Acute Mechanical Stress in Primary RPE Cells Induces Angiogenic Factor Expression and in vitro Angiogenesis

Background Choroidal neovascularization (CNV) is a major cause of blindness in patients with age-related macular degeneration. CNV is characterized by new blood vessel growth and subretinal uid accumulation, which results in mechanical pressure on retinal pigment epithelial (RPE) cells. Overexpressing RPE-derived angiogenic factors plays an important role in inducing CNV. Results The goal of this study was to determine whether low levels of acute mechanical stress during early CNV can induce the expression of angiogenic factors in RPE cells and accelerate angiogenesis. Using a novel device, acute mechanical stress was applied to primary porcine RPE cells and the resulting changes in the expression of major angiogenic factors were examined using immunocytochemistry, qRT-PCR, and ELISA. An in vitro tube formation assay was used to determine the effect of mechanical stress on RPE cells on angiogenesis. Our results showed an increase in the expression of major angiogenic factors in response to mechanical stress, resulting in increased in vitro angiogenesis. Abnormal epithelial-mesenchymal transition (EMT) in RPE cells is also associated with CNV and further retinal degeneration. Our qRT-PCR results veried an increase in the expression of EMT genes in RPE cells. Conclusions In conclusion, we showed that acute mechanical stress induces the expression of major angiogenic and EMT factors and promotes in vitro angiogenesis, suggesting that mechanical stress plays a role in promoting aberrant angiogenesis in AMD. Fisher Scientic) was used to obtain cDNA from 1 µg of isolated RNA. qRT-PCR was performed using PowerUp™ SYBR Green Master Mix (Thermo Fisher Scientic) in an Eppendorf RealPlex4 real-time mastercycler (Hamburg, Germany). Data were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) values and fold change expression was calculated using the 2 -ΔΔCt method. disrupted (c, h, m) due to increased mechanical stress. Z-stack images conrm the deformation of the RPE monolayer after adding mechanical stress (f, k, p). qRT-PCR results show increased expression of VEGF isoforms, VEGF121 and VEGF165, HIF-1α, ANG2, IL-6, IL-8, TNF-α and the antiangiogenic factor, PEDF (q). An increase in the expression of EMT genes, VIM and CDH2 and brosis gene, FN1, and a decrease in the expression of RPE-specic RPE65 was also observed (r). ELISA results showed increased apical expression of IL-6 and basal expression of VEGF, ANG2, IL-6 and IL-8 (s, t). * p<0.05 ** p<0.01. Error bars represent one standard deviation.


Background
The choroid is a vascular layer underneath the retinal pigment epithelium (RPE) that supplies blood to the RPE and the retina. The abnormal overgrowth of choroidal blood vessels creates a condition called choroidal neovascularization (CNV). In age-related macular degeneration (AMD), CMV damages the overlying RPE and photoreceptors, resulting in irreversible blindness. The etiology of CNV remains to be fully outlined, but RPE cells do produce higher levels of angiogenic proteins in response to mechanical stress, promoting angiogenesis and contributing to CNV development [1,2]. Choroidal blood vessel invasion and sub-RPE uid accumulation are potential sources of mechanical stress during AMD. As new blood vessels form, creating spatial crowding and hemorrhages, RPE cells elongate from < 10% to ~ 60%, uniaxially [3][4][5]. However, little is known about the resulting angiogenic factor expression in RPE cells.
In addition, we used nite element analysis and immunocytochemistry to nd correlations between strain levels and increased expression of VEGF, IL-6 and IL-8. We also assessed the angiogenic potential of the stress-induced RPE secretome using an in vitro angiogenesis assay. Our results showed that mechanical stress during early CNV induces the expression of angiogenic and EMT factors, further promoting in vitro angiogenesis.
Increased strain leads to increased VEGF, IL-6, and IL-8 expression Finite element analysis showed that the mechanically stressed area's center experiences the highest strain level (Fig. 3a). These results were compared to the ICC results to determine whether strain levels and increased angiogenic protein expression are correlated.
Confocal images of mechanically-stressed RPE cells showed that in non-stressed cells, F-actin localized to cell-cell junctions (Fig. 2b), while in mechanically stressed cells, the actin laments distributed diffusely in the cytoplasm (Fig. 3c, h, m). ICC results also showed that the mechanical stress-induced disruption of the F-actin cytoskeleton was associated with increases in VEGF (Fig. 3b-e), IL-6 ( Fig. 3g-j), and IL-8 ( Fig. 3l-o) expression. The RPE monolayer also began to deform after stretching the Transwell membranes, based on z-stack images and (Fig. 3f, k, p), as predicted by FEA (Fig. 3a).
qRT-PCR results showed a concurrent increase in pro-angiogenic gene expression, as mentioned above, as well as the anti-angiogenic factor, PEDF (Fig. 3q). To determine whether the increase of pro-angiogenic factors overrides the anti-angiogenic activity of PEDF, an in vitro angiogenesis assay was performed.
Used media from RPE cultures activated endothelial tube formation, while limited angiogenic activity was observed in HUVECs cultured in fresh RPE media (Fig. 4). Exposure to apical and basal conditioned media from mechanically stressed RPE cells resulted in a signi cant ~ 2-and 1.4-fold increase in endothelial tube length (p = 0.001, 0.017, respectively) and ~ 3.6-and 2-fold increase in node number respectively (p = 0.003, 0.018, respectively; Fig. 4f, g).

