The mechanisms of class 1A PI3K and Wnt/β-catenin coupled signaling in breast cancer

The class IA PI3K signaling pathway is activated by growth factor stimulation and regulates a signaling cascade that promotes diverse events including cell growth, proliferation, migration and metabolism. PI3K signaling is one of the most commonly hyperactivated pathways in breast cancer, leading to increased tumor growth and progression. PI3K hyperactivation occurs via a number of genetic and epigenetic mechanisms including mutation or amplification of PIK3CA, the gene encoding the p110α subunit of PI3Kα, as well as via dysregulation of the upstream growth factor receptors or downstream signaling effectors. Over the past decade, extensive efforts to develop therapeutics that suppress oncogenic PI3K signaling have been undertaken. Although FDA-approved PI3K inhibitors are now emerging, their clinical success remains limited due to adverse effects and negative feedback mechanisms which contribute to their reduced efficacy. There is an emerging body of evidence demonstrating crosstalk between the PI3K and Wnt/β-catenin pathways in breast cancer. However, PI3K exhibits opposing effects on Wnt/β-catenin signaling in distinct tumor subsets, whereby PI3K promotes Wnt/β-catenin activation in ER+ cancers, but paradoxically suppresses this pathway in ER− breast cancers. This review discusses the molecular mechanisms for PI3K–Wnt crosstalk in breast cancer, and how Wnt-targeted therapies have the potential to contribute to treatment regimens for breast cancers with PI3K dysregulation.


PI3K hyperactivation drives breast cancer initiation and progression
Hyperactivation of class IA PI3K signaling is one of the most common molecular events in human cancers and occurs in more than 70% of human breast cancers [28].PIK3CA, the gene encoding the p110α catalytic subunit of PI3Kα, is mutated in 20-40% of breast cancers, and most frequently in estrogen receptor-positive (ER + ) breast cancers [29,30].PIK3CA gene amplification also occurs less commonly in ∼9% of breast cancers [31].The most common PIK3CA mutations are gain-of-function missense mutations that occur at two hotspot regions; H1047R in the kinase and E542K/E545K in the helical domain.PIK3CA H1047R promotes constitutive p110α plasma membrane association [32] whereas PIK3CA E542K or PIK3CA E545K mutations preclude p110α binding to the inhibitory p85 subunit [3].Pik3ca H1047R conditional mammary knock-in promotes de novo tumor formation with long latency and incomplete penetrance [33,34], suggesting PIK3CA mutations are weakly oncogenic in isolation.A subset of PIK3CA-mutant cancers harbor multiple copies of mutant PIK3CA in cis (same allele) or trans (separate alleles) that additively affect downstream signaling and cellular phenotypes in a dose-dependent manner [35][36][37] or elicit biphasic effects [35][36][37][38].In particular, double PIK3CA mutations in cis occur in 8-13% of breast cancers, which enhance cell proliferation and tumor growth compared with single hotspot mutations [35].Single or multiple PIK3CA mutations do not independently correlate with human breast cancer prognosis when corrected for ER-positivity or other favorable-risk variables [35,39].
Class IA PI3K signaling hyperactivation in breast cancer also occurs via other mechanisms including the dysregulation of upstream RTKs that activate PI3Ks, or the lipid phosphatases that regulate downstream phosphoinositide signaling.EGFR is rarely amplified or mutated in breast cancer, but increased gene copy number due to polysomy can occur in triple negative breast cancers, leading to activation of multiple signaling pathways including PI3K [40].The RTK Met is frequently up-regulated in triple negative breast cancers, and mice with mammary-specific transgenic expression of mutationally activated Met develop tumors with moderate penetrance and long latency [41].Germline PTEN mutations cause Cowden disease, which predisposes individuals to tumor development in multiple tissues including breast, thyroid, renal, endometrial and brain [42,43].PTEN is also sporadically deleted or mutated or exhibits promoter hypermethylation in all breast cancer subtypes [44,45].Pten +/− mice develop a range of tumors in multiple organs including mammary, prostate, intestine, endometrial and lymphatic tissues [46][47][48][49].Interestingly, PTEN-deficient prostate, colorectal and triple negative breast cancers exhibit dependence on PI3Kβ rather than PI3Kα signaling [50][51][52][53].Pten G129E (lipid phosphatasedead) or Pten C124S (lipid and protein phosphatase-dead) mutant knock-in mice are also prone to mammary, adrenal and thyroid tumors [54].Proline-rich inositol polyphosphate 5-phosphatase (PIPP), which hydrolyzes PI(3,4,5)P 3 to PI(3,4)P 2 , exhibits decreased expression in triple negative breast cancers, associated with reduced relapse-free and overall survival [55].Pipp ablation accelerates mammary tumor initiation and growth but reduces metastasis in PyMT breast cancer model mice [55].Loss of heterozygosity (LOH) of the chromosomal region encoding the INPP4B gene (4q31.21)occurs in ∼55% of triple negative breast cancers [56,57].Murine Inpp4b knockout enhances mammary tumor incidence in mammary-specific Tp53 −/− ;Brca1 −/− mice and enhances thyroid tumor formation and metastasis in Pten +/− mice [21,23,58].However, INPP4B expression is increased in a subset of ER + breast cancers where it paradoxically promotes cell proliferation and tumor growth [25].Interestingly, PIK3CA mutations frequently co-occur with PTEN mutations or INPP4B overexpression, which accelerate tumor development [25,59].
