Changes in endothelial glycocalyx layer protective ability after inflammatory stimulus

Abstract

Leukocyte adhesion to the endothelium is an important early step in the initiation and progression of sepsis. The endothelial glycocalyx layer (EGL) has been implicated in neutrophil adhesion and barrier dysfunction, but studies in this area are few. In this report, we examine the hypothesis that damage to the structure of the EGL caused by inflammation leads to increased leukocyte adhesion and endothelial barrier dysfunction. We used human umbilical vein endothelial cells enzymatically treated to remove the EGL components hyaluronic acid (HA) and heparan sulfate (HS) as a model for EGL damage. Using atomic force microscopy, we show reductions in EGL thickness after removal of either HA or HS individually, but the largest decrease, comparable with TNF-α treatment, was observed when both HA and HS were removed. Interestingly, removal of HS or HA individually did not affect neutrophil adhesion significantly, but removal of both constituents resulted in increased neutrophil adhesion. To test EGL contributions to endothelial barrier properties, we measured transendothelial electrical resistance (TEER) and diffusion of fluorescently labeled dextran (10 kDa molecular weight) across the monolayer. Removal of EGL components decreased TEER but had an insignificant effect on dextran diffusion rates. The reduction in TEER suggests that disruption of the EGL may predispose endothelial cells to increased rates of fluid leakage. These data support the view that damage to the EGL during inflammation has significant effects on the accessibility of adhesion molecules, likely facilitates leukocyte adhesion, and may also contribute to increased rates of fluid transport into tissues.

INTRODUCTION

Systemic inflammatory response syndrome, or sepsis, is the number one cause of death in intensive care units in the United States (1–3). Sepsis causes multisystem organ failure due to severe vascular damage. The vascular degradation is associated with extensive leukocyte interaction with the endothelium triggered by inflammatory mediators (4, 5). Despite a significant established body of work on the immune system’s response to inflammation, there are few specialized treatments for the severe inflammatory response that occurs during sepsis (6, 7). This is in turn because mechanisms by which the endothelium’s innate barrier properties change under a systemic infection are not well understood.

An important structure on the apical side of the endothelium that helps modulate leukocyte-endothelial interactions is the endothelial glycocalyx layer (EGL). The EGL is composed of a soft, dense mesh of proteoglycans and glycoproteins that function as a layer of protection on the endothelial surface (8–12). In vivo, the EGL forms an endothelial surface layer that influences inflammation and endothelial permeability, processes that are highly relevant to sepsis pathophysiology. It is well established that the EGL contains a variety of proteoglycans, glycoproteins, embedded soluble components, and glycosaminoglycans (GAGs) (11, 13, 14). The principal GAGs are heparan sulfate (HS), chondroitin sulfate, and hyaluronic acid (HA), with HS and HA being the most abundant of the GAGs (15). Literature indicates that loss of the EGL is an early event in the inflammatory response in vivo (16–19). Normally, the EGL is degraded during inflammation, and this allows leukocytes to bind to the appropriate adhesion molecules. This is accompanied by increased leukocyte extravasation and followed by loss of endothelial barrier function. Evidence for EGL degradation comes from measurements of patient serum following admittance to the intensive care unit where the blood was analyzed for EGL components following diagnosis of sepsis (17, 20–22). Plasma levels of HS and HA were increased fourfold for patients with septic shock relative to healthy patients. TNF-α has been shown to activate endothelial cells through an NF-κB pathway that leads to the production of enzymes that remove glycoproteins from the luminal surface (23). This activation upregulates enzymes such as hyaluronidase and heparinase that will cleave associated molecules in the EGL. The metalloprotease ADAM-17 has also been shown to be activated and cleave CD44, the binding site for HA during inflammation (24). High levels of EGL plasma components correlate with severe inflammation, but to what degree and how EGL loss contributes to the propagation of the inflammatory response is unknown. In the following, we provide evidence that loss of EGL components alone increases the susceptibility of uninflamed endothelium to neutrophil attachment. This finding is particularly relevant to systemic propagation of an inflammatory response and points to a potential contributing mechanism by which local inflammation can propagate to other tissues.

