Primary human aortic endothelial cells (HAOECs) were purchased from PromoCell (Heidelberg, Germany, c-12271, LOT#4082102.16). Cells were cultured at 37°C and 5% CO₂ in a humidified incubator in endothelial cell growth medium MV (PromoCell, Heidelberg, Germany, C-22020) and endothelial cell supplement mix (PromoCell, Heidelberg, Germany, c-39225). For all experiments, the cells were used at passage 4 to 6. Cells were grown to 100% confluence and then stimulated with Ang-II (Sigma Merck, A9525), sodium butyrate (Millipore, 567430), PP2 (Sigma, 529573), GLPG0974 (Sigma, SML2443), and ß-HB (β-hydroxybutyric acid) (Sigma, 166898). Sodium butyrate was resolved in pH-buffered endothelial cell basal medium (PromoCell, c-22220) without supplements (pH ±SD at 37°C and 5% CO₂ measured in triplicate by a pH meter: pH_{0.1mM BUT} = 7.36 ± 0.04; pH_{1mM BUT} = 7.38 ± 0.02; and pH_{5mM BUT} = 7.35 ± 0.05). Control cells were treated only by basal medium (pH = 7.36 ± 0.03) without a vehicle.

Tissue of human thoracic aorta was obtained during surgery and fixed in 4% PFA immediately after resection for immunohistochemistry (see Immunohistochemistry). Patient consent for the collection and use of the samples was obtained in advance, based on a positive vote by the ethics committee.

We used the following commercially available antibodies for western blotting and immunofluorescence microscopy: monoclonal mouse anti-p-Tyr (PY99, Santa Cruz, sc-7020), polyclonal rabbit anti-phospho-VEC (against phospho-Tyr731, Invitrogen 44-1145G) (Luo et al., 2017), polyclonal rabbit anti-phospho-VEC (against phospho-Tyr685, Abcam, ab119785) (Liu et al., 2021), polyclonal rabbit anti-phospho-VEC (against phospho-Tyr658, Invitrogen, 44-1144G) (Hahn et al., 2015), monoclonal mouse anti-human VEC (clone F-8, Santa Cruz, sc-9989), monoclonal mouse anti-human VEC (BV6, Millipore, MABT134), monoclonal mouse anti-human SRC (Invitrogen, AHO1152), monoclonal rabbit anti-human phospho-SRC (against phospho-Tyr 416 (D49G4), Cell Signaling, 6943), monoclonal rabbit anti-human VEC (E6N7A, Cell Signaling, 93467S), and monoclonal mouse anti-human ß-actin (Cell Signaling, 3700). The secondary antibodies include goat anti-rabbit IgG HRP-linked antibody (Cell Signaling, 7074), horse anti-mouse IgG HRP-linked antibody (Cell Signaling, 7076), polyclonal goat anti-mouse IgG (H + L) cross-adsorbed Alexa Fluor 488 (Invitrogen, A11001), goat anti-mouse IgG (H + L) Cy3 (Jackson ImmunoResearch, 111-165-062), and donkey anti-rabbit IgG (H + L) Alexa Fluor 488 (Invitrogen, A21206).