There has been significant interest in developing PI3K-targeted therapies for breast cancer due to the high prevalence of PI3K pathway hyperactivation.However, the clinical efficacy of PI3K inhibitors for breast cancer remains limited, as these therapies can elicit significant adverse effects.In addition, feedback signaling mechanisms can reduce drug efficacy over a sustained treatment period.Early inhibitors targeted multiple PI3K isoforms, but few of these successfully progressed through clinical trials.Further clinical progress has been made via specifically targeting the PI3Kα isoform.The PI3Kα inhibitor, alpelisib, was initially found to up-regulate ER-dependent transcriptional activity in PIK3CA-mutant ER + breast cancer xenografts, which was circumvented by co-treatment with the selective estrogen receptor degrader (SERD), fulvestrant [60].Alpelisib was recently approved for advanced PIK3CA-mutant ER + breast cancer in combination with fulvestrant, which prolongs relapse-free survival but does not prevent disease recurrence [61].Studies have identified a number of feedback mechanisms that limit alpelisib efficacy, including PI3K reactivation by p110β in HER2-amplified and PIK3CA-mutant breast cancers [62].In addition, increased pancreatic insulin secretion drives alpelisib resistance in murine mammary tumor models which is overcome by dietary and pharmacological approaches that reduce insulin signaling and restore alpelisib sensitivity [63].Acquired alterations in other PI3K signaling components can circumvent alpelisib efficacy, such as allelic PTEN loss [64].A current focus is to identify compounds that show higher selectivity for mutant PI3Kα versus the wild-type protein, such as inavolisib which promotes HER2-dependent degradation of mutant PI3Kα [65].
The PI3K and Wnt/β-catenin pathways share several core signaling components and it has long been predicted that crosstalk exists between these pathways (reviewed in [76]).However, as both PI3K and Wnt/ β-catenin networks exhibit a high degree of complexity with numerous feedback loops and context-dependent signaling dynamics [77][78][79], dissecting the mechanisms of PI3K-Wnt crosstalk has been challenging.Multiple independent studies show that activation of PI3K signaling by insulin stimulation alone is insufficient to concomitantly activate Wnt/β-catenin signaling [80,81].This is further complicated by the multifaceted role of GSK3β, a core component of the β-catenin destruction complex, which is also independently inactivated via phosphorylation of its Ser9 residue by AKT [82,83].This has led to speculation that AKT-mediated inactivation of GSK3β promotes Wnt/β-catenin signaling.However, the GSK3β pool bound to the destruction complex is protected from AKT phosphorylation, as its molecular arrangement within the complex shields its Ser9 residue from phosphorylation by AKT [80].Gsk3b S9A knock-in mice, which harbor a mutation in the AKT-dependent Ser9 phosphorylation site of GSK3β, show no difference in Wnt3a-dependent activation of β-catenin or Wnt gene transcription [84], revealing that AKT phosphorylation of GSK3β does not significantly contribute to Wnt/β-catenin signaling.
Interestingly, PI3K/AKT signaling also suppresses Wnt/β-catenin signaling via the activation of mTORC1 (Figure 2).AKT phosphorylates and inactivates TSC2, which alleviates the inhibitory effect of TSC2 on mTORC1 leading to its activation [90].Murine Tsc2 deletion, which enhances mTORC1 activity, decreased Wnt/β-catenin signaling by reducing cell surface FZD levels and LRP6 phosphorylation in intestinal stem cells, leading to reduced stemness [91].FZD mRNA levels were unchanged following treatment with the mTORC1 inhibitor, RAD001, suggesting that mTORC1 signaling regulates FZD protein degradation, rather than its gene expression [91].Thus, PI3K has the potential to activate and suppress Wnt/β-catenin via its regulation of distinct downstream effectors.This dichotomy is also observed in different breast cancer subsets, where PI3K promotes Wnt/β-catenin activation in ER + breast cancers, but suppresses this pathway in ER − breast cancers, as discussed in more detail below.

