- Email: [email protected]
The effects of vascular endothelial growth factor on endothelial cells: A potential role in preeclampsia
The effects of vascular endothelial growth factor on endothelial cells: A potential role in preeclampsia Jeremy C. Brockelsby, MB, BS,a Frederick W. Anthony, PhD,b Ian R. Johnson, DM,a and Philip N. Baker, DMa Nottingham and Southampton, United Kingdom OBJECTIVES: Preeclampsia is primarily a disorder of the maternal endothelium. An as yet unidentified circulating factor causes widespread alteration in endothelial function, and levels of vascular endothelial growth factor are elevated in preeclampsia. We hypothesized that vascular endothelial growth factor is involved in the alteration of endothelial function and set out to find further evidence for this contention. STUDY DESIGN: Bovine microvascular endothelial cells (B-88) were cultured in vitro. These cultured cells were then stimulated with vascular endothelial growth factor and with plasma from women with preeclampsia in the presence and absence of anti–vascular endothelial growth factor antibody. Prostacyclin, nitric oxide, and lactate dehydrogenase levels were measured. RESULTS: Vascular endothelial growth factor induced a significant concentration-dependent increase in prostacyclin production but not nitric oxide production. Cells stimulated with plasma from women with preeclampsia showed increases in production of both prostacyclin and nitric oxide. Vascular endothelial growth factor concentration in plasma was correlated with prostacyclin production by stimulated cells. The increase in prostacyclin production that usually followed the addition of plasma did not occur when anti–vascular endothelial growth factor antibody was present. CONCLUSIONS: Vascular endothelial growth factor has the ability to alter endothelial cell function in a manner analogous to that of plasma from women with preeclampsia. (Am J Obstet Gynecol 2000;182:176-83.)
Key words: Endothelium, preeclampsia, prostacyclin, vascular endothelial growth factor
Preeclampsia, a multisystem disorder characterized by hypertension and proteinuria, is among the leading causes of maternal mortality. The disease is also responsible for considerable perinatal mortality and morbidity, including the occupancy of a sixth of the neonatal hospital beds in Nottingham, United Kingdom. Although the etiology is unclear, there is accumulating evidence for a pathogenic model of preeclampsia whereby a deficiency in trophoblastic invasion of the placental bed leads to a poorly perfused fetoplacental unit. This results in the release of a factor or factors into the maternal circulation that cause an alteration of
From the Division of Obstetrics and Gynaecology, School of Human Development, Nottingham City Hospital,a and the Department of Obstetrics and Gynaecology, Princess Anne Hospital.b J.C.B. is a National Birthday Trust Fund and Wellbeing Research Training Fellow. Support was provided by Wellbeing (grant RTFB1/98) and Action Research (grant S/P/2792). Genentech, Inc, South San Francisco, Calif, supplied the vascular endothelial growth factor reagents. Gensia Sicor Pharmaceuticals, Irvine, Calif, supplied the B-88 cell line. Received for publication November 16, 1998; revised February 8, 1999; accepted July 29, 1999. Reprints not available from the authors. Copyright © 2000 by Mosby, Inc. 0002-9378/2000 $12.00 + 0 6/1/101828
176
the vascular endothelium, with the clinical syndrome resulting from widespread changes in endothelial function.1 The intact endothelium has anticoagulant properties and modifies the response of vascular smooth muscle to agonists. However, endothelial cells that have undergone alteration in their function as a result of a disease process promote coagulation and increase sensitivity to vasopressor agents, two major features of preeclampsia.2 Evidence of endothelial dysfunction in preeclampsia includes the characteristic changes in glomerular capillary endothelial morphologic characteristics, increased capillary permeability, elevated blood levels of molecules associated with endothelial activation (endothelin and fibronectin), and raised levels of soluble adhesion molecules. 1, 2 Moreover, studies examining the effects of plasma or serum from women with preeclampsia on endothelial cells in vitro have demonstrated increases in plateletderived growth factor messenger ribonucleic acid and protein production, intracellular triglyceride concentration, cellular fibronectin release, nitric oxide production, and prostacyclin generation.3 However, the nature of any circulating factors involved and the mechanism through which the endothelium is activated remain as yet undetermined.
Volume 182, Number 1, Part 1 Am J Obstet Gynecol
We are currently investigating the hypothesis that placentally derived vascular endothelial growth factor (VEGF) is the circulating factor responsible for the endothelial alteration of preeclampsia. There is strong evidence for a role of cytokines in the endothelial activation of preeclampsia. Several angiogenic growth factors, including basic fibroblast growth factor, platelet-derived growth factor, and VEGF, are expressed in placental tissue.4 However, basic fibroblast growth factor and plateletderived growth factor lack the hydrophobic signal sequence that govern secretion5 and are therefore unlikely to be the sole mediators of endothelial activation. In contrast, VEGF contains a hydrophobic secretory signal sequence and exerts effects specific to vascular endothelial cells.5 In addition to promotion of endothelial growth these include promotion of calcium ion entry into cells, induction of increased expression and release of von Willebrand factor, and activation of the phospholipase C enzyme.6 These effects have been reported at in vitro concentrations similar to in vivo circulating plasma levels.5, 7 VEGF exerts its biologic effect through 1 of 2 receptors, fms-like tyrosine kinase receptor and kinase insert domain–containing receptor.5 There is much circumstantial evidence to support our hypothesis that an elevation in VEGF concentration mediates the endothelial cell activation of preeclampsia. In vitro levels of VEGF are dramatically increased within a few hours of exposure to hypoxia.8 Moreover, VEGF has been demonstrated to promote coagulation and induce vascular permeability, 2 characteristic features of preeclampsia.2 Uchida et al9 suggested that proteinuria may result from VEGF-induced proteolytic disruption of glomerular endothelial cell basement membranes. Mitogenic activity, typical of growth factors such as VEGF, is also found in the plasma of women with preeclampsia well before the disease is clinically evident.3 In addition, VEGF has a synergistic effect with cytokines implicated in the endothelial dysfunction and in the pathogenesis of preeclampsia, such as tumor necrosis factor α and transforming growth factor β.10 We previously reported increased circulating levels of VEGF in women with preeclampsia.11 Although this finding has been challenged,12 studies by ourselves and others7, 13 have confirmed these increased levels. In this study we investigated the in vitro effects of VEGF on cultured endothelial cells, with the hypothesis that VEGF would produce effects analogous to those of plasma from women with preeclampsia. We further investigated this hypothesis by determining the effect of antibodies to VEGF on the action of plasma from women with preeclampsia. Material and methods Subjects. Plasma samples were collected from 12 patients with the diagnosis of preeclampsia, as defined by international criteria. 14 These subjects had systolic
Brockelsby et al 177
