Insulin Induces Relaxation and Decreases Hydrogen Peroxide-Induced Vasoconstriction in Human Placental Vascular Bed in a Mechanism Mediated by Calcium-Activated Potassium Channels and L-Arginine/Nitric Oxide Pathways

HIGHLIGHTS Short-term incubation with insulin increases the L-arginine transport in HUVECs. Short-term incubation with insulin increases the NO synthesis in HUVECs. Insulin induces relaxation in human placental vascular bed. Insulin attenuates the constriction induced by hydrogen peroxide in human placenta. The relaxation induced by insulin is dependent on BKCa channels activity in human placenta. Insulin induces relaxation in umbilical veins, increasing the expression of human amino acid transporter 1 (hCAT-1) and nitric oxide synthesis (NO) in human umbilical vein endothelial cells (HUVECs). Short-term effects of insulin on vasculature have been reported in healthy subjects and cell cultures; however, its mechanisms remain unknown. The aim of this study was to characterize the effect of acute incubation with insulin on the regulation of vascular tone of placental vasculature. HUVECs and chorionic vein rings were isolated from normal pregnancies. The effect of insulin on NO synthesis, L-arginine transport, and hCAT-1 abundance was measured in HUVECs. Isometric tension induced by U46619 (thromboxane A2 analog) or hydrogen peroxide (H2O2) were measured in vessels previously incubated 30 min with insulin and/or the following pharmacological inhibitors: tetraethylammonium (KCa channels), iberiotoxin (BKCa channels), genistein (tyrosine kinases), and wortmannin (phosphatidylinositol 3-kinase). Insulin increases L-arginine transport and NO synthesis in HUVECs. In the placenta, this hormone caused relaxation of the chorionic vein, and reduced perfusion pressure in placental cotyledons. In vessels pre-incubated with insulin, the constriction evoked by H2O2 and U46619 was attenuated and the effect on H2O2-induced constriction was blocked with tetraethylammonium and iberiotoxin, but not with genistein, or wortmannin. Insulin rapidly dilates the placental vasculature through a mechanism involving activity of BKCa channels and L-arginine/NO pathway in endothelial cells. This phenomenon is related to quick increases of hCAT-1 abundance and higher capacity of endothelial cells to take up L-arginine and generate NO.


INTRODUCTION
In the human placenta, an organ lacking innervation, endothelial factors released in response to shear stress, oxygen level, paracrine, or endocrine signals, are the main mechanisms that maintain low resistance and blood flow in the placental vascular bed (Wareing, 2014). In this regard, insulin induces relaxation in human umbilical veins via a mechanism involving increased transcriptional activity of SLC7A1 (coding for human Cationic Amino Acid Transporter-1, hCAT-1) and L-arginine transport (González et al., 2011). In addition, insulin also increases nitric oxide (NO) synthesis through activation of phosphatidylinositol 3-kinase (PI3K) and endothelial NO synthase (eNOS) in HUVECs (González et al., 2004). We reported that insulin may also generate hyperpolarization in this cell type (González et al., 2004), which enhances vasomotor activity of this hormone in the placental macrocirculation and microcirculation. Despite this evidence, detailed mechanisms underlying the vasomotor activity of insulin are still unclear.
Other vascular tone regulators are reactive oxygen species (ROS) and nitrogen species (RNS), which constitute a family of radical and non-radical derivatives of molecular oxygen (O 2 ) and nitrogen (N 2 ), respectively (Klandorf and Van Dyke, 2012). Specifically, hydrogen peroxide (H 2 O 2 ) and peroxynitrite (ONOO − ) induce rapid and transient contraction or relaxation in human placental chorionic plate arteries (Mills et al., 2009). In addition, H 2 O 2 and ONOO − have higher stability than their respective precursors (Beckman and Koppenol, 1996), and may cause deleterious effects in vascular beds when insufficiently buffered/neutralized (González et al., 2011(González et al., , 2015. In this regard, several reports have shown that ROS reduces NO availability and vascular relaxation in human placenta vasculature, during either healthy or pathological conditions such as preeclampsia (PE; Bernardi et al., 2008;Catarino et al., 2012), intrauterine growth restriction (IUGR; Takagi et al., 2004), and gestational diabetes mellitus (GDM; Coughlan et al., 2004).
However, it is unknown whether H 2 O 2 might control vascular tone in the chorionic plate veins or in the placental microcirculation. It is also unknown whether insulin can regulate the vascular tone in these placental vessels. Therefore, the present study aimed to determine whether insulin attenuates the vascular response induced by H 2 O 2 in the placental vasculature, and elucidate whether BKCa channel activity, endothelial expression, and activity of hCAT-1 are involved in this process.

