Inflammation disrupts vascular barrier function in a microfluidic model of acute joint injury

Introduction:: In patients with joint injury (ex. anterior cruciate ligament (ACL) tear), joint-wide inflammation is the highest in the first 48 h post-injury [1]. After this acute injury ( >1 week later), inflammatory cytokine levels typically decrease. However, in a certain subset of patients, the initial pro-inflammatory phase is uncontrolled, leading to chronic inflammation, tissue damage, and post-traumatic osteoarthritis (PTOA) [2]. Human clinical data has shown heightened concentrations of the proinflammatory cytokine IL-1β in ACL-deficient knees 6 months to even 12 years after injury [3]. It remains unknown how and why some patients this initial acute inflammatory stage, whereas others progress to develop PTOA. The answer may lay in part in the vascular system, particularly as the joint is an interconnected organ lined by the vessel-rich synovium. These vessels contact the synovial fluid, the primary ‘sink’ where inflammatory molecules accumulate. Inflammation-induced alterations in vessel structure could trigger further release of catabolic factors. In order to probe this crosstalk between endothelial cells within the synovium and pro-inflammatory molecules from the synovial fluid, the objective of this study was to develop a humanized microfluidic model of acute joint injury. Endothelialized vessels were engineered within collagen gels (the primary matrix component of the synovium), exposed to IL-1β (inflammation/injury), and functional outcomes (vascular structure, permeability, and cell-cell contacts) were assessed over 5 days. Our hypothesis was that treatment of the engineered vessels with IL-1β would increase vascular permeability.


Materials and Methods::

Device fabrication. Polydimethylsiloxane (PDMS) microfluidic devices were fabricated using soft-lithography as described previously [4]. PDMS rods were formed via extrusion through 23-gauge needles, resulting in ~340μm diameter lumens. The top and bottom PDMS layers were aligned, and a PDMS rod was inserted through the inner channel of each device. Device endothelialization and culture. Bovine type I collagen (3 mg/mL) was injected into the hexagonal chamber of the devices and polymerized (37°C for 1 h). After removing the PDMS rods, human umbilical vein endothelial cells (HUVECs) (P4; 30,000/μL) were injected into each hollow channel. The devices were flipped every 2 min for 30 min, and cultured in endothelial growth media (EGM, Lonza). After 1 day of culture, the devices (n = 8-10/group) were cultured in EGM (Cntrl) or EGM + 10 ng/mL IL-1β (IL1) for 4 additional days with media changed once daily. Permeability. Vessel permeability was quantified, as in [5], by injecting 70-kDa FITC-dextran into a channel, and serially imaging the device midplane for 10 min. Image series were uploaded into a MATLAB script [5] to quantify P_d. Immunofluorescence. Devices were fixed (4% paraformaldehyde), permeabilized (0.5% Triton-X), blocked (10% donkey serum), and incubated in rhodamine phalloidin (1:500) and VE-cadherin primary antibody (1:300) overnight at 4°C. Subsequently, devices were rinsed and stained with DAPI (1:1000), and Alexa Fluor 647 donkey anti-rabbit IgG (1:500) for 1 h before imaging using confocal microscopy. Statistics. An unpaired t-test (GraphPad Prism) was performed on the permeability data.




Results, Conclusions, and Discussions:: Results/Discussion: Microfluidic devices were fabricated in a high-throughput manner (Fig. 1A, B), enabling investigation into the impact of joint inflammation on vascular integrity. As shown, the seeded HUVECs attached along the circumference of the lumens (Fig. 1C), spanning ~ 300μm in diameter. Over 5 days in culture, the vessels retained their shape (Fig. 1D), and the adherent HUVECs formed a confluent monolayer (Fig. 1E). To model inflammation during acute joint injury, we introduced IL-1β to the engineered vessels (after HUVEC attachment, on Day 1). After exposing the vessels to IL-1β for 4 days, the vessel walls appeared discontinuous with cells protruding from the vessel (Fig. 2A, B). To quantify the impact of this disruption in vessel shape on subsequent barrier function, we tracked the outward diffusion of FITC-dextran from control and IL1-treated vessels into the surrounding collagen gel using time-lapse confocal microscopy (Fig. 2C). Ten minutes after injecting FITC-dextran into each vessel, there was a stark difference between the control and IL1 groups, as quantified using the equation shown (Fig. 2D). On average, the vessels treated with IL-1β were approximately twice as permeable as the control vessels (Fig. 2E). The range of permeability values achieved in this assay for the control vessels compare similarly to those achieved by Polacheck et al. (P_d ~ 2-7 x 10^-6 cm/sec) who created smaller 160μm diameter channels surrounded by 3 mg/mL collagen [5], supporting our results. To investigate how inflammation disrupts vascular barrier function, we stained the control and IL1-treated vessels to visualize cell-cell contacts (VE-cadherin) (Fig. 2F). As expected, VE-cadherin staining localized along cell peripheries in the healthy vessels. However, interestingly, this endothelial specific adhesion molecule was largely absent in the IL1-treated vessels, indicating decreased cell-cell contacts in the presence of inflammatory signals.

Conclusion: This model is the first-of-its-kind to provide a potential explanation towards how inflammation may disrupt vascular barrier function in a collagen-rich tissue, alike the human synovium. To further enhance the physiologic relevance of our model, ongoing studies are integrating both healthy and osteoarthritic human synoviocytes into the outer collagen gel.


Acknowledgements (Optional): : The authors would like to thank the Rhodes Trust, Schmidt Futures, and the Schmidt Science Fellowship.


References (Optional): : [1] Sward, P., et al., Osteoarthritis and Cart., 20:1302-1308, 2012.
[2] Lieberthal, J., et al., Osteoarthritis and Cart., 23:1825-1834, 2015.
[3] Marks, P. H., et al., Arthroscopy: The J. of Arthroscop. & Rel. Surg., 21:1342-1347, 2005.
[4] Jimenez-Torres, J. A., et al., Adv Healthcare Mat, 5:198-204, 2016.
[5] Polacheck, W. J., et al., Nature Protocols, 14:1425-1454, 2019.