blood pressure measurements of ≥140 mm Hg and diastolic blood pressure measurements of ≥90 mm Hg on ≥2 separate occasions after the 20th week of pregnancy and had previously been normotensive and without a history of renal disease. In addition they had significant proteinuria (either >500 mg protein in a 24-hour collection or 2+ proteinuria on a voided random urine sample) in the absence of urinary tract infection. Both the hypertension and the proteinuria were resolved by the sixth week post partum in all cases. To act as a control preparation plasma was collected from 12 gestationally matched women with normotensive pregnancies without complications or underlying illnesses. All the women with preeclampsia were nulliparous and had no other complications of pregnancy. The local ethical committee approved the project, and subjects provided written consent before taking part in the study. Samples. Blood samples were taken within 24 hours of admission. Blood was collected into previously cooled glass tubes containing ethylenediaminetetraacetic acid (EDTA) with a Vacutainer (Becton Dickinson and Company, Franklin Lakes, NJ) blood collecting tube. Blood was centrifuged at 2500 rpm for 15 minutes at 4°C and then divided into 200-µL aliquots that were stored at –80°C until use. Cell culture. Most studies that have examined the effects of plasma from women with preeclampsia on endothelial cells have used human umbilical vein endothelial cells or bovine microvascular endothelial cells (B-88). We previously demonstrated differential effects of plasma from patients on different endothelial cell types,15 and we therefore chose to use B-88 cells because they produced the most reliable bioassay. Cells were supplied as a cryopreserved aliquot of 500,000 cells (Gensia Sicor Pharmaceuticals, Irvine, Calif). Since the establishment of this cell line, the phenotype of these cells has been maintained for >180 passages. Cellular characteristics include growth as a monolayer, a cobblestone morphologic appearance at confluence, and positive immunostaining for both vimentin and von Willebrand factor–related antigen. Cells were cultured in medium consisting of αmodified minimum essential medium (Gibco alphaMEM; Life Technologies, Inc, Rockville, Md) containing 10% horse serum, 2-mmol/L glutamine, 5 µg/mL gentamicin, 20 µg/mL kanamycin, and 10 U/mL nystatin. Cells were passaged with 0.025% trypsin and 0.01% EDTA. Cells were grown to confluence in 24-well plates and were maintained in a quiescent state for 24 hours before stimulation by the addition of serum-free medium. In preliminary studies we demonstrated that all cells were able to tolerate this quiescent protocol, as assessed by using lactate dehydrogenase (LDH) level as a measure of cell viability. Preliminary experiments in which cell num-
178 Brockelsby et al
ber was measured by a hemocytometer demonstrated that each well contained approximately the same number of cells. The consistency of the cell number of each well was confirmed by measurement of protein content. Experimental results are thus expressed per milligram of protein. Experimental design Addition of VEGF. Cells were stimulated for 24 hours in serum-free medium containing 0.05% bovine albumin and incremental concentrations of VEGF (0.01-1.0 nmol/L; PeproTech EC Ltd, London, United Kingdom). Effects were studied in quadruplicate wells. Similar experiments were performed in the presence of anti-VEGF antibody (MAB293; R&D Systems Europe Ltd, Abingdon, United Kingdom) at a final concentration of 10 ng/mL to confirm that the anti-VEGF antibody could block the observed effect of 1-nmol/L VEGF (46 ng/mL) on these endothelial cells. After this stimulation, medium was collected for determination of 6-keto-prostaglandin F1α (6-keto-PGF1α) and nitrite (stable metabolites of prostacyclin and nitric oxide, respectively) and LDH concentrations. Cells were collected for protein determination. In preliminary experiments we demonstrated that VEGF had no significant mitogenic effect on cell number or protein content once the cells had formed a monolayer. Addition of plasma from women with preeclampsia and from women with normal pregnancy. Cells were then stimulated for 24 hours in serum-free medium containing 0.05% bovine albumin and 10% plasma. First, plasma from women with preeclampsia or from women with normal pregnancies was added to endothelial cells. After this stimulation, medium was collected and the concentrations of prostaglandin F1α, nitrite, LDH, and cellular protein were determined. These experiments were repeated in the presence of anti-VEGF antibody (MAB293) at a final concentration of 10 ng/mL. Radioimmunoassay for VEGF. Measurements of the total VEGF concentration (bound and free) in the plasma samples were made by means of competitive radioimmunoassay with recombinant human VEGF 165; details of this technique have been described elsewhere.16 Briefly, recombinant human VEGF was labeled with iodine 125 to act as a tracer and known quantities of unlabeled VEGF were used to construct a standard curve. Standards, control preparations, and unknown samples were incubated overnight with tracer and rabbit polyclonal antiserum (lot No. 11094-70B; Genentech, Inc, South San Francisco, Calif). Then, after 30 minutes of incubation with donkey antirabbit–coated cellulose suspension, the mixture was suspended in 1 mL distilled water and centrifuged. After decantation the radioactivity of the pellet was counted and analyzed by means of the RIA-CALC, LKB-Wallac (LKB-Wallac, Sweden) package. The sensitivity of the assay was 0.125 µg/L and the interassay coefficient of variation was 6.9% for a mean level of 2.86 µg/L.
January 2000 Am J Obstet Gynecol
Nitrite assay. Nitrite production was determined by means of the spectrophotometric Griess reaction. An aliquot of medium (100 µL) from each well was mixed with 100 µL Griess reagent (1% sulfanilamide and 0.1% naphthethylenediamine dihydrochoride in 2% phosphoric acid). The mixture was then incubated at room temperature for 10 minutes and absorbance at 550 nm was read in a Vmax kinetic microplate reader (Labsystems Mulitscan MS; Labsystems Oy, Helsinki, Finland). Concentrations were determined by comparison with a standard solution of nitrite in plasma-free medium. The reaction curve was linear from 0.25 to 64 nmol/mL. The intra-assay and interassay coefficients of variation were both <10%. 6-Keto-PGF1α assay. Production of 6-keto-PGF1α, the stable metabolite of prostacyclin, was measured with a commercially available enzyme-linked immunosorbent assay (Cayman Chemical Company, Ann Arbor, Mich). The concentration of the 6-keto-PGF1α in the sample was calculated by comparison with a standard curve for 6keto-PGF1α and the absorbance was measured at 412 nm. Assay sensitivity was 20 pg/mL and the intra-assay and interassay coefficients of variation were both <10%. Protein determination. The concentration of protein was determined by means of a commercially available spectrophotometric assay (Bio-Rad Laboratories Inc, Hercules, Calif). This assay, which is based on the Lowry assay, uses the reaction of protein with alkaline copper tartrate solution and Folin reagent to produce a characteristic blue color. The concentration of the protein in the sample was calculated by comparison with a standard curve for bovine albumin and the absorbance was measured at 750 nm. Assay sensitivity was 5 µg/mL and the intra-assay and interassay coefficients of variation were both <10%. LDH determination. Cellular viability was assessed by measurement of LDH levels. The assay was based on the measurement of reduced nicotinamide adenine dinucleotide formed from nicotinamide adenine dinucleotide plus lactate. Medium (50 µL) was added to 1 mL reagent (Sigma-Aldrich Company Ltd, Poole, United Kingdom). LDH concentration was measured at 25°C by the change in absorbance at 340 nm by means of a UNICAM (Unicam Software, Inc, Portsmouth, NH) ultraviolet and visible spectrophotometer. For assessment of maximal LDH release the LDH concentrations were also measured after cells had been exposed to 1% Triton X-100 (Union Carbide Corporation, Danbury, Conn) for 24 hours. Results Subjects. Table I summarizes the demographic data and the VEGF concentrations in the cohorts of women with preeclampsia and normotensive pregnant control subjects. As anticipated, the mean arterial blood pressure was significantly greater in the preeclampsia group than in the control group (113 mm Hg vs 90 mm Hg; P > .01
Brockelsby et al 179
Volume 182, Number 1, Part 1 Am J Obstet Gynecol
Fig 1. Dose response of 6-keto-PGF1α production by B-88 cells stimulated with incremental doses of VEGF. Bars, Mean; error bars, SEM. P < .05, by repeated-measures analysis of variance. Effect of monoclonal antibody for VEGF (10 ng/mL) is represented by filled bar. Asterisk, P < .05, by Mann-Whitney U test against cells stimulated by 1.0-nmol/L VEGF.