MATERIALS AND METHODS

Cell Culture and Seeding Flow Chambers

Human umbilical vein endothelial cells (HUVECS) were purchased from VEC Technologies (Rensselaer, NY) and grown to confluence in MCDB-131 medium. After reaching confluence, the cells were harvested using mild trypsin treatment and seeded into flow chamber devices. Cells were allowed to settle and adhere to the glass substrate for 2 h before flow commenced. After the cells were grown to confluency under flow of 10 dyn/cm² for 48 h at 37°C, they were removed from flow and tested. See Supplemental Materials, Section 1, for additional details (all Supplemental material is available at https://doi.org/10.6084/m9.figshare.12398036.v3).

Enzyme treatment.

Cells were also treated with hyaluronidase and heparinase III to enzymatically remove specific components of the EGL, following published methods (28). The flow channel above the cells was removed (see schematic in Supplemental Materials, Section 1), and 1 U/mL of heparinase III (Millipore Sigma, St. Louis, MI) was incubated with the cells for 1 h at 37°C in imaging buffer. For hyaluronidase treatments, cells were incubated with 10 U/mL of hyaluronidase (Millipore Sigma, St. Louis, MI) for 1 h at 37°C.

TNF-α treatment.

To observe changes in the EGL upon endothelial cell activation, cells were stimulated with 50 ng/mL of TNF-α under flow. The cells were grown under a flow of 10 dyn/cm² for 48 h, and TNF-α was injected into the flow reservoirs to produce a total TNF-α concentration of 50 ng/mL in the circulating medium. The cells were subjected to this stimulus for 4 h at 37°C and then measured in the same manner as control cells.

Immunofluorescence Microscopy

The MCDB-131 medium was unsuitable for imaging experiments because it contains phenol red and requires CO₂ to maintain its buffered pH, so for all labeling, adhesion, and mechanical measurements, we used an imaging buffer [modified version of McCoy’s medium without phenol red, with 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid added to buffer the solution, and 10% fetal calf serum]. After the cells had grown under flow for 48 h, they were removed from flow and the residual culture medium was removed from the dish. The cells were washed gently three times with imaging buffer before they were labeled with 10 μg/mL mouse anti-human heparan sulfate antibody (US Biological, 10E4 Epitope, Cat. No. H1890, RRID:AB_10013601) for 15 min at room temperature. Excess antibody label was washed off of the cells three times with imaging buffer followed by labeling with 10 μg/mL goat anti-mouse IgM Alexa Fluor 546 secondary antibody (Thermo Fisher Scientific, Cat. No. A-21045, RRID:AB_2535714) for 15 min at room temperature. For hyaluronic acid labeling, HUVECs were fixed with paraformaldehyde and incubated with 50 μg/mL of biotinylated hyaluronic acid binding protein (HABP; Sigma-Aldrich) overnight, after which they were washed and incubated with secondary Alexa Fluor 546 anti-biotin antibody (Santa Cruz Biotechnology, Cat. No. sc-101339, RRID:AB_1119609). For ICAM-1 labeling, cells were incubated with PE anti-human ICAM-1 antibody (BioLegend, Cat. No. 353105, RRID:AB_10899575) at 10 µg/mL for 15 min at room temperature. In addition, the nuclei of cells were stained with Hoechst (BD Biosciences, Cat. No. 561908, RRID:AB_2869394) at 1 μg/mL for 15 min. After all incubations with secondary antibody, cells were washed three times with imaging buffer and transferred to the microscope for imaging. All control and treatment groups were collected at the same time under the same conditions.

Atomic Force Microscopy Indentation

Elastic layer model equations are as follows:

$$F_{cell}=\frac{4}{3}{E}^{*}{R}_b^{\frac{1}{2}}{\delta }_c^{\frac{3}{2}}$$ (1)

$$E^{\text{*}}=E_{GL}+\left(E_{Cell}-E_{GL}\right)\left(\frac{P{\xi }^n}{1+P{\xi }^n}\right)$$ (2)

$$\xi =\frac{\sqrt{R\delta }}{t}{\left(\frac{E_{GL}}{E_{Cell}}\right)}^m\left(\frac{1-B_C{\nu }_C^2}{1-B_G{\nu }_G^2}\right)$$ (3)

