Regulatory, Integrative and Comparative Physiology

Time-dependent physiological regulation of ovine placental GLUT-3 glucose transporter protein

Utpala G. Das, Jing He, Richard A. Ehrhardt, William W. Hay Jr., Sherin U. Devaskar


We immunolocalized the GLUT-3 glucose transporter isoform versus GLUT-1 in the late-gestation epitheliochorial ovine placenta, and we examined the effect of chronic maternal hyperglycemia and hypoglycemia on placental GLUT-3 concentrations. GLUT-3 was limited to the apical surface of the trophoectoderm, whereas GLUT-1 was on the basolateral and apical surfaces of this cell layer and in the epithelial cells lining the placental uterine glands. GLUT-3 concentrations declined at 17–20 days of chronic hyperglycemia (P < 0.05), associated with increased uterine and uteroplacental net glucose uptake rate, but a normal fetal glucose uptake rate was observed. Chronic hypoglycemia did not change GLUT-3 concentrations, although uterine, uteroplacental, and fetal net glucose uptake rates were decreased. Thus maternal hyperglycemia causes a time-dependent decline in the entire placental glucose transporter pool (GLUT-1 and GLUT-3). In contrast, maternal hypoglycemia decreases GLUT-1 but not GLUT-3, resulting in a relatively increased GLUT-3 contribution to the placental glucose transporter pool, which could maintain glucose delivery to the placenta relative to the fetus when maternal glucose is low.

  • placenta
  • glucose transporters

glucose supply to the fetus is derived from the maternal circulation via placental transport of substrate (22). Maternal glucose is transported across the placental barrier into placental trophoblast cells and subsequently into the fetal circulation via facilitative diffusion (5, 28). This transport process is mediated predominantly by two major isoforms of glycoproteins that belong to the family of facilitative glucose transporters (4, 16). GLUT-1, with a transplacental glucose concentration gradient (K m) of 2–5 mM, is the predominant isoform in human (3, 24), sheep (12, 16), rat (13, 34), and mouse (15) placentas. GLUT-3, a more efficient transporter with a K mof 1.8 mM, also has been found, albeit in lower