Adhesion of malignant mammary tumor cells MDA-MB-231 to microvessel wall increases microvascular permeability via degradation of endothelial surface glycocalyx

Abstract

To investigate the effect of tumor cell adhesion on microvascular permeability (P) in intact microvessels, we measured the adhesion rate of human mammary carcinoma MDA-MB-231, the hydraulic conductivity (L_p), the P, and reflection coefficient (σ) to albumin of the microvessels at the initial tumor cell adhesion and after ∼45 min cell perfusion in the postcapillary venules of rat mesentery in vivo. Rats (Sprague-Dawley, 250–300 g) were anesthetized with pentobarbital sodium given subcutaneously. A midline incision was made in the abdominal wall, and the mesentery was gently taken out and arranged on the surface of a glass coverslip for the measurement. An individual postcapillary venule was perfused with cells at a rate of ∼1 mm/s, which is the mean blood flow velocity in this type of microvessels. At the initial tumor cell adhesion, which was defined as one adherent cell in ∼100- to 145-μm vessel segment, L_p was 1.5-fold and P was 2.3-fold of their controls, and σ decreased from 0.92 to 0.64; after ∼45-min perfusion, the adhesion increased to ∼5 adherent cells in ∼100- to 145-μm vessel segment, while L_p increased to 2.8-fold, P to 5.7-fold of their controls, and σ decreased from 0.92 to 0.42. Combining these measured data with the predictions from a mathematical model for the interendothelial transport suggests that tumor cell adhesion to the microvessel wall degrades the endothelial surface glycocalyx (ESG) layer. This suggestion was confirmed by immunostaining of heparan sulfate of the ESG on the microvessel wall. Preserving of the ESG by a plasma glycoprotein orosomucoid decreased the P to albumin and reduced the tumor cell adhesion.

tumor metastasis is the leading cause of cancer-related death among cancer patients (24, 51). Tumor cell adhesion to, and extravasation out of, the microvessel wall are critical steps in tumor metastasis (51). To search for effective anti-metastatic therapies, many in vivo, ex vivo, and in vitro studies have been conducted to understand underlying molecular mechanisms by which tumor cells interact with endothelial cells lining the microvessel wall for the adhesion and extravasation (8, 11, 17, 18, 25, 26, 33, 47, 48). Development of advanced intravital and multiphoton microscopy and fluorescence labeling techniques has enabled the observation of dynamic tumor adhesion and extravasation processes in vivo in the target organs after injecting tumor cells through systemic administration (27, 30). Although recent in vivo study (30) reported that single cancer cell MDA-MB-435 adhesion and extravasation during brain metastasis induced endothelial remodeling, to date, however, little has been learned about whether tumor cell adhesion affects microvessel wall integrity and how. Therefore, the objective of this study was to investigate the structural mechanisms by which tumor cell adhesion affects microvessel wall integrity [or permeability (P)].

The luminal surfaces of endothelial cells (ECs) that line our vasculature are coated with a glycocalyx of membrane-bound macromolecules composed of sulfated proteoglycans, hyaluronic acid, sialic acids, glycoproteins, and plasma proteins that adhere to this surface matrix (38, 47). This endothelial surface glycocalyx (ESG) plays an important role in regulating vascular P, attenuating interactions between circulating blood cells and the ECs, as well as sensing the hydrodynamic changes in the blood flow (12–14, 40, 42, 44, 53). Damage to and modification of the ESG were found in many diseases, such as diabetes, ischemia, myocardial edema, chronic infectious diseases, and atherosclerosis (7, 9, 44, 54, 56). Previous studies showed that a tumor secretion, vascular endothelial growth factor (VEGF), enhanced microvessel P by increasing the gap between adjacent ECs and degrading the ESG (5, 21). VEGF also increased the tumor cell adhesion to the endothelium in static conditions (18, 33) and under flow (47). In addition to ESG, the junction strands in the cleft between adjacent ECs, the junction pores, and the gap spacing of the cleft are the determinants of the microvessel wall integrity (or P) (2, 37). To adhere to the ECs forming the microvessel wall, a circulating tumor cell must first overcome the ESG barrier, which is 0.1- to 1-μm thick (1, 3, 23, 49, 58). To further extravasate to the surrounding tissue, an arrested tumor cell has to overcome the microvessel wall barriers, either the paracellular barrier (junction strands between adjacent ECs) or the transcellular barrier (ECs).

Prior studies have shown that the microvessel wall integrity can be represented by the microvessel P, which can be directly measured in vivo (20, 21, 34, 37, 47). There are three types of P coefficients: the hydraulic conductivity (L_p) is the microvessel P to water; the solute P is the vessel P to a solute, and the reflection coefficient σ represents the selectivity of the microvessel wall to a solute. In addition to direct measurements from the experiments, these three coefficients can be predicted from mathematical models for the transport through the interendothelial cleft between adjacent ECs (22, 34, 52).

On the basis of our laboratory's previous studies, which showed that VEGF increases microvessel P and tumor cell adhesion by possibly degrading ESG (18–21, 47), we hypothesized that the most likely structural mechanism by which tumor cell adhesion affects microvessel wall integrity is to degrade the ESG. To test this hypothesis, we first quantified the initiation time and adhesion rate of human malignant breast cancer cells MDA-MB-231 in individually cannulated postcapillary venules under the perfusion velocity of ∼1 mm/s, which is the mean blood flow velocity in this type of vessels. The perfusion rate of tumor cells is about one cell per second. Second, we measured the microvessel L_p, solute P, and σ to albumin (Stokes radius ∼3.5 nm) at the initial tumor cell adhesion and after 45-min perfusion of tumor cells. Third, we applied a mathematical model for the interendothelial cleft (revised from Ref. 22) to predict the possible changes in L_p and P to albumin by changing the structural components of the microvessel wall, e.g., degrading the surface glycocalyx, increasing the gap between ECs, and increasing the size or the number of junction pores. Fourth, we compared the measured L_p and P to albumin with those predicted and found the most likely structural changes induced by the tumor cell adhesion. Finally, the predicted structural change was confirmed by the immunostaining of heparan sulfate (HS), the most abundant glycosaminoglycans (GAGs) in the ESG (53).

Orosomucoid, or α₁-acid glycoprotein, is a plasma protein essential for the maintenance of stable solute P of microvessels by enhancing the charge and organization of the ESG (35, 46, 59). To further test that preserving of the ESG can reduce tumor cell adhesion to the microvessel wall, we pretreated the microvessel wall with orosomucoid and found that it decreased the P to albumin and indeed reduced the tumor cell adhesion to the vessel wall. Understanding the structural mechanisms by which tumor cell adhesion acts on the microvessel integrity (or P) is important for finding new strategies against tumor metastasis.

MATERIALS AND METHODS

Animal preparation.