Table I. Patient details for samples used to stimulate endothelial cells at 10% plasma concentrations Preeclampsia group
Normal pregnancy group
Characteristic
Median
Interquartile range
Median
Interquartile range
Mean arterial blood pressure at <20 wk (mm Hg) Mean arterial blood pressure at venipuncture (mm Hg) Proteinuria (g/L) Platelet count (×109 cells/L) Gestational age at sampling (wk) Gestational age at delivery (wk) Infant birth weight (kg ) Plasma concentration of VEGF (ng/mL)
86 113* 0.73* 217* 36 37 2.5* 5.2*
83-93 109-119 0.51-1.64 168-255 35-37 35-38 2.1-2.8 4.6-5.3
87 90 —† 317 36 40 3.5 3.9
83-93 89-93 —† 268-359 34-38 37-40 3.2-3.9 3.1-4.1
*P < .05, by Mann-Whitney U test, versus corresponding control values. †Not measurable.
by Mann-Whitney U test). In accord with the definition of preeclampsia all women with preeclampsia displayed significant proteinuria. As anticipated, the plasma VEGF concentrations were significantly higher in the women with preeclampsia (5.2 ng/mL vs 3.9 ng/mL; P > .01 by Mann-Whitney U test). The birth weights of infants from women with preeclampsia were significantly lower (2.51 kg vs 3.53 kg; P < .01 by Mann-Whitney U test). The platelet counts were significantly lower in the group of women with preeclampsia (217 × 109 cells/L vs 317 × 109 cells/L; P < .01 by Mann-Whitney U test). Prostacyclin production after addition of VEGF. B-88 endothelial cell prostacyclin production increased with
increasing concentrations of VEGF (P < .05 by repeated measures analysis of variance; Fig 1). Anti-VEGF antibody (10 ng/mL) had the effect of inhibiting the prostacyclin production of stimulated cells back to the observed baseline production (P < .05 by Mann-Whitney U test; Fig 1). Nitric oxide production after addition of VEGF. B-88 cells produced measurable quantities of nitrite. However, there was no significant increase in nitric oxide production, as indicated by nitrite concentrations, on stimulation with VEGF (P > .05). Correlation of plasma VEGF levels with prostacyclin and nitric oxide production. There was no significant correlation between plasma VEGF concentrations and nitric
180 Brockelsby et al
January 2000 Am J Obstet Gynecol
Fig 2. B-88 6-keto-PGF1α production of cells stimulated with plasma from women with preeclampsia (filled circles; r = 0.64; P = .02, by Spearman rank correlation) and normotensive control subjects (open squares). Lines, Linear regression.
Table II. LDH production after stimulation with incremental doses of VEGF VEGF (nmol/L)
LDH (U/mL)
0.01 0.1 1.0 Triton X-100
5.0 ± 2.1 4.22 ± 0.84 6.7 ± 1.3 42.2 ± 0.85
Data are expressed as mean ± SEM. Triton X-100 indicates the maximal release of LDH on total cell lysis.
oxide production in the medium from the stimulated B88 cell line for plasma from either the women with preeclampsia (r = 0.180; P > .05 by Spearman rank correlation) or normotensive control subjects (r = 0.223; P > .05 by Spearman rank correlation). For the plasma of women with preeclampsia, however, there was a significant correlation between prostacyclin production and plasma VEGF levels (r = 0.64; P = .02 by Spearman rank correlation; Fig 2). This correlation was not present when the normotensive group was studied. (r = 0.339; P > .05 by Spearman rank correlation; Fig 2). Effect of anti-VEGF antibody on plasma-induced endothelial cell prostacyclin and nitrite production. When endothelial cells were stimulated with plasma from women with preeclampsia, there were significant increases in the nitric oxide and prostacyclin production by cells with respect to production seen after stimulation with plasma from normotensive control subjects (P = .01; Fig 3). When anti-VEGF antibody was added to plasma
from both groups, there was no effect on the nitrite production (preeclampsia plasma, 3.10 nmol/24 hours per well; range, 2.7-4.0 nmol/24 hours per well; preeclampsia plasma plus anti-VEGF antibody, 2.83 nmol/24 hours per well; range, 2.25-3.52 nmol/24 hours per well; P > .47 by Mann-Whitney U test). However, prostacyclin production by cells stimulated with plasma from women with preeclampsia was significantly reversed by anti-VEGF (P = .01), and there was no significant differential effect of plasma plus anti-VEGF between the 2 groups (P > .05). Determination of protein and assessment of cellular viability with the use of LDH levels. There were no significant differences in the protein concentration per well with increasing VEGF concentrations (P > .05 by repeated-measures analysis of variance). The B-88 cell type showed no significant increase in LDH levels with increasing VEGF concentrations (P > .05 by repeated-measures analysis of variance; Table II). Experiments also demonstrated that the anti-VEGF antibody (5.0 ± 1.4 U/L) had no significant effect on LDH levels compared with the control preparations (5.0 ± 2.1 U/L). Comment We demonstrated that VEGF was capable of increasing the endothelial cell production of prostacyclin and that anti-VEGF antibody was capable of blocking this effect. We also showed that plasma VEGF concentrations were correlated with the increased production of prostacyclin induced by plasma from patients with preeclampsia. The observed increase in prostacyclin production induced by plasma from women with preeclampsia could be inhib-
Volume 182, Number 1, Part 1 Am J Obstet Gynecol
Brockelsby et al 181
Fig 3. B-88 6-keto-PGF1α production after stimulation with plasma from 12 normotensive women (NT) and 12 women with preeclampsia (PE) with and without anti-VEGF antibody. Horizontal lines, Median. There was a significant difference between groups. Asterisk, Significant difference at P < .05, by Mann-Whitney U test.
