(I-329) Development of a model of sex-specific inflammation induced diabetic vasculopathy

Diabetes mellitus (DM) is an ongoing health crisis; an estimated 537 million adults live with DM and another 240 million are believed to be undiagnosed. Unfortunately, complications, especially cardiovascular disease (CVD), are common in DM patients. Microvascular complications, including diabetic retinopathy, neuropathy, and nephropathy, arise from increased vascular permeability due to compromised tight junctions. Importantly, microvascular CVD, particularly diabetic retinopathy, does not affect all patients in the same way. Some studies show that male patients are at an increased risk of developing diabetic vasculopathy. This is in line with data that shows that pre-menopausal levels of estradiol (E2) in females may be protective against inflammation. However, contrasting studies suggest women are actually more susceptible to microvascular CVD secondary to DM. Hence, more examination of the role of sex in progression of inflammatory diseases of the microvasculature is warranted. Utilization of microphysiological systems (MPS) that allow the examination of human cells in engineered arrangements are an attractive option to study sex differences in vascular inflammatory disease. Here we report the development and validation of an MPS that may be utilized to examine difference in vascular permeability between XX and XY models in response to inflammatory insult and E2.

A 4-layer PDMS MPS containing upper and lower media reservoirs, a stromal mimetic layer, and an endothelial channel all separated by polyester membranes was designed for use in these experiments. UV sterilized endothelial channels and stromal compartments were pretreated with polydopamine (5mg/ml) for 2 hours. Prior to loading, endothelial channels were loaded with Geltrex diluted in serum free endothelial cell media. Primary human umbilical vein endothelial cells (HUVEC) from XX donors (ATCC, ages 20-45, P2 to P8) were loaded into prepared channels at 8x10⁶ cells/ml and the stromal layer was loaded with 2.5 mg/ml collagen I hydrogels. Devices were flipped at 1 hour to ensure complete coating of the channel to develop a continuous 3D endothelial model. Unadhered HUVEC were removed 2 hours post seeding, media was added, and devices were again flipped for incubation. To develop more mature vessels, devices were cultured on a rocker at 15^o incline. After 48 hours, devices were changed to hormone starvation media. At 72 hours, devices were exposed to lipopolysaccharides from E. coli (LPS, 1ug/ml), E2 (10nM), or both LPS and E2. At 96 hours, media containing FITC-dextran was loaded into the upper media channel, and devices were cultured for 24 more hours to allow FITC-dextran to perfuse through the endothelial layer. At 120 hours, media from the bottom chamber was collected and devices were fixed. Media collected was processed via spectrophotometry via a plate reader, and devices were stained with DAPI and phalloidin to visualize nuclei and the cytoskeletal f-actin respectively.

The device was validated to show that a 3D vessel construct was formed by endothelial cells coating the entire channel, and endothelial layers were shown to impede FITC-dextran perfusion compared to bare membranes. Rocking seems to have matured endothelial layers as perfusion decreased in rocked conditions. In devices treated with 1ug/ml LPS for 48 hours, endothelial cells displayed increased gaps in the monolayers and increased permeability to FITC-dextran compared to control devices. Conversely, devices treated with 10nM E2 only resulted in decreased permeability. Devices cultured with both LPS and E2 displayed less disruption to the monolayer and a permeability between that seen in LPS and E2 alone. Actin staining showed that devices exposed to LPS were more prone to non-localized actin staining with higher positivity of cytosolic actin as compared to control or E2 treated devices which displayed more localized actin at cell-cell junctions.

Based on these results, we concluded that the device is successful in modeling an endothelial layer and the effects of simulated inflammation on said layer. The increase in non-localized actin and in FITC permeability in LPS exposed devices suggests that the LPS mimics inflammation seen in DM which is known to cause “leaky” vasculature. The reduction to this disrupted actin staining and permeability caused by co-culture with E2 hints that E2 may be partially able to counteract the pro-inflammatory effects of LPS in models loaded with XX cells.

Future directions include expansion of these studies to evaluate both XX and XY cells with both male and female hormones. Additionally, we aim to build upon these results by incorporation of endothelial tight junction specific stains, such as zonula occludens-1 (ZO-1). Further, we hope to use this device to mimic more clinical causes of inflammation such as utilizing hypoxia and hyperglycemia to mimic the pro-inflammatory environment seen in DM instead of using chemical means such as LPS. Overall, the results presented here establish a foundation for an MPS that may serve as a valid model of DM associated vasculopathies.