1. We have investigated the interaction of plasma proteins with the endothelial cell using cationized ferritin as a marker of the cell surface glycocalyx. 2. Single microvessels of the frog mesentery were sequentially perfused using glass micropipettes with solutions containing cationized ferritin (CF, 6.7 mg ml-1) in 0.10 M-NaCl and then with either frog plasma or bovine serum albumin (BSA; 50 or 10 mg ml-1), or protein-free Ringer solution, before suffusion fixation in 2.5% glutaraldehyde. 3. A layer of CF, usually two to four molecules thick, was associated with the luminal endothelial cell surface. In vessels post-flushed with protein-free Ringer solution the CF layer was closely adherent to all regions of the luminal endothelium, including the plasma membrane, vesicle diaphragms, coated pits and the entrances to clefts. However, when plasma was present during fixation the CF layer was separated from the cell surface by up to 100 nm over all regions. In vessels post-flushed with BSA the CF layer was also separated from the membrane but the effect was less striking. 4. The association of cationized ferritin with the endothelial cell surface was assessed quantitatively using electron micrographs of transverse sections (approximately 50 nm thick) of the perfused vessels, and expressed in terms of the depth of the layer of CF associated with the endothelial cell surface, its separation from the plasma membrane of the luminal endothelium, and the concentration of CF in the layer. The mean (+/- S.D.) separation in the presence of plasma, 32.3 +/- 10.5 nm (n = 12), was significantly greater (P < 0.01) than that with either protein-free Ringer solution, 3.0 +/- 1.4 nm (n = 9), or BSA in Ringer solution, 8.3 +/- 3.0 nm (n = 8). The separation seen with BSA in Ringer was also significantly greater than that measured with a final Ringer solution perfusion (P < 0.01). The effects of 10 and 50 mg ml-1 BSA were not different from one another. The total glycocalyx thickness, defined as the sum of the separation layer and depth of CF layer, with plasma present, 56.2 +/- 13.7 nm, was twice the value seen with Ringer solution, 28.0 +/- 9.1 nm (P < 0.01), while the total thickness with BSA, 30.9 +/- 5.4 nm, was not different from the Ringer solution value. 5. The concentration of ferritin within the CF binding layer was not significantly affected by the protein content of the final perfusate solution and ranged between mean values of 14.4 x 10(4) (Ringer solution reperfusion) and 17.1 x 10(4) (50 mg ml-1 BSA reperfusion) CF molecules mu-m-3. The depth of the CF layer at the endothelial cell surface also was unaffected by the composition of the final perfusate with values of 24.8 +/- 8.8 nm, 22.6 +/- 3.0 nm, and 23.9 +/- 7.0 nm for reperfusion with Ringer solution (n = 8), BSA in Ringer solution (n = 9) and plasma (n = 12), respectively. 6. We conclude that addition of plasma proteins results in a separation of the CF binding layer from the plasma membrane of the endothelial cell surface. Our interpretation of this is that the increased separation of the CF binding layer is a result of a change in the orientation of surface glycoproteins to which CF binds. Our observation that whole plasma is more effective than serum albumin alone is consistent with previous observations that plasma is more effective at reducing microvascular permeability than its constituent proteins alone.