ited by anti-VEGF antibody. These observations are consistent with our original hypothesis that VEGF may be among the circulating factors that cause the alteration in endothelial function that is characteristic of preeclampsia.1 The effects of VEGF on cellular permeability are well documented.5 Recently Haller et al17 added further credence to our hypothesis that VEGF may be among the circulating factors in preeclampsia through their observation that plasma from women with preeclampsia increased the permeability of cultured endothelial cells by means of prostaglandin production. Our previous observation that prostacyclin concentration was significantly increased in the medium collected from B-88 cells stimulated by plasma from women with preeclampsia compared with plasma from women with normal pregnancies18 was extended by Davidge et al.19 They demonstrated that if plasma was fractionated it produced 2 distinct components, a low-molecular-weight fraction that increased prostacyclin production and a high-molecular-weight fraction that increased nitric oxide production. The low-molecular-weight fraction that stimulated prostacyclin production was in the range 42 to 62 kd, a range that corresponds to that of the VEGF family. Davidge et al19 also noted that fractionation increased the effect of plasma on prostacyclin production, and they postulated that the fractionation process may have increased the activity by causing the separation of an inhibitory protein from the active fraction. There is a protein in the serum of pregnant women that binds VEGF16; this has recently been demonstrated to be a soluble fmslike tyrosine kinase receptor.20 The increased circulating levels of VEGF in preeclampsia are probably the result of
bound rather than free VEGF.7 In this experiment we found evidence to suggest that VEGF in the plasma from women with preeclampsia stimulated endothelial cell prostacyclin production because this stimulation could be inhibited by the addition of anti-VEGF antibody. We also showed that the plasma VEGF concentrations were correlated with the increased production of prostacyclin only in the plasma from women with preeclampsia. If the 2 groups were combined, however, there remained a strong correlation of increasing prostacyclin production with increasing VEGF levels. The fact that only the group with preeclampsia was so correlated individually may well indicate either that the VEGF has a critical concentration above which it has an effect or that there were insufficient numbers in the study groups to demonstrate a correlation. Although preeclampsia is a disease that is associated with a decrease in prostacyclin production in vivo, it is well established that in vitro endothelial cells acutely exposed to plasma from women with preeclampsia demonstrate an increased production of prostacyclin. We further confirmed this in vitro finding. Most studies examining this effect (including this one) used either bovine endothelial cells (B-88) or human umbilical vein endothelial cells; however, neither of these cell types represents the human pregnant microvascular endothelium, where the pathologic process of this disease is postulated to occur.1 The extrapolation of these in vitro results to the in vivo situation must therefore be treated with caution. The results presented here serve to demonstrate that plasma from women with preeclampsia does cause an acute in vitro effect on endothelial cell prostacyclin production and that this effect can be inhibited by anti-
182 Brockelsby et al
VEGF antibody. Further work is required to confirm that this observed in vitro effect is relevant to the situation in vivo. One further caveat to be considered is our finding that the demonstrated effects of VEGF occurred at concentrations above reported circulating levels.7, 11, 13 This finding is consistent with the previous report that VEGF causes a stimulation of prostacyclin production by human umbilical vein endothelial cells at nonphysiologically relevant concentrations.21 Many cytokines have been demonstrated to enhance the effect of VEGF in vitro, and such enhancement may explain the discrepancies in these levels. Further work is required to fully elucidate the possible spectrum of cytokine interactions for this observed increase in prostacyclin production. We also previously reported that nitric oxide production was significantly increased in the medium collected from endothelial cells.22 It is interesting that in this study we were unable to demonstrate any correlation between VEGF and nitric oxide production, although we would expect nitric oxide to be involved in VEGF action. It is important to note that VEGF has been reported to increase nitric oxide production by human endothelial cells.23 Our inability to demonstrate this was probably related to the B-88 cells that we used and to the sensitivity of our assay. However, our cellular model has provided us with an ideal system for studying prostacyclin production by plasma from pregnant women. We have postulated that the elevated VEGF concentration in plasma from pregnant women originates in the placenta. Cooper et al24 (1996) reported that messenger ribonucleic acid expression of VEGF was significantly reduced in placentas from women with preeclampsia compared with expression in normotensive pregnant women. That study appears to contradict our hypothesis that placentally derived VEGF is responsible for the increased circulating levels of VEGF7, 11, 13 and thus for the altered endothelial function seen in preeclampsia. There was much overlap between the groups in the study of Cooper et al,24 however, which reflects the difficulties inherent in such a study. One experimental detail that may have been important was the selection of biopsy site. Cooper et al24 chose a “central region” of the placenta. Various studies demonstrate that the histologic appearance of the chorionic villi varies according to the position of the villus in the placental cotyledon and may reflect differences in the local environment; thus a single area of biopsy may lead to discrepancies in interpretation. Alternatively, the VEGF could have come from another source, for instance, the uterus or vascular smooth muscle cells, in response to events occurring in utero.25 We demonstrated that VEGF has an effect on the production of prostacyclin in cultured B-88 endothelial cells comparable to that of plasma from women with pregnan-
January 2000 Am J Obstet Gynecol
cies complicated by preeclampsia. This study provides circumstantial evidence that VEGF is among the circulating factors that are involved in the altered endothelial function characteristic of this disease, although the effects obtained by VEGF without plasma occurred at supraphsysiologic concentrations.
REFERENCES
1. Roberts JM, Taylor RN, Musci TJ, Rodgers GM, Hubel CA, McLaughlin MK. Preeclampsia: an endothelial cell disorder. Am J Obstet Gynecol 1989;161:1200-4. 2. Rodgers GM. Hemostatic properties of normal and perturbed vascular cells. FASEB J 1988;2:116-23. 3. Roberts JM, Redman CW. Pre-eclampsia: more than pregnancyinduced hypertension [published erratum appears in Lancet 1993;342:504]. Lancet 1993;341:1447-51. 4. Jackson MR, Carney EW, Lye SJ, Ritchie JW. Localization of two angiogenic growth factors (PDECGF and VEGF) in human placentae throughout gestation. Placenta 1994;15:341-53. 5. Ferrara N, Houck K, Jakeman L, Leung DW. Molecular and biological properties of the vascular endothelial growth factor family of proteins. Endocr Rev 1992;13:18-32. 6. Brock TA, Dvorak HF, Senger DR. Tumor-secreted vascular permeability factor increases cytosolic Ca2+ and von Willebrand factor release in human endothelial cells. Am J Pathol 1991;138:213-21. 7. Brockelsby JC, Anthony F, Johnson IR, Wheeler T, Baker PN. Elevated serum vascular endothelial growth factor concentrations in preeclampsia. Hypertens Pregnancy 1998;17(3): 283-91. 8. Shweiki D, Itin A, Soffer D, Keshet E. Vascular endothelial growth factor induced by hypoxia may mediate hypoxia-initiated angiogenesis. Nature 1992;359:843-5. 9. Uchida K, Uchida S, Nitta K, Yumura W, Marumo F, Nihei H. Glomerular endothelial cells in culture express and secrete vascular endothelial growth factor. Am J Physiol 1994;266(1 Pt 2):F81-8. 10. Stark JM. Pre-eclampsia and cytokine induced oxidative stress. Br J Obstet Gynaecol 1993;100:105-9. 11. Baker PN, Krasnow J, Roberts JM, Yeo KT. Elevated serum levels of vascular endothelial growth factor in patients with preeclampsia. Obstet Gynecol 1995;86:815-21. 12. Lyall F, Greer IA, Boswell F, Fleming R. Suppression of serum vascular endothelial growth factor immunoreactivity in normal pregnancy and in pre-eclampsia. Br J Obstet Gynaecol 1997;104:223-8. 13. Sharkey AM, Cooper JC, Balmforth JR, McLaren J, Clark DE, Charnock-Jones DS, et al. Maternal plasma levels of vascular endothelial growth factor in normotensive pregnancies and in pregnancies complicated by pre-eclampsia. Eur J Clin Invest 1996;26:1182-5. 14. Davey DA, MacGillivray I. The classification and definition of the hypertensive disorders of pregnancy. Am J Obstet Gynecol 1988;158:892-8. 15. Wellings RP, Brockelsby JC, Baker PN. Activation of endothelial cells by plasma from women with preeclampsia: differential effects on four endothelial cell types. J Soc Gynecol Investig 1998;5:31-7. 16. Anthony FW, Evans PW, Wheeler T, Wood PJ. Variation in detection of VEGF in maternal serum by immunoassay and the possible influence of binding proteins. Ann Clin Biochem 1997;34(Pt 3):276-80. 17. Haller H, Hempel A, Homuth V, Mandelkow A, Busjahn A, Maasch C, et al. Endothelial-cell permeability and protein kinase C in pre-eclampsia. Lancet 1998;351:945-9. 18. Baker PN, Davidge ST, Barankiewicz J, Roberts JM. Plasma of preeclamptic women stimulates and then inhibits endothelial prostacyclin. Hypertension 1996;27:56-61.
Volume 182, Number 1, Part 1 Am J Obstet Gynecol
19. Davidge ST, Signorella AP, Hubel CA, Lykins DL, Roberts JM. Distinct factors in plasma of preeclamptic women increase endothelial nitric oxide or prostacyclin. Hypertension 1996;28: 758-64. 20. Clark DE, Smith SK, He Y, Day KA, Licence DR, Corps AN, et al. A vascular endothelial growth factor antagonist is produced by the human placenta and released into the maternal circulation. Biol Reprod 1998;59:1540-8. 21. Bikfalvi A, Sauzeau C, Moukadiri H, Maclouf J, Busso N, Bryckaert M, et al. Interaction of vasculotropin/vascular endothelial cell growth factor with human umbilical vein endothelial cells: binding, internalization, degradation, and biological effects. J Cell Physiol 1991;149:50-9.