Cells were lysed in RIPA lysis buffer (150 mM NaCl (AppliChem, 141659.1211), 50 mM Tris-HCl (pH 8.0) (Roth, 4855,1), 1% Triton X-100 (AppliChem, A4975), 0.5% w/v sodium deoxycholate (AppliChem, A1531.0100), 0.1% w/v SDS (AppliChem, A1112.0500) supplemented with protease inhibitor cocktail (Sigma, P8340), and sodium orthovanadate (Sigma, 6508)). After incubating on ice for 1 h, the cell lysates were cleared by centrifugation at 20,817 rpm for 10 min at 4°C. The concentration of total proteins was measured using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23225). The lysates were mixed with the sample buffer (250 mM Tris-HCl, pH 6.8) (Roth, 4855.2), 30% v/v glycerin (Fisher Scientific, 10021083), 10% w/v SDS, 0.5 M 1,4-dithiothreitol (DTT, Roth, 6908.1), and 0.5% w/v Bromphenolblau (Serva, 15375.02)) and heated at 100°C for 10 min. 10 µg of cell lysates were loaded on 10% polyacrylamide gel (SDS-PAGE) (resolving gel (2.5 mL running buffer (1.5 M Tris-HCL, pH 9.0), 4% v/v TEMED (Roth, 2367.1), 0.4% w/v SDS), 4.02 mL Millipore water, 3.38 mL Rotiphorese^® Gel 30 (Roth, 3029.2), 45 µL 10% v/v ammonium persulfate (Roth, 9592.5), stacking gel (4.44 mL stacking buffer (140 mM Tris-HCl (PH 6.8), 0.1% v/v TEMED, 0.1% w/v SDS), and 650 µL Rotiphorese^® Gel 30, 100 µL 10% v/v ammonium persulfate) and electro-transferred into a 0.45-µm nitrocellulose membrane (Roth, 9201.1). The membranes were then incubated with 10% low fat milk (MILSANI, Germany) in TBST (TBS containing 0.05% v/v Tween-20 (AppliChem, A4974) at room temperature for 1 h. Primary antibodies (phospho-Tyr731(1 μL/mL), phospho-Tyr658 (2 μL/mL), phospho-Tyr658 (2 μL/mL), VEC (F-8) (0.2 μg/ml), ß-actin (0.5 μL/mL), and SRC (1 μg/ml) were diluted in 10% w/v low fat milk in TBST and incubated at 4°C overnight. After washing three times for 10 min in TBST, the membranes were incubated with secondary antibodies for 45 min at room temperature, followed by detection by chemiluminescence (Clarity ECL; BioRad, 1705061) and imaging on a ChemiDoc Imaging System (BioRad). Protein abundance was quantified by densitometry (Image Lab software version 6.0.1) and subsequently normalized to a loading control. Phosphorylated VEC was detected by an HRP-linked secondary antibody (0.5 μL/mL), while the total VEC was detected by the cross-adsorbed Alexa Fluor 488 secondary antibody (4 μg/ml).

HAOECs were seeded on top of a polyester membrane transwell insert on 12-well-plates (12 mm Transwell^® with 0.4-μm Pore Polyester Membrane Insert, Corning, #3460/TC-Inserts, 0.4 mM Pore, Sarstedt, 83.3931.041 for knockdown experiments) at a density of 0,8 × 10⁵ cells/chamber and cultured for 72 h. After discarding the medium, the cells in the upper chamber were incubated with different concentrations of BUT as indicated. Subsequently, the medium in the apical chamber was replaced by endothelial cell basal medium containing 0.05 mg/ml of 10 kDa dextran (Invitrogen, D22910) followed by incubation for 30 min at 37°C in a CO₂ incubator, after which the fluorescence (F) intensity in the upper and lower chambers was determined with a 96-well plate reader (Synergy HTX Multifunction Detector, BioTek, United States: excitation 485/20 and emission 528/20). All independent experiments were performed in triplicate. Permeability was presented as F_basal/F_apical (F_b/F_a). After VEC was knocked down by specific siRNA, permeability was measured as described previously and experiments were performed in duplicate.

In brief, HAOECs were seeded into a 24-well plate (1 × 10⁵ cells/well) on glass coverslips for 48 h. Confluent monolayers of HAOECs were treated with 1 mM of BUT for 1 h. Afterward, HAOECs were fixed with 4% v/v paraformaldehyde in PBS (Roth, 0335.3) for 1 h and incubated with 50 mM NH₄CL (Roth, K298.1) for 45 min at room temperature. HAOECs were incubated with the anti-VEC antibody (E6N7A and BV6, respectively) diluted 1:100 in non-permeabilizing blocking solution (5% w/v Albumin Fraction V (Roth, T844.1) in PBS) or anti-phospho-tyrosine antibody (PY99) diluted 1:100 in permeabilizing blocking solution (containing 0.3% v/v Triton^® X-100, AppliChem, A4975,0500) for 60 min at room temperature. Coverslips were washed three times with PBS. HAOECs were incubated with a secondary antibody of Alexa Flour 488 anti-mouse IgG (H + L) or Alexa Fluor 488 donkey anti-rabbit IgG (H + L) and goat anti-rabbit IgG (H + L) Cy3 diluted 1:300 for 45 min, washed three times with PBS, and then incubated for 5 min at room temperature with a mounting medium (Dako, S3023). Images were acquired using a fluorescence microscope (Nikon, ECLIPSE Ti2) and NIS-Elements AR 5.02.03 and NIS-Elements Viewer 5.21. For quantification of junction phenotypes, we used the program junction mapper (Brezovjakova et al., 2019). After the map was created, all cell boundaries of representative cells were measured by Junction Mapper. The thresholds for the VEC signal were set to zero signals in the cytoplasm. Only completely sharp boundaries with continuous contact with adjacent cells were included into the analysis. Data were shown from at least three independent experiments.

