p85α regulatory subunit isoform controls PI3-kinase and TRPC6 membrane translocation

Abstract

Activation of phosphatidylinositol 3-kinase (PI3K) by lipid oxidation products, including lysophosphatidylcholine (lysoPC), increases the externalization of canonical transient receptor potential 6 (TRPC6) channels leading to a subsequent increase in intracellular calcium that contributes to cytoskeletal changes which inhibit endothelial cell (EC) migration in vitro and impair EC healing of arterial injuries in vivo. The PI3K p110α and p110δ catalytic subunit isoforms regulate lysoPC-induced TRPC6 externalization in vitro, but have many other functions. The goal of the current study is to identify the PI3K regulatory subunit isoform involved in TRPC6 externalization to potentially identify a more specific treatment regimen to improve EC migration and arterial healing, while minimizing off-target effects. Decreasing the p85α regulatory subunit isoform protein levels, but not the p85β and p55γ regulatory subunit isoforms, with small interfering RNA inhibits lysoPC-induced translocation of the PI3K catalytic subunit to the plasma membrane, dramatically decreased phosphatidylinositol (3,4,5)-trisphosphate (PIP3) production and TRPC6 externalization, and significantly improves EC migration in the presence of lysoPC. These results identify the important and specific role of p85α in controlling translocation of PI3K from the cytosol to the plasma membrane and PI3K-mediated TRPC externalization by oxidized lipids. Current PI3K inhibitors block the catalytic subunit, but our data suggest that the regulatory subunit is a novel therapeutic target to promote EC migration and healing after arterial injuries that occur with angioplasty.

Graphical Abstract

1.0. INTRODUCTION¹

EC migration is essential for healing intimal injuries such as those that occur with angioplasty, but hypercholesterolemia reduces endothelial healing [1, 2]. Oxidized low-density lipoprotein (oxLDL), and lysophosphatidylcholine (lysoPC), the major lysophospholipid of oxLDL, accumulate in atherosclerotic arteries [3] and at regions of injury. LysoPC inhibits EC migration in vitro [4, 5], and increased lysoPC levels are associated with decreased endothelial healing in vivo [1, 2, 6]. The effect of lysoPC is mediated, at least in part, through its activation of canonical transient receptor potential 6 (TRPC6) and TRPC5 channels, but not voltage-gated Ca²⁺ channels [7, 8]. Our previous in vitro studies and in vivo studies in Trpc6^−/− mice provide compelling evidence of the importance of the TRPC6 to TRPC5 activation cascade in impaired EC migration and arterial healing in hypercholesterolemic mice [7, 9].

Oxidized lipids and lysoPC induce TRPC6 externalization by activating phosphatidylinositol 3-kinase (PI3K) [10]. PI3K activation increases production of phosphatidylinositol (3,4,5)-trisphosphate (PIP3). PIP3 binds to TRPC6, anchoring it in the plasma membrane, and thus promoting increased Ca²⁺ influx[11]. The rise in intracellular calcium ion concentration ([Ca²⁺]_i) leads to alterations of the cytoskeleton and inhibits EC migration across an area of injury [8].

PI3 kinases are divided into four classes on the basis of structure and function, and are composed of a catalytic subunit and a regulatory subunit. Class I PI3K isoforms catalyze the conversion of phosphatidylinositol (4,5)-bisphosphate (PIP2) into PIP3. Class I PI3Ks are composed of a p110 catalytic subunit (p110α, β, δ, γ), and a regulatory subunit (p85α, p85β, p50α, p55α or p55γ), and they are the most well-characterized in vascular biology [12]. We have shown that the p110α and p110δ catalytic subunit isoforms of PI3K, but not the p110β or p110γ isoforms, are responsible for the lysoPC-induced increase in PIP3 production and TRPC6 externalization, and decreased migration of ECs in the presence of lysoPC [13]. Pharmacologic inhibition of PI3K and its p110α and p110δ isoforms partially preserves arterial healing in hypercholesterolemic mice, but the multiple functions of PI3K suggest that a more specific inhibition of the effect PI3K on TRPC6 would be desirable to preserve EC migration in the presence of lipid oxidation products.

The regulatory subunit of PI3K influences subcellular location, binding partners and activity of the catalytic subunit [14]. Previously, we showed that lysoPC induces the phosphorylation of calmodulin at Tyr99, which leads to the association of calmodulin with the PI3K p85 regulatory subunit and subsequent increase in PI3K activity [10], but the specific isoform of the regulatory subunit was not identified. The p85α and p85β isoforms of the PI3K regulatory subunit are ubiquitously expressed, but p85β is expressed at lower levels in the vasculature [14, 15]. p55γ is expressed predominantly in brain, testes, and embryonic tissue, but is present in heart and aorta [16, 17]. p50α and p55α are splice-variants of the gene encoding p85α and have limited tissue distribution. Therefore, the focus of the study is on the p85α, p85β, and p55γ regulatory subunit isoforms. The goal of this study is to identify the PI3K regulatory subunit isoform(s) activated by lysoPC and determine if decreasing the level of the specific regulatory isoform(s) inhibits TRPC6 externalization and preserves EC migration in the presence of lipid oxidation products.

2.0. MATERIALS AND METHODS

2.1. EC Harvest and culture

After collagenase treatment, bovine aortic ECs (BAECs) (unknown gender) were isolated from fresh bovine aortas (from an abattoir) and cultured as previously described [7]. All experiments were performed on BAECs from passage 4 to 10. To minimize contamination of these primary cell cultures, cloning disc were utilized to subculture only ECs. EC identity was confirmed by immunostaining subsets of each culture with anti-CD-31 monoclonal antibody, as described below[7].

