Positive versus negative effects of VEGF₁₆₅ on Ca²⁺ signaling and NO production in human endothelial cells

In this manuscript, we show that VEGF levels associated with preeclampsia are a net negative contributor to potential vasodilator production in both a human ex vivo and in vitro endothelial cell model. Therefore, pharmacological targeting of VEGF-stimulated signaling pathways could be a novel treatment modality for preeclampsia-related hypertension.

Abstract

The role increased vascular endothelial growth factor (VEGF) plays in vascular function during normal vs. preeclamptic pregnancy has been a source of some controversy of late. In this study, we seek to understand how VEGF₁₆₅ influences vasodilator production via Ca²⁺ signaling mechanisms in human endothelial cells. We utilize human umbilical vein endothelial cells (HUVEC) as well as intact ex vivo human umbilical vein (HUV Endo) to address direct stimulation of Ca²⁺ and NO by VEGF₁₆₅ alone, as well as the effect of VEGF₁₆₅ on subsequent ATP-stimulated Ca²⁺ signaling and NO production. We show that VEGF₁₆₅ stimulates Ca²⁺ responses in both HUVEC and HUV Endo, which results in a corresponding increase in NO production in HUV Endo. Longer-term VEGF₁₆₅ pretreatment then inhibits sustained Ca²⁺ burst responses to ATP in HUVEC and HUV Endo. This is paralleled by a corresponding drop in ATP-stimulated NO production in HUV Endo, likely through inhibition of Cx43 gap-junction function. Thus, although VEGF₁₆₅ makes a small initial positive impact on vasodilator production via direct stimulation of Ca²⁺ responses, this is outweighed by the greater subsequent negative impact on Ca²⁺ bursts and vasodilator production promoted by more potent agonists such as ATP. Overall, elevated levels of VEGF₁₆₅ associated with preeclampsia could contribute to the endothelial dysfunction by preventing Ca²⁺ bursts to other agonists including but not limited to ATP.

NEW & NOTEWORTHY In this manuscript, we show that VEGF levels associated with preeclampsia are a net negative contributor to potential vasodilator production in both a human ex vivo and in vitro endothelial cell model. Therefore, pharmacological targeting of VEGF-stimulated signaling pathways could be a novel treatment modality for preeclampsia-related hypertension.

pregnancy is a time of dramatic changes in vascular physiology. Processes of angiogenesis and vasodilation both contribute to the decrease in vascular resistance and the redirection of blood to the uteroplacental unit. Vascular endothelial growth factor (VEGF) has been implicated to varying degrees in each of these processes. To further complicate matters, the exact role VEGF signaling plays in the vascular physiology of pregnancy is still poorly understood, not least because of the many isoforms of VEGF that may be present. Much of the confusion in the VEGF literature also has to do with imprecisely defined VEGF measurement criteria. Although “VEGF” is often used in shorthand for VEGFA, it should be noted that multiple variants of VEGFA occur naturally, including VEGF₁₆₅ (the most abundant variant in humans), VEGF₁₂₁, VEGF₁₈₉, and VEGF₂₀₆ (reviewed in Ref. 14). Often the literature on circulating VEGF in pregnancy and PE lack specifics on which forms of VEGF are detected and therefore reported.

A large proportion of more recent VEGF-related studies in the context of pregnancy adaptation and vascular disorders of pregnancy revolve around the soluble VEGFR1/FLT-1 receptor splice variant, known more commonly as sFLT-1, and its level relative to placental growth factor (PlGF), a member of the VEGF family of peptides. Although there is consensus that sFlt-1 is elevated, controversy still surrounds the specific role of sFLT-1 in preeclampsia (PE), a hypertensive disorder of pregnancy. High levels of circulating sFLT-1 have been correlated with PE (26, 44). Some believe that erroneous sFLT-1 production lowers circulating VEGF levels to an extent that inhibits the ability of VEGF to promote angiogenesis (33); others believe that sFLT-1 is a protective reaction to an overabundance of circulating VEGF in preeclampsia (13). There are also reports that VEGF production in preeclampsia outpaces sFLT-1 production (Ref. 27; reviewed in Ref. 18). Because of this, some have focused on “free VEGF” levels in PE, as opposed to total VEGF. Those studies, which have focused specifically on free VEGF, have shown that both free VEGF and PlGF levels are decreased in the circulation during PE pregnancies (24, 26, 31). Vascular pathologies related to decreased bioavailability of VEGF are currently being studied. However, there is evidence of greatly increased production and secretion of VEGF isoforms other than PlGF in certain tissues in PE, most notably in the decidua (13, 32, 43). Thus, although some tissues experience decreased VEGF bioavailability, others may simultaneously be experiencing local elevations of VEGF in the same patient. This raises the question of whether signaling mechanisms stimulated by excessive VEGF contributes to endothelial dysfunction of preeclampsia or whether local increases in VEGF are necessary to compensate for preexisting endothelial dysfunction caused by generally decreased circulating levels of free VEGF.