All experiments were performed on female Sprague-Dawley rats (250–300 g, age 3–4 mo), supplied by Hilltop Laboratory Animals (Scottsdale, PA). All procedures were approved by the Animal Care and Use Committees at the City College of the City University of New York. The methods used to prepare rat mesenteries, perfusate solutions, and micropipettes for microperfusion experiments have been described in detail elsewhere (20, 47). A brief outline of the methods was given below, with emphasis on the special features of the present experiments. At the end of experiments, the animals were euthanized with excess anesthetic. The thorax was opened to ensure death.

Rats were first anaesthetized with pentobarbital sodium given subcutaneously. The initial dosage was 65 mg/kg, and an additional 3 mg/dose was given as needed. After a rat was anesthetized, a midline surgical incision (2–3 cm) was made in the abdominal wall. The rat was then transferred to a tray and kept warm on a heating pad. The mesentery was gently taken out from the abdominal cavity and spread on a glass coverslip, which formed the base of the observation platform as previously described (47). The gut was gently pinned out against a silicon elastomer barrier to maintain the spread of the mesentery. The upper surface of the mesentery was continuously superfused by a dripper with mammalian Ringer solution at 35–37°C, which was regulated by a controlled water bath and monitored regularly by a thermometer probe. The microvessels chosen for the study were straight nonbranched postcapillary venules, with diameters of 35–50 μm. All vessels had brisk blood flow immediately before cannulation and had no marginating white cells.

Cell culture.

Human breast carcinoma (MDA-MB-231) cells were purchased from ATCC (Manassas, VA) and cultured in a humidified atmosphere of 5% CO₂ and 95% air at 37°C in 75-cm² plastic tissue culture flasks in DMEM, supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 mg/ml streptomycin sulfate, all from Sigma (St. Louis, MO). Cells were seeded at 10⁶ cells/ml and grown to confluence (∼90%) in ∼3 days and routinely passaged using trypsin/EDTA (ATCC) at ratio 1:4 (17). Human nontumorigenic breast epithelial cell line MCF-10A (ATCC) were cultured in 75-cm² plastic tissue culture flasks with MEGM Bullet Kit (Lonza Walkersville, MD, catalog no. CC-3150) supplemented with 100 ng/ml cholera toxin.

On the day of experiment, cells were collected by brief trypsinization, then counted and suspended in phosphate-buffered saline (PBS; Sigma). To remove any remaining cell clumps, the cell suspension was filtered through a 40-μm nylon mesh. Then tumor cells were fluorescently labeled using 1 μM calcein acetoxymethyl ester (AM) or Cell Tracker Orange (Invitrogen, Eugene, OR) in serum-free medium for 30 min. Concentration of cell suspension was adjusted for the final perfusate ∼4 million/ml in 1% BSA mammalian Ringer. The cell survival rate was >95% before perfusion and >90% after ∼1.5-h perfusion at driving pressures of 1–10 cmH₂O in perfusing micropipettes.

Solutions and reagents.

Mammalian Ringer solution was used for all dissections, perfusate, and superfusate. The solution composition was (in mM) 132 NaCl, 4.6 KCl, 1.2 MgSO₄, 2.0 CaCl₂, 5.0 NaHCO₃, 5.5 glucose, and 20 HEPES. Its pH was balanced to 7.4 by adjusting the ratio of HEPES acid to base. In addition, the perfusate into the microvessel lumen contained bovine serum albumin (BSA) (Sigma) at 10 mg/ml (1% BSA-Ringer solution). Calcein-AM and Cell Tracker Orange were purchased from Invitrogen (Eugene, OR). Tetramethylrhodamine isothiocyanate-bovine serum albumin (TRITC-BSA, molecular weight 67,000, Stokes radius ∼3.5 nm, A2289, Sigma-Aldrich, St. Louis, MO) was dissolved in 1% BSA-Ringer solution at concentration 0.75 mg/ml. The orosomucoid (G3643, α₁-acid glycoprotein, Sigma) concentration used in the experiments was 0.1 mg/ml in 1% BSA-Ringer. FITC-conjugated mouse anti-human HS (anti-HS, 10e4 epitope) was purchased from United States Biological (Swampscott, MA). It was diluted to 1:50 (20 μg/ml) in 1% BSA-Ringer solution for labeling HS in the microvascular ESG. F. heparinum heparinase III (IBEX) is selectively active only toward HS (10). Heparinase III (50 mU/ml) was used to digest the ESG. A blocking solution was made of 5% goat serum (Invitrogen, Eugene, OR) in 1% BSA-Ringer. All of the solutions described above were made at the time when the experiment was performed and were discarded at the end of the day.

MDA-MB-231 cell adhesion in individually perfused microvessels.

To measure the tumor cell adhesion rate under a controlled condition, a single straight postcapillary venule (35- to 50-μm diameter) was cannulated with a micropipette (∼30 μm tip diameter, WPI, Sarasota, FL) filled with 1% BSA-Ringer solution containing ∼4 million cells/ml. The venule was perfused with a driving pressure of ∼10 cmH₂O to maintain a flow velocity of ∼1 mm/s, which is a typical blood flow velocity in the postcapillary venules of rat mesentery. The perfusion flow velocity was determined by the driving pressure and was calculated from the movement of a marker tumor cell (47). We also measured the rate of cells out of the micropipette, which was ∼1 cell/s at the perfusion velocity of ∼1 mm/s. The adhesion process was recorded at ∼2 frames/s in a ∼1-min interval for ∼60 min in each experiment. A single experiment was carried out in one microvessel per animal. Since we used ×20/numerical aperture (NA) 0.75 objective lens to observe the cell adhesion, which has a depth of light collection ∼100 μm (60), the cells adhering at the top and bottom of the vessel can also be observed when we focus at the midplane of the vessel. Therefore, we used the midplane area (diameter × length) as the unit for the cell adhesion instead of the length of the vessel. The 5,000-μm² midplane area is equivalent to 100- to 143-μm length for 35- to 50-μm diameter vessels. Due to the limitation of the two-dimensional (2D) images, we used the midplane area as the projection of the surface area of the vessel where the cells adhere.

To test the effect of surface glycocalyx on tumor cell adhesion to the microvessel wall, we pretreated the microvessels with 0.1 mg/ml orosomucoid for 30 min before perfusing the cells. Orosomucoid has been shown to alter the microvessel P by enhancing the charge of ESG (52). Metzger et al. (36) reported that orosomucoid concentration in the rat serum of normal Sprague-Dawley rats is 0.09–0.8 mg/ml, nearly 10 times fluctuation. Curry et al. (16) also found that there was no significant difference in the vessel P to charged solutes between 0.1 and 1 mg/ml orosomucoid treatments. We thus used 0.1 mg/ml orosomucoid in the present study.

Measurement of microvascular L_p.