Ethics Statement
This investigation conforms to the principles outlined in the Declaration of Helsinki, and has received approval from the Ethics Committee of the Faculty of Biological Sciences of Universidad de Concepción, the Hospital Regional Guillermo Grant Benavente, Concepción Chile, and National Research Ethics System (NRES ref; 08/H1010/55), UK, and the Comisión Nacional de Investigación en Ciencia y Tecnología (CONICYT grant number 11100192, Chile). All women signed written informed consent. The maternal and newborns clinical parameters are summarized in Table 1.

Human Placenta and Umbilical Cords Collection
Placentas with their umbilical cords were collected after delivery from 75 full-term normal pregnancies from the Hospital Regional Guillermo Grant Benavente in Concepción (Chile) and St. Mary's Hospital in Manchester (UK). All pregnancies were single births. The pregnant women did not smoke or consume drugs or alcohol, had no intrauterine infection or any medical or obstetrical complications, were normotensive and exhibited a normal response to the oral glucose tolerance test. They were under a normal food regimen during the whole pregnancy period and newborns were at term, born by vaginal delivery or cesarean section. Placentas were transferred in a sterile container (4 • C) to the laboratory. Sections of umbilical cords (10-20 cm length) were collected into sterile 200 ml phosphate-buffered saline (PBS) solution [(mM): 130 NaCl, 2.7 KCl, 0.8 Na 2 HPO 4 , 1.4 KH 2 PO 4 (pH 7.4, 4 • C)] and used for isolation of umbilical vein endothelial cells (HUVECs) between 6-12 h after delivery.

Cell Culture
HUVECs were isolated by collagenase digestion (0.25 mg/ml Collagenase Type I from Clostridium histolyticum; Gibco Life Technologies, Grand Island, NY, USA) as previously described (González et al., 2004). In brief, cells were cultured (37 • C, 5% CO 2 ) up to passage 3 in medium 199 (M199) (Gibco Life Technologies, Grand Island, NY, USA) containing 5 mM D-glucose, 10% newborn calf serum (NBCS), 10% fetal calf serum (FCS), 3.2 mM L-glutamine, and 100 U/ml penicillinstreptomycin (primary culture medium, PCM). Experiments were performed on cells incubated (0-30 min) in M199 in the absence or presence of insulin (1 nM). Cell viability estimated by LGA 10 (13%) Women with normal pregnancies (n = 75) were included in the study (see Methods). The maternal parameters were registered before delivery. OGTT, oral glucose tolerance test; SGA, small for gestational age; AGA, appropriate for gestational age; LGA, large for gestational age. All the values are mean ± SD, unless otherwise stated.
Trypan blue exclusion was higher than 97% (not shown). Sixteen hours prior, the experimental incubation medium was changed to sera-free M199 (González et al., 2015).

DAF Fluorescence
HUVECs were grown on microscope coverslips and intracellular NO was determined in cells incubated with insulin (1 nM, 1-30 min) and exposed (45 min, 37 • C) to 10 µM of 4-amino-5-methylamino-2 ′ ,7 ′ -difluorofluorescein (DAF-FM) (Molecular Probes, Leiden, The Netherlands). The fluorescence was observed in fixed cells by fluorescence microscopy (Olympus IX81) and the signal density was analyzed by Image J software (Java-based imaging processing program, National Institute of Health, USA).