Immortalized human umbilical vein ECs (EA.hy926 cells) (unknown gender) were cultured in Dulbecco's Modified Eagle's Medium (DMEM) containing 10% fetal bovine serum (FBS) (Hyclone Laboratories, Logan, UT) with 1 μg/mL gentamicin.

2.2. Immunofluorescence imaging for the endothelial cell marker CD-31

ECs were grown on 25 mm coverslips. At 60% confluence, the cells were fixed in 100% methanol for 10 min at 4°C. The cells were permeabilized with Triton-X-100 (0.1%) for 10 min, then blocked with bovine serum albumin (3%) for 1 h. The EC were incubated with anti-CD-31 monoclonal antibody conjugated with Alexa Fluor^™ 488 (1:100; Invitrogen, Brookfield, WI; Catalog No. MA5-18135) overnight at 4°C. After washing the cells, the nuclei were counterstained with propidium iodide (2 μg/mL; Sigma-Aldrich, St. Louis, MO; Catalog No. P4170). The coverslips were mounted with Vectashield Reagent (Vector Laboratories, Burlingame, CA) and imaged using a Leica fluorescence microscope (λ_ex 495 nm, λ_em 519 nm; Heidelberg, Germany).

2.3. Immunoprecipitation and immunoblot analysis of proteins

Confluent ECs were made quiescent in serum-free DMEM for 12 h before specified agents were added. Target proteins were immunoprecipitated as previously described [18]. Cell lysates were stored at −20° C until analyzed.

Using a 4-20% gradient SDS-PAGE, lysate proteins (50 μg/lane of total protein) were separated, as previously described [18]. Target proteins were detected by antibodies specific for TRPC6 (1:500; Cell Signaling Technology, Danvers, MA Catalog No. 16716), p85α regulatory subunit of PI3K (1:500; Abcam, Boston, MA; Catalog No. AB40755), p85β regulatory subunit of PI3K (1:500; R&D Systems, Minneapolis, MN; Catalog No. MAB6777-sp), p55γ regulatory subunit of PI3K (1:500; GeneTex, Irvine, CA; Catalog No. GTX10779 ), p110α catalytic subunit of PI3K (1:500; Cell Signaling Technology, Danvers, MA; Catalog No. 4255S), p110β catalytic subunit of PI3K (1:500; Novus Biologicals, Centennial, CO; Catalog No. NBP1 33116), the p110δ catalytic subunit of PI3K (1:500; Cell Signaling Technology, Danvers, MA; Catalog No. 34050S), or phosphotyrosine (1:1000; Invitrogen, Brookfield, WI; Catalog No. 615800). Blots were incubated with a horseradish peroxidase-conjugated secondary antibody (1:1000; Cell Signaling Technology, Danvers, MA; Catalog No. 7074) and the signal was developed using a chemiluminescent reagent (Invitrogen, Brookfield, WI). The blots were stripped and reprobed using an anti-actin antibody (1:1000; Santa Cruz Biotechnology, Santa Cruz, CA; Catalog No. MA1-744) to verify equal loading. After acquiring the images on HyBlot CL autoradiography film (Denville Scientific, Metuchen, NJ), the images were scanned with an Epson Perfection V600 Photo using EPSON scan Software. ImageJ software (NIH, Bethesda, MD) was used to quantitate protein band density.

2.4. Reducing target protein using small interfering RNA (siRNA)

ECs at 80% confluence were transiently transfected with the siRNA duplex specific to p85α (30nmol/L, Dharmacon, Lafayette, CO), p85β (30nmol/L, Dharmacon), p55γ (30 nmol/L, Dharmacon), p110α (10-20 nmol/L, Dharmacon), or p110δ (20-30 nmol/L, Dharmacon) using a transfection kit (RNAiFect; Qiagen, Germantown, MD) following the manufacturer’s protocol. Decrease of the endogenous target protein level was verified after 48 h by immunoblot analysis. NsiRNA (20–40 nmol/L; Ambion, Austin, TX) with no known homology to any gene sequence was used as a negative control.

2.5. Separation of membrane and cytosolic fractions

Confluent ECs were made quiescent in serum-free DMEM for 12 h before specified agents were added. The Mem-Per Plus protein extraction kit (ThermoFisher Scientific, Waltham, MA; Catalog No. 89842) was used to isolate cell membrane proteins, following the manufacturer’s protocol with minor modifications. Briefly, the ECs were scraped from the plates with cell wash solution. The cell suspension was centrifuged at 2000 x g for 5 min. The supernatant was discarded and the pelleted cells were washed twice with Cell Wash Solution. Permeabilization Buffer was added and the cells were permeabilized while mixing continuously at 4°C for 20 min. An aliquot was separated for later use to determine total target protein levels. The permeabilized cells were centrifuged at 16,000 x g for 15 minutes. The supernatant containing the cytosolic proteins was removed without disrupting the pellet containing the membrane fraction. Solubilization Buffer was added to the membrane fraction and the sample was mixed for 45 min at 4°C. After centrifuging the sample at 16,000 x g for 15 min, the supernatant containing those solubilized membrane proteins was collected and stored at −80°C for later use. To determine total target protein level, the aliquot separated after the cells were permeabilized was mixed with lysis buffer. Total target protein level was determined by immunoblot analysis as described above.

2.6. PIP3 production by ELISA

Confluent ECs were made quiescent in serum-free medium for 12 h and agents added as indicated. A PIP3 Mass ELISA kit (Echelon Biosciences, Salt Lake City, UT) was used to extract PIP3, following the manufacturer’s protocol. A Spectramax 190 microplate reader (Molecular Devices, San Jose, CA) was used to measure PIP3 levels at 450 nm.

