Plasma of Preeclamptic Women Stimulates and Then Inhibits Endothelial Prostacyclin
Abstract We propose that the dichotomy between the in vivo reduction in intravascular prostacyclin production that occurs in preeclampsia and the in vitro stimulatory effect of plasma from preeclamptic patients on endothelial cell prostacyclin production is due to differential effects of chronic versus acute exposure to the plasma. We studied the acute versus chronic effects of 2% plasma from healthy pregnant and preeclamptic subjects by measuring endothelial prostacyclin production at different time periods after exposure to plasma. To determine whether such effects were specific to prostacyclin, we also measured prostaglandin E2 production. To determine whether chronic changes in prostacyclin production resulted from altered cellular responsiveness, we stimulated cells that had been exposed to plasma for 72 hours with arachidonic acid and measured prostaglandin production. Preliminary characterization of the plasma factor or factors responsible for alterations in prostaglandin production was performed. After 24 hours cells exposed to plasma from preeclamptic women produced more prostacyclin and prostaglandin E2 than cells exposed to plasma from healthy pregnant women. In contrast, after 72 hours exposure to plasma from preeclamptic women resulted in less endothelial cell prostacyclin production than exposure to plasma from healthy pregnant women, but there were no such differences in prostaglandin E2 production. Cells that had been exposed to plasma from preeclamptic women for 72 hours produced less prostacyclin but the same quantity of prostaglandin E2 after stimulation with arachidonic acid than cells exposed to plasma from healthy pregnant women. The plasma factor or factors responsible for altered prostacyclin production were sensitive to heat, acid, and proteases. In contrast to acute exposure, chronic exposure to plasma from preeclamptic women alters endothelial cells to result in decreased prostacyclin production, an observation consistent with in vivo findings.
There is accumulating evidence for a pathogenic model for preeclampsia, whereby a hypoxic fetoplacental unit secretes a factor into the maternal plasma/serum that causes dysfunction of vascular endothelial cells.1 Many of the pathophysiological changes of preeclampsia can be explained by activation or injury of endothelial cells. Although intact endothelium has anticoagulant properties and modifies the response of vascular smooth muscle to agonists, the functional properties of endothelial cells are greatly altered when activated,2 with promotion of coagulation and increased sensitivity to vasopressor agents, which are major features of preeclampsia. Further evidence of endothelial dysfunction in the disorder includes the characteristic change in glomerular capillary endothelial morphology,3 increased capillary permeability,4 and elevated blood levels of molecules associated with endothelial cell activation, such as endothelin5 and cellular fibronectin.6 Support for this concept comes from studies examining the effect of sera from preeclamptic women on endothelial cells in vitro. Sera from preeclamptic women stimulate platelet-derived growth factor mRNA and protein production,7 increase intracellular triglycerides,8 and stimulate cellular fibronectin release.9
Prostacyclin (prostaglandin I2 [PGI2]), the major prostanoid produced by endothelial cells, is a potent vasodilator and inhibits platelet aggregation. Its production by the endothelium is thought to be relevant to the control of vascular resistance and the maintenance of the nonthrombogenic surface of the vascular endothelium.10 Preeclampsia is associated with deficient intravascular production of PGI2.10 11 Paradoxically, we and other investigators have reported that serum or plasma from women with preeclampsia increases in vitro endothelial cell generation of PGI2 after 24 hours compared with the concentrations generated after exposure to serum or plasma from normal pregnancies.12 13 With the use of a different experimental paradigm, in which cells were exposed to serum that was then removed, endothelial cell PGI2 production was reduced 48 hours after exposure to serum from preeclamptic women compared with that after exposure to serum from healthy pregnant women.8 These differences are unlikely to be due to the variety of different cell types that have been studied, because when the effects of exposure to plasma from preeclamptic and healthy pregnant women on different cell types were compared, similar patterns of PGI2 production were observed.13
Preeclampsia is associated with altered lipid metabolism, including elevated circulating triglyceride levels.14 Interestingly, exposure to hyperlipidemic serum is also associated with an initial increase in endothelial cell PGI2 release, but chronic exposure results in reduced PGI2 release.15 Time course experiments examining the effects on endothelial cell PGI2 production of increasing exposure times to preeclamptic serum or plasma have not previously been reported. We thus compared the effects of plasma from healthy pregnant and preeclamptic women on PGI2 production by microvascular endothelial cells at different times of exposure to the plasma. To determine whether such effects were specific to PGI2, we also measured the production of PGE2 (produced by endothelial cells and thought to have a vasodilator effect in vivo10 ). On finding that exposure to plasma from preeclamptic women for 72 hours resulted in reduced PGI2 production, we determined whether this chronic change resulted from altered cellular responsiveness by stimulating cells that had been exposed to plasma for 72 hours with arachidonic acid and measuring prostaglandin production. We also performed preliminary characterization of the plasma factor or factors responsible for alterations in prostaglandin production.