Brockelsby et al 183
22. Baker PN, Davidge ST, Roberts JM. Plasma from women with preeclampsia increases endothelial cell nitric oxide production. Hypertension 1995;26:244-8. 23. Hood JD, Meininger CJ, Ziche M, Granger HJ. VEGF upregulates ecNOS message, protein, and NO production in human endothelial cells. Am J Physiol 1998;274(3 Pt 2):H1054-8. 24. Cooper JC, Sharkey AM, Charnock-Jones DS, Palmer CR, Smith SK. VEGF mRNA levels in placentae from pregnancies complicated by pre-eclampsia. Br J Obstet Gynaecol 1996;103: 1191-6. 25. Ni Y, May V, Braas K, Osol G. Pregnancy augments uteroplacental vascular endothelial growth factor gene expression and vasodilator effects. Am J Physiol 1997;273(2 Pt 2):H938-44.
Key words: Endothelium, preeclampsia, prostacyclin, vascular endothelial growth factor
Preeclampsia, a multisystem disorder characterized by hypertension and proteinuria, is among the leading causes of maternal mortality. The disease is also responsible for considerable perinatal mortality and morbidity, including the occupancy of a sixth of the neonatal hospital beds in Nottingham, United Kingdom. Although the etiology is unclear, there is accumulating evidence for a pathogenic model of preeclampsia whereby a deficiency in trophoblastic invasion of the placental bed leads to a poorly perfused fetoplacental unit. This results in the release of a factor or factors into the maternal circulation that cause an alteration of
From the Division of Obstetrics and Gynaecology, School of Human Development, Nottingham City Hospital,a and the Department of Obstetrics and Gynaecology, Princess Anne Hospital.b J.C.B. is a National Birthday Trust Fund and Wellbeing Research Training Fellow. Support was provided by Wellbeing (grant RTFB1/98) and Action Research (grant S/P/2792). Genentech, Inc, South San Francisco, Calif, supplied the vascular endothelial growth factor reagents. Gensia Sicor Pharmaceuticals, Irvine, Calif, supplied the B-88 cell line. Received for publication November 16, 1998; revised February 8, 1999; accepted July 29, 1999. Reprints not available from the authors. Copyright © 2000 by Mosby, Inc. 0002-9378/2000 $12.00 + 0 6/1/101828
176
the vascular endothelium, with the clinical syndrome resulting from widespread changes in endothelial function.1 The intact endothelium has anticoagulant properties and modifies the response of vascular smooth muscle to agonists. However, endothelial cells that have undergone alteration in their function as a result of a disease process promote coagulation and increase sensitivity to vasopressor agents, two major features of preeclampsia.2 Evidence of endothelial dysfunction in preeclampsia includes the characteristic changes in glomerular capillary endothelial morphologic characteristics, increased capillary permeability, elevated blood levels of molecules associated with endothelial activation (endothelin and fibronectin), and raised levels of soluble adhesion molecules. 1, 2 Moreover, studies examining the effects of plasma or serum from women with preeclampsia on endothelial cells in vitro have demonstrated increases in plateletderived growth factor messenger ribonucleic acid and protein production, intracellular triglyceride concentration, cellular fibronectin release, nitric oxide production, and prostacyclin generation.3 However, the nature of any circulating factors involved and the mechanism through which the endothelium is activated remain as yet undetermined.
Volume 182, Number 1, Part 1 Am J Obstet Gynecol
We are currently investigating the hypothesis that placentally derived vascular endothelial growth factor (VEGF) is the circulating factor responsible for the endothelial alteration of preeclampsia. There is strong evidence for a role of cytokines in the endothelial activation of preeclampsia. Several angiogenic growth factors, including basic fibroblast growth factor, platelet-derived growth factor, and VEGF, are expressed in placental tissue.4 However, basic fibroblast growth factor and plateletderived growth factor lack the hydrophobic signal sequence that govern secretion5 and are therefore unlikely to be the sole mediators of endothelial activation. In contrast, VEGF contains a hydrophobic secretory signal sequence and exerts effects specific to vascular endothelial cells.5 In addition to promotion of endothelial growth these include promotion of calcium ion entry into cells, induction of increased expression and release of von Willebrand factor, and activation of the phospholipase C enzyme.6 These effects have been reported at in vitro concentrations similar to in vivo circulating plasma levels.5, 7 VEGF exerts its biologic effect through 1 of 2 receptors, fms-like tyrosine kinase receptor and kinase insert domain–containing receptor.5 There is much circumstantial evidence to support our hypothesis that an elevation in VEGF concentration mediates the endothelial cell activation of preeclampsia. In vitro levels of VEGF are dramatically increased within a few hours of exposure to hypoxia.8 Moreover, VEGF has been demonstrated to promote coagulation and induce vascular permeability, 2 characteristic features of preeclampsia.2 Uchida et al9 suggested that proteinuria may result from VEGF-induced proteolytic disruption of glomerular endothelial cell basement membranes. Mitogenic activity, typical of growth factors such as VEGF, is also found in the plasma of women with preeclampsia well before the disease is clinically evident.3 In addition, VEGF has a synergistic effect with cytokines implicated in the endothelial dysfunction and in the pathogenesis of preeclampsia, such as tumor necrosis factor α and transforming growth factor β.10 We previously reported increased circulating levels of VEGF in women with preeclampsia.11 Although this finding has been challenged,12 studies by ourselves and others7, 13 have confirmed these increased levels. In this study we investigated the in vitro effects of VEGF on cultured endothelial cells, with the hypothesis that VEGF would produce effects analogous to those of plasma from women with preeclampsia. We further investigated this hypothesis by determining the effect of antibodies to VEGF on the action of plasma from women with preeclampsia. Material and methods Subjects. Plasma samples were collected from 12 patients with the diagnosis of preeclampsia, as defined by international criteria. 14 These subjects had systolic
Brockelsby et al 177
blood pressure measurements of ≥140 mm Hg and diastolic blood pressure measurements of ≥90 mm Hg on ≥2 separate occasions after the 20th week of pregnancy and had previously been normotensive and without a history of renal disease. In addition they had significant proteinuria (either >500 mg protein in a 24-hour collection or 2+ proteinuria on a voided random urine sample) in the absence of urinary tract infection. Both the hypertension and the proteinuria were resolved by the sixth week post partum in all cases. To act as a control preparation plasma was collected from 12 gestationally matched women with normotensive pregnancies without complications or underlying illnesses. All the women with preeclampsia were nulliparous and had no other complications of pregnancy. The local ethical committee approved the project, and subjects provided written consent before taking part in the study. Samples. Blood samples were taken within 24 hours of admission. Blood was collected into previously cooled glass tubes containing ethylenediaminetetraacetic acid (EDTA) with a Vacutainer (Becton Dickinson and Company, Franklin Lakes, NJ) blood collecting tube. Blood was centrifuged at 2500 rpm for 15 minutes at 4°C and then divided into 200-µL aliquots