Research ArticleCell Biology

Phosphoinositide 3-Kinase p110β Activity: Key Role in Metabolism and Mammary Gland Cancer but Not Development

See allHide authors and affiliations

Science Signaling  09 Sep 2008:
Vol. 1, Issue 36, pp. ra3
DOI: 10.1126/scisignal.1161577

Abstract

The phosphoinositide 3-kinase (PI3K) pathway crucially controls metabolism and cell growth. Although different PI3K catalytic subunits are known to play distinct roles, the specific in vivo function of p110β (the product of the PIK3CB gene) is not clear. Here, we show that mouse mutants expressing a catalytically inactive PIK3CBK805R mutant survived to adulthood but showed growth retardation and developed mild insulin resistance with age. Pharmacological and genetic analyses of p110β function revealed that p110β catalytic activity is required for PI3K signaling downstream of heterotrimeric guanine nucleotide–binding protein (G protein)–coupled receptors as well as to sustain long-term insulin signaling. In addition, PIK3CBK805R mice were protected in a model of ERBB2-driven tumor development. These findings indicate an unexpected role for p110β catalytic activity in diabetes and cancer, opening potential avenues for therapeutic intervention.

Introduction

Phosphoinositide 3-kinases and their lipid product phosphatidylinositol-3,4,5-trisphosphate [PtdIns(3,4,5)P3] are involved in signaling events influencing a large number of cellular processes (13). Class IA PI3Ks are mainly activated by receptor tyrosine kinases (RTKs) and form heterodimers composed of a catalytic subunit (p110α, β, or δ, which are encoded by the PIK3CA, PIK3CB, and PIK3CD genes, respectively) and an Src homology 2 domain–containing adaptor protein (p85α, p50α, p55α, p85β, or p55γ) (3). Class IA PI3Ks are activated by either p85 recruitment to phosphorylated RTKs and adaptors such as insulin receptor substrates 1 and 2 (IRS-1 and IRS-2) or by binding to Ras (2). Whereas p110α is known to play a major role in insulin signaling (4, 5), and p110δ in lymphocyte activation (6), the role of p110β has remained elusive. Pharmacological studies have suggested a role for p110β in platelet aggregation (7); however, the early embryonic lethal phenotype caused by its genetic ablation (8) has prevented an accurate characterization of its in vivo function.

Aberrant regulation of the PI3K signaling pathway is frequently associated with cancer (9). Mutations in the class IA p110 gene PIK3CA can be detected in a number of human tumors (10, 11). Although mutations in p110β corresponding to oncogenic p110α mutations fail to elicit the same oncogenic behavior (12), overexpression of wild-type p110β induces transformation in cultured cells (13), and various human cancers show increased abundance of p110β (14). Nonetheless, whether targeting p110β could be effective in cancer therapy is unknown.

Here, we show that mice expressing a catalytically inactive form of p110β survive to adulthood and develop a mild insulin resistance with age. Mutant mice were protected in an in vivo model of ERBB2-driven cancer development in the mammary gland, identifying p110β as a promising drug target with tolerable side effects.

Results

Generation of mice expressing a kinase-dead p110β mutant

A mouse strain was engineered to carry a mutation in the PIK3CB gene (fig. S1) in which Lys805 of the p110β adenosine 5′-triphosphate (ATP)–binding site was replaced with arginine (PIK3CBK805R allele), leading to the production of a catalytically inactive form of p110β (p110βK805R). Unexpectedly, homozygous PIK3CBK805R/K805R mice were viable and reached adulthood. However, 50% fewer PIK3CBK805R/K805R mice were born from heterozygous crosses than expected (50 mutant homozygotes out of 372 mice analyzed; P < 0.0001 by χ2), indicating embryonic lethality with incomplete penetrance. This phenotype could not be associated to expression of aberrant p110β variants still retaining partial catalytic activity (fig. S2, A to E).

