Gestational Diabetes Reduces Adenosine Transport in Human Placental Microvascular Endothelium, an Effect Reversed by Insulin

Gestational diabetes mellitus (GDM) courses with increased fetal plasma adenosine concentration and reduced adenosine transport in placental macrovascular endothelium. Since insulin modulates human equilibrative nucleoside transporters (hENTs) expression/activity, we hypothesize that GDM will alter hENT2-mediated transport in human placental microvascular endothelium (hPMEC), and that insulin will restore GDM to a normal phenotype involving insulin receptors A (IR-A) and B (IR-B). GDM effect on hENTs expression and transport activity, and IR-A/IR-B expression and associated cell signalling cascades (p42/44 mitogen-activated protein kinases (p42/44mapk) and Akt) role in hPMEC primary cultures was assayed. GDM associates with elevated umbilical whole and vein, but not arteries blood adenosine, and reduced hENTs adenosine transport and expression. IR-A/IR-B mRNA expression and p42/44mapk/Akt ratios (‘metabolic phenotype’) were lower in GDM. Insulin reversed GDM-reduced hENT2 expression/activity, IR-A/IR-B mRNA expression and p42/44mapk/Akt ratios to normal pregnancies (‘mitogenic phenotype’). It is suggested that insulin effects required IR-A and IR-B expression leading to differential modulation of signalling pathways restoring GDM-metabolic to a normal-mitogenic like phenotype. Insulin could be acting as protecting factor for placental microvascular endothelial dysfunction in GDM.

Insulin activates plasma membrane insulin receptor (IR) isoforms A (IR-A) and B (IR-B) [5,11]. These transcripts relative abundance is tissue-specific [11], suggesting that IR-A and IR-B functional differences might underlie tissue-specific insulin effect in vivo. IR-A is preferentially expressed in HUVEC from GDM [4], complementing similar information and increased IR-B mRNA expression in patients with diseases characterized by insulin resistance such as type 2 diabetes mellitus (T2DM) [12,13] or myotonic dystrophy [14]. Increased mitogen-activated protein kinases 1/2 (p42/44 mapk )/protein kinase B (Akt) ratio, i.e., higher mitogenic/metabolic-like signalling ratio [15,16], is characteristic of higher IR-A/IR-B ratios. Since GDM is associated with umbilical vein blood hyperinsulinemia [17,18], we hypothesize that GDM alters hENT2-mediated transport in hPMEC, and that a GDM-like phenotype will be restored to a normal-like phenotype by insulin involving IR-A or IR-B activation. Results show reduced hENT2-adenosine transport and expression, and SLC29A2 (for hENT2) promoter activity, and reduced IR-A/IR-B expression ratio paralleled by p42/44 mapk /Akt activation ratios in GDM, all phenomena reversed by insulin. These findings could be determinant in diseases of pregnancy associated with abnormal insulin signalling and endothelial dysfunction such as GDM [5,7].

Ethics statement
The investigation conforms to the principles outlined in the Declaration of Helsinki. Ethics Committee approval from the Faculty of Medicine of the Pontificia Universidad Católica de Chile, the Comisión Nacional de Investigación en Ciencia y Tecnología (CONICYT, Chile) and patient informed written consent were obtained.