Isolated Cotyledon Perfusion
Techniques for perfusion of the placental cotyledon was that of Penfold et al. (1981), which was modified by perfusing only the fetal vascular compartment, instead of the dual perfusion model (Acevedo et al., 1995). After delivery (15-30 min), a fetal vein and artery pair on surface of chorionic plate, leading to peripheral cotyledon, was cannulated with plastic tubing. Each cotyledon was perfused with Krebs-Ringer solution at a constant flow rate (7 mL/min), maintained with oxygen levels similar to physiological conditions for placental vessels in situ. The perfusion pressure was continuously monitored and the viability of preparation was controlled as previously described (Acevedo et al., 1999).

Wire Myography
Chorionic plate veins, identified as branches of the umbilical vein, were dissected from biopsies and placed in an ice-cold physiological saline solution (PSS). Veins were mounted on a myograph (610 M; Danish Myotechnology, Aarhus, Denmark) and normalized to 0.9 L 5.1kPa as described (Mills et al., 2009). Vessels were bathed in PSS and maintained with oxygen levels similar to physiological conditions for chorionic vessels in situ (Mills et al., 2009). After the vein rings were stabilized for isometric force measurements with optimal diameter (∼310 µm), the maximal active response was determined with modified PSS containing 90 mM KCl. Two different protocols were designed to determine the effects of insulin consisting of incubation of pre-constricted (U46619) veins with the hormone (in the presence of inhibitors); or preincubation (30 min) with insulin (in the presence of inhibitors) prior to U46619 (10 −10 -10 −5 M) or H 2 O 2 (10 −5 -10 −3 M) exposure.

Statistical Analysis
Values are mean ± S.E.M., where n indicates the number of different cell cultures (three to four replicates). Comparisons between two or more groups were performed by means of Student's unpaired t-test and analysis of variance (ANOVA), respectively. If the ANOVA demonstrated a significant interaction between variables, post-hoc analyses were performed by the multiple-comparison Bonferroni correction test. Values of p < 0.05 were considered statistically significant.

Insulin Increases the L-Arginine/NO Pathway in HUVECs
Apparent K m was maintained in a range between 54 ± 16 to 153 ± 18 µM ( Figure 1A, Table 2) in all experimental conditions. Insulin increased the V max of L-arginine transport with maximal effect (5.4 ± 0.9-fold) after 30 min of treatment and lower but significant (p < 0.005) increases after 3 min (2.2 ± 0.8-fold) and 5 min (2 ± 0.5-fold) of incubation (Figures 1A,B, Table 2). No changes in V max were detected after 10 or 20 min of treatment with insulin. The V max /K m was significantly increased 2.6 ± 0.8 and 2.5 ± 0.9-fold after 5 min and 30 min incubation with 1 nM insulin, respectively ( Figure 1B, Table 2). Insulin (1 nM, 30 min) increased hCAT-1-associated fluorescence in permeabilized (1.8 ± 0.3-fold) and non-permeabilized (2.8 ± 0.3fold) cells, compared to control ( Figure 1C). The fold of increase induced by insulin in hCAT-1-associated fluorescence is 55% higher (2.8 vs. 1.8-fold) in non-permeabilized cells compared with permeabilized cells. From total fluorescence, in insulintreated cells the 78% (27.8 vs. 35.6 arbitrary units of fluorescence) correspond to hCAT-1 expression on cell surface, meanwhile this percentage decreased until 50% (10 vs. 20 arbitrary units of fluorescence) in control cells ( Figure 1D). Insulin also increased NO levels (Figure 2A) with maximal effect (2.5 ± 0.2-fold) ( Figure 2B) after 30 min of incubation. No changes were detected in control cells incubated by 30 min in medium without insulin (Figure 2A). The basal level of DAF fluorescence in HUVECs was largely detected in a vesicular-like form; however, in cells incubated with insulin, the fluorescence was diffused throughout the cytoplasm (Figure 2C).