2.7. Biotinylation of proteins on EC cell surface to detect TRPC6 externalization

Biotinylation assay was used to determine TRPC6 channel externalization, as previously described [7]. ECs were treated with Sulfo-NHS-Biotin (2 mg/mL; Calbiochem, La Jolla, CA) to biotinylate externalized proteins. Following lysis of the ECs, the lysates were incubated with streptavidin-agarose beads. After collecting the beads, the biotinylated proteins were released from the beads, separated and resolved by SDS-PAGE, and detected by immunoblot analysis. Immunoblot analysis was performed on an aliquot of cell lysate removed prior to incubation with the streptavidin-agarose beads to determine total TRCP6 protein in the lysates.

2.8. Measurement of [Ca²⁺]_i

ECs at 80-90% confluence were loaded with the FITC-conjugated fluorophore Calbryte^™ 520 AM dye (AAT Bioquest, Sunnyvale, CA; Catalog No. 36310) following the manufacturer’s protocol. Calbryte^™ 520 AM is a cell-permeable indicator for the measurement of intracellular calcium that is hydrolyzed by intracellular esterase when it enters the cell to become activated. The activated indicator is a polar molecule that is unable to diffuse through the cell or ER membrane, trapping it in the cytosol. After 35 min incubation period, the ECs were suspended and loaded into the sort chamber of a BD FACSMelody^™ cell sorter (BD Biosciences, San Jose, CA) maintained at 37°C. After adjusting the baseline, lysoPC (12.5 μmol/L; 1-palmitol-2-hydroxy-sn-glycero-3-phosphocholine; Avanti Polar Lipids, Alabaster, AL) was added. The relative changes in [Ca²⁺]_i were read using the kinetic reading mode at Ex/Em 490/525 nm. FlowJo^™ v10 software (BD Biosciences) was used to analyze the kinetics data.

2.9. EC migration assay

Confluent ECs were made quiescent in serum-free DMEM for 16-18 h prior to the migration assay. The ECs were incubated with specified agents, and migration was assessed by the razor scrape assay, as previously described [8]. After 24 h, an observer blinded to the experimental condition quantitated migrating ECs using NIH ImageJ (NIH).

2.10. Proliferation Assay

ECs were cultured to 70-80% confluence in DMEM containing 10% fetal calf serum. The medium was removed, ECs washed twice with PBS, and DMEM without FBS was applied. Cell proliferation was assessed using the Quick Cell Proliferation Kit II (Abcam, Boston, MA; Catalog No. ab65475), following the manufacturer’s protocol. Absorbance was measured at 450 nm using a Spectramax 190 microplate reader (Molecular Devices). Fibroblast growth factor (FGF; 20 nM) and camptothecin (8 μM) were used as positive and negative controls, respectively.

2.11. EC viability assay

ECs at 80-90% confluence were loaded with the Calcein-AM fluorophore (5μM, AAT Bioquest, Sunnyvale, CA; Catalog No. 21905) for 30 minutes. Calcein-AM is a hydrophobic compound that easily penetrates intact and live cells and remains in the cell cytoplasm. EC were harvested using trypsin, suspended in DMEM without FBS , and loaded into the sort chamber of a BD FACS Aria^™ Fusion Cell Sorter (BD Biosciences, San Jose, CA) maintained at 37 °C. Calcein-AM fluorescence were measured at Ex/Em = 485/530 nm. The measured fluorescence intensity is proportional to the number of viable cells. DMEM and H₂O₂ (3mM) were used as a positive and negative controls, respectively. FlowJo^™ v10 software (BD Biosciences, San Jose, CA) was used to analyze the viability data.

2.12. Statistical methods

Experimental results are represented as mean ± SD. Unless otherwise specified, experiments were performed in triplicate with cells cultured from three different animals or independent biological samples. All Western blots and EC migration images are representative images of at least three separate experiments. GraphPad Prism 9.1 (GraphPad Software, Inc., San Diego, CA) was used to perform the statistical analysis. Analysis of variance (ANOVA) with Tukey’s multiple comparison’s test, ANOVA with Dunnett’s multiple comparison test, or Student’s t-test was performed. Differences were considered statistically significant at P<0.05.

3.0. RESULTS

3.1. PI3K regulatory subunit isoform-specific siRNA selectively decreases target protein levels without affecting catalytic subunit protein levels.

Isoform-specific siRNA was used to selectively decrease the protein levels of the PI3K p85α, p85β, or p55γ regulatory subunits. Transient transfection of BAECs (Fig. S1) with p85α siRNA decreased p85α protein levels to 30% of basal level, but had no effect on the level of p85β or p55γ protein (Fig. 1A). Similarly, transient transfection of BAECs with p85β siRNA decreased p85β protein levels to 36% of basal level, but had not effect on the level of p85α or p55γ protein (Fig. 1B). Transfection of BAECs with p55γ siRNA decreased p55γ protein levels to 35% of basal level, but had not effect on the level of p85α or p85β protein (Fig. 1C). Decreasing PI3K regulatory protein levels with siRNA did not alter protein levels of the PI3K catalytic subunits p110α (Fig. 2A) or p110δ (Fig 2B), which are the catalytic subunit isoforms that mediate activation of PI3K in the presence of lysoPC[13].

Fig. 1. siRNA for the PI3K regulatory subunit isoforms selectively reduces the corresponding regulatory subunit isoform protein levels.

Bovine aortic ECs were transiently transfected with control siRNA (NsiRNA) (40 nmol/L), p85α siRNA (30 nmol/L), p85β siRNA (30 nmol/L), or p55γ siRNA (30 nmol/L) for 24 hours. The siRNA was then removed. At 48 h after siRNA removal, the cells were lysed and p85α (A), p85β (B), or p55γ (C) was identified by immunoblot analysis. Actin served as the loading control (n=3 independent biological samples). Black lines indicate lanes rearranged from the same gel. All bands are from the same gel. Densitometry measurements of the appropriate regulatory subunit protein are represented in scatter plots after decreasing p85α (A, right panel), p85β (B, right panel), or p55γ (C, right panel) protein levels (n=3 independent biological samples, * P< 0.001 compared to NsiRNA, † P < 0.001 compared to p85α siRNA, ‡ P < 0.001 compared to p85β siRNA, and § P < 0.001 compared to p55γ siRNA). Statistical analysis was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test.

