Observation and characterisation of the glycocalyx of viable human endothelial cells using confocal laser scanning microscopy

This paper describes the use of confocal laser scanning microscopy (CLSM) to observe and characterise the fully hydrated glycocalyx of human umbilical vein endothelial cells (HUVECs).

Viable HUVECs in primary culture were studied at room temperature in HEPES-buffered, phenol red- and serum-free CS-C cell culture medium.

A fluorescein isothiocyanate-linked wheat germ agglutinin (WGA-FITC) (2 μg ml−1, 30 min) was used to detect N-acetylneuraminic (sialic) acid, which is a significant component of the endothelial glycocalyx.

Single confocal sections, less than 1.3 μm thick, were collected at intervals of 0.5 μm, scanning through the entire z-axis of a series of cells.

Cell-surface associated staining was observed, which enabled the glycocalyx thickness to be deduced as 2.5 ± 0.5 μm.

This dimension is significantly greater than that measured by electron microscopy, for glutaraldehyde-fixed cells (0.10 ± 0.04 μm).

The specificity of WGA-FITC staining was demonstrated by treatments with several enzymes, known to degrade glycocalyx (heparatinase, chondroitinase, hyaluronidase and neuraminidase), of which neuraminidase (1 U ml−1, 30–60 min) was the most effective, removing up to 78 ± 2% of WGA-FITC binding to HUVECs.

Cell viability was assessed simultaneously with ethidium homodimer-1 staining and confirmed by standard colorimetric 3-[4,5]dimethylthiazol-2,5diphenyltetrazolium bromide (MTT) test.

CLSM thus provides a useful approach for in situ visualisation and characterisation of the endothelial glycocalyx in viable preparations, revealing a thickness that is an order of magnitude greater than found in ex situ measurements on fixed cells.


A single layer of endothelial cells forms a critical barrier between the blood and the interstitial fluid of most organs.

The luminal surface of the endothelium is exposed directly to the blood and is covered by an elaborate carbohydrate-containing, polyanionic surface coat called the glycocalyx.

It includes plasmalemmal components such as the glycosylated ectodomains of integral membrane proteins, proteoglycans and glycolipids, as well as adsorbed plasma proteins such as albumin.1

Many important normal and pathological vascular functions, such as capillary permeability,2 receptor-mediated transcytosis,3 transendothelial cell migration,4 thrombosis,5 inflammation,6 blood coagulation,6 oxygen transport7 and sensing of wall shear stress,8 are dependent on interactions occurring at the level of the glycocalyx.

Albumin, located within the luminal glycocalyx, has been shown to increase the anionic charge and volume exclusion of the glycocalyx, resulting in increased steric and electrostatic exclusion forces that reduce vascular permeability.9

Several studies have used electron microscopy to demonstrate the presence of the glycocalyx in the microcirculation.10–12

Conventional electron microscopy requires that cell specimens be dehydrated during the fixation procedure so that the total volume of extracellular material is significantly reduced or collapsed and compressed, leading to the distortion of the glycocalyx architecture.

To eliminate this problem, we have chosen to develop methodology centred on confocal laser scanning microscopy (CLSM) to study the hydrated glycocalyx of viable, cultured endothelial cells derived from human umbilical vein.

CLSM is a well established physical method for collecting three-dimensional images of cellular structures of living cells without destroying important functional features of studied specimens.13,14

However, there are no reports on in situ observation of living endothelial glycocalyx with CLSM.

The glycocalyx layer has been visualised by lectin binding to living cells.

We used a fluorescein isothiocyanate-linked wheat germ agglutinin (WGA-FITC), which binds selectively to N-acetyl-d-glucosaminyl residues, a constituent of heparan sulfate, the major saccharide of cell surface proteoglycans15 and to N-acetylneuraminic (sialyl) residues, making a significant contribution to the surface charge and representing a major component of the endothelial glycocalyx.16

Stringent concurrent screening for viability of the cells was undertaken for all experimental conditions used.

The CLSM analysis provided evidence of the extensive expression of the glycocalyx in cultured macrovascular endothelial cells in physiologically relevant conditions.