We have recently shown in an ovine model that uterine artery endothelial cells from pregnant ewes normally display a pregnancy adapted increase in Ca²⁺ signaling to heptahelical receptor agonists such as ATP that are linked to phospholipase C beta (39). Such pregnancy adaptation (reprogramming) enhances and sustains Ca²⁺ bursting and so nitric oxide (NO) output far above that in nonpregnancy, and this is dependent on increased connexin 43 (Cx43) gap-junction communication (40). An emerging role of VEGF₁₆₅ signaling in preeclampsia is the role it plays in acutely phosphorylating Cx43 gap junctions at inhibitory residues, thus reducing cell-cell communication. We have recently shown that prior exposure to VEGF₁₆₅ can reduce pregnancy-adapted function to levels equivalent to cells derived from non-pregnant ewes (6, 41), and this is apparently due to Cx43 inhibitory phosphorylation (6).

More recently, we have extended our studies to humans and shown enhanced Ca²⁺ bursting is also observable in human umbilical vein endothelium (HUV Endo) of cords from normal pregnancy. Of note, such bursting is blunted in cords from PE women, but we do not yet know whether VEGF₁₆₅ exposure can reproduce this phenomenon or not. We also do not know to what extent VEGF₁₆₅ may have a positive compensatory effect on these cells, since VEGF has been reported to be a vasodilator in its own right through activation of phospholipase C (PLC) gamma-mediated Ca²⁺ responses and NO production (1, 5, 17, 41).

In summary, to fully understand this dichotomy, we must be able to understand the relative contributions of VEGF to the inhibition of general vasodilator production capacity of the cell and VEGF-stimulated vasodilator production in human cells and tissues to be able to compare any inhibitory response to our prior data on PE subjects. Our translation of knowledge from former studies in animal models into human counterparts will also allow us to assess the possible impact of human clinical therapies based on VEGF signaling pathways in the future. We hypothesized that, in humans as in sheep, VEGF₁₆₅ will stimulate Ca²⁺ responses in a subset of cells. We also hypothesized that VEGF₁₆₅ will inhibit gap-junction function and therefore inhibit ATP-stimulated Ca²⁺ bursts responses far more, so the net effect is loss of function. If we are correct, then our study should support the notion that VEGF₁₆₅ itself can be a damaging agent in certain circumstances and indirectly support the notion that sFLT-1 elevation in some PE subjects may be an attempt to deal with local VEGF excess but is insufficient to cope.

MATERIALS AND METHODS

Human Subjects Enrollment and Tissue Collection

The Institutional Review Boards of the University of Wisconsin Hospital and Clinics and Meriter Hospital (both located in Madison, WI) approved this study. All subjects signed a formal consent. Subjects used were the same control patients as in Krupp et al. (22). Data were obtained from intact HUV Endo of six subjects. Human umbilical vein endothelial cells (HUVEC) were pooled and cultured from five subjects for experimentation and data analyses.

Materials

Fura-2-AM, DAF2-DA, Minimum Essential Medium (MEM), Medium 199 (M199), fetal bovine serum (FBS), and all other general cell culture reagents were purchased from Life Technologies (Grand Island, NY) unless otherwise noted. The endothelial cell agonist ATP (disodium salt) and all other general laboratory chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless stated otherwise. Endothelial cell growth supplement was from Millipore (Billerica, MA), heparin (sodium salt, grade 1A from porcine intestinal mucosa, no. H3149) was from Sigma, and CaCl2 was from EMD Chemicals (San Diego, CA). Glass-bottom microwell dishes (35 mm) for [Ca²⁺]_i imaging studies were from MatTek Corporation (Ashland, MA). Gap peptides GAP27(37,43) (sequence SRPTEKTIFII), GAP27 (37, 43) (sequence REKIITFIPT), GAP27(40) (sequence SRPTEKNVFIV), GAP26(43) (sequence VCYDKSFPISHVR), and GAP26(37,40) (sequence VCYDQAFPISHIR) were from Anaspec (San Jose, CA), purity of >95% by HPLC.