HAOECs and tissue of human thoracic aorta were fixed in 4% PFA/PBS and embedded in paraffin. Before paraffin embedding, 2.5 × 10⁶ HAOEC in 50 µL PBS were mixed in an Eppendorf tube with 50 µL of 10 mg/mL agarose (Sigma, A9539-25G) in PBS at 42°C. Agarose-embedded HAOECs were placed in a histological cassette for further paraffinization. Sections of 4 µm were cut with a rotary microtome (Leica, Germany). Histological sections were dried overnight at 37°C. After deparaffinization and dehydration, the sections were pretreated with target retrieval solution (Dako, S1699, pH 6.1) for 40 min in a steamer. Intrinsic peroxidase was blocked with 3% H2O2/PBS. FFAR3 primary antibodies (GPR41 Rabbit anti-human, Polyclonal, Invitrogen, and PA599629) and FFAR2 (GPR43 Rabbit Polyclonal Antibody, Invitrogen, and PA5-100944) were diluted in antibody diluent (Dako REAL, S2022). Sections were incubated with a primary antibody overnight at 4°C in a humid chamber. Sections were then incubated with a polymer-enhancer and HRP-polymer from the SuperVision 2 HRP Kit (DCS, PD000POL) for 1 h, respectively. Diaminobenzidine tetrahydrochloride hydrate (DAB, Sigma, D5637) was used as a color substrate and nucleus staining was performed with hematoxylin. Finally, the sections were mounted using the Eukitt^® mounting medium (Kindler). Histological sections were analyzed using an upright microscope (Nikon, ECLIPSE Ci) and images were acquired using NIS-Elements F 4.60.00 and NIS-Elements Viewer 5.21 software. For quantification of FFAR3 expression, images were converted into 8-bit TIFs and cell borders of cells in the focal plane were outlined manually, defining the ROI. Mean signal intensities of ROIs were extracted using ImageJ.

The MTT assay was used to determine the cell viability after adding BUT using the MTT Assay Kit by Promega (G4000, including Solubilization Solution/Stop Mix G401A and Dye Solution G402A). HAOECs were seeded into a 96-well plate at a density of 1 × 10⁴ cells per well in 100 μL of the complete medium and grown to confluence. Each group contained three replicates. The cells were treated with various doses of BUT (0.1 mM, 1 mM, and 5 mM) for 1 h at 37°C and 5% CO₂ in a humidified incubator. Afterward, 15 μL dye solution was added to each well and the cells were incubated for 4 h at 37°C in a CO₂ incubator. Crystals were solubilized by adding 100 µL Solubilization Solution/Stop Mix for 1 h. The content of each well was mixed to obtain a uniformly colored solution by using a multichannel pipette. Absorbance at 570 nm wavelength was measured using a plate reader (BioTek).

HAOECs were seeded into a 12-well plate (1.2 × 10⁵ cells/well) and the cell density reached 70%–80% confluence before transfection with siRNA. The following reagents were used for transfection: control siRNA-A (Santa Cruz, sc-37007), siRNA transfection medium (Santa Cruz, sc-36868), siRNA transfection reagent (Santa Cruz, sc-29528), siRNA dilution buffer (sc-29517), VE-cadherin siRNA (h) (Santa Cruz, sc-36814), GPR41 siRNA(h) (Santa Cruz, sc-97148), and siRNA c-Src (h) (Santa Cruz, sc-29228). The cells were transfected according to the manufacturer’s instructions (Protocol “siRNA Mediated Inhibition of Gene Expression” by Santa Cruz). In brief, control siRNA-A and target siRNA were diluted in 66 μL and 330 µL siRNA dilution buffer, respectively. 4 µL of each siRNA duplex solution and transfection reagent, respectively, were added into 100 µL of the transfection medium, gently mixed, and incubated for 30 min. 792 µL of the transfection medium was added to each tube containing the siRNA and reagent, mixed gently, and incubated with cells. After culturing at 37°C and 5% CO₂ for 5.5 h, the transfection medium was replaced by endothelial cell growth medium, and experiments were performed between 12 and 24 h after transfection.

