (94)).

PI-3,4,5-P 3
Regulation-The majority of the PtdIns-3,4,5-P 3 synthesized in response to extracellular signals is, most likely, generated by phosphorylation of PtdIns-4,5-P 2 at the 3-position of the inositol ring (25). The Class I PI 3-Ks are the only enzymes that can use PtdIns-4,5-P 2 as a substrate to synthesize PtdIns-3,4,5-P 3 (Fig. 1). Activation of class I A PI 3-Ks by growth factor stimulation of cells is mediated in part by interaction of their SH2 domain with tyrosine-phosphorylated proteins and results in a rapid elevation of PtdIns-3,4,5-P 3 levels (reviewed by Ref. 2). Class I A PI 3-Ks can also be regulated by the GTP-bound form of the small G protein Ras (59). The Class I B PI 3-K can be directly activated by the ␤␥ subunits of heterotrimeric G proteins (reviewed in Ref. 58). In addition, one of the Class I A enzymes (p110␤) can be activated synergistically by phosphotyrosine peptides plus ␤␥ subunits (60).
Targets-In addition to Akt, PDK1, and PKC⑀ (discussed above), many targets for PtdIns-3,4,5-P 3 have now been described. Several of these proteins have PH domains that mediate binding (for a review on PH domains see Ref. 65). Many PtdIns-3,4,5-P 3 -binding PH domains can also bind to PtdIns-4,5-P 2 , and only those that have at least a 10-fold higher affinity for PtdIns-3,4,5-P 3 than for PtdIns-4,5-P 2 will be considered here.
The PH domain of the Bruton's tyrosine kinase (Btk) was shown to interact with PtdIns-3,4,5-P 3 and its head group, inositol 1,3,4,5-P 4 , with high affinity (66 -68). Substituting cysteine for arginine 28 (R28C) in the PH domain of Btk, a natural mutation that causes X-linked immunodeficiency in mice, significantly affects the binding of PtdIns-3,4,5-P 3 and inositol 1,3,4,5-P 4 . In vivo, overexpression of the Class I A PI 3-K enzyme p110* (a constitutively active form of PI 3-K) or Class I B PI 3-K ␥ was shown to enhance Btk autophosphorylation and Src family kinase-mediated tyrosine phosphorylation of Btk (69,70). This effect was inhibitable by wortmannin and required the PH domain of Btk (69,70). Phosphorylation of Btk leads to its activation, as measured by its ability to regulate tyrosine phosphorylation of PLC␥2. Overexpression of Btk and PI 3-K enhanced production of inositol 1,4,5-P 3 in response to cross-linking of surface immunoglobulin in B cells. Engagement of SHIP to the inhibitory receptor Fc␥RIIB1 blocks B cell receptorinduced PtdIns-3,4,5-P 3 elevation and Btk activation. Altogether, these results indicate that PtdIns-3,4,5-P 3 , but not PtdIns-3,4-P 2 , is involved in activation of Btk by Src family kinases through a mechanism that resembles activation of Akt by PDK1.
The protein Grp1 (general receptor for phosphoinositides) was cloned based on its ability to bind PtdIns-3,4,5-P 3 in vitro (6). Analysis of its amino acid sequence revealed a PH domain and a Sec7 homology domain. The Grp1 PH domain is very selective for PtdIns-3,4,5-P 3 compared with other phosphoinositides (73). The Sec7 homology domain of Grp1 was shown to function as a guanine nucleotide exchange factor for the small G proteins Arf1 and Arf5. Grp1 exchange activity toward myristoylated Arf1 can be enhanced in vitro by PtdIns-3,4,5-P 3 -containing micelles, suggesting that PtdIns-3,4,5-P 3 regulates Grp1 by recruiting it to membranes where Arf is localized (73).
The presence of PH domains in a wide range of guanine nucleotide exchange factors for small G proteins (74) suggests that phosphoinositide regulation of these proteins may be widespread. PDGFinduced binding of GTP to the small G protein Rac depends on PI 3-K activation (75). This result suggests that PI 3-K lipids may directly affect the exchange factors for Rac. Indeed, PtdIns-3,4,5-P 3 and Pt-dIns-3,4-P 2 were shown to bind to the nucleotide exchange factor Vav and stimulate its exchange activity toward Rac, Cdc42, and RhoA (76). Interestingly, PtdIns-4,5-P 2 was also able to bind to Vav, but in this case, the exchange activity of Vav was inhibited by this lipid. Because water-soluble (short chain fatty acid) lipids were used in these experiments, it was suggested that PtdIns-3,4,5-P 3 binds to the PH domain of Vav and allosterically activates it.
The SH2 domains of Src and p85 (the PI 3-K Class I A regulatory subunit) can bind PtdIns-3,4,5-P 3 in competition with phosphotyrosine-containing proteins (72). More recently, PtdIns-3,4,5-P 3 was shown to bind to the SH2 domains of PLC␥ and to enhance its phospholipase activity toward PtdIns-4,5-P 2 in vitro (77,78). Inhibition of PI 3-K activity (by wortmannin treatment, mutation of PI 3-K-binding sites in the PDGF receptor, or overexpression of dominant-negative enzyme) partially inhibits PDGF-dependent inositol-1,4,5-P 3 production in intact cells, implicating PtdIns-3,4,5-P 3 as a positive regulator of PLC␥ in vivo. Another report showed that the PLC␥ PH domain also binds PtdIns-3,4,5-P 3 and mediates PLC␥ translocation to the cell membrane in response to growth factors (79).
Cellular Functions-With the identification of several PtdIns-3,4,5-P 3 targets, many of the cellular functions attributed to PI 3-Ks can now be understood at the molecular level. Elevation in cytosolic calcium in response to B cell stimulation appears to be modulated by PI 3-K, based on studies with PI 3-K inhibitors. This result can be explained by PtdIns-3,4,5-P 3 -dependent activation of Btk and perhaps also by direct effects of PtdIns-3,4,5-P 3 in recruitment of PLC␥ to the membrane (69,70,77,79).
A role for PI 3-K in vesicle recruitment to the plasma membrane has been proposed based on the observation that wortmannin and dominant-negative PI 3-K block GLUT 4 translocation to the plasma membrane in response to insulin (80,81). The observation that Pt-dIns-3,4,5-P 3 mediates recruitment of Grp1 to membranes and enhances Grp1 nucleotide exchange activity toward Arf1 provides an explanation for how PtdIns-3,4,5-P 3 may regulate coating and budding of intracellular vesicles (6, 73). As discussed above, a role for PtdIns-3-P and FYVE domain proteins in vesicle fusion is also likely. PI 3-K recruitment and activation is also necessary for PDGFinduced chemotaxis and membrane ruffling (82,83). PtdIns-3,4,5-P 3 activation of Vav2 (or other Rac exchange factors) (76) and consequently binding of Rac to GTP may explain the mechanism by which PI 3-K is involved in growth factor and Ras-stimulated cytoskeleton rearrangements that lead to cell migration.
Several studies support the idea that PI 3-K is necessary for growth factor and oncogene-induced cell proliferation. Recently, a natural oncogenic form of PI 3-K, v-p3k, was isolated from a chicken retrovirus that causes hemangiosarcomas, ASV16 (84). Expression of v-p3k protein as well as its cellular counterpart, the chicken p110␣ PI 3-K, causes elevation in PtdIns-3,4-P 2 and Ptd-Ins-3,4,5-P 3 levels, activation of Akt, and transformation of chicken embryo fibroblasts. Another oncogenic form of PI 3-K that consists of a truncated version of p85 (p65) associated with the p110 catalytic subunit has been isolated from transformed lymphoid cells (85). In cells expressing this constitutively active PI 3-K, Akt is also up-regulated. Strong evidence indicating that PtdIns-3,4,5-P 3 is involved in cell proliferation came with the recent finding that the tumor suppressor protein, PTEN, is a 3-phosphatase that dephosphorylates PtdIns-3,4,5-P 3 (62). The PTEN gene is deleted or mutated in a wide variety of human cancers, and it is capable of suppressing the growth of glioma cells (86 -88).
The gene encoding the mouse PI-3K adapter subunit, p85␣, has now been disrupted by two independent groups (89,90). Defects in B cell development and proliferation were observed in both studies. This phenotype resembles the phenotype of Btk-deficient mice. These data support the hypothesis that PtdIns-3,4,5-P 3 activation of Btk is likely to mediate B cell functions in animals (89,90).