Discussion
While the mechanisms leading to increased expression of RPE-derived angiogenic factors and the resulting angiogenesis during CNV in AMD are not entirely clear, physical changes in the RPE contribute to the elevated expression of angiogenic factors [1,18,19]. To better understand the role that mechanical stress plays in AMD pathogenesis, we investigated the effect of mechanical stress on the expression of angiogenic and EMT factors and in vitro angiogenesis using a novel in vitro device (Fig. 1).

First, we produced a realistic in vitro model of the RPE by growing freshly isolated porcine RPE cells on
Transwell membranes, which demonstrated several characteristics similar to native RPE, such as junctional localization of β-catenin, F-actin, and ZO-1, expression of the RPE-speci c RPE65, and high TEER in 4-week-old RPE cultures (Fig. 2), indicating that the substrate supports proper maturation of RPE cells. Next, using a novel device, mechanical stress was added to the RPE monolayer to determine changes in gene and protein expression of major angiogenic and EMT factors. Lastly, an in vitro angiogenesis assay was performed to determine if pro-angiogenic factor production outweighed antiangiogenic factor production and induced angiogenesis.
Our results showed elevated mRNA expression of angiogenic factors, VEGF, HIF-1α, IL-6, IL-8, ANG2, and anti-angiogenic PEDF (Fig. 3q), and increased protein expression of VEGF, IL-6, IL-8, and ANG2 in mechanically stressed RPE cells (Fig. 3b-p, s, t). ICC results also revealed a remarkable disruption of the actin cytoskeleton in RPE sites with higher VEGF, IL-6 and IL-8 expression. The stressed areas of the RPE showed diffusely distributed actin bers, while in the non-stressed areas, actin was localized in cell-cell junctions (insets in Fig. 3c, h, m). According to the FEA (Fig. 3a), in these areas of impact, strain distribution pattern correlated with the disruption of actin structure and increased VEGF, IL-6, and IL-8 expression, supporting the hypothesis that the increase in mechanical stress is responsible for elevated angiogenic factor expression.
The increase in IL-8 and VEGF expression (Fig. 3q, s, t) was consistent with previous reports of IL-8 and VEGF induction in human RPE cells after RPE injury [20,21]. It has previously been shown that actin polymerization activation induces IL-8 expression and different VEGF isoforms [22,23]. It is therefore possible that disrupting the actin cytoskeleton activates actin polymerization, leading to the increased expression of IL-8 and different VEGF isoforms. Unlike VEGF121 and VEGF165, VEGF189 was not sensitive to the mechanical stress levels used in our experiments (Fig. 3q). The two soluble VEGF isoforms, VEGF121 and VEGF165, are regulated by low frequency stress while insoluble VEGF189 is more susceptible to high frequency mechanical stress [23]. The single pulse of mechanical stress used in our experiment may be lower than required for VEGF189 mRNA overexpression. Both soluble VEGF isoforms, VEGF121 and VEGF165, have been implicated in in vitro and in vivo neovascularization [24][25][26][27]. Hence, their overexpression induced by mechanical stress may contribute to CNV development.
We also observed an increase in the expression of HIF-1 α in response to mechanical stress (Fig. 3q). HIF-1α is an inducer of VEGF, IL-6 ,and IL-8 under hypoxic conditions [28,29], and its mRNA expression was activated by mechanical stress, suggesting that the mechanisms of mechanical stress-induced VEGF, IL-6, and IL-8 overexpression may be similar to hypoxic conditions. While HIF-1α does not directly regulate ANG2, previous studies have shown that HIF-1α-induced VEGF is a potential activator of ANG2 [30], which may explain the increased ANG2 expression (Fig. 3q, t) upon activation of HIF-1α and VEGF expression due to mechanical stress.
Our gene expression results also showed an increase in the expression of TNF-α in response to mechanical stress (Fig. 3q). However, ELISA results showed undetectably low levels of TNF-α in RPE supernatants before or after applying mechanical stress, suggesting that mechanical stress alone does not activate TNF-α protein expression in RPE cells.
The qRT-PCR results showed that mechanical stress promoted an EMT-like phenotype in RPE cells, as demonstrated by an increase in the expression of mesenchymal markers, CDH2, VIM, and FN1, and a decrease in the expression of the RPE-speci c RPE65 (Fig. 3r). Previous studies have shown that VEGF, IL-6 and IL-8 can trigger EMT [15][16][17]. Our experiments demonstrated that mechanical stress induced all three of these cytokines, con rming the hypothesis that mechanical stress may induce EMT in the RPE through the induction of EMT promoters during stages of CNV development.
The expression of PEDF, a major anti-angiogenic gene, also increased due to mechanical stress (Fig. 3q). Previous work has shown that the balance between VEGF and PEDF must change for choroidal angiogenesis initiation [31], The in vitro angiogenesis assay determined if pro-angiogenic protein levels outweigh anti-angiogenic factors and promote angiogenesis. The increased angiogenic response of the HUVECs to the mechanically stressed RPE secretome (Fig. 4) suggests that the increase in the PEDF expression was either not su cient to neutralize the angiogenic activity of the overexpressed angiogenic factors or did not lead to signi cant protein secretion. However, the higher increase in the angiogenic potential of the apical supernatant compared to that of the basal supernatant (Fig. 4) could be due to the elevated PEDF expression, as the expression of PEDF is mainly basolateral. Further protein expression analyses are needed to better understand the dynamics of the VEGF/PEDF angiogenic switch under mechanical stress. The choroid is adjacent to the basal side of the RPE and our experiments showed that media from both sides of mechanically-stressed RPE cultures increased in vitro angiogenesis (Fig. 4), suggesting that mechanical stress promotes angiogenesis by inducing the overexpression of angiogenic factors.

