Enhanced responsiveness of endothelium in the growing/motile state to tumor necrosis factor/cachectin [published erratum appears in J Exp Med 1989 Nov 1;170(5):1793]

Some in vivo observations have suggested that growing or perturbed endothelium, such as that which occurs during angiogenesis, is more sensitive to the action of cytokines (TNF/cachectin, TNF, or IL-1) than normal quiescent endothelial cells. This led us to examine the responsiveness of endothelium to TNF as a function of the growth/motile state of the cell. TNF-induced modulation of endothelial cell surface coagulant function was half-maximal at a concentration of approximately 0.1 nM in subconfluent cultures, whereas 1-2 nM was required for the same effect in postconfluent cultures. Perturbation of endothelial cell shape/cytoskeleton was similarly more sensitive to TNF in subconfluent cultures. Consistent with these results, radioligand binding studies demonstrated high affinity TNF binding sites, Kd approximately 0.1 nM on subconfluent cultures, whereas only lower affinity sites (Kd approximately 1.8 nM) were detected on postconfluent cultures. The mechanisms underlying this change in the affinity of endothelium for TNF were studied in four settings. Crosslinking experiments with 125I-TNF and endothelium showed additional bands corresponding to Mr approximately 66,000 and approximately 84,000 with subconfluent cultures that were not observed with postconfluent cultures. Experiments with X-irradiated endothelium, whose growth but not motility was blocked, indicated that proliferation was not required for induction of high affinity TNF sites. Postconfluent endothelium, triggered to enter the proliferative cycle by microbutuble poisons, expressed high affinity TNF binding sites together with changes in cell shape/cytoskeleton well before their entry into S phase. Using wounded postconfluent monolayers, cells that migrated into the wound and those close to the wound edge displayed enhanced TNF binding and modulation of coagulant properties. These results suggest a model for targetting TNF action within the vasculature; regulation of high affinity endothelial cell binding sites can direct TNF to activated cells in particular parts of the vascular tree.     

Endothelial receptor-mediated binding of glucose-modified albumin is associated with increased monolayer permeability and modulation of cell surface coagulant properties

Advanced glycosylation end products (AGE) of proteins accumulate in the vasculature with diabetes and aging, and are thought to be associated with vascular complications. This led us to examine the interaction of AGE-BSA as a prototype of this class of nonenzymatically glycosylated proteins subjected to further processing, with endothelium. Incubation of 125I-AGE-BSA with cultured bovine endothelium resulted in time-dependent, saturable binding that was half-maximal at a concentration of approximately 100 nM. Although unlabeled normal BSA was not a competitor, unlabeled AGE-BSA was an effective competitor of 125I-AGE-BSA-endothelial cell interaction. In addition, AGE modification of two alternative proteins, hemoglobin and ribonuclease, rendered them inhibitors of 125I-AGE-BSA binding to endothelium, although the native, unmodified forms of these proteins were not. At 37 degrees C, binding of 125I-AGE-BSA or gold-labeled AGE-BSA was followed by internalization and subsequent segregation either to a lysosomal compartment or to the endothelial-derived matrix after transcytosis. Exposure of endothelium to AGE-BSA led to perturbation of two important endothelial cell homeostatic properties, coagulant and barrier function. AGE-BSA downregulated the anticoagulant endothelial cofactor thrombomodulin, and induced synthesis and cell surface expression of the procoagulant cofactor tissue factor over the same range of concentrations that resulted in occupancy of cell surface AGE-BSA binding sites. In addition, AGE-BSA increased endothelial permeability, resulting in accelerated passage of an inert macromolecular tracer, [3H]inulin, across the monolayer. These results indicate that AGE derivatives of proteins, potentially important constituents of pathologic vascular tissue, bind to specific sites on the endothelial cell surface and modulate central endothelial cell functions. The interaction of AGE-modified proteins with endothelium may play an important role in the early stages of increased vascular permeability, as well as vessel wall-related abnormalities of the coagulation system, characteristic of diabetes and aging.     