AFM experiments were performed using experimental protocols from a previous manuscript (25). Briefly, tipless AFM cantilevers (NanoWorld, Switzerland) with spring constants in the range of 30 pN/µm were affixed with a 6-µm-diameter glass bead. The cells were indented 10 times at the same location in the perinuclear region at a 4 µm/s indentation rate to a maximum force of 5 nN. We have adopted an analytical approximation to finite element modeling of a thin soft layer on top of an elastic half space developed by Clifford and Seah (26, 27). The authors used experimental data and finite element modeling to calculate the stiffness of both the thin film layer and the underlying material. We applied their model, treating the EGL as a uniform thin soft film on the surface of the cell body (25). With this approach, a spherical indenter with radius R_bapplying a force on the cell F_cell can be described by Eq. 1. Applying the approach used by Clifford and Seah, we obtain the expression for the reduced modulus E^* (Eq. 2), where E_GL is the modulus of the EGL, E_cell is the modulus of the cell body, P and n are constants that have been empirically determined by Clifford and Seah from the fits to the finite element results, and ξ is given by Eq. 3, where t is the thickness of the EGL, v_G is the Poisson ratio of the EGL, v_cis the Poisson ratio of the cell, and m, B_G, and B_c are constants determined from the finite element fits (27). In a previous report (25), we have shown that the values of the Poisson ratios have little effect on the fits, and these were both fixed at a value of 0.3. This leaves undetermined coefficients that are obtained by least squares regression to the data: the elastic moduli E_GL and E_cell and the thickness t.

Neutrophil Isolation and Adhesion Assay

Whole blood samples were obtained from healthy donors, who provided written informed consent, under a protocol approved by the Institutional Review Board at the University of Rochester. For each neutrophil capture experiment, cells from a single donor were used. Donors typically donated cells for more than one experiment. There were four different donors in total, two male and two female, ages 25–55 yr. Neutrophils were isolated from the blood using 1-step polymorphs (Accurate Chemical and Scientific Corp., Westbury, NY). Neutrophils were then counted and resuspended to a final desired concentration of 10⁶ cells/mL. A suspension of neutrophils was perfused over endothelial monolayers at a physiological shear stress of 0.5 dyn/cm² for 5 min. After 5 min, the upstream fluid was switched to cell-free buffer for 1 min at the same flow rate to wash out nonadherent cells. Then, flow was stopped, and the number of adhered neutrophils per unit area was quantified from digitized images of the monolayer.

Permeability and Resistance Experiments

Human umbilical vein endothelial cells were seeded onto commercially available Transwell inserts (Corning, Oneonta, NY) and grown for 48 h. Using an EndOhm chamber (WPI Inc., Sarasota, FL), transendothelial electrical resistance (TEER) was measured by immersing the Transwell inserts into the chamber and recording the resistance values. Following TEER measurements, 1 mg/mL of 10-kDa-molecular weight fluorescein isothiocyanate-dextran (FITC-dextran; TCI, Tokyo, Japan) was added to the insert and allowed to diffuse through the endothelial monolayer for 1 h at 37°C. Transwells were then removed, and media from the bottom well were collected in triplicate for each experimental condition, and fluorescence intensity was measured using a TECAN microplate reader (TECAN Trading AG, Switzerland). The amount of FITC-dextran passing through the monolayer was calculated by reference to a set of standard dilutions.

RESULTS

Heparan Sulfate and Hyaluronic Acid Signal on HUVECs Is Altered after Enzymatic Removal of EGL Components and TNF-α Exposure

To test the effects of the different inflammatory mediators on the EGL, we used fluorescent labeling to detect the presence of heparan sulfate and hyaluronic acid on the surface of cultured HUVECs under different treatment conditions. The results of these different imaging treatments are shown in Fig. 1. Both heparan sulfate and hyaluronic acid were detectable on cells grown at a physiological shear stress of 10 dyn/cm². As one would expect, when the cells were treated with heparinase III and hyaluronidase, the apparent signal decreased for the respective component. Treatment with TNF-α decreased the presence of both heparan sulfate and hyaluronic acid. Although not evident from the images, treatment with heparinase III resulted in a slight but statistically significant decrease (−15%) in the hyaluronic acid signal, whereas treatment with hyaluronidase caused a variable but statistically significant increase (+25% on average) in the heparan sulfate signal. Quantification of the intensity data is described in Supplemental Materials, Section 2.

Figure 1.

