Increased circulating lipid peroxides in severe preeclampsia activate NF-kappaB and upregulate ICAM-1 in vascular endothelial cells.

Preeclampsia is a systemic disease of pregnancy characterized by maternal hypertension, proteinuria, and edema. These clinical pathological findings may be attributed to abnormalities in vascular endothelial activation secondary to increased oxidative stress. To test the hypothesis that increased circulating lipid peroxides in preeclamptic women activate vascular endothelial cells, we determined NF-kappaB transcriptional activity and ICAM-1 expression in human umbilical vein endothelial cells (HUVEC) cultured with plasma from women with severe preeclampsia (preeclamptic plasma, N = 12) or plasma from normal pregnancies (normal plasma, N = 12). Preeclamptic women had increased circulating lipid peroxides compared with normal pregnant women, as demonstrated by a 4.5-fold higher concentration of plasma malondialdehyde (PkB luciferase reporter construct transfected into HUVEC, preeclamptic plasma was found to up-regulate HUVEC NF-kappaB activity by 2.5-fold when compared with normal plasma (PkB activation in response to preeclamptic-plasma by 77% (PkB activation and ICAM-1 expression on HUVEC, which can be inhibited by vitamin E and N-acetyl-cysteine.

Sphingosine induces phospholipase D and mitogen activated protein kinase in vascular smooth muscle cells.

The enzymes phospholipase D and diacylglycerol kinase generate phosphatidic acid which is considered to be a mitogen. Here we report that sphingosine produced a significant amount of phosphatidic acid in vascular smooth muscle cells from the rat aorta. The diacylglycerol kinase inhibitor R59 949 partially depressed sphingosine induced phosphatidic acid formation, suggesting that activation of phospholipase C and diacylglycerol kinase can not account for the bulk of phosphatidic acid produced and that additional pathways such as phospholipase D may contribute to this. Further, we have shown that phosphatidylethanol was produced by sphingosine when vascular smooth muscle cells were stimulated in the presence of ethanol. Finally, as previously shown for other cell types, sphingosine stimulated mitogen-activated protein kinase in vascular smooth muscle cells.

Heat shock provides delayed protection against oxidative injury in cultured human umbilical vein endothelial cells.

During both mild and severe ischemia, vascular endothelial cells lining large and small vessels of the ischemic organ are exposed to oxygen-derived free radicals resulting in oxidative damage to the organ. Heat shock has been shown to induce thermotolerance and also protect against ischemic injury, possibly via increased synthesis of heat shock proteins (HSPs). We hypothesized that heat shock preconditioning may protect human endothelial cells against oxidative damage. Cultured human umbilical vein endothelial cells (HUVEC) were subjected to heat shock (42 degrees C, 1 h) and allowed to recover for 2 or 20 h, at which times the cells were oxidatively stressed for 1 h by exposing them to 100-200 mumol/l of hydrogen peroxide (H2O2). Cellular damage was assessed immediately and 18 h later by morphology and release of lactate dehydrogenase (LDH). No protection of HUVEC was seen using the 2-hour recovery interval, but a significant protection (P < 0.05) was observed after the 20-hour delay. Northern blot analysis at 1 and 2 h after heating showed induction of HSP-70 mRNA. Western blot analysis demonstrated a significant increase in HSP-72 protein after 2 as well as 20 h of recovery from heat shock, although the amounts of protein at the two times were not significantly different. Furthermore, no differences in the activity of the antioxidant enzyme catalase were observed between heated and unheated HUVEC at 2 and 20 h after heat preconditioning. Thus, heat shock preconditioning induces delayed protection against oxidative injury in HUVEC, and the mechanism of protection appears to involve more than the expression of HSP-72 or activity of catalase.

Delayed administration of inhaled nitric oxide preserves alveolar-capillary membrane integrity in porcine gram-negative sepsis.