Modulation of endothelial cell hemostatic properties by tumor necrosis factor

Tumor necrosis factor/cachectin (TNF) is a mediator of the septic state, which involves diffuse abnormalities of coagulation throughout the vasculature. Since previous studies have shown that endothelial cells can play an active role in coagulation, we wished to determine whether TNF could modulate endothelial cell hemostatic properties. Incubation of purified recombinant TNF with cultured endothelial cells resulted in a time- and dose-dependent acquisition of tissue factor procoagulant activity. Concomitant with enhanced procoagulant activity, TNF also suppressed endothelial cell cofactor activity for the anticoagulant protein C pathway; both thrombin-mediated protein C activation and formation of functional activated protein C-protein S complex on the cell surface were considerably attenuated. Comparable concentrations of TNF (half-maximal affect at approximately 50 pM) and incubation times (half-maximal affect by 4 h after addition to cultures) were required for each of these changes in endothelial cell coagulant properties. This unidirectional shift in cell surface hemostatic properties favoring promotion of clot formation indicates that, in addition to leukocyte procoagulants, endothelium can potentially be instrumental in the pathogenesis of the thrombotic state associated with inflammatory and malignant disorders.     

Oxygen tension regulates the nitric oxide pathway. Physiological role in penile erection.

Relaxation of the trabecular smooth muscle of the corpus cavernosum (the erectile tissue) of the penis is mediated by nitric oxide released by the nerves and endothelium. We have investigated the physiological role of oxygen tension in the regulation of trabecular smooth muscle tone. In human subjects, measurement of intracavernosal PO2 in blood drawn from corpus cavernosum in the flaccid state was comparable to that of venous blood (25-43 mmHg). Vasodilatation of the resistance arteries and trabecular smooth muscle relaxation by intracavernosal injection of papaverine and phentolamine caused oxygen tension to rise rapidly to arterial levels (PO2 approximately 100 mmHg). Isolated human and rabbit corpus cavernosum tissue strips in organ baths, exposed to arterial-like PO2 relaxed to the endothelium-dependent dilator acetylcholine and to electrical stimulation of the autonomic dilator nerves. These nitric oxide-mediated responses were progressively inhibited as a function of decreasing PO2 to levels measured in the flaccid penis. Reverting to normoxic conditions readily restored endothelium-dependent and neurogenic relaxation. Relaxation to exogenous nitric oxide was not impaired in low PO2. In rabbit corpus cavernosum, low PO2 reduced basal levels of cGMP and prevented cGMP accumulation induced by stimulation of dilator nerves. Furthermore, low PO2 inhibited nitric oxide synthase activity in corpus cavernosum cytosol. It is concluded that physiological concentrations of oxygen modulate penile erection by regulating nitric oxide synthesis in corpus cavernosum tissue.     

Acceleration of the thrombin inactivation of single chain urokinase-type plasminogen activator (pro-urokinase) by thrombomodulin.

The in vitro effects of thrombomodulin on the inactivation of single chain urokinase-type plasminogen activator (scu-PA) by thrombin were investigated by incubating scu-PA with varying concentrations of human thrombin, in both the absence and presence of soluble rabbit thrombomodulin. 50% inactivation of scu-PA occurred in 45 min at 160 ng/ml thrombin in the absence of thrombomodulin and at 4.6 ng/ml thrombin in the presence of thrombomodulin. No difference was found in either the absence or the presence of thrombomodulin between the inactivation rates of high molecular weight scu-PA, and a low molecular weight scu-PA which lacked the growth factor and kringle domains. Enzyme kinetic experiments with varying concentrations of scu-PA showed that thrombomodulin decreased the Km of thrombin for scu-PA from 7.8 to 0.43 microM and increased the kcat from 0.30 to 1.2 s-1, corresponding to a 70-fold increase in the second-order rate constant kcat/Km. SDS-polyacrylamide gel electrophoresis showed that scu-PA was cleaved into two chains upon inactivation by thrombin, and confirmed the acceleration effect of thrombomodulin on inactivation of scu-PA. Thrombomodulin thus not only has anticoagulant properties but is also antifibrinolytic. The acceleration may imply a new mechanism for the regulation of local plasminogen activator activity on the cell surface.     