Methods

The protocols for isolating the intact human umbilical vein endothelium and for culturing the human umbilical vein endothelial cells were as in Krupp et al. (22). All standard culture media used for cell isolation or culture (MEM or M199) routinely contained 1% penicillin-streptomycin solution (Life Technologies) and 4 μg/ml gentamycin (Life Technologies). The methods of dual imaging of intracellular free Ca²⁺ concentration ([Ca²⁺]_i) and simultaneous NO production utilized in intact UA Endo and the [Ca²⁺]_i imaging alone were optimized in Krupp at el. (22) and repeated identically in this study.

Simultaneous imaging of [Ca²⁺]_i and NO in intact HUV Endo.

Umbilical veins of ~6-mm length were loaded with both DAF2-DA (10 μM) and Fura-2-AM (10 μM) in standard Krebs buffer (in mM: 125 NaCl, 5 KCl, 1 MgSO4, 1 KH2PO4, 6 glucose, 2 CaCl2, 25 HEPES, pH 7.4) for 90 min. Dye hydrolysis was allowed to proceed for 30 min in Krebs buffer. The vessel placed on an inverted microscope (Diaphot 150; Nikon, Melville, NY) with the ×20 phase/fluor objective focused on the endothelium after mounting in imaging chamber. Individual endothelial cells (>30 cells/field) were visualized using dual excitation (switching at 340 and 380 nm) for Fura-2 and 485-nm excitation DAF-2 (high-speed Lambda 10-2; Sutter Instruments, Novato, CA). Emission was uniformly measured at 535 nm for all excitation wavelengths. Images were recorded in real time by a PixelFly camera (Cooke Corporation, Romulus, MI) using InCyt Im3 software (Intracellular Imaging, Cinncinati, OH). Relative fluorescence (F/F0) was then calculated as the relative measure of intracellular NO, whereas [Ca²⁺]_i was calculated ratiometrically against a pre-recorded standard curve.

HUVEC cell isolation and culture.

HUVEC cells from normal (control) pregnancies were collected, cultured, and validated as in Krupp et al. (22) and Anaya et al. (2). Individual cell preps were isolated from freshly harvested umbilical cords and expanded to passage 2. Five individual preps were grown up and pooled to be frozen down for experimental usage at passage 3.

HUVEC imaging.

Cultured HUVEC frozen at passage 3 were plated to six 35-mm glass-bottom dishes in HEH media [MEM-based media containing 20% FBS, endothelial cell growth supplement (3.75 mg/100 ml), and porcine intestinal mucosa heparin (10 mg/100 ml)]. Cells were grown for 7 days and imaged at 95% confluence. HUVEC were incubated in 10 μM of Fura-2 AM with 0.05% Pluronic acid F127 (Life Technologies) dissolved in 1 ml HEH media for 1 h at 37°C. The cells were then incubated in Krebs buffer for 30 min at room temperature for ester hydrolysis. The dish was then placed in the field of view, and Fura-2 loading was verified by viewing at 380-nm UV excitation on a Nikon inverted microscope (above) and 70–90 cells were selected for imaging. An initial 5-min baseline recording was performed before subsequent addition of 100 μM ATP. The [Ca²⁺]_i for each cell was calculated in real time against an established ratiometric standard curve using the InCyt Im2 software. ATP-stimulated Ca²⁺ bursts were counted before and after VEGF pretreatment.

Statistical analysis.

For vessels, data from six or more individual subject cell preparations were collected. For HUVEC, an average of ~80 cells/dish were analyzed for four to six dishes of pooled cells. For Ca²⁺ burst inhibition experiments, cells that initially responded with 3+ bursts were counted, since prior experiments have shown that complete conversion to capacitative entry has taken place by the third burst. Data were subsequently analyzed by Student’s t-test for single-stimulation experiments. For repeat ATP stimulation experiments, changes in Ca²⁺ burst numbers were analyzed by Rank sum test on differences against internal control. A value of P < 0.05 was considered statistically significant.