OBJECTIVE: To determine the effect of delayed administration of inhaled nitric oxide (NO) on acute lung injury after the onset of gram-negative sepsis. DESIGN: Nonrandomized controlled study. SETTING: University medical center laboratory. SUBJECTS: Yorkshire swine. INTERVENTIONS: Five groups of swine were anesthetized, mechanically ventilated, and studied for 5 hours. After surgical preparation, control (n = 10) and NO-treated control (n = 6) animals received a 1-hour infusion of sterile saline solution. Sepsis was induced with a 1-hour intravenous infusion of live Pseudomonas aeruginosa. Untreated animals with sepsis (n = 10) received no treatment. Inhaled NO at 20 ppm was administered to NO30-treated animals with sepsis (n = 7) and NO60-treated animals with sepsis (n = 8) beginning at 30 and 60 minutes after bacterial infusion was begun, respectively. MAIN OUTCOME MEASURES: Systemic and pulmonary hemodynamics, arterial blood gas determination, bronchoalveolar lavage protein and neutrophil content, neutrophil oxidant burst, lung myeloperoxidase content, and scanning electron micrographic studies. RESULTS: A progressive, significant (P < .05) decline in PaO2 developed in untreated animals with sepsis, which was prevented in NO30- and NO60-treated animals with sepsis. A significant (P < .05) increase in bronchoalveolar lavage protein and neutrophil counts compared with baseline values was observed in untreated animals with sepsis, indicating acute lung injury. These variables exhibited no notable increase in NO30- and NO60-treated animals with sepsis and were significantly (P < .05) reduced compared with untreated animals with sepsis. The lung myeloperoxidase content was significantly (P < .05) elevated at 5 hours in all groups with sepsis compared with baseline values and the control and NO-treated control groups. The total phorbol myristate acetate-induced polymorphonuclear leukocyte oxidant burst at 5 hours was significantly (P < .05) decreased in the NO30- and NO60-treated animals with sepsis compared with untreated animals with sepsis. Untreated and NO30- and NO60-treated animals with sepsis showed a significant (P < .05) increase in pulmonary artery pressure at 30 minutes, followed by a progressive decline. These changes were significant (P < .05) compared with baseline values and the control groups. No significant (P < .05) difference in pulmonary artery pressure or systemic arterial pressure was found at any time between untreated and NO30- and NO60-treated animals with sepsis. CONCLUSIONS: The delayed administration of inhaled NO preserves alveolar-capillary membrane integrity in this porcine model of gram-negative sepsis. The inhibition of neutrophil transendothelial migration, rather than neutrophil rolling or tight adhesion, may be a critical mechanism by which inhaled NO produces this effect. Decreased oxidant production by activated neutrophils may be a secondary mechanism by which inhaled NO reduces acute lung injury.

A dual-binding antibody to E- and L-selectin attenuates sepsis-induced lung injury.

Many studies indicate a pivotal role for neutrophil adhesion in sepsis-associated lung injury. Neutrophil adhesion to endothelium depends on activation and expression of selectin and integrin adhesion receptors. We studied the effects of pretreatment with a dual-binding porcine anti-E- and anti-L-selectin monoclonal antibody (EL-246) on a porcine model of sepsis-induced lung injury. Four groups were studied for 5 h. Group 1 (control animals) received intravenous saline only. Group 2 (septic) received a 1-h infusion of Pseudomonas aeruginosa. Group 3 (EL-246 pretreatment) received EL-246 (1 mg/kg) prior to Pseudomonas infusion. Group 4 (EL-246 controls) received EL-246 infusion only. Group 2 animals showed rapid, significant decline in arterial pH and oxygen tension whereas, in Group 3, physiologic deterioration was significantly attenuated. Bronchoalveolar lavage at 5 h showed a significant increase in neutrophil count and protein content in Group 2. Group 3, however, showed no significant differences in these parameters compared with control animals. Despite severe neutropenia, lung myeloperoxidase content at 5 h was significantly reduced in Group 3 compared with Group 2. There was no significant difference in pulmonary and systemic hemodynamics between Groups 2 and 3. Group 4 animals exhibited a transient neutropenia, but otherwise no other differences in measured parameters were found compared with Group 1 control animals. In conclusion, EL-246 significantly reduced neutrophil accumulation in lung and attenuated sepsis-induced lung injury, but failed to attenuate deranged pulmonary and systemic hemodynamics.(ABSTRACT TRUNCATED AT 250 WORDS)

Beneficial cardiopulmonary effects of pentoxifylline in experimental sepsis are lost once septic shock is established.