Endothelial function and hemostasis

Summary The vascular endothelium influences not only the three classically interacting components of hemostasis: the vessel, the blood platelets and the clotting and fibrinolytic systems of plasma, but also the natural sequelae: inflammation and tissue repair. Two principal modes of endothelial behaviour may be differentiated, best defined as an anti- and a prothrombotic state. Under physiological conditions endothelium mediates vascular dilatation (formation of NO, PGI₂, adenosine, hyperpolarising factor), prevents platelet adhesion and activation (production of adenosine, NO and PGI₂, removal of ADP), blocks thrombin formation (tissue factor pathway inhibitor, activation of protein C via thrombomodulin, activation of antithrombin III) and mitigates fibrin deposition (t- and scu-plasminogen activator production). Adhesion and transmigration of inflammatory leukocytes are attenuated, e.g. by NO and IL-10, and oxygen radicals are efficiently scavenged (urate, NO, glutathione, SOD). When the endothelium is physically disrupted or functionally perturbed by postischemic reperfusion, acute and chronic inflammation, atherosclerosis, diabetes and chronic arterial hypertension, then completely opposing actions pertain. This prothrombotic, proinflammatory state is characterised by vasoconstriction, platelet and leukocyte activation and adhesion (externalisation, expression and upregulation of von Willebrand factor, platelet activating factor, P-selectin, ?ICAM-1, IL-8, MCP-1, TNFα, etc.), promotion of thrombin formation, coagulation and fibrin deposition at the vascular wall (expression of tissue factor, PAI-1, phosphatidyl serine, etc.) and, in platelet-leukocyte coaggregates, additional inflammatory interactions via attachment of platelet CD40-ligand to endothelial, monocyte and B-cell CD40. Since thrombin formation and inflammatory stimulation set the stage for later tissue repair, complete abolition of such endothelial responses cannot be the goal of clinical interventions aimed at limiting procoagulatory, prothrombotic actions of a dysfunctional vascular endothelium. Received: 15 September 1999/Accepted: 16 December 1999     

Synthesis and release of interleukin 1 by reoxygenated human mononuclear phagocytes.

To examine the possible involvement of cytokines in reperfusion injury, we have studied production of IL-1 by human vascular cells, including smooth muscle and mononuclear phagocytes. Exposure of cells to hypoxia (pO2 approximately 14 torr) followed by reoxygenation led to significant release of IL-1 only from the mononuclear phagocytes. Elaboration of IL-1 was dependent on the oxygen tension and duration of hypoxia (optimal at lower pO2s, approximately 14-20 torr, and after 9 h), as well as the time in reoxygenation (maximal IL-1 release at 6-9 h). Although a period of hypoxia was necessary for subsequent IL-1 production during reoxygenation of either peripheral blood monocytes or cultured monocyte-derived macrophages, no IL-1 release occurred during the hypoxic exposure. IL-1 released during reoxygenation was newly synthesized, and its production was triggered by the generation of oxygen free radicals, as it could be blocked by the addition of either allopurinol or free radical scavengers to cultures and could be stimulated in part by low concentrations of hydrogen peroxide or xanthine/xanthine oxidase. The potential pathophysiological effects of IL-1-containing supernatants from reoxygenated macrophages was shown by their induction of endothelial tissue factor and enhancement of endothelial adhesiveness for neutrophils, both of which could be blocked by anti-IL-1 antibody. The relevance of IL-1 to hypoxia/reoxygenation in vivo was suggested by the presence of circulating nanogram amounts of this cytokine in the plasma of mice during the reoxygenation period following a hypoxia.     

