For this prospective study, 27 male patients were recruited at the Interdisciplinary Fabry Disease Center (IFAZ) from the University Hospital Muenster. All analyses were performed after approval of the Medical Association of Westphalia-Lippe and the Ethical Committee of the Medical Faculty of the University of Muenster (project-nos.: 2011-347-f; 2011-186-f). Patients gave written informed consent for data analysis and publication. All investigations conformed to the principles outlined in the Declaration of Helsinki.

Blood pressure was measured in a quiet room and sitting position after 5 min of rest with a Mobil-O-Graph device (IEM, Stolberg, Germany), according to current guidelines (15–17). Pulse wave velocity (PWV) measurements were performed according in a quiet room after a 15 min rest in a sitting position by two well trained operators (18, 19). This device is validated according to the British Hypertension Society and European Society of Hypertension recommendations (15, 16, 20, 21). Cuffs were applied at the left upper arms in a sitting position. Microvascular imaging was performed in a supine position in a quiet room by two experienced operators (baseline and follow-up) as reported previously (22). To visualize the sublingual microvasculature, a SDF camera (CapiScope HVCS; KK Technology, Honiton, UK) was used to document at least 10 areas close to the frenulum, as described previously (22) and according to recent guidelines (23). Image acquisition and analysis were performed using the GlycoCheck software (Microvascular Health Solutions, Salt Lake City, UT). Three consecutive sets of measurements were performed for every patient, and individual mean values were calculated for analysis. Vessel density was calculated according to previous reports (22, 24). Patients’ mean PWV, augmentation indices, glycocalyx thickness, perfused boundary region (PBR5-25) and red blood cell (RBC) filling were determined out of three consecutive measurements.

Anti-drug antibodies (ADAs) were determined using the methodology previously described by Lenders and colleagues (25, 26).

Human vascular endothelial cell line EA.hy926 and CRISPR/Cas9-mediated AGAL-deficient EA.hy926 (EA.hy926-KO) generated and described previously (27) were cultivated in Dulbecco’s Modified Eagle Medium (DMEM; Thermo Fisher Scientific) containing 10% fetal calf serum (FCS, Sigma-Aldrich, St. Louis, Missouri, USA), 100 U/ml penicillin, 100 ng/ml streptomycin, and 2 mM L-glutamine (all Capricorn Scientific, Ebsdorfergrund, Germany). THP1 monocytes were cultivated in Roswell Park Memorial Institute 1640 (RPMI; Thermo Fisher Scientific) containing 10% FCS, 100 U/ml penicillin, 100 ng/mL streptomycin, 2 mM/ml L-glutamine, and 1% MEM non-essential amino acid solution (Sigma-Aldrich). CRISPR/Cas9-mediated AGAL knockout in THP1 (THP1-KO) cells was performed as previously described (28). The AGAL knockout resulted in less than 2% of wild-type AGAL activity in both cell lines (EA.hy926-KO and THP1-KO), representing a classical FD phenotype ( Supplemental Figure 1 ).

For nuclear detection of NF-κB 2 × 10⁵ cells/ml were seeded for 3 days and stimulated as stated. In short, cells were harvested in PBS and resuspended in buffer A (1 mM DTT; 10 mM KCl; 10 mM HEPES, pH 7.9; 1.5 mM MgCl2; protease inhibitor cocktail). After 15 min incubation on ice, 10% NP40 was added and centrifuged at 5,000 g for 5 min. Subsequently pellets were resuspended in buffer B (1.5 mM MgCl2; 1 mM DTT; 420 mM NaCl; 0.5 mM PMSF; 0.2 mM EDTA; 20 mM HEPES, pH 7.9; 25% (v/v) glycerol) and incubated for 30 min under frequent vortexing. After a final centrifugation step at 14,000 rpm, the nuclear faction (supernatant) was obtained. 20 µl of the nuclear fraction were separated on a 10% SDS gel. Subsequently, the samples were blotted onto a PVDF membrane and blocked overnight in 5% semi-skimmed milk powder (Roth, Karlsruhe, Germany) in Tris-buffered saline supplemented with 0.1% Tween 20 (TBST; AppliChem, Darmstadt, Germany). Anti-p105/50 (Cell Signaling Technology, Danvers, Massachusetts, USA; 13586, 1:3,000) was used to detect NF-κB and visualized with a secondary horseradish peroxidase-conjugated goat anti-rabbit antibody (Merck Millipore, Burlington, Massachusetts, USA, 12-348, 1:10,000). Loading control was performed with monoclonal PCNA (Santa Cruz Biothechnology, Dallas, Texas USA; sc-56, 1:500).

