BOECs were isolated from the peripheral blood of two healthy adult volunteers. Normal human serum (NHS) was derived from three healthy adult volunteers.

BOECs were isolated as previously described and cultured in endothelial cell growth medium (cEGM-2: Endothelial Basal Medium 2 [EBM-2] supplemented with growth factors [EGM-2 BulletKit]) (Lonza, Walkersville, MD), 10% fetal bovine serum (FBS) (Sigma-Aldrich, St. Louis, MO), and 1% antibiotic–antimycotic (Gibco, Invitrogen, Life Technologies, Carlsbad, CA) (36, 37). Cells were maintained at 37°C, in an environment with 5% CO₂. Passages 3–12 were used.

BOECs grown to confluence were exposed to cyclosporine 10, 20, 50, or 100 μg/ml in media for up to 24 h. Cyclosporine stock solution (Sandimmune IV, Novartis Pharmaceuticals Canada Inc., Dorval, QC, 50 mg/ml) was kindly provided by the SickKids pharmacy. For experiments involving complement fixation on BOECs, cells were sequentially exposed to 50% NHS (1:1 in serum-free media [SFM]) for 30 min. In experiments utilizing an antibody specifically blocking the membrane-anchored complement regulator CD59 (BRIC229, IgG2b, International Blood Group Reference Laboratory, NHS Blood and Transplant, Bristol, UK), cells were incubated for 30 min with the blocking antibody (5 μg/ml) prior to exposure to 50% NHS (36, 38).

C3b and C5b-9 deposition on BOEC surfaces were demonstrated by flow cytometry using a C3c antibody detecting the C3c portion of native C3 and C3b (C3c-FITC conjugated antibody, Abcam, ab4212, 1:50 dilution) and C9 (Complement Technologies Inc, TX, A226, 1:100 dilution). Cells were grown to confluence and exposed to cyclosporine treatment and complement fixation as described. Cells were washed with phosphate-buffered saline (PBS) and incubated with Fixable Viability Dye eFluor780 (eBioscience, San Diego, CA, 1:1,000 dilution reconstituted in PBS) at 4°C for 30 min. For flow cytometry, cells were harvested by scraping and washed with PBS before use (Supplementary material).

von Willebrand Factor release from BOECs was detected via immunofluorescence as described previously (36). BOECs treated with media for 24 h, followed by incubation with anti-CD59 blocking antibody for 30 min and 50% NHS/50% SFM for 30 min, were used as positive control and compared to cells kept in media (negative control). Cells were then washed and fixed with 2% paraformaldehyde and permeabilized with 0.2% Triton in PBS, followed by incubation with rabbit anti-VWF (Dako, Carpinteria, CA, A0082, 1:1,000) and goat anti-VE-cadherin (Santa Cruz Biotechnology, Dallas, TX, sc-6458, 1:250) for 4 h. Alexa Fluor 488- and Alexa Fluor 555-conjugated species-specific secondary antibodies were used at a dilution of 1:1,000. Nuclei of cells were stained with 0.12 μg/ml Hoechst stain (Thermo Fisher Scientific, Waltham, MA) for 5 min.

To determine the expression level of the membrane-anchored complement regulators MCP/CD46, decay-accelerating factor (DAF/CD55), and CD59 on BOECs, BOEC lysates were utilized for flow cytometry and Western blotting analysis (Supplementary material).

The binding of CFH to BOEC surfaces was demonstrated by flow cytometry as described previously (39), using purified CFH (CSL Behring, Marburg, Germany) tagged with Alexa Fluor 488 succinimidyl ester (10 μg/mL, Life Technologies) for 1 h at room temperature before being dialyzed overnight in PBS. Cells exposed to 500 mU/mL neuraminidase (MilliporeSigma; N2876) were used as the positive control. Cells were washed two times with PBS and scraped off. Cells were then incubated with Fixable Viability Dye eFluor780 at 4°C for 30 min. They were then washed with PBS and resuspended in 100 μL PBS. Each sample was then incubated with 4 μg of Alexa Fluor 488-tagged CFH for 1 min, after which 500–1,000 μL of Attune focusing fluid (Thermo Fisher Scientific, 4449791) was added and assessed by flow cytometry (Supplementary material).