PI3K signaling enhances Wnt/β-catenin activation in ER + breast cancer
Wnt/β-catenin signaling is hyperactivated in greater than 50% of human breast cancers [92].Mutations in Wnt pathway components (eg APC, AXIN1/2, β-catenin) occur frequently in other cancers, but are rarely detected in breast cancer suggesting other mechanisms likely contribute to Wnt activation in this context (reviewed in [66,93]).Increased nuclear β-catenin accumulation is observed more frequently in primary human triple negative breast cancers [94,95], whereas ER + breast cancers exhibit an RNA profile consistent with enhanced Wnt/ β-catenin signaling [96].
AKT is a central effector of class I PI3K signaling, however, PIK3CA-mutant ER + breast cancers show little AKT activation compared with alterations in other PI3K pathway components, and frequently exhibit AKT-independent tumor growth [97,98].Several independent groups have undertaken RNA profile analysis of primary human breast cancers and mouse models of PIK3CA mutant ER + breast cancer and uncovered AKT-independent downstream signaling events that contribute to tumor progression [25,96,99].PIK3CA mutant ER + breast cancers exhibit an RNA expression profile of enhanced Wnt/β-catenin signaling, including up-regulation of Wnt target genes (AXIN2, LEF1, MYCN) and other components such as transcriptional regulators (TCF7L1, TCF7L2, CTNNB1), receptors (FZD4, FZD7) and ligands (WNT5A).Mammary glands of mice with transgenic expression of human PIK3CA H1047R or PIK3CA E545K exhibit increased activated-β-catenin levels compared with wild-type mice [99].Activated-β-catenin and AXIN2 mRNA levels are also increased in ER + mouse mammary tumor cells expressing human PIK3CA H1047R compared with transgenic MMTV-Her2/neu mouse mammary tumor cells [99].However, it is unknown whether multiple PIK3CA mutations affect Wnt/ β-catenin signaling in a dose-dependent or biphasic manner.Treatment with buparlisib (PI3Kα/β/δ/γ inhibitor) reduced the expression of Wnt target genes AXIN2, LEF1, and MYCN in human PIK3CA-mutant ER + breast cancer cells [25].Thus, PI3K signaling promotes activation of Wnt/β-catenin signaling in ER + breast cancer (summarized in Table 1), although whether other class IA PI3K isoforms contribute to Wnt activation in this context remains unclear.The Tankyrase1/2 inhibitor XAV939, which stabilizes the β-catenin destruction complex and suppresses Wnt/β-catenin signaling, sensitized PIK3CA H1047R expressing mammary tumor cells derived from a doxycycline inducible-PIK3CA H1047R mouse to LY294002, a non-selective PI3Kα/δ/β inhibitor that also inhibits other PI3K-related kinases as well as unrelated proteins, although little effect was observed in mammary tumor cells derived from PIK3CA H1047R transgenic mice or PIK3CA H1047R ER − breast cancer cells for unknown reasons [99].
β-catenin destruction complex within endosomes [72,100,101].INPP4B localizes to early endosomes in a variety of cell types including mouse fibroblasts and cancer cells [22,23].In PIK3CA-mutant ER + breast cancer cells, INPP4B localizes prominently to late endosomes via its interaction with the small GTPase Rab7 [25].This interaction enhances PI(3,4)P 2 conversion to PI(3)P and thereby promotes HRS-dependent endosome maturation and cargo trafficking [25].As a consequence, INPP4B overexpression increases the endosomal sequestration and lysosomal degradation of GSK3β, leading to increased activated-β-catenin and Wnt target gene transcription [25] (Figure 2).Intriguingly, INPP4B increased SGK3 and Wnt/β-catenin signaling despite suppressing AKT activation.However, whether INPP4B expression affects AKT-dependent β-catenin S552 phosphorylation has not been reported.The porcupine O-acyltransferase (PORCN) inhibitors LGK-974 or IWP-2, which prevent Wnt ligand secretion, rescued the increased proliferation of INPP4B-overexpressing ER + breast cancer cells [25].INPP4B-overexpressing PIK3CA-mutant ER + breast cancer cells were selectively sensitive to nanomolar concentrations of pyrvinium [102], an FDA-approved anthelmintic drug that binds and activates CK1α to suppress Wnt/β-catenin signaling [103], and were significantly more sensitive to pyrvinium in combination with the standard-of-care treatment 4-hydroxytamoxifen (4-OHT) [102].Further investigation of pyrvinium and other FDA-approved compounds that suppress Wnt/β-catenin signaling in human breast cancers is required to determine whether these therapeutics improve ER + breast cancer outcomes.Furthermore, testing of potential combination therapies in non-cancerous versus preclinical breast cancer models would allow evaluation of on-target toxicity and efficacy of simultaneous versus sequential treatment regimens.