Human placentas and study groups
Placentas were collected after delivery from 64 full-term normal or 64 full-term gestational diabetic pregnancies. Patients between the 24-28 weeks of gestation with basal glycaemia ,90 mg/dL (i.e., overnight starvation) and .140 mg/dL at 2 hours after an oral glucose load (75 g) were diagnosed as gestational diabetes mellitus (GDM) [4,6]. Patients with GDM were treated with diet (1500 kcal/day and 200 g of carbohydrates as maximum per day). All pregnancies were singleton and pregnant women were normotensive, non-smoking, non-alcohol or drug consuming, and without intrauterine infection or any other medical or obstetrical complications (Table 1). Patients with GDM exhibit increased maternal glycosilated hemoglobin A 1c , altered oral glucose tolerance test (OGTT), insulinemia, increased insulin resistance and reduced ß-cell function. Newborn from DGM exhibit increased insulin resistance and ponderal index compared with normal pregnancies.
A single Michaelis-Menten equation was used to obtain maximal velocity (V max ) and apparent Michaelis-Menten constant (K m ) of transport at initial rates (i.e., lineal uptake up to 20 seconds). Relative contribution of hENT1 and hENT2 ( hENT1/2 F) to total transport (i.e., hENT1 + hENT2 mediated) was defined by: where hENT1 V max and hENT1 K m are kinetic parameters for hENT1saturable transport, and hENT2 V max and hENT2 K m for hENT2saturable transport. Relative effect of GDM (GDM) compared with normal (N) pregnancies on transport activity via hENT1 (1/ N/GDM-hENT1 F) or hENT2 (1/ N/GDM-hENT2 F) was estimated from maximal transport capacity (V max /K m ) for transport by: where N-hENT1 V max , N-hENT2 V max , N-hENT1 K m and N-hENT2 K m are kinetic parameters for transport via hENT1 and hENT2, respectively, in normal pregnancies, and GDM-hENT1 V max , GDM-hENT2 V max , GDM-hENT1 K m and GDM-hENT2 K m for transport via hENT1 and hENT2, respectively, in GDM. Relative contribution of insulin to saturable transport kinetic parameters was estimated from V max /K m by: where X is hENT1 or hENT2-mediated transport, Y is normal or GDM pregnancy, C V max and C K m are transport kinetics parameters in control (basal-insulin), and Ins V max and Ins K m are in presence of insulin.
In experiments where sodium was replaced by N-methylglucamine-HCl (Sigma) or choline chloride (Sigma), adenosine transport was unaltered (not shown) as previously reported [1,4]. Cell viability was assayed by Trypan blue exclusion and was not significantly altered (,97% of viable cells) by addition of the molecules used in this study. Rinsing the monolayers with ice-cold Krebs containing 10 mmol/L NBTI and 2 mmol/L hypoxanthine terminated tracer uptake. Radioactivity in formic acid cell digests was determined by liquid scintillation counting, and uptake was corrected for [ 14 C or 3 H] mannitol (NEN) disintegrations per minute (d.p.m.) in the extracellular space [1,4]. Adenosine measurements by high-performance liquid chromatography (hplc) Adenosine concentration was measured in whole umbilical blood (vein + arteries), umbilical arteries or veins by hplc [1,4,20]. For collection of vein blood, arteries from placenta-attached umbilical cord were clamped and vein blood was collected. For artery blood collection, umbilical cord was double-clamped and detached from the placenta. One end of umbilical arteries was unclamped and blood drained out. Blood samples (3.5 mL) were collected into a syringe containing (250 mL) 10 mmol/L erythro-9-(2-hydroxy-3-nonyl) adenine (EHNA, adenosine deaminase inhibitor), 10 mmol/L NBTI and 1 mmol/L dilazep (inhibitors of adenosine transport), 2 mg/mL indomethacin (inhibitor of nucleotides release from platelets), and 40 mmol/L O,O'-bis(2-aminoethyl)ethyleneglycol-N,N,N'-N'-tetraacetic acid (G-EDTA, inhibitor of adenosine release from platelets) immediately after birth, as described [4,19]. Samples were centrifuged (14,000 g, 1 minute) and aliquots (1 mL) of plasma were deproteinated (100 mL, 50% trichloroacetic acid) and centrifuged again (5 minutes). Super-nantant (750 mL) was neutralized by addition of 100 mL of 3.3 N potassium hydroxide. Adenine nucleotides were extracted by adding 500 mL of 1 mol/l zinc sulfate and 1 mL of saturated barium hydroxide, and vortex mixed for 10 seconds, centrifugated (14,000 g, 5 minutes). Adenosine was finally converted to ethenoadenosine by mixing the sample with chloroacetaldehyde to a final concentration of 440 mmol/L and incubating it at 80uC for 1 hour. Derivatized samples were mixed and stored at 4uC until use 60-120 minutes later for hplc analysis.
For hplc analysis of samples, aliquots (200 mL) were collected and mixed with 10 mL of 0.5 mol/L acetate-buffer, 10 mL of 1 mmol/L internal standard (adenosine), and 10 mmol/L of 50% aqueous chloroacetaldehyde. After incubation (80uC, 1 hour) and centrifugation (14,000 g, 4 minutes), aliquots (80 mL) were injected into an Isco hplc system (pump model 2350, gradient programmer model 2360, 4.66250 mm C 18 reverse-phase column, 5-mm particle size) (Chemical Research Data Management System, Lincoln, NE, USA). Mobile phase was 10 mmol/L citrate-buffer with 4.5% acetonitrile and was run isocratically at 1 mL/minute. Fluorescence detection was achieved at an excitation wavelength of 275 nm and an emission wavelength of 420 nm using a Waters M-470 fluorescence detector. Ratio of the area under the adenosine peaks to the area under the internal standard peak was compared with a standard curve [1,4]. The concentration of adenosine was calculated from the peak area, using the standard line. Pearson's correlation coefficient for the standard line of standard adenosine solution was more than 0.999 from 2.997 to 431 nmol/L, and the recovery of plasma adenosine was 82.761.0% (n = 425). Plasma adenosine concentrations were calculated by dividing the amount of adenosine in the samples by the volume of plasma assayed, as described [4].