L_p was determined using the modified Landis technique (15, 34), which measured the volume flux of water across the microvessel wall. Briefly, a micropipette with the tip diameter of 10–15 μm, filled with the marker cells (red blood cells from another rat) in 1% BSA-Ringer, was used to cannulate a single microvessel of rat mesentery. The micropipette was connected to a water manometer. The pressure was set to 30–70 cmH₂O for perfusion. For each measurement, the perfused vessel was occluded with a glass rod at downstream at least 800 μm from the cannulating site. L_p was calculated from the Starling equation, L_p = (J_v/S)₀/Δp_eff, where Δp_eff is the effective hydrostatic and oncotic pressure difference across the microvessel wall, and (J_v/S)₀ is the initial water flux across the microvessel wall, which was calculated from the velocity of the marker cell after the vessel was occluded. For the size of postcapillary venules (35- to 50-μm diameter) in our study, we calculated the mean fluid velocity by the centerline marker cell velocity divided by 2(1 − r_m²/2R_v²) due to the velocity profile deviation from the parabolic Poiseuille flow (38). Here r_m is the radius of the marker cell (rat red blood cells), and R_v, the radius of the vessel. The bulk flow rate (J_v) during the microocclusion was determined by the corrected mean velocity multiplied by the cross-sectional area of the vessel. The details for how to determine (J_v/S)₀ were described in Refs. 15, 34, 38. We measured L_p under control without tumor cell adhesion, at initial tumor cell adhesion, and after 45-min tumor cell perfusion and adhesion. To ensure that any changes in L_p after cannulation and perfusion were not caused by the cannulation procedure alone, in a matched sham control group, we measured the baseline L_p of a vessel perfused with 1% BSA-Ringer and then recannulated the same vessel and measured L_p in the same control solution for ∼10 min. During the replacement of the pipette, the pressure was dropped to <1 cmH₂O, so flow at the tip during recannulation could be neglected. The pressure was then set back to 30–60 cmH₂O for perfusion.

Due to the limitation of our technique, we could only cannulate once on each microvessel for L_p measurement during tumor cell adhesion. To determine the effect of tumor cell adhesion on L_p, we thus measured L_p at the initial tumor cell adhesion and after ∼45 min tumor cell perfusion and adhesion in different vessels. The initial adhesion was defined as the moment when there was one cell adhering in ∼100- to 145-μm vessel segment (∼5,000 μm² midplane area). At the initial cell adhesion, or after a vessel was perfused for ∼45 min with the 1% BSA-Ringer containing ∼4 million/ml tumor cells, a micropipette containing 1% BSA-Ringer and the marker red blood cells was used to cannulate the microvessel and L_p was measured immediately and then every 1–3 min for ∼10 min.

Measurement of apparent microvascular solute P.

To investigate the structural mechanisms by which tumor cell adhesion increases microvascular P, we also measured apparent vascular solute P to albumin (TRITC-BSA). Measurement of P was taken on the individual postcapillary venules under control without tumor cell adhesion, at the initial adhesion and after 45 min tumor cell perfusion and adhesion. The detailed method using θ pipette for P measurement was previously described in Refs. 20, 34, 47. Briefly, when TRITC-BSA was perfused into the vessel and the vessel was exposed to a 540-nm wavelength light, the images were recorded simultaneously by a high-performance digital 12-bit charge-coupled device (CCD) camera (SensiCam QE, Cooke, Romulus, MI) with a Super Fluor ×20 objective lens (NA = 0.75, Nikon). Then the P was determined offline. The total fluorescence intensity (I) in the lumen of a straight vessel and surrounding tissue was determined by image analysis software (Intracellular Imaging, Cincinnati, OH). The measuring window was 300- to 500-μm long and 100- to 200-μm wide and was set at least 100 μm from the cannulation site and from the base of the bifurcation to avoid solute contamination from the cannulation site and from the side arms. P was calculated by P = (1/ΔI₀)(dI/dt)₀(r/2), where ΔI₀ was the step increase in fluorescence intensity in the measuring window when the perfused dye just filled up the vessel lumen, (dI/dt)₀ was the initial rate of increase in fluorescence intensity after the dye filled the lumen and began to accumulate in the tissue, and r was the vessel radius. The assumption for using the above equation for determining the P was that the fluorescence intensity is linearly related to the fluorescence concentration. We did in vitro calibration experiments to test this assumption, as described Liu et al. (34). We used the same instrument settings in the calibration experiments as those used in the P measurement. The linear range of TRITC-BSA concentrations was from 0 to 1.25 mg/ml under our settings. We thus chose 0.75 mg/ml TRITC-BSA in our experiments.

To determine the effect of tumor cell adhesion on P, we measured P at the initial tumor cell adhesion and after ∼45 min of tumor cell perfusion and adhesion. At the beginning of tumor cell adhesion, or after a vessel was perfused for ∼45 min with the 1% BSA Ringer containing ∼4 million/ml tumor cells, a θ pipette containing 1% BSA Ringer and TRITC-BSA was used to cannulate the microvessel and P was measured immediately and then every 1–3 min for 10 min. To determine the effect of orosomucoid on P to albumin, the perfusate additionally contained 0.1 mg/ml orosomucoid in both washout and dye sides of the θ pipette. P was measured after pretreatment of the vessel with orosomucoid for 30 min.

To test whether increases in P were purely due to tumor cell secretions, after 1-h incubation of 4 million/ml tumor cells, the supernatant was collected to test its effect on microvessel P to albumin. In the same vessel, after the control measurement with 1% BSA-Ringer, the θ pipette was replaced with another one with the supernatant in both washout and dye sides and the P was measured every 1–5 min up to 60 min.

Measurement of microvascular σ to albumin.

The technique for measuring microvascular σ to a solute was detailed in Ref. 15, which is similar to that for measuring L_p. A micropipette containing 5% BSA-Ringer was used to cannulate and perfuse a microvessel. After tumor cell initial adhesion or after 45-min perfusion, the marker cell velocity was recorded under three perfusion pressures with ∼10-cmH₂O increment when the downstream of the vessel was completely occluded. The marker cell velocity (y-axis) vs. perfusion pressure (x-axis) was plotted. Extrapolation of a straight line passing those three points intercepts x-axis. From the equation J_v/S = L_p (Δp − σΔπ), the value at the intercept gives σΔπ = Δp. Therefore, σ = Δp/Δπ. Here, Δp is the perfusion pressure at the intercept when the marker cell velocity is zero; Δπ is the osmotic pressure of the perfusate. For 5% BSA-Ringer at 37°C, Δπ = 20 cmH₂O.

Determination of diffusive solute P from measured apparent P.