To detect angiopoietin-2 (Angpt-2) expression, 2 × 10⁵ cells/ml were seeded onto 6 wells. Intracellular angiopoietin 2 expression was detected using an anti-angiopoetin-2 antibody (Abcam, Cambridge, UK; ab155106) combined with a HRP-coupled secondary antibody (anti-rabbit IgG [Merck; no: 12-348]).

Fucose stainings were performed to determine the glycocalyx on the cellular surface. Stainings were quantified by a plate reader and by flow cytometric analyses and visualized by fluorescence microscopy. For flow cytometric analyses, EA.hy926 and AGAL-deficient EA.hy926 cells were seeded at a density of 1.5 × 10⁵ cells/ml on 48-well plates for 72 h. For fluorescence microscopy, 2 × 10⁵ cells/ml of cells were seeded on glass cover slips in 24-well plates for 96 h until confluence. To determine the glycocalyx by a plate reader, 2 × 10⁵ cells/ml were seeded on a 96-well plate (Greiner Bio One, Frickenhausen, Germany; 655090). Treatments of AGAL-deficient EA.hy926 cells were carried out at seeding for indicated times. Cells were treated with 20 µg/ml agalsidase-beta (Sanofi-Genzyme, Amsterdam, Netherlands; 11541667), 0.4 U/ml heparin (Sigma-Aldrich), 10 µM AKB-9778 (razuprotafib; MED Chem Express, Monmouth Junction, New Jersey, USA) for 48 h. Additionally, for long-term treatment AGAL-deficient EA.hy926 cells were cultivated for four to five weeks with 0.4 U/ml heparin or 10 µM AKB-9778 before subsequent experiments. Treatments with 20 nM dexamethasone (dexa; Sigma-Aldrich, D4902), 60 µM N-acetylcysteine (NAC; Sigma-Aldrich; A9165), 100 µM acetylsalicylic acid (ASA; Sigma- Aldrich; A5376) were performed for 96 h. To quantify the amount of glycocalyx by plate reader analyses, cells were co-stained with 5 µg/ml UEA-1-fluorescein isothiocyanate (UEA-1-FITC; Thermo Fisher Scientific, Eugene, Oregon, USA; L32476) and 5 µg/ml Hoechst (Thermo Fisher Scientific, H1399) for 20 min in HBSS at 37°C. After removing the staining solution, 100 µl HBSS per well were added and fluorescence was measured at 459 nm and 515 nm for UEA-1-FITC and at 350 nm and 461 nm for Hoechst (Tecan, Infinite M200, Maennedorf, Switzerland). The fluorescence intensity of UEA-1-FITC was normalized against Hoechst.

To visualize the glycocalyx by fluorescence microscopy, cells cultivated on glass cover slips were washed with HBSS followed by a staining with 5 µg/ml UEA-1 FITC for 20 min at 37°C in HBSS. For microscopy-based NF-κB detection, cells were seeded on glass cover slips. After stimulation, cells were fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton X-100 (Carl Roth, Karlsruhe, Germany; 3051.2) in PBS. Samples were incubated overnight with primary rabbit anti-p105/50 antibody (Cell Signaling Technology; 1:1,000). The secondary antibody was an Alexa594-labeled anti-rabbit IgG (Thermo Fisher Scientific; A11012, 1:2000). Fluorescence microscopy was performed using an Axio Observer Z1 fluorescence microscope (Carl Zeiss, Oberkochen, Germany) with a 63-fold oil objective. Images were processed using the Axiovision software 2.3 (Carl Zeiss).