Plasma samples (prepared from blood collected into EDTA) were studied to avoid the confounding effects of cellular products released into serum during blood coagulation. Twenty-five nulliparous pregnant subjects were recruited from the Obstetric Service of the Medical Center at the University of California–San Francisco and from the Magee-Womens Hospital, Pittsburgh, with the use of protocols approved by the respective hospital Ethics Committees. Fifteen subjects had preeclampsia, defined by the criteria of hypertension, proteinuria, hyperuricemia, and reversal of hypertension and proteinuria after pregnancy.16 Hypertension was defined as an increase of 30 mm Hg systolic or 15 mm Hg diastolic pressure compared with values obtained before 20 weeks of gestation, or an absolute blood pressure greater than or equal to 140/90 mm Hg after 20 weeks. (If blood pressure recordings in the first half of pregnancy were unknown, all subjects included in the study were healthy when reviewed in the puerperium.) Proteinuria was defined as greater than 500 mg per 24-hour collection or greater than or equal to 2+ on a voided or 1+ on a catheterized random urine specimen. Hyperuricemia was defined as greater than or equal to 0.33 mmol/L at term (≥1 SD above the normal mean concentration for gestational age).17 Ten healthy pregnant subjects were also recruited.
No subject was known to have chronic hypertension or renal or metabolic disease. All subjects gave informed consent for their inclusion in the study after the nature of the research had been explained to them. The characteristics of each group are detailed in the Table⇓.
Blood samples were obtained within 24 hours before delivery. Since samples were collected by the same investigative team, collection, preparation, and storage were similar in San Francisco and Pittsburgh. Briefly, samples were maintained at 26°C for 2 to 10 hours before centrifugation at 2000g for 20 minutes and then were aliquoted under sterile conditions and stored at −80°C. Samples obtained in San Francisco were transported at −60°C to Pittsburgh, where all experiments were performed. Before the experiments samples were thawed, realiquoted, and refrozen. To test the effects of these handling procedures, we determined that there were no correlations in prostaglandin production (at any time interval) after exposure of cells to plasma with (1) time from venipuncture to centrifugation (r2=.01 to .16, P>.05) and (2) time of storage (up to 30 months) (r2=.03 to .14, P>.05), and that up to five episodes of freezing and thawing resulted in no alteration in PGI2 production (P>.1).
Prostaglandin production after exposure of cells to plasma samples collected in Pittsburgh was similar to that after exposure to samples from San Francisco. There were no significant differences in time of storage or time from venipuncture to centrifugation between the subject groups (P>.05).
The endothelial cell culture was an endothelial cell line from a bovine coronary microvessel (B88) (Gensia Pharmaceutical). Since the establishment of this cell line, the phenotype of the endothelium has been maintained over 180 passages. The cellular characteristics include growth in a monolayer; “cobblestone” morphology at confluence; positive immunostaining for von Willebrand factor–related antigen; presence of receptors for acetylated low-density lipoprotein; and secretion of prostaglandins, tissue-type plasminogen activator, and plasminogen activator inhibitor-1.