RESULTS

VEGF₁₆₅-Stimulated Ca²⁺ Response in Primary HUVEC

Stimulation of primary HUVEC with 10 ng/ml VEGF₁₆₅ revealed a slightly delayed onset but overall prolonged Ca²⁺ response. In Fig. 1, we show the mean Ca²⁺ response to VEGF₁₆₅ stimulation. Consistent with what others have reported (3, 4, 12, 25), we see a moderately rapid rise in Ca²⁺ levels, followed by a sustained plateau phase, which continues for the duration of the 2,100-s experiment. This contrasts with the repetitive transient Ca²⁺ burst pattern resulting from a typical ATP stimulation (light gray dotted line). There was no significant difference in proportion of cells responding to VEGF₁₆₅ (81.47 ± 5.94%) and ATP (90.56 ± 2.06%), P = 0.116. We also examined the role of gap junctions in VEGF₁₆₅-stimulated Ca²⁺ responses using the classic pharmacological gap-junction inhibitor 12-O-tetradeconylphorbol 13-acetate (TPA). There was no change in proportion of cells responding to VEGF₁₆₅ (86.46 ± 3.63%) and TPA+VEGF₁₆₅ (90.64 ± 3.17%), P = 0.487; or area under the curve (AUC) of the VEGF₁₆₅-stimulated Ca²⁺ response when pretreated with TPA (Fig. 1, inset).

Fig. 1.

Mean Ca²⁺ response to 30 min of VEGF₁₆₅ treatment in human umbilical vein endothelial cells (HUVEC). Fura-2-loaded HUVEC were placed on an inverted fluorescent microscope. After a 5-min baseline intracellular free Ca²⁺ concentration ([Ca²⁺]_i) recording, cells were exposed to 10 ng/ml vascular endothelial growth factor (VEGF)₁₆₅ for 30 min. The mean response ± SE is depicted in black over the duration of the experiment. For comparison, a typical 100 μM ATP-stimulated HUVEC [Ca²⁺]_i recording is shown in gray. For VEGF₁₆₅ treatment, n = 6 dishes. Inset: mean area under the curve (AUC) of 10 ng/ml VEGF₁₆₅-stimulated Ca²⁺ responses with and without 10 nM TPA pretreatment for 30 min, normalized to VEGF₁₆₅ alone (VEGF₁₆₅, n = 5 dishes; TPA + VEGF₁₆₅, n = 5 dishes; P = 0.221).

VEGF₁₆₅-Stimulated Ca²⁺ Response in HUV Endo

Intact HUV Endo from normal pregnancies respond to stimulation with 10 ng/ml VEGF₁₆₅ with an initial transient Ca²⁺ peak followed by a relatively short (<10 min) sustained small elevation in [Ca²⁺]_i (Fig. 2A). Figure 2A also shows that this is accompanied by modest NO production through the first 10–15 min of the 30-min recording, resulting in an increase amounting to 1.1-fold NO production compared with basal levels. Figure 2B shows there is no effect of TPA on VEGF₁₆₅-stimulated Ca²⁺ or NO responses in HUV Endo, for mean peak Ca²⁺ (P = 0.640), mean Ca²⁺ at 900 s (P = 0.644), or total NO (P = 0.639).

Fig. 2.

Mean [Ca²⁺]_i and DAF/nitric oxide (NO) recordings after VEGF₁₆₅ stimulation in UV Endo. Dual imaging of [Ca²⁺]_i and NO was performed in ex vivo intact UV Endo using simultaneous loading of Fura-2 and DAF-2 fluorescent dyes. A: 10 ng/ml VEGF₁₆₅ was added at time 0 and recorded for 30 min. B: UV Endo are pretreated with 10 nM TPA for 30 min before 10 ng/ml VEGF₁₆₅ addition at time 0. Values are means + SE (n = 6 vessels).

Role of Gap Junctions in Sustained Ca²⁺ Bursts in HUVEC

To test the role of gap junctions in regulating sustained ATP-stimulated Ca²⁺ bursts in primary HUVEC, we exposed cells to TPA. Figure 3, A and B, show representative single-cell Ca²⁺ tracings in response to 100 μM ATP treatment before and after exposure to TPA. Number of ATP-stimulated Ca²⁺ bursts are marked above each respective burst, showing inhibition of ATP-stimulated Ca²⁺ bursts. Figure 3C shows the quantification of the effect of 1 or 10 nM TPA on ATP-stimulated Ca²⁺ bursts. Pretreatment with TPA caused a significant reduction in Ca²⁺ bursts at 1 nM (58% of control) or 10 nM (32% of control). A repeat ATP stimulation control was carried out, and, in the absence of TPA pretreatment, no change in Ca²⁺ burst numbers were observed between first and second stimulation, so tachyphylaxis is not a concern.