The total RNA of 80×10⁵ HAOECs was extracted using the RNeasy Mini-Kit (50) (QIAGEN, 74104) according to the manufacturer’s instructions. 100 ng of total RNA was reverse-transcribed using the iScript™ Advanced cDNA Synthesis Kit (Bio-Rad, 1725037) at 46°C for 20 min and 95°C for 1 min in a thermal cycler (Mastercycler nexus gradient, Eppendorf). 100 ng of cDNA was used for qPCR performed with SsoAdvanced Universal SYBR^® Green Supermix (Bio-Rad, 1725270) according to the manufacturer’s instructions with 1 µL of primers for FFAR2 (Bio-Rad, qHsaCED0044139), FFAR3 (Bio-Rad, qHsaCED0037214), and Glyceraldehyde 3-phosphate dehydrogenase (GAPDH, Bio-Rad, qHsaCED0038674). Samples were measured using the CFX96 system (Bio-Rad) and analyzed using BioRad Analysis Software (Maestro). Relative mRNA levels of FFAR3 were normalized to GAPDH. PCR products were subjected to agarose gel electrophoresis (75V, 60 min) on a 1% agarose gel stained with GelStain-Red dye (Roth, 0984.1).

Statistical analysis was carried out with GraphPad Prism 9 software (GraphPad Software, San Diego, CA, United States). All data are presented as the mean ± SEM. Data were analyzed for normality and equal variance by the Shapiro–Wilk test. To compare between multiple experimental groups, non-parametric data were analyzed with the Kruskal–Wallis test, while parametric data were analyzed with one-way analysis of variance (ANOVA), followed by Tukey’s post hoc analysis. An unpaired t-test (Welch´s t-test) was used to compare two independent samples (Figure 3C). Statistical significance was considered for p ≤ 0.05 (*), p ≤ 0.005 (**), p ≤ 0.001 (***), and p ≤ 0.0001 (****).

Since previous studies have demonstrated the impact of BUT on the endothelial barrier function, we first tested the capacity of BUT to alter the permeability of the HAOEC monolayer. Therefore, HAOECs were exposed for 1 h to different concentrations of BUT from 0.1 to 5 mM in serum-free growth medium (Figure 1). Of note, BUT induced a measurable increase in EC monolayer permeability at 1 mM characterized by a significant increase in FITC-Dextran diffusion (10 kDa) from the upper to the lower compartment of the filter inserts. At 1 mM and 5 mM BUT, diffusion was increased by about 1.24 ± 0.05 fold and 1.18 ± 0.06 fold, respectively, whereas 0.1 mM BUT moderately decreased EC permeability, compared to the medium (Figure 1A). A knockdown (KD) of VEC expression by siRNA (Figure 1B) significantly increased EC permeability per se in contrast to cells transfected with non-targeting siRNA (Figure 1C), indicating that the integrity of the HAOEC monolayer relies on VEC expression. Of note, under the VEC knockdown condition, BUT was still able to increase the permeability, but the magnitude given by the difference between BUT-treated and –untreated cells (0.1663) is lower than that of cells transfected with non-targeting siRNA (0.2656) (Figure 1C). These findings indicate that BUT exerts its effect on aortic endothelial permeability via VEC probably in addition to other pathways.

Since VEC mediates BUT-induced effects on HAOEC permeability, we next checked the localization of VEC at the cell junctions by fluorescence microscopy to determine whether junctional VEC was modified in response to BUT. We observed an altered VEC localization in HAOECs in response to BUT-treated cells, which resembled interrupted, plaque-like patterns for VEC, suggesting a remodeling of junctional VEC by BUT (Figure 1D).