DISCUSSION
The mechanism previously reported for vascular properties of insulin involves higher expression and activity of eNOS in HUVECs (Montagnani et al., 2002;González et al., 2004), human aortic coronary endothelial cells (HAECs; Federici et al., 2002) and bovine aortic endothelial cells (BAECs; Kuboki et al., 2000). The effect of long-term insulin incubation (several hours) on eNOS activity is known to be dependent on PI3K signaling (Montagnani et al., 2002;González et al., 2011). However, mechanisms underlying a rapid stimulation of vasodilatation induced by insulin (González et al., 2011) are still unclear, and have been associated with NO-mediated blood flow in healthy subjects (Scherrer et al., 1994;Steinberg et al., 1994;Lind et al., 2002). In this study, we demonstrated that insulin induces relaxation in pre-constricted vessels and attenuates constriction (when hormone is pre-incubated) in a mechanism mediated by BKCa channel activity and related to increased L-arginine transport and endothelial NO synthesis.

Mechanism of Relaxation Induced by Insulin
Insulin caused 42% relaxation in pre-constricted human umbilical vein (González et al., 2011). The EC 50 of insulin was 1.8 ± 0.2 nM and the effect was abolished by preincubation with Nethylmaleimide (NEM) and L-lysine, both competitive inhibitors of transport system y + for cationic amino acid (González et al., 2011). Now, we confirmed that insulin induced relaxation in chorionic plate veins and in fetal-side of placental vascular bed, decreasing the vasoconstriction induced by U46619 and H 2 O 2 , stimuli that had previously been shown to have vasoconstrictor effects in chorionic plate arteries (Beckman and Koppenol, 1996;Hayward et al., 2013). Relaxation induced by insulin in placental vessels was related to L-arginine/NO pathway in HUVECs. Previously, insulin has been noted to increase L-citrulline synthesis and L-arginine transport using concentrations between 0.1 and 10 nM in longterm (8 h) incubation assays, via a mechanism that involves high expression of hCAT-1 (Sobrevia et al., 1996;González et al., 2004). In isolated umbilical vein rings, insulin induced relaxation after 30 min of incubation, an effect disrupted by co-incubation with L-lysine, NEM or by endothelial denudation (González et al., 2011). This study confirms the effect of insulin in human placenta, and also suggests that this phenomenon is related to high NO synthesis, L-arginine transport and higher abundance of hCAT-1 in HUVECs. Regarding cell surface expression of hCAT-1, it has been determined that activation of protein kinase C (PKC) reduces cell surface expression of hCAT-1 and L-arginine transport in Xenopus laevis oocytes and U373MG glioblastoma cells (Rotmann et al., 2004a,b), reinforcing the idea that a rapid mobilization of hCAT-1 from plasma membrane to cytoplasm reduces L-arginine transport. Moreover, our work suggests that high V max of L-arginine transport mediated by insulin is associated with mobilization of hCAT-1 from cytoplasm to cell surface in HUVECs considering that insulin treatment results in a larger percentage of the total fluorescence from hCAT-1 expression on the surface of the cells. Further studies are necessary to determine the functional co-localization of hCAT-1 with relevant proteins for endothelial cell function like caveolin 1 or eNOS, as has been shown in porcine aortic endothelial cells (PAECs) (McDonald et al., 1997) or baby hamster kidney (BHK) cells (Lu and Silver, 2000).
One of the main mechanisms triggered by insulin is related to activation of PI3K in the human endothelium (González et al., 2004;Muniyappa et al., 2008). However, our results showed that inhibition of PI3K induces vessel relaxation; which was potentiated by insulin. Similarly, genistein induced relaxation in an endothelium-independent mechanism in rat aorta pre-constricted by sodium fluoride (NaF). In addition, when vessels were pre-treated with genistein, there was a decrease in constriction induced by U46619 (Je and Sohn, 2009) and an increase of eNOS phosphorylation in serine 1179 after 10-30 min of incubation via a protein kinase A (PKA)-dependent mechanism in BAEC (Liu et al., 2004). In the endothelial cell   line ECV-304, the pre-incubation with genistein improved cell viability and reversed apoptosis induced by H 2 O 2 , which was associated with enhanced antioxidant capacity (Jin et al., 2015). On the other hand, the activity of protein kinase B/Akt plays a central role in the PI3K-dependent activation of eNOS catalyzing the phosphorylation of serine 1179 and serine 617, increasing the sensitivity of eNOS by Ca 2+ /calmodulin (CaM) complex (Tran et al., 2009). Previously, it has been shown that 100 nM FIGURE 5 | The effect of pre-incubation with insulin on H 2 O 2 constriction is dependent on BKCa channels. Chorionic vein rings were pre-incubated (30 min, 5%CO 2 , 37 • C) in absence (control) or presence of 10 nM insulin and/or 1 mM tetraethylammonium (TEA; A), 100 nM iberiotoxin (IbTx; B), or 30 nM wortmannin (C). After these treatments, vessels were exposed to hydrogen peroxide (H 2 O 2 ). In ( insulin induces the phosphorylation of PKB/Akt after 2 or 10 min of treatment in HUVEC, increasing the eNOS activity in a mechanism dependent on insulin receptor substrate 1 (IRS1; Federici et al., 2004). With these evidences accounted for, our results show that inhibition of tyrosine kinases induces relaxation in pre-constricted placental vessels and, more importantly, the relaxation induced by insulin could be independent of PI3K pathway. It will still be important to explore the effects of insulin in PKB/Akt activity and CaM-dependent activation of eNOS, especially if the insulin-induced relaxation is associated with activity of K + channels activated by Ca 2+ . In regard to this association, recently it has been shown that genistein, in combination with magnesium, induces relaxation in rat mesenteric arteries in a mechanism dependent on eNOS activity (blocked by L-NAME) and associated with high BKCa currents in rat mesenteric smooth muscle cells (Sun et al., 2015). These results suggest that genistein has a dual effect both in endothelial cells and VSMCs. In our experiments it is possible that the effects of genistein in the chorionic veins were not due to inhibition of protein kinases mainly, because we cannot discard the activation of BKCa by genistein directly in chorionic vein smooth muscle cells.