Since solute flux can be coupled to water flow (solvent drag), the albumin P measured in our experiments (apparent P) tends to overestimate the true diffusive P (P_d) of albumin. Using the L_p and σ to albumin σ, which were measured under control and tumor cell adhesions, we calculated the P_d to albumin by employing the following formula in Refs. 15, 20, 21.

$$P=P_{\text{d}}\frac{{\text{P}}_{\text{e}}}{\exp \left({\text{P}}_{\text{e}}\right)-1}+\text{Lp}\left(1-\sigma \right)\Delta {\text{p}}_{\text{eff}}$$ (1)

$${\text{P}}_{\text{e}}=\frac{\text{Lp}\left(1-\sigma \right)\Delta {\text{p}}_{\text{eff}}}{P_{\text{d}}}$$ (2)

$$\Delta {\text{p}}_{\text{eff}}=\Delta \text{p}-{\sigma }^{\text{albumin}}\Delta {\pi }^{\text{albumin}}-{\sigma }^{\text{TRITC}-\text{albumin}}\Delta {\pi }^{\text{TRITC}-\text{albumin}}$$ (3)

Here Δp_eff is the effective filtration pressure across the microvessel wall; Δp and Δπ are the hydrostatic (perfusion) and osmotic pressure drops across the microvessel wall, respectively; and P_e is the Peclet number. We assumed σ^albumin = σ^{TRITC-albumin} = σ in our study.

Immunolabeling and quantification of microvessel ESG.

To investigate the effect of tumor cell adhesion on the ESG of the microvessel wall, FITC-conjugated antibody was used to label HS, one of the most abundant GAGs forming the ESG (44, 53). Similar to the method in Yen et al. (58), a postcapillary venule (35- to 50-μm diameter) of rat mesentery was cannulated by a θ micropipette. The upper surface of the mesentery was continuously superfused by a dripper with a mammalian Ringer solution at 4°C, which was regulated by a controlled water bath with ice and monitored using a thermometer probe. The vessel was first perfused for 15 min with a blocking solution of 5% goat serum containing 1% BSA-Ringer through one lumen of θ pipette. Then the perfusion was switched to another lumen of the pipette to inject FITC-conjugated anti-HS in 1% BSA-Ringer (20 μg/ml) into the microvessels for ∼2.5 h. The 2.5 h was long enough to allow FITC-anti-HS to infiltrate the entire depth of the ESG (58). After 15-min perfusion of the first perfusate to wash away the free dye, the vessel with fluorescently labeled glycocalyx (focused at the midplane of a vessel) was imaged by the same imaging system used in the P measurement. The intensity of the fluorescently labeled glycocalyx in the vessel segment was measured by InCyt Im imaging and analyzing system (Intracellular Imaging, Cincinnati, OH). To test the assumption that the fluorescence intensity is linearly related to the amount of the fluorescently labeled glycocalyx, we did in vitro calibration experiments. We used the same instrument settings in the calibration experiments as those used in the in vivo measurement of the fluorescently labeled glycocalyx. The linear range of FITC-anti-HS concentrations was from 0 to 50 μg/ml under our settings. We thus chose 20 μg/ml FITC-anti-HS in our experiments. We determined the amount of the fluorescently labeled glycocalyx in the vessels under control, at the initial tumor cell adhesion, after 45-min tumor cell perfusion, and pretreated with 0.1 mg/ml orosomucoid in 1% BSA Ringer for 30 min.

To show the location of adherent tumor cells, in some of the vessels, we used confocal microscopy for the three-dimensional scanning. After wash away the free dye, the vessel was fixed by superfusing the tissue with ice-cold 1% paraformaldehyde for 3–5 min. The animal was killed by anesthetic overdosing. The vessel was perfused until no blood circulation was observed in the tissue, and the cannulating micropipette was pulled out of the vessel. The tissue (∼1 cm × 1 cm) surrounding the vessel was then dissected, rinsed with ice-cold PBS, and mounted on a glass coverslip. A secured-seal spacer (Invitrogen, Eugene, OR) was used to surround the tissue and to make a well about 120-μm deep between two coverslips to retain the three-dimensional structure of the vessels.

Intravital and confocal microscopy.

A Nikon Eclipse TE2000-E inverted fluorescent microscope was used to observe the mesentery. A ×10 lens (NA 0.3, Nikon) gave a field of view of ∼2 mm in diameter. The tissue was observed with either transmitted white light from a light pipe suspended above the preparation, or with fluorescent light from an illumination system (the monochromator with a xenon lamp; FSM150Xe, Bentham Instrument). The monochromator can generate the light of wavelength from 200 to 700 nm. The observation of fluorescently labeled glycocalyx and fluorescently labeled tumor cells and measurement of P to TRITC-BSA were done by a high-performance digital 12-bit CCD camera (SensiCam QE, Cooke, Romulus, MI) with a Super Fluor ×20 objective lens (NA = 0.75, Nikon) and recorded by InCyt Im imaging and analyzing system (Intracellular Imaging, Cincinnati, OH). The observation of the marker cell movement in L_p measurement was by a CCD video camera (CV-M50, JAI, Japan) and recorded on a VCR.

The cross-sectional view of a vessel with the fluorescently labeled glycocalyx and adherent tumor cells were observed using 12-bit laser scanning confocal microscopy (Zeiss LSM 510 Confocal Microscope System) with a ×40/NA 1.3 objective lens. Excitation/emission wavelength (nm) = 490/525 for FITC and 548/576 for Cell Tracker Orange. Images were collected from the top (near lens) to the bottom (z-direction) for each sample, forming a stack of images along the z-direction. The thickness of each image was 0.2–0.3 μm. The image stacks were analyzed with the public domain National Institutes of Health IMAGE J program. Stacks were reconstructed as a three-dimensional view and resliced into ∼1-μm-thick cross sections along the vessel axis.

Analysis and statistics.

Measurements of L_p or P during the control period in a vessel were averaged to establish a single value for a baseline L_p or P. This value was then used as a reference for all subsequent measurements on that vessel. Results for vessel P and σ, cell adhesion, and fluorescently labeled glycocalyx are presented as the means ± SE, unless specified otherwise. Statistical significance of the treatment over time was tested with a nonparametric Wilcoxon signed-rank test applied to the averaged P data. The Mann-Whitney U-test was applied to between-group data to test for differences at specific times. Significance was assumed for probability levels P < 0.05.

RESULTS

MDA-MB-231 adhesion to the microvessel wall.

Figure 1 summarizes MDA-MB-231 cell adhesion to the microvessel wall as a function of time. At a specific time, the cell adhesion was expressed as the number of adherent cells per 5,000 μm² of the midplane area in a vessel segment (∼100- to 145-μm long). The initial adhesion time was defined as the moment when there was one adherent cell in 5,000-μm² vessel segment. The initial adhesion time was 6.6 ± 0.8 (SE) min in 14 vessels. Figure 1 also demonstrates that, under ∼1 mm/s perfusion velocity, the mean blood flow velocity in this type of vessels, the cell adhesion increased linearly with the time during 60-min perfusion. The adhesion rate (the slope of the adhesion vs. time curve) was 1.2 cells per 10 min in ∼100- to 145-μm long vessel segment (R² > 0.99).