Treatment effect on glycocalyx signal for HUVECs. A and B: heparan sulfate fluorescent antibody labeling and hyaluronic acid binding protein labeling of untreated cells. C and D: treatment with heparinase III. E and F: treatment with hyaluronidase. G and H: treatment with TNF-α. When the glycocalyx is degraded with an enzymatic treatment, the fluorescence corresponding to the respective EGL component is significantly decreased. Both EGL components are significantly reduced when the cell is stimulated with TNF-α (3 replicate experiments). EGL, endothelial glycocalyx layer ; HUVECs, human umbilical vein endothelial cells.

Localization of HS and HA on HUVEC Surfaces

It is important to know not only whether a certain EGL component is present but also where it is located. Using confocal microscopy, we find that in our cultures, HS tends to have significant presence on the abluminal side of the cell, as well as the luminal side, whereas HA appears to be present only on the luminal surface (Fig. 2, Supplemental Figs. S3 and S4; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.12398036.v3). Yao et al. (29) reported that following 24 h of fluid flow stimulation, HS was more prominent in the junctional region of the cell than in the central region, consistent with our own findings that the abluminal HS appeared predominantly in regions away from the cell nucleus (Fig. 2A). Similarly, Wang (30) demonstrated that HUVECs form a robust HA signal over the apical side of cells following application of shear stress. (Additional orthogonal views are shown in Supplemental Fig. S4.) Our fluorescence data are consistent with those findings.

Figure 2.

Representative confocal images of HUVECs labeled for heparan sulfate or hyaluronic acid. A: 2-D image of HUVECs stained for heparan sulfate and nucleus. B: 2-D image of HUVECs stained for hyaluronic acid and nucleus. C: 3-D side view of a HUVEC monolayer stained for heparan sulfate and nucleus. The majority of the heparin sulfate localizes near the junctions of cells and closer to the bottom near the microscope slide, whereas only a smaller proportion of the heparan sulfate is localized on the top. D: 3-D side view of HUVEC monolayer stained for hyaluronic acid and nucleus. Hyaluronic acid localizes near the apical side of HUVECs. E: orthogonal slice of HUVEC monolayer stained for heparan sulfate. Heparan sulfate is located near the boundary of the cell and the glass surface. F: orthogonal slice of HUVEC monolayer stained for hyaluronic acid. The hyaluronic acid signal is localized on the apical side of the cell away from the glass surface (3 replicate experiments). HUVECs, human umbilical vein endothelial cells; 2-D, two-dimensional; 3-D, three-dimensional.

Effects of Selected Enzymatic Degradation and Inflammatory Mediators on the Thickness and Stiffness of the EGL

Changes in EGL thickness caused by enzymatic or inflammatory mediator treatment were quantified using AFM (Fig. 3). Under control conditions, the EGL had a mean thickness of 113 ± 7 nm (n = 80 cells). When the cells were stimulated with TNF-α, the mean EGL thickness was reduced to 75 nm ± 8 nm (n = 30 cells). When the EGL of the cells was enzymatically digested with a combination of heparinase III and hyaluronidase, the mean thickness was reduced to 60 ± 10 nm (n = 30 cells). (The ranges given for the average thickness and modulus are the standard errors of the mean.) As previously reported, treatment with either hyaluronidase or heparinase alone caused intermediate reductions in thickness: heparinase, 87 nm ± 9 nm (n = 59 cells); hyaluronidase, 78 nm ± 8 nm (n = 50 cells) (25). The EGL modulus trended toward higher values after treatment, but the differences in the values were not statistically significant (ANOVA).

Figure 3.

A and B: AFM data for glycocalyx properties. Error bars represent standard error. ANOVA was used to test for significance, P < 0.05. A: AFM data show a reduced EGL thickness after TNF-α treatment, and after treatment with both hyaluronidase and heparinase III. B: there was no significant effect of the combination of enzymes or TNF-α treatment on the EGL modulus. ANOVA followed with Tukey’s HSD was used to test for significance: P < 0.05. AFM, atomic force microscopy; EGL, endothelial glycocalyx layer; HSD, honestly significant difference.

