Objective
Obese women are at increased risk to deliver a large infant, however, the underlying mechanisms are poorly understood. Fetal glucose availability is critically dependent on placental transfer and is linked to fetal growth by regulating the release of fetal growth hormones such as insulin. We hypothesized that (1) umbilical vein glucose and insulin levels and (2) placental glucose transporter (GLUT) expression and activity are positively correlated with early pregnancy maternal body mass index and infant birthweight.
Study Design
Subjects in this prospective observational cohort study were nondiabetic predominantly Hispanic women delivered at term. Fasting maternal and umbilical vein glucose and insulin concentrations were determined in 29 women with varying early pregnancy body mass index (range, 18.0–54.3) who delivered infants with birthweights ranging from 2800–4402 g. We isolated syncytiotrophoblast microvillous and basal plasma membranes from 33 placentas and determined the expression of GLUT-1 and -9 (Western blot) and glucose uptake (radiolabeled glucose).
Results
Birthweight was positively correlated with umbilical vein glucose and insulin and maternal body mass index. Umbilical vein glucose levels were positively correlated with placental weight and maternal body mass index, but not with maternal fasting glucose. Basal plasma membranes GLUT-1 expression was positively correlated with birthweight. In contrast, syncytiotrophoblast microvillous GLUT-1 and -9, basal plasma membranes GLUT-9 expression and syncytiotrophoblast microvillous and basal plasma membranes glucose transport activity were not correlated with birthweight.
Conclusion
Because maternal fasting glucose levels and placental glucose transport capacity were not increased in obese women delivering larger infants, we speculate that increased placental size promotes glucose delivery to these fetuses.
Obesity in pregnancy is linked to a multitude of short- and long-term adverse fetal outcomes. For example, obese women are more likely to give birth to a macrosomic infant, which is associated with increased perinatal morbidity and risk of developing obesity and metabolic syndrome later in life. Fetal growth is highly dependent on the availability of nutrients such as glucose; however, the mechanisms underlying fetal overgrowth in nondiabetic pregnancies of obese women remain to be fully established.
Glucose is the primary energy substrate for the placenta and the fetus and fetoplacental glucose needs are met entirely by uptake from the maternal circulation. Approximately 55% of the glucose taken up from the uteroplacental circulation is metabolized by the placenta and the remaining 45% is transferred to the fetus. Glucose stimulates fetal secretion of insulin and insulin-like growth factor-I (IGF-I), the 2 primary fetal growth hormones, providing a direct link between fetal glucose availability and fetal growth. In addition, glucose can be converted to fat consistent with the possibility that increased glucose availability could increase fetal adiposity.
Placental glucose transport occurs via facilitated diffusion, mediated by glucose transporters (GLUT). GLUT-1 is highly expressed in the 2 plasma membranes of the syncytiotrophoblast (microvillous membrane, [MVM] and basal membrane, [BM]), the transporting epithelium of the human placenta. Because BM has been shown to have a much lower expression of GLUT-1 than MVM, the transfer across BM has been suggested to be the rate limiting step in maternofetal glucose transport. This model is supported by recent mathematical modeling of placental glucose transport and studies in in vitro perfused placenta. BM GLUT-1 expression and glucose transporter activity are increased in pregnancies complicated by diabetes, in particular in association to increased fetal growth. In addition, GLUT-9 is also expressed in term placentas and MVM and BM GLUT-9 expression has been reported to be increased in pregnancies complicated by diabetes.
Increased fetal glucose availability could contribute to fetal overgrowth in obese women without diabetes. Because of the facilitated nature of maternal-fetal glucose transfer even a modest increase in maternal glucose levels may enhance glucose supply to the fetus, which is consistent with the continuous association between maternal glucose levels (below those diagnostic of diabetes) and increased birthweight reported by the HAPO Study. Obese women are at greater risk for glucose intolerance in pregnancy because of their markedly lower insulin sensitivity as compared with lean controls and intermittent minor elevations of maternal blood glucose could contribute to stimulate fetal growth in these women. Alternatively, an enhanced placental glucose transport capacity could increase fetal glucose availability in obese women. We hypothesized that (1) umbilical vein glucose and insulin levels and (2) placental glucose transporter expression and activity are positively correlated with early pregnancy maternal BMI and infant birthweight. To address this hypothesis we collected maternal and umbilical vein plasma samples and placentas from pregnancies of women with varying BMI giving birth to infants across the growth spectrum, from appropriate for their gestational age to large infants with birthweights greater than 4000 g. We determined maternal and fetal plasma glucose levels and studied GLUT-1 and GLUT-9 protein expression and glucose transport activity in isolated MVM and BM.