Fig. 1.

Adhesion of MDA-MB-231 cells as a function of time. The cells at a concentration of ∼4 million/ml in 1% BSA-Ringer solution were perfused into a single microvessel at a velocity of ∼1 mm/s, which is the normal mean blood flow velocity in postcapillary venules. The cell perfusion rate was ∼1 cell/s. The initial adhesion time in n = 14 vessels was 6.6 ± 0.8 min (mean ± SE). The initial adhesion time was defined as one adherent cell in a vessel segment of 5,000-μm² midplane area (∼100–143 μm long).

MDA-MB-231 adhesion increases microvascular L_p.

To investigate whether MDA-MB-231 adhesion increases microvascular L_p, we directly measured the L_p after tumor cell adhesion. Due to the limitation of our measurement technique, we can only measure L_p at one specific time for a vessel during tumor cell adhesion. We thus measured L_p at the initial adhesion in one group of vessels and after ∼45-min cell perfusion and adhesion in another group. Figure 2A shows the effect of tumor cell adhesion on L_p. At tumor cell initial adhesion, i.e., after 6.3 ± 0.6 (SE) min cell perfusion, L_p was 1.6 ± 0.1 × 10⁻⁷ cm·s⁻¹·cmH₂O⁻¹ in 15 vessels, and no significant change in another 10 min (P > 0.3). After 45-min perfusion and adhesion, L_p increased to 2.8 ± 0.1 × 10⁻⁷ cm·s⁻¹·cmH₂O⁻¹ in 12 vessels (P < 0.001). As a sham control, we measured L_p with only 1% BSA-Ringer perfusate in the third group of vessels. After ∼5-min perfusion, L_p was 1.2 ± 0.03 × 10⁻⁷ and was 1.1 ± 0.08 × 10⁻⁷ cm·s⁻¹·cmH₂O⁻¹ after 45-min perfusion in nine vessels. There was no significant change in L_p measured after 5 min and after 45-min 1% BSA-Ringer perfusion compared with the control, which was 1.1 ± 0.07 × 10⁻⁷ cm·s⁻¹·cmH₂O⁻¹ (P > 0.3). However, compared with the control, the initial tumor cell adhesion (after ∼6-min perfusion) induced a 1.5 ± 0.06-fold increase, and 45-min tumor cell perfusion and adhesion induced a 2.6 ± 0.1-fold increase in L_p (P < 0.001, Fig. 2B).

Fig. 2.

A: hydraulic conductivity (L_p) vs. time, measured under control, at the initial tumor cell adhesion and after ∼45-min tumor cell perfusion. B: comparison of L_p under control, at the initial tumor cell adhesion, and after ∼45-min tumor cell perfusion. The L_p was normalized by a mean value of that under the control condition. Values are means ± SE. *P < 0.001.

The normal rat serum colloid osmotic pressure is 20–27 cmH₂O (43), which is higher than 1% BSA mammalian Ringer (osmotic pressure ∼4 cmH₂O) used as the perfusate in our experiments. To test if a higher colloid osmotic pressure would affect the glycocalyx and the P, we measured the L_p from another group of vessels perfused with 5% BSA mammalian Ringer (osmotic pressure ∼20 cmH₂O). In six vessels, L_p = 1.2 ± 0.08 × 10⁻⁷ cm·s⁻¹·cmH₂O⁻¹, which was not significantly different from L_p (1.1 ± 0.07, n = 9) × 10⁻⁷ cm·s⁻¹·cmH₂O⁻¹ under 1% BSA perfusate (P = 0.34). No change in L_p indicates no change in the structural components of the microvessel wall and, consequently, no change in the glycocalyx for this range of osmotic pressures. Some previous studies found that L_p of frog mesenteric venular capillaries was sensitive to the wall shear stress (57), while other studies found there was no relationship between L_p and the wall shear stress (41). In the present study, before perfusing the tumor cells into a postcapillary venule, we checked that the blood flow in the vessel was normal, ranging from 0.8 to 1.1 mm/s in the mean velocity. For this range of blood flow velocity or wall shear stress, we did not find significant change in the L_p.

MDA-MB-231 adhesion increases microvascular solute P.

To find the most likely microvessel wall structural changes induced by tumor cell adhesion, we also measured the microvascular solute P to albumin (TRITC-BSA) under the same conditions as for measuring L_p. Figure 3A shows the effect of tumor cell adhesion on P. At tumor cell initial adhesion, i.e., after 6.7 ± 0.9 (SE) min perfusion, P was 1.9 ± 0.3 × 10⁻⁶ cm/s in eight vessels, and there was no significant change in another 10 min (P > 0.2). After 45-min perfusion and adhesion, P increased to 4.7 ± 0.5 × 10⁻⁶ cm/s in 10 vessels (P < 0.001). As a sham control, we measured P with only 1% BSA-Ringer perfusate in another group of vessels. The control P was 0.82 ± 0.02 × 10⁻⁶ cm/s. After ∼5-min perfusion, P was 0.86 ± 0.10 × 10⁻⁶ and was 0.88 ± 0.13 × 10⁻⁶ cm/s after 45-min perfusion, in 10 vessels. There was no significant change in P measured after 5 min and after 45 min of 1% BSA-Ringer perfusion (P > 0.6). However, compared with the control, the initial tumor cell adhesion (after ∼7-min perfusion) induced a 2.3-fold increase, and 45-min tumor cell perfusion and adhesion induced a 5.7-fold increase in P (P < 0.001).

Fig. 3.

A: apparent permeability to BSA (P^BSA) vs. time, measured under control, at the initial tumor cell adhesion, and after ∼45-min tumor cell perfusion. B: comparison of reflection coefficient to albumin (σ) under control, at the initial tumor cell adhesion, and after ∼45-min tumor cell perfusion. C: comparison of P_d^BSA under control, at the initial tumor cell adhesion, and after ∼45-min tumor cell perfusion. The P_d^BSA was normalized by a mean value of that under the control condition. Values are means ± SE. #P < 0.02. *P < 0.001.

MDA-MB-231 adhesion decreases microvascular σ to albumin.

Since the microvascular σ to albumin is a measure for selectivity of the microvessel wall. It also contributes to the solvent drag component (convection) of apparent P of albumin (measured P); it was thus determined under the same control and tumor cell adhesion conditions as for L_p and P measurements. At the initial tumor cell adhesion (after 8.2 ± 0.5 min), σ to albumin was 0.64 ± 0.07 (n = 5), and, after 45-min tumor cell perfusion and adhesion, σ to albumin was 0.42 ± 0.06 (n = 7). Compared with the control value of 0.92 ± 0.03 (n = 6), tumor adhesion decreased the σ to albumin, starting at initial adhesion (P < 0.02, Fig. 3B), indicating a reduced selectivity of the microvessel wall to solutes.