Cell adherence of monocytes on endothelial cell was additionally visualized by fluorescence microscopy. After removal of non-adherent THP1 monocytes cells were fixed with 4% para-formaldehyde solved in PBS. After permeabilization with 0.2% Triton X100 in PBS, cells were incubated with phalloidin-FITC (Enzo Life Sciences, ALX-350-268-MC01; working concentration: 0.4 µg/ml) and mouse anti-human CD32 allophycocyanin (CD32-APC; BD Bioscience, 559769; working concentration 1:5) in 0.5% bovine serum albumin (BSA) in PBS for 30 min at room temperature to stain filamentous actin (F-actin) and CD32-APC. Fluorescence microscopy was performed using Axio Observer Z1 fluorescence microscope (Carl Zeiss, Oberkochen, Germany) with a 40-fold oil objective. Images were processed using the Axiovision software 2.3 (Carl Zeiss) and ImageJ (v1.51n) (29).

For flow cytometry analyses, EA.hy926 cells were dissolved with enzyme-free cell dissociation buffer (Life Technologies, Bleiswijk, Netherlands; 13151-014) and washed with PBS after indicated cultivation times. After staining of fucose with 5 µg/ml UEA-1-FITC (Thermo Fisher Scientific) in PBS for 20 min at room temperature, cells were washed again. Finally, cells were analyzed using a fluorescence-activated cell sorting (FACS) Canto II flow cytometer (BD Bioscience, Franklin Lakes, New Jersey, USA) and the data were evaluated using BD FACSDiva software (BD Bioscience) and FlowJo (FlowJo LLC, Ashland, Oregon, USA).

To quantify heparanase 1 expression via quantitative real-time-PCR in EA.hy926 and AGAL-deficient EA.hy.926-KO cells, these were seeded in a cell density of 1.5 x 10⁵ cells in 6-well plates and grown for 48 h. Wild-type EA.hy926 cells were stimulated with TNFα (10 ng/ml) for the last 24 h as positive controls. RNA was extracted using the NucleoSpin RNA extraction kit (740955, Macherey-Nagel, Dueren, Germany) according to manufacturer’s protocol. mRNAs were reverse-transcribed into cDNA by using cDNA synthesis kit (Thermofischer, K1612), 1 µg total RNA and Oligo-dT primers according to manufacturer’s instructions. mRNA levels were determined by using QuantiNova SYBR Green PCR kit (QIagen, 208044) on a BioRad CFX384 Real-Time-System C1000 Touch Thermal Cycler. β-actin was used as a reference mRNA and relative expression was calculated with the ΔΔct method. The following primer were used: hACTP sense 5’-CGT GTG GAT CGG CGG CTC-3´, hACTB antisense 5´-TAG GTT TTG TCA AGA AAG GGT G-3´, hHPSE sense 5´-GCA CAA ACA CTG ACA ATC CAA GG-3´, hHPSE antisense 5´-AGG TCC CAA AGG TCT TAG AAG GT-3´.

