INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE INNERMOST COMPONENT of the blood-tissue barrier consists of a monolayer of endothelial cells. Their membrane facing the vascular lumen in direct contact with the blood phase is coated with a glycocalyx entailing cell-born proteoglycans and glycoproteins (16). In electron microscopic studies, this layer exhibits a thickness mostly ranging between 50 and 100 nm (4, 20). However, intravital microscopic studies suggest that a more complex endothelial surface layer (ESL) of 0.5-1 µm is present in vivo (30, 43). Histological studies attempting to conserve gel-like structures by employing nonaqueous fixation methods provide additional evidence for a thicker surface layer (36). The discrepancy between washed and dehydrated glycocalyx accessible to most electron microscopic investigations and functional ESL in vivo has been explained by contribution of blood-borne substances (29, 31). Therefore, the present study was aimed at investigating the interaction of plasma proteins with the endothelial glycocalyx.

The most abundant plasma protein, albumin, is a likely candidate for blood-borne components of the ESL. Binding studies of albumin on cultured endothelial cells showed slow saturation indicating either negative cooperativity or a heterogenous population of binding sites with different affinities (34). Adhering albumin influences ESL properties and has, for example, a marked effect on capillary permeability for water (15). This could suggest a firm attachment of albumin to membrane-bound ligands. On the other hand, albumin is an essential transport protein underlined by its ability to bind poorly water-soluble substances such as fatty acids, endogenous metabolites, and drugs (2, 3, 22, 27). This function could be severely compromised by a loss of mobility at the surface of endothelial cells.

Possible types of interaction between albumin and the endothelial surface may be distinguished by analyzing changes in albumin molecular mobility in the presence and absence of endothelial cells. If albumin affinity to macromolecules present in the surface layer matrix is high, freedom not only of translocation but also of rotational movements will be restricted. This effect is mechanically comparable to that of a highly viscous environment (32). On the other hand, if albumin does not interact strongly with the endothelial surface, only limited effects on its rotational motion should be observed.

An established way to determine molecular mobility is electron spin resonance (ESR) spectroscopy using markers with nitroxyl moieties as spin labels (14, 17). Well-established molecular probes for studies on biomembranes (5, 13, 23), on model membranes (1, 7), and in lipoprotein research (9) are fatty acids with a nitroxide substitution at various C-positions (17, 35). This class of compounds binds avidly to albumin (3, 44) where the spin label shows restricted mobility depending on its position along the backbone of the fatty acid (26).

Spectroscopical techniques encompass a vast range of molecular mobility determined as rotational correlation times ranging from 10¹¹ to 10³ s mainly employing two different experimental setups: conventional ESR for rotational correlation time ranging between 10¹¹ and 10⁸ s and saturation transfer ESR (ST-ESR) between 10⁷ and 10³ s (11). However, the usual approach of interpreting only out-of-phase ST-ESR spectra is not suited to reliably determine rotational correlation time values in the intermediate range between ~10⁸ and 10⁷ s, i.e., typical values for plasma proteins in aqueous solutions at body or ambient temperature. In the present study, this problem has been addressed by developing an extended approach analyzing both the in-phase and out-of-phase spectral information. The extended ST-ESR technique was employed to assess changes of albumin mobility in the presence of endothelial cells and thus obtain information on the interaction of a typical plasma protein with the endothelial surface as a critical step for the generation of the ESL.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cell preparation. Cultured umbilical cord endothelial cells (ECV304; American Type Culture Collection) were grown in medium 199 with Earle's salts (Biochrom, Berlin, Germany) containing 10% FCS, 1% L-glutamine, and 0.1% penicillin-streptomycin (10,000 E/10 mg/ml) on glass coverslips cut to the size of 6 × 60 mm and placed in petri dishes. Confluent cultures were transferred into protein-free Earle's medium 199 buffered with 20 mM HEPES (without L-glutamine and NaHCO₃) for transportation. Whole blood samples (5 ml) were obtained from a healthy volunteer by puncture of an antecubital vein; heparin (50 international units) was added to avoid clotting. Procedures were performed according to the principles outlined in the Declaration of Helsinki. Plasma and blood cells were separated by 15-min centrifugation at 1,500 g, and the plasma fraction was subsequently cleared by 5-min centrifugation at 20,000 g. Influence of whole blood plasma on the cell surface layer was studied with cultures incubated in freshly prepared human plasma for 1 h. Incubation was performed at room temperature to minimize pinocytosis and possible metabolic effects.