Employing the measured values for L_p, P, and σ to albumin, we used Eqs. 1–3 to calculate the P_d to albumin P_d, which was 0.78 ± 0.05 (n = 10) under control, 1.6 ± 0.2 (n = 8) at initial adhesion, and 3.9 ± 0.4 × 10⁻⁶ cm/s (n = 10) after 45-min tumor cell perfusion and adhesion. Compared with the control, the initial tumor cell adhesion induced a 2.0 ± 0.3-fold increase, and 45-min tumor cell perfusion and adhesion induced a 4.9 ± 0.5-fold increase in P_d (P < 0.001, Fig. 3C).

Supernatant from cultured MDA-MB-231 cells does not increase P.

To test whether increases in microvascular P were due to the secretion of the cultured MDA-MB-231 cells alone, we did the paired measurement of the microvascular solute P to albumin under control and under the treatment of supernatant collected from 60-min culture of 4 million/ml MDA-MB-231 cells, either labeled with the fluorescent cell tracker or not. The concentration of 4 million/ml was used in our tumor cell perfusion and adhesion experiments. Figure 4 shows that the supernatant without labeling insignificantly increased the P from a control of 8.1 ± 0.7 to 8.9 ± 0.6 × 10⁻⁷ cm/s (P > 0.4, n = 9) (the line with diamonds). It also shows that the P did not change over 60 min under the treatment of the supernatant. The line with the circles is the result for the supernatant with labeling, which insignificantly increased the P from a control of 8.4 ± 1.0 to 9.0 ± 0.7 × 10⁻⁷ cm/s (P > 0.5, n = 8). In parallel, a sham control was performed in another six vessels to confirm that there was no significant change in P due to the manipulation and other factors during the measurement (the line with triangles).

Fig. 4.

The line with the diamonds is for the paired measurement of P^BSA, measured under control with 1% BSA-Ringer perfusate and under the treatment of supernatant from the 4 million/ml nonfluorescently labeled MDA-MB-231 cell culture after 1-h incubation. The line with the circles is for the paired measurement of P^BSA under the treatment of supernatant from the 4 million/ml fluorescently labeled MDA-MB-231 cell culture after 1-h incubation. The line with the triangles is the sham control with 1% BSA-Ringer perfusate. Values are means ± SE.

Model predictions.

As shown above, MDA-MB-231 adhesion to the microvessel wall increased microvascular P to water (L_p) and solute (P) and decreased the microvessel wall selectivity (σ to albumin). To further investigate the underlying structural mechanisms by which the tumor cell adhesion modulates microvascular P, we adapted the mathematical model for the interendothelial transport developed by Fu and colleagues (22, 34). Figure 5A shows a three-dimensional (3D) model geometry for the interendothelial cleft in the wall of a rat mesenteric microvessel [revised from Fu et al. (22)]. There are junction strands with discontinuous breaks within the cleft (2) and a surface glycocalyx layer (1, 3, 49, 58) at the luminal entrance of the cleft. Under normal physiological conditions, electron microscopy study on rat mesenteric microvessels (2) revealed that, on average, the cleft width 2B = 18 nm, break width 2d = 315 nm, spacing between adjacent large breaks 2D = 3,590 nm, and cleft depth L = 411 nm. The thickness, L_f, of the molecular sieve part of the surface glycocalyx is from 100 to 400 nm, with an average of 250 nm (1, 3, 23, 49). The glycocalyx fiber radius a = 6 nm, and the gap spacing between fibers Δ = 8 nm (3). Our predictions for the P increase were based on these baseline parameters. Figure 5, B–E, shows our model predictions for increasing L_p and P to albumin by changing the structural components of the interendothelial cleft. If tumor cell adhesion degrades the surface glycocalyx (decreasing L_f), P to albumin would increase largely, while L_p would only have moderate increase (Fig. 5B); if tumor cell adhesion increases the gap (2B) between ECs (ECs contract or shrink), L_p would increase greatly, while P to albumin would have a moderate increase (Fig. 5C); if tumor cell adhesion increases the size (2d) (Fig. 5D) or number (decreasing 2D) (Fig. 5E) of junction breaks in the cleft, L_p would have a moderate increase, while P to albumin would only have a negligible increase.

Fig. 5.

A: three-dimensional view of the model geometry for the interendothelial cleft in the wall of a rat mesenteric microvessel [revised from Fu et al. (22) and Liu et al. (34)]. A junction strand with periodic breaks lies parallel to the luminal front of the cleft. An endothelial surface glycocalyx layer with thickness L_f is at the cleft entrance; L, the total depth of the cleft; 2D, the distance between two adjacent breaks in the junction strand; 2d, the width of the junction break; and 2B, the width of the cleft. B–E: model predictions for the effect of changing structural components of the interendothelial cleft on L_p (solid lines) and diffusive P (P_d) of albumin (dashed lines). B: decreasing the surface glycocalyx layer with thickness L_f. C: increasing the cleft width 2B. D: increasing the junction break width 2d. E: increasing the number of junction breaks (decreasing 2D). L_{p control}, P_control, L_{f control}, B_control, d_control, and D_control: control values of L_p, P, L_f, 2B, 2d, and 2D, respectively.

From the measured L_p and P data, we noticed that tumor cell adhesion induced a larger increase in P than in L_p. Out of all the structural changes predicted by the model for the interendothelial cleft shown in Fig. 5, B–E, this trend can only be accounted for by degrading the ESG (Fig. 5B). Therefore, in Fig. 6, we have compared our measured P data with the model predictions when decreasing the ESG L_f. The solid line is the model prediction for L_p, and the dashed line for P_d to albumin. To account for the measured ratio of L_p at the initial adhesion to the control value, 1.5 ± 0.06 (open square), and that of P_d at the initial adhesion to the control value, 2.0 ± 0.3 (open triangle), the predicted L_f/L_{f control} would be ∼0.45, suggesting that ∼55% of ESG is degraded. To account for the measured ratio of L_p after 45-min tumor cell perfusion and adhesion to the control value, 2.6 ± 0.1 (filled square), and that of P_d after 45-min tumor cell perfusion to the control value, 4.9 ± 0.5 (solid triangle), the predicted L_f/L_{f control} would be ∼0.1, suggesting that ∼90% ESG is degraded.

Fig. 6.

Comparison of the experimental results (symbols) with model predictions for L_p (solid line) and P_d of albumin (dashed line) when decreasing the surface glycocalyx layer with thickness L_f. Open square and triangle are for measured L_p and P_d^BSA at the initial tumor cell adhesion, respectively, whereas solid ones are for those after 45-min tumor cell perfusion and adhesion. Values are means ± SE.