For immunofluorescence experiments, cells were cultured to a minimum of 80% confluency on collagenized coverslips and exposed to cyclosporine A as described. Cells exposed to 500 mU/mL Neuraminidase for 1 h and 0.5 U/mL Heparinase III (H8891-5UN, Sigma-Aldrich, St. Louis, MO) for 30 min were used as positive controls. Cells were washed and fixed with 4% paraformaldehyde, blocked for 1 h with 3% BSA, followed by incubation with goat anti-Factor H (1:100, Complement Technology Inc., TX; A237) and mouse anti-heparan sulfate (1:50, US Biological Life Sciences, Salem, MA; H1890) overnight at 4°C. Goat Alexa Fluor 488 and Mouse Alexa Fluor 555 secondary antibodies were used, respectively, at a dilution of 1:200 for 1 h at room temperature. The nuclei of the cells were stained with 0.12 μg/ml Hoechst stain (Thermo Fisher Scientific, Waltham, MA) for 5 min. Confocal microscopy was performed as detailed in Supplementary material, and total fluorescence intensity was measured using ImageJ software.

To determine CFH cofactor activity on BOEC surfaces, cells exposed to 500 mU/mL neuraminidase for 1 h (Millipore Sigma; N2876) were used as the positive control. Cofactor activity of surface-bound CFH was detected as previously described (40). Cells were incubated with 10 μg/ml CFH (CSL Behring, Marburg, Germany) at 37°C for 1 h, 10 μg/ml CFI (EMD Millipore Corp., MA, 341280) and with 3.3 μg/ml C3b (EMD Millipore Corp., MA, 204860). The supernatant was collected at baseline and various subsequent time points (up to 180 min), and the samples were transferred to a reduced sample buffer and separated by 10% SDS-PAGE. The appearance of C3b degradation fragments was detected by Western blotting (Figure 6). Primary goat anti-C3, 1:1,000 dilution (Complement Technology Inc., TX, A213) with corresponding secondary HRP-conjugated antibody at a dilution of 1:5,000 was used for detection.

To image the endothelial glycocalyx, an Alexa Fluor 594-conjugated wheat germ agglutinin (WGA, Thermo Fischer Scientific, W11262, dilution 2 μg/ml), anti-heparan sulfate antibody (Abcam, Cambridge, UK, ab23418, 1:100), and peanut agglutinin (PNA, Vector Labs, Ontario, CA, FL-1071-5, 1:200) were used. Cells were cultured to confluence on coverslips and exposed to cyclosporine as described. Cells exposed to 500 mU/mL neuraminidase for 1 h were used as a positive control in WGA and PNA experiments. Cells exposed to 0.5 U/mL Heparinase III (H8891-5UN, Sigma-Aldrich, St. Louis, MO) for 30 min were used as a positive control in heparan sulfate experiments. Cells were incubated with Alexa Fluor 594-conjugated WGA for 5 min on ice and washed two times with ice-cold HBSS, and the coverslips were mounted in a Chamlide magnetic chamber (Life Cell Instrument, Seoul, Korea) and overlaid with media. Confocal microscopy was performed as detailed in Supplementary material, and total fluorescence intensity was measured using ImageJ software. For experiments using anti-heparan sulfate and PNA, cells were washed and fixed with 2% paraformaldehyde, followed by incubation with mouse anti-heparan sulfate (1:100) and anti-PNA (1:100) for 1 h. Alexa Fluor 488-conjugated species-specific secondary antibodies were used at a dilution of 1:1,000. Nuclei of cells were stained with 0.12 μg/ml Hoechst stain (Thermo Fisher Scientific, Waltham, MA) for 5 min.

Figures were generated with GraphPad Prism (Version 6.0c; GraphPad Software, La Jolla, CA) and displayed as the mean and standard deviation. Statistical analysis was performed via paired t-test or two-way ANOVA with post-hoc analysis. A p < 0.05 was considered statistically significant. In the figure legends, p-values are presented as follows: ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001.

