Flow shear stress regulates endothelial barrier function and expression of angiogenic factors in a 3D microfluidic tumor vascular model

In vitro tumor vascular model

Co-culture of breast tumor and endothelial cells within microfluidic type I collagen hydrogels was conducted using a previously developed tumor vascular model.³⁸ Briefly, 1 × 10⁶ MDA-MB-231 cells were resuspended in 8 mg/ml neutralized collagen, poured into FEP tubing fit concentrically with a 22G needle (711 μm), and capped with PDMS sleeves (Fig. 1). After polymerization for 20 minutes at 37°C, the needle was removed and 10 × 10⁶ endothelial cells/ml were carefully injected into the central microchannel 2X at 10 min intervals and slowly rotated to completely cover the luminal surface. Flow was then immediately introduced through the endothelialized microchannel by perfusing EBM-2 growth media at a low flow rate of 3 μl/min. A graded increase in flow that subjected the endothelium to a low shear stress of τ = 0.01 dynes/cm² up to τ = 0.1 dyne/cm² over 72 hours was utilized to encourage the establishment of a viable and functional endothelialized channel.³⁸ Following this initial preconditioning time period, the flow was increased to induce one of 3 target WSS (τ_w = 1, 4, or 10 dyn/cm²) for 6 hours, after which analysis of angiogenic activity (expression of angiogenic factors and vascular permeability) was conducted. The 6h time point was selected to probe the initial and transient endothelial response to flow as well as the subsequent changes in tumor expressed angiogenic genes. Flow rates, Q, were initially estimated based on an assumption of Poiseuille flow, ${\displaystyle {\tau }_w=4Q\mu /\pi r^3}$, using the measured viscosity of EBM-2 culture media at 37°C (μ = 0.78 ± 0.0057 cP) and then validated using micro-particle image velocimetry as described previously.³⁸ Mono-cultures of MDA-MB-231 cells or TIME cells alone were used as controls.

VEGF protein secretion

Total VEGF protein was isolated from the perfusate of endothelial mono-cultures compared to co-culture with tumor cells during the 72 hr preconditioning period. Analysis shows a statistically significant increase in VEGF secretion from day 1 to day 3 (Fig. 2). While a greater amount of VEGF protein was measured in co-culture samples relative to endothelial mono-cultures, the difference was not significant. This may be due to matrix binding or endothelial VEGFR association of tumor-secreted VEGF, such that measurable changes in total VEGF secretions were undetected during co-culture. In addition, measured VEGF protein secretion from tumor-monocultures (data not shown) was slightly greater than that of endothelial mono-culture or co-culture, though the difference was not statistically different. While total VEGF secretion was significantly increased after 6 hr exposure to each τ_W relative to precondition conditions (τ_W < 0.01 dyn/cm²), no trends were observed as a function of increasing τ_W immediately post-shear, as VEGF concentration remained relatively constant for each flow condition (data not shown).

Our current findings indicate that under flow shear stress conditions (0.1<τ_W<10 dyn/cm²) and short culture durations (<3 days), VEGF protein secretion is slightly increased by co-culture conditions relative to tumor or endothelial mono-cultures. Additional studies investigating endothelial-originated VEGF in the microfluidic collagen hydrogel are needed to fully understand the role of flow shear stress on autocrine signaling mechanisms. The dynamic interplay between VEGF and its downstream signaling pathways has only recently been implicated in studies investigating shear stress-mediated angiogenesis.^39-42 Although hypoxia is the most well characterized factor that causes VEGF upregulation,^43,44 emerging mechanisms by which mechanical forces regulate VEGF expression are now being discovered. For example, a dynamic regulation pattern of VEGF gene and protein expression in endothelial cells was first demonstrated by Gan et al., in which high shear stress (25 dynes/cm²) decreased VEGF gene expression after exposure for 1.5 hours, but increased VEGF gene expression after exposure for 3 hours.⁴⁰ A later study by Conklin et al. showed that 24 hours of low shear stress (1.5 dynes/cm²) significantly increased VEGF gene and protein expression in endothelial cells.³⁹ In addition, we have previously shown that VEGF protein expression is enhanced in 2D, static tumor-endothelial co-cultures compared to monocultures,³⁷ though this response was observed over long-term culture durations (up to 28 days).