Fig. 3.

TPA pretreatment inhibits sustained ATP-stimulated Ca²⁺ bursts in HUVEC. HUVEC were loaded with Fura-2, stimulated with 100 μM ATP for 30 min, and Ca²⁺ bursts were counted. Cells were then washed and pretreated with TPA for 30 min, with focus maintained on the cells initially recorded. After pretreatment, a 5-min baseline [Ca²⁺]_i was recorded, and 100 μM ATP was added for an additional 30 min, with Ca²⁺ bursts counted a second time. A: a single-cell representative tracing shows the number of ATP-stimulated Ca²⁺ bursts before and after 1 nM TPA treatment in the same cell. B: the effect of 10 nM TPA is also shown for a single representative cell. C: the quantification of Ca²⁺ bursts for both TPA treatment types for all cells that gave three or more Ca²⁺ bursts upon initial ATP stimulation. For Control, n = 143 cells; for 1 nM TPA, n = 147 cells; for 10 nM TPA, n = 231 cells. Results are shown as means +SE, normalized to initial ATP stimulation. *Significant difference (P < 0.001).

To more specifically address which connexin proteins regulate ATP-stimulated Ca²⁺ bursts, we administered various short peptides designed to mimic extracellular regions of specific connexin proteins. Figure 4 shows the quantification of the effect of these treatments on ATP-stimulated Ca²⁺ bursts. Two peptides [Gap26(37,40) and Gap26(43)] significantly reduced Ca²⁺ bursts compared with control. One peptide [Gap27(40)] significantly improved Ca²⁺ burst responses.

Fig. 4.

Connexin isoform specificity on ATP-stimulated Ca²⁺ burst inhibition using GAP peptides in HUVEC. Bar graph shows the effect of 2 h of pretreatment with 300 μM GAP peptide on subsequent ATP-stimulated Ca²⁺ bursts. Cells were treated initially with ATP for 30 min, and bursts were counted. Cells were then washed and incubated with GAP peptides before subsequent stimulation of the same cells for a second time with ATP. Ca²⁺ bursts were again counted. Initial ATP-stimulated Ca²⁺ burst numbers for cells giving 3+ bursts were taken as a mean and set to 100% in black bars. Any change in mean Ca²⁺ bursts after treatment was plotted in white bars. For Control, n = 66 cells; for Gap26(37,40), n = 149 cells; for Gap27(37,43), n = 119 cells; for Gap27(37,43) Scr, n = 147 cells; for Gap27(40), n = 101 cells; for Gap26(43), n = 116 cells. Data are means +SE. Statistics were performed on raw data. Significant differences with and without the corresponding Gap peptide are shown by *Inhibition vs. ⁺Control stimulation (P < 0.05).

VEGF₁₆₅ Inhibits Sustained Ca²⁺ Burst Responses in HUVEC

Pretreatment of HUVEC with 10 ng/ml VEGF₁₆₅ for 30 min has an inhibitory effect on sustained ATP-stimulated Ca²⁺ bursts. Those cells, which produced three or more Ca²⁺ bursts upon initial ATP stimulation, were included in the analysis because the mechanistic switch between initial and sustained Ca²⁺ bursts has been fully realized by the third burst. Cells pretreated with VEGF₁₆₅ produced only 55% (45% reduction in Ca²⁺ bursts) of initial control Ca²⁺ bursts upon subsequent ATP treatment (Fig. 5).

Fig. 5.

Effect of VEGF₁₆₅ pretreatment on ATP-stimulated Ca²⁺ bursts in HUVEC. Cells were initially treated with ATP (100 μM) for 30 min, and Ca²⁺ bursts were counted. After being washed and pretreated with Krebs buffer (Control) or VEGF₁₆₅ for 30 min, the same cells were again treated with ATP (100 μM), and Ca²⁺ bursts were counted. Initial Ca²⁺ burst numbers for those cells that gave three or more bursts was normalized to 100%. For Control, n = 143; for VEGF₁₆₅, n = 112. Results are shown as means +SE. *Significant difference (P < 0.001).