Unexpected noncatalytic function of p110β

At embryonic day 13.5, two distinct groups of PIK3CBK805R/K805R littermate embryos were found: about 70% appeared normal, whereas about 30% were abnormally small and moribund. The existence of these two distinct groups appeared to be related to genetic background because increased contribution of the C57/BL6J background resulted in a marked decrease in the percentage of normal homozygous embryos. In murine embryonic fibroblasts (MEFs) derived from these two mutant populations, the protein and messenger RNA (mRNA) abundance of p110βK805R was different (Fig. 1, A and B, and fig. S2G); it reached 60% to 80% of the control amounts in normal embryos (PIK3CBK805R/K805R High) but only 5% to 20% of the wild-type p110β amounts in abnormal embryos (PIK3CBK805R/K805R Low). In MEFs from PIK3CBK805R/K805R High, the enzymatic activity of p110βK805R did not increase above background (Fig. 1, C and D); however, p110α and p85 expression was not altered from that in wild-type MEFs (Fig. 1, A and B). Similarly, total class IA PI3K activity—precipitated with beads linked to a phosphopeptide that mimicked an activated growth factor receptor (Fig. 1D and fig. S2F)—and p110α activity—immunoprecipitated with antibodies directed against p110α (fig. S2F)—were normal, showing that the in vitro activity of other class IA PI3Ks was unaltered. Furthermore, decreased p110β abundance did not cause any substantial increase in free p85 (fig. S3).

Fig. 1

Kinase-dependent and kinase-independent functions of p110β in wild-type (PIK3CBWT/WT) and PIK3CBK805R/K805R MEFs. (A) PIK3CBK805R gene dosage inversely correlates with phenotype severity. MEFs were derived from apparently normal (PIK3CBK805R/K805R High) or abnormal (PIK3CBK805R/K805R Low) PIK3CBK805R/K805R embryos 13.5 days after conception. Homogenates of MEFs of the described phenotypes were analyzed by SDS-PAGE and immunoblotted (WB) with the indicated antibodies. (B) Analysis of p85 association to p110 in MEFs. Protein extracts were immunoprecipitated (IP) with anti–pan p85 antibodies and immunoblotted (WB) with the indicated antibodies. (C) Analysis of p110β catalytic activity. Representative lipid kinase assay with p110β immunoprecipitated from wild-type or PIK3CBK805R/K805R High MEFs. Background activity (Bkg) is the lipid kinase activity measured on protein extracts before immunoprecipitation. (D) PI3K activity in PIK3CBK805R/K805R High (KR) MEFs relative to that in wild-type controls (WT), as measured after pull down with anti-p110β antibodies or with phosphopeptide-bound beads associating with all class IA PI3Ks. ***P < 0.001. (E) Proliferation curve of mutant MEFs with high and low p110βK805R expression levels compared to that of wild-type MEFs with or without 100 nM TGX-221 treatment (TGX). Statistical significance: PIK3CBK805R/K805R Low cells versus all other conditions (*P < 0.05, **P < 0.01); other pairs of data sets are not significant.

Lack of p110β leads to early embryonic lethality associated with defective cell proliferation (8) and, in agreement, impaired cell proliferation was found in PIK3CBK805R/K805R Low MEFs. In contrast, PIK3CBK805R/K805R High cells unexpectedly proliferated at a rate similar to that of wild-type MEFs (Fig. 1E). Consistent with this, cell proliferation of wild-type MEFs did not change after treatment with the p110β-selective inhibitors TGX-221 (7) (Fig. 1E) or TGX-155 (15) (fig. S4), suggesting a noncatalytic function of p110β. p110β is known to associate with Rab5, a monomeric small guanosine triphosphatase involved in the fusion of clathrin-coated vesicles and in growth factor receptor endocytosis (16, 17). We thus investigated the possible role of p110β in these processes. We found that although the amount of plasma membrane epidermal growth factor (EGF) receptor (EGFR) was similar in mutant and control cells, internalization of EGF-activated EGFR in PIK3CBK805R/K805R Low cells was impaired compared with that in wild-type and PIK3CBK805R/K805R High cells (Fig. 2A).

Fig. 2

Analysis of EGF receptor endocytosis. (A) Immunofluorescence of cells of the given genotype incubated with EGF for 1 hour at 4°C (0 min) and then shifted to 37°C for 15 min (15 min). Merged images show EGFR in white and DAPI in blue. Exposure times were identical for the different genotypes but longer at 0 min to better show cell surface EGFR. (B) Immunofluorescence of cells of the given genotype with anti-clathrin (top) or anti-EEA1 antibodies (bottom). Merged images show clathrin or EEA1 in red and DAPI in blue.

Overexpression of a dominant-active Rab5 mutant (Rab5Q79L) did not rescue EGFR internalization in the PIK3CBK805R/K805R Low MEFs (fig. S5), suggesting a role of p110β at the early steps of endocytosis. Consistently, PIK3CBK805R/K805R Low MEFs showed a decrease in the numbers of clathrin-positive vesicles beneath the plasma membrane compared with control cells. In contrast, clathrin staining in the perinuclear region was not affected (Fig. 2B). Moreover, recruitment of the Rab5 effector early endosomal antigen 1 (EEA1) to early endosomes was reduced in PIK3CBK805R/K805R Low MEFs (Fig. 2B), likely as a consequence of alterations in the endocytic route.