Tumor necrosis factor/cachectin interacts with endothelial cell receptors to induce release of interleukin 1

Tumor necrosis factor/cachectin (TNF) has been implicated as a mediator of the host response in sepsis and neoplasia. Recent work has shown that TNF can modulate endothelial cell hemostatic properties, suggesting that endothelium is a target tissue for TNF. This led us to examine whether endothelial cells have specific binding sites for TNF and augment the biological response to TNF by elaborating the inflammatory mediator, IL-1. Incubation of 125I-recombinant human TNF with confluent, cultured human umbilical vein endothelial cells resulted in time-dependent, reversible, and saturable binding. Binding was half-maximal at a TNF concentration of 105 +/- 40 pM, and at saturation 1,500 molecules were bound per cell. Heat-treated TNF, which is biologically inactive, did not bind to endothelium. In addition to surface binding, TNF induced the elaboration of IL-1 activity by endothelial cells in a time-dependent manner. Generation of IL-1 activity required protein synthesis and was half-maximal at a TNF concentration of 50 +/- 20 pM. IL-1 activity from TNF-treated endothelium could be adsorbed by an immobilized antibody to IL-1. Heat-treated TNF was ineffective in eliciting endothelial cell IL-1. These data indicate that TNF can bind specifically to endothelium and initiate a cascade of inflammatory and coagulant events on the vessel surface potentially central to the host response to neoplasia and sepsis.     

Tumor necrosis factor/cachectin increases permeability of endothelial cell monolayers by a mechanism involving regulatory G proteins

Endothelium is an important target of tumor necrosis factor/cachectin (TNF), a central mediator of the host response in endotoxemia and Gram-negative sepsis. In this report, TNF is shown to increase the permeability of endothelial cell monolayers to macromolecules and lower molecular weight solutes by a mechanism involving a pertussis toxin-sensitive regulatory G protein. Within 1-3 h of exposure to TNF (5 nM), changes in cell shape/cytoskeleton occurred that led to disruption of monolayer continuity with the formation of intercellular gaps. Correlated with these structural changes was an increase in endothelial permeability to macromolecular and lower molecular weight tracers; time-dependent, reversible increases in passage of these tracers, evident by 1-3 h, were observed after addition of TNF to cultures. Perturbation of barrier function by TNF also depended on the dose of TNF added being half-maximal by approximately 0.4 nM. Only a brief exposure (15 min) of TNF to endothelium was required to induce an increase in permeability, and this was not prevented by the presence of cycloheximide or actinomycin D. Preincubation of monolayers with pertussis toxin blocked in parallel TNF-induced increased passage of solutes and cell shape/cytoskeletal perturbation, indicating the close correlation between these changes in endothelial cell function. In contrast, pertussis toxin did not alter TNF-induced modulation of two endothelial cell coagulant properties. These data provide evidence for two intracellular pathways of TNF action that are distinguishable by pertussis toxin and provide insight into a mechanism underlying loss of solute from the intravascular space mediated by TNF: alteration in endothelial cell barrier function.     

Hypoxic ventilatory response in subjects with normal and high oxygen affinity hemoglobins.

It has still not been shown unequivocally whether a decrement of arterial oxygen content or tension governs the ventilatory response to hypoxia. In an attempt to discriminate between the two possibilities, we have measured the ventilatory response to isocapnic progressive hypoxia in two healthy children with a high oxygen affinity hemoglobin (Hb Andrew-Minneapolis) and in their age- and sex-matched normal siblings. Hypoxic ventilatory response was identical in all subjects, there being no difference in minute ventilation at PAo2 = 40 mm Hg or in k (decrement of PO2 required to increase ventilation by a factor of 2.718). In contrast, at PAo2 = 40 mm Hg, hemoglobin oxygen saturation decreased markedly in controls but only slightly in high affinity subjects. Furthermore the increase in heart rate at PAo2 = 40 mm Hg was significantly less in high affinity subjects, suggesting a concomitant difference in oxygen delivery. Thus, with identical decrements in PAo2 but widely divergent changes in arterial oxygen content and oxygen delivery, controls and high affinity subjects showed virtually identical ventilatory response to hypoxia. We conclude that decrements of oxygen tension are the major stimulus for hypoxic ventilatory response.     