We further characterized the patterns of VEC by analyzing the images obtained by immune fluorescence microscopy using Junction Mapper software (Brezovjakova et al., 2019). Junction Mapper identifies cellular junction semi-automated and analyzes the length, intensity, and distribution of junctional protein staining. To compare morphology in BUT-treated cells to the control, we used indices not affected by the intensity to prevent influence of variance of staining. The Coverage Index measures the level of stained fragments along the cellular junction, the Junction Linearity Index measures the deviation from a straight-line junction, and the Interface Occupancy measures the share of stained area in a fixed area around the junction.

The analysis confirmed a significantly higher degree of fractionated, junctional VEC (Coverage Index) in BUT-treated cells (1 mM) than in control cells (Figure 1E). Moreover, the linearity of VEC-distribution at cellular junctions (Junction Linearity Index) of BUT-treated cells (1 mM) is decreased compared to control cells (Figure 1E).

Interestingly, increased permeability and junction remodeling occurred not due to potential toxic effects of BUT since under all experimental conditions we did not measure relevant effects on cell viability and toxicity (Supplementary Figure S1). Because these results hint on a direct impact of BUT on VEC, we subsequently examined whether distinct phosphorylation sites of VEC, which are critical for its regulation, are affected by BUT treatment.

We analyzed the phosphorylation status of three tyrosine residues of VEC, which have been reported to regulate endothelial permeability: tyrosine-731, -685, and -658. Therefore, we incubated HAOECs under the same conditions described for the permeability experiments (Figure 1A) with 0.1 mM, 1 mM, and 5 mM BUT and analyzed the tyrosine-specific phosphorylation of VEC in cell lysates by western blotting followed by densitometry quantification. Thereby, we used validated antibodies which detect specific phosphorylation of VEC at tyrosine-731, -685, and -658 (Hahn et al., 2015; Luo et al., 2017; Liu et al., 2021). We found that BUT affects the phosphorylation levels of all target tyrosines to varying degrees. The largest effects could be observed for tyrosine-731, where BUT led to elevated phosphorylation compared to medium-only treated cells (Figure 2A). VEC phosphorylation for tyrosine-731 peaked at 1 mM of BUT (Figure 2A, far left panel). Following the analysis of the time course of phosphorylation revealed that tyrosine-731 was transiently phosphorylated in response to BUT (1 mM) which peaked at 1 h of treatment by 2.40 ± 0.35 fold, whereas the phospho-level at 0.5 h and 2 h was just moderately increased compared to the baseline level (red dashed line, Figure 2B). The phospho-VEC-levels of tyrosine-685 and -658 did not significantly change during the observation period compared to medium-only treated cells (Supplementary Figure S2). This suggests that BUT exerts a site-specific effect on VEC. Interestingly, phosphorylation of VEC did not result in the decreased amount of full-length VEC at approx. 140 kDa, which would indicate significant degradation (Figure 2B, lower right panel).

Ang-II is an important angiogenic molecule with a broad impact on vascular (patho-) physiology. Among others, Ang-II influences EC permeability and VEC phosphorylation (Wu et al., 2016). Therefore, we analyzed whether BUT may interfere with or enhance the effects of Ang-II regarding VEC phosphorylation. Indeed, we observed an elevated level of phospho-tyrosine-731 in Ang-II (100 ng/ml)-treated cells (Figure 2C) in line with previous reports (Jeong et al., 2019). Interestingly, BUT was able to increase this level when co-incubated with Ang-II (Figure 2C), suggesting a synergistic rather than an interfering effect of BUT on Ang-II function with respect to VEC phosphorylation.

In summary, BUT stimulated a measurable increase in tyrosin-731 phosphorylation at VEC in a time- and dose-dependent manner, which correlated with an increased permeability of the HAOEC monolayer (Figure 1).