B88 cells were grown on uncoated plastic culture dishes in α-minimum essential medium supplemented by 10% heat-inactivated horse serum, 2 mmol/L l-glutamine, 20 U/mL nystatin, 5 μg/mL gentamicin, and 20 μg/mL kanamycin at 37°C in 5% CO2. Cell cultures were dispersed with trypsin (0.05%)/EDTA (0.53 mmol/L), plated in six-well dishes, and grown as confluent monolayers. Preliminary experiments in which cell number was measured by a hemacytometer demonstrated that each well contained approximately 106 cells when a confluent monolayer had been established. Experimental results are thus expressed per 106 cells. The consistency of the cell number of each well was confirmed by measurement of protein content.
The cell monolayers were made quiescent in serum-free medium for 24 hours. This serum-free medium was then replaced with heparinized plasma. Heparinized (500 U/mL) plasma from each subject was added to duplicate wells. It was necessary to heparinize the EDTA-prepared plasma to prevent the diluted samples from clotting when added to the cells. Heparin alone had previously been found to have no effect on human umbilical vein endothelial cell PGI2 production,13 and a preliminary experiment confirmed that at this concentration heparin did not affect B88 PGI2 production. In further preliminary experiments similar effects were observed at 2% and 10% plasma concentrations; the former concentration was thus used.
PGI2 (stable metabolite, 6-keto-PGF1α) and PGE2 concentrations were measured by enzyme immunoassay (Cayman Co). Prostaglandin levels were determined in medium containing 2% plasma before the B88 cells were exposed to medium and in the aliquots of the medium removed from the cells after each time interval. Intra-assay variations were 3% and 5%, respectively; interassay variations were 8% and 10%. In addition, we tested for cross-reactivity of arachidonic acid (up to 200 mmol/L) and found no interference with the assays.
Calculations and Statistics
Results are expressed as prostaglandin production per 24 hours calculated by subtracting the amount of prostaglandin produced before the time interval from the total produced at the end of the time interval.
Parametric statistics were used to compare the characteristics of the subjects studied (Student’s unpaired t tests), and replicates of prostaglandin production were obtained after exposure to pooled samples (ANOVA with Bonferroni/Dunn post hoc test), with values expressed as mean±SE.
Nonparametric statistics (Mann-Whitney U test) were used to compare levels of prostaglandin production after exposure to individual plasma samples, as the data for the values of these parameters were found to be significantly skewed. Medians thus were used to measure central tendency.
Subject characteristics are summarized in the Table⇑. As anticipated from the definition criteria, the systolic and diastolic pressures at term were significantly greater in the preeclamptic than healthy pregnant women (P<.001). Gestational ages at delivery were significantly lower in the preeclamptic group compared with the healthy pregnant group (P<.05), as were infant birth weights (P<.001). Within each group there was no correlation between prostaglandin production from cells exposed to plasma and gestational age at delivery (Spearman correlation). None of the infant birth weights in the healthy pregnant group were below the 10th percentile for gestational age (small for gestational age); in contrast, five of the infants in the preeclampsia group were small for gestational age. No differences between the groups were found when maternal age, prepregnant weight, or prepregnant blood pressures were compared, and no correlations were found between these subject characteristics and prostaglandin production from cells exposed to plasma at any time interval.
Time Course Study
Duplicate wells were incubated with 2% plasma from each of the 25 subjects. Biovariability was minimized by study of the effects of all plasma samples on the same generation of B88 cells. Medium was removed for measurement of prostaglandin production after 24, 48, and 72 hours.
Fig 1⇓ illustrates the endothelial cell PGI2 production (stable metabolite, 6-keto-PGF1α) of the two groups of women over the 72 hours of the time course study. Production in the first 24 hours was greater in cells exposed to plasma from preeclamptic women than cells exposed to plasma from healthy pregnant women (P<.01, Mann-Whitney U test). PGI2 production did not differ between the groups in the second 24 hours. Between 48 and 72 hours after cells were exposed to plasma, PGI2 production was significantly lower in cells exposed to plasma from preeclamptic women compared with cells exposed to plasma from healthy pregnant women (P<.01).