that were stored at –80°C until use. Cell culture. Most studies that have examined the effects of plasma from women with preeclampsia on endothelial cells have used human umbilical vein endothelial cells or bovine microvascular endothelial cells (B-88). We previously demonstrated differential effects of plasma from patients on different endothelial cell types,15 and we therefore chose to use B-88 cells because they produced the most reliable bioassay. Cells were supplied as a cryopreserved aliquot of 500,000 cells (Gensia Sicor Pharmaceuticals, Irvine, Calif). Since the establishment of this cell line, the phenotype of these cells has been maintained for >180 passages. Cellular characteristics include growth as a monolayer, a cobblestone morphologic appearance at confluence, and positive immunostaining for both vimentin and von Willebrand factor–related antigen. Cells were cultured in medium consisting of αmodified minimum essential medium (Gibco alphaMEM; Life Technologies, Inc, Rockville, Md) containing 10% horse serum, 2-mmol/L glutamine, 5 µg/mL gentamicin, 20 µg/mL kanamycin, and 10 U/mL nystatin. Cells were passaged with 0.025% trypsin and 0.01% EDTA. Cells were grown to confluence in 24-well plates and were maintained in a quiescent state for 24 hours before stimulation by the addition of serum-free medium. In preliminary studies we demonstrated that all cells were able to tolerate this quiescent protocol, as assessed by using lactate dehydrogenase (LDH) level as a measure of cell viability. Preliminary experiments in which cell num-
178 Brockelsby et al
ber was measured by a hemocytometer demonstrated that each well contained approximately the same number of cells. The consistency of the cell number of each well was confirmed by measurement of protein content. Experimental results are thus expressed per milligram of protein. Experimental design Addition of VEGF. Cells were stimulated for 24 hours in serum-free medium containing 0.05% bovine albumin and incremental concentrations of VEGF (0.01-1.0 nmol/L; PeproTech EC Ltd, London, United Kingdom). Effects were studied in quadruplicate wells. Similar experiments were performed in the presence of anti-VEGF antibody (MAB293; R&D Systems Europe Ltd, Abingdon, United Kingdom) at a final concentration of 10 ng/mL to confirm that the anti-VEGF antibody could block the observed effect of 1-nmol/L VEGF (46 ng/mL) on these endothelial cells. After this stimulation, medium was collected for determination of 6-keto-prostaglandin F1α (6-keto-PGF1α) and nitrite (stable metabolites of prostacyclin and nitric oxide, respectively) and LDH concentrations. Cells were collected for protein determination. In preliminary experiments we demonstrated that VEGF had no significant mitogenic effect on cell number or protein content once the cells had formed a monolayer. Addition of plasma from women with preeclampsia and from women with normal pregnancy. Cells were then stimulated for 24 hours in serum-free medium containing 0.05% bovine albumin and 10% plasma. First, plasma from women with preeclampsia or from women with normal pregnancies was added to endothelial cells. After this stimulation, medium was collected and the concentrations of prostaglandin F1α, nitrite, LDH, and cellular protein were determined. These experiments were repeated in the presence of anti-VEGF antibody (MAB293) at a final concentration of 10 ng/mL. Radioimmunoassay for VEGF. Measurements of the total VEGF concentration (bound and free) in the plasma samples were made by means of competitive radioimmunoassay with recombinant human VEGF 165; details of this technique have been described elsewhere.16 Briefly, recombinant human VEGF was labeled with iodine 125 to act as a tracer and known quantities of unlabeled VEGF were used to construct a standard curve. Standards, control preparations, and unknown samples were incubated overnight with tracer and rabbit polyclonal antiserum (lot No. 11094-70B; Genentech, Inc, South San Francisco, Calif). Then, after 30 minutes of incubation with donkey antirabbit–coated cellulose suspension, the mixture was suspended in 1 mL distilled water and centrifuged. After decantation the radioactivity of the pellet was counted and analyzed by means of the RIA-CALC, LKB-Wallac (LKB-Wallac, Sweden) package. The sensitivity of the assay was 0.125 µg/L and the interassay coefficient of variation was 6.9% for a mean level of 2.86 µg/L.
January 2000 Am J Obstet Gynecol
Nitrite assay. Nitrite production was determined by means of the spectrophotometric Griess reaction. An aliquot of medium (100 µL) from each well was mixed with 100 µL Griess reagent (1% sulfanilamide and 0.1% naphthethylenediamine dihydrochoride in 2% phosphoric acid). The mixture was then incubated at room temperature for 10 minutes and absorbance at 550 nm was read in a Vmax kinetic microplate reader (Labsystems Mulitscan MS; Labsystems Oy, Helsinki, Finland). Concentrations were determined by comparison with a standard solution of nitrite in plasma-free medium. The reaction curve was linear from 0.25 to 64 nmol/mL. The intra-assay and interassay coefficients of variation were both <10%. 6-Keto-PGF1α assay. Production of 6-keto-PGF1α, the stable metabolite of prostacyclin, was measured with a commercially available enzyme-linked immunosorbent assay (Cayman Chemical Company, Ann Arbor, Mich). The concentration of the 6-keto-PGF1α in the sample was calculated by comparison with a standard curve for 6keto-PGF1α and the absorbance was measured at 412 nm. Assay sensitivity was 20 pg/mL and the intra-assay and interassay coefficients of variation were both <10%. Protein determination. The concentration of protein was determined by means of a commercially available spectrophotometric assay (Bio-Rad Laboratories Inc, Hercules, Calif). This assay, which is based on the Lowry assay, uses the reaction of protein with alkaline copper tartrate solution and Folin reagent to produce a characteristic blue color. The concentration of the protein in the sample was calculated by comparison with a standard curve for bovine albumin and the absorbance was measured at 750 nm. Assay sensitivity was 5 µg/mL and the intra-assay and interassay coefficients of variation were both <10%. LDH determination. Cellular viability was assessed by measurement of LDH levels. The assay was based on the measurement of reduced nicotinamide adenine dinucleotide formed from nicotinamide adenine dinucleotide plus lactate. Medium (50 µL) was added to 1 mL reagent (Sigma-Aldrich Company Ltd, Poole, United Kingdom). LDH concentration was measured at 25°C by the change in absorbance at 340 nm by means of a UNICAM (Unicam Software, Inc, Portsmouth, NH) ultraviolet and visible spectrophotometer. For assessment of maximal LDH release the LDH concentrations were also measured after cells had been exposed to 1% Triton X-100 (Union Carbide Corporation, Danbury, Conn) for 24 hours. Results Subjects. Table I summarizes the demographic data and the VEGF concentrations in the cohorts of women with preeclampsia and normotensive pregnant control subjects. As anticipated, the mean arterial blood pressure was significantly greater in the preeclampsia group than in the control group (113 mm Hg vs 90 mm Hg; P > .01
Brockelsby et al 179
Volume 182, Number 1, Part 1 Am J Obstet Gynecol
Fig 1. Dose response of 6-keto-PGF1α production by B-88 cells stimulated with incremental doses of VEGF. Bars, Mean; error bars, SEM. P < .05, by repeated-measures analysis of variance. Effect of monoclonal antibody for VEGF (10 ng/mL) is represented by filled bar. Asterisk, P < .05, by Mann-Whitney U test against cells stimulated by 1.0-nmol/L VEGF.