Changes in EGL Barrier Function

Endothelial barrier function is known to be affected by inflammatory stimulus, and there are reports that loss of the EGL can contribute to these changes (31–34). We addressed this question by measuring the permeability of our cultured HUVECs using TEER and FITC-dextran. (Measurements to ensure the continuity of our monolayers are presented in Supplemental Materials, Section 6.) TEER measurements were made immediately after each treatment, and when TEER measurements were completed, dye permeability measurements were performed. The effects of thrombin were tested as a positive control. The results for the TEER measurements and permeability experiments are given in Fig. 4. Removal of both the HS component of the EGL by heparinase III and HA by hyaluronidase has a significant effect on the passage of small solutes as assessed by TEER, and the magnitude of those changes was comparable with that observed after TNF-α treatment but slightly less than the change produced by thrombin. In contrast, whereas both TNF-α and thrombin treatments resulted in large increases in dye permeability, enzymatic treatment to remove EGL components did not have a statistically significant effect.

Figure 4.

Effect of treatments on TEER and permeability of cultured HUVECs. A: after treatment with heparinase III and hyaluronidase, the TEER was significantly decreased relative to control. Both thrombin and TNF-α reduced TEER significantly relative to control. A combination of the two enzymes did not cause larger changes in resistance than the individual components. B: treatment with heparinase III and hyaluronidase did not significantly increase FITC-dextran signal relative to control. Treatment with TNF-α and thrombin increased FITC-dextran signal relative to control as expected. Significance was determined using ANOVA followed by Tukey’s HSD, where *P ≤ 0.05 and **P ≤ 0.01, 3 replicate experiments. HSD, honestly significant difference; HUVECs, human umbilical vein endothelial cells; TEER, transendothelial electrical resistance.

An Enzyme Cocktail but Not Heparinase or Hyaluronidase Individually Increases Neutrophil Adhesion on Cultured HUVECs

Finally, we investigated leukocyte adhesion after removal of different EGL components. During an inflammatory response, the endothelium sheds the EGL, presumably to enable closer interaction between leukocytes and the endothelium, enabling capture, slow rolling, adhesion, spreading, crawling, and (in some cases) transcellular migration (35). The ability of leukocytes to roll and adhere to the endothelial surface is a critical step in the inflammatory response. We tested the effects of EGL degradation on neutrophil capture on endothelial monolayers. Neutrophils were perfused over endothelial cells at physiological shear stress (0.5 dyn/cm²) for 5 min, after which, the number of adhering leukocytes was counted (Fig. 5). (We did not observe significant rolling behavior before arrest, and the distribution of cell velocities before neutrophil capture was not affected by enzymatic treatment. See Supplemental Materials, Section 5.) The total number of adhering cells was then averaged for each treatment and compared with the average of the control for each experimental day and reported as a ratio. Treatment with TNF-α significantly increased neutrophil adhesion relative to untreated cells. This was expected based on the observed removal of the EGL (Fig. 1) and increases in ICAM-1 expression (Fig. 6). Heparinase III or hyaluronidase treatment alone did not affect neutrophil capture. When a combination of the two enzymes was used, we observed that the number of captured neutrophils increased, and the size of the increase became larger with the time elapsed since neutrophil isolation (see Supplemental Materials, Section 4). The increase in neutrophil avidity over time was not evident for untreated monolayers and was slight for monolayers treated with hyaluronidase, but it was clearly evident for monolayers treated with both heparinase III and hyaluronidase.

Figure 5.

Flow adhesion assay results. Neither heparinase III treatment nor hyaluronidase treatment affects neutrophil capture. A combination of the two enzymes resulted in a significant increase in neutrophil adhesion. Treatment with TNF-α maximizes cell adhesion as expected with an average increase of 14.6-fold. (Note scale difference between the two panels.) Error bars represent standard error. Significance was determined using ANOVA with Tukey’s HSD where *P < 0.05 and **P < 0.01. HSD, honestly significant difference.

Figure 6.

Treatment effect on ICAM expression on HUVECs. ICAM-1 fluorescent antibody labeling with and without heparinase III treatment or TNF-α treatment. We observed significant differences in ICAM-1 only after treatment with TNF-α (3 replicate experiments). HUVECs, human umbilical vein endothelial cells.

