Substrate stiffening promotes VEGF-A functions via the PI3K/Akt/mTOR pathway

Abstract

While it is now well-established that substrate stiffness regulates vascular endothelial growth factor-A (VEGF-A) mediated signaling and functions, causal mechanisms remain poorly understood. Here, we report an underlying role for the PI3K/Akt/mTOR signaling pathway. This pathway is activated on stiffer substrates, is amplified by VEGF-A stimulation, and correlates with enhanced endothelial cell (EC) proliferation, contraction, pro-angiogenic secretion, and capillary-like tube formation. In the settings of advanced age-related macular degeneration, characterized by EC and retinal pigment epithelial (RPE)-mediated angiogenesis, these data implicate substrate stiffness as a novel causative mechanism and Akt/mTOR inhibition as a novel therapeutic pathway.

Introduction

During growth, development, and disease, new blood vessels arise from existing vessels through a process called angiogenesis. Angiogenesis is initiated and sustained by several cytokines and growth factors, especially, the vascular permeability factor also known as the vascular endothelial growth factor-A (VEGF-A) [1,2]. VEGF-A binds with two receptor tyrosine kinases on endothelial cells (ECs), VEGF receptor (VEGFR)-1 and VEGFR-2, to trigger a cascade of molecular signaling events that culminates in enhanced EC proliferation, paracellular disruption, migration, and ultimately, new vessel formation. Several aspects of this functional cascade have recently been linked to stiffness of the substrate upon which the ECs are adherent.

These data have shown that substrate stiffness, independent of matrix composition or density, modulates binding, internalization, trafficking, and secretion of VEGF in the EC [3–5]. Substrate stiffness also modulates EC adhesion [6], cytoskeletal assembly [7], deformation [8], proliferation [9], barrier function [10], collective migration [11,12], and capillary-like network formation [13–17]. In mediating these changes, EC-generated contractile forces also known as traction forces, play a pivotal role (e.g. [8,10,18,19]). However, no specific molecule has yet been identified that links VEGF-A stimulation, substrate stiffness sensing, and EC traction force generation.

Here, we examined an underlying role for PI3K/Akt/mTOR signaling. This pathway is activated in the EC by VEGF-A stimulation [20], and its inhibition blocks early- and mid-stages of VEGF-A driven angiogenesis [1,21]. PI3K/Akt/mTOR signaling also modulates EC adhesion, proliferation, migration, metabolism, invasion, and survival [22,23]. While substrate stiffness dependent activation of PI3K/Akt signaling has been reported in glioma cells [24], implications to the EC, and its relationship to VEGF-A mediated functions, remains unknown.

Our primary finding is that PI3K/Akt/mTOR signaling in the EC is substrate stiffness dependent. Our secondary finding is that this mechano-sensation regulates VEGF-A mediated angiogenic changes. We implicate these findings in the pathogenesis of advanced age-related macular degeneration (AMD).

Materials and Methods

Cell Isolation and Culture:

Human dermal microvascular blood endothelial cells (HDMECs) were isolated from the neonatal dermal foreskin as previously described [25]. Briefly, the foreskin was mechanically minced and subsequently digested enzymatically with commercially available collagenase and dispase. HDMECs in cell suspension were then labeled with antibodies against specific EC surface antigens and separated from the remainder of the cells by capture with magnetic beads [26,27]. HDMECs were expanded and maintained in collagen-coated plastic dishes with MCDB media (Cellgro) containing 5% Micro Vascular Growth Supplement (MVGS, GIBCO) and Penicillin-Streptomycin (100 mg/ml each), at 37°C in 5% CO₂. The cells were grown till 80–90% confluence before passaging and were used up to passage number 6. Human foreskin use was approved by the Institutional Review Board for the Beth Israel Deaconess Medical Center. The purified human blood endothelial cells were generously provided by the laboratory of Prof. Harold F. Dvorak. The human RPE cell line, ARPE-19, was obtained from ATCC (CRL-2302) and cultured in DMEM/F12 medium containing 10% FBS (Atlanta Biologicals) 1% Glutamax (Lonza) and 1% penicillin-streptomycin (Lonza) at 37°C, 10% CO₂. For low serum conditions, the medium was replaced with DMEM/F12 plus 1% FBS.