Fig. 2. siRNA for the PI3K regulatory subunit isoforms does not reduce the p110 catalytic subunit protein levels.

Bovine aortic ECs were transiently transfected with control siRNA (NsiRNA) (40 nmol/L), p85α siRNA (30 nmol/L), p85β siRNA (30 nmol/L), p55γ siRNA (30 nmol/L), p110α siRNA (20 nmol/L) (positive control), or p110δ siRNA (20 nmol/L) (positive control) for 24 hours. The siRNA was then removed. At 48 h after siRNA removal, the cells were lysed and p110α (A), or p110δ (B) was identified by immunoblot analysis. Actin served as the loading control (n=3 independent biological samples). Black lines indicate lanes rearranged from the same gel. All bands are from the same gel. Densitometry measurements of the p110α (A, right panel) or p110δ (B, right panel) catalytic subunit protein are represented in scatter plots after decreasing p85α, p85β, or p55γ protein levels (n=3 independent biological samples, * P < 0.001 compared to NsiRNA, † P < 0.001 compared to p110α siRNA, and ‡ P < 0.001 compared to p110δ siRNA). Statistical analysis was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test.

3.2. PI3K p85α subunit phosphorylation is induced by lysoPC.

To identify the specific regulatory subunit isoform involved in the lysoPC-induced activation of PI3K, we evaluated the impact of lysoPC on the activation of the p85α, p85β, and p55γ regulatory subunit isoforms. Previous reports showed that tyrosine phosphorylation was associated with activation of the regulatory subunit [19], but serine phosphorylation of the regulatory subunit was associated with inhibition of PI3K activity [20]. In the presence of lysoPC, the p85α isoform was phosphorylated at tyrosine residues (Fig. 3A), but the p85β and the p55γ isoforms were not tyrosine phosphorylated (Fig. 3B-C). The results suggested that the p85α isoform was the only regulatory subunit isoform activated in the presence of lysoPC.

Fig. 3. p85α is phosphorylated in the presence of lysoPC.

Bovine aortic ECs were incubated with lysoPC (12.5 μmol/L) for 15 min. p85α (A), p85β (B), or p55γ (C) were immunoprecipitated and immunoblot analysis for phosphotyrosine was performed. Immunoblot analysis of total p85α (A), p85β (B), or p55γ (C) was performed for control. Densitometry measurements of phospho-p85α (A, right panel), phospho-p85β (B, right panel), or phospho-p55γ (C, right panel) are represented in dot-whisker plots (n=3 independent biological samples, * P < 0.001 compared to medium). Statistical analysis was performed using unpaired Student’s t-test.

3.3. p85α activation leads to translocation of the p110α and p110δ catalytic subunits to the plasma membrane.

Only PI3K Class I produce PIP3 [21] in the inner leaflet of the plasma membrane, so we investigated the impact of tyrosine phosphorylation of the regulatory subunit on the translocation of the PI3K catalytic subunit from the cytosol to the plasma membrane. Based on our prior studies, the p110α and p110δ catalytic subunit isoforms, but not the p110β or p110γ isoforms, were responsible for the increase in PIP3 production in the presence of lysoPC [13]. To assess translocation, EC were lysed under the various treatment conditions and separated into cytosolic and membrane fractions. Under basal conditions, p110α localized to the cytosolic fraction, but in the presence of lysoPC, p110α translocated from the cytosolic fraction to the membrane fraction (Fig. 4A-C). Transfection of ECs with p85α siRNA inhibited the lysoPC-induced translocation of the p110α subunit from the cytosolic fraction to the membrane fraction, but did not decrease total p110α protein levels (Fig. 4A). In contrast, transfection of ECs with p85β siRNA (Fig 4B) or p55γ siRNA (Fig. 4C) did not block the lysoPC-induced translocation of p110α to the plasma membrane. p85β siRNA (Fig 4B) and p55γ siRNA (Fig. 4C) also did not alter total p110α protein levels. To ensure the effect of siRNA-mediated decrease in p85α on p110α translocation was not specific to the p110α isoform, we assessed the impact on the p110δ and p110β catalytic subunit isoforms. Similar to p110α, decreased p85α levels blocked the lysoPC-mediated translocation of p110δ from the cytosolic fraction to the membrane, while not altering total p110δ protein levels (Fig. 5A). As expected based on our prior studies, neither lysoPC nor p85α protein levels impacted the localization of the p110β isoform or total p110β protein levels (Fig. 5B). These results showed that the translocation of the catalytic subunits (p110α and p110δ) associated with increased PI3K activity in the presence of lysoPC was dependent on the p85α regulatory subunit isoform.

Fig. 4. Decreasing p85α regulatory subunit protein levels with siRNA blocks the translocation of the p110α catalytic subunit from the cytoplasm to the plasma membrane in the presence of lysoPC.

