PI3K Signaling in Cancer: Beyond AKT

Abstract

The phosphoinositide 3-kinase (PI3K) signaling pathway is one of the most frequently altered pathways in human cancer and has a critical role in driving tumor initiation and progression. Although PI3K and its lipid product phosphatidylinositol-3,4,5-trisphosphate (PIP₃) have been shown to activate multiple downstream signaling proteins, the vast majority of studies have focused on the protein kinase AKT as the dominant effector of PI3K signaling. However, recent studies have demonstrated many contexts under which other PIP₃-dependent signaling proteins critically contribute to cancer progression, illustrating the importance of understanding AKT-independent signaling downstream of PI3K. Here, we highlight three PI3K-dependent, but AKT-independent, signaling branches that have recently been shown to have important roles in promoting phenotypes associated with malignancy. First, the PDK1 -mTORC2-SGK axis can substitute for AKT in survival, migration, and growth signaling and has emerged as a major mechanism of resistance to PI3K and AKT inhibitors. Second, Rac signaling mediates the reorganization of the actin cytoskeleton to regulate cancer cell migration, invasion, and metabolism. Finally, the TEC family kinase BTK has a critical role in B cell function and malignancy and represents a recent example of an effective therapeutic target in cancer. These mechanisms highlight how understanding PI3K-dependent, but AKT-independent, signaling mechanisms that drive cancer progression will be crucial for the development of novel and more effective approaches for targeting the PI3K pathway for therapeutic benefit in cancer.

Introduction

Phosphoinositide 3-kinase (PI3K) signaling plays a central role in cellular physiology, coordinating insulin signaling during organismal growth and mediating critical cellular processes such as glucose homeostasis, protein synthesis, cell proliferation, and survival. This pathway has been an intense area of investigation, particularly in light of cancer genetics studies that have revealed it to be one of the most frequently altered pathways in human malignancies that controls most hallmarks of cancer, including cell proliferation, survival, genomic instability, and metabolism [1]. Consequently, PI3K signaling has emerged as an attractive target for cancer therapy, and many drugs that inhibit various pathway components are currently in clinical trials [2, 3].

Class I PI3K transduces upstream signals from receptor tyrosine kinases (RTKs) and G protein-coupled receptors (GPCRs) by phosphorylating the 3′-hydroxyl group of the inositol ring of phosphatidylinositol-4,5-bisphosphate (PI-4,5-P₂) to generate phosphatidylinositol-3,4,5-trisphosphate (PIP₃) [4, 5]. PIP₃ serves as a critical lipid second messenger that recruits cytosolic proteins containing pleckstrin homology (PH) domains to the plasma membrane to promote either their activation or co-localization with other effector proteins [6–8]. It should be noted that only a small subset of PH domains in the human genome are thought to bind PIP₃ with high affinity and specificity (10–20% out of ~290 PH domains have been shown to robustly bind phosphoinositides, with some of these robustly binding PI-3,4-P₂ or PI-4,5-P₂ but not PIP₃) [9, 10]. Of the PH domain-containing proteins that do bind PIP₃, the serine/threonine AGC-family protein kinase AKT has received the greatest attention, especially for its multi-faceted roles in promoting glucose metabolism and cancer [11, 12]. However, recent advances have demonstrated critical mechanisms by which other proteins with PIP₃-binding PH domains contribute to cancer progression. Understanding the role of AKT-independent signaling downstream of PI3K is important because: a) AKT is not always hyperactivated in the context of mutations in PI3K pathway components such as PIK3CA and PTEN that elevate PIP₃ levels in cancer; b) many critical cellular processes are driven by PI3K-dependent but AKT-independent signaling to promote malignant phenotypes, and; c) mechanisms of resistance to PI3K pathway inhibitors can involve the activation of PI3K-dependent signaling proteins that can substitute for AKT signaling. To illustrate this, in this review we highlight three AKT-independent signaling branches downstream of PI3K that have recently been shown to have critical roles in promoting cancer progression: the PDK1-mTORC2-SGK axis, Rac signaling, and the TEC family kinases.