PGE2 production over the first 24 hours of the incubation was also greater in cells exposed to plasma from preeclamptic women than in cells exposed to plasma from healthy pregnant women (P<.01) (Fig 2⇓). No differences in PGE2 production between the groups were demonstrable in the second 24-hour period. In contrast to the PGI2 data, PGE2 production in the third 24-hour incubation period also did not differ (P>.2).
Studies to Determine Whether the Chronic Reduction in PGI2 Production Resulted From Altered Cellular Responsiveness
The following experiment used B88 cells that had been exposed to 2% pooled plasma from either the previously described preeclamptic women (n=15) or the healthy pregnant women (n=10) for 72 hours. Plasma lipid concentrations in the pooled samples were determined enzymatically.18 19 Similar cholesterol levels were found in the pooled samples from preeclamptic (7.58 mmol/L) and healthy pregnant (6.39 mmol/L) women. In contrast, triglyceride concentrations were higher in the pooled samples from the preeclamptic (>4.87 mmol/L) than healthy pregnant (2.03 mmol/L) women.
Medium was removed from the cells exposed to plasma from subjects with preeclampsia for 72 hours and replaced with 1 mL of 100 μmol/L arachidonic acid (Sigma Chemical Co). Endothelial cell PGI2 and PGE2 production after 15 minutes, 1 hour, and 4 hours was then measured.
Fig 3⇓ illustrates prostaglandin production in the first 15 minutes after arachidonic acid stimulation. PGI2 production was significantly less by cells that had been exposed to plasma from preeclamptic women than by cells exposed to plasma from healthy pregnant women (P<.01), whereas no such differences in PGE2 production were found. Similar differences in PGI2 production (but not in PGE2 production) between cells exposed to plasma from the two subject groups were found after 1 hour and 4 hours of exposure to arachidonic acid (P<.01, data not shown).
Studies to Characterize the Responsible Factor or Factors in Preeclamptic Plasma
The biochemical characteristics of the factor or factors in the plasma from subjects with preeclampsia that altered prostaglandin production were evaluated for sensitivity to proteases, heat, and acid. In the following experiments untreated and treated 2% pooled plasma samples from the preeclamptic women (n=15) or healthy pregnant women (n=10) were incubated with the cells: (1) Plasma was treated with heat to 60°C for 2 hours before incubation with cells; (2) plasma was acidified with 1 mol/L HCl to pH 3 for 15 minutes and then neutralized with 1 mol/L NaOH to pH 7 before incubation with cells; and (3) protease sensitivity was tested with the cysteine protease papain (Boehringer Mannheim Biochemicals, 0.2 U/mL). Plasma samples were incubated with papain and then incubated with 20% (vol/vol) α2-macroglobulin (Boehringer Mannheim), a protease inhibitor that irreversibly binds proteases via covalent bonds.20 α2-Macroglobulin was removed by centrifugation (5 minutes, 800g).
Medium was removed for measurement of prostaglandin production after 24, 48, and 72 hours.
Production in the First 24 Hours of Incubation With Plasma
When the plasma was from preeclamptic women, PGI2 production was significantly lower after incubation with pooled plasma samples treated with protease than untreated samples (P<.05), and the effects of treatment with heat or acid approached significance (.1>P>.05). When the plasma was from healthy pregnant women, no differences in PGI2 production were found between incubations with untreated samples and samples treated with heat, acid, or protease (Fig 4⇓, top).
The significant difference between PGI2 production by cells incubated with plasma from preeclamptic women compared with plasma from healthy pregnant women (P<.01) was eliminated by heat, acid, and protease treatment.
Production Between 24 and 72 Hours of Incubation With Plasma
When the plasma was from preeclamptic women, PGI2 production was significantly greater after incubation with pooled plasma samples treated with heat than untreated samples (P<.05), and the effects of treatment with acid or protease approached significance (.1>P>.05). When the plasma was from healthy pregnant women, no differences in PGI2 production between incubations with untreated samples and samples treated with heat, acid, or protease were found (Fig 4⇑, bottom).