Role of Potassium Channels Activity
Using hippocampal neurons, O' Malley and Harvey (2004) showed that insulin (10 nM) increased (∼3.8-fold) the mean channel activity (NfP o ) of BKCa after 15 min incubation in the bath solution. Meanwhile, in similar experiments but via patch pipette solution, the effect was faster (2-8 min postinsulin). Authors also showed that insulin increased the mean open time (τ o ) of BKCa from ∼0.76 ms (at 2-4 min) to ∼2.01 ms (at 15-17 min). Similar effects were observed in HEK293 cells expressing hSlo (pore-forming α subunit of BKCa channel) and direct activation of BKCa channels through application of its selective channel opener, NS-1619 (O'Malley andHarvey, 2004). Similar to our results, this last study showed that the effect of insulin on BKCa activity is mediated by a mechanism independent of PI3K (but dependent of Ras/Raf/MEK/ERK pathway). Therefore, our study shows that the effects of insulin on placental vasculature are independent of PI3K activity, but mainly dependent of BKCa activity. We acknowledge that further experiments are required in order to elucidate underling intracellular pathway linked with this effect of insulin on BKCa channels, including potential participation of MAPK pathway.  (Wareing et al., 2006). More recently, in chorionic plate artery smooth muscle cells (CPASMCs), the incubation with TEA and iberiotoxin demonstrated that BKCa are the main channels responsible for outward currents in CPASMCs (Brereton et al., 2013). Importantly, immunohistochemistry assays in placental tissue, have demonstrated a strong expression of BKCa in endothelium with similar localization of endothelial cell marker CD31 (Sand et al., 2006). In our study, we cannot discern if BKCa expression/activity is more important in endothelium or VSMCs, but the relaxation induced by TEA and iberiotoxin reveals a role of BKCa in the vascular tone regulation of placenta. In rabbit basilar artery, similar pre-incubation (30 min) with iberiotoxin and TEA reduced the vascular tone induced by sodium acetate (Cho et al., 2007). This result is similar to our finding about the reduction of vascular tone induced by H 2 O 2 ; in both cases the mechanism of constriction is not via a receptor-mediated signaling pathway. The mechanism of vascular tone modulation by H 2 O 2 in placenta is still unclear, but previous report of Mills showed a transient constriction induced by H 2 O 2 in chorionic plate arteries, reversed by catalase (Mills et al., 2009). In umbilical artery, similar concentration (10-100 µM) of H 2 O 2 enhanced the tension induced by prostanglandin F2α without change in sensitivity to calcium chloride (Okatani et al., 1997 -induced constriction in placental vessels. At moment, we still cannot explain with direct evidence the effect of TEA or iberiotoxin on H 2 O 2 -induced constriction, but we speculate that the alteration in vascular homeostasis after 30 min of BKCa inhibition could reduce the constriction associated with oxidative stress. Although in our study we did not determine the changes in plasma membrane potential, is possible that a change of resting potential (both in endothelium and VSMCs) induced by iberiotoxin alters the vascular response to oxidative stress.