Requirement for p110β catalytic function downstream of RTKs and G protein–coupled receptors

Comparison of wild-type and PIK3CBK805R/K805R High MEFs revealed that the catalytic function of p110β was not required for Akt activation 5 min after stimulation with insulin, insulin-like growth factor 1 (IGF-1), EGF, or platelet-derived growth factor (PDGF) (Fig. 3A and fig. S6). Nonetheless, p110β catalytic activity in MEFs was required for lysophosphatidic acid (LPA)– and sphingosine-1-phosphate (S1P)–dependent phosphorylation of Akt (Fig. 3, B and C), providing genetic evidence for the involvement of p110β in heterotrimeric guanine nucleotide–binding protein (G protein)–coupled receptor (GPCR) signaling, as previously suggested (5, 18, 19).

Fig. 3

In vitro and in vivo consequences of p110β kinase-dead expression. (A) IGF-1 and insulin-dependent Akt (also known as protein kinase B, PKB) and extracellular signal regulated kinase 1 and 2 (Erk1 and 2) phosphorylation in wild-type (WT) and kinase-dead (KR) MEFs derived from normal embryos. Shown is a representative Western blot of eight independent experiments analyzing the response without agonist (Co) and at 5 min after stimulation. (B) Analysis of Akt phosphorylation after stimulation with LPA. Wild-type (WT) and PIK3CBK805R/K805R High MEFs were stimulated with 10 μM LPA in the absence or presence of 100 nM TGX-221. (C) Analysis of Akt phosphorylation 5 min after stimulation with S1P of cells of the indicated genotype in the presence or absence of the p110β inhibitor TGX-221. Shown is a representative Western blot of three independent experiments. (D) Growth analysis of PIK3CBK805R/K805R (KR) mice. The figure shows the external appearance of 6-week-old wild-type (WT) and PIK3CBK805R/K805R (KR) mice. (E) Weight gain curve of wild-type (WT) and PIK3CBK805R/K805R (KR) mice from 3 to 24 weeks after birth (males, n = 13; females, n = 22; *P < 0.05, **P < 0.01, ***P < 0.001).

Despite the normal growth of PIK3CBK805R/K805R High MEFs in culture, PIK3CBK805R/K805R mice showed growth retardation, suggesting p110β involvement in growth control in vivo. PIK3CBK805R/K805R mice were born smaller than controls and showed an average of 20% growth retardation that was compensated only after 24 weeks of age, at which time they did not show a significant reduction in muscle weight or alteration in fat mass (Fig. 3, D and E, and fig. S7, A and B). The abundance of p110βK805R in various tissues, including liver, fat, and skeletal muscle, appeared lower than in controls but was never as low as in MEFs derived from abnormal embryos (fig. S7C). Mutant livers also displayed normal amounts of IRβ and IRS-1 (fig. S7D). In PIK3CBK805R/K805R liver extracts, p110β enzymatic activity was undetectable; however, PI3K activity precipitated by either anti-p110α antibodies or pan-p85 unselective antibodies or by phosphopeptide-linked beads (fig. S2F) was unaltered, suggesting normal in vitro activity of other class IA PI3Ks. PIK3CBK805R/K805R mice developed increased blood glucose concentrations and signs of mild insulin resistance that were detectable from 6 months of age (Fig. 4, A to F). This phenotype was accompanied by pancreatic islet hyperplasia and increased insulin secretion (Fig. 4G). Furthermore, mutant mice showed reduced hepatic glycogen deposits and defective insulin-mediated inhibition of gluconeogenesis (Fig. 4, H to J), indicating that the mutation affects insulin-mediated control of liver metabolism. Similarly, mutant mice showed decreased expression of sterol regulatory element–binding protein factor 1c (SREBF1C), a transcription factor that regulates the lipogenic program, as well as reduced amounts of serum triglyceride and cholesterol, showing abnormal lipid metabolism in the liver and consequent dyslipidemia (Table 1).