MDA-MB-231 adhesion degrades microvessel ESG.

To confirm the above model prediction that tumor cell adhesion degrades the ESG, we quantified the microvessel ESG by immunolabeling HS, one of the most abundant GAGs in the ESG. Figure 7A1 shows the midplane view of the FITC-anti-HS-labeled ESG in a control vessel without tumor cell adhesion; Fig. 7A2 is the cross-sectional view for the vessel segment between dotted lines in Fig. 7A1. Correspondingly, Fig. 7, B1 and B2, demonstrates those of the FITC-anti-HS labeled ESG in a vessel at the initial tumor cell adhesion, and Fig. 7, C1 and C2, for those after 45-min tumor cell perfusion and adhesion. The green is the FITC-anti-HS labeled ESG and the red are the adherent tumor cells. Figure 7D compares the amount of FITC-anti-HS in a control vessel, a vessel with the initial tumor adhesion, and that with 45-min perfusion and adhesion. We can see from Fig. 7D that even the initial tumor cell adhesion degraded the ESG to 44 ± 8% of the control (P = 0.002). Forty-five-minute tumor cell perfusion and adhesion further decreased the ESG amount to 18 ± 3% of the control (P < 0.001).

Fig. 7.

Staining of FITC-anti-heparan sulfate (HS) on the luminal surface of a postcapillary venule under control: midplane view (A1) and cross-sectional view (A2) over a ∼10-μm-thick vessel segment (between dotted lines in A1). In addition to FITC-anti-HS (green), adherent cell Tracker Orange-labeled tumor cells (red) were shown in a postcapillary venule after initial adhesion: midplane view (B1) and cross-sectional view (B2) over a ∼10-μm-thick vessel segment (between dotted lines in B1); and in a postcapillary venule after ∼45-min tumor cell perfusion: midplane view (C1) and cross-sectional view (C2) over a ∼60-μm-thick vessel segment (between dotted lines in C1). D: comparison of the normalized intensity of FITC-anti-HS at the luminal surface of the vessels under control, after initial tumor cell adhesion, and after ∼45 min tumor cell perfusion. Values are means ± SE. #P < 0.01.

Orosomucoid decreases microvascular solute P and tumor cell adhesion.

Orosomucoid, a glycoprotein in the plasma, was suggested to be essential for stabilizing microvessel solute P by modulating the charge of the microvessel wall (16, 46, 59). To investigate the effect of orosomucoid on microvessel P and tumor cell adhesion, we measured the P to albumin and tumor cell adhesion under the treatment of 0.1 mg/ml orosomucoid. Figure 8A shows that 0.1 mg/ml orosomucoid significantly decreased P to albumin to 55 ± 9% of the control (P = 0.008) in the paired measurement on seven vessels. Figure 8B demonstrates that 30-min 0.1 mg/ml orosomucoid pretreatment also significantly retarded MDA-MB-231 adhesion to the microvessel wall by ∼46% during 60-min adhesion under flow (P < 0.02).

Fig. 8.

A: comparison of vessel P_d^BSA under control and after 30-min treatment of 0.1 mg/ml orosomucoid. *P < 0.01. B: comparison of adhesion of MDA-MB-231 tumor cells to the microvessel wall under control and after 30-min pretreatment of 0.1 mg/ml orosomucoid. *P < 0.02 starting at ∼4 min after tumor cell perfusion. Values are means ± SE.

Orosomucoid preserves microvessel ESG.

To further investigate the structural mechanisms by which orosomucoid decreased P to albumin and tumor cell adhesion to the microvessel wall, we quantified the ESG by immunolabeling the HS under control and under orosomucoid treatment. Figure 9, A and B, shows the midplane view of FITC-anti-HS in a control vessel and in a vessel pretreated with 0.1 mg/ml orosomucoid for 30 min. Because the depth of the light collection of our ×20 (NA = 0.75) objective lens is ∼100 μm (60), the fluorescence shown in Fig. 9, A and B, should come from the entire vessel for our vessels with diameter 35–50 μm. Figure 9C shows that orosomucoid did increase the amount of the ESG to 1.4 ± 0.1-fold that of the control (P = 0.014).

Fig. 9.

Staining of FITC-anti-HS at the luminal surface of postcapillary venules under control (A) and after 30-min treatment of 0.1 mg/ml orosomucoid (B). C: comparison of the normalized intensity of FITC-anti-HS at the luminal surface of the vessels under control and that after 30-min treatment of 0.1 mg/ml orosomucoid. Values are means ± SE. *P < 0.01.

MCF-10A adhesion to the microvessel wall.

To compare the adhesion of malignant and normal mammary epithelial cells to the microvessel wall and the effect of their adhesion on microvessel P, we also measured the adhesion of a nontumorigenic breast epithelial cell MCF-10A in our single-vessel perfusion experiments under the same condition as for MDA-MB-231. We found that the initial adhesion time for MCF-10A was delayed to ∼40 min, compared with ∼7 min for MDA-MB-231. In addition, the adhesion rate of MCF-10A was only one-fifth that of MDA-MB-231 (data not shown) in 60-min adhesion under flow. However, 45-min MCF-10A cell perfusion and adhesion significantly increased P to albumin from a control of 0.81 ± 0.07 (n = 10) to 1.6 ± 0.21 × 10⁻⁶ cm/s (n = 7, P = 0.02) (data not shown).

DISCUSSION

Circulating tumor cell adhesion to the microvascular wall is one of the critical steps in tumor metastasis (51). Understanding the mechanisms by which tumor cell adhesion to endothelium lining the vessel wall affects microvessel integrity is important in developing anti-metastatic strategies. Thus the aim of this study was to determine the effect of tumor cell adhesion on microvessel integrity in intact microvessels. Comparison of the measured P data and the model predictions (Fig. 6) shows that, at the initial tumor cell adhesion (one adherent cell in ∼100- to 145-μm vessel segment), ∼55% ESG is diminished, while a longer time adhesion (five adherent cells in ∼100- to 145-μm vessel segment) further diminishes the ESG by ∼90%. These numbers are consistent with the HS amount detected by the immunostaining method (Fig. 7). Figure 7D demonstrates that, compared with the control, the HS amount is only 44% at the initial tumor cell adhesion and decreases to 18% at a longer time adhesion. To further confirm the effect of ESG on L_p and P, we used 50 mU/ml heparinase III to pretreat the vessel for 1 h. After the pretreatment, L_p increased by 2.3 ± 0.15-fold (n = 7), and P_d to albumin by 4.6 ± 0.5-fold (n = 7) compared with their baselines (1% BSA-Ringer). These values were not significantly different from those measured after 45-min tumor cell adhesion (P > 0.15). We also measured the intensity of the FITC-anti-HS after 1-h 50 mU/ml heparinase III treatment, which was 17 ± 1% (n = 3) of the baseline. This was consistent with that observed after 45-min tumor cell adhesion.