To assess monocyte adhesion on endothelial cell layers, wild-type EA.hy926 cells were incubated with wild-type THP1 monocytes and AGAL-deficient EA.hy926 cells with AGAL-deficient THP1 monocytes to simulate a healthy or FD situation. Endothelial cells were seeded at a density of 1.5 × 10⁵ cells/ml on glass cover slips in 24-well plates and cultured for 72 h or 96 h until confluence. Treatments of AGAL-deficient EA.hy926 cells were carried out at seeding for indicated times. Treatments with 20 µg/ml agalsidase-beta (Sanofi-Genzyme), 1 µM atorvastatin (PZ0001), 1 µM mevastatin (M2537), or 1 µM simvastatin (S6196) (all Sigma-Aldrich) were conducted for 96 h including a medium change 48 h after seeding. Treatment with 20 nM dexamethasone (Sigma-Aldrich; D4902), 60 µM NAC (Sigma-Aldrich), 100 µM acetylsalicylic acid (ASA; Sigma-Aldrich; A5376), 10 µM AKB-9778 (MEDChemExpress), or 0.4 U/ml heparin (Sigma-Aldrich) was performed for 72 h. Furthermore, AGAL-deficient EA.hy926 cells were treated for four to five weeks with 10 µM AKB-9778 or 0.4 U/ml heparin before appropriate experiments. After indicated treatment time 1.25 × 10⁵ THP1 or AGAL-deficient THP1 monocytes in DMEM containing 10% FCS, 100 U/ml penicillin, 100 ng/ml streptomycin, and 2 mM L-glutamine were added to the endothelial monolayers. Cells were incubated for 30 min at 37°C before non-adherent monocytes were removed by gently transferring the cover slips twice to new wells containing fresh PBS.

To quantify the amount of adherent THP1 monocytes (wild-type or AGAL-deficient) flow cytometry analyses were performed. Therefore, adherent THP1 monocytes and EA.hy926 cells were dissociated using enzyme-free cell dissociation buffer (Life Technologies, Bleiswijk, Netherlands; 13151-014). Subsequently, cells were washed with 0.5% BSA in PBS and incubated with mouse anti-human CD32 APC (BD Bioscience) in 0.5% BSA in PBS for 30 min at 4°C. After a final washing step with 0.5% BSA in PBS, stained monocytes were detected using a FACS Canto II flow cytometer. Data were analyzed using BD FACSDiva software and FlowJo.

To quantify the amount of sulfated glycosaminoglycans (GAG) in the urine dimethylmethylene blue (DMMB) assays modified from Sun and colleagues (30) were performed. Urine samples were thawed on ice. Chondroitin 4 sulfate (Sigma-Aldrich; 27042) served as standard with a calibration curve ranged from 0 to 50 µg/ml solved in H₂O. 25 µl of the urine samples per 96-well were incubated with 200 µl of DMMB solution (for 1 liter: 16 mg DMMB (Sigma-Aldrich; 341088), 3.05 g glycine, 1.6 g sodium chloride, 544 µl acetic acid, pH 4). The GAG concentration of each urine sample was calculated from the chondroitin 4 sulfate standard curve and then normalized to urine creatinine concentration. Samples out of range were repeated in serial dilution.

All experiments were performed at least three times. Continuous variables were expressed as mean with standard deviation (SD). Unpaired two-tailed Student’s t-test or one-way analysis of variance (ANOVA) with correction for multiple testing was used for statistical analysis where appropriate. Analyses and visualizations were performed with GraphPad PRISM v8.0 software (GraphPad Software Inc., La Jolla, CA, USA).

To analyze potential vascular and endothelial damage in patients with FD, we performed pulse wave velocity (PWV), augmentation index and glycocalyx measures in ERT- (n=16) and migalastat-treated (n=6) male patients at two consecutive follow-up visits within 12 months (visit T1 and visit T2). Untreated males (n=5) carrying genetic variants of unknown significance (GVUS) within the GLA gene (p.S126G and p.A143T) with enzymatic AGAL activities within the reference range, normal lyso-Gb₃ values, lacking FD-typical manifestations and symptoms served as controls. Baseline characteristics are presented in Table 1 . Patients under ERT were not treatment-naïve and treated with ERT for a mean duration of 102 ± 64 months at first (T1) visit. In contrast, two patients (33.3%) treated with migalastat were treatment-naïve at T1 and the remaining four patients (66.7%) received ERT for 32 ± 29 months before being switched to migalastat ( Table 1 ). Migalastat-treated FD patients were significantly older than ERT-treated patients (p<0.05; Table 1 ). No significant differences for eGFR were observed between the three groups. The interventricular septum thickness was significantly increased in migalastat-treated patients compared to untreated controls and ERT-treated patients ( Table 1 ). This cardiac involvement in migalastat-treated patients might be due to the increased age and higher central systolic blood pressure (cSBP) as well as the high frequency of the cardiac variant p.N215S (66.7%) in this group ( Table 1 ). PWV and augmentation index did not differ between the three groups and were stable over time ( Figures 1A, B ). The perfused boundary region (PBR) as an inverse marker for glycocalyx thickness in microvascular vessels remained stable in untreated control patients, but decreased significantly in ERT- and migalastat-treated patients, demonstrating an increase in glycocalyx thickness over time ( Figure 1C ). Increase of glycocalyx was more prominent in migalastat-treated patients compared to those under ERT ( Figure 1D ). Red blood cell (RBC) filling also improved over time in treated patients ( Figure 1E ) and was more prominent in patients receiving migalastat, too ( Figure 1F ).