OBJECTIVE: To determine the effects of pretreatment and posttreatment with pentoxifylline in a porcine model of gram-negative sepsis. DESIGN: Nonrandomized controlled trial. STUDY SUBJECTS: Young Yorkshire swine. INTERVENTIONS: Six groups of ventilated swine were studied for 5 hours. Group 1 swine (control, n = 8) received saline solution only. Group 2 swine (sepsis, n = 8) received a 1-hour infusion of Pseudomonas aeruginosa. Groups 3, 4, and 5 swine received the P aeruginosa infusion and a 20 mg/kg bolus followed by a 6 mg/kg per hour infusion of pentoxifylline. Group 3 swine (n = 6) received pentoxifylline prior to the onset of sepsis; group 4 swine (n = 6) received pentoxifylline at 1 hour and group 5 swine (n = 4) at 2 hours after the onset of the P aeruginosa infusion. Group 6 swine (control pentoxifylline, n = 3) received pentoxifylline only. OUTCOME MEASURES: Hemodynamic variables, neutrophil counts and CD18 expression, tumor necrosis factor activity, and arterial blood gases were measured hourly. Bronchoalveolar lavage was performed at 0 and 5 hours to measure neutrophil and protein content. RESULTS: All variables remained unchanged in the control and control pentoxifylline groups. Both pretreatment and posttreatment with pentoxifylline significantly attenuated lung injury and improved arterial PaO2. The cardiac index was significantly improved by administration of pentoxifylline in groups 3 and 4. Administration of pentoxifylline to group 5 animals in established septic shock caused uncontrolled, fatal systemic hypotension in two of the four animals. Plasma tumor necrosis factor activity, blood polymorphonuclear leukocyte counts, and CD18 expression were unaffected by the administration of pentoxifylline. CONCLUSIONS: Pentoxifylline protects against pulmonary and systemic hemodynamic derangements in fulminant sepsis and protects against pulmonary dysfunction. Pentoxifylline has a "therapeutic window" when given in established sepsis; if administration is delayed until overt septic shock occurs, it may then have deleterious effects.

Dehydroepiandrosterone and related steroids induce multilamellar lipid structures in cultured human endothelial cells.

Pharmacologic doses of dehydroepiandrosterone (DHEA), a steroid hormone produced naturally by the adrenal cortex, may lower plasma lipoprotein levels in humans and reduce the severity of experimental atherosclerosis in rabbits. Effects of DHEA on cells of the vascular wall, particularly endothelial cells (EC), which are in direct contact with the plasma, have not been documented. The authors have found that micromolar doses of DHEA induce a consistent and reversible morphologic change in cultured EC derived from the human umbilical vein. During 24 hours of exposure to DHEA, cultured EC became loaded with phase-dense, perinuclear cytoplasmic granules, which persisted while DHEA remained in the culture medium. Certain steroids related to DHEA, particularly 17-ketosteroids, also induced perinuclear cytoplasmic granules. The granules lost their phase-density after fixed monolayers were extracted using ethanol or methanol. The granules did not form in media made with lipoprotein-deficient serum, suggesting that serum lipoproteins were involved in formation of the granules. Ultrastructurally, the granules were identical to multilamellar lipid structures, a type of pleomorphic lipid-containing lysosome found in foam cells. The granules were identified as lysosomes by positive reaction for acid phosphatase. The mechanism by which DHEA induces formation of lysosomal lipid structures remains to be determined.

Nonspecific cytotoxicity of recombinant interleukin-2 activated lymphocytes.

The administration of interleukin-2 (IL-2) and lymphokine activated killer (LAK) cells to patients with advanced metastatic cancer has yielded encouraging results. The purported ability of LAK cells to be discriminatively tumoricidal, thus sparing normal host tissue, represents a major advance over conventional chemotherapy. However, IL-2 adoptive immunotherapy results in dose-limiting toxicity characterized by weight gain, dyspnea, ascites, and peripheral-pulmonary edema suggestive of a vascular leak syndrome. It is unclear whether the observed toxicity is directly related to IL-2 and/or LAK cells. The authors examined the cytolytic nature of human LAK cells against human endothelial, epithelial, and fibroblast cell lines. Bovine endothelial cells also were studied. Using a 51Cr release assay, the cytolytic potential, time course, and effect of reactive oxygen intermediate inhibitors were studied. LAK cells were uniformly toxic against all cell lines, in contrast to high dose rIL-2 and excipient. Significant cytolysis was observed within 30 minutes and increased over the first 2 hours of LAK cells coming in contact with target cells. Reactive oxygen intermediate inhibitors did not reduce cytolytic activity. The authors thus found human LAK cells to be rapidly cytolytic against a variety of human and bovine cell lines. This cytolysis was independent of reactive oxygen intermediates.