Materials and methods


Unless stated otherwise, materials were obtained from Sigma Chemicals (Poole, UK).


Human umbilical cords were obtained with full ethical approval within 12 h of birth from normal pregnancies.

Human umbilical vein endothelial cells (HUVECs) were isolated by collagenase (0.3 mg ml−1) digestion17 and cultured in 35 mm diameter Petri dishes in medium 199 (M199) containing 20% fetal bovine serum, 0.05 mg ml−1 gentamycin, 20 μg ml−1 endothelial growth factor supplement and 90 μg ml−1 heparin at 37 °C in 5% CO2/air.

The identity of endothelial cells in culture was confirmed by the presence of typical “cobblestone” monolayer morphology (shown by the phase contrast image, Fig. 1) as well as by the expression of von Willebrand factor, VCAM, and angiotensin converting enzyme (ACE), as determined by immunocytochemical staining and Western blotting.

After 3 days in primary culture, at 70–80% confluency, the cells were washed twice with phosphate buffered saline (PBS), and WGA-FITC (2 μg ml−1), prepared in HEPES-buffered, phenol red- and serum-free CS-C (Sigma, C 1556) cell culture medium, was applied for 30–40 min.

Confocal laser scanning microscopy

Lectin binding (as WGA-FITC binding) was observed with a confocal laser scanning microscope (LSM 510, Axioplan 2, Carl Zeiss, Jena, Germany).

All CLSM images (1024 × 1024 pixels, 12 bit pixel depth) were acquired using a water immersion objective lens (Zeiss, Achroplan 63x/0.95NA W) with a 10× tube lens.

To observe WGA-FITC binding, an argon laser (λ = 488 nm) was used in conjunction with a long-pass filter (λ = 505 nm).

For transmission images a HeNe laser was used (λ = 543 nm).

For simultaneous assessment of cell viability, WGA-FITC and ethidium homodimer-1 were excited using Ar laser lines at λ = 488 and 514 nm, respectively, and fluorescence emission was measured with filter sets, band-pass λ = 505–550 nm and long-pass λ = 560 nm, respectively.

Black level (background offset) was adjusted to eliminate autofluorescence from unstained cells.

To achieve the optimum compromise between resolution and image intensity, the confocal aperture was set to give an optical slice of <1.3μm (data from Zeiss).

Under these conditions lateral resolution was sub-micron and axial resolution was ±0.65 μm.

Images were processed using the LSM Image Browser software (Zeiss).

Average pixel intensities were calculated over the area corresponding to single cells (typically equivalent to 10–40 000 square pixels) using Scion Image software (Scion Corporation, USA).

To discriminate between binding of WGA to sialic acid residues and binding to N-acetyl-d-glucosaminyl residues, a range of known glycocalyx-digesting enzymes12,18,19 namely neuraminidase (1 U ml−1 in CS-C, 30–60 min), hyaluronidase (from Streptomyces hyalurolyticus, 50 U ml−1 in CS-C, 30–60 min), chondroitinase (1 U ml−1 in CS-C, 60 min) and heparitinase (5 U ml−1 in CS-C, 60 min), were added after a standard staining procedure with WGA-FITC (2 μg ml−1 in CS-C, 30 min).

All the confocal microscopy analyses were carried out in an air-conditioned room, at 23 ± 0.5 °C.

Cell viability was simultaneously assessed with ethidium homodimer-1 (2.5 μM), which undergoes enhancement of fluorescence upon binding to DNA and RNA, which is only possible for cells with damaged plasma membranes.20

Additionally, a colorimetric 3-[4,5]dimethylthiazol-2,5diphenyltetrazolium bromide (MTT) viability test was performed on cells growing as primary culture in 96-well plates, under conditions identical to those of the confocal analyses.21

The tetrazolium ring of MTT is cleaved in active mitochondria, and so the reaction occurs only in living cells.

Electron microscopy

HUVECs grown on cover slips were incubated in the presence or absence of neuraminidase (1 U ml−1) for one hour.