Tumor-expressed angiogenic genes in response to WSS

Tumor co-cultures and tumor mono-cultures were maintained under the 72 hr low flow preconditioning scheme, followed by exposure to one of 3 target WSS (τ_W = 1, 4, or 10 dynes/cm²) for 6 hrs, after which tumor mRNA was isolated for gene expression analysis. qRT-PCR results indicate that in the presence of an endothelialized microchannel, high WSS (τ_W = 10 dynes/cm²) significantly down-regulates all MDA-MB-231-expressed angiogenic factors probed in this experiment (Fig. 3). In contrast, WSS did not have an effect on tumor mono-cultures, suggesting fluid forces acting on the endothelium regulate cross-talk with surrounding tumor cells, thus altering gene expression activity. Furthermore, the decreasing trend in gene expression as a function of increasing WSS was not observed in tumor-monocultures compared to tumor co-cultures, thereby suggesting that this phenomenon is not a function of flow-mediated transport of oxygen and nutrients through the collagen matrix.

A conditioned media experiment was conducted in which MDA-MB-231 cells were cultured in collagen hydrogels polymerized in 48-well plates under static conditions. Cells were grown for 3 days in media collected from the flow shear stress experiments to determine if down-regulation of angiogenic proteins was due to soluble factors present in the media. Results showed no change in MDA-MB-231 gene expression after culture in conditioned media (data not shown). This suggests a physical, flow-mediated mechanism exists in the microfluidic collagen hydrogel, which plays a role in transcriptional regulation of the angiogenic factors probed in these experiments.

In a recent study by Song et al. investigating the role of VEGF gradients and shear stress on in vitro endothelial morphogenesis, results demonstrated that physiological shear stress (3 dyn/cm²) attenuates VEGF-driven endothelial sprouting.²² Even in the presence of high VEGF concentrations (50 ng/ml) endothelial invasion was minimal under flow relative to static conditions. These studies further confirmed that inhibition of VEGF-induced sprouting requires nitric oxide (NO) signaling, an important shear stress-responsive signaling molecule. Results from this work may provide an explanation by which high shear stress down-regulates tumor expressed angiogenic factors via NO secretion. For example, several studies have shown that NO can have inhibitory effects on tumorigenesis, in which DNA damage induced by NO causes tumor regression and inhibition of metastasis.^45,46 Angiogenic factors such as VEGF and the angiopoietins have also been shown to activate eNOS in vascular endothelial cells.^47,48 As such, co-culture conditions within the microfluidic tumor vascular model, which increase VEGF, may stimulate endothelial NO secretion. Flow shear stress may further increase NO signaling, such that the combined effect of flow shear stress and co-culture-induced VEGF decreases tumor angiogenic activity. Further analysis is needed to better understand thses underlying flow-mediated effects.

Endothelial barrier function

The permeability of the endothelium was measured following the 72 hrs precondition period, and after 6 hrs exposure to one of 3 target WSS (τ_W = 1, 4, or 10 dyn/cm²). The measured effective permeability coefficient, P_d, of the endothelialized microchannel as a function of WSS indicates that P_d of the 70 kDa Oregon Green-conjugated dextran decreases with increasing WSS for both endothelial mono-culture and co-culture with tumor cells (Fig. 4A). This decreased permeability is likely due to increased endothelial alignment and spreading in the direction of flow,³⁸ in which high WSS (τ_W = 10 dyn/cm²) significantly decreased P_d relative to the preconditioned endothelium. These findings are consistent with others showing shear stress-mediated endothelial morphological adaptations^{22,38, 49-52} and permeability.^{14,17,19,21,23,53} The presence of tumor cells also increased P_d relative to endothelial mono-cultures, with a significantly greater P_d under high WSS conditions. Images taken at the end of the permeability assay for the case of τ_W = 10 dyn/cm² qualitatively show greater distribution of dextran across the endothelium during co-culture with tumor cells, relative to dextran distribution immediately post 72 hrs preconditioning (Fig. 4B). A more disorganized endothelium at the channel wall are also observed during co-culture, with initial signs of migration off the luminal surface, which may contribute to increased permeability.

Vessel destabilization and reduced endothelial barrier function increase endothelial cell exposure to angiogenic growth factors such as VEGF and the angiopoietins, and leads to vessels that are more amenable to sprouting and angiogenesis.⁵⁴ Studies by Holash et al. in a rat glioma model have also demonstrated that mature vessels must first destabilize to allow for subsequent sprouting and vessel growth.³¹ Endothelial tight junctions and adherens junctions are known to play a role in the paracellular permeability of the endothelium; therefore, a more thorough investigation of tight junction protein distribution in response for flow shear stress and co-culture with tumor cells is needed to determine if down-regulated tight junction protein expression explains the decreased permeability observed in the current study. Several reports have provided evidence of flow shear stress-mediated distribution of tight junction proteins, in which exposure to shear stress leads to a transient decrease (<4 hrs post shear) in VE-cadherin or occludin protein expression, as endothelial cells migrate and align in the direction of flow.^{14,15,39, 55-59} Studies investigating the effect of shear stress on permeability of macromolecules across the endothelium also revealed that endothelial permeability is acutely sensitive to shear stress in vitro. ⁵³ Exposure of both 1 and 10 dyne/cm² dramatically increased permeability of FITC-BSA after 30 and 60 minutes; however, permeability returned to pre-shear values after 120 minutes.⁵³ Similar studies demonstrated that when subjected to low shear stress, endothelial cells cultured in collagen hydrogels were visibly more permeable to FITC-BSA.⁶⁰ These findings are consistent with permeability results obtained from this study, suggesting that the endothelial cells migrate and align to varying extents in response to different magnitudes of flow shear stress.