Reverse transcription and quantitative RT-PCR
Total RNA was isolated using the Quiagen RNAeasy kit (Quiagen, Crawley, UK). RNA quality and integrity were insured by gel visualization and spectrophotometric analysis (OD 260/280 ), quantified at 260 nm and precipitated to obtain 4 mg/mL. Aliquots (1 mg) of total RNA were reversed transcribed into cDNA as described [4,21].

hENT2 mRNA and protein levels half-life
Total RNA and protein from hPMEC exposed (0-24 hours) to culture medium without or with 1.5 mmol/L actinomycin D (transcription inhibitor) or 1 mmol/L cycloheximide (protein synthesis inhibitor) was measured as described [21]. hENT2 and 28S mRNA were amplified by qRT-PCR, and hENT2 and ßactin proteins detected by western blot.
The resulting sticky-end inserts were subcloned into pGL3-Basic

Luciferase assay
Electroporated cells were lysed in 200 mL passive lysis buffer (Promega), and Firefly and Renilla luciferase activity was measured using Dual-LuciferaseH Reporter Assay System (Promega) in a Sirius luminometer (Berthold Detection System; Oak Ridge, TN, USA) as described [4,21].  In addition, hENT2 protein abundance is reduced leading to lower removal (dotted arrow) of extracellular adenosine (Ado), thus leading to extracellular accumulation of adenosine. Altered IR-A and IR-B expression associates with preferential reduction of p42/44 mapk (p42/44) activation compared with Akt activation, respectively, favouring a metabolic-rather than a mitogenic-like phenotype. This phenomenon leads to reduced signalling (segmented light blue arrows) mediated by p42/44 and Akt, respectively, reducing promoter activity between 21491 and 2602 bp from ATG of SLC29A2 (for hENT2). Thus, lowered mRNA expression and hENT1 protein abundance could explain a reduced hENT2 availability at the plasma membrane to take up adenosine. When hPMEC cultures are exposed to insulin (Insulin), GDM-reduced IR-A and increased IR-B is reversed (orange arrows) to similar expression levels (+, increase; -, decrease), reaching comparable values to normal pregnancies under basal-insulin. Furthermore, IR-A and IR-B altered expression associates with restoration of p42/44 mapk and Akt activation, respectively, resulting in p42/44 mapk /Akt ,1 (i.e., similar contribution of p42/44 mapk and Akt signalling pathways), thus recovering the equilibrium between mitogenic and metabolic phenotype characteristic of hPMEC from normal pregnancies. These changes lead to restored signalling (solid light blue arrows) mediated by p42/44 and Akt, respectively, causing activation of SLC29A2 promoter activity and normal hENT2 mRNA expression and protein synthesis increasing this transporter availability at the plasma membrane. The latter restores hENT2-mediated transport of adenosine from the extracellular space reestablishing extracellular levels of this nucleoside. doi:10.1371/journal.pone.0040578.g006 Insulin receptor isoforms suppression with small interference RNA To suppress IR-A and IR-B expression an adenoviral-based siRNA delivering system (pSilencer TM adeno 1.0-CMV System Kit, Ambion) was used. Complementary oligonucleotides for IR-A (sense: 59-GTTTTCGTCCCCAGGCCATCTTTCAAGAGAA-GATGGCCTGGGGACGAAAAC-39, anti-sense: 59-CAAAAG-CAGGGGTCCGGTAGAAAGTTCTCTTCTACCG-GACCCCTGCTTTTG-39) and IR-B (sense: 59-GACCCTAGGCCATCTCGGAAATTCAAGAGATTTCCGA-GATGGCCTAGGGTC-3, anti-sense: CTGGGATCCGGTA-GAGCCTTTAAGTTCTCTAAAGGCTCTACCGGATCC-CAG-39) encoding for a siRNA hairpin targeting human IR (GenBank accession: NM_001079817.1 for IR-A and NM_000208.2 for IR-B) were designed, annealed and cloned into a pShuttle vector. Ligation products were amplified in E. Coli DH5a competent cells by bacterial transformation and plasmid constructs