The use of cyclosporine is associated with a vascular injury in pathophysiological situations. We, therefore, tested whether cyclosporine treatment of cultured BOECs caused endothelial cell toxicity using an established lactate dehydrogenase (LDH) assay for lytic cell death. We found that cyclosporine caused cytotoxicity of BOEC cultures in a dose- and time-dependent fashion (Supplementary Figure S1). Specifically, the acute (1 h) treatment of BOECs with low concentrations (<50 μg/mL) of cyclosporine did not cause cell lysis, while a 24-h treatment of the cells with cyclosporine used above 250 μg/mL led to lysis of nearly the entire culture. Intermediate concentrations of cyclosporine (50 μg/mL) caused ~60% of the cells to rupture, and lower concentrations of 10 μg/ml did not lead to any detectable LDH release (Supplementary Figure S1). We, therefore, chose to treat BOECs within the range of 10 μg/mL (non-lethal) and 50 μg/mL (~half-maximal lysis) concentrations of cyclosporine in subsequent experiments.

To that end, confluent monolayers of BOECs were treated with these concentrations of cyclosporine in medium containing 10% fetal bovine serum (FBS) for 24 h and subsequently exposed to 50% NHS in serum-free medium (SFM) for 30 min as established in Supplementary Figure S1. Under these conditions, we found that the treatment of BOECs with 50 μg/ml of cyclosporine caused a significant increase in complement C3 deposition (Figure 1A, MFI: cyclosporine 50 μg/mL 441.1 ± 67.1 vs. control 265.8 ± 50.1, n = 4, p = 0.023). Using lower doses of cyclosporine (10 μg/ml), we determined that the increased deposition of C3c was enhanced in the absence of serum. Factors in the serum prevented C3c deposition on the cyclosporine-treated BOEC cultures: >2.5% FBS prevented C3c deposition, while at <0.5% FBS, significantly increased C3c was detected on the surface of cyclosporine-treated cells (Figure 1B, MFI: cyclosporine 10 μg/mL in serum-free media 622.5 ± 32.72 vs. control 343.1 ± 65.84, n = 6, p < 0.01).

Inhibiting the function of CD59, a membrane-anchored complement regulator, is an established means of sensitizing complement fixation on endothelial cells. Blocking CD59 with antibodies has the dual effect of complement induction mainly via sensitization (classical pathway) but also through complement amplification (alternative pathway) (36, 38, 41–43). We were interested to examine whether cyclosporine had general effects on the membrane topology that impact C3c deposition or if its effect was via CD59. Using the same flow cytometry approach used in Figure 1B, we found that blocking CD59 indeed led to a large increase in C3c associated with the endothelial cells (Figure 1C). However, cyclosporine treatment further increased C3 deposition ~2-fold beyond the level achieved by blocking CD59 alone. This effect was also observed for C5b-9 to an even greater extend fold increase (Figure 1D). Thus, BOECs treated with cyclosporine had a dose-dependent injury concomitant with increased complement deposition that could be enhanced by the removal of serum or complement regulators.

Weibel–Palade bodies (WPBs) are endothelial storage granules containing pro-hemostatic and pro-inflammatory molecules, including VWF, P-selectin, interleukin-8, endothelin-1, and angiopoietin-2 (44–46). As previously demonstrated by us and others, WPBs are exocytosed upon endothelial cell injury and activation to release their contents, which potentiates inflammatory responses, vascular leakage, and leukocyte adhesion (36, 45, 47). Given that cyclosporine resulted in endothelial cell injury and complement deposition, we hypothesized that cyclosporine treatment may also lead to the endothelial release of VWF.

Using the previously established protocol, we first showed that complement activation indeed caused the release of intracellular VWF (Supplementary Figure S2—positive control using anti-CD59 sensitization) (36). We then found that BOECs treated with cyclosporine 10 μg/mL had less intense staining of intracellular VWF (Figure 2). Taken together, our results showed that cyclosporine induces VWF release from BOECs.

The regulation of the alternative pathway of complement activation is executed by a combination of fluid-phase (CFH and CFI) and membrane-bound regulators (mainly MCP/CD46, DAF/CD55, and CD59) that maintain the balance between complement activation and inhibition (8, 13). Given that cyclosporine caused an increase in complement activation on the surface of BOECs, it was conceivable that cyclosporine decreased the expression of membrane-bound complement regulators. We, therefore, assessed the expression of MCP/CD46, DAF/CD55, and CD59 on the surface of BOECs after their treatment with cyclosporine using flow cytometry, and the total cell expression of these regulators by probing cell lysates with Western blotting. We found that treatment of the BOECs with low concentrations (10 μg/mL) of cyclosporine resulted in the increased surface and total cell expression of MCP/CD46, DAF/CD55, and CD59 (Figure 3). Incubation with cyclosporine at higher concentrations of cyclosporine (50 μg/mL) resulted in a similar effect (data not shown). Thus, the increased complement deposition on the surface of cyclosporine-treated cells was not the result of the lost expression of membrane-bound complement regulators.