Spin labeling. Stearic acids labeled with 4,4-dimethyloxazolidine-N-oxyl (DOXYL; Sigma, Deisenhofen, Germany) at the positions C5 or C12 were used as spin probes. All labeling procedures started with transferring the nitroxide-labeled stearic acid to a glass surface. This was achieved by dissolving 5- or 12-DOXYL-stearic acid in ethanol (1 mg/300 µl) in a glass test tube. Ethanol was subsequently evaporated with a stream of dry and cold nitrogen, leaving labeled stearic acid as a coat on the glass surface. Coated glass tubes were stored at 20°C.

ESR spectroscopy. Coverslips with labeled cell cultures were attached to a spatula-shaped quartz holder for ESR measurements (Fig. 1) with a rim (height 0.3 mm) to allow tight adhesion of a 6 × 60-mm coverslip without further mechanical support. This arrangement allowed an aqueous layer of only 300 µm on top of the cells suitable for ESR-spectroscopy in the X-band (9.4 GHz). The cuvette with cultured cells was then placed in the center of a H₁₀₂ resonator (rectangular cavity) with parallel orientation to the Zeeman field of an X-band ESR spectrometer (model ESR 231; Academy of Science, Berlin, Germany). The same cuvette with a blank coverslip was used for reference measurements with solutions of albumin and experiments with whole blood preparations. Instrument settings common to all experiments were 9.4 GHz microwave frequency, 336 mT magnetic field center, 17 mT field sweep, 1.7 min scanning time, and 0.1 s time constant. For conventional ESR, microwave power was adjusted to 20 mW to achieve maximum sensitivity without spin saturation, and the magnetic field was modulated with 100 kHz and an amplitude of 0.1 mT.

View larger version (26K): [in this window] [in a new window]   Fig. 1.   Cuvette with holder and coverslip. Sample liquid fills the volume between the hollowed quartz holder (300 µm deep) and the coverslip (6 × 60 mm) coated with cultured endothelial cells.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Complex saturation transfer electron spin resonance (ESR) spectra. A: baseline corrected and smoothed in-phase (solid lines) and out-of-phase spectra (dashed lines) of 12-4,4-dimethyloxazolidine-N-oxyl (DOXYL)-stearic acid bound to albumin at 3°C ( = 98 ns) and at 37°C ( = 48 ns). B: phase diagrams (out-of-phase spectra plotted against corresponding in-phase values) of the same spectra. Aspect ratio of the enclosing box with minimal area in the phase diagrams (arrows, phase ratio) was used to estimate the rotational correlation time () in addition to the amplitude ratio.

Data processing. Recorded spectra were interactively processed with a program written in Pascal (Delphi, Borland Software). Conventional ESR spectra were smoothed by a rectangular filter that best conserves peak heights. Only in cases of a smooth transition from signal peaks to noise was a Gauss filter preferred, which tends to damp peak heights but avoids oscillation artefacts typical for rectangular curve filtering in such cases. The filter parameter was adjusted according to the transition between the slope of the signal power and the noise plateau in the power spectrum. Smoothing of spectra was essential to precisely determine the location of maxima and minima indicating parallel and perpendicular tensors (T and T) and to allow quantification of the degree of anisotropy. From the distances in the spectrum corresponding to these tensors, the order parameter (S) was automatically derived using values of A = 0.61, 0.61, or 3.24 mT for nitrogen hyperfine splitting (14, 35). Order parameter values have been converted into rotational correlation time () by an empirical relationship (38). The dimensionless order parameter S (35), with theoretical values between 0 and 1, indicates the degree of mobility of the spin probe between total immobilization (S = 1, glass configuration, no motion) and maximal freedom of rotation (S = 0, ideal isotropic spectrum). However, the order parameter S is only applicable in a range of rotational correlation time between ~10¹⁰ and 10⁸ s. This is a typical range for spin probes incorporated into biomembranes, and respective studies often express molecular mobility in terms of the order parameter. However, especially for high values of the order parameter, these conversions bear a large degree of uncertainty. Therefore, the present study uses rotational correlation time instead of the order parameter S wherever possible.