were confirmed by restriction enzyme digestion. The pShuttle-IRA/B siRNA and the adenoviral LacZ backbone were linearized following PacI digestion and were transfected into HEK-293 cells to generate the IR siRNA recombinant adenovirus(Ad-siIRA and Ad-siIRB). The negative control pShuttle vector (encoding a scramble siRNA sequence not found in human, mouse or rat genome databases) was used to generate the negative control adenovirus. A positive siGAPDH provided by the kit was also used to generate an Ad-siGAPDH. Recombinant adenoviral vectors were expanded by serial infection of HEK-293 cells, harvested by a three freeze-thaw procedure and used to infect primary cultures (passage 1) of hPMEC. Adenoviral particles were purified and quantified before experiments using a commercial kit (Viral-Bind TM Adenovirus Purification kit, Cell Biolabs, USA). Cells at 50-60% confluence were seeded 24 hours before adenovirus infection. Viral stocks were diluted to reach the desired multiplicity of infection (MOI) in serum-free medium and added to the cell monolayer. Infected cells were incubated with serum free medium for 8 hours. After this period the infective medium was changed to complete culture medium and cells were incubated for a further 48 hours under standard culture conditions. Isolation of total RNA and protein, and functional assays were then performed as above.

Statistical analysis
Values are mean 6 SEM, with different cell cultures (2-4 replicates) from normal (n = 64) or GDM (n = 64) pregnancies. Since the yield of hPMEC from one single placenta was not enough to proceed with all the experimental strategies included in this study, the reported n values is variable and corresponds to paired cell cultures from normal and GDM pregnancies. Data reported in this study describe a normal standard distribution and comparison between two and 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. The statistical software GraphPad Instat 3.0b and Graphpad Prism 5.0b (GraphPad Software Inc., San Diego, CA, USA) were used for data analysis. P,0.05 was considered statistically significant.

Patients and newborns
Patients with GDM exhibit increased maternal glycosilated hemoglobin A 1c , altered OGTT, insulinemia, increased insulin resistance [21] and reduced ß-cell function (Table 1). In addition, increased fetal insulin resistance and ponderal index in GDM compared with normal pregnancies was found.

Adenosine transport
Overall adenosine transport was lower in GDM compared with normal pregnancies (Fig. 1a). NBTI reduced transport in cells from normal pregnancies to values in GDM, and hypoxanthine reduced adenosine transport only in normal pregnancies. NBTI reduced adenosine transport in GDM to values in normal or GDM pregnancies in presence of NBTI + hypoxanthine. Fig. 1a also shows that insulin did not alter overall adenosine transport in normal pregnancies, but reduced GDM-inhibition of transport. NBTI and hypoxanthine reduced adenosine transport only in normal pregnancies, an effect that was comparable to GDMinhibition in presence of insulin. However, hypoxanthine reduced transport in GDM to values in normal or GDM pregnancies with NBTI + hypoxanthine.
Knowing that hENT1 and hENT2 are differentially regulated by insulin in placental macrovascular endothelium from GDM [5,9], we assayed whether GDM and insulin effect regards selective hENTs modulation in hPMEC. GDM reduced hENT1and hENT2-adenosine transport compared with normal pregnancies (Fig. 1b). Insulin did not alter hENT1-or hENT2-adenosine transport in normal pregnancies or the GDM-inhibited hENT1 transport; however, insulin reversed GDM-inhibited hENT2adenosine transport to values in normal pregnancies.