DISCUSSION

Localization and Contribution to the Thickness of EGL Components

Characteristics of the EGL on the surface of endothelial cells have been well described in several recent reviews (15, 36, 37). The EGL network is made up of proteoglycan side chains, core proteins, glycoproteins, and glycolipids, with the components HA and HS making up the majority of its mass (38). Although it is widely known that these are major components of the EGL, it is less well understood how they are distributed on the cell surface and whether they might serve different functions. One recent study has examined the lateral distribution of HS and HA on the luminal surface of cultured bEnd3 cells using high-resolution stochastic optical reconstruction microscopy (STORM), characterizing the localizations as “high,” “middle,” or “low” for HS and HA in the nuclear, perinuclear, or edge regions (39). It is important to distinguish those descriptions from our findings which reveal that a large proportion (but not all) of the HS appears to be on the abluminal surface of the cells in our HUVEC cultures. Consistent with the findings of Fan et al. (39), we find that HA appears to be predominantly located on the luminal surface. Long chains of hyaluronan, attached to endothelial membrane-bound receptors, namely, CD44, are presumed to contribute to the layered structure above the endothelial surface at much greater distances than other glycosaminoglycans (39, 40). Bai and Wang (41) used wheat germ agglutinin (WGA) labeled with FITC (WGA-FITC) to label EGL components and observed that the intensity and distribution of labeling increased from the edge of the cell to apical regions over periods of up to 3 wk, becoming fully distributed after ∼2 wk. Although our observations of HS on the abluminal surface near the cell edge resemble the distribution of WGA-FITC Bai and Wang observed after ∼5 days of culture, there are important differences in both the labeling protocols and the growth conditions between the two studies. WGA lacks the specificity of the HS and HA labels used here, as evidenced by the significant residual WGA labeling in our hands after the enzymatic removal of both HS and HA (Supplemental Materials). Moreover, cells were grown under static conditions in those prior studies, conditions known to produce a less robust EGL than cells grown under flow. Although it is possible that the EGL in our studies might have become more fully developed over time, technical difficulties prevented us from observing the development of the EGL in cells grown under flow over longer periods. We did take care to ensure that our monolayers were fully confluent, as described in our Supplemental Materials, Section 6. Our observation that a high proportion of HS is found on the abluminal surface of the cells creates a significant challenge in trying to assess HS contributions to barrier properties using epifluorescence intensity because high spatial resolution is needed to distinguish between luminal and abluminal EGL localizations.

In recently published work, we have demonstrated that individual treatments with heparinase and hyaluronidase reduce the EGL thickness by 20% and 30%, respectively (25). It is unlikely that the reduction in EGL thickness resulting from heparinase III treatment is due to collateral effects on the presence of HA on the surface for two reasons. First, because we see only a 15% reduction in the amount of HA on the surface after heparinase III treatment. Given that complete removal of HA results in a 30% thickness reduction, loss of 15% of the HA would not be sufficient to cause a 20% reduction in thickness. Second, when a combination of the enzymes is used, the thickness is reduced by 50%, indicating that the effects of heparinase III and hyaluronidase are additive. Importantly, the reduction in thickness observed when both components are removed is comparable with what is observed after inflammatory stimulus. Indeed, after TNF-α treatment, the EGL thickness determined using AFM was reduced by 40%, an amount not statistically different from the change observed when heparinase III and hyaluronidase were used in combination. This implies that HS and HA together constitute the majority of the luminal barrier and that their removal should enable us to understand the role that this barrier plays per se as part of the changes in endothelial function accompanying inflammation.

Permeability of the EGL

Understanding of fluid transport from the vasculature to the tissues has evolved in recent decades as shortcomings in the original theories of Starling were recognized (42), and the EGL has been recognized as an important component of the regulation of tissue-vascular fluid balance. Theoretical descriptions of fluid transport across the endothelium vary in complexity but generally include two parallel pathways for transport: one so-called large-pore pathway that allows the passage of plasma proteins, and one designated the small-pore pathway that only allows passage of water and very small solutes (42, 43). The EGL’s primary role in this context appears to be limiting the permeability of plasma proteins, thus reducing transport through “large pores,” but it likely affects the resistance to flow of smaller solutes as well. Evidence for this was first obtained by Adamson (44), who showed that digestion of the EGL with pronase led to increased rates of fluid transport out of mesenteric vessels in the frog with no evidence of changes in the endothelial ultrastructure. Contradictory evidence exists with regard to increased leakage of proteins after EGL removal. Treatment of isolated porcine arterioles with pronase or heparinase I (100 U/mL) increased leakage of both 14- and 65-kDa protein tracers (45), and cultured glomerular endothelial cells also became leakier to albumin after treatment with either neuraminidase or heparinase III. In contrast, an in vivo study in the hamster found no effect of hyaluronidase treatment on leakage of plasma proteins (46). A number of studies have examined penetration of the EGL by both dextrans and proteins (13, 33), but increased EGL penetration does not necessarily correlate with increased leakage, as the endothelium itself likely serves as a barrier to large solute transport. Our present results support this view. Decreases in TEER likely reflect increases in hydraulic conductivity, but the small but statistically insignificant increase in permeability of dextran (10 kDa) would be consistent with the presence of a residual barrier to protein-sized molecules in the endothelium itself. To this point, when we apply the EGL label WGA-FITC on monolayers treated with heparinase III and hyaluronidase in combination, residual labeling is observed, particularly in the cell junctions (Supplemental Fig. S8). Furthermore, treatment with the enzyme combination appears to have little effect on VE cadherin label, in marked contrast to the effects of TNF-α, which showed significant disruption of VE cadherin at the cell junctions (Supplemental Fig. S7). These findings suggest that it is the integrity of the cell junctions that is most critical for maintaining impermeability to dextran transport.