Substituting for AKT signaling: The PDK1-mTORC2-SGK axis

PDK1 (3-phosphoinositide-dependent protein kinase 1) and the multi-protein complex mTORC2 (mechanistic target of rapamycin complex 2) are PI3K-dependent, PH domain-containing kinases that coordinately activate several growth factor-sensitive AGC kinases, including AKT (also known as protein kinase B), SGKs (serum and glucocorticoid-regulated kinase), and certain PKCs (protein kinase C), by phosphorylating their activation loops and hydrophobic motifs (HM), respectively [13]. PDK1 is a constitutively active kinase with two major regulatory domains: a C-terminal PH domain that binds PIP₃, and a “PIF-binding pocket” within its catalytic domain that docks on the phosphorylated HM of AGC kinases, a region also known as the PDK1-interacting fragment (PIF) [14–17]. The PH domain allows PDK1 to co-localize with AKT at the plasma membrane and phosphorylate its activation loop upon PI3K activation. SGKs and PKCs, however, lack PH domains, and the PDK1 PH domain is not required for their phosphorylation; rather, PDK1 docks on the phosphorylated HM of these kinases through its PIF-binding pocket in order to phosphorylate their activation loops (atypical PKC isoforms contain acidic residues in their HMs that mimic constitutive phosphorylation) [18]. Two other growth factor-sensitive AGC kinases, S6K and RSK, are also activated by PDK1 in this manner but their HMs are not mTORC2 substrates and any dependence on PI3K is indirect. Therefore, at least for SGKs and some PKCs, activation loop phosphorylation by PDK1 is primed by phosphorylation of the HM by mTORC2, which is a multi-protein complex consisting of mTOR, Rictor, mSin1, mLST8, Protor1/2, and Deptor [19]. mTORC2 activity can be stimulated by PI3K activation and PIP₃ [20], though the precise underlying mechanisms has long remained elusive. However, the mSin1 subunit of mTORC2 has been reported to contain a putative PH domain [21], and a recent study proposes that autoinhibition of mTOR kinase activity by the mSin1 PH domain is disrupted when mSin1-PH binds PIP3, resulting in mTORC2 activation downstream of PI3K [22••].

The roles of PDK1 and mTORC2 in cancer are frequently linked to their activation of AKT, but recent evidence has demonstrated that other downstream AGC kinases also contribute to cancer malignancy. This is highlighted by the finding that reduced anchorage-independent growth and increased apoptosis caused by PDK1 silencing cannot necessarily be rescued by the expression of constitutively active AKT [23]. Recent studies have identified the SGKs, which consist of three isoforms SGK1, SGK2, and SGK3, as critical mediators of AKT-independent signaling downstream of PDK1 and mTORC2 in cancer (Figure 1). The SGK isoforms share high sequence similarity, and isoform-specific functions remain an open question. SGK knockout mice have revealed little about their physiological functions: SGK1-null mice have sodium balance intolerances, SGK3 knockout mice have defective hair follicle development, and SGK2 knockout mice have not been studied [24, 25]. Since SGKs and AKT share highly similar substrate specificities and have overlapping substrates [26–28], SGKs can substitute for AKT signaling in survival, migration, and growth signaling (Figure 1). For instance, a subset of PIK3CA mutant breast cancer cell lines that exhibit minimal AKT activation were found to rely on the PDK1-mTORC2-mediated activation of SGK3, and not AKT, for anchorage-independent growth [29]. Interestingly, although the SGKs lack PH domains, SGK3 possesses a phox homology (PX) domain that binds phosphatidylinositol-3-phosphate (PI-3-P), which is predominantly produced at the endosome by the Class III PI3K hVps34 [30]. In addition to activation at the endosome, SGK3 can also be activated at the plasma membrane by PI3K-dependent PI-3-P production through the sequential dephosphorylation of PIP₃ to PI-3,4-P₂ and PI-3-P by the lipid phosphatases SHIP1/2 (SH2 domain-containing inositol phosphatase) and INPP4B (Inositol polyphosphate 4-phosphatase type II), respectively (Figure 1) [31–34••]. In PIK3CA mutant breast cancer cells, SGK3 activation correlates with elevated expression of INPP4B, which positively regulates SGK3 activity to promote cell proliferation, anchorage-independent growth, cell migration, and tumor growth in vivo [35••].

Figure 1. The PDK1-mTORC2-SGK signaling axis.