Umbilical blood adenosine concentration
Adenosine concentration in umbilical whole (arteries + veins) and vein blood was higher in GDM (Fig. 1c). However, adenosine concentration in umbilical arteries blood was similar in GDM compared with normal pregnancies.
Adenosine transport kinetic parameters hENT1-and hENT2-mediated transport was saturable (Fig. 2a,  c) and lineal in a Eadie-Hofstee analysis (Fig. 2b, c). The V max for hENT1-or hENT2-mediated transport was lower in GDM compared with normal pregnancies, without significant changes in apparent K m ( Table 2). Insulin blocked GDM effect on hENT2, but did not alter kinetic parameters for hENT1-adenosine transport in GDM or normal pregnancies. hENT1 and hENT2 expression hENT1 and hENT2 protein abundance was lower in GDM versus normal pregnancies (Fig. 3a). Insulin blocked GDM effect on hENT2 protein abundance (Fig. 3b) or mRNA number of copies (Fig. 3d). However, insulin did not alter hENT1 expression in both cell types (Fig. 3a, c). hENT2 protein and mRNA number of copies half-life were unaltered in cells from GDM in absence or presence of insulin compared with cells from normal pregnancies (not shown).

SLC29A2 promoter activity
Reporter luciferase activity in cells from GDM transfected with pGL3-hENT2 21491 , but not pGL3-hENT2 2602 construct was lower compared with normal pregnancies in basal-insulin (Fig. 3e). Insulin did not alter reporter activity in pGL3-hENT2 21491transfected cells from normal pregnancies, but reversed GDMreduced pGL3-hENT2 21491 reporter activity to values determined in cells from normal pregnancies in basal-insulin.

IR isoforms expression
IR-A and IR-B mRNA expression was comparable in both cell types (Fig. 4c). IR-A mRNA expression was reduced, but IR-B mRNA expression was increased, leading to lower IR-A/IR-B ratio, in GDM compared with normal pregnancies. GDM effect was blocked by insulin, but IR-A/IR-B mRNA expression was unaltered by this hormone in normal pregnancies.

Discussion
This study shows for the first time that GDM is a syndrome associated with reduced overall adenosine transport due to reduced hENT1 and hENT2 activity in primary cultures of human placental microvascular endothelial cells (hPMEC). Since there is increasing experimental evidence establishing that macrovascular compared with microvascular endothelial cell function is different within a same vascular bed (in this case, the human placental vasculature), a characteristic that is critical for the vascular hemostasis, we believe that our present study contributes to a better understanding of the differential biological functions of hPMEC in health and disease. The results of this study also show increased umbilical vein, but not arteries blood adenosine concentration compared with normal pregnancies. Insulin reestablishes GDM-reduced hENT2, but not hENT1 expression and activity. hPMEC from GDM exhibit lower IR-A/ IR-B mRNA expression and p42/44 mapk /Akt activity ratios compared with normal pregnancies, an effect also blocked by insulin. IR-A or IR-B knockdown blocked insulin-restored GDMreduced hENT2 protein abundance, mRNA expression and activity; however, only IR-A knockdown reduced hENT2 expression and activity in normal pregnancies. Regarding hENTs-mediated adenosine transport, insulin reverses a GDM-(preferentially metabolic) to a normal (preferentially mitogenic)phenotype via IR-B and IR-A differential expression and activation in hPMEC.

