Evidence that inositol polyphosphate 4-phosphatase type II is a tumor suppressor that inhibits PI3K signaling

Abstract

We report that knocking down the expression of inositol polyphosphate 4-phosphatase type II (INPP4B) in human epithelial cells, like knockdown of PTEN, resulted in enhanced Akt activation and anchorage-independent growth and enhanced overall motility. In xenograft experiments, overexpression of INPP4B resulted in reduced tumor growth. INPP4B preferentially hydrolyzes phosphatidylinositol-3,4-bisphosphate (PI(3,4)P(2)) with no effect on phosphatidylinositol-3.4.5-triphosphate (PI(3,4,5)P(3)), suggesting that PI(3,4)P(2) and PI(3,4,5)P(3) may cooperate in Akt activation and cell transformation. Dual knockdown of INPP4B and PTEN resulted in cellular senescence. Finally, we found loss of heterozygosity (LOH) at the INPP4B locus in a majority of basal-like breast cancers, as well as in a significant fraction of ovarian cancers, which correlated with lower overall patient survival, suggesting that INPP4B is a tumor suppressor.

Figure 1. A model indicating the roles of PTEN and INPP4B in regulation of signaling downstream of Class I PI3Ks

Upon stimulation of Class 1 PI3Ks two major phospholipid pools are generated: PI(3,4,5)P₃ and PI(3,4)P₂. PTEN hydrolyzes the 3′-phosphate of PI(3,4,5)P₃ to terminate PI3K signaling. However, SHIP family members hydrolyze the 5′-phosphate of PI(3,4,5)P₃ to generate PI(3,4)P₂, which, like PI(3,4,5)P₃, can facilitate PDK1-dependent phosphorylation and activation of AKT. INPP4B converts PI(3,4)P₂ to PI(3)P. Loss of PTEN or loss of INPP4B results in prolonged activation of Akt, and subsequently in increased cell proliferation, cell migration and invasion.

Figure 2. Knockdown of INPP4B results in anchorage-independent growth and increased cell proliferation

(A) Semi-quantitative RNA-levels of stable HMEC shRNA-INPP4B knockdown colonies grown in anchorage-independent growth assays determined by RT-PCR. Both shRNA hairpins, INPP4B-1 and INPP4B-2, directed against INPP4B resulted in approximately 65% knockdown at the RNA level (n = 3). (B) Knockdown of INPP4B and PTEN in MCF-7 cells. The human mammary epithelial cancer cell line MCF-7 was infected with shRNA hairpins directed against Renilla, PTEN, and INPP4B (‘INPP4B-1’ and ‘INPP4B-2’). Protein levels of INPP4B and PTEN in cells expressing the indicated shRNA were determined by using immunoblotting. Protein loading was shown using antibody directed against GSK3beta. (C) Anchorage-independent growth assay of stable HMEC cell pools infected with retrovirus expressing shRNA directed against Renilla, PTEN, and two hairpin constructs directed against INPP4B (‘INPP4B-1’ and ‘INPP4B-2’). Results are representative of four independent experiments with similar results. Scale bar, 1mm. (D) Quantification of anchorage-independent growth assay of stable HMEC shRNA cell pools (***p < 0.001, n = 3). (E) Proliferation assay of stably infected HMEC shRNA cell pools. Cells from shRNA-Renilla, shRNA-PTEN, shRNA-INPP4B-1 and shRNA-INPP4B+INPP4B^R (rescue cell pool expressing knockdown resistant INPP4B) were counted on day 2, 3, 4 and 5 in a Coulter counter. Averages of three independent experiments are represented. All data are shown as mean +/− SD.

Figure 3. Knockdown of INPP4B results in strong migratory and invasive behavior

(A) Knocking down INPP4B or PTEN with shRNA enhances wound closure following a scratch of confluent HMEC cells. Cells were photographed at time points t = 0 hours and t = 8 hours after wound introduction. Samples were observed under a light microscope. Scale bars, 50μm. (B) Quantification of wound closure. Displayed are the percentages of wound closure at 8 hours after wound introduction determined using NIH Image J (Averages from 3 experiments like those in part A are presented **p < 0.01 and ***p < 0.001). (C) Expression of an INPP4B sequence (INPP4B^R) that circumvents the knockdown of endogenous INPP4B reverses enhanced wound closure induced by INPP4B-1 shRNA. Expression of a catalytically inactive form of INPP4B (INPP4B^R-CD) did not suppress wound closure induced by INPP4B-1 shRNA. The experiment was analogous to that in part A (Averages from three experiments are presented, ***p<0.001). (D) The PI3K inhibitor, LY294002 (10 μM), inhibit accelerated wound closure induced by knocking down INPP4B or PTEN in HMEC. The experiments were analogous to those presented in Part A. (Averages from 3 experiments are presented: ***p < 0.001, **p < 0.01). (E) Knocking down the expression of PTEN or INPP4B enhances the ability of MCF-10A cells to cross a matrigel barrier in a Boyden chamber assay. MCF-10A were stably infected with shRNA directed against Renilla, PTEN, and INPP4B (INPP4B-1 and INPP4B-2). Cells were seeded in transwell chambers containing matrigel as a barrier. No stimulus was applied to the bottom chamber, in order to examine non-directional invasive behavior in knockdown cell pools. 24 hours after seeding, invaded cells were fixed, stained and quantified. (Average from three experiments are presented: ***p < 0.001). All data are shown as mean +/− SD.