Substrate preparation:

Soft elastic substrates (Youngs’ Modulus ranging from 0.4 to 73kPa) were prepared using silicone elastomers [28]. Briefly, a compliant silicone elastomer (NuSil® 8100, NuSil Silicone Technologies, Carpinteria, CA) was mixed at a ratio of 1:1 as per manufacturer instructions. To this mixture, we added additional crosslinker (Sylgard 184 curing agent, Dow Corning, Midland, MI) to create gels of desired final stiffness. The final mixture was poured into 35mm cell culture dishes and then baked at 60°C for 3 days to produce an elastic gel approximately 2mm thick. The gel surfaces were activated using the crosslinker Sulfo-SANPAH (N-Sulfosuccinimidyl-6-[4’-azido-2’-nitrophenylamino] hexanoate) (Proteochem, Hurricane, UT). Briefly, Sulfo-SANPAH was dissolved in 0.1M HEPES buffer at a final concentration of 0.4mM, added to wells, and then exposed to UV light (Wavelength=254nm, Power=40W, Philips, USA) for 6 minutes. The entire Sulfo-SANPAH process was then repeated to provide two treatments, ensuring adequate functionalization. The wells were then washed with 0.1M HEPES and incubated overnight with 0.05 mg/ml of type 1 collagen at 4°C (5005; Advanced Biomatrix, Carlsbad, CA). After incubation, the wells were washed twice with PBS and plated with cells in serum-containing medium at full confluence (150,000 cells/cm² for HDMECs and 170,000 cells/cm² for ARPE-19).

Measurement of pro-angiogenic cytokine secretion – ELISA:

HDMECs or ARPE-19 cells were cultured to confluence in medium containing 1% serum upon collagen-coated elastic substrates (see a previous section for substrate preparation). Culture supernatant were collected after 24 hours of cell plating. Standard procedures entailed by the human pro-angiogenesis ELISA kit (Signosis) was used to measure pro-angiogenic cytokine secretion.

Measurement of cell proliferation:

HDMECs or ARPE-19 cells were cultured to confluence at a density of 25,000 cells/well on collagen-coated elastic substrates prepared in a 96-well plate format (see a previous section for substrate preparation). After 24 hours of plating, the wells were stimulated with 50ng/ml of human VEGF 165 (R&D systems), together with matched controls, and incubated with Bromodeoxyuridine (BrdU) solution at a final working concentration of 1X at 37°C for 36 hours. Medium was removed, and after denaturation of cells, the antibody solution (1X) was added to each well and plate was incubated at room temperature for 1 hour. After 3 washes, 1X Horseradish peroxidase (HRP)-conjugated secondary antibody solution was added and again incubated at room temperature for 30 min. Solution was removed followed by 3 washes with 1X Wash Buffer. Tetra-methybenzidine (TMB) substrate (100ul) was added for 30 minutes at room temperature followed by a sulfuric acid stop solution (100ul) and kept for 5 minutes. The dye absorbance was measured at 450 nm to indicate the extent of cell proliferation.

Measurement of endothelial tube formation:

HDMVECs – Elastic substrates were prepared in a 24-well tissue culture dish to create approximately 2mm thick substrates (see ‘substrate preparation‘). The substrate was then coated with 180 ul of growth factor reduced Matrigel (BD Biosciences, USA). Approximately 100×10³ endothelial cells were added on it as a drop and visualized for their ability to assemble into tubes. Phase-contrast imaging of cells was performed 8 hours after plating using a Leica inverted microscope equipped with a camera (Leica DFC 350 FX, USA). Tube area, branching points and overall tube length were measured using the Wimasis software [29].

RPE cells –

Culture supernatant was collected from ARPE-19 cells after 48 hours of plating and used as conditioned media (CM) in a standard endothelial tube formation assay. Briefly, each well of a 24-well tissue culture dish (without additional elastic substrate) was coated with 180 ul of growth factor reduced Matrigel (BD Biosciences, USA). Approximately 100×10³ HDMVECs were added upon it as a drop, suspended in CM, and visualized for their ability to assemble into tubes. Phase-contrast imaging of cells was performed 8 hours after plating using a Leica inverted microscope equipped with a camera (Leica DFC 350 FX, USA). Tube area, branching points and overall tube length were measured using the Wimasis software [29].