Table I. Patient details for samples used to stimulate endothelial cells at 10% plasma concentrations Preeclampsia group
Normal pregnancy group
Characteristic
Median
Interquartile range
Median
Interquartile range
Mean arterial blood pressure at <20 wk (mm Hg) Mean arterial blood pressure at venipuncture (mm Hg) Proteinuria (g/L) Platelet count (×109 cells/L) Gestational age at sampling (wk) Gestational age at delivery (wk) Infant birth weight (kg ) Plasma concentration of VEGF (ng/mL)
86 113* 0.73* 217* 36 37 2.5* 5.2*
83-93 109-119 0.51-1.64 168-255 35-37 35-38 2.1-2.8 4.6-5.3
87 90 —† 317 36 40 3.5 3.9
83-93 89-93 —† 268-359 34-38 37-40 3.2-3.9 3.1-4.1
*P < .05, by Mann-Whitney U test, versus corresponding control values. †Not measurable.
by Mann-Whitney U test). In accord with the definition of preeclampsia all women with preeclampsia displayed significant proteinuria. As anticipated, the plasma VEGF concentrations were significantly higher in the women with preeclampsia (5.2 ng/mL vs 3.9 ng/mL; P > .01 by Mann-Whitney U test). The birth weights of infants from women with preeclampsia were significantly lower (2.51 kg vs 3.53 kg; P < .01 by Mann-Whitney U test). The platelet counts were significantly lower in the group of women with preeclampsia (217 × 109 cells/L vs 317 × 109 cells/L; P < .01 by Mann-Whitney U test). Prostacyclin production after addition of VEGF. B-88 endothelial cell prostacyclin production increased with
increasing concentrations of VEGF (P < .05 by repeated measures analysis of variance; Fig 1). Anti-VEGF antibody (10 ng/mL) had the effect of inhibiting the prostacyclin production of stimulated cells back to the observed baseline production (P < .05 by Mann-Whitney U test; Fig 1). Nitric oxide production after addition of VEGF. B-88 cells produced measurable quantities of nitrite. However, there was no significant increase in nitric oxide production, as indicated by nitrite concentrations, on stimulation with VEGF (P > .05). Correlation of plasma VEGF levels with prostacyclin and nitric oxide production. There was no significant correlation between plasma VEGF concentrations and nitric
180 Brockelsby et al
January 2000 Am J Obstet Gynecol
Fig 2. B-88 6-keto-PGF1α production of cells stimulated with plasma from women with preeclampsia (filled circles; r = 0.64; P = .02, by Spearman rank correlation) and normotensive control subjects (open squares). Lines, Linear regression.
Table II. LDH production after stimulation with incremental doses of VEGF VEGF (nmol/L)
LDH (U/mL)
0.01 0.1 1.0 Triton X-100
5.0 ± 2.1 4.22 ± 0.84 6.7 ± 1.3 42.2 ± 0.85
Data are expressed as mean ± SEM. Triton X-100 indicates the maximal release of LDH on total cell lysis.
oxide production in the medium from the stimulated B88 cell line for plasma from either the women with preeclampsia (r = 0.180; P > .05 by Spearman rank correlation) or normotensive control subjects (r = 0.223; P > .05 by Spearman rank correlation). For the plasma of women with preeclampsia, however, there was a significant correlation between prostacyclin production and plasma VEGF levels (r = 0.64; P = .02 by Spearman rank correlation; Fig 2). This correlation was not present when the normotensive group was studied. (r = 0.339; P > .05 by Spearman rank correlation; Fig 2). Effect of anti-VEGF antibody on plasma-induced endothelial cell prostacyclin and nitrite production. When endothelial cells were stimulated with plasma from women with preeclampsia, there were significant increases in the nitric oxide and prostacyclin production by cells with respect to production seen after stimulation with plasma from normotensive control subjects (P = .01; Fig 3). When anti-VEGF antibody was added to plasma
from both groups, there was no effect on the nitrite production (preeclampsia plasma, 3.10 nmol/24 hours per well; range, 2.7-4.0 nmol/24 hours per well; preeclampsia plasma plus anti-VEGF antibody, 2.83 nmol/24 hours per well; range, 2.25-3.52 nmol/24 hours per well; P > .47 by Mann-Whitney U test). However, prostacyclin production by cells stimulated with plasma from women with preeclampsia was significantly reversed by anti-VEGF (P = .01), and there was no significant differential effect of plasma plus anti-VEGF between the 2 groups (P > .05). Determination of protein and assessment of cellular viability with the use of LDH levels. There were no significant differences in the protein concentration per well with increasing VEGF concentrations (P > .05 by repeated-measures analysis of variance). The B-88 cell type showed no significant increase in LDH levels with increasing VEGF concentrations (P > .05 by repeated-measures analysis of variance; Table II). Experiments also demonstrated that the anti-VEGF antibody (5.0 ± 1.4 U/L) had no significant effect on LDH levels compared with the control preparations (5.0 ± 2.1 U/L). Comment We demonstrated that VEGF was capable of increasing the endothelial cell production of prostacyclin and that anti-VEGF antibody was capable of blocking this effect. We also showed that plasma VEGF concentrations were correlated with the increased production of prostacyclin induced by plasma from patients with preeclampsia. The observed increase in prostacyclin production induced by plasma from women with preeclampsia could be inhib-
Volume 182, Number 1, Part 1 Am J Obstet Gynecol
Brockelsby et al 181
Fig 3. B-88 6-keto-PGF1α production after stimulation with plasma from 12 normotensive women (NT) and 12 women with preeclampsia (PE) with and without anti-VEGF antibody. Horizontal lines, Median. There was a significant difference between groups. Asterisk, Significant difference at P < .05, by Mann-Whitney U test.
ited by anti-VEGF antibody. These observations are consistent with our original hypothesis that VEGF may be among the circulating factors that cause the alteration in endothelial function that is characteristic of preeclampsia.1 The effects of VEGF on cellular permeability are well documented.5 Recently Haller et al17 added further credence to our hypothesis that VEGF may be among the circulating factors in preeclampsia through their observation that plasma from women with preeclampsia increased the permeability of cultured endothelial cells by means of prostaglandin production. Our previous observation that prostacyclin concentration was significantly increased in the medium collected from B-88 cells stimulated by plasma from women with preeclampsia compared with plasma from women with normal pregnancies18 was extended by Davidge et al.19 They demonstrated that if plasma was fractionated it produced 2 distinct components, a low-molecular-weight fraction that increased prostacyclin production and a high-molecular-weight fraction that increased nitric oxide production. The low-molecular-weight fraction that stimulated prostacyclin production was in the range 42 to 62 kd, a range that corresponds to that of the VEGF family. Davidge et al19 also noted that fractionation increased the effect of plasma on prostacyclin production, and they postulated that the fractionation process may have increased the activity by causing the separation of an inhibitory protein from the active fraction. There is a protein in the serum of pregnant women that binds VEGF16; this has recently been demonstrated to be a soluble fmslike tyrosine kinase receptor.20 The increased circulating levels of VEGF in preeclampsia are probably the result of
bound rather than free VEGF.7 In this experiment we found evidence to suggest that VEGF in the plasma from women with preeclampsia stimulated endothelial cell prostacyclin production because this stimulation could be inhibited by the addition of anti-VEGF antibody. We also showed that the plasma VEGF concentrations were correlated with the increased production of prostacyclin only in the plasma from women with preeclampsia. If the 2 groups were combined, however, there remained a strong correlation of increasing prostacyclin production with increasing VEGF levels. The fact that only the group with preeclampsia was so correlated individually may well indicate either that the VEGF has a critical concentration above which it has an effect or that there were insufficient numbers in the study groups to demonstrate a correlation. Although preeclampsia is a disease that is associated with a decrease in prostacyclin production in vivo, it is well established that in vitro endothelial cells acutely exposed to plasma from women with preeclampsia demonstrate an increased production of prostacyclin. We further confirmed this in vitro finding. Most studies examining this effect (including this one) used either bovine endothelial cells (B-88) or human umbilical vein endothelial cells; however, neither of these cell types represents the human pregnant microvascular endothelium, where the pathologic process of this disease is postulated to occur.1 The extrapolation of these in vitro results to the in vivo situation must therefore be treated with caution. The results presented here serve to demonstrate that plasma from women with preeclampsia does cause an acute in vitro effect on endothelial cell prostacyclin production and that this effect can be inhibited by anti-
182 Brockelsby et al
VEGF antibody. Further work is required to confirm that this observed in vitro effect is relevant to the situation in vivo. One further caveat to be considered is our finding that the demonstrated effects of VEGF occurred at concentrations above reported circulating levels.7, 11, 13 This finding is consistent with the previous report that VEGF causes a stimulation of prostacyclin production by human umbilical vein endothelial cells at nonphysiologically relevant concentrations.21 Many cytokines have been demonstrated to enhance the effect of VEGF in vitro, and such enhancement may explain the discrepancies in these levels. Further work is required to fully elucidate the possible spectrum of cytokine interactions for this observed increase in prostacyclin production. We also previously reported that nitric oxide production was significantly increased in the medium collected from endothelial cells.22 It is interesting that in this study we were unable to demonstrate any correlation between VEGF and nitric oxide production, although we would expect nitric oxide to be involved in VEGF action. It is important to note that VEGF has been reported to increase nitric oxide production by human endothelial cells.23 Our inability to demonstrate this was probably related to the B-88 cells that we used and to the sensitivity of our assay. However, our cellular model has provided us with an ideal system for studying prostacyclin production by plasma from pregnant women. We have postulated that the elevated VEGF concentration in plasma from pregnant women originates in the placenta. Cooper et al24 (1996) reported that messenger ribonucleic acid expression of VEGF was significantly reduced in placentas from women with preeclampsia compared with expression in normotensive pregnant women. That study appears to contradict our hypothesis that placentally derived VEGF is responsible for the increased circulating levels of VEGF7, 11, 13 and thus for the altered endothelial function seen in preeclampsia. There was much overlap between the groups in the study of Cooper et al,24 however, which reflects the difficulties inherent in such a study. One experimental detail that may have been important was the selection of biopsy site. Cooper et al24 chose a “central region” of the placenta. Various studies demonstrate that the histologic appearance of the chorionic villi varies according to the position of the villus in the placental cotyledon and may reflect differences in the local environment; thus a single area of biopsy may lead to discrepancies in interpretation. Alternatively, the VEGF could have come from another source, for instance, the uterus or vascular smooth muscle cells, in response to events occurring in utero.25 We demonstrated that VEGF has an effect on the production of prostacyclin in cultured B-88 endothelial cells comparable to that of plasma from women with pregnan-
January 2000 Am J Obstet Gynecol
cies complicated by preeclampsia. This study provides circumstantial evidence that VEGF is among the circulating factors that are involved in the altered endothelial function characteristic of this disease, although the effects obtained by VEGF without plasma occurred at supraphsysiologic concentrations.