The mechanisms by which PI3K-Wnt crosstalk occurs in ER − breast cancer are poorly understood.As mTOR suppresses FZD levels on the plasma membrane of intestinal stem cells [91], it is intriguing to consider whether PI3K/mTOR inhibition may increase FZD surface expression in ER − breast cancer.In addition, buparlisib (PI3Kα/β/δ/γ inhibitor) treatment increased mRNA/protein expression of PORCN, an O-acetyltransferase required for Wnt ligand palmitoleoylation and secretion [106].Therefore, class I PI3K down-regulation of Wnt ligand processing may also suppress Wnt/β-catenin signaling.It is also interesting to speculate that PI(4,5)P 2 or PI(3,4,5)P 3 , the substrate and product of class I PI3K function respectively, may also contribute to Wnt suppression.For example, the PI(3,4,5)P 3 effector BTK suppresses Wnt/β-catenin activation in human colorectal cancer and B cells by up-regulating the Wnt repressor, CDC73 [109].In contrast, PI(4,5)P 2 is required for Wnt/β-catenin activation by facilitating LRP phosphorylation and Dishevelled recruitment [110,111].Further investigation is required to determine the molecular mechanisms of PI3K-Wnt crosstalk in this context.
The induction of Wnt/β-catenin signaling in ER − breast cancers with PI3K/mTOR inhibition suggests that Wnt activation may reduce the sensitivity of these breast cancers to PI3K inhibitors.Pictilisib (strong PI3Kα/δ, modest PI3Kβ/γ inhibitor) treatment had little effect on the proliferation of ER − breast cancer cells, whereas combined pictilisib and LGK-974 (PORCN inhibitor) treatment synergistically reduced ER − breast cancer cell proliferation [107].LGK-974 buparlisib (PI3Kα/β/δ/γ inhibitor) treatment also synergistically reduced ER − breast cancer cell viability and xenograft tumor growth [106].Similar effects have been observed in other cancers such as pancreatic cancer, where combined ETC-159 (PORCN inhibitor) and pictilisib treatment synergistically reduced pancreatic cancer cell proliferation and xenograft tumor growth [112].Recently, phase I clinical trials were conducted in individuals with metastatic ER − breast cancer using the dual PI3Kα/γ and mTOR inhibitor, gedatolisib, in combination with cofetuzumab pelidotin, an antibody-drug conjugate with an auristatin payload targeting the Wnt co-receptor, protein tyrosine kinase 7 (PTK7) [113].This PI3K/Wnt combination therapy had manageable toxicity, and some individuals exhibited a partial clinical response with disease stabilization [113].However, further trials are required to assess whether PI3K/Wnt combination therapies can improve ER − breast cancer outcomes.

Conclusion and future directions
Class I PI3K signaling is one of the most frequently hyperactivated pathways in human breast cancer, and there have been significant advances in developing therapeutics that target this pathway in breast cancer.However, the clinical success of PI3K-targeted therapies remains limited, in part due to the complex interplay between PI3K and other signaling networks including estrogen or insulin.More recent findings show that PI3K signaling also undergoes crosstalk with the Wnt/β-catenin pathway with opposing effects observed in different breast cancer subsets.In PIK3CA-mutant ER + breast cancer, PI3Kα signaling promotes INPP4B-dependent lysosomal degradation of GSK3β, which enhances Wnt/β-catenin activation.Conversely, in ER − breast cancer, inhibition of class I PI3K and/or mTOR leads to Wnt/β-catenin activation.However, the molecular mechanisms and the contribution of individual PI3K isoforms are unknown.This may be influenced by the genetic or epigenetic profile of these distinct breast cancer subsets, such as the presence of PIK3CA mutations and increased INPP4B, or the expression of effectors that mediate PI3K suppression of Wnt signaling.