Cells on cover slips were then fixed by immersion in freshly prepared solution comprising 2.5% glutaraldehyde and 0.3% ruthenium red in 0.1 M cacodylate buffer (pH 7.3), using a technique adapted from Hayat.22

After primary fixation, on ice for six hours, cells were washed in three changes of 0.1 M cacodylate buffer at 4 °C and post-fixed in 1% osmium tetroxide, 0.3% ruthenium red in 0.1 M cacodylate buffer for 30 min at 4 °C.

Cells were washed again in three changes of cacodylate buffer and then distilled water, prior to dehydration with ethanol.

This was followed by infiltration and embedding in Araldite resin.

Resin surrounding the coverslips was trimmed and the block placed in liquid nitrogen to enhance cleavage between the glass coverslip and the overlying resin (which have different thermal contraction rates).

Glass fragments were gently removed from the overlying resin containing the cells.

The cell layer was trimmed and 0.5 μm survey sections were cut and stained with 1% toluidine blue in 1% aqueous borax for light microscopy.

Cell rich areas of the block were trimmed and 50 nm thick sections were cut (perpendicular to the culture surface) and stained with 3% aqueous uranyl acetate and Reynolds' lead citrate solution.23

Digital micrographs were taken on a Philips 100CS microscope to show the disposition of any ruthenium red deposits on the surface of the HUVECs.


WGA-FITC binding

Cells incubated with 2 μg ml−1 WGA-FITC showed clearly detectable fluorescence on the cell surface which appeared after 5 min and which increased to a maximum intensity after approximately 30 min of WGA-FITC incubation.

Representative WGA-FITC staining and a corresponding transmission image of HUVECs are illustrated in Fig. 2.

It can be seen that staining occurs heterogeneously over an area corresponding to that occupied by cells.

Confocal microscopy allows a quasi-optical sectioning of fluorescing cells by eliminating fluorescence from sources below and above the focal plane.

A number of single confocal sections, less than 1.3 μm thick, were collected at intervals of 0.5 μm through the entire z-axis of cells, labelled with WGA-FITC for 30 min.

A typical set of results is shown in Fig. 3.

The shape and intensity of the fluorescent pattern differed as a function of the level of cross-section scanning, from a small clearly defined area near to the top of the cell, to the diffused pericellular matrix of weaker intensity on the bottom, indicating that the staining is likely to be related to the cell surface.

For the laser intensity and scan time used, no significant photobleaching was observed.

Fig. 4(a) shows a single optical slice near to the surface of the cell and a re-construction of the z-stack images as orthogonal projections in the xz and yz planes, through a cell, along planes passing through the green and red lines shown in Fig. 4.

From these projections it can be seen that staining occurs in an arc over the cells, indicating surface localisation.

Measurement of the glycocalyx thickness of the four cells visible in Fig. 4(a) from the orthogonal projections gave a value of 2.5 ± 0.5 μm (n = 4).

The range in the thickness of glycocalyx measured for more than 20 cells in four separate experiments was 2.0 ± 1.2 μm.

Enzyme treatment

Enzyme digestion was used to elucidate the specificity of WGA-FITC staining, and to obtain information about the types of WGA binding sites.

Neuraminidase cleaves the O-glycosidic linkages between the terminal neuraminic (sialic) acids and the subterminal sugars.24

Incubation of WGA-labelled HUVECs with a medium containing 1 U ml−1 neuraminidase resulted in progressive reduction of the fluorescence (Fig. 4(b)).

Reconstruction of z-axis intensity profile of cells after neuraminidase digestion reflected a clear diminishing of both the surface staining and its thickness (Fig. 4(b)).

The variation in intensity of staining of FITC-WGA without enzyme treatment under similar conditions (60 min, same laser settings and number of confocal slices recorded) was less than 5% (data not shown).

Therefore, we can be confident that the decrease in the intensity of staining with enzyme treatment is not significantly affected by photobleaching of the fluorophore or changes in the intensity of the laser during the time course of the experiment.

Fig. 5(a and b) show WGA-FITC staining before, (a), and after one hour incubation with neuraminidase, (b).

In Fig. 5(c) the decrease in staining is illustrated quantitatively along a line taken through a single cell, as shown in Fig. 5(a) and 5(b).

The significant effect of neuraminidase treatment on the fluorescence intensity can clearly be seen.

