We have combined an ex vivo whole blood and a Transwell insert system to evaluate the biological relevance of antibiotic-induced lipopolysaccharide (LPS) release, and have shown that this event results in a brisk release of TNFα in whole blood (Arditi M., Kabat W., Yogev R. Antibiotic-induced bacterial killing stimulates tumor necrosis factor-α release in whole blood. J Infect Dis 1993; 167: 240-244). We now investigated the activation of vascular endothelial cells in response to antibiotic-induced bacterial killing and free LPS release. We have previously shown that the endothelial cells respond to LPS in a serum- or soluble CD14-dependent manner. We used LPS-induced IL-6 release from human umbilical vein endothelial cells (HUVEC) and human blood cells as two readout systems. Escherichia coli O111:B4 (106 CFU/ml) was placed into the upper chamber of a Transwell filter insert (0.1 μm) and incubated for 6 h with or without ceftriaxone (40x MIC) in 24-well plates containing HUVEC monolayers in the presence of 5% serum, or diluted human blood cells in the lower chamber. Filter inserts were removed at 6 h and plates were incubated for an additional 12 h and supernatants were collected for IL-6 measurement. Bacterial kill curves were generated. Antibiotic-induced killing of E. coli resulted in activation of human endothelial and blood cells as assessed by IL-6 release. Addition of 5 μg/ml of bactericidal permeability factor (BPI) to the media containing ceftriaxone did not influence the kill curve but resulted in over 96% decrease in IL-6 release from both endothelial and blood cells, suggesting that free LPS was the predominant stimulus for IL-6 release. Furthermore, when the endothelial cell experiments were conducted in serum-free conditions, antibiotic-induced bacterial killing did not induce IL-6 release from HUVEC, and addition of recombinant human soluble CD14 into the media reconstituted the serum effect, consistent with the concept that serum or soluble CD14 is absolutely required for LPS-induced endothelial cell responses, including LPS-induced IL-6 release. IL-10 was able to significantly inhibit LPS-induced IL-6 production by human blood cells but had no effect on LPS-mediated IL-6 release from HUVEC. These data suggest that antibiotic-induced free LPS release leads to activation of both human vascular endothelial cells and blood cells, which can be readily inhibited by the addition of BPI. The results support the hypothesis that free endotoxin released from E. coli following bacteriolysis is the major contributor to the enhancement of the inflammation associated with the initiation of antibiotic therapy. The dichotomy of IL-10 effects on LPS-induced cytokine production in blood mononuclear cells versus endothelial cells may play a role in the pathophysiology of TH-2-directed responses.