EGL as a Barrier to Neutrophil Adhesion

Leukocyte adhesion to the endothelium is an important early step in the initiation and progression of sepsis. Determining the molecular mechanisms underlying this process is key to better understanding the development of the disease itself. Our observations of the effects of EGL removal on neutrophil adhesion provide an important insight in this regard. Interestingly, the reduction in HS and HA individually did not affect adhesion significantly, and it was only when there was complete removal of both components that we saw robust increases in adhesion. Therefore, both HS and HA contribute to the prevention of neutrophil adhesion, as prior studies have surmised (24, 31, 47, 48), and their contributions appear to be additive. The observation that removal of both HA and HS, but neither one individually, causes significant increases in neutrophil capture should not be surprising, as the protection provided by the EGL is expected to be highly nonlinear with thickness. When the thickness of the EGL is sufficient to prevent adhesion molecule engagement, little or no adhesion is expected, but as soon as the thickness is reduced below some critical value, molecules on the opposing surfaces can bind to each other and adhesion ensues. The thickness measurements of the present study indicate that this critical minimum thickness is on the order of 60–70 nm.

Another important observation is that the level of adhesion increased over time (see Supplemental Materials, Section 4), not because of changes in the endothelium, but because the neutrophils interacting with the endothelium developed a partially activated phenotype with time after they were obtained from donors. This is relevant to the systemic propagation of sepsis. The increased serum levels of EGL components suggest that there is degradation of the EGL by circulating enzymes. In addition, neutrophils exposed to inflammatory stimulus in one region of the body are transported to others, where EGL degradation combined with mild neutrophil activation would result in increased neutrophil adhesion and propagation of the inflammatory response. It is also relevant that the largest increases in adhesion in our studies occurred after treatment with TNF-α, which not only reduced the EGL barrier but also increased the expression of adhesion molecules on the endothelial surface. This suggests a mechanism for the increased susceptibility of patients with preexisting inflammation-associated diseases, such as diabetes or obesity. When the EGL is removed, increased presence of endothelial adhesion molecules is expected to increase neutrophil adhesion and the spread of the inflammatory response.

Conclusions

These studies document a role for the EGL in preventing adhesion of mildly activated neutrophils to uninflamed endothelium and support the pursuit of potential therapeutic strategies using HA and HS to reduce propagation of an inflammatory response in patients with localized infections. Prior studies exploring the protection of EGL in treating inflammation have focused on protecting syndecans or repairing HS (48–50). The present findings support protection of CD44 and reparation of HA as complementary strategies, as both EGL components contribute to maintenance of barrier function and prevention of unwanted neutrophil adhesion.

GRANTS

This work was supported by the US Public Health Service under NIH Award No. 5 R01 HL125265 from National Heart, Lung, and Blood Institute.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

L.F.D. and R.E.W. conceived and designed research; L.F.D., E.B.L., and J.K. performed experiments; L.F.D., E.B.L., J.K., and R.E.W. analyzed data; L.F.D., E.B.L., and R.E.W. interpreted results of experiments; L.F.D. and R.E.W. prepared figures; L.F.D. drafted manuscript; L.F.D., E.B.L., and R.E.W. edited and revised manuscript; L.F.D., E.B.L., J.K., and R.E.W. approved final version of manuscript.