Comparison of the neovascular effects of stimulated macrophages and neutrophils in autologous rabbit corneas.

The ability of peritoneal exudate macrophages and neutrophils to induce neovascularization was tested in autologous rabbit corneas. Macrophages and neutrophils elicited by proteose peptone or glycogen and macrophages activated by C parvum were purified, pelleted, and implanted 1-2 mm from the corneal limbus. Neovascular responses were evaluated by daily slit-lamp observations and terminal whole-mount and histologic examinations of colloidal carbon-perfused vessels. Pellets of elicited macrophages (5 X 10(5) - 6 X 10(6) cells, 92-98% macrophages) or activated macrophages (2 X 10(6) cells, 87-97% macrophages) induced neovascularization by 4 days in 72-82% of cases. In contrast, pellets of neutrophils (1 X 10(5) - 8 X 10(6) cells, 92-98% neutrophils) did not induce neovascularization in any case. Histologic examinations at 4-24 hours revealed diapedesis and substantial infiltration of peripheral blood neutrophils in response to implants of either macrophages or neutrophils. Infiltration was diminished by 48 hours and negligible at later times. The finding that neovascular responses were not evoked by implantation of neutrophils or by the accompanying infiltration of neutrophils indicates that neutrophils do not initiate neovascularization in this model. Under similar test conditions, neovascular responses were initiated by implantation of either elicited or activated macrophages.

Endothelial mitosis during the initial stages of corpus luteum neovascularization in the cycling adult rat.

The initial stages of neovascularization of the corpus luteum were studied in cycling adult rats using light-microscopic autoradiography. The aim of this analysis was to determine whether endothelial mitosis is a factor in this vascular growth and whether there are differences in the amount of mitotic activity in various regions of the ovary. Ovaries were examined at two time intervals: 1-2 hr and 7-8 hr following ovulation. Animals received an intraperitoneal injection of tritiated-thymidine 20 min prior to perfusion fixation of the ovaries. Autoradiographic demonstration of tritiated-thymidine labeling in endothelial nuclei was considered an indication of DNA synthesis preceding mitosis. The percentage of labeled endothelial cells in the ovaries at both time intervals varied according to the region of tissue examined and the stage of differentiation of that region. Stromal vessels were less heavily labeled than thecal vessels. Thecal vessels surrounding growing follicles were more heavily labeled than those surrounding atretic follicles. The heaviest labeling was seen in the developing corpora lutea 7-8 hr following ovulation. Minimal labeling was evident in the corpora lutea which were formed in previous cycles. A regional difference was also detected in the ovarian mesothelium. The portion of the mesothelium overlying ovulated follicles and developing corpora lutea had the greatest percentage of labeled cells. The major findings of this study were: endothelial mitosis was elevated in the initial stages of luteal neovascularization; the heightened endothelial labeling was confined to specific regions of the ovary; and mesothelium in close proximity to the developing corpora lutea also displayed heightened DNA synthesis.

A comparative study of osteoblasts: in situ versus isolated specimens.

Osteoblasts and osteoclasts are known to adhere tenaciously to bone surfaces even when the fibrous periosteal covering is removed. Brush-smear isolates and in situ, whole bone mounts of the osteogenic periosteum on the inner and outer calvarial tables of young adult rats were examined at the light microscopic level after exposing the cells by removing the fibrous periosteum. Based on staining, enzymatic activity, and overall morphology, most of the cells could be classified as osteogenic in nature. However, detailed analysis revealed that some cells exhibited certain unique morphologic appearances that suggested the existence of subpopulations or variant forms of the conventional or prototypic osteoblast. Three variant forms were the large nucleate, small nucleate, and multinucleate osteoblast-like cells, the latter with nuclear numbers ranging from 2-10. The finding of all such forms in smears, short-term cell cultures, and in situ specimens greatly lessened the probability that the unusual forms were artifactual. Phase contrast microscopic studies and sectioning of cellular isolates for study at both the light and electron microscopic levels provided additional support for the existence of the multinucleate osteoblast-like cell. The small size of these osteoblast-like subpopulations and the orientation of the cells to the commonly used plane of section of most histologic bone specimens, particularly with the multinucleate type, could explain why they had escaped earlier detection.

Mechanisms of neovascularization. Vascular sprouting can occur without proliferation of endothelial cells.