To further assess an impact of AGAL-deficiency on endothelial cell function, we used a stable AGAL-deficient endothelial cell line (EA.hy926-KO) (27) to analyze the glycocalyx composition and its function. AGAL-deficient cells demonstrated significantly less L-fucose and N-acetylglucosamine (GlcNAc) on their cellular surface (p<0.001, Figures 2A, B ), which are major components of the glycocalyx. In line with the FD patient-derived data, treatment of EA.hy926-KO cells with recombinant AGAL (20 mg/ml) increased L-fucose and GlcNAc on AGAL-deficient cells by 8.0% and 10.0%, respectively (both p<0.001, Figure 2B ). An important process for glycocalyx degradation is the expression and release of heparanases due to an inflammatory stimulus. To test whether heparanases might be involved in glycocalyx degradation in our model system, we analyzed heparanase I expression via qPCR. Indeed AGAL-deficient EA-hy926-KO cells showed an increased (6-fold) heparanase I expression compared to wild-type ( Supplemental Figure 2A ). Consequently, AGAL-deficient cells were treated with heparin, which is known to inhibit heparanase activity to analyze whether heparanases might be responsible for the observed glycocalyx reduction in FD. Indeed, treatment of AGAL-deficient cells with heparin for 48 h resulted in an increase of the glycocalyx by 11% (p=0.0013; Figure 2C ). In addition, treatment with anti-inflammatory drugs (NSAIDs, dexamethasone) as well as N-acetylcysteine (NAC) as an antioxidant for 72 h showed comparable effects (p<0.001 and p=0.0154, respectively; Figure 2D ). To further demonstrate that secreted enzymes (i.e. heparanases) are involved in the observed glycocalyx degradation, wild-type EA.hy926 cells were treated with supernatants from AGAL-deficient EA.hy926 cells ( Supplemental Figure 2B ). Treatment for 30 minutes with supernatants from AGAL-deficient cells resulted in a significant decrease of glycocalyx (p=0.0007), which could be reversed by a co-incubation with heparin (p=0.0225; Supplemental Figure 2B ).