Measurement of gene and protein expression – Western Blotting and Antibodies:

Cell lysates from HDMECs were prepared in Cell signaling lysis buffer supplemented with cocktail of protease and phosphatase inhibitors (Cell Signaling Technology) diluted to 1X in lysis buffer (Cell Signaling Technology). Protein concentration was measured in cell lysates, and equal amounts of protein with SDS loading buffer were heated at boiling temperature for 5 minutes. Then, proteins in denatured form were loaded on 12% Tris-Glycine Polyacrylamide SDS gel followed by transfer on nitrocellulose membrane (Invitrogen). After blocking with 5% non-fat dry milk (Bio-Rad) in tris buffer saline (TBS) for 2 hours at room temperature, the membrane was washed with TBS + 0.1% Tween (TBST) and incubated with primary antibodies overnight at 4°C. Primary antibodies (Cell Signaling Technology) for ph-Akt (S473), ph-mTOR (S2448), ph-S6 Ribosomal protein, ph-4EBP1 were used at 1:1000 dilution. For total Akt, mTOR, S6 Ribosomal protein, 4EBP1, and GAPDH, antibodies were used at 1:2000 dilution. Following primary antibody incubation, membranes were washed 3 times with TBST, and incubated for an additional 90 min at room temperature with HRP-linked anti-rabbit or anti-mouse IgG (Cell Signaling Technology) at 1:2000 dilution. This was followed by three additional washes with TBST. Protein levels were detected through chemiluminescence using a Western blot detection system (ThermoFisher Scientific, Inc.).

Measurement of EC contractile force:

Fourier transform traction microscopy was implemented in a 96-well format using the methodology of contractile force screening [28]. Briefly, the compliant silicone elastomer (NuSil® 8100, NuSil Silicone Technologies, Carpinteria, CA, Youngs Modulus = 1.9kPa) was prepared by spincoating on custom cut glass slides (109.6 mm × 78mm × 1mm, Hausser Scientific, Horsham PA), and baked at 100°C for 2 hours, to yield a final gel that was approximately 100μm thick. A second layer of gel with pre-mixed fluorescent bead markers was spun coat on the first substrate and then baked for an additional 2 hours to create a bead layer that was approximately 1μm thick. Finally, a 96-well insert (2572; Corning, Tewksbury, MA) was bonded to the gel surface to subdivide the gel into many independent wells. The wells were then ligated and prepared for cell culture as described in a previous section.

The 96-well plate was placed in a heated chamber (37°C) on a computer-controlled motorized stage of an inverted microscope (DMI 6000B, Leica Inc., Germany) and imaged at 10x magnification using a monochrome camera (Leica DFC365 FX). The images comprised of phase imaging of cells and fluorescent imaging of bead markers embedded in the gel substrate directly underneath the cells. Such image pairs were obtained at (i) baseline (0 minutes), (ii) after treatment (15, 30, and 60 minutes), and (iii) after detachment of cells with trypsin (reference). By comparing the position of bead markers at reference with those at baseline, and, at reference and with those after treatment, we obtained baseline and treatment maps of bead displacement (spatial resolution = ~15μm). Using these maps together with the pre-defined substrate modulus (1.9kPa), and thickness (100 μm), we calculated baseline and treatment maps of contractile forces and the root-mean squared value, on a well-by-well basis, using the well-established approach of Fourier-transform traction cytometry [30] adapted to cell monolayers [31]. The measurements were performed in parallel over the 96-well plate, with each run requiring ~15 minutes. Every drug treatment condition was tested in at least n=8–10 separate wells of the 96 well plate. The forces were normalized on a well-by-well basis as a % change relative to baseline, and then pooled from separate wells corresponding to a given drug treatment condition. The drug conditions tested were vehicle, 100ng/ml VEGF 165, 50nM Rapamycin, and 10μM Y27632.