The significant difference between PGI2 production by cells incubated with plasma from preeclamptic women compared with plasma from healthy pregnant women (P<.01) was eliminated by heat, acid, and protease treatment.
No effects of heat, acid, or protease treatment on PGE2 production after incubation with plasma from preeclamptic or healthy pregnant women were found in either time interval (data not shown).
After 24 hours production of both PGI2 and PGE2 was greater by endothelial cells exposed to plasma from preeclamptic women than to plasma from healthy pregnant women. These findings concur with previous studies of endothelial cell PGI2 production.12 13 In contrast, by 72 hours PGI2 production was diminished in cells exposed to plasma from preeclamptic women compared with cells exposed to that from healthy pregnant women (a finding in accord with in vivo observations10 11 ), whereas no differences in PGE2 production were found.
These findings suggest that the differential effect of chronic exposure to plasma from preeclamptic women may be specific to PGI2 production. When the plasma was removed after 72 hours and the cells were stimulated with arachidonic acid, PGI2 production by cells that had been exposed to plasma from preeclamptic women was lower than the production by cells exposed to plasma from healthy pregnant women; however, there was no differential stimulation of PGE2 production. Chronic exposure to a factor or factors in plasma from subjects with preeclampsia appears to alter the responsiveness of the endothelial cells to arachidonic acid. It is difficult to extrapolate the in vivo situation (where the half-life of endothelial cells extends for months) to a cell culture environment. However, the period of 72 hours may be more relevant to the in vivo situation than to the previously reported shorter time periods.
It is unlikely that the differences in PGI2 production were due to differential cytotoxic effects of plasma from preeclamptic and healthy pregnant women. Similar PGE2 levels were produced after 72 hours of exposure to plasma. Moreover, we have found no differences in B88 endothelial cellular viability (assessed by lactate dehydrogenase determination) between cells exposed to plasma from preeclamptic women and cells exposed to that from healthy pregnant women at time intervals ranging from 24 to 120 hours (unpublished data, 1994). However, considerations of cytotoxicity are complicated by the known cytoprotective effect of prostaglandin,21 which might mask an initial effect of plasma from preeclamptic women.
There is a close parallel between these results and those obtained after bovine aortic endothelial cells were exposed to hyperlipidemic serum,15 with increased PGI2 production at 24 hours and reduced production after 72 hours. Preeclampsia is associated with abnormal lipid metabolism,14 and plasma triglycerides were elevated in the present study. In addition, a number of reports indicate that blood levels of lipid peroxidation products are elevated in women with preeclampsia (reviewed in Reference 22). Lipid peroxidation is a process that occurs in all cells and involves the oxidative conversion of unsaturated fatty acids to lipid peroxides.22 Normal levels of lipid peroxides play an essential physiological role as activators of prostaglandin endoperoxide synthase in the prostaglandin cascade.23 However, as lipid peroxide levels become elevated under certain pathological conditions, PGI2 synthase is specifically impaired.23 The diminution in PGI2 production seen after exposure to plasma from women with preeclampsia may thus be occurring as a consequence of elevated lipid peroxidation associated with the disease.
An alternative possible mechanism, which would account for the acute stimulatory effect of plasma from preeclamptic women relative to that from healthy pregnant women, concerns the interaction between nitric oxide and prostaglandin synthesis. Under different conditions nitric oxide has been reported as having either an activating or an inhibiting effect on cyclooxygenase enzymes, which are key regulators of prostaglandin synthesis.24 25 We have previously reported that nitric oxide synthase inhibition by l-arginine antagonists results in reduced production of both nitric oxide and prostaglandins by B88 endothelial cells,26 suggesting that nitric oxide is acting to increase cyclooxygenase activity. Exposure to plasma from preeclamptic women increases B88 cell nitric oxide production,27 and this might also contribute to the initial elevation in prostaglandin synthesis.