Related to the connection between the ex vivo and in vitro results, previously it has been demonstrated that HUVECs express Ca 2+ -activated potassium currents blocked by iberiotoxin (Wiecha et al., 1998;Watanapa et al., 2012). More importantly, 0.6 nM (100 µUI/ml) insulin increased the openstate probability (NPo) of BKCa after 3 min incubation in this cell type (Wiecha et al., 1998), in a similar fashion that 10 nM insulin increased the activity of BKCa in O'Malley and Harvey study. A different stimulus, quercetin, induces hyperpolarization, high concentration of intracellular Ca 2+ , cGMP synthesis and reduction of proliferation of HUVECs, and each of these effects are blocked when the cells are incubated with iberiotoxin (Kuhlmann et al., 2005). In addition, HUVECs treated with plasma samples from preeclamptic pregnancies exhibit a higher fraction of cells expressing outward currents associated with KCa channels, showing a compensatory mechanism attributed to some factors secreted in preeclampsia (Watanapa et al., 2012). These evidences allow us to propose that the insulin signaling induces the activation of KCa channels, changing the plasma membrane polarity for activation of hCAT-1 reflected in higher V max for L-arginine transport and, finally, higher NO synthesis. In this regard, a study published by Kavanaugh showed that the influx of L-arginine is increased by membrane hyperpolarization in Xenopus laevis oocytes expressing CAT-1 (Kavanaugh, 1993). Oppositely, the depolarization induced by increased extracellular concentration of K + reduced the Larginine transport in HUVECs (Sobrevia et al., 1995). Also in Xenopus laevis oocytes expressing hCAT-1, the incubation (6 h) with high concentration of K + reduced the intracellular accumulation of L-[ 3 H]arginine (Rotmann et al., 2004a). These findings show the dependency between hCAT-1 activity and voltage of the plasma membrane, but further studies are necessary to establish direct evidence of regulation of hCAT-1 activity through membrane hyperpolarization in placental endothelial cells.
Also, it is important to note that previously it has been shown that insulin evokes hyperpolarization (from −65.5 ± 0.4 to −82.3 ± 0.4 mV) and high intracellular concentration of Ca 2+ (from 40 ± 3 nM to 372 ± 29 nM) in HUVECs (González et al., 2004). Although the mechanism of intracellular Ca 2+ regulation by insulin in endothelium still is not clear, other agonists or stimuli that activate eNOS have been studied. VEGF-A and shear stress increase the activity of eNOS and NO synthesis in a mechanism dependent of enhancement of Ca 2+ (Devika and Jaffar Ali, 2013). Importantly, Anaya et al., showed that the Ca 2+ mobilization and NO availability (detected with DAF) were reduced in intact endothelium from umbilical vein isolated of GDM samples, without changes in the protein abundance of eNOS. Additionally, in HUVECs isolated from GDM, the response of Ca 2+ mobilization to ATP is lower than control cells (Anaya et al., 2015). From these data, we suggest that the effects of insulin on L-arginine/NO pathway in HUVECs and BKCadependent relaxation in chorionic vein, reported in this study, could be significantly altered in GDM due to reduced capacity of endothelial cells for increased Ca 2+ mobilization in response to some agonists.
In conclusion, we found that insulin induces a rapid relaxation in placental vascular bed through a mechanism associated with high activity of BKCa channels and L-arginine/NO pathways in endothelial cells.