Materials and Methods
Study subjects
We obtained coded placental tissue and plasma samples and deidentified relevant medical information from a tissue repository approved by the University of Texas Health Science Center, San Antonio institutional review board (HSC20100262H), to which pregnant women were recruited following written informed consent. We used samples from healthy women with normal term pregnancies. Samples were selected randomly with the exception that we included all overweight/obese women giving birth to large infants from whom samples were available. Samples from a total of 52 women who were lean (BMI 18.0-24.9, n = 20) or overweight/obese (BMI 25-54.3, n = 32) during early pregnancy (<20 weeks’ gestational age) and had uncomplicated term pregnancy (>37 weeks of gestation) were used. In 10/52 study subjects both blood and placental samples were available. Thus, of the 29 women in whom we had blood samples and the 33 where we had placental samples (29 + 33 = 62) 10 overlap, corresponding to the 52 unique study subjects. Study subjects were delivered by cesarean section (n = 47) or vaginally (n = 5). Seventy-five percent of the study subjects were Hispanic (Mexican-American), 21% were nonHispanic whites, 2% were Asian, and 2% were African-American.
Collection of blood and determination of glucose concentrations
After at least an 8-hour self-reported fast, maternal blood samples were obtained from the antecubital vein before cesarean delivery. Immediately after delivery, blood was collected from the umbilical vein. All blood samples were collected in a purple top vacutainer blood collection tube containing ethylenediaminetetraacetic acid and within 30 minutes of collection, were centrifuged at 2500 rpm. Plasma was flash frozen in liquid nitrogen and subsequently stored at −80°C. Maternal and fetal plasma glucose levels were measured in triplicate using an Analox Glucose Analyzer GM9 (Analox Instruments USA Inc., Lunenburg, MA).
Placenta collection and immunohistochemistry
Placentas were placed on ice immediately after delivery and several small villous tissue pieces were rinsed in cold physiologic saline before being fixed in formalin and embedded in paraffin. Immunohistochemical analysis was performed on 5 μm sections. The slides were heated to 60-70°C for 20 minutes and thereafter cooled to room temperature. The paraffin was removed by xylene, and the tissue rehydrated with ethanol and then placed in phosphate buffered saline. The slides were boiled for 10 minutes in citrate buffer (H-3300 Antigen Unmasking Solution; Vector Laboratories, Burlingame, CA). The slides were cooled for 30 minutes in room temperature and thereafter washed 3 times in phosphate buffered saline for 5 minutes. The immunochemical staining was done with the Vectastain Elite ABC kit (Vector Laboratories) with some minor modifications. Briefly, endogenous peroxidase was quenched in 3% H 2 O 2 for 20 minutes. The sections were incubated with primary antibodies overnight at 4°C in a humidified chamber. Primary antibodies targeting GLUT-1 and GLUT-9 were purchased from Millipore, Temecula, CA and Abcam, Cambridge, MA, respectively. Subsequently, sections were incubated in secondary antibodies for 60 minutes at room temperature. Staining was visualized with a peroxidase substrate kit (DAB SK-4100; Vector Laboratories). Sections were then counterstained with hematoxylin, cleared, dehydrated, and mounted.
Isolation of syncytiotrophoblast plasma membranes
Syncytiotrophoblast MVM and BM were isolated according to a well-characterized protocol. The isolated MVM and BM vesicles were stored at −80°C in buffer D containing protease and phosphatase inhibitors (Sigma-Aldrich). Protein content of the vesicles was determined using Pierce BCA Protein Assay Kit (Pierce Biotechnology, Rockford, IL). Purity of isolated MVM was measured by enrichment of alkaline phosphatase activity. At least a 10-fold enrichment of enzyme activity in isolated vesicles compared with placental homogenate was used as a marker for successful enrichment. The mean enrichments of alkaline phosphatase in MVM was 13.86 ± 0.89. The enrichment of BM was determined using the protein expression of the iron transporter ferroportin-1 (SLC40A1), which is almost exclusively expressed on the BM to facilitate unidirectional iron transport from mother to fetus. The mean enrichment of ferroportin expression in BM was 15.7 ± 2.5.
Western blot
Protein expression of GLUT-1 and GLUT-9 in MVM and BM was determined using commercial antibodies (Millipore and Abcam) and Western blotting, which was performed as described previously, with the modification that amido black was used to quantify the total protein for normalization. GLUT-1 was used at 1:20,000 and GLUT-9 at 1:1000 dilution. Relative GLUT-1 and GLUT-9 densities were analyzed by densitometry by means of ImageJ software (National Institutes of Health). To facilitate comparisons between groups, the expression of the target protein in each individual sample (target/total protein density) was then calculated as percentage of the mean of all samples.