Interestingly, at 72 hours levels of PGI2 production were very low after exposure of cells to plasma from approximately half the subjects with preeclampsia, and production was similar to control levels in the other half. Although this suggests different subgroups, there were no evident differences in the clinical expression of preeclampsia in subsets of subjects whose plasma resulted in the extremes of PGI2 release.
Similar differences in prostaglandin production were found with pooled and individual plasma samples. Absolute levels measured after exposure to pooled and individual samples were different, which is explained by the fact that different passages of the B88 cells were used. The differences were typical of those found with different passages and were found when plasma from healthy pregnant women and preeclamptic women was studied.
This study provides further support for the concept that a circulating factor or factors are responsible for differences in endothelial cell function between preeclamptic and healthy pregnant women, with differences being found at a plasma concentration as low as 2%. Preliminary characterization studies suggested that the factor or factors were sensitive to heat, acid, and protease treatment. Whatever the underlying mechanism, the time course experiment resolved the dichotomy between in vitro stimulatory effects on PGI2 production of plasma from preeclamptic women12 13 and the deficient in vivo PGI2 production that has previously been reported.10 11
This study was supported by National Institutes of Health grant HD-30367. P.N.B. was the 1994 Wellbeing (UK)/American College of Obstetrics and Gynecology Exchange Research Fellow. This article is dedicated to the late Zbigniew Rymaszewski, Department of Endocrinology, University of Cincinnati (Ohio), without whose collaboration this study would not have been possible. We gratefully acknowledge the help of the Clinical Data Core and nursing staff of both the Magee-Womens Hospital, Pittsburgh, and the Medical Center at the University of California–San Francisco for invaluable assistance in sample collection.
- Received January 18, 1995.
- Revision received March 14, 1995.
- Accepted September 20, 1995.
Rodgers GM. Hemostatic properties of normal and perturbed vascular cells. FASEB J. 1988;2:116-123.
Roberts JM, Edep ME, Goldfein A, Taylor RN. Sera from preeclamptic women specifically activate human umbilical vein endothelial cells in vitro. Am J Reprod Immunol. 1992;27:101-108.
Taylor RN, Musci TJ, Rodgers GM, Roberts JM. Prepartum pre-eclamptic sera stimulate platelet-derived growth factor mRNA and protein production by cultured human endothelial cells. Am J Reprod Immunol. 1991;25:105-108.
Myatt L. Eicosanoids and blood pressure regulation. In: Sharp F, Symonds EM, eds. Hypertension in Pregnancy: Proceedings of the 16th Study Group RCOG. New York, NY: Perinatology Press; 1987:167-182.
Allain CC, Poon LS, Chan CSG, Richmond W, Fu PC. Enzymatic determination of total serum cholesterol. Clin Chem. 1974;20:470-475.
Bucolo G, David H. Quantitative determination of serum triglycerides by the use of enzymes. Clin Chem. 1973;19:476-482.
James K. Alpha2 macroglobulin and its possible importance in immune systems. Trends Biochem Sci. 1980;5:43-47.
Savemini D, Misko TP, Masferrer JL, Seibert K, Curries MG, Needleman P. Nitric oxide activates cyclooxygenase enzymes. Proc Natl Acad Sci U S A. 1993;90:7240-7244.
Stadler J, Harbrecht BG, Di Silvio MD, Curran RD, Jordan ML, Simmons RL, Billiar TR. Endogenous nitric oxide inhibits the synthesis of cyclooxygenase products and interleukin-6 by rat Kupffer cells. J Leukoc Biol. 1993;53:165-172.
Davidge ST, Baker PN, McLaughlin MK, Roberts JM. Nitric oxide produced by endothelial cells increases production of eicosanoids through activation of prostaglandin H synthase. Circ Res. 1995;77:274-283.
Davidge ST, Baker PN, Roberts JM. Nitric oxide synthase expression is increased in endothelial cells exposed to plasma from women with preeclampsia. Am J Physiol. 1995;269:1106-1112.