SGK3 is activated at either the plasma membrane or the endosome, where its PX domain binds to PI-3-P. PI-3-P is produced both at the plasma membrane through the sequential dephosphorylation of PIP₃ by SHIP1/2 and INPP4B and at the endosome by the Class III PI3K hVps34. After PI-3-P binding, SGK3 is first phosphorylated on the hydrophobic motif (HM) by mTORC2. PDK1 then binds to the phosphorylated HM through its PIF-binding pocket and phosphorylates SGK3 on its activation loop. SGK1 is not regulated by PI-3-P because it lacks a PX domain, but it is phosphorylated by mTORC2 and PDK1 in the same manner as SGK3. Upon full activation, SGK1/3 phosphorylates its substrates with an optimal consensus motif of R-X-R-X-X-S/T to regulate cellular processes including cell growth, proliferation, migration, invasion, and resistance to PI3K and AKT inhibitors.

Activation of SGK signaling has also recently emerged as an important mechanism of resistance to PI3K and AKT inhibitors. For example, treatment of breast cancer cells with PI3K or AKT inhibitors results in increased expression and activation of SGK3, which under these conditions depends on hVps34 for activation by PDK1 and mTORC2. SGK3 subsequently substitutes for AKT by phosphorylating TSC2 and inhibiting the TSC complex to activate the kinase complex mTORC1 [36••, 37]. Similarly, elevated SGK1 expression and activation by PDK1/mTORC2 is also observed in breast cancer cell lines resistant to PI3K/AKT inhibitors. In this setting, SGK1 maintains signaling downstream of AKT by phosphorylating TSC2 to activate mTORC1 while also phosphorylating and inhibiting the transcription factor FOXO3 [38, 39••]. As mTORC1 and FOXO are major coordinators of anabolic metabolism and growth, their regulation by SGK can allow cells to overcome PI3K/AKT inhibition [40]. These results suggest that inhibitors of PDK1 or SGKs, listed in Table 1, may synergize with PI3K/AKT inhibitors to induce more robust antitumoral responses.

Table 1.

Selected list of inhibitors of AKT-independent PI3K signaling

Inhibitor	Target	Refs
PDK1/mTORC2-SGK signaling
GSK2334470	PDK1	[82, 83]
Torin 1	mTOR kinase	[84]
PP242	mTOR kinase	[85]
AZD8055	mTOR kinase	[86]
AZD2014	mTOR kinase	[87]
MLN0128 (formerly INK128)	mTOR kinase	[88]
VPS34-IN1	hVps34	[34]
SAR405	hVps34	[89•]
GSK650394	SGK1/2/3	[90]
14h	SGK1/3	[36, 91•]
SI113	SGK1	[92–95•]
EMD638683	SGK1/2/3	[96]
PI3K/Rac signaling
NSC23766	Rac	[97, 98]
EHop-16	Rac	[99, 100•]
EHT 1864	Rac	[101–103]
BTK and PI3Kδ inhibitors
Ibrutinib (PCI-32765)	BTK (1st gen)	[75, 76, 104, 105]
Acalabrutinib (ACP-196)	BTK (2nd gen)	[77, 78]
ONO/GS-4059	BTK (2nd gen)	[78, 106]
BGB-3111	BTK (2nd gen)	[78, 107]
CC-292	BTK (2nd gen)	[108, 109]
Idelalisib (GS-1101; CAL-101)	PI3Kδ	[79, 80]
AMG-319	PI3Kδ	[80, 110]
TGR-1202	PI3Kδ	[80, 111]
INCB050465	PI3Kδ	[80]

Taken together, these studies demonstrate that AKT-independent signaling initiated by PDK1 and mTORC2 downstream of PI3K contributes significantly to cancer progression. Understanding the specific contexts under which the PDK1-mTORC2-SGK axis provides an alternative mechanism for PI3K signaling to drive tumorigenesis will be critical for the development of more effective strategies for targeting the PI3K pathway for cancer treatment.

Actin cytoskeleton remodeling: Rac signaling

One of the most important PI3K-dependent cellular processes that primarily occurs independently of AKT is remodeling of the actin cytoskeleton. A major effector of this process is the Rac proteins (RAC1, 2, and 3), which are a subfamily of the Rho family of small GTPases (Figure 2) [41]. Like all G proteins, Rac alternates between its inactive GDP-bound and active GTP-bound states, which are promoted by GTPase-activating proteins (GAPs) and guanine-nucleotide exchange factors (GEFs), respectively. PI3K functions as an upstream activator of Rac by stimulating PIP₃-sensitive Rac-GEFs, which include PREX1, PREX2, Vav1, Sos1, and SWAP-70 (Figure 2). These Rac-GEFs each contain a PH domain that interacts with PIP₃, which subsequently relieves the inhibition of the GEF catalytic guanine-nucleotide exchange domain [42]. Upon activation, Rac-GTP then regulates cytoskeleton reorganization by binding and activating relevant target proteins, most notably p21-activated kinase (PAK) and WAVE [43–46].