Figure 4. Knocking down INPP4B or PTEN in MCF-10A cells results in dysmorphic cell clusters upon growth in three-dimensional culture, and overexpression of INPP4B suppresses tumor growth in the SUM149 xenograft mouse model

(A) Stable MCF-10A shRNA cell pools were grown on a matrigel layer under differentiating conditions and observed at day 21 of culture. Left panels display grown, untreated spheres of Renilla, PTEN, and INPP4B knockdown cells. The same stable shRNA cell pools were treated starting day 1 with 1 nM Rapamycin (middle left panels), 10 μM LY294002 (middle right panels), or 10 μM UO126, respectively (right panels). Representative pictures are shown. The experiment was repeated three times with similar results. Scale bar, 50μm. (B) Nude mice were injected with the human invasive ductal carcinoma cell line SUM149 expressing Flag-INPP4B (n=4) or empty Flag control (n=4). Mice were sacrificed 4 weeks post injection and tumor growth and size evaluated. All mice injected with SUM149 cells expressing empty Flag vector showed tumor growth (n=4), while mice injected with SUM149 cells expressing Flag-INPP4B showed reduced tumor growth and size (n=2). (C) Quantification of tumor diameter in xenograft experiment. All mice in the control group (SUM149+Flag Empty) showed tumor growth (n=4). Overexpression of Flag-INPP4B in SUM149 resulted in reduced tumor growth (n=4, p=0.011). (D) Protein levels of INPP4B in SUM149 cells expressing empty Flag vector or Flag-INPP4B were determined by using immunoblotting. Protein loading was shown using antibody directed against p85.

Figure 5. Substrate specificity of INPP4B

(A) FLAG-tagged INPP4B was immunoprecipitated from 293T cells and incubated with either [³²P]-PI(3)P, or [³²P]-PI(3,4)P₂ or [³²P]-PI(3,4,5)P₃. As control, the phosphoinositides were incubated with Flag-immunoprecipitate from transfected 293T cells expressing an empty Flag-expression construct. The percent hydrolysis of each lipid was determined by chloroform/methanol/HCL extraction, thin layer chromatography and Phosphoimager analysis. (n=3: ***p < 0.001). (B) Overexpression of human INPP4B causes a reduction in cellular PI(3,4)P₂ levels in vivo. 3T3-L6 cells were transiently transfected with empty Flag or Flag-INPP4B constructs and labeled with [³²P]-inorganic phosphate. Lipids were extracted, deacylated and the headgroups separated by HPLC. The radioactivity in the glyceroyl-phosphoryl-inositol moieties of each of the D-3 phosphorylated phosphoinositides was then normalized to the total radioactivity in the more abundant phosphoinositides, PI(4)P and PI(4,5)P₂ (which did not significantly vary between experiments). The bars indicate per cent changes in these ratios in INPP4B transfected cells compared to control cells. The actual ratios of radioactivity in each lipid to total radioactivity in PI(4)P plus PI(4,5)P₂ in the control 3T3-L6 cells were: PI(3)P – 3.2%, PI(3,5)P₂ – 0.035%, PI(3,4)P₂ – 0.17%, PI(3,4,5)P₃ – 0.69%. (Averages from three experiments are presented: *p<0.1). Protein expression of Flag-INPP4B in 3T3-L6 cells was demonstrated by Immunoblotting. Protein loading was shown in using antibody directed against PTEN. (Inset). All data are shown as mean +/− SD.