Statistics. Rotational correlation times resulting from the ST-ESR experiments were treated separately for the 5-DOXYL- and 12-DOXYL-stearic acid groups. Dependency of rotational correlation time on temperature was investigated by linear or quadratic polynomial regression analysis. Differences between cell-free albumin solution and cell preparations with and without preincubation with albumin were tested by comparing results of the regression analysis employing the F-test based on the average vertical distance between regression lines and the sum of squared differences from the respective regression line at each temperature. The existence of differences was additionally tested for three temperature levels (37°C, 20°C, and 3°C) separately using the F-test and the t-test in cases where the F-test indicated nonhomogeneity. Significance was assumed at P > 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Conventional ESR of labeled endothelial cells. For endothelial cells prelabeled with 5-DOXYL-stearic acid in protein-free culture medium, Fig. 3 shows the ESR-signal intensity as a parameter for stability of the nitroxide label with time at 37°C. In the absence of albumin, results indicate a slow signal decay consistent with a half life of the label of ~30 min (Fig. 3A). Replacing the supernating medium with albumin solution further stabilized the signal intensity rendering a decay hardly determinable in the time course of the experiment and resulted in a step increase of the order parameter indicating a decreased mobility of the spin probe. Subsequent flushing with ascorbic acid reduced the signal intensity to nearly noise level between the two consecutive measurements (<3 min). In contrast, ascorbic acid did not rapidly extinguish the nitroxide label incorporated into cell membranes incubated in albumin-free medium (Fig. 3B) or label bound to albumin in the absence of cells (Fig. 3C).

View larger version (19K): [in this window] [in a new window]   Fig. 3.   A: results of conventional ESR for endothelial cells loaded with 5-DOXYL-stearic acid in an albumin-free environment at 37°C. Signal intensity (left ordinate) decreased slightly with time. Order parameter (S, right ordinate) exhibits typical values for lipid membrane bilayers (0.6-0.7) that after the addition of albumin increase to ~0.7-0.8, typical for the spin probe bound to protein. Subsequent addition of ascorbate nearly eliminated the specific signal. Ascorbate added to labeled endothelial cells in the absence of albumin (B) or to labeled albumin in the absence of cultured cells (C) only led to a gradual reduction of signal intensity.

ST-ESR measurements of spin-labeled albumin in presence and absence of cells. Temperature dependencies of rotational correlation time for albumin labeled with either 5- or 12-DOXYL-stearic acid are summarized in Fig. 4. As expected, rotational correlation time correlated inversely with temperature in all cases with a correlation coefficient (r) of 0.764 for 5-DOXYL- stearic acid and of 0.887 for 12-DOXYL-stearic acid-labeled albumin, but on different levels. From 37°C down to 3°C, rotational correlation time values for 5-DOXYL-stearic acid increased from 26 to 64 ns (Fig. 4A) and those for 12-DOXYL-stearic acid, from 48 to 98 ns (Fig. 4B). The rotational correlation time for 12-DOXYL-labeled albumin was for all temperatures similar to values estimated from physical principles (Stokes-Einstein relation) based on medium viscosity and the molecular radius of albumin. This suggests a fairly rigid coupling of 12-DOXYL-stearic acid mobility to those of the albumin molecule.

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Rotational correlation times (; means ± SE) of spin-labeled albumin in presence and absence of cell surfaces. A: 5-DOXYL-stearic acid bound to albumin showed a decrease of rotational mobility (increasing rotational correlation time values) with decreasing temperature. Presence of endothelial cells caused a shift to higher rotational correlation time values that was enhanced by preincubation of the cells with plasma. B: 12-DOXYL-stearic acid bound to albumin exhibits lower mobility that was not significantly altered by presence of endothelial cells.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In the present study, ESR spectroscopy using nitroxide spin probes was used to characterize molecular interactions of albumin with the surface of endothelial cells. Two distinct spectroscopical effects caused by the anisotropy of the nitroxyl moiety are usually employed to measure molecular mobility: distortion of spectral lines and phase shift due to spin saturation. The range of rotational motions accessible with standard implementation of these two techniques was enlarged by introducing a two-dimensional interpretation of ST-ESR spectra. To our knowledge, this is the first study using ST-ESR spectroscopy to investigate molecular rotational motion in the range between 10⁸ and 10⁷ s typical for plasma proteins like albumin in aqueous solution at body temperature.