Fig. 4

Insulin-dependent glucose metabolism in 6-month-old wild-type (WT) and PIK3CBK805R/K805R mice (KR). Statistical significance: *P < 0.05, **P < 0.01, ***P < 0.001. (A) Blood glucose concentration in random fed mice (n = 9 per genotype). (B) Insulin concentrations in the serum of random fed mice (n = 9 per genotype). (C) Blood glucose in fasted mice (WT, n = 9; KR, n = 6). (D) Glucose tolerance test (n = 5 per genotype). (E) Insulin tolerance test (n = 7 per genotype). (F) Insulin concentration in serum of glucose-treated fasted animals. (G) Analysis of pancreatic islet size. Left: histological sections of hematoxylin and eosin (H & E)–stained pancreata. Center: quantification of islet area (n = 4 per genotype). Right: immunofluorescence with antibodies directed against insulin (Ins, red) and against glucagon (Gcn, green). Nuclei were stained with bisbenzimide (DNA, blue). Bars represent 100 μm. (H) Histology of a representative liver section of the indicated mice stained with the periodate–Schiff (PAS) stain recognizing carbohydrates. Measurement of percentage of PAS-positive pixels: WT: 7.3 ± 0.9; KR: 2.27 ± 0.4. (I) Glycogen concentrations in the liver of randomly fed 24-week-old mice (n = 7). (J) Pyruvate challenge of wild-type (WT) and PIK3CBK805R/K805R (KR) mice. A bolus of 2 g/kg was administered intraperitoneally in 24-week-old mice fasted for 16 hours, and the amount of blood glucose was measured at the indicated time points (n = 5).

Table 1 Metabolic measurements in PIK3CBK805R/K805R mice. AST, aspartate aminotransferase; ALT, alanine transaminase; SREBF1C, sterol regulatory element–binding protein factor 1c; PCK1, phosphoenolpyruvate carboxykinase-1; G6PC, glucose-6-phosphatase, catalytic subunit; FBP1, fructose 1,6-bisphosphatase; NS, not significant.
View this table:

Because these findings appear inconsistent with a major role of p110α in insulin signaling (4, 5), we sought to define the molecular mechanism involved by analyzing the signaling complex at the activated insulin receptor. In agreement with p110α being crucial for short-term insulin receptor signaling, at early time points after insulin stimulation, 8-week-old PIK3CBK805R/K805R livers showed normal recruitment of p85 and p110α to IRS-1 (Fig. 5A) as well as unaltered p110 activity (fig. S7E) and Akt phosphorylation (Fig. 5B). However, in livers of wild-type mice treated with TGX-155 or of PIK3CBK805R/K805R mice, insulin-evoked Akt activation declined significantly faster than in untreated wild-type controls (Fig. 5B). Similar results were observed in insulin-stimulated HepG2 hepatoma cells treated with TGX-221 (fig. S8).

Fig. 5

Role of p110β in insulin signaling. (A) Insulin-induced recruitment to IRS-1 of p85 and p110α, determined by IRS-1 immunoprecipitation followed by immunoblot, 5 min after insulin stimulation of livers. (B) Phosphorylation of Akt (on Thr308 and Ser473) determined by immunoblot in livers of mice of the indicated genotype with (+) and without (−) TGX-155 treatment. Lower panel: quantification of Akt phosphorylation on Ser473 (n = 5 mice; **P < 0.01).

Requirement for p110β in ERBB2-mediated mammary carcinogenesis

The observation that p110β catalytic function was required for insulin signaling suggested that this isoform could also be involved downstream of other growth factor receptors or their oncogenic forms. Because p110β can be activated downstream of EGFR (17, 20) and is homogeneously expressed in the epithelium of mammary ducts (Fig. 6A), we studied the PIK3CBK805R mutation in a model of breast cancer triggered by activated ERBB2 (also known as HER-2 or neu) (21, 22), an oncogene known to signal through unidentified PI3K isoforms (23). PIK3CBK805R/K805R mice were intercrossed with BalbC transgenic mice expressing activated ERBB2 (neuT) in the mammary gland. A cohort of 7 PIK3CBK805R/K805R/neuT and 10 PIK3CBWT/WT/neuT female mice was followed for 350 days. In agreement with whole-mount preparations showing a strong reduction of the side buds at 10 weeks of age (Fig. 6C), development of the first tumor was substantially delayed in PIK3CBK805R/K805R/neuT mice (Fig. 6B and Table 2). Furthermore, at least up to 380 days of life, PIK3CBK805R/K805R/neuT mice showed a reduced number of tumors (fig. S9A), which grew at a significantly lower rate than in PIK3CBWT/WT/neuT controls (Table 2 and fig. S9B). Immunohistochemistry showed expression of activated ERBB2 in both genotypes; however, the foci of transformation and the high numbers of proliferating cells that were positive for the protein proliferating cell nuclear antigen (PCNA), which completely filled the duct lumina of PIK3CBWT/WT/neuT mammary glands, were reduced in mutant samples, which showed empty and scarcely proliferating structures (Fig. 6D). To determine whether this protection was intrinsic to the function of p110β in mammary gland epithelium, cells from multiple primary tumors were bulk cultured in vitro. Immunostaining with antibodies directed against E-cadherin confirmed that these cultures consisted of a homogeneous epithelial population (fig. S10). PIK3CBK805R/K805R/neuT cells showed an average of one-third reduction in total p110β abundance but normal amounts of p110α and p85 (Fig. 7A). The development of tumors in compound mutant mice correlated with mutations downstream of PI3K because a reduction in the amount of the PtdIns(3,4,5)P3 phosphatase PTEN was detected in polyclonal PIK3CBK805R/K805R/neuT tumors (Fig. 7A). Despite the decrease in PTEN abundance, PIK3CBK805R/K805R/neuT cell populations grew significantly slower than controls (Fig. 7B). To test if this effect was a result of either the lack of the kinase activity or the reduction in p110βK805R expression, cells of both genotypes were cultured in the presence of p110β-selective inhibitors TGX-155 or TGX-221 (7, 15, 24) (Fig. 7B). This treatment did not show effects in PIK3CBK805R/K805R/neuT cells but caused a significant reduction in proliferation in wild-type tumor cells, showing that oncogenic ERBB2 drives tumor growth largely through p110β catalytic activity.