Conclusions
In this work, we present a novel method of introducing and understanding mechanical changes in the RPE during early stages of CNV development and reported, for the rst time, that acute mechanical stress induces the expression of angiogenic and EMT factors. In vitro angiogenesis results con rmed the main hypothesis that mechanical stress in RPE cells can induce angiogenesis. This result suggests that mechanical stretching of the RPE accelerates angiogenesis during CNV.

RPE isolation and culture
RPE cells were isolated from locally-sourced pig eyeballs using a previously described method [32]. Isolated cells (passage 0) were cultured on Dulbecco's modi cation of Eagle medium (DMEM) 1x (Corning, Manassas, VA) supplemented with 10% premium grade fetal bovine serum (FBS; VWR, Radnor, PA) on 2.4 mm Transwell inserts (pore size: 0.4 µm, Corning) and incubated at 37°C, 5% CO 2 in a humidi ed incubator until the inserts were con uent (approximately 3-5 days). After reaching con uency, FBS concentration was dropped to 1% and cells were grown on Transwell inserts for 4 weeks to promote RPE differentiation before the experiments.
Transepithelial electrical resistance (TEER) TEER was performed using an EVOM2 voltohmmeter (World Precision Instrument, Sarasota, FL) connected to an ENDOHM-24SNAP measurement chamber (World Precision Instrument). TEER of RPE monolayers was measured after 1 week and 4 weeks of culture on Transwell membranes.