Cell culture

A human breast carcinoma cell line (MDA-MB-231) (American Type Culture Collection (ATCC)) and a telomerase-immortalized human microvascular endothelial cell line (TIME) were used in this study. TIME cells were provided as a generous gift from Dr. Shay Soker at the Wake Forest Institute for Regenerative Medicine (Winston-Salem, NC). A lentiviral vector system was used to genetically modify the TIME cells to stably express a red fluorescent protein (RFP) for real-time visualization of endothelial morphology during culture. Imaging was performed using a fluorescence microscope (Leica AF6000) to monitor cell growth during culture.

MDA-MB-231 cells were cultured in Dulbecco's Modified Eagle Medium: nutrient mixture F-12 (DMEM/F12) (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich, St. Louis, MO), and 1% penicillin-streptomycin (Invitrogen, Carlsbad, CA). TIME cells were cultured in EBM-2 (Lonza, Rockland, ME) media supplemented with a growth factor BulletKit (Lonza CC-4176). All cells were maintained in a humidified 5% CO₂/95% air atmosphere at 37°C within an incubator for all experiments.

Enzyme-linked immunosorbent assay

Protein levels of VEGF during the 72 hr preconditioning time period as well as post shear were measured using an enzyme-linked immunosorbent assay (ELISA) (Quantikine Human Immunoassay kits (R & D Systems, Minneapolis, MN)) according to the manufacturer's protocol. Samples were collected from conditioned perfusate of tumor-endothelial co-cultures and compared to tumor and endothelial mono-cultures at 24, 48 and 72 hrs (n = 4). Samples were also collected after the 6 hr exposure to each τ_W to assess flow-mediated translation of VEGF.

Quantitative RT-PCR

To determine the effect of WSS on tumor-endothelial cross-talk, expression levels of target genes in MDA-MB-231 during co-culture or mono-culture after the 6 hr exposure to each τ_W were determined quantitatively by real-time RT-PCR. To isolate tumor cell mRNA from co-cultures, the endothelialized microchannel was de-nuded by carefully scraping the cells off the luminal surface and subsequently washing with PBS. Total RNA from the microfluidic collagen scaffolds was then isolated by the phenol-chloroform extraction method as described previously.³⁸ Reverse transcription and PCR amplification of gene-specific TaqMan PCR primers: VEGF-A (NM_001025366.2), ANG-1 (NM_001146.3), ANG-2 (NM_001118887), HIF1-α (NM_001243084.1) and MMP-9 (NM_004994.2) (Applied Biosystems, Foster City, CA) were then conducted. Evaluation of 2^−ΔΔC_T indicates the fold change in gene expression, normalized to GAPDH (NM_002046.3) housekeeping gene and relative to the control group (n = 4).

Endothelial permeability assay

A permeability assay to quantify barrier function of the endothelialized microchannel was conducted as described previously.⁶¹ Briefly, 70 kDa Oregon green-conjugated dextran (10 μg/ml in serum-free medium) was perfused through the endothelialized microchannel at 26 μl/min (τ = 0.1 dyn/cm²) for 1 hr, as selectively permeable mature capillaries are known to be impermeable to dextrans over a molecular weight of 65 kDa.⁶² Images were captured every 4 minutes using an inverted fluorescence microscope (Leica AF6000). ImageJ (National Institutes of Health, Bethesda, MD) was then used to measure average fluorescence intensity in a ROI that spanned the width of the collagen, including the endothelialized microchannel. The effective permeability coefficient, P_d, which describes the ability of a solute to escape uniformly from the vascular lumen,⁶¹ was calculated using the following equation:

where d is the microchannel diameter, I_b is the average background intensity, I₁ is the average initial intensity, and I₂ is the average intensity after the recovery time Δt.⁶¹ P_d was calculated as a function of each target WSS (τ_W = 1, 4, or 10 dynes/cm²) from at least 5 consecutive measurements in both endothelial mono-culture as well as co-culture with tumor cells (n = 3).

Statistical analysis

Experimental groups were tested and analyzed independently and the data are expressed as mean value ± standard deviation. Significance of the RT-PCR results, ELISA results and permeability assay were verified using Student's t-test. A 95% confidence criterion will be used to determine statistically significant differences between experimental groups.