Fig. 6

p110β is required for ERBB2-driven breast cancer development. (A) p110β expression in mammary gland epithelium. Cryostat sections of mammary glands derived from PIK3CBK805R/K805R/neuT (KR/neuT) and PIK3CB+/+/neuT (WT/neuT) virgin females were stained with the anti–p110β 5g9 monoclonal antibody directly coupled to Alexa 488 (described in Fig. S2D; green) as well as with the nuclear stain propidium iodide (red) and analyzed by confocal microscopy. Bar corresponds to 20 μm. (B) Kinetics of tumor appearance in PIK3CBWT/WT/neuT (WT/neuT; black lines; n = 10) and PIK3CBK805R/K805R/neuT (KR/neuT; red lines; n = 7) compound mutant mice (P = 0.01). (C) Whole-mount preparations of PIK3CBWT/WT/neuT and PIK3CBK805R/K805R/neuT mammary glands at 10 weeks. Upper panel: PIK3CBK805R/K805R/neuT mammary gland shows a reduction of duct side buds constituted by atypical hyperplastic lesions. Lower panel: magnification of the side buds revealing the empty aspect of PIK3CBK805R/K805R/neuT hyperplastic lesions. L, lymph node; mh, atypical mammary hyperplasia. Bar corresponds to 1 mm. (D) Histology of mammary glands. Ducts were stained with anti-ERBB2 and with anti-PCNA antibodies to show transgene expression and proliferating cells, respectively. Bar represents 100 μm. Arrowheads indicate PCNA-positive cells in the mutant sample.

Fig. 7

p110β kinase activity is required for ERBB2-driven breast cancer cell proliferation. (A) Western blot analysis of protein expression in cultured mammary tumors of the two genotypes with the indicated antibodies. (B) Proliferation curves, reported as relative increase in the number of cells compared to the initial seeding density of cultured tumor cells of the two genotypes in the absence or presence of the p110β-selective inhibitors TGX-155 (10 μM) and TGX-221 (100 nM). Statistical significance: wild-type cells versus all other conditions (*P < 0.05, **P < 0.01); other pairs of data sets are not significant.

Table 2 Autochthonous carcinogenesis in PIK3CBK805R/K805R/neuT mice.
View this table:

Discussion

Whereas PI3Ks-mediated signaling has been implicated in the control of cell proliferation, survival, and metabolism, the specific role of the p110β isoform has long remained elusive. We showed here that, unexpectedly, p110β has both kinase-dependent and kinase-independent functions. Indeed, our results indicate that p110β may have a scaffolding function, as already shown for the G protein–coupled p110γ (25). The kinase-independent functions of p110β are sufficient for embryonic development because low abundance of p110β protein led to embryonic lethality, and the presence of p110β—even in a catalytically inactive form—was sufficient to allow embryonic development and viability to adulthood.

The existence of a multiprotein complex containing both p110β and Rab5 (16, 17) suggests a potential function of p110β in clathrin-mediated endocytosis. Our results are consistent with p110β being mainly associated with clathrin-coated vesicles (16) and support the requirement for p110β kinase-independent activity early in the endocytic pathway. Indeed, our findings suggest that GTP-bound active-Rab5 recruits p110β to clathrin-coated pits or vesicles and that the scaffolding activity of p110β contributes to their assembly. In the absence of such scaffolding function, clathrin-coated vesicles seem to be inefficiently formed. In turn, this appears to affect the Rab5-dependent endocytic pathway, resulting in a decrease of EEA1-positive endosomes.