Although animal models have improved our understanding of cancer biology, there are limitations in extrapolating results from these model systems to humans.Human tumor organoids and patient-derived tumor xenografts are powerful, preclinical models for mechanistic and therapeutic testing with potential to circumvent the limitations of mouse models.Examination of PI3Kα, β, δ, or γ inactivation in normal and ER + versus ER − human breast cancer 3D organoids would help elucidate whether specific PI3K isoforms elicit opposing effects on Wnt signaling in preclinical models.Integrating Wnt/β-catenin reporter assays with PI3K signaling effector activity could provide further clarification of the mechanistic differences in PI3K and Wnt/β-catenin crosstalk in different breast cancer subtypes.However, comparing Wnt/β-catenin activation across different breast cancer subtypes can be challenging.Nuclear β-catenin localization is not evident across all subtypes [94,95] and β-catenin transcriptional reporter assays require co-expression of multiple probes that can be challenging in preclinical models.This may be improved by the development of quantitative live cell imaging biosensors to examine effector activation across subcellular compartments, such as those described for β-catenin or AKT [114,115].Furthermore, studies of human model systems would benefit from comparison of PI3K-Wnt crosstalk in stem versus non-stem breast cancer cell states which exhibit distinct differences in PI3K and Wnt signaling mechanisms [78,116] (reviewed in [117,118]).
Wnt-targeted therapies may be a credible therapeutic strategy for treating ER + or ER − breast cancers.Although these therapeutics are yet to be evaluated in human breast cancer trials, data obtained from cell culture and mouse models suggests ER + breast cancers with PIK3CA-mutations and/or increased INPP4B expression may benefit from Wnt therapies in combination with current standard-of-care treatments, whereas ER − breast cancers may respond to combination therapy of Wnt and PI3K/mTOR inhibition to prevent Wnt reactivation.There has been substantial interest in developing Wnt therapeutics including small molecule inhibitors, peptides and antibodies that suppress Wnt/β-catenin signaling [119].However, most Wnt-targeted inhibitors elicit significant adverse effects and/or do not effectively suppress Wnt/β-catenin signaling in human tumors, and there are currently no approved Wnt inhibitors for clinical use.Phase I clinical trials with the antibody-drug conjugate cofetuzumab pelidotin suggest this treatment exhibits partial efficacy with limited toxicity in ER − breast cancers in combination with gedatolisib [113].The repurposing of existing FDA-approved NSAIDs and anti-parasitic drugs that suppress Wnt/β-catenin signaling is emerging as a more promising clinical strategy [120], such as pyrvinium which was selectively cytotoxic to INPP4B-overexpressing ER+ breast cancer cells in combination with 4-OHT [102].Further investigation of Wnt-targeted therapies in preclinical and clinical models, including careful assessment of the efficacy and on-target toxicity of these therapeutics in cancerous versus non-cancerous models will help to whether targeting the Wnt pathway is a viable therapeutic strategy for human breast cancers.

Perspectives
• Class I PI3K signaling is one of the most frequently hyperactivated pathways in breast cancer, and its crosstalk with other signaling pathways contributes to the limited efficacy of PI3K-targeted therapeutics.
• Recent findings show that PI3K signaling enhances Wnt/β-catenin signaling in ER + breast cancer by promoting GSK3β lysosomal degradation, but paradoxically suppresses Wnt/ β-catenin activation in ER − breast cancer by unknown mechanisms.
• Understanding the complex interplay between PI3K and Wnt signaling in breast cancer will elucidate novel mechanisms of crosstalk and determine whether Wnt-targeted therapeutics will play a future role in breast cancer treatment.
activated-β-catenin to accumulate.β-catenin translocates into the nucleus, where it binds to TCF/LEF transcription factors to promote Wnt target gene expression.PI3K is proposed to undergo crosstalk with Wnt/β-catenin signaling via at least three mechanisms: (1) AKT phosphorylates the serine 552 residue of β-catenin at cellular junctions, leading to its accumulation in the cytosol and translocation to the nucleus.(2) AKT phosphorylates and inactivates TSC2, which alleviates the inhibitory effect of TSC2 on mTORC1 leading to mTORC1 activation and down-regulation of FZD levels.(3) INPP4B-generated PI(3)P downstream of PI3Kα facilitates HRS-dependent endosome maturation, which promotes GSK3β endosomal sequestration and lysosomal degradation leading to β-catenin activation.Created with BioRender.com.

Table 1
Summary of PI3K and Wnt/β-catenin in breast cancer