Bovine aortic ECs were transiently transfected with control siRNA (NsiRNA) (40 nmol/L), p85α siRNA (30 nmol/L), p85β siRNA (30 nmol/L), or p55γ siRNA (30 nmol/L) for 24 h before incubation with lysoPC (12.5 μmol/L) for 15 min. The ECs were permeabilized and separated in the cytosol and membrane fractions. Location of p110α in the cytosol and plasma membrane fractions was determined by immunoblot analysis after decreasing p85α (A), p85β (B), or p55γ (C) protein levels. Total p110α in the permeabilized cell solution aliquot obtained prior to separation into the fractions was determined by immunoblot analysis. Actin served as loading control (n=3 independent biological samples). Black lines indicate lanes rearranged from the same gel. All bands are from the same gel. Densitometry measurements of cytosol and membrane p110α fractions are represented in dot-whisker plots after decreasing p85α (A, middle panel), p85β (B, middle panel), or p55γ (C, middle panel) protein levels (n=3 independent biological samples). Densitometry measurements of total p110α prior to separation into cytosol and membrane fractions are represented in dot-whisker plots after decreasing p85α (A, right panel), p85β (B, right panel), or p55γ (C, right panel) protein levels (n=3 independent biological samples). * P < 0.01 compared to NsiRNA, † P < 0.01 compared to NsiRNA + LysoPC, ‡ P < 0.01 compared to p85β NsiRNA, and § P < 0.01 compared to p55γ NsiRNA). Statistical analysis was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test.

Fig. 5. Decreasing p85α regulatory subunit protein levels with siRNA blocks the translocation of the p110δ, but not the p110β, catalytic subunit from the cytoplasm to the plasma membrane in the presence of lysoPC.

Bovine aortic ECs were transiently transfected with control siRNA (NsiRNA) (40 nmol/L) or p85α siRNA (30 nmol/L) for 24 h before incubation with lysoPC (12.5 μmol/L) for 15 min. The ECs were permeabilized and separated in the cytosol and membrane fractions. Localization of p110δ (A) and p110β (B) in the cytosol and plasma membrane fractions was determined by immunoblot analysis. Total p110δ (A) or p110β (B) in the permeabilized cell solution aliquot obtained prior to separation into the fractions was determined by immunoblot analysis. Actin served as loading control (n=3 independent biological samples). Black lines indicate lanes rearranged from the same gel. All bands are from the same gel. Densitometry measurements of cytosol and membrane p110δ (A, middle panel) or p110β (B, middle panel) fractions are represented in dot-whisker plots (n=3 independent biological samples). Densitometry measurements of total p110δ (A, right panel) or p110β (B, right panel) prior to separation into cytosol and membrane fractions are represented in dot-whisker plots (n=3 independent biological samples). * P < 0.01 compared to NsiRNA, † P < 0.01 compared to NsiRNA + LysoPC). Statistical analysis was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test.

3.4. PI3K p85α regulatory subunit is needed for lysoPC-induced PI3K activation.

PIP3 production was measured to determine the effect of siRNA-mediated decrease of the PI3K regulatory subunit isoforms on lysoPC-induced PI3K activation. Basal PIP3 production was similar in control BAECs and BAECs transfected with NsiRNA, p85α siRNA, p85β siRNA, or p55γ siRNA (Fig. 6A-B). LysoPC significantly increased PIP3 production in control BAECs (Fig. 6A-B) and in cells transfected with NsiRNA (Fig. 6A-B). The lysoPC-induced increase in PIP3 production by lysoPC was significantly reduced by siRNA-mediated decrease in p85α protein (Fig. 6A). The siRNA-mediated decrease in p85β or p55γ did not alter the increased PIP3 production in response to lysoPC (Fig. 6B).

Fig. 6. Decreasing p85α regulatory subunit protein levels with siRNA reduces lysoPC-induced PIP3 production.

(A) Bovine aortic ECs were transiently transfected with NsiRNA (40 nmol/L) or p85α siRNA (30 nmol/L). PIP3 production in the presence or absence of lysoPC (12.5 μmol/L) was determined by ELISA and represented by scatter plot (n=3 independent biological samples, * P<0.001 compared to medium, † P<0.001 compared to NsiRNA, ‡ P<0.001 compared to p85α siRNA, § P<0.001 compared to NsiRNA + lysoPC). (B) Bovine aortic ECs were transiently transfected with NsiRNA (40 nmol/L), p85β siRNA (30 nmol/L), or p55γ siRNA (30 nmol/L). PIP3 production in the presence or absence of lysoPC (12.5 μmol/L) was determined by ELISA and represented by scatter plot (n=3 independent biological samples, * P < 0.001 compared to medium, † P < 0.001 compared to NsiRNA, # P < 0.001 compared to p85β siRNA, ∞ P < 0.001 compared to p55γ siRNA). Statistical analysis for (A-B) was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test.

3.5. PI3K p85α regulatory subunit mediates lysoPC-induced TRPC6 externalization.

The effect of decreased PI3K regulatory subunit protein levels with siRNA transfection on lysoPC-induced TRPC6 externalization was determined by biotinylation assay. In the presence of lysoPC, TRPC6 externalization was significantly increased in NsiRNA transfected cells, but TRPC6 externalization was inhibited in p85α siRNA transfected cells (Fig. 7A). In contrast, transfection of p85β siRNA (Fig. 7B) or p55γ siRNA (Fig. 7C) did not alter TRPC6 externalization in the presence of lysoPC.

Fig. 7. Decreasing p85α regulatory subunit protein levels with siRNA inhibits lysoPC-induced TRPC6 externalization.

Bovine aortic ECs were transiently transfected with control siRNA (NsiRNA) (40 nmol/L), p85α siRNA (30 nmol/L), p85β siRNA (30 nmol/L), or p55γ siRNA (30 nmol/L) for 24 h before incubation with lysoPC (12.5 μmol/L) for 15 min. Externalization of TRPC6 after decreasing p85α (A), p85β (B), or p55γ (C) protein levels was detected by biotinylation assay and total TRPC6 by immunoblot analysis. Actin served as loading control (n=3 independent biological samples). Black lines indicate lanes rearranged from the same gel. All bands are from the same gel. Densitometry measurements of Biotin-TRPC6 are represented in dot-whisker plots after decreasing p85α (A, right panel), p85β (B, right panel), or p55γ (C, right panel) protein levels (n=3 independent biological samples, * P < 0.001 compared to NsiRNA, † P < 0.001 compared to NsiRNA + LysoPC, ‡ P < 0.001 compared to p85β NsiRNA + LysoPC, and § P < 0.001 compared to p55γ NsiRNA + LysoPC). Statistical analysis was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test.