Effects of thrombomodulin and coagulation Factor Va-light chain on protein C activation in vitro.

Protein C activation by thrombin is significantly accelerated by the endothelial cell surface protein thrombomodulin, Factor Va, or its light chain. In this study we have compared the activation of protein C in the presence of either cofactor and examined the possibility that thrombomodulin and Factor Va-light chain act together to regulate protein C activation by thrombin. At all concentrations of protein C used, thrombomodulin was 20 times more efficient than Factor Va-light chain in accelerating protein C activation by thrombin. Protein C treated with chymotrypsin to remove the amino-terminal 41 amino acids that contain the gamma-carboxyglutamyl residues was activated by the thrombin-thrombomodulin complex at an identical rate to native protein C, whereas the modified protein C was activated by Factor Va-light chain and thrombin at only 5% of the rate obtained by using native protein C. Increasing concentrations of Factor Va-light chain, greater than or equal to 30 nM, inhibited thrombin-thrombomodulin catalyzed protein C activation with complete inhibition observed at 90 nM Factor Va-light chain. On the other hand, increasing thrombomodulin concentrations did not inhibit protein C activation by Factor Va-light chain and thrombin. These reactions in solution mimic, in part, those obtained on endothelial cells where protein C lacking the gamma-carboxyglutamyl domain is activated poorly and Factor Va-light chain at concentrations greater than 50 nM inhibited the activation of native protein C. The results of this study suggest that thrombomodulin and Factor Va-light chain may act in concert to regulate protein C activation by thrombin.     

Inhibition of thrombomodulin surface expression and protein C activation by the thrombogenic agent homocysteine.

Elevated levels of plasma homocysteine are associated with both venous and arterial thrombosis. Homocysteine inhibits the function of thrombomodulin, an anticoagulant glycoprotein on the endothelial surface that serves as a cofactor for the activation of protein C by thrombin. The effects of homocysteine on thrombomodulin expression and protein C activation were investigated in cultured human umbilical vein endothelial cells and CV-1(18A) cells that express recombinant human thrombomodulin. Addition of 5 mM homocysteine to endothelial cells produced slight increases in thrombomodulin mRNA and thrombomodulin synthesis without affecting cell viability. In both cell types, thrombomodulin synthesized in the presence of homocysteine remained sensitive to digestion with endoglycosidase H and failed to appear on the cell surface, suggesting impaired transit along the secretory pathway. In a cell-free protein C activation assay, homocysteine irreversibly inactivated both thrombomodulin and protein C in a process that required free thiol groups and was inhibited by the oxidizing agents diamide or N-ethylmaleimide. By inhibiting both thrombomodulin surface expression and protein C activation, homocysteine may contribute to the development of thrombosis in patients with cystathionine beta-synthase deficiency.     

Inhibition of angiotensin converting enzyme activity in cultured endothelial cells by hypoxia.

Endothelial cells in tissue culture degrade bradykinin and convert angiotensin I to angiotensin II. These are both functions of a single dipeptidyl hydrolase, angiotensin converting enzyme. Monolayer cultures were prepared from human, rabbit, pig, and calf vessels. Angiotensin converting enzyme activity was assessed by adding either bradykinin or angiotensin I to the cells in culture flasks, and measuring residual peptide over time by radioimmunoassay. Peptide degradation was inhibited by the specific converting enzyme inhibitor, SQ 20881. The flasks were equilibrated with varying hypoxic gas mixtures: hypoxia rapidly (less than 2 min) decreased enzyme activity and room air restored it as rapidly. The extent to which activity was reduced was a direct function of PO2 (r = 0.93, P less than 0.001), and there was no enzyme activity below a PO2 of 30 mm Hg. Four preparations were studied with respect to decrease in enzyme activity by hypoxia: (a) intact cells in monolayer, (b) sonicated cells, (c) sonicated cells from which converting enzyme was partially dissolved by a detergent, and (d) purified converting enzyme. Hypoxia had progressively less of an inhibiting effect on the enzyme activity of the preparations as the degree of cell integrity decreased. Hypoxia inhibits angiotensin converting enzyme activity in cultured endothelial cells, but the effect of hypoxia is not on the enzyme per se, but appears to be a unique characteristic of the endothelial cell.     