Mechanical stress
We have previously shown that the plastic properties of the porous polyester Transwell membranes can be permanently stretched to convey mechanical stress to adherent cells [2]. In this work, we fabricated a device to convey controlled and localized stress to RPE monolayers instead. The tip of a ballpoint pen was attached to an in-house reciprocating motion mechanism. A Transwell membrane was placed upside down on a custom-made stage (a laser-cut acrylic sheet) directly under the pen tip and rotated as the reciprocating motion was engaged to simulate 60 non-overlapping impacts (mechanical stresses) on the membrane over 2 minutes (Fig. 1). The height of the stage was adjusted so that the tip's pressure created 10% strain, mimicking low levels of strain on the RPE during early stages of CNV.

Finite element analysis
Finite element analysis (FEA) was performed using Ansys V19.1 to evaluate strain distribution on Transwell membranes due to mechanical stress. Plastic strain in the membranes was simulated by pushing a 3D model of the pen tip (500 µm diameter) on a 10 µm-thick polyester membrane (same as Transwell membranes) to create 10% strain.

RNA isolation and qRT-PCR
Three and six hours after adding mechanical stress, RPE cells from 8 independent cultures (4 unstressed and 4 mechanically stressed) were lysed directly on the membranes and RNA was isolated using an Illustra RNAspin Mini RNA Isolation kit (GE Healthcare, Chicago, IL). A High Capacity RNA-to-cDNA kit (Thermo Fisher Scienti c) was used to obtain cDNA from 1 µg of isolated RNA. qRT-PCR was performed using PowerUp™ SYBR Green Master Mix (Thermo Fisher Scienti c) in an Eppendorf RealPlex4 real-time mastercycler (Hamburg, Germany). Data were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) values and fold change expression was calculated using the 2 -ΔΔCt method.
Enzyme-linked immunosorbent assay 50 µL samples of spent apical and basal media from control and mechanically stressed RPE cultures were collected after 24 hours of applying mechanical stress. The expression of VEGF, ANG2, FGF2, IL-6, IL-8 and TNF-α was tested using a multiplex human enzyme-linked immunosorbent assay (ELISA) kit according to manufacturer's instructions (Quansys Biosciences, Logan, UT).

Tube formation assay
Tube formation assays were performed using an in vitro angiogenesis kit (Gibco) according to manufacturer's instructions. Brie y, the wells of a 48-well plate were coated with 100 µL of reduced growth factor Geltrex matrix (Gibco) and incubated at 37˚C for 30 minutes. Human umbilical vein endothelial cells (HUVECs) were diluted in spent media from unstressed or mechanically stressed RPE cultures to a concentration of 10 6 cells/mL. 200 µL of cell suspensions were seeded on Geltrex matrices and incubated for 6 hours at 37˚C in a humidi ed incubator with 5% CO 2 to induce endothelial tube formation. Next, HUVECs were stained with Calcein AM dye (Thermo Fisher Scienti c) and imaged using an Eclipse TS100 uorescence microscope (Nikon Instrument Inc., Melville, NY). Tube length and node numbers were quanti ed using ImageJ software.

Statistical analysis
The data are presented as the mean ± standard deviation (SD) of at least three independent experiments.
Comparisons between two groups were analyzed using two-tailed Student's t-test and p-values were adjusted using the Benjamini-Hochberg method. P < 0.05 was considered statistically signi cant.

Declarations
Ethics approval and consent to participate N/A Consent for publication N/A

Availability of data and materials
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.
FF designed the study, collected data, performed analyses, and wrote the manuscript. AA and SO collected data, performed analyses, and wrote the manuscript. EV provided equipment and funding, supervised the project and edited and approved the manuscript for publication. Figure 1 Schematic of the mechanical stress device. RPE cells were grown on porous membranes of Transwell inserts (a). After 4 weeks, Transwells were placed upside down on a custom-made stage under a pen tip and controlled with a rotor (b). The pen tip was pushed against the Transwell membrane to permanently stretch focal regions of the membrane (c). The Transwell insert was rotated between impacts to produce 60 non-overlapping bumps (d, e)   In vitro angiogenesis results. Mechanical stress increased the endothelial tube formation response of HUVECs to used media from RPE cultures. The length of the endothelial tubes and the number of nodes increased when HUVECs were exposed to conditioned apical (a, d) and basal (b, e) media from mechanically stressed RPE cultures for 6 hours (f, g). HUVECs incubated with fresh medium resulted in smaller endothelial tubes and fewer nodes compared to those grown with media from both unstressed and mechanically stressed RPE cultures (c, f, g). * p<0.05; ** p<0.01; Control groups were compared to all treatment groups. Error bars represent one standard deviation.