Figure 2. PI3K-Rac signaling regulates actin cytoskeleton reorganization.

Upon binding to PIP₃ via their PH domains, PIP₃-sensitive Rac-GEFs catalyze the exchange of GDP for GTP on Rac. Rac activation is negatively regulated by Rac-GAPs, which facilitate GTP hydrolysis by Rac. When Rac-GTP is activated, it regulates cytoskeleton reorganization by binding and activating relevant target proteins, most notably PAK and WAVE. Remodeling of the actin cytoskeleton modulates cellular processes such as cell migration, invasion, and metabolism.

While many studies have demonstrated that AKT regulates cancer cell invasion and migration [47, 48], Rac signaling also promotes this phenotype independently of AKT by remodeling the actin cytoskeleton [49, 50]. Recent studies have focused on the Rac-GEFs PREX1 and PREX2, which have been implicated in multiple cancers. For example, PREX1 is up-regulated in breast cancer, metastatic prostate cancer, and melanoma, and it activates RAC1 to promote cell migration and invasion through the formation of lamellipodia. PREX1 overexpression is sufficient to increase the metastatic potential of non-metastatic prostate cancer cells in vivo, and Prex1^−/− mice are resistant to metastasis in an Nras/INK4a-driven mouse melanoma model [51–53]. Likewise, PREX2 is up-regulated in many cancers and is also frequently mutated, especially in melanoma [54–56]. These oncogenic PREX2 mutations increase its Rac-GEF activity, which contributes to NRAS-driven melanoma development and promotes breast cancer cell migration [57••, 58••]. Interestingly, PREX2 is not only activated by PIP₃, but also directly binds to PTEN and inhibits its lipid phosphatase activity to amplify PI3K signaling [59]. As a potential mechanism to prevent a feed-forward circuit in which PREX2-induced PI3K signaling results in unrestrained PIP₃-mediated PREX2 activation, PREX2 is reciprocally inhibited by PTEN independently of its lipid phosphatase activity and PIP₃ [58••]. Remarkably, oncogenic PREX2 mutants are able to escape PTEN-mediated inhibition, allowing them to both amplify PI3K signaling and activate Rac to promote tumor growth and cell invasion to drive cancer progression [57••, 58••].

Rac signaling has also recently been shown to contribute to PI3K-dependent regulation of cellular metabolism, which has typically been thought to be predominantly controlled by AKT [60–62]. A recent study reported that PI3K acutely stimulates glycolysis independently of AKT by regulating aldolase A in a Rac1-dependent manner. In quiescent cells, aldolase A is bound to the actin cytoskeleton, which inhibits its enzymatic activity. Upon growth factor-induced PI3K activation, activation of Rac and its effectors PAK and WAVE leads to actin cytoskeleton reorganization that releases aldolase A, and thus increases its activity [63••]. Interestingly, another study demonstrated that inhibition of PI3K, but not AKT, selectively inhibits the non-oxidative branch of the pentose phosphate pathway (non-oxPPP), resulting in decreased ribose 5-phosphate and nucleotide production [64•]. Since glyceraldehyde 3-phosphate, a product of aldolase A, serves as an entry point into the non-oxPPP, PI3K/Rac-mediated aldolase A activation may promote the PI3K-dependent, but AKT-independent, stimulation of the non-oxPPP to contribute to nucleotide synthesis. Given recent interest in therapeutically exploiting cancer-associated metabolic vulnerabilities, a deeper understanding of how PI3K-Rac signaling regulates glycolysis and other metabolic pathways may reveal novel strategies for targeting the PI3K pathway for cancer therapy.

These studies highlight the importance of Rac signaling in modulating cellular processes associated with actin cytoskeleton remodeling to promote cancer malignancy. Importantly, Rac also controls other processes independently of cytoskeleton reorganization such as proliferation and survival, which are not discussed in this review [41]. Therefore, developing inhibitors of Rac (Table 1) and exploring their potential as therapeutic options for cancer treatment is an active area of investigation [65].