PI 3-K as a Protein Kinase
PI 3-K is a dual specificity kinase that can phosphorylate serine and threonine residues in addition to phosphoinositide lipids (91,92). p110␣ can phosphorylate itself, the associated p85 regulatory subunit, and the insulin receptor substrate (IRS1) (reviewed in Ref. 58). Because phosphorylation of p85 by PI 3-K decreases the lipid kinase activity of the complex, it was proposed that the protein kinase intrinsic to PI 3-K has a regulatory function. The possibility that the protein kinase activity of PI 3-K plays a role in signaling is suggested by a recent study. Bondeva et al. (93) have now demonstrated MAPK activation by a PI 3-K␥ hybrid protein that has protein kinase activity but lacks lipid kinase activity. Moreover, membrane-bound PI 3-K␥ was unable to stimulate MAPK activation, indicating that the substrate for PI 3-K protein kinase is not at the cell membrane. On the other hand, this enzyme failed to stimulate Akt consistent with Akt activation being dependent on PtdIns-3,4-P 2 and/or PtdIns-3,4,5-P 3 synthesis (93). These results show that PI 3-K-mediated signaling involves independent pathways that lead to MAPK activation or Akt activation. Activation of the MAPK pathway may be an additional mechanism by which PI 3-K mediates the transduction of proliferation signals.

Concluding Remarks
Given that PI 3-K is involved in so many different cellular responses to a variety of different signals and that several different proteins are direct targets for 3-phosphorylated phosphoinositides (Fig. 2), an important question is: how is specificity in downstream signaling maintained?
One level of specificity may be obtained by synthesis of different lipid products by different PI 3-K isoforms. A second level of specificity can be obtained by recruitment of PI 3-K to specific subcellular compartments and a consequent increase in local production of lipids where a specific target may be available (for example, EEA1 recruitment by PtdIns-3-P and Rab5). Finally, it is possible that the specificity of the response is determined by convergence of two parallel pathways triggered by a specific signal (for example, Btk activation by PtdIns-3,4,5-P 3 and Lyn).
As other targets for PI 3-K lipids are unveiled, the job of untangling the intricate network of PI 3-K signaling will continue.