Interleukin 1 induces endothelial cell synthesis of plasminogen activator inhibitor

Human endothelial cells activated with IL-1 express a surface membrane-oriented procoagulant generating system characterized by increased tissue factor synthesis and decreased thrombomodulin activity. We now report that IL-1 also stimulates endothelial cell synthesis of plasminogen activator inhibitor. This array of IL-1-induced activities shifts the balance at the endothelial cell surface to a prothrombotic influence and may reflect an early response of the blood vessel wall to injury.     

Thrombomodulin expression by human keratinocytes. Induction of cofactor activity during epidermal differentiation.

Thrombomodulin is an endothelial cell surface glycoprotein that inhibits the procoagulant activities of thrombin and accelerates activation of the anticoagulant protein C. Because protein C deficiency is associated with cutaneous thrombosis, we investigated the expression of thrombomodulin in human skin. Thrombomodulin was detected by immunohistochemical staining both in dermal endothelial cells and in epidermal keratinocytes. Within the epidermis, thrombomodulin staining was limited to keratinocytes of the spinous layer, suggesting that thrombomodulin is induced when basal keratinocytes begin to terminally differentiate. Thrombomodulin expression also correlated with squamous differentiation in epidermal malignancies; little or no thrombomodulin staining was seen in five basal cell carcinomas, whereas strong thrombomodulin staining was observed in each of five squamous cell carcinomas. Human foreskin keratinocytes cultured in medium containing 0.07 mM calcium chloride synthesized functional thrombomodulin with cofactor activity comparable to thrombomodulin in human umbilical vein endothelial cells. Stimulation of keratinocyte differentiation with 1.4 mM calcium chloride for 48 h produced 3.5-, 3.2-, and 5.6-fold increases in thrombomodulin cofactor activity, antigen, and mRNA, respectively. These observations suggest that thrombin is regulated by keratinocyte thrombomodulin at sites of cutaneous injury, and indicate a potential role for thrombomodulin in epidermal differentiation.     

Effect of thrombin on the fibrinolytic activity of cultured bovine endothelial cells.

The vascular endothelium is a rich source of plasminogen activator (PA) and thus of blood vessel-associated fibrinolytic activity. Cultured bovine aortic endothelial cells were employed to determine if components of the coagulation system interact with the endothelium to modify expression of this activity. The addition of thrombin to these cultures led to a rapid decline in intracellular PA activity, with as little as 3 ng/ml, or 0.1 nM thrombin causing a 50% decrease within 30 min. Thrombin inactivated with diisopropylflurophosphate or hirudin did not elicit the response. Although control cultures secreted high levels of PA, no PA activity could be detected in the media surrounding the thrombin-treated cells. This loss of activity did not appear to result from direct inactivation of PA by thrombin. These observations indicate that the fibrinolytic potential of cultured endothelial cells is rapidly suppressed by trace amounts of thrombin. The generation of thrombin at sites of vascular injury may have a similar effect on the endothelium.     