Activation of non-receptor tyrosine kinase signaling: TEC family kinases

Of the 90 tyrosine kinases in the human genome, 58 are membrane-spanning receptor tyrosine kinases (RTKs), while the remaining 32 lack transmembrane domains and are classified as nonreceptor tyrosine kinases (NRTKs). NRTKs function within the nucleocytoplasm to trigger signaling cascades downstream of cell surface receptors and tend to play an especially important role in propagating signals downstream of receptors that do not contain intrinsic enzymatic domains [66]. For instance, in immune cells members of the SRC and TEC subfamilies of NRTKs mediate critical signaling events downstream of the T cell receptor (TCR), B cell receptor (BCR), Toll-like receptors (TLRs), and integrins [67]. Like RTKs, NRTKs can activate PI3K by phosphorylating its regulatory subunit or creating phospho-tyrosine docking sites on adaptors and receptors that recruit and activate PI3K. However, members of one NRTK subgroup, the TEC family, conversely rely on PI3K for their activation, distinguishing them from other NRTKs [68].

The five-member TEC kinase family includes: TEC (tyrosine kinase expressed in hepatocellular carcinoma), BMX (bone marrow tyrosine kinase gene on chromosome X; also known as ETK), BTK (bruton’s tyrosine kinase), ITK (interleukin-2-inducible T-cell kinase; also known as EMT or TSK), and TXK (also known as RLK). These share a common domain architecture including a C-terminal tyrosine kinase domain, SH2 and SH3 protein-binding domains, and, with the exception of TXK, an N-terminal PH domain that binds PIP₃ [68]. The binding of PI3K-produced PIP₃ results in the accumulation of TEC, BMX, BTK, and ITK at the plasma membrane where they are phosphorylated and activated by other NRTKs (as described for BTK in more detail below). Although PI3K may simultaneously activate AKT in the same cell, TEC kinase activation is AKT-independent. Hence, PI3K activity can dynamically promote membrane localization and activation of TEC kinases in contrast to, for instance, SRC family members which constitutively associate with membranes via their myristoyl and palmitoyl lipid modifications. Although the TEC kinases are expressed in multiple organ systems and preclinical studies have implicated them in multiple cancers including breast, colorectal, prostate, and glioblastoma, their role in the immune system has received the greatest attention [67, 68]. Here we highlight the role of the TEC family member BTK and its regulation by the delta isoform of PI3K in B lymphocyte (B cell) function and malignancy.

The B cell receptor (BCR) plays a pivotal role in both the development and activation of B cells. In naïve B cells, antigen binding to the BCR induces clustering of BCRs and associated CD79A/CD79B heterodimers that triggers tyrosine kinase signaling by NRTKs (Figure 3). BCR-associated LYN, a SRC family member that tethers to the membrane via dual lipid modifications, phosphorylates tyrosine motifs on the CD79A/CD79B heterodimer. The NRTK SYK then binds tyrosine-phosphorylated CD79A/CD79B, is activated, and subsequently phosphorylates a tyrosine motif on the co-receptor CD19 to which Class I PI3K binds via its SH2 domain-containing regulatory subunit. Production of PIP₃ from PI-4,5-P₂ by PI3K recruits BTK (via its PH domain) to the plasma membrane, where it can undergo activating phosphorylation by LYN and SYK. PI3K-dependent AKT activation also occurs in this context and regulates certain B-cell functions, but does not directly regulate BTK and is not discussed further here [69]. Through its SH2 domain, BTK participates in a signaling complex nucleated by a scaffolding protein BLNK (B cell linker; also known as SLP65) that is phosphorylated on multiple tyrosines by other NRTKs. PLCγ and the Rho family GEF Vav1 also dock on tyrosine-phosphorylated BLNK via their SH2 domains. Phosphorylation and activation of PLCγ by BTK triggers calcium mobilization and PKC signaling that ultimately turns on Ras signaling along with NF-κB and NFAT transcriptional programs. BTK-dependent activation of Vav1 turns on the Rho family small GTPases Cdc42 and Rac, which control actin remodeling. Together, these and other signals cooperate to govern the survival, differentiation, proliferation, and migration responses that underlie B cell development and activation [70].

Figure 3. PI3K and BTK Signaling in Response to BCR activation in B cells.