VEGF₁₆₅ Inhibits Sustained Ca²⁺ Burst Response in HUV Endo

When pretreated with 10 ng/ml VEGF₁₆₅ for 30 min, 100 μM ATP-stimulated HUV Endo show fewer sustained Ca²⁺ bursts than those exposed to ATP alone. Mean tracings for ATP and VEGF₁₆₅ + ATP are shown in Fig. 6, A and B, and quantification of the data is shown in Fig. 6C. VEGF₁₆₅ pretreatment has no effect on initial ATP-stimulated Ca²⁺ peak height, but significant decreases of sustained Ca²⁺ and NO are apparent. At 900 s, ATP-stimulated Ca²⁺ responses remain ~40 nM above basal, whereas VEGF₁₆₅ + ATP have returned to basal levels. DAF signal as a measure of NO production is at 0.25-fold above basal at 1,500 s in ATP-stimulated vessels, whereas VEGF₁₆₅ + ATP-treated vessels fall short of 0.15-fold change above basal. Figure 6D shows a significant decrease in Ca²⁺ burst frequency by the second burst and is maintained through the third and fourth bursts.

Fig. 6.

Effect of VEGF₁₆₅ pretreatment on ATP-stimulated Ca²⁺ and NO responses in HUV Endo. HUV Endo were loaded with Fura-2 and DAF-2 to simultaneously record changes in [Ca²⁺]_i and NO. A: a 30-min recording is shown with mean levels of [Ca²⁺]_i and NO for 100 μM ATP treatment. B: the vessels were pretreated with 10 ng/ml VEGF₁₆₅ for 30 min before subsequent 100 μM ATP addition and recording for 30 min. C: the quantification of the data from A and B for peak and sustained (900 s) [Ca²⁺]_i and total NO (at 1,500 s). Significant reductions after VEGF₁₆₅ pretreatment occur at sustained [Ca²⁺]_i and total NO. D: the proportion of cells yielding each successive Ca²⁺ burst (peaks) are plotted. For ATP, n = 6 vessels; for VEGF₁₆₅ + ATP, n = 6 vessels. Results are shown as means +SE. *Significant difference (P < 0.05).

DISCUSSION

Regulation of vascular tone during pregnancy is critical for the health of both the mother and the developing fetus. A key regulator of vascular tone is vasodilator production in the vascular endothelium. Endothelial vasodilator production is controlled at many levels of cell signaling, but perhaps the most important and best understood is Ca²⁺ signaling driving NO production. Many circulating factors promote increases in [Ca²⁺]_i, so examining the relative contributions of these factors is critical to understanding how they might support or impair pregnancy-related vasodilator production.

VEGF₁₆₅ Stimulation of Ca²⁺ Responses

Although both ATP and VEGF₁₆₅ promote increases in [Ca²⁺]_i, they do so in different ways. Ultimately, this results in differing capacities for NO production. Both promote Ca²⁺ responses in a large number of cells exposed to each agonist, but the periodic and transient burst pattern typical of ATP-stimulated cells is less apparent in VEGF₁₆₅-stimulated cells at the maximal dose of VEGF₁₆₅ used. Instead, we observed a short initial Ca²⁺ peak followed by a sustained elevation of [Ca²⁺]_i. No portion of the VEGF₁₆₅-stimulated Ca²⁺ response appears to be gap junction dependent, since pretreatment with TPA had no effect on any portion of this response in either HUVEC or HUV Endo. In HUV Endo, there was also no effect of TPA on corresponding NO production. This is consistent with VEGF₁₆₅ signaling coupling to PLCgamma and Ca²⁺ entry occurring via TRPC6 (5, 30). It should also be noted that, in our ovine model, PlGF was unable to reliably stimulate any Ca²⁺ response and therefore was not a focus of this study, although others have reported NO-dependent vasorelaxation in rat and human vessels when treated with PlGF (29). Further study into the role of PlGF in human endothelial cell Ca²⁺ responses may be warranted, but Cunningham et al. (12) failed to show any stimulation of Ca²⁺ responses in HUVEC at a very high dose of PlGF.