Fig. 5(d) shows the average intensity over an area of 210 000 square pixels, recorded from images taken during 60 min of neuraminidase treatment.

The reduction in fluorescence intensity demonstrates a smooth trend.

Fluorescence was found to diminish by 64 ± 3% (n = 5) after 30 min and by 78 ± 2% (n = 2) after 60 min.

Electron microscopy examination of fixed endothelial cells stained with ruthenium red indicated the presence of glycocalyx, visible as a layer of dark material on the apical surface (Fig. 6(a)), with a thickness of 0.10 ± 0.04 μm (n = 6).

In agreement with the in situ observations reported above, neuraminidase-treated cells which have been fixed lack any obvious glycocalyx (Fig. 6(b)).

The results of fluorescence intensity measurements, after incubation with a series of enzymes, are summarised in Table 1.

The distribution of WGA-FITC signal after heparitinase (typical data shown in Fig. 7), hyaluronidase and chondroitinase ABC revealed relatively low levels of glycocalyx removal, so that reduction in average intensity was 21 ± 8% (n = 9), 13 ± 5% (n = 7) and 29 ± 3% (n = 9) of the control level (100%) after 60 min of heparitinase, chondroitinase and hyaluronidase treatment, respectively.

Cell viability

The aim of this study was to examine cells in a physiologically relevant environment, where the cells are likely to have retained normal functional characteristics.

Therefore, the maintenance of cell viability was of paramount importance, particularly given the potential cytotoxic effect of laser illumination25 and FITC-labelled lectins.26

Cell viability was tested using two approaches.

Firstly, the preincubation and continued presence of ethidium homodimer-1 (2.5 μM, 20 min), throughout all labelling and digestion procedures, demonstrated that all cells monitored maintained intact plasma membranes.

Secondly, cell viability assessed by MTT revealed no statistically significant differences between control and WGA-FITC, or control and WGA-FITC followed by neuraminidase-treated cells (P = 1, Mann–Whitney test, three independent cell populations, data not shown).

These data confirmed that HUVEC monolayers retained viability throughout all the procedures of WGA-FITC binding and neuraminidase treatment.


It is well established that the luminal surface of the endothelium is lined with glycocalyx, a network of carbohydrate-rich membrane-bound molecules.1

Electron microscopy studies have assessed the thickness of the endothelial glycocalyx as less than 100 nm.12,27

Recently, developed methods, such as in vivo visualisation by intravital microscopy, provided information about structures extending much further from the cell surface, averaging 0.6 μm.28

Although it is still a matter of debate,29 it is possible that collapse of highly hydrated surface structures during chemical fixation of specimens for electron microscopy preparation may lead to a considerable underestimation of the real thickness of the glycocalyx.

The present study focused on using CLSM applied to isolated HUVECs, as a model to demonstrate non-invasively the in vitro expression of the glycocalyx.

By this approach we have shown that the apical surface of HUVECs is covered with an extensive layer of glycocalyx, with an estimated thickness on the micron scale.

It is worth noting that the CLSM procedure involved several washing steps, which may have removed plasma proteins associated with, and contributing to, the formation of the endothelial surface layer, which suggests that the thickness determined is unlikely to be overestimated.

The relatively small molecular size of WGA (molecular weight 36 kDa, volume ca. 64 nm330) used as a probe for the dissection of cell surface associated molecules, rules out the possibility of it having a significant role in the observed glycocalyx thickness.

Interestingly, the comparative detailed electron microscopy examination with the cationic dye, ruthenium red, demonstrated considerably lower thickness of glycocalyx structures in HUVECs (0.10 ± 0.04 μm) consistent with earlier electron microscopy studies of endothelial cells.12,27

This suggests that the preparation and measurement routine for electron microscopy results in significant peturbation of the glycocalyx layer thickness.

The much larger dimension of the glycocalyx in HUVECs observed with CLSM may therefore be considered as a characteristic of the unperturbed monolayer.

In addition to the clear consequences of sample preparation procedures, the discrepancies in the measurements of the dimension of the glycocalyx in the present study, and data published earlier in vivo,18 might also partly arise from the structural heterogeneity of the endothelial glycocalyx in various species and vessels.