3.6. PI3K p85α regulatory subunit required for lysoPC-induced calcium influx.

The effect of PI3K regulatory subunit isoform siRNA on [Ca²⁺]_i in the presence of lysoPC was evaluated. LysoPC significantly increased [Ca²⁺]_i in NsiRNA transfected cells (Fig. 8A). p85α siRNA prevents the rise in [Ca²⁺]_i in response to lysoPC compared to NsiRNA-transfected cells (Fig. 8B & 8E). In contrast, p85β and p55γ did not alter the lysoPC-induced rise in [Ca²⁺]_i compared to ECs transfected with NsiRNA (Fig. 8C-E). These results supported the important role of the p85α regulatory subunit isoforms of PI3K in lysoPC-induced Ca²⁺ influx.

Fig. 8. Decreasing p85α regulatory subunit protein levels with siRNA inhibits lysoPC-induced increase in [Ca²⁺]_i.

Bovine aortic ECs were transiently transfected with control siRNA (NsiRNA) (40 nmol/L), p85α siRNA (30 nmol/L), p85β siRNA (30 nmol/L), or p55γ siRNA (30 nmol/L) for 24 h. ECs were loaded with the FITC-conjugated fluorophore Calbryte^™ 520 AM dye. The ECs were suspended and loaded into the sort chamber of a BD FACSMelody^™ cell sorter maintained at 37°C. After adjusting the baseline, lysoPC (12.5 μmol/L) was added. Using the kinetic reading mode at Ex/Em 490/525 nm, relative changes in [Ca²⁺]_i after transfection with NsiRNA (A), p85α (B), p85β (C), or p55γ (D) siRNA were determined (Representative images of n=3 independent biologic samples). (E) Fold increase of [Ca²⁺]_i measured by difference in mean [Ca²⁺]_i at baseline and after addition of lysoPC (12.5 μmol/L) are presented as a dot whisker plot (n=3 independent biological samples, * P < 0.01 compared to NsiRNA). Statistical analysis was performed using Student’s t-test.

3.7. Decreased p85α protein levels partially preserved EC migration in the presence of lysoPC.

Based on the above results indicating the role of the p85α regulatory subunit isoform of PI3K in lysoPC-induced TRPC6 externalization, the effect of decreased p85α protein levels on EC migration in the presence of lysoPC was evaluated. Migration of control BAECs and BAECs transfected with NsiRNA, p85α siRNA, p85β siRNA, or p55γ siRNA was similar (Fig. 9A-C, Fig. S2A-C). In the presence of lysoPC, migration in control BAECs decreased by 71-72% compared to no lysoPC (Fig. 9A-C, Fig. S2A-C) and by 70-72% in BAECs transfected with NsiRNA compared to NsiRNA with no lysoPC (Fig. 9A-C, Fig. S2A-C). In BAECs transfected with p85α siRNA, however, lysoPC inhibited migration by only 44% compared to p85α siRNA and no lysoPC (Fig. 9A, Fig. S2A). In contrast, transfection of BAECs with p85β or p55γ siRNA did not impact EC migration in the presence of lysoPC (Fig. 9B-C, Fig. S2B-C). The partial preservation in BAEC migration suggested the p85α isoform was involved in the lysoPC-induced inhibition of BAEC migration.

Fig. 9. Decreasing p85α regulatory subunit protein levels with siRNA partially preserves EC migration in presence of lysoPC.

Bovine aortic ECs were transiently transfected for 24 h with control siRNA (NsiRNA) (40 nmol/L), p85α siRNA (30 nmol/L) (A), p85β siRNA (30 nmol/L) (B), or p55γ siRNA (30 nmol/L) (C) and then made quiescent for 12 h. The migration assay was initiated and migration in the presence or absence of lysoPC (12.5 μmol/L) was assessed after 24 h. (A) (Upper) Representative images are shown at 40x magnification (Scale Bar, 200 μm). Arrow indicates the starting line of EC migration. (Lower) EC migration is represented in scatter plot for p85α protein experiments (n=3 independent biological samples, * P<0.001 compared to medium, † P<0.001 compared to NsiRNA, ‡ P<0.001 compared to p85α siRNA, and § P<0.001 compared to p85α siRNA + lysoPC). (B) EC migration is represented in scatter plot for p85β protein studies (n=3 independent biological samples, * P<0.001 compared to medium, † P<0.001 compared to NsiRNA, # P<0.001 compared to p85β siRNA). (C) EC migration is represented in scatter plot for p55γ protein experiments (n=3 independent biological samples, * P<0.001 compared to medium, † P<0.001 compared to NsiRNA, ∞ P<0.001 compared to p55γ siRNA). Statistical analysis was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test.

Viability assays were performed to verify the results were not related to changes in BAEC viability. BAEC viability in the presence or absence of lysoPC was unchanged in BAECs transfected with NsiRNA or p85α siRNA (Fig. S3).

Proliferation assays were performed to verify the changes in BAEC migration were not related to changes in BAEC proliferation. BAEC proliferation in the presence or absence of lysoPC was unchanged in BAECs transfected with NsiRNA or p85α siRNA (Fig. S4).

3.8. Effect of decreased p85α protein levels on p110α membrane translocation, TRPC6 externalization, [Ca2+]_i, and EC migration in human ECs are consistent with effects seen in BAECs.