Despite the unexpected finding of its noncatalytic function, p110β kinase activity is required downstream of RTKs and GPCRs. For example, GPCR signaling triggered by S1P and LPA required p110β to trigger Akt phosphorylation. This provides a simple mechanistic explanation for the apparently paradoxical ability of GPCRs to trigger PI3K signaling in cells not expressing p110γ, the prototypical GPCR-activated PI3K (26). Our findings are in agreement with both recent (27, 28) and earlier pioneering studies indicating GPCR-mediated p110β activation (19) by direct association of p110β with the βγ dimer of heterotrimeric G proteins (18).

On the other hand, our data also indicate that p110β catalytic function is required downstream of RTKs and contributes to insulin receptor signaling. PIK3CBK805R/K805R mice show a mild increase in blood glucose concentration and peripheral insulin resistance, which is accompanied by increased insulin secretion and pancreatic islet hyperplasia. This is consistent with a compensation typical of the hyperinsulinemia observed in patients with impaired glucose tolerance (29). In addition, decreased hepatic glycogen deposits and defective insulin-mediated inhibition of gluconeogenesis indicate that the absence of p110β catalytic activity affects insulin-mediated control of liver metabolism. Similarly, p110β is required for lipid metabolism and for the regulation of the lipogenic program as shown by the decreased expression of SREBF1C as well as reduced serum triglyceride and cholesterol concentrations. These findings were unexpected because p110α is the main PI3K isoform known to be involved in insulin signaling (4, 5). Nonetheless, consistent with our findings, reduced p110β expression correlates with the incidence of type 2 diabetes associated with low birth weight (30). Although p110α plays a major role in the insulin signaling pathway, we show that p110β activity is essential to sustain prolonged insulin stimulation. Mechanistically, this effect could perhaps be explained through the autoinhibition of p110α, but not of p110β, after insulin stimulation (31). Although the limited extent of Akt phosphorylation impairment suggests a possible combination of the kinase-dependent and kinase-independent functions of p110β in insulin signaling, our data show that the catalytic activity of p110β supports p110α function. Therefore, whereas the spike in PI3K signaling after insulin and growth factor stimulation depends on p110α (5), our results provide conclusive genetic evidence that p110β supports the PtdIns(3,4,5)P3 production that sustains this response.

A similar mechanism could account for the reduced tumorigenicity associated with oncogenic ERBB2 signaling. Previous studies have shown that PI3K signaling is involved in the ERBB2-mediated proliferation of mammary gland epithelium (32) and that mutations of the p110α-encoding PIK3CA gene are associated with ERBB2 amplification, suggesting a link between p110α and ERBB2 signaling (33). However, because decreased proliferation is detected in PIK3CBK805R/K805R/neuT compound mutant mammary glands, it is possible that instead of p110α, p110β is the critical PI3K involved in the proliferation of ERBB2-positive cancer cells. PIK3CA mutation is mutually exclusive with the loss of the PtdIns(3,4,5)P3 phosphatase PTEN (33), suggesting that p110α might play a little role when PTEN is absent. In agreement, prostate cancer development triggered by PTEN loss is blocked by the lack of p110β and not p110α (27). Our data confirm and extend this idea, providing the first conclusive evidence that inhibition of the catalytic activity of p110β reduces the proliferation of mammary gland cancer cells even under conditions of low PTEN abundance.

In conclusion, the data presented here indicate not only that the kinase-independent functions of p110β are sufficient for embryonic development but also that p110β kinase activity is needed for normal insulin receptor signaling and for the growth of ERBB2-dependent mammary gland cancers (Fig. 8). These results suggest highly tissue-specific functions for p110β and indicate that p110β targeting could be a promising option for treatment of selected tumors such as ERBB2-driven breast cancer.

Fig. 8

Model of the catalytic and noncatalytic functions of p110β. The catalytic activity of p110β is triggered by both RTK and GPCR signaling and cooperates in Akt activation, cancer development, and glucose homeostasis (left). p110β also functions as a scaffold protein (right inset) required for the organization of clathrin-coated pits or vesicles at the plasma membrane, thus controlling RTK endocytosis.