The kinetics of endothelial cells during microvascular growth were studied using a model of inflammation-induced neovascularization of the rat cornea. Inflammation was produced by central silver nitrate cauterization, cellular proliferation was assessed by tritiated-thymidine autoradiography and nuclear counts on plastic sections, and formation of new vessels was studied on whole-mount preparations after vascular perfusion with colloidal carbon. The 3H-thymidine-labeling index of endothelial cells was significantly higher than normal at 1 day following cauterization, although neither mitotic figures nor vascular sprouts were present. The labeling index reached a peak at 2 days, when cell division was evidenced by mitotic figures and doubling of the number of nuclei per section. Actual vascular sprouting also began during the 1- to 2-day interval. To determine whether vascular sprouting was dependent upon endothelial cell division, proliferation was suppressed by X-irradiation (2000 or 8000 rads) prior to cauterization. In irradiated corneas displaying no cellular proliferation, vascular sprouting at 2 days was similar to that in contralateral shielded corneas. Vascular growth continued in irradiated corneas between 2 and 4 days, but at 4 days the length of the vascular ingrowth was reduced to 66.7 and 53.4% of control after 2000 and 8000 rads, respectively. Vascular ingrowth did not progress between 4 and 7 days. This study demonstrates that initial vascular sprouting does not require proliferation of endothelial cells, although under ordinary circumstances DNA synthesis has been stimulated and is in progress at the time of sprouting. After initial sprouting without proliferation, limited vascular growth can continue for about 2 more days but subsequently ceases. Ultrastructural evaluation suggested that migration and redistribution of existing endothelial cells from the limbal vessels enable vascular sprouting and elongation without cellular proliferation.

An ultrastructural analysis of endothelial change paralleling platelet aggregation in a light/dye model of microvascular insult.

Those microvascular endothelial events that parallel the evolution of platelet aggregation were evaluated in a well-controlled animal model. Cat pial microvessels were observed through a cranial window while local platelet aggregation was produced by intravenous injection of sodium fluorescein and simultaneous exposure of the pial vessels to light from a filtered mercury lamp that excited the fluorescein. The vessels were fixed in situ when the in vivo observations of a preselected vessel indicated early, intermediate, or advanced aggregation in that vessel. The preselected vessel was then harvested for ultrastructural study together with adjacent vessels from the illuminated field. These vessels and appropriate controls were compared in semiserial thin sections. The onset of platelet aggregation in both venules and arterioles was accompanied by focal endothelial lucency, vacuole formation, luminal membrane rupture, and swelling of the nuclear envelope. These changes were not found in control material. With intermediate aggregation these changes were more common, while with advanced aggregation these abnormalities occurred together with focal endothelial denudation. Thus, in this model denudation occurred only with advanced aggregation and was not a prerequisite for aggregation.

Cytoplasmic crystalloids in irradiated rat parotid glands.

Cytoplasmic crystalloids were found in parotid acinar cells of rats given a large (6400 R) single exposure of X-rays to the head and neck. The crystalloids were first observed 1 day after irradiation and became numerous at 3-4 days. They were associated with autophagic vacuoles, which were seen in acinar cells as early as 3-6 h. Crystalloids sometimes appeared to be forming within autophagic vacuoles, which also contained membranous residues and apparently degenerating secretory material. They were bounded by a single smooth membrane and had a substructure consisting of dense, parallel longitudinal striations. They crystalloids were also seen in macrophages associated with the basal surface of acinar cells. At 3-4 days macrophages were numerous and many contained crystalloids, degenerated secretory droplets, and other cellular debris, which they apparently had phagocytosed. By 6-8 days crystalloids and macrophages were seen infrequently. Regarding mode of formation, removal by macrophages, and ultrastructure, the crystalloids resembled those described by others after ethionine intoxication. Ethionine-induced crystalloids have cytochemical characteristics consistent with a lysosomal identity. The crystalloids in irradiated parotid glands probably reflect a variant type of lysosome, which is a nonspecific manifestation of severe cellular injury and can be elicited by a variety of injurious agents.

Histamine-mediated delayed permeability response after scald burn inhibited by cimetidine or cold-water treatment.