To ensure the effects of siRNA-mediated decrease in p85α protein levels on lysoPC-induced p110α membrane translocation, TRPC6 externalization, [Ca²⁺]_i, and EC migration are not species-specific to bovine ECs, the results were verified in human EA.hy926 ECs. Transient transfection of EA.hy926 ECs with p85α siRNA significantly decreased p85α protein levels (data not shown). In EA.hy926 ECs, p110α translocated from the cytosol fraction to the membrane fraction in the presence of lysoPC, similar to the results in BAECs. Decreasing p85α protein levels also inhibited lysoPC-induced p110α translocation to the plasma membrane (Fig. S5A). Similarly, decreasing p85α levels in EA.hy926 ECs inhibited the lysoPC-induced externalization of TRPC6 (Fig. S5B) and increase in [Ca²⁺]_i (Fig. S5C), and partially preserved EC migration in the presence of lysoPC (Fig. S5D), as seen in BAECs. These results indicated that the changes in PI3K catalytic subunit translocation to the membrane, TRPC6 externalization, and EC migration in response to lysoPC were not species specific and strengthen case that p85α is the PI3K regulatory subunit isoform involved in externalization of TRPC6 channels in the presence of lysoPC.

3.9. LysoPC-induced increase in [Ca²⁺]_i, p85a phosphorylation, and TRCP6 externalization not related to release of calcium from the endoplasmic reticulum.

Prior studies in our lab suggest the increase in cytoplasmic calcium concentration in response to lysoPC is associated with increased calcium influx from the extracellular space [8]. The impact of lysoPC on release of calcium from the endoplasmic reticulum (ER) and the potential role of ER calcium release on lysoPC-induced p85α activation and TRPC6 externalization has not previously been reported.

Thapsigargin, a non-competitive ER Ca2+ ATPase (SERCA) inhibitor, leads to increased cytosolic calcium concentration by preventing the cell from transferring calcium into the ER. The effect of thapsigargin and lysoPC on [Ca²⁺]_i in calcium-free buffer was evaluated. Calcium-free buffer was used to ensure any changes in [Ca²⁺]_i were associated with release of calcium from the ER as opposed to influx from the extracellular space. Thapsigargin induced increased [Ca²⁺]_i, but lysoPC did not increase [Ca²⁺]_i (Fig. 10A). These results indicate that lysoPC does not induce release of calcium from the ER in EC.

Fig. 10. LysoPC induced p85a phosphorylation and TRPC6 externalization are not mediated by release of calcium from the endoplasmic reticulum.

(A) Bovine aortic ECs were loaded with the FITC-conjugated fluorophore Calbryte^™ 520 AM dye. The ECs were suspended in calcium-free buffer and loaded into the sort chamber of a BD FACSMelody^™ cell sorter maintained at 37°C. After adjusting the baseline, thapsigargin (1 μmol/L) or lysoPC (12.5 μmol/L) was added. Using the kinetic reading mode at Ex/Em 490/525 nm, relative changes in [Ca²⁺]_i after the addition of thapsigargin (Left) or lysoPC (Middle) were determined (Representative images of n=3 independent biologic samples). (Right) Fold increase of [Ca²⁺]_i measured by difference in mean [Ca²⁺]_i at baseline and after addition of thapsigargin or lysoPC are presented as a dot whisker plot (n=3 independent biological samples, * P < 0.002 compared to thapsigargin). Statistical analysis was performed using Student’s t-test. (B) Bovine aortic ECs were incubated with thapsigargin (1 μmol/L) and/or lysoPC (12.5 μmol/L) for 15 min. p85α was immunoprecipitated and immunoblot analysis for phosphotyrosine was performed. Immunoblot analysis of total p85α was performed for control. Actin served as a loading control (n=3 independent biological samples). Black lines indicate lanes rearranged from the same gel. All bands are from the same gel. Densitometry measurements of phospho-p85α are represented in dot-whisker plots (n=3 independent biological samples, * P < 0.003 compared to medium, † P<0.004 compared to thapsigargin alone). Statistical analysis was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test. (C) Bovine aortic ECs were incubated with thapsigargin (1 μmol/L) and/or lysoPC (12.5 μmol/L) for 15 min. Externalization of TRPC6 was detected by biotinylation assay and total TRPC6 by immunoblot analysis. Actin served as loading control (n=3 independent biological samples). Black lines indicate lanes rearranged from the same gel. All bands are from the same gel. Densitometry measurements of Biotin-TRPC6 are represented in dot-whisker plots after lysoPC and/or thapsigargin treatment (n=3 independent biological samples, * P < 0.001 compared to Medium, † P < 0.001 compared to LysoPC alone, ‡ P < 0.001 compared to thapsigargin alone). Statistical analysis was performed using One-Way ANOVA with Tukey’s Multiple Comparisons Test.

The impact of ER calcium release on lysoPC-induced p85α activation was evaluated. Pretreatment with thapsigargin to deplete ER calcium levels did not alter p85α phosphorylation in the presence of lysoPC (Fig. 10B), suggesting ER calcium release is not required for lysoPC-induced p85α activation.

The impact of ER calcium release on lyso-PC induced TRPC6 externalization was evaluated. Thapsigargin has been shown to increase TRPC6 externalization in HEK cells because store-depletion can secondarily activate plasma membrane calcium channels, allowing an influx of calcium into the cytosol [22]. In BAEC, thapsigargin alone significantly increased TRPC6 externalization to a level similar to lysoPC (Fig. 10C). Pretreatment with thapsigargin followed by treatment with lysoPC lead to a significant increase in TRPC6 externalization compared to lysoPC or thapsigargin alone (Fig. 10C). These results indicate that lysoPC-induced TRPC6 externalization occurs through a different mechanism than thapsigargin-induced TRPC6 externalization. These results also suggest that release of calcium from the ER does not have a role in lysoPC-induced PI3K activation, TRPC6 externalization, and inhibition of EC migration.