According to Tran et al. (35), the ability of endothelial NO synthase (eNOS) to produce NO is determined by a combination of eNOS phosphorylation state and [Ca²⁺]_i elevation. Availability of substrate and cofactors such as l-arginine and tetrahydrobiopterin can also impact eNOS activity, but in our cell models and culture conditions, this has been shown not to be an issue (Refs. 9, 34 for UAEC; I. M. Bird, unpublished observations for HUVEC). Tran et al. showed that, in the case of eNOS failing to achieve optimal phosphorylation, a [Ca²⁺]_i of 200 nM or higher is necessary for any significant degree of eNOS activation. In the event of optimal eNOS phosphorylation, only 100 nM [Ca²⁺]_i is necessary to achieve 30% eNOS activation. We have previously shown in an ovine model that VEGF₁₆₅ stimulates both s617 and s1179 phosphorylation on eNOS (16). Both VEGF₁₆₅-stimulated phosphorylation events yielded only slight increases (even if statistically significant) over basal conditions, and therefore the majority of eNOS protein in any given cell would be unlikely to have both phosphorylation events occur simultaneously. This is in contrast to the effect of ATP stimulation of eNOS phosphorylation at s617 and s1179 in the same ovine model, where both residues are strongly phosphorylated after ATP treatment (34). Given that VEGF₁₆₅ typically fails to stimulate [Ca²⁺]_i above 200 nM in HUVEC, eNOS activation and therefore NO production will predictably be weak. This contrasts with ATP-stimulated Ca²⁺ signaling, where [Ca²⁺]_i tends to remain elevated for only short periods of time as a series of bursts but easily crosses into the range of eNOS activation and does so when optimum phosphorylation is also achieved.

When the corresponding experiments are performed using intact HUV Endo such that NO production can be directly imaged with DAF-2, it is clear that this relationship holds true, and NO output is effected in parallel. As reported by Krupp et al. (22), HUV Endo Ca²⁺ signaling is optimized to sustain ATP-stimulated elevations in [Ca²⁺]_i and maintain strong NO output, and this is lacking in PE subjects. Here, we show that, in comparison, VEGF₁₆₅-stimulated Ca²⁺ signaling alone fails to sustain elevated [Ca²⁺]_i in vessels when compared with the primary cells in culture at passage 4, and the resulting NO production is significantly less than for ATP-stimulated vessels. The period of time that corresponds to the greatest NO production in response to either agonist is when the mean ATP response is >150 nM [Ca²⁺]_i, consistent with observations in our laboratory's ovine model (41, 42).

Role of Cx43 in Ca²⁺ Signaling

Apart from direct stimulation of Ca²⁺ signaling and NO production, VEGF₁₆₅ also has been shown to activate signaling pathways (Src and MEK/ERK), which result in inhibitory phosphorylation of Cx43 (23, 38), and this has been confirmed in our ovine model (6). To that end, we first investigated the role of gap-junction subtypes in supporting ATP-stimulated Ca²⁺ bursting in our HUVEC model. Although confirmation that ATP-stimulated Ca²⁺ bursts in HUVEC are gap-junction-dependent is implied by TPA-mediated loss of Ca²⁺ bursts, it also appears to depend on a mix of gap-junction isoforms. By using GAP peptides designed to compete with the binding regions of the extracellular loop domains between adjacent cells, the effectiveness of multiple peptides gives us reason to believe Cx43 remains important, but Cx37 and Cx40 may also be involved in Ca²⁺ signaling in some capacity, perhaps through heteromeric or heterotypic gap junctions (10, 11, 21). Van Rijen et al. have previously shown that all three isoforms are expressed to some degree in cultured HUVEC (37), but electrical communication is predominantly mediated by Cx43. Certainly, the effectiveness of Gap26(43) to inhibit sustained ATP-stimulated Ca²⁺ bursts supports this conclusion. Surprisingly, Gap27(37,43) did not inhibit ATP-stimulated Ca²⁺ bursts, in contrast to the strong inhibition observed in the ovine model. Nonetheless, in the ovine model, Cx43 is the exclusive gap-junction peptide mediating sustained ATP-stimulated Ca²⁺ bursts, likely as a homodimer.

A key role for Cx43 in maintaining cell-cell communication would suggest VEGF₁₆₅ will be able to inhibit Ca²⁺ bursting on subsequent ATP stimulation. VEGF₁₆₅ treatment has been shown to phosphorylate Cx43 on multiple inhibitory residues by activation of MEK/ERK and Src kinase pathways in ovine uterine artery endothelial cells (6). Kinase regulation of the additional connexin isoforms present in HUVEC are still poorly understood compared with Cx43. It was thus unclear what effect VEGF₁₆₅ may have on ATP-stimulated Ca²⁺ bursts on the mixed connexin isoform environment of HUVECs. Since VEGF₁₆₅ pretreatment in HUVEC and HUV Endo did reduce ATP-stimulated Ca²⁺ bursts to a degree comparable to that observed in the ovine model (6, 41), our findings are consistent with Cx43 as the dominant isoform or, alternatively, imply kinase pathways stimulated by VEGF₁₆₅ are similarly capable of inhibiting Cx37 and Cx40, despite our lack of understanding of these proteins. Given the report from van Rijen et al. (37), it would again appear that Cx43 is most likely playing a leading role in determining cell-cell communication, even if involved in heterotypic coupling with Cx37 and Cx40. Morchauser et al. (28) showed that, in ovine uterine arteries, like in the corresponding ovine primary UAEC (40), Cx43 is the mediator of sustained Ca²⁺ signaling and NO production.