Glucose transport activity measurements
Uptake of radiolabeled D- and L-glucose into plasma membranes vesicles was measured at 23°C as described previously with some modifications. With 2 investigators performing the experiment, one investigator promptly mixed vesicles at 23°C (40 μL) with 20 μL D-glucose (12 mmol/L), mannitol (288 mmol/L), and HEPES-Tris (10 mmol/L; pH 7.4) containing (14C) D-glucose (0.015 μCi/μL) and tritiated L-glucose (0.048 μCi/μL). Using a metronome, after 0.6 seconds, uptake of radiolabeled glucose was terminated by the addition of 2 mL of ice cold stop solution (250 μmol phloretin per liter in buffer A, plus 2% ethanol) by a second investigator using a precalibrated bottle top automatic dispenser. Vesicles were separated from the substrate medium by filtration on mixed ester filters (0.45 μm pore size, Millipore) and washed with 6 mL of rinse solution (100 μmol phloretin per liter in buffer A, plus 2% ethanol). Measurements were carried out in triplicate for each sample. Filters were placed in 2 mL liquid scintillation fluid and counted. Uptake at 0.6 seconds was taken to approximate the initial rate. Net (carrier-mediated) D-glucose uptake was calculated by subtracting the L-glucose rate from the total D-glucose uptake rate. To slow the rate of glucose uptake and confirm the initial findings, the experiment was duplicated at +4°C. Uptake at 1.5 seconds was taken to approximate the initial rate after establishing a time course for +4°C.
Data presentation and statistics
Our sample sizes represent a convenience sample as we did not have sufficient data for a power calculation or sample size determination. There were a total of 52 unique study subjects across the range of maternal BMIs; 29 plasma and 33 placental samples with 10 study subjects’ samples overlapping between the 2 groups. Summary data are presented as means ± SEM. Because our data did not significantly deviate from normality for most variables (as tested using Shapiro-Wilk test) and the well-established robustness of t tests even with relatively small sample sizes and with significant deviations from normal distribution, statistical significant differences between groups were determined using Student t test. In addition, because the approximate normal distribution of data, variables were analyzed using Pearson’s correlation coefficients as continuous across the range of birthweights, placental weights, maternal BMI, and maternal fasting glucose. A P < .05 value was considered significant.
Results
Demographic data and glucose concentrations
Maternal age and gestational age at delivery did not differ between the normal BMI (BMI <25) and overweight/obese (Ow/Ob) (BMI ≥25) groups ( Table ). Ninety-six percent of the study subjects were multiparous and 87% of the women had not experienced labor. Per study design, there was a difference in early pregnancy maternal BMI and infant birthweight between the 2 groups ( P < .0001, P < .0001; Table ). Maternal BMI remained significantly higher in the Ow/Ob group at delivery as compared with the normal BMI group ( P < .0001, Table ). Gestational weight gain was not significantly different between the 2 groups ( Table ). Fifty-eight percent of the newborns were male and 42% were female. The placental weight of Ow/Ob women was significantly greater than normal BMI women ( P = .006; Table ). No difference was observed between maternal fasting glucose, maternal/ fetal insulin and homeostasis model assessment of insulin resistance values between the 2 groups ( Table ). Birthweight was positively correlated with umbilical vein glucose ( P = .008; Figure 1 , A), insulin ( P = .04; Figure 1 , B), and maternal BMI ( P = .03; Figure 1 , C). Placental weight positively correlated with umbilical vein glucose ( P = .001; Figure 2 ). Maternal BMI ( P = .04; Figure 3 , A), but not gestational weight gain (not shown), was positively correlated with umbilical vein glucose. However, there was no relationship between maternal fasting glucose and umbilical vein glucose ( P = .94; Figure 3 , B).
| Descriptions | BMI <25 | BMI ≥25 | Total/ P value |
|---|---|---|---|
| n | 20 | 32 | 52 |
| Age, y | 29.1 ± 1.5 | 27.6 ± 1.0 | .39 |
| Gestational age, wks | 39.1 ± 0.2 | 39.5 ± 0.1 | .16 |
| Nulliparous | 1 | 1 | 2 |
| Labor | 3 | 4 | 7 |
| Early pregnancy BMI | 21.3 ± 0.4 | 34.3 ± 1.0 | < .0001 |
| Birthweight, g | 3191 ± 50 (2800-3699) | 3768 ± 66 (3167-4402) | < .0001 |
| BMI at delivery | 27.4 ± 1.0 | 36.8 ± 1.3 | < .0001 |
| Gestational weight gain, lbs | 24.2 ± 2.4 | 18.2 ± 3.5 | .21 |
| Male/Female | 13/7 | 17/15 | 30/22 |
| Placental weight, g | 661 ± 18 | 782 ± 32 | .006 |
| Maternal fasting glucose, mg/dL | 73.4 ± 3.2 | 74.6 ± 1.5 | .71 |
| Maternal insulin, pg/mL | 342.0 ± 59 | 532 ± 60 | .05 |
| Maternal HOMA-IR | 1.65 ± 0.33 | 2.5 ± 0.29 | .09 |
| Fetal insulin, pg/mL | 160 ± 30.8 | 238 ± 51.0 | .23 |
| Fetal HOMA-IR | 0.50 ± 0.10 | 0.82 ± 0.20 | .21 |