REFERENCES
1. Roberts JM, Taylor RN, Musci TJ, Rodgers GM, Hubel CA, McLaughlin MK. Preeclampsia: an endothelial cell disorder. Am J Obstet Gynecol 1989;161:1200-4. 2. Rodgers GM. Hemostatic properties of normal and perturbed vascular cells. FASEB J 1988;2:116-23. 3. Roberts JM, Redman CW. Pre-eclampsia: more than pregnancyinduced hypertension [published erratum appears in Lancet 1993;342:504]. Lancet 1993;341:1447-51. 4. Jackson MR, Carney EW, Lye SJ, Ritchie JW. Localization of two angiogenic growth factors (PDECGF and VEGF) in human placentae throughout gestation. Placenta 1994;15:341-53. 5. Ferrara N, Houck K, Jakeman L, Leung DW. Molecular and biological properties of the vascular endothelial growth factor family of proteins. Endocr Rev 1992;13:18-32. 6. Brock TA, Dvorak HF, Senger DR. Tumor-secreted vascular permeability factor increases cytosolic Ca2+ and von Willebrand factor release in human endothelial cells. Am J Pathol 1991;138:213-21. 7. Brockelsby JC, Anthony F, Johnson IR, Wheeler T, Baker PN. Elevated serum vascular endothelial growth factor concentrations in preeclampsia. Hypertens Pregnancy 1998;17(3): 283-91. 8. Shweiki D, Itin A, Soffer D, Keshet E. Vascular endothelial growth factor induced by hypoxia may mediate hypoxia-initiated angiogenesis. Nature 1992;359:843-5. 9. Uchida K, Uchida S, Nitta K, Yumura W, Marumo F, Nihei H. Glomerular endothelial cells in culture express and secrete vascular endothelial growth factor. Am J Physiol 1994;266(1 Pt 2):F81-8. 10. Stark JM. Pre-eclampsia and cytokine induced oxidative stress. Br J Obstet Gynaecol 1993;100:105-9. 11. Baker PN, Krasnow J, Roberts JM, Yeo KT. Elevated serum levels of vascular endothelial growth factor in patients with preeclampsia. Obstet Gynecol 1995;86:815-21. 12. Lyall F, Greer IA, Boswell F, Fleming R. Suppression of serum vascular endothelial growth factor immunoreactivity in normal pregnancy and in pre-eclampsia. Br J Obstet Gynaecol 1997;104:223-8. 13. Sharkey AM, Cooper JC, Balmforth JR, McLaren J, Clark DE, Charnock-Jones DS, et al. Maternal plasma levels of vascular endothelial growth factor in normotensive pregnancies and in pregnancies complicated by pre-eclampsia. Eur J Clin Invest 1996;26:1182-5. 14. Davey DA, MacGillivray I. The classification and definition of the hypertensive disorders of pregnancy. Am J Obstet Gynecol 1988;158:892-8. 15. Wellings RP, Brockelsby JC, Baker PN. Activation of endothelial cells by plasma from women with preeclampsia: differential effects on four endothelial cell types. J Soc Gynecol Investig 1998;5:31-7. 16. Anthony FW, Evans PW, Wheeler T, Wood PJ. Variation in detection of VEGF in maternal serum by immunoassay and the possible influence of binding proteins. Ann Clin Biochem 1997;34(Pt 3):276-80. 17. Haller H, Hempel A, Homuth V, Mandelkow A, Busjahn A, Maasch C, et al. Endothelial-cell permeability and protein kinase C in pre-eclampsia. Lancet 1998;351:945-9. 18. Baker PN, Davidge ST, Barankiewicz J, Roberts JM. Plasma of preeclamptic women stimulates and then inhibits endothelial prostacyclin. Hypertension 1996;27:56-61.
Volume 182, Number 1, Part 1 Am J Obstet Gynecol
19. Davidge ST, Signorella AP, Hubel CA, Lykins DL, Roberts JM. Distinct factors in plasma of preeclamptic women increase endothelial nitric oxide or prostacyclin. Hypertension 1996;28: 758-64. 20. Clark DE, Smith SK, He Y, Day KA, Licence DR, Corps AN, et al. A vascular endothelial growth factor antagonist is produced by the human placenta and released into the maternal circulation. Biol Reprod 1998;59:1540-8. 21. Bikfalvi A, Sauzeau C, Moukadiri H, Maclouf J, Busso N, Bryckaert M, et al. Interaction of vasculotropin/vascular endothelial cell growth factor with human umbilical vein endothelial cells: binding, internalization, degradation, and biological effects. J Cell Physiol 1991;149:50-9.
Brockelsby et al 183
22. Baker PN, Davidge ST, Roberts JM. Plasma from women with preeclampsia increases endothelial cell nitric oxide production. Hypertension 1995;26:244-8. 23. Hood JD, Meininger CJ, Ziche M, Granger HJ. VEGF upregulates ecNOS message, protein, and NO production in human endothelial cells. Am J Physiol 1998;274(3 Pt 2):H1054-8. 24. Cooper JC, Sharkey AM, Charnock-Jones DS, Palmer CR, Smith SK. VEGF mRNA levels in placentae from pregnancies complicated by pre-eclampsia. Br J Obstet Gynaecol 1996;103: 1191-6. 25. Ni Y, May V, Braas K, Osol G. Pregnancy augments uteroplacental vascular endothelial growth factor gene expression and vasodilator effects. Am J Physiol 1997;273(2 Pt 2):H938-44.