4.0. DISCUSSION

Lipid oxidation products, including lysoPC, increase PI3K activity in ECs, which results in TRPC6 translocation to the membrane. LysoPC induces the phosphorylation of calmodulin at Tyr99, which leads to the association of calmodulin with a PI3K p85 regulatory subunit and subsequent translocation of PI3K to the plasma membrane and increased PIP3 production [10]. The p110α and p110δ catalytic subunit isoforms of PI3K are responsible for the lysoPC-induced increase in PIP3 production and TRPC6 externalization [13]. Here we show that the p85α isoform is involved in this process. The p85α regulatory subunit isoform, but not the p85β or the p55γ isoforms, is tyrosine phosphorylated in the presence of lysoPC. Decreasing the level of the p85α isoform protein, but not the p85β or the p55γ isoform protein, inhibits the translocation of PI3K to the cell membrane in the presence of lysoPC. PIP3 anchors TRPC6 in the membrane, where it is activated, allowing increased [Ca²⁺]_i [11, 23, 24] via increased influx from the extracellular space[8], as opposed to release from the endoplasmic reticulum. This increase in [Ca²⁺]_i leads to cytoskeletal changes that inhibit EC migration needed to restore cellular continuity after monolayer disruption [8], such as that which occurs due to an angioplasty. Downregulation of p85α does not affect basal EC migration, suggesting that PI3K, specifically the p85α regulatory subunit isoform, is not required for EC migration under normal physiological conditions. p85α is only involved in the activation of TRPC6 with the subsequent inhibition of EC migration under pathologic conditions (i.e. presence of lipid oxidation products).

Inhibition of PI3K p110α or p110δ catalytic subunits preserves EC migration in the presence of lysoPC in vitro and promotes arterial healing after injury in hypercholesterolemic animals [13, 25]. In addition to their role in EC migration, Class I PI3K isoforms also participate in a variety of intracellular signaling processes, cell growth, cell proliferation, cell survival, metabolic processes, and autophagy, endothelial cell-leukocyte interaction, angiogenesis, and endothelial progenitor cell biology [12, 21, 26]. Class II and Class III PI3K isoforms are associated with membrane trafficking [21], but this has not been reported with Class I PI3K isoforms. Although the role of PI3K in TRPC6 externalization makes PI3K a candidate for therapeutic intervention to improve EC migration and arterial healing after injury, targeting PI3K is not desirable due to its 50-100 downstream effectors and involvement in multiple intracellular signaling pathways [26]. Therefore, identification of the isoform of the regulatory subunit activated by lysoPC opens the possibility of its inhibition to prevent lysoPC-induced translocation of PI3K without affecting cellular viability.

The PI3K regulatory subunits are important for controlling the activity and stability of the catalytic subunit. In the absence of a regulatory subunit, the p110α catalytic subunit is unstable and rapidly degraded [27]. The stability of the p110δ catalytic subunit isoform has not been reported. The p110α subunit will form a dimer with a p85α or p85β regulatory subunit under basal conditions, but p110δ dimerizes specifically with the p85α subunit [28]. Interestingly, our study does not show a decrease in the p110α or p110δ protein level when the p85α regulatory subunit is decreased using siRNA. This may be a reflection of the incomplete depletion of the p85α regulatory subunit to 30% of baseline levels by siRNA. The p85 regulatory subunit is expressed at higher levels than the p110 catalytic subunit [29], and the residual p85α regulatory subunit along with the p85β and p55γ regulatory subunits may be sufficient to ensure the stability of the p110δ and p110α subunits, respectively.

PI3K regulatory subunits possess adaptor functions in cellular signaling through interactions with their N-terminal SH3 domains, which are largely independent of their role in regulating PI3K stability or activity [14]. One such adaptor function is in the regulation of the actin cytoskeleton. p85α mediates PDGF receptor-induced cytoskeletal changes and cell migration of fibroblasts by linking receptor stimulation at the cell membrane with actin dynamics [30]. p85α is also associated with endocytosis and down-regulation of membrane-bound receptor tyrosine kinases and interacts with the Rab family of proteins to regulate trafficking of the internalized receptor tyrosine kinases [29]. The role of p85α in lysoPC-induced TRPC6 externalization, calcium influx, and inhibition of EC migration appears to be related to its role in regulating PI3K location and activity, however the function of p85α regulatory subunit as an adaptor protein cannot be ruled out.

5.0. CONCLUSIONS

The current study shows that the p85α regulatory subunit regulates the translocation of the p110α and p110δ catalytic subunits to the plasma membrane. siRNA-mediated decrease in p85α protein levels not only reduce PIP3 production, it also blocks the translocation of the catalytic subunit from the cytosol to the plasma membrane in the presence of lysoPC. Identifying the specific PI3K regulatory subunit isoform involved in lysoPC-induced TRPC6 externalization provides the opportunity to target the specific isoform for therapeutic intervention. Pan-class I PI3K catalytic subunit inhibitors and isoform-specific catalytic subunit inhibitors of p110α and p110δ are FDA approved or currently in late-stage clinical trials for various solid tumor cancers, lymphoma, and leukemia [31], but no PI3K inhibitors target the regulatory subunit. Targeting p85α directly, the mechanism responsible for activation of p85α, or the interaction between p85α and calmodulin [10] provides novel therapeutic approaches to promote arterial healing after vascular intervention, such as angioplasty or implantation of vascular bypass grafts.

Highlights:

• Oxidized lipids selectively activate the PI3K p85α regulatory subunit isoform

• p85α isoform selectively controls lysoPC-induced PI3K membrane translocation

• p85α-mediated PI3K activation by lysoPC induces TRPC6 translocation and activation

• p85α siRNA blocks lysoPC-induced Ca2+ influx via TRPC6 and improves EC migration

• p85α is a novel target to promote EC migration and arterial healing after injury