Interpretation of the rotational correlation time of spin probes necessitates knowledge about the location of the label within the system. Two initial locations for spin probes were used here: binding sites for fatty acids at the albumin molecule and endothelial cell plasma membranes. The result obtained with labeled endothelial cells (Fig. 3) was typical for biomembranes (41). Slow reduction of signal intensity with time can be explained by losses of the nitroxide spin probe due to inward translocation (28), distribution throughout the cell (25), and subsequent metabolic reduction (39). This behavior changed significantly on introduction of albumin into the system. Albumin has a significant affinity to bind fatty acids such as the spin probes used in this study (44, 45), and it removes spin labels from a variety of lipid membranes (19, 21). Accordingly, mobility of the spin label is slowed down after the addition of albumin as would be expected from uptake of the label from endothelial cells. The fast change in order parameter indicates that the label had been predominantly located in the outer cell membrane accessible to albumin. Furthermore, the label was apparently more stable with time when bound to albumin compared with cell membranes probably due to the absence of cellular degrading mechanisms.

For plain albumin incubated with ascorbic acid, a slow decay of the spin label with time (and a slight increase of the order parameter that might be linked to a slow protein precipitation process) is observed. In sharp contrast, the label was destroyed immediately on the addition of ascorbate in the presence of endothelial cells (Fig. 3). This could indicate that the spin probe is not permanently bound to albumin but in a fast mutual exchange with the cellular membrane (Fig. 5). Because albumin affinity is much higher than that of the plasma membrane, most of the label will be attached to the protein at any time. During fast transition between the two locations, the nitroxide would become vulnerable to destruction by the hydrophilic ascorbate.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Schematic interpretation of obtained results. Labeled stearic acids incorporated into plasma membranes (not drawn to scale) or albumin are protected from destruction by ascorbic acid. If both plasma membranes and albumin are present (bottom), most of the label is bound to albumin at any time. The fast destruction of the label by ascorbate, however, indicates a frequent exchange between the plasma membrane and the albumin.

The data shown in Fig. 3 also provide evidence that albumin added to the culture medium penetrates the ESL to take up the spin probe from the plasma membrane within the sample preparation time (~2-3 min) as indicated by the rapid change of order parameter after the addition of albumin. This finding may be compared with observations by Vink and Duling (42) in intravital microscopic studies observing diffusion of anionic macromolecules into the ESL. Albumin and fibrinogen equilibrated between blood and the ESL with a half time of ~40 min. The much shorter interval needed in the present approach for removal of fatty acids from the endothelial membranes may relate to the fact that no complete equilibration is needed for this process. Alternatively, it could indicate that the ESL present in vivo is more complex than that on the cultured endothelial cells (even in the presence of blood plasma) despite the fact that the existence of a glycocalyx on ECV304 cells has been reported in the literature (18, 33).

Interactions of albumin with the ESL were further investigated by comparing results obtained with two different spin labels. The nitroxyl moities of 5- and 12-DOXYL-stearic acid are differently attached to the protein resulting in different rotational correlation times (24, 26) over a large temperature range (Fig. 4). Differences in rotational freedom relate to the binding characteristics for fatty acids in hydrophobic pouches inside the albumin molecule (3, 10, 45). Nitroxide of the 12-DOXYL probe appears to be sufficiently immobilized to report rotational motion of the entire albumin molecule, whereas the label of the 5-DOXYL probe located at the surface of the albumin molecule is more susceptible to intermolecular contacts. As a consequence, the presence of an ESL matrix may influence the rotation of the peripheral 5-DOXYL label, whereas the 12-DOXYL label in a deeper position with restricted local rotation would mainly be affected by comparatively stable bonds of the carrier molecule albumin. The data for 12-DOXYL may hint at a slight change in the rotational dynamics on contact with the ESL expressed in an altered temperature dependency rather than in an increase of rotational correlation time. Taken together, the experimental results support the concept of albumin being able to diffuse through the ESL while superficially interacting with the gel matrix macromolecules.

These results could help to understand the physiological double role of albumin at the endothelial surface: interaction with gel-like surface structures and shuttling poorly water-soluble substances (22) that would be excluded by persistent incorporation into local macromolecular networks. However, despite the lack of static links to a macromolecular network, albumin may contribute to the establishment of the spatial dimensions of the ESL (30), e.g., by transiently filling hydrophilic pores of the ESL. Weak and short-lived links sufficient to explain such effects do not necessarily impair rotation of the protein significantly. Without permanent binding to stationary molecules, albumin could frequently exchange with the luminal fluid phase and thus assume its transport function for hydrophobic substrates through the ESL.