VEGF₁₆₅ Inhibition of ATP-Stimulated Ca²⁺ Responses

Yi et al. (41) observed that VEGF₁₆₅ also inhibits ATP-stimulated Ca²⁺ bursts and NO production in the ovine uterine arteries. It therefore stands to reason that if VEGF₁₆₅ acts through kinase signaling pathways to inhibit Cx43 function and Cx43 function is critical to ATP-stimulated Ca²⁺ bursts and NO production in HUV Endo, then VEGF₁₆₅ will similarly block these Ca²⁺ and NO responses in HUV Endo. Our study now shows this is indeed the case. It is also important to consider that the degree to which VEGF₁₆₅ inhibits ATP-stimulated Ca²⁺ and NO responses is greater than the degree to which VEGF₁₆₅ stimulates Ca²⁺ and NO responses in its own right. The net result of VEGF₁₆₅ presence locally in the vasculature at the doses tested would be a negative contribution toward vasodilation. These are, of course, prototypical agonists at prototypical doses used to tease out molecular mechanisms, but it should be noted that neither of these doses are beyond that which can be seen in a physiological or pathophysiological context (7, 8, 15, 19, 20, 36). Especially when the case of PE is considered, doses of 10 ng/ml total VEGF₁₆₅ have been associated with the circulating range in the diseased state (although circulating sFLT-1 consistently reduces systemic bioavailability) and are likely to be easily achieved locally in tissues such as decidua. The data presented here also make the case that, since high circulating levels of VEGF₁₆₅ can be deleterious to endothelial cell function (possibly even promoting endothelial dysfunction), elevated circulating sFLT-1 in PE may indeed be more of a response to high local VEGF concentrations rather than a cause of endothelial dysfunction through reduced systemic VEGF bioavailability but does not discount any further role reduced VEGF bioavailability may have on endothelial cell function. This is an important consideration, because even a local disruption in uterine perfusion, especially in the decidua, may instigate downstream hypoxia and upstream hypertension.

Ultimately, to translate these observations to therapy for hypertension associated with PE, we must either better understand how to adjust VEGF to levels more in line with normal pregnancies or better understand the signaling mechanisms downstream of VEGFR to allow us to target them pharmacologically. In the ovine cell culture model, we have shown that the specific isomer 10,12 CLA can target Src kinase signaling pathways to restore ATP-stimulated sustained Ca²⁺ signaling after VEGF₁₆₅ pretreatment (6). However, we still lack sufficient knowledge in the human condition to know whether it will be effective in ameliorating hypertension associated with excessive VEGF in preeclampsia. Future studies will be needed to address that point. This study provides critical insight into subtle differences between ovine and human models, but, more importantly, it shows that the key regulators remain in place and behave in a similar fashion, such that therapies targeted to signaling pathways impacting on Cx43 function first elucidated in ovine will likely translate to human models and hopefully provide an option for clinical care for preeclampsia that target the endothelium directly.

GRANTS

This work was funded by an R&D Grant award from the Department of Obstetrics and Gynecology, University of Wisconsin Madison School of Medicine and Public Health and by National Institute of Child Health and Human Development Grants R21 HD-069181, P01 HD-38843, and R03 HD-079865.

DISCLOSURES

D.S.B. and I.M.B. have a patent pending for the therapeutic use of 10,12 CLA in preeclampsia.

AUTHOR CONTRIBUTIONS

D.S.B., J.K., F.-X.Y., and N.K. performed experiments; D.S.B., J.K., F.-X.Y., and N.K. analyzed data; D.S.B., J.K., F.-X.Y., D.M.S., and I.M.B. interpreted results of experiments; D.S.B. and F.-X.Y. prepared figures; D.S.B. drafted manuscript; D.S.B., J.K., D.M.S., and I.M.B. edited and revised manuscript; D.S.B., J.K., F.-X.Y., N.K., D.M.S., and I.M.B. approved final version of manuscript.