Adenosine transport
Overall adenosine transport was reduced in hPMEC from GDM compared with normal pregnancies, an effect likely due to reduced hENT1-and hENT2-mediated transport. These results complement the elevated umbilical vein blood adenosine concentration recently reported by our group in umbilical vein blood in GDM [4]. We here show that plasma adenosine concentration in umbilical arteries was unaltered in GDM, suggesting that increased adenosine detected in umbilical vein blood may result from reduced adenosine uptake in hPMEC. In addition, since umbilical artery carries blood from the fetus to the placenta, the latter findings could be the result of a defective placental rather than fetal vasculature adenosine handling in GDM. However, a higher placental adenosine release and/or elevated fetal extracellular adenosine catabolism in GDM can not be rulled out.
Kinetics of adenosine transport show that relative basal maximal transport capacity (V max /K m ) values for hENT1 and hENT2 are similar in normal or GDM ( hENT1/2 F ,1.1) pregnancies (see Table 2). Therefore, the reported comparable contribution to overall adenosine transport of these proteins in hPMEC from normal pregnancies [1] is apparently unaltered by GDM. The V max /K m values for hENT1 and hENT2 are similarly reduced (,55%) (see 1/ N/GDM-hENT1 F and 1/ N/GDM-hENT2 F values in Table 3) due to lower V max in GDM compared with normal pregnancies. Since this GDM effect was associated with similar reduction (,65%) in hENT1 and hENT2 protein abundance, with no changes in the half-life of these proteins, GDM-reduced transport may result from hENT1 and hENT2-lower availability rather than reduced affinity of a fix number of transporters at the plasma membrane [22].
GDM is also associated with lower (,62%) hENT1 and hENT2 mRNA expression, and unaltered mRNA half-lifes, suggesting that reduced availability of these membrane transporters may results from reduced SLC29A1 (for hENT1) and SLC29A2 (for hENT2) transcription. Supporting the latter a correlation between GDMreduced adenosine transport and transporters expression (transport activity/protein abundance, transport activity/mRNA expression, and protein abundance/mRNA expression ratios were ,1.04) was found. Results on SLC29A2 transcriptional activity suggest that a repressor consensus sequence is feasible to exist within 21491 and 2602 bp from ATG of this gene in hPMEC from GDM pregnancies. On the contrary, the 2602 bp promoter fragment is apparently not involved in the GDM-reduced expression of SLC29A2 in this cell type.

Insulin effect on adenosine transport
GDM-reduced hENT2-, but not hENT1-mediated adenosine transport was reversed by insulin, suggesting differential modulation of these membrane transporters. Since insulin action was measured in hPMEC cultured in presence of insulin concentrations equivalent to those found in whole umbilical blood from GDM or normal pregnancies at birth, the possibility that insulin effect was due to this hormone withdrawal in culture is unlikely. Restoration of GDM-reduced hENT2 activity by insulin resulted from a higher V max /K m (,1.7 fold, from (1/ C/Ins-hENT2 F GDM )/(1/ C/ Ins-hENT2 F N ) in Table 3). Most likely insulin effect on hENT2 transport results from restablishment of hENT2 expression since insulin increased the abundance of this protein (,2 fold) and mRNA expression (,3.5 fold). The finding that insulin effect on hENT2 mRNA expression is higher (,1.7 fold) than the protein abundance, suggests that hPMEC from GDM could require higher SLC29A2 transcriptional activity to sustain hENT2 protein content. In addition, because insulin restores GDM-reduced SLC29A2 transcriptional activity and modulates different tran-scription factors [23], this hormone could also be acting as potential inhibitor of repressive transcription factor(s) in SLC29A2 in GDM. Interestingly, insulin was innefective in reversing GDM effects on hENT1 expression and activity in hPMEC. This finding suggests that this nucleoside transporter isoform is likely not under regulation by insulin in this cell type. This is a result complemented by findings in rat B lymphocytes where rat ENT1 (rENT1) is unaltered, but rENT2 is increased by insulin [24]. Thus, a differential modulation of hENT1 and hENT2 by insulin is not exclusive of human placental endothelial cells. Furthermore, insulin was also innefective in modulating the expression or activity of hENT1 in cells from normal pregnancies. The latter could be a characteristic of hPMEC that may be associated with its microvascular origin compared with cells derived from the macrovasculature of the human placenta.