Since enhanced complement deposition induced by cyclosporine occurred in the context of increased expression of membrane-bound complement regulators, we hypothesized that cyclosporine may instead impair CFH-mediated complement regulation. CFH is the central circulating alternative pathway inhibitor, which competitively prevents C3b deposition on cell surfaces, acts as a cofactor to CFI to cleave surface-bound C3b, and accelerates the decay of the C3bBb complex (48–50). To exert these functions, CFH is known to be closely associated with endothelial surfaces via its multiple glycosaminoglycan/sialic acid-binding domains (51–55).

To test whether cyclosporine impaired CFH binding, we pre-treated BOECs with cyclosporine and then assessed the ability of the cells to secure Alexa Fluor 488-conjugated CFH from the culture medium. The Alexa Fluor 488-labeled CFH was added for 1 min to live cells before their analysis by flow cytometry. We found that incubation of BOECs with cyclosporine at 10 μg/mL for 24 h caused a significant reduction in CFH binding (Figures 4A, B, MFI control 386.3 ± 97.8 vs. cyclosporine 10 μg/mL 78.3 ± 45.8, n = 3, p = 0.0078). A brief (1 h) treatment of the cells with neuraminidase used at 500 mU/mL, an enzyme that cleaves terminal sialic acid groups from glycoproteins, was used as a positive control. The functionality of neuraminidase in cleaving sialic acid was confirmed by live imaging with wheat germ agglutinin (WGA; see the section below) and by CFH binding. Removal of sialic acids inhibited CFH binding to the endothelium to nearly the same extent as cyclosporine treatment (Figure 4B). This reduction in CFH binding on cells treated with cyclosporine 10 μg/mL for 24 h was also confirmed by immunofluorescence (Figures 4C–L, MFI control 6.43 ± 0.44 vs. cyclosporine 10 μg/mL 3.03 ± 0.26, n = 3, p < 0.001).

Locally concentrating CFH to the membrane of the vascular endothelium is critical for the protection of the membrane from complement deposition. The activity of the CFH, once docked to the endothelial surface, can subsequently be measured by assays that determined the degradation of complement. We assessed the functional consequences of the cyclosporine-induced reduction in CFH binding to BOECs by employing a previously established CFH surface cofactor activity assay (40). In this assay, endothelial cell-bound CFH was used as the sole source of CFH. The incubation of C3b with CFI and CFH results in C3b degradation with the appearance of C3b fragments with molecular weights of 68 kDa (C3b α'1), 43/46 kDa (C3b α'2), all of which can be detected via the same Western blotting approach. We first assessed the endogenous cofactor activity of the membrane-bound complement regulator MCP/CD46 in the absence of CFH when exposed to media (control) and various concentrations of cyclosporine (10, 50, and 100 μg/mL). Degradation products were detectable after 90 min, with no detectable significant differences between cyclosporine concentrations (Figure 5A, Supplementary Figures 3A, C, E, G, I). Pre-incubation of BOECs with CFH resulted in the appearance of C3b degradation products after 15 min (Figure 5B), demonstrating the expected significantly higher cofactor activity of CFH on endothelial surfaces. However, when BOECs were pre-incubated with neuraminidase followed by incubation with CFH and subsequently with C3b and CFI in the absence of additional CFH, degradation products were detectable only after 60 min. This result was in keeping with a lack of surface CFH in cells devoid of sialic acids (Figure 5C).

We then assessed the effect of cyclosporine exposure on the cofactor activity of surface-bound CFH. BOECs exposed to increasing doses of cyclosporine (10, 50, and 100 μg/mL for 24 h) demonstrated a dose-dependent decrease in CFH cofactor activity as evidenced by the later appearances of C3b degradation products: cyclosporine 10 μg/mL after 45 min, cyclosporine 50 μg/mL after 45 min, and cyclosporine 100 μg/mL after 90 min (Figures 5D–F). Taken together, we found decreased cofactor activity of CFH on BOECs pre-treated with cyclosporine (Figure 5G).