Recombinant adeno-associated virus vector-transduced vascular endothelial cells express the thrombomodulin transgene under the regulation of enhanced plasminogen activator inhibitor-1 promoter

We were able to facilitate plasminogen activator inhibitor 1 (PAI-1) promoter activity approximately by 14-fold using an enhancer element. This enhanced PAI-1 promoter has a strong basal activity, comparable to CAG promoter activity, and has a response similar to the PAI-1 promoter with respect to TGFbeta 1 and TNFalpha stimulation. The characteristics of the enhanced PAI-1 promoter are thought to be suited to timely and tissue-specific expression of anticoagulant molecules in the vascular cells. Thus, we developed recombinant adeno-associated virus (rAAV) vectors using the enhanced PAI-1 promoter and were successful in transducing vascular endothelial cells to express the thrombomodulin transgene under the regulation of the enhanced PAI-1 promoter in vitro. Thromobomodulin transgene expression driven by the enhanced PAI-1 promoter in rAAV vector-transduced cultured endothelial cells was between 600- and 1000-fold higher than constitutive thrombomodulin gene expression in cultured human umbilical vein endothelial cells and was up-regulated by TGFbeta1 and TNFalpha stimulation which may down-regulate endogenous thrombomodulin gene expression in endothelial cells. The brain vascular endothelial cells of Mongolian gerbils could also be transduced by the same rAAV vector in vivo. Transduction of endothelial cells by rAAV vectors to express enhanced PAI-1 promoter-driven transgenes may be a useful gene therapy approach for vascular diseases.     

Utilization of arachidonic and linoleic acids by cultured human endothelial cells.

When cultured human umbilical vein endothelial cells are supplemented with linoleic acid, the arachidonic acid content of the cellular phospholipids is reduced approximately 35%. Most of the fatty acid compositional change occurs during the first 24 h. One factor responsible for this effect is the inability of the endothelial cells to convert appreciable amounts of linoleic to arachidonic acid, due to a fatty acid delta 6-desaturase deficiency. By contrast, these endothelial cultures contain delta 5- and delta 9-desaturase activity and are able to elongate long-chain polyunsaturated fatty acids. The other factor that contributes to the decrease in arachidonic acid is that high concentrations of linoleic acid reduce the incorporation of arachidonate into cellular phospholipids. Stearic acid, a long-chain saturate, does not produce any reduction, whereas eicosatrienoic acid is an even more effective inhibitor than linoleic acid. In spite of the fact that high concentrations of these polyunsaturates produced inhibition, the endothelial cells were found to efficiently incorporate exogenous arachidonic acid into cellular phospholipids and triglycerides. This may serve to compensate for the inability of these cells to synthesize arachidonic acid from linoleic acid. These findings suggest that the endothelium obtains arachidonic acid from an extracellular source, that this cannot be provided in the form of linoleic acid and, in fact, that high concentrations of linoleic acid actually may interfere with the ability of the endothelium to maintain an adequate supply of intracellular arachidonic acid.     

Estimation of changes in alveolar-arterial oxygen gradient induced by hypoxia

The alveolar-arterial oxygen tension difference provides a useful clinical indication of ventilation-blood flow mismatching in the lungs. In some clinical situations involving alveolar hypoxia (e.g., patients with chronic obstructive lung disease flying in commercial aircraft or normal humans at high altitudes) it would be useful to know this tension difference to predict the likely arterial PO2 under such potentially stressful conditions. Such estimates would require multiple arterial punctures performed under a variety of trying circumstances, conditions usually far distant from a suitable analytic facility. Consequently, we induced controlled hypoxia in 23 healthy humans and calculated changes in the alveolar-arterial oxygen tension difference during the hypoxic challenge test. We plotted this difference as a function of the alveolar oxygen tension over a range from 35 to 110 mm Hg. In addition to a series of control studies in which multiple arterial blood samples were obtained, we calculated arterial PO2 by converting the arterial oxyhemoglobin saturation (measured with an ear oximeter) into partial pressure of oxygen. During hypoxic procedures in which levels of oxygenation fell on the steep section of the oxyhemoglobin dissociation curve, fixing PCO2 at constant predetermined levels allowed accurate predictions of arterial PO2. We were able to demonstrate that the alveolar-arterial oxygen tension difference narrowed with decreasing alveolar oxygen tension, and that measurement with an ear oximeter provided data that allowed a reasonable estimate of the tension difference during hypoxic conditions.