Antigen engagement of the BCR receptor induces clustering of the BCR, associated CD79A/CD79B heterodimers, and the non-receptor tyrosine kinase (NRTK) LYN which is tethered to the membrane by dual lipid modifications. LYN phosphorylates tyrosine motifs on CD79A/CD79B and the co-receptor CD19 to which the NRTK SYK and PI3K dock respectively and become activated. PIP₃ produced by PI3K recruits BTK to the membrane where it is phosphorylated and activated by SYK and LYN. BTK also forms a complex with the adaptor protein BLNK, the Rho family GEF Vav, and the phospholipase PLCγ. Following its BTK-dependent activation, Vav promotes GTP loading and activation of Cdc42 and Rac which then trigger actin cytoskeleton remodeling through a variety of effectors. Simultaneously, BTK phosphorylates and activates PLCγ which cleaves PI-4,5-P₂ to diacylglycerol (DAG) and inositol 1,4,5-trisphosphate (IP3). DAG is bound by PKCβ which triggers IKK signaling and NFkB-dependent transcription plus activation of Ras-Raf-MEK-ERK signaling. IP3 stimulates calcium (Ca²⁺) mobilization which results in NFAT-dependent transcription. These signaling events cooperate to control survival, proliferation, and adhesion signals that underlie B cell development and activation.

In humans, inactivating mutations in BTK are the molecular basis for the immunodeficiency disorder X-linked agammaglobulinemia which is characterized by severe defects in B cell development and function [71, 72]. BTK knockout mice or mice treated with small molecule inhibitors of BTK have similar B cell deficiencies, illustrating the critical role played by BTK in B cell biology [73–75]. Aberrant BCR signaling can promote B cell transformation, and due to the tissue-specific importance of BTK and PI3Kδ in B cells, there has been an intense effort to therapeutically target these for the treatment of B cell malignancies [70]. It is worth noting that the role of BTK is more specific to B cells compared with PI3Kδ, which is broadly important in leukocytes. A first generation irreversible BTK inhibitor Imbruvica (Ibrutinib) is FDA-approved for treatment of chronic lymphocytic leukemia (CLL), mantle cell lymphoma (MCL), and Waldenström macroglobulinemia (Table 1) [75, 76]. As Imbruvica inhibits multiple Tec kinases that are important in other immune cell types, more potent and selective second generation BTK inhibitors, such as Acalabrutinib, have been developed and are now showing promise in Phase I and II clinical trials (Table 1) [77••, 78]. A PI3Kδ inhibitor, Zydelig (Idelalisib), has also been FDA-approved for the treatment of relapsed CLL, follicular B cell non-Hodgkin lymphoma (FL) and small lymphocytic lymphoma (SLL), again underscoring the critical role played by PI3K in B cells [79, 80]. Interestingly, there is some evidence that these inhibitors may also prove useful in treating solid tumors. In a mouse model of pancreas ductal adenocarcinoma (PDAC), inhibitors of BTK and, in this case, PI3Kγ, were shown to alter lymphocyte-macrophage cross-talk that ultimately promoted immunosuppression of the tumor [81••]. Therefore, PI3K-dependent BTK signaling is a key example of an AKT-independent branch of the PI3K network that has emerged as an effective therapeutic target in cancer. The clinical results achieved with BTK and PI3Kδ inhibitors represent one of the recent success stories for rationally developed, precision medicine.

Summary and outlook

As one of the most frequently altered pathways in human cancer, PI3K signaling has clearly emerged as a highly attractive drug target for the treatment of human cancers. Given the predominant focus on studying AKT as the primary effector downstream of PI3K, much of the effort in targeting this pathway has centered around the development of drugs that effectively inhibit PI3K/AKT signaling. However, as this review has illustrated, it is becoming increasingly clear that other PI3K-dependent but AKT-independent signaling branches are just as important, if not more important, in promoting cancer initiation and progression. Indeed, a growing recognition of this idea has led to the development of drugs that target these other pathways downstream of PI3K (Table 1). Given that PI3K and AKT inhibitors currently in the clinic have shown limited efficacy with dose-limiting toxicities, these drugs may provide opportunities to rationalize new therapeutic combinations that may target the PI3K pathway more effectively. It is also important to note that the three pathways highlighted in this review constitute only a subset of the PH domain-containing proteins that are regulated by PIP₃ (or other phosphoinositides derived from PI3K-produced PIP₃). Therefore, we believe that there is still much to learn about how these various PI3K-dependent signaling proteins contribute to cancer malignancy, and that this knowledge will greatly strengthen our ability to develop novel and more effective strategies for targeting the PI3K pathway for the treatment of human cancers.