Since a reduction of the glycocalyx facilitates monocyte adhesion and invasion, we additionally analyzed monocyte adhesion on endothelial cells. To simulate a comprehensive AGAL-deficient background comparable to a situation in vivo, we generated CRISPR/Cas9-mediated AGAL-deficient THP1 monocytes based on a previously established method (28). Of note, only THP1 monocytes were positive for CD32 expression and showed no differences between wild-type and AGAL-deficient cells, allowing a proper quantification by flow cytometry analysis ( Supplemental Figure 3A ). In an AGAL-deficient background (i.e. AGAL-deficient EA.hy926 cells combined with appropriate THP1-KO monocytes), monocyte adhesion was increased compared to wild-type (p<0.0001; Figures 2A, E ). Furthermore, the stimulation of wild-type as well as AGAL-deficient endothelial cells with TNFα demonstrated that both cell lines were yet able to react on an external inflammatory stimulus ( Supplemental Figure 3B ). Treatment of AGAL-deficient EA.hy926 cells with recombinant AGAL for 96 h significantly reduced monocyte adhesion (p=0.0045; Figure 2E ). AGAL-treatment of both AGAL-deficient cell lines (EA.hy926-KO and THP1-KO) did not result in a more prominent monocyte adhesion, suggesting that the endothelial cells are responsible for the observed increased adhesion, rather than the monocytes (p=0.9348; Figure 2E ). Treatment with heparin for 4 or 72 h resulted in a decreased monocyte adhesion to AGAL-deficient EA.hy926 cells (both p<0.001; Figure 2F ). Vice versa, the incubation of wild-type EA.hy926 cells with conditioned supernatants of AGAL-deficient cells resulted in an increased (p=0.0005), heparin-treatment-reversible monocyte adhesion (p=0.0005) on wild-type EA.hy926 cells, also pointing towards an involvement of secreted heparanases ( Supplemental Figure 2C ). Incubation with anti-inflammatory drugs (dexamethasone and acetylsalicylic acid [ASA]) and the antioxidant NAC all resulted in a decreased monocyte adhesion in an AGAL-deficient background, too ( Figure 2G ). However, monocyte adhesion is not only mediated by glycocalyx thickness, but also by monocyte-adhesion receptors. Thus, we analyzed if statins, which are known to reduce monocyte adhesion by for example CD11b inhibition (31), also decrease cell adhesion in a FD background. Treatment with atorvastatin, mevastatin, and simvastatin for 96 h reduced monocyte adhesion, significantly (all p<0.01; Figure 2H ).

Recent studies demonstrated that heparanase release is mediated by angiopoietin-1-receptor (Tie2) signaling, in that Angpt-2 mediates the binding of vascular endothelial protein tyrosine phosphatase (VE-PTP) to Tie2, followed by a release of heparanases. Razuprotafib (AKB-9778) is a novel Tie2 activator preventing the binding of VE-PTP to Tie2 and thus the release of heparanases (32, 33). Consequently, we analyzed whether AKB-9778 also inhibits a heparanase release in an AGAL-deficient background. Indeed, treatment of AGAL-deficient EA.hy926 cells with 10 µM AKB-9778 resulted in an increase of L-fucose (48 h) and a decreased monocyte adhesion (72h), too (both p<0.001; Figures 2I, J ), indicating that Tie2 and Angpt-2 are also involved in the observed glycocalyx degradation in FD. An involvement of Angpt-2 was supported in that AGAL-deficient EA.hy926 cells revealed an increased Angpt-2 expression compared to wild-type (p=0.0272, Figure 2K ).

Previous studies demonstrated that lyso-Gb₃ may trigger inflammatory signaling in cell culture (34). To analyze a potential direct effect of lyso-Gb₃ on glycocalyx thickness, wild-type EA.hy926 cells were treated with a pathological lyso-Gb₃ concentration of 200 nM. Interestingly, treatment for 4 or 24 h resulted in a decrease of glycocalyx (both p<0.001; Figure 3A ), suggesting a rapid, lyso-Gb₃-mediated heparanase release. Since AGAL-deficient EA.hy926 cells showed an increased NF-κB (p50) translocation ( Supplemental Figure 4 ), we further analyzed whether pathologic lyso-Gb₃ levels may activate nuclear translocation of NF-κB, too. Strikingly, treatment of wild-type EA.hy926 cells with 200 nM lyso-Gb₃ for 4 h resulted in increased nuclear p50 signals determined by fluorescence microscopy ( Figure 3B ) and western blots ( Figure 3C ).

Our cell culture experiments pointed to an increased glycocalyx degradation by inflammatory processes due to FD. Recent studies demonstrated measurable changes in urinary GAG levels in patients suffering from sepsis. To analyze if a continuous degradation of the glycocalyx in FD patients under FD-specific treatment might also be detectable, we additionally measured GAGs in urine samples from a subset of treated patients at two time-points. At the first time-point (T0) all patients were naïve to any FD-specific therapy. The second measurements were performed in urine samples from the control group and ERT- or migalastat-treated patients from the last (current) visit (T2), which is the same visit as for glycocalyx measurements. Urinary GAG levels were in all groups and all time points comparable and thus constant ( Figure 4 ).