All known glycocalyx estimates have previously been carried out on animal blood vessels rather than human, and have shown considerable variations in composition and thickness, which depended on vascular tree, age of animals, and even the specific region of the vessel.12,31

It would be reasonable to expect the glycocalyx morphology of cultured human endothelial cells to have its own unique characteristics.

It is also likely that glycocalyx thickness would be influenced by a variety of environmental factors such as density of cells, culture conditions and shear stress.

Henry et al. have estimated the glycocalyx thickness in vivo as 0.5 μm,18 which is clearly less than our studies.

Many factors could contribute to this difference eg, different species, in vivovs.in vitro conditions.

Additionally, the measurements of Henry et al. were made under physiological flow and it is not fully understand how this may affect the thickness of the glycocalyx, although it has been shown that flow-induced shear stress enhances glycosaminoglycan synthesis in vascular endothelial cells in vitro.32

WGA recognises sialyl residues, which are reported to be expressed at high concentration in umbilical endothelium.16

Sialic acids are 9C monosaccharides that link to the terminal galactose, N-acetylgalactosamine, or other sialic acid residues in carbohydrate chains that are attached to glycoproteins or glycolipids.

The hydrolysis of this linkage by neuraminidase is used as a common tool to identify the specific presence of sialic residues on a cell surface.12

This standard procedure of using glycocalyx-degrading enzymes to demonstrate the targeted sites for WGA binding was considered appropriate to validate the new experimental approach.

Due to their terminal location, sialic acids on the cell surface are among the first molecules encountered by other cells coming in contact with the cell.33

In addition, the carboxylate group at C-1 is deprotonated at physiological pH,34 making sialic acid the only sugar in glycoproteins which bears a net negative charge.

Sialic acids have also been shown to inhibit interactions between molecules and contribute to the anti-adhesive nature of luminal endothelium, including HUVECs.34

In the experiments reported herein, neuraminidase removed WGA-FITC labelling effectively, but not completely.

The residual labelling could be ascribed to interaction with N-acetylglucosamine, a constituent of glucosaminoglycans such as heparan sulfate or hyaluronate, which are both expressed in HUVECs.35

This is supported by the observation of glycocalyx removal by heparitinase and hyaluronidase, which act on glycosidic bonds of N-acetylglucosamine, in heparan sulfate and hyaluronate molecules, respectively.

Hyaluronidase also caused a marked removal of the fluorescent staining.

Interestingly, however, the kinetics of the enzyme's action was different from that described by Henry et al.in vivo, where hyaluronidase induced a highly pronounced effect in hamster vasculature, with the maximum achieved after one hour of systemic infusion.18

Variation in the properties of the glycocalyx in different species, as well as different incubation temperatures, could explain the discrepancy.

Low levels of chondroitin sulfate may explain the absence of chondroitinase-induced glycocalyx disruption on the cell surface in our study.

The lack of enzymatic effect may also be due to steric hindrance of the molecules in the heterogeneous glycocalyx structure.


In conclusion, we have developed a methodology for high resolution CLSM coupled with direct staining with fluorescently labelled lectin, WGA-FITC, to enable the visualisation of the sialic acid-enriched glycocalyx in viable HUVECs.

The specificity of the WGA-FITC staining was demonstrated by treatment with several enzymes known to degrade glycocalyx, from which neuraminidase was found to be most effective.

By recording single confocal sections through the z-axis of cells the thickness of the hydrated glycocalyx was determined to be an order of magnitude greater than that found in ex situ measurements on fixed cells.

Viability was confirmed in the present studies using double labelling with ethidium homodimer-1 as an indicator of plasma membrane integrity, the MTT assay and demonstration of morphologically undisturbed monolayers throughout the imaging process.

These tests allow us to be confident that the measurements of glycocalyx relate to viable cell monolayers.

To our knowledge, this is the first report using CLSM for quantitative in situ analysis of the fully-hydrated structurally-intact glycocalyx in human macrovascular endothelium.

There is scope to now use this technique to further investigate the effect of fluid-flow generated shear stress on the glycocalyx.