Scald injury to one ear of the hairless mouse induced significant (P < .05) delayed edema formation in remote, uninjured skin. This remote edema formation was completely inhibited by immediate cold-water treatment of the scalded ear. Cold-water treatment significantly reduced histamine loss from the scalded ear, and the edema-inhibiting effect of the treatment could be mimicked by treating the animal prior to injury with the H2-histamine receptor antagonist cimetidine or a drug that causes histamine depletion. These observations suggest (i) that a histamine-mediated, delayed permeability response occurs after thermal injury that causes remote edema formation and (ii) that one mechanism of remote edema inhibition by cold-water treatment is the prevention of histamine release from thermally injured tissues.

Endothelial proliferation in inflammation. II. Autoradiographic studies in x-irradiated leukopenic rats after thermal injury to the skin.

The effect of leukocyte depletion on endothelial proliferation in the microvasculature of skin sites of acute inflammation was studied. Leukocytes were suppressed by 800 rad of whole-body irradiation 2 or 4 days prior to producing necrotizing thermal injuries (60 C, 20 seconds) on a shielded area of skin. Endothelial proliferation was assayed 3 days after thermal injury by quantitating the labeling index after injection of 3H-thymidine. Circulating mononuclear cells were depressed to 1.3% of pre-irradiation levels by 2 days and remained at similar levels at 5 days. Lesions developing over this interval were devoid of mononuclear infiltrate, although neutrophils emigrated as usual. Three-day lesions without mononuclear infiltrate had a mean endothelial-labeling index of 8.97%, and this was not significantly different control controls (9.42%). Lesions induced at 4 days, when circulating neutrophils were also suppressed, had reduced infiltration of neutrophils, but endothelial-labeling indexes were similar to those of controls. The results indicate that infiltration by monocytes is not a necessary stimulus for endothelial proliferation of new vessel growth in sites of nonimmunologic acute inflammation.

The effects of leukocyte depletion on corneal neovascularization.

The role of leukocytic infiltration in the initiation and maintenance of corneal neovascularization was studied in rats. Vascular ingrowth was induced by central cauterization of the cornea with silver nitrate and assessed after intraarterial perfusion of colloidal carbon. In normal rats, the mean vascular length was 0.33 mm. at 3 days and 0.63 mm. at 4 days. Whole-body x-irradiation with 800 rads depressed peripheral blood leukocyte counts to 1% of normal, prevented infiltration of monocytes, and reduced infiltration of neutrophils, but did not alter the neovascular response at 4 days. Combined treatment with radiation (800 rads) and repeated injections of antineutrophil serum (ANS) reduced peripheral leukocyte counts to nearly zero and eliminated infiltration of both monocytes and neutrophils. Despite the absence of leukocytes, neovascularization occurred in all corneas. However, the mean vascular length was reduced to 67% of control at 3 days and to 33% at 4 days. The results indicate that vascular growth can be initiated in the absence of leukocytic infiltration. The reduced neovascularization in totally leukopenic animals might be due to either the ability of neutrophils to facilitate or augment vascular growth or the nonspecific effects of treatment with radiation and antineutrophil serum.

Endothelial proliferation in inflammation. I. Autoradiographic studies following thermal injury to the skin of normal rats.

Endothelial prolifertion was studied in sites of acute inflammation induced by necrotizing (60 C for 20 seconds) or mild (54 C for 20 seconds) thermal injury to the skin of rsts. DNA synthesis in endothelial cells was assayed 6 hours to 10 days following injury by quantitation of the (3)H-thymidine labeling indices on 2-mu Epon section autoradiographs. In lesions induced at 60 C for 20 seconds, increase in DNA synthesis in small vessels around the necrotic tissue began at 1 day and became significant at 2 and 3 days (10 to 12% for endothelial cells, 9% for perivascular cells). This increased endothelial replication resulted in the formation of new blood vessels by 5 to 7 days. Endothelial labeling diminished progressively after 3 days, as the epidermis regenerated. Foci completely covered by new epidermis consistently showed lower labeling indices than those which were not reepithelialized. Mild thermal injury (54 C for 20 seconds) also resulted in significant increases in endothelial labeling (6%), but the labeling was present mainly in superficial vessels and was not followed by neovascularization. The findings with mild injury are consistent with data that vascular leakage from superficial vessels is due to direct, albeit delayed, endothelial damage. Electron microscopic studies confirmed labeling in endothelial cells and indicated that ultrastructural alterations that were previously ascribed to activation, recovery, or regenerative transformation of endothelium represent, in the main, endothelial proliferation.