Although the cause for the ESG degradation by the tumor cell adhesion has not been clarified, we postulate that both tumor cell secretions and tumor-endothelium interactions play a role in this process. As found in many prior studies, breast cancer cells, including MDA-MB-231, express VEGF to a high degree (33), while the microvascular endothelium has abundant VEGF receptors (39). VEGF was reported to increase microvascular P, partially by degrading the ESG (20, 21). Degradation of the ESG enables VEGF receptors at the endothelium (or ligands of cell adhesion molecules at tumor cells) to interact with the VEGF secreted by the tumor cells (or cell adhesion molecules at tumor cells), consequently increasing the tumor cell adhesion. Anti-VEGF and inhibition of a VEGF receptor (VEGFR2, kinase insert domain-containing receptor/fetal liver kinase-1) have been reported to almost completely abolish the tumor cell adhesion to the microvascular endothelium in vitro (18, 33) and in vivo (47). These studies also showed that antibodies blocking cell adhesion molecules at tumor cells or at endothelium can reduce tumor cell adhesion to a variety of levels. In addition to VEGF, cancer cells produce some special enzymes that are known to degrade the ESG, such as heparinase and hyaluronidase (55).

To examine the effect of tumor cell secretions on microvascular P, we used the supernatant from 1-h cell culture of 4 million/ml MDA-MB-231 and did not see a significant increase in the microvascular P (Fig. 4). The plausible reason for this is that the supernatant was too diluted to reach the threshold for increasing the vessel P. The local tumor cell secretions near adhesion sites should have a higher concentration to degrade the ESG and result in more tumor cell adhesion. This can be an explanation for more ESG degradation at longer time tumor adhesion. Another reason is that the supernatant in our experiments was from the tumor cell culture under static conditions instead of under flow, which may also affect tumor cell secretions. In addition to tumor cell secretions, tumor cell interactions with endothelium can be another reason for adhesion-induced P increase, since previous studies showed that a VEGFR2 inhibitor, SU1498, can abolish VEGF-induced increases in vascular P and tumor cell adhesion (18, 47), but it does not modulate the basal microvascular P and tumor cell adhesion.

Most recently, Yen et al. (58) reported a heterogeneous ESG coverage on the microvessels. They found that about one-half of the microvessel wall is covered with the significant ESG, and one-half is not, in the capillaries and postcapillary venules of rat mesentery and mouse cremaster muscles. The uncovered spacing is from 12 to 44 μm, with an average of 30 μm in the postcapillary venules of rat mesentery, the same type of vessels as used in this study. The initial tumor cell arrest may be due to this heterogeneous ESG coverage. The perfused tumor cells of size ∼14-μm diameter are initially arrested at the vessel wall segments, which lack the ESG. The secretions, such as VEGF from the adherent tumor cells, can increase the vessel P within a couple of minutes as reported before, by possibly opening up the gap between adjacent ECs and degradation of ESG (4, 20, 21). The adhesion between tumor cells and the ECs and that between tumor cells and the extracellular matrix proteins (18, 47) can anchor the initially arrested tumor cells, which produce more secretions to degrade the ESG and result in more tumor cell adhesion.

Our results and the above explanations suggest that ESG protection or the restoration of an already damaged ESG can be a promising anti-metastatic approach. Agents that can specifically increase the synthesis of ESG components, refurbish it, or selectively prevent its enzymatic degradation do not seem to be available at this time, but may be available in the future. Fortunately, our body creates its own defensive agent. For example, albumin, the most abundant plasma protein, was reported to maintain the molecular sieve of the microvessel wall by organizing and stabilizing the ESG components (1, 37). Orosomucoid is another plasma protein, which is needed for the maintenance of normal permselectivity of the capillary walls (50), because it can strengthen the ESG, as shown in Fig. 9, and thus decrease the microvascular P and reduce the tumor cell adhesion (Fig. 8). In addition to albumin and orosomucoid, angiopoietin-1, a 70-kDa glycoprotein, which is a ligand for the tyrasine kinase receptor Tie2, expressed by ECs throughout the vasculature, was reported to decrease microvascular P by increasing the ESG thickness in intact microvessels (45). Furthermore, pharmacological blockers of radical production may be useful to diminish the oxygen radical stress on the ESG. Examples include direct inhibition of the cytokine tumor necrosis factor-α, use of antithrombin III (lowering susceptibility to enzymatic attack), the application of hydrocortisone (inhibiting mast-cell degranulation), and avoidance of the liberation of natriuretic peptides (as in volume loading and heart surgery) (6).

Prior studies demonstrated that leukocyte adhesion, emigration, and increased protein leakage are closely related (29, 32). More recent studies, however, reported that leukocyte adhesion and migration are uncoupled from increases in microvascular P (28, 31). Zeng et al. (61) found that, although inflammatory mediator-induced increases in P may promote leukocyte adhesion, tumor necrosis factor-α-induced leukocyte adhesion does not increase microvascular P. Despite the morphological similarities of leukocyte and tumor cell adhesion, there are distinct differences between inflammatory cells and cancer cells due to the available surface adhesion molecules and intracellular signaling cascades (62). This may be one explanation for the coupled tumor cell adhesion, but uncoupled leukocyte adhesion, from the increases in microvascular P. However, leukocyte adhesion is indeed inhibited in venules perfused with HS and heparin, which is most likely due to the reconstitution of the ESG caused by the attachment of HS and heparin to the vessel wall (12). Protection of ESG by suppressing matrix metalloprotease activity was also reported to inhibit leukocyte-endothelial adhesion in postcapillary venules (40).

In summary, we have shown in this study that adhesion of human mammary carcinoma MDA-MB-231 increases microvascular P and reduces its selectivity via degradation of the ESG at the microvessel wall. We further examined this effect of tumor cell adhesion on the microvessel integrity by enhancing the ESG using a plasma protein, orosomucoid, which decreased microvascular P as well as tumor cell adhesion. Our results thus suggest a new anti-tumor metastatic therapy via ESG protection.

GRANTS

This work was supported by National Institutes of Health (National Cancer Institute) Grants CA153325-01 and CA137788-01 and the National Science Foundation Grant CBET-0754158. B. Cai was a research scholar from the Department of Human Anatomy and Histology, Southern Medical University, Guangzhou, China.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: B.C. and B.M.F. conception and design of research; B.C., J.F., M.Z., and L.Z. performed experiments; B.C. analyzed data; B.C. and B.M.F. interpreted results of experiments; B.C. and B.M.F. prepared figures; B.C. drafted manuscript; B.M.F. edited and revised manuscript; B.M.F. approved final version of manuscript.