FFAR2 and FFAR3 bind SCFAs including BUT and trigger intracellular signaling pathways. Therefore, we experimentally addressed the question of whether FFAR2 and FFAR3 are involved in the BUT-induced VEC phosphorylation in HAOECs. Detection of FFAR2 and FFAR3 by immunohistochemistry and RT-PCR using commercially available antibodies and primers confirmed the expression of these receptors in HAOECs (Figure 3A left panel) and in the intima of human aorta (Figure 3A right panel, red arrowheads). We initially perturbed the receptor function using an antagonist of FFAR2 (GLPG0974) and FFAR3 (β-HB) (Figure 3B). In the presence of GLPG, phospho-tyrosine-731 significantly dropped in HAOEC, whereas in the absence of the antagonist, BUT increased the phospho-tyrosine-731 level (Figure 3B, left panel). When BUT was co-incubated with β-HB, the resulting phospho-tyrosine-731 level was decreased to the basal level as observed for GLPG (Figure 3B, right panel). Since BUT has higher affinity for FFAR3 than for FFAR2 (Brown et al., 2003), we subsequently performed a siRNA-mediated knockdown of FFAR3 and tested if BUT was still able to elevate the phospho-tyrosine-731 level of VEC as observed previously (Figure 2A). Even when FFAR3 expression was only moderately knocked down (Figure 3C, Supplementary Figure S3), BUT failed to efficiently induce the phosphorylation of tyrosine-731, in contrast to cells properly expressing FFAR3 (Figure 3D). This raises the question if BUT-induced junctional remodeling of VEC (Figure 1D) also relies on FFAR3 expression. Indeed, we observed that BUT was less efficiently able to remodel junctional VEC in cells treated with FFAR3 siRNA. Under such a condition, VEC localization principally resembled those in BUT-unstimulated HAOECs (Figure 3E), indicating a role for FFAR3 not only in BUT-induced phosphorylation but also for junctional remodeling of VEC.

These findings suggest that FFAR2/3 is involved in the signaling pathway engaged by BUT that finally leads to the phosphorylation and junctional remodeling of VEC.

c-Src kinase has been reported to phosphorylate specific tyrosine residues at VEC to regulate the stability of VEC complexes and endothelial permeability. We initially detected that the basal phosphorylation of VEC at tyrosine-731 was significantly lowered upon the application of the broadband inhibitor of Src family kinases PP2 (Figure 4A). This finding confirmed that c-Src kinases are mainly involved in the phosphorylation of VEC also in aortic EC. When PP2 was applied in combination with BUT, phospho-tyrosine-731 remained at the level observed for PP2-only treated HAOECs (Figure 4A), whereas BUT alone was still able to induce VEC phosphorylation (Figure 4A) to a similar extent as observed previously (Figure 2). This suggests that BUT likely activates Src family kinases to phosphorylate VEC. Interestingly, we found by immune-fluorescence microscopy that an altered junctional VEC pattern in response to BUT (Figure 4B middle panel; Figure 1D) was partially reversed and resembled VEC junctions in the control approach, when BUT was co-incubated with PP2 (Figure 4B, lower panel). Moreover, BUT increased the general phospho-tyrosine level in a c-Src-dependent manner, which was partially co-localized with junctional VEC compared to PP2-untreated cells, (Figure 4B, zoom-in panel). Finally, we tested by specific KD of c-Src expression (Figure 4C) if c-Src indeed mediates BUT-induced VEC phosphorylation. In HAOECs not stimulated with BUT, the c-Src KD condition per se led to a similar decrease in the basal phosphorylation of VEC at tyrosine-731 as observed for PP2 (Figure 4D). When HAOECs were stimulated with BUT upon KD of c-Src, BUT was still able to increase the phosphorylation of VEC at tyrosine-731 but to a significantly lower level than in cells transfected with non-targeting siRNA (Figure 4D). The findings demonstrate that c-Src is, indeed, engaged by BUT to phosphorylate VEC at tyrosine 731. Since our data suggest that Src kinase activation is required to phosphorylate VEC, we next analyzed if Src becomes activated downstream of FFAR3. Therefore, we probed cell lysates of FFAR3 KD and control cells (Figure 3D) for Src activation by detecting phosho-Src^Y416 (Adam et al., 2010). Thereby, we found an increased level of phospho-Src^Y416 in response to BUT treatment (Figure 4E, CTRL siRNA), whereas the phosphorylation of Src was less pronounced by BUT in FFAR3 KD cells (Figure 4E, FFAR3 siRNA). This indicates that FFAR3 is required for Src activation.

In summary, our results suggest that BUT may influence specific VEC phosphorylation by activating a signaling cascade in aortic EC triggered by the binding of BUT to FFAR2/3 which results in the activation of c-Src as an effector kinase to phosphorylate and remodel junctional VEC. These processes presumably lead to the observed increased endothelial permeability of aortic EC monolayer under BUT treatment.