Figure 6. Double knockdown of INPP4B and PTEN in HMEC cells results in increased cellular senescence

(A) Proliferation assay of stably infected HMEC shRNA cell pools including double knockdown of INPP4B and PTEN. Cells were counted on indicated days. Results are representative of three independent experiments with similar results. (B) Morphology of anchorage-independent growth assays of stable HMEC cell pools infected with retrovirus expressing shRNA directed against Renilla, PTEN, INPP4B, and INPP4B + PTEN. Results are representative of three independent experiments with similar results. Scale bar, 1mm. (C) Quantification of anchorage-independent growth assay of stable HMEC shRNA cell pools (***p<0.001, n=3). (D) SA-beta-Gal activity of stable shRNA HMEC cell pools without or with knockdown of Trp53 was assessed. The error bars indicate the standard deviation of three replicates (n = 3: *p<0.1, ***p<0.001). Data are shown as mean +/− SD. (E) Protein levels of INPP4B, PTEN, and Trp53 in cells expressing the indicated shRNA were determined by using immunoblotting. Protein loading was shown using antibody directed against p85. All data are shown as mean +/− SD.

Figure 7. Knocking down the expression of INPP4B increases the duration of AKT activation in response to insulin

(A) Stable HMEC cell pools expressing shRNA vectors directed against Renilla, PTEN, INPP4B (‘INPP4B-1’) or HMEC cell pools expressing the knockdown resistant INPP4B expression construct in the shRNA-INPP4B-1 cell pool (‘INPP4B-1 + INPP4B^R’) were serum-starved and stimulated with 100nM insulin for the indicated time periods. Cell lysates were separated by SDS-PAGE and phospho-Akt detected with specific antibodies in Western blot analysis. INPP4B knockdown cell pools display increased and prolonged phospho-Thr308 and phospho-Ser473 Akt compared to Renilla control knockdown cell pools. Expression of a Flag-tagged INPP4B construct harboring two silent mutations at Ser487 to render it resistant to shRNA knockdown in shRNA-INPP4B-1 cell pools (INPP4B^R) reversed prolonged Akt phosphorylation. The results are representative of 4 separate experiments. (B) Quantification of phospho-Thr308 Akt normalized to total Akt levels (n = 4, *p< 0.1, **p<0.01, ***p<0.001). Quantification of each single time point was determined using NIH ImageJ (n = 4). Data are shown as mean +/− SEM.

Figure 8. Loss of heterozygosity at the INPP4B locus occurs frequently in BRCA1-mutant and basal-like breast cancers and ovarian cancers and correlates with decreased overall patient survival and lymph node involvement

(A) Frequency of loss of heterozygosity (LOH) in the 4q31 region of Chromosome 4 from 43 high-grade human breast tumors is shown. In the data set, five tumors were carrying germline BRCA1 mutations, 18 additional tumors displayed basal-like features and 20 tumors were non-basal-like. The data come from 10K SNP array and were analyzed using dChip software. LOH of INPP4B, located at 4q31.21, occurred in 60%, 55.6%, and 5% of BRCA1-mutant, basal-like, and non-basal-like tumors, respectively (p<0.001). (B) Differential expression of INPP4B in subtypes of breast cancers analyzed by gene expression array. Sample subsets are indicated across the top as follows: BRCA1 germline mutation-associated tumors (B1, pink), sporadic basal-like tumors (basal-like, red), Luminal predominantly ER positive and HER2 negative tumors (ER+/HER2-, blue), and HER2 positive tumors (HER2+, green), and normal breast samples from mammographic reductions (Normal, yellow). Each column represents an individual sample. Each row represents the relative expression level measured from two different probes to INPP4B. Relative expression levels are represented as follows: mean levels are shown in white, expression levels above mean in progressively darker shades of red, and expression levels below mean in progressively darker shades of blue. The relative expression color scale is shown at the bottom. (C) Percent survival graph over time (in months) for breast cancer tissue stained for INPP4B expression (n = 112). Tissues from breast cancer patients were stained for INPP4B and scored for expression levels (0 = no expression, 1 = low expression, 2 = high expression). Breast cancer patients with no expression of INPP4B in tumor tissue showed significantly shorter over all survival compared to patients with low and high INPP4B expression in breast tumor tissue (p = 0.0018). (D) Percent survival graph over time (in months) for ovarian cancer tissue stained for INPP4B expression (n = 50). Tissues from ovarian cancer patients were stained for INPP4B and scored for protein expression levels (0 = no expression, 1 = low expression, 2 = high expression). Ovarian cancer patients with no expression of INPP4B in tumor tissue showed poor over all survival compared to patients with low or high INPP4B expression (p<0.0001). (E) Ovarian cancer tissues were analyzed for correlation of loss of INPP4B protein expression and lymph node involvement (LN Met). Lymph node metastases were found significantly more frequent in ovarian cancer tissues, which lost INPP4B protein expression (p = 0.043). The graph displays percent of patients with positive lymph nodes over INPP4B expression levels (0 = no expression, 1 = low expression, 2 = high expression).