Materials and Methods

Mice

The targeting strategy used to generate the PIK3CBK805R allele was similar to that previously used to generate the PIK3CGK833R allele (25). Briefly, the point mutation leading to the substitution with an arginine of the Lys805 crucial for kinase activity was generated by site-directed mutagenesis of the PIK3CB complementary DNA (cDNA) and fused in-frame with the PIK3CB fifth coding exon, followed by an internal ribosome entry site–enhanced green fluorescent protein cassette and a pA signal. A Neor/herpes simplex virus thymidine kinase selection cassette flanked by loxP sites was placed upstream of the mutant cDNA. To allow conditional expression of the kinase-dead p110β mutant, a duplicated PIK3CB fifth coding exon fused in-frame with the PIK3CB wild-type cDNA and flanked by loxP sites was placed upstream of the selection cassette. The construct was electroporated in E14 embryonic stem cells, and two independent recombinant clones were isolated. Heterozygous mice obtained from germline chimeras were bred with Balancer Cre mice to delete the wild-type cDNA cassette, and the progenies were intercrossed to obtain PIK3CBK805R/K805R mice. Phenotypic analysis was carried out on two lines derived from independent clones. Results were obtained by studying wild-type and mutant littermates derived from heterozygous crosses of mixed 50% 129/Sv-C57Bl/6J–50% BalbC genetic background as well as in ninth-generation C57Bl/6J (for diabetes studies) and third-generation BalbC (for cancer studies) backcrossed mice.

Reagents

The antibody directed against p110β was from Santa Cruz Biotechnology (#sc-602). Antibodies directed against total Akt and P-Ser473 Akt were from a mouse monoclonal clone and from Cell Signaling Technology (#9271), respectively. Antibodies directed against p85 (#4257) and p110α (# 4254) were from Cell Signaling Technology. Antibodies directed against IRS-1 (#05-699) and IRβ subunit (#07-724) were from Upstate Biotechnology Inc.

Protein analysis

Tissues were removed, frozen in liquid nitrogen, and homogenized in lysis buffer (50 mM Tris–HCl, pH 8, and 150 mM NaCl) supplemented with 2 mg/ml aprotinin, 1 mM pepstatin, 1 ng/ml leupeptin, 50 mM NaF, 2 mM sodium orthovanadate, 1 mM sodium pyrophosphate, and 1% Triton X-100. The same buffer was used to solubilize protein extracts from cultured cells. Homogenates were clarified by centrifugation in a microcentrifuge at 4°C. Supernatants were analyzed for immunoblotting or immunoprecipitated either for 1 hour or overnight with the indicated antibodies. Immunocomplexes were bound to 30 μl of a suspension of 50% protein A– or protein G–Sepharose beads and washed with lysis buffer. Beads were resolved on SDS–polyacrylamide gel electrophoresis (PAGE) and transferred onto polyvinylidene difluoride membranes. Blots were probed with the indicated antibodies and developed with enhanced chemiluminescence (ECL, Millipore).

Lipid kinase assay

p110α, p110β, and pan-p85 were immunoprecipitated by use of specific antibodies. Alternatively, all class IA PI3Ks were pulled down with a phosphopeptide YpVPMLG corresponding to the consensus sequence next to tyrosine-751 of the human PDGF receptor β. Proteins were then incubated with 10 mg of phosphatidyl inositol and radiolabeled ATP (10 mM cold ATP, 5 μCi [32P]ATP). PtdIns(3)P products were resolved by thin-layer chromatography and visualized with autoradiography.

Endocytosis assays

Endocytosis assay was performed as described (34). Briefly, MEF cells were serum-starved for 2 hours in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 0.3% bovine serum albumin (BSA). Cells were incubated with purified EGF (100 ng/ml, Upstate Biotechnology Inc.) or fluorescently labeled EGF in DMEM supplemented with 0.3% BSA at 4°C for 1 hour. The EGF-containing medium was then replaced with warm DMEM, and cells were incubated at 37°C for a further 15 min. Cells were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 10 min and permeabilized with 0.1% Triton X-100 in PBS for 10 min. Cells were extensively washed and stained with monoclonal anti-EGFR antibody (Oncogene Science, USA, catalog no. Ab-1), anti-EEA1 antibody (Santa Cruz Biotechnology, catalog no. SC6415), and anti-clathrin antibody (Affinity Bioreagents, CO, USA, catalog no. MA1-065) followed by Cy3-conjugated goat antimouse immunoglobulin G (1:400 in PBS). Nuclei were stained with 4′,6′-diamidino-2-phenylindole (DAPI) (Sigma). Time point 0 was obtained by fixing the cells after incubation with EGF at 4°C for 1 hour. Cells were stained without permeabilization as described above.

Cell culture and proliferation

For determination of relative growth, MEFs were seeded in triplicate at 2.5 × 105 cells per 6-cm tissue-culture dish with DMEM containing GlutaMAX and 4.5 g/l glucose (Invitrogen, USA) supplemented with 10% fetal bovine serum. Cell numbers were evaluated with an automatic cell counter (Casy Technology, Germany). For measurement of tumor proliferation, cells were seeded in triplicate at 5 × 103 cells per 96-well dish and counted with a 3-(4,5-dimethyl thiazol-2-yl)-2,5-diphenyl tetrazolium bromide–based colorimetric assay (Roche, Germany). Inhibitors of p110β were added to the culture 24 hours after plating. Data are the average of at least three independent experiments.

Metabolic studies

Measurements of body weight and length, glucose tolerance tests, and determination of plasma insulin concentration were performed as previously described (35). Insulin tolerance tests were conducted on randomly fed (fed ad libitum) mice by intraperitoneal injection of 1 U of recombinant human insulin per kilogram of body weight, followed by measurement of blood glucose concentration at various time points. Plasma insulin was dosed with the use of a radioimmunoassay kit (#SRI-13K; LINCO Research, USA) following the manufacturer’s instructions. Measurement of other hematological parameters and quantitative polymerase chain reaction were performed following standard protocols (36). Glycogen was measured as described (37).

Histological analysis

For pancreatic islet morphometric analysis, tissue was fixed in 10% neutral-buffered formalin, embedded in paraffin, and cut into 5-μm-thick sections. Sections were stained with hematoxylin and eosin following standard protocols. Islet area was measured with the Metamorph software in eight sections 200 μm apart. Nuclei in each islet were counted and normalized to the measured area.

Histology, immunohistochemistry, and mammary gland whole-mount preparations were performed as previously described (38).

For mammary gland immunofluorescence analysis, mammary glands were fixed in PLP (4% paraformaldehyde, 0.2% periodate, 1.2% lysine in 0.1 M phosphate buffer), embedded in optimum cutting temperature freezing medium and cut into 5-μm-thick sections. Fluorescently labeled 5g9 anti-p110β monoclonal antibody was obtained by coupling the purified Ig with Alexa Fluor 488 (Pierce Biotechnology, # 46403).

Statistical analysis

Statistical significance was calculated with Student’s t test and one- or two-way analysis of variance tests followed by Bonferroni’s post hoc analysis, or Mantel–Haenszel log-rank test where appropriate. Values are reported as the mean ± standard error of the mean.

Acknowledgments

We thank V. Spaziani, L. Virgili, and L. Braccini for technical help; P. Shepherd for inhibitors; and G. Tarone and B. Vanhaesebroeck for constructive discussions. This work was supported by a grant from University of Torino (ex 60%), Programmi di ricerca di Rilevante Interesse Nazionale (PRIN), Telethon, Italian Association for Cancer Research (to E.H. and G.F.), the Sixth Framework Programme EUGeneHeart and Fondation Leducq (to E.H), NIH GM55692 (to J.M.B.), the European Union FP6 grant BBW 03.0441-3/LSHG-CT-2003-502935 and Swiss National Foundation grant 3100A0-109718 (to M.P.W.).

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/1/36/ra3/DC1

Supplementary Materials and Methods

Fig. S1. Description of the gene targeting strategy

Fig. S2. Expression analysis of the products of the PIK3CBK805R allele

Fig. S3. Expression of PIK3CBK805R/K805R does not affect abundance of free p85

Fig. S4. Pharmacological inhibition of p110β by TGX-155 does not impair proliferation of wild-type MEFs

Fig. S5. Transfection of PIK3CBK805R/K805R low MEFs with Rab5Q79L constitutively active mutant does not rescue defective endocytosis of ligand-bound EGFR

Fig. S6. Analysis of Akt phosphorylation by EGF or PDGF

Fig. S7. Body size and protein expression in adult PIK3CBK805R/K805R mice

Fig. S8. Role of PI3Kβ downstream of the insulin receptor in human hepatocellular liver carcinoma cell line (HEPG2)

Fig. S9. Tumor multiplicity and growth in PIK3CBWT/WT/neuT and PIK3CBK805R/K805R/neuT mice

Fig. S10. Homogeneous epithelial characteristics in both mutant and wild-type tumor-derived mammary gland cancer cell lines

Supplementary References

References and Notes

  1. A. J. Robertson, S. Jackson, V. Kenche, C. Yaip, H. Parbaharan, P. Thompson (2001). International Patent Application (Patent WO 0153266 A1).
View Abstract

Navigate This Article