CFH has been reported to bind to endothelial surfaces via its glycosaminoglycan/sialic acid-binding domains (51–55). Since removing sialic acids with neuraminidase ablated CFH binding to the same extent as cyclosporine treatment, we assessed whether cyclosporine exerted its inhibitory effects on CFH binding via remodeling of the glycocalyx. We first stained glycans/polysaccharides containing sialic acid and N-acetyl-D-glucosamine using the lectin wheat germ agglutinin (WGA) conjugated to Alexa Fluor 594. Of note, we imaged the cells live as fixation resulted in a dramatic decrease in overall fluorescence. To prevent endocytosis of the lectin, incubation with Alexa Fluor 594-WGA was performed in the cold (4°C). We determined a decrease in Alexa Fluor 594-WGA staining in BOECs treated with neuraminidase used at 500 mU/mL for 1 h, with conditions identical to those that inhibited CFH binding (Figures 5C, 6A, B: MFI neuraminidase 18,459 ± 6,154 vs. control 32,525 ± 8,990, p < 0.0001). Treatment with cyclosporine at 10 μg/mL also resulted in less intense staining with Alexa Fluor 594-WGA when compared to control (Figures 6C, D: MFI cyclosporine 10 μg/mL 18,752 ± 6,154 vs. control 32,525 ± 8,990, p < 0.0001). The decrease in the WGA signal in cyclosporine was more apparent in the clusters on the apical surface of the endothelial cells and less visible at cell–cell junctions (Figures 6A, C).

We further assessed whether cyclosporine had additional effects on the endothelial glycocalyx, specifically on the surface density of heparan sulfates. Heparan sulfates are covalently attached to the proteoglycans process in the Golgi apparatus (e.g., syndecans and glypicans). These side chains can be detected by immunostaining: While the polysaccharides may not be immunogenic on their own, in the context of proteoglycans, good antibodies have been generated and made commercially available. We, therefore, immunostained non-permeabilized control or cyclosporine-treated endothelial cells with anti-heparan sulfate antibodies. When compared to control, treatment with cyclosporine used at 10 μg/mL resulted in an ~60% decrease in the intensity of heparan sulfate per cell (Figures 6E–N: MFI cyclosporine 10 μg/mL 3.01 ± 0.36 vs. control 9.30 ± 1.04, p < 0.0001). Heparinase III, a polysaccharide lyase that degrades heparan sulfate, was used as the positive control, which led to a similar decrease in the intensity of heparan sulfate (Figures 6H–J, N: MFI heparinase III 3.19 ± 0.41 vs. control 9.30 ± 1.04, p < 0.0001). Treatment with cyclosporine 10 μg/mL and heparinase III led to a similar decrease in CFH (Supplementary Figure S4: MFI cyclosporine 10 μg/mL 3.03 ± 0.26 vs. heparinase III 4.11 ± 0.20 vs. control 6.43 ± 0.44, p < 0.0001).

Finally, the modifications to the glycocalyx upon cyclosporine treatment could be the result of overactive hydrolases (i.e., glycosidases or proteases) or the result of mistrafficking and expression of proteoglycans and glycoproteins. To determine whether surface glycoproteins in cyclosporine-treated cells were devoid of sialic acids, we used a lectin, peanut agglutinin (PNA), that recognizes exposed, terminal galactose sugars. We found that cyclosporine-treated cells did not have cleaved sialic acids from surface glycoproteins as evidenced by no PNA signal observed on the surface of the cells (Figures 6 O–X: MFI cyclosporine 10 μg/mL 13.6 ± 4.1 vs. control 16.1 ± 4.9, p = ns). As a positive control, we showed increased PNA staining in BOECs treated with neuraminidase (Figures 6R–T, X: MFI neuraminidase 35.8 ± 6.7 vs. control 16.1 ± 4.9, p < 0.001). The findings with PNA were also confirmed on flow cytometry (Supplementary Figure S5).

Taken together, our findings suggest that cyclosporine treatment results in endothelial glycocalyx breakdown via the loss of surface glycoproteins and heparan sulfates, which leads to impaired CFH surface binding.