Statistics:

All statistical comparisons were performed using a standard student t-test. Differences were considered significant for p<0.05.

Results and Discussion

Substrate stiffening promotes pro-angiogenic changes in the EC:

Substrate stiffness enhanced EC proliferation (Figure 1A), and VEGF-A secretion (Figure 1B). While a relationship between substrate stiffness and EC proliferation has been reported previously [5,9], here we report a novel amplifying effect for VEGF-A co-stimulation that is operative even in the case of confluent endothelial monolayers.

Figure 1. Substrate stiffening and VEGF co-operatively promote EC angiogenesis.

A) Compared to 0.4kPa stiff substrates, ECs plated on 1.9kPa or 73.3kPa stiff substrates proliferate more. Their proliferation is further enhanced by stimulation with VEGF165. B) Compared to 0.4kPa stiff substrates, ECs plated on a 73.3kPa stiff substrate secrete more VEGF-A. For all measurements, data were pooled from n=3–4 wells per condition and reported as mean and standard error; ‘ns’ indicates not significant, *** indicates p<0.001, and **** indicates p<0.0001. C) Substrate stiffening enhanced capillary-like tube formation. This enhancement was amplified by co-stimulation with VEGF165.

An in vitro correlate of angiogenesis is the spontaneous organization of ECs into capillary-like tubes. Robust tube formation was induced in ECs cultured upon Matrigel-coated stiff substrates; in comparison tubes were fewer in both number and density on the softest substrate (Figure 1C). These differences were still maintained in the combined presence of exogenous VEGF165.

Substrate stiffening activates PI3K/Akt/mTOR signaling:

An unbiased protein screen in ECs revealed that substrate stiffening activates PI3K/Akt/mTOR signaling. Specifically, it activates phosphorylation of Akt, S6, mTOR and 4EBP1 (eukaryotic initiation factor 4E binding protein 1) (Figure 2A and Supplementary Figure 1). Each of these pathways is further amplified by VEGF-A stimulation (Figure 2B and Supplementary Figure 2).

Figure 2. Substrate stiffening promotes PI3K/Akt/mTOR signaling.

A) Phosphorylation of Akt and other mTOR-targets were enhanced by substrate stiffening in HDMECs. B) These targets are further enhanced by VEGF. C) Phosphorylation of the proline-rich Akt substrate of 40 kDa (PRAS40) and mTOR were enhanced by substrate stiffening in RPEs. For each measurement, cell lysates were pooled from at least n=3 wells.

mTOR inhibition reduces EC traction forces:

To sense and respond to substrate stiffening, ECs exert traction forces upon their substrate [10]. As expected, these forces were substantial and spatially-heterogeneous upon stiff substrates (1.9kPa) (Figure 3) [18]. While VEGF-A further enhanced the forces, as has been reported previously [32], mTOR inhibition using rapamycin ablated them, commensurate to the well-known relaxant, the rho-kinase inhibitor, Y27632. To our knowledge, this is the first direct elucidation of a cell mechanical impact of mTOR signaling in the EC. Because traction forces guide EC angiogenesis [24], its inhibition by rapamycin suggests a novel and additional therapeutic benefit to Akt/mTOR inhibition.

Figure 3. mTOR inhibition reduces EC traction forces:

Fourier transform traction microscopy was implemented using a soft elastic substrate (Youngs Modulus = 1.9kPa) prepared in a 96-well format. A-C) Representative maps and average value (inset) from three independent wells after 30 minutes of treatment. Bar = 80μm. D) While VEGF enhanced EC tractions, mTOR and rho-kinase inhibition ablated them. Traction values were normalized relative to baseline, on a well-by-well basis, and then pooled according to treatment. Results are plotted as mean and standard error over at least n=8–10 separate wells per treatment; **,**** indicates p<0.01, and 0.0001, respectively.

Implications:

To assess further the generality of these findings, we examined the retinal pigment epithelium (RPE). We discovered that not only do stiffer substrates activate PI3K/Akt/mTOR signaling (Figure 2C and Supplementary Figure 3) but they also induce pro-angiogenic cytokine secretion (Figure 4A). Furthermore, when RPE-derived culture supernatant from stiffer substrates is used in a standard Matrigel assay, it promotes robust EC tube formation (Figure 4B). Taken together, these data implicate an altogether new factor, substrate stiffness, and a potential mechanism, Akt/mTOR signaling, to explain RPE pro-angiogenic changes as is observed in advanced age-related macular degeneration (AMD) [33].

Figure 4. Substrate stiffening promotes pro-angiogenic changes in the RPE.

A) Compared to 0.4kPa stiff substrates, RPE cells plated on 1.9kPa stiff substrates secreted more VEGF-A, IL6, and EGF while RPE cells plated on 73.3kPa stiff substrates secreted more IL6 and EGF. B) Compared to RPE supernatants from the 0.4kPa stiff substrate, RPE supernatants obtained from the 1.9kPa stiff substrate induced greater and denser endothelial tubes. For each experiment, data were pooled from n=4 wells per condition and reported as mean and standard error; *,**,*** indicates p<0.05, 0.01, and 0.001, respectively.

Our choice of RPE substrate stiffness (in the kPa range) is based on direct previous measurements of explanted Bruch’s membranes from young and old eyes [34]. Other measurements that include AMD eyes have reported integrated stiffness values of the entire Bruch’s membrane (BM)-choroid complex [35] which are not to be confused with the local (i.e. functional) stiffness of the BM upon with the RPE is adherent. A particularly important future experiment is to directly measure age- and disease-related changes in local BM stiffness.

Although our approaches are generalizable, we considered a commonly reported range of in vivo substrate stiffness values, the widely used extracellular matrix protein, collagen 1, and a routinely studied endothelial (HDMVECs), and retinal epithelial cell (ARPE-19). Despite the limitations of this model system, our data show a potent role for substrate stiffness in VEGF-mediated angiogenesis and suggest reconsideration of current methodologies that use plastic or glass substrates whose stiffness is several orders of magnitude greater that the physiological range.

We propose two important directions for future investigation. First, it remains to be determined how substrate stiffness, VEGF-A, and PI3K/Akt/mTOR signaling might co-regulate EC migration, sprouting, activation, and extracellular matrix production and degradation. Second, it would be of interest to study PI3K/Akt/mTOR signaling in the context of known VEGF-dependent EC mechanotransduction pathways including the RhoGTPase Cdc42 [15] and the Rho inhibitor p190RhoGAP [4], as well as the Extracellular signal-regulated kinase (ERK) pathway, since ERK is a key regulatory pathway of EC proliferation [36], and has been implicated is both VEGF-A and PI3K/ Akt mediated signaling [37]. However, we note that when endothelial cells are studied in confluence, as is the case here, VEGF-induced ERK 1/2 phosphorylation is unaffected by substrate stiffening [5].

Conclusion:

By studying EC and RPE monolayers plated over a range of substrate stiffness gels, we elucidated underlying changes in PI3K/Akt/mTOR signaling and discovered that activation of this pathway by stiffer substrates correlated with enhanced pro-angiogenic secretion, proliferation, and induction of capillary-like tubes. In advanced AMD that is characterized by substrate stiffening, VEGF-A stimulation, and choroidal neovascularization, Akt/mTOR inhibition emerges as a key therapeutic hub.

Supplementary Material

1

Supplementary Figure 1. Densitometry analysis of western blots for Ph-MTOR, Ph-AKT-S473, Ph-AKT-T308, Ph-S6, Ph-4EBP1, and total levels, normalized to GAPDH protein.

Supplementary Figure 2. Densitometry analysis of western blots for Ph-AKT and Ph-S6, with and without VEGF, normalized to GAPDH protein.

Supplementary Figure 3. Densitometry analysis of Immunoprecipitated PRAS40 with mTOR, and total levels, normalized to total PRAS40 levels in cell lysates from soft and stiff substrates.

Highlights.

• PI3K/Akt/mTOR signaling in the endothelial cell and the retinal pigmented epithelial cell is substrate stiffness dependent.

• This mechano-sensation regulates VEGF-A mediated angiogenic changes.

• We implicate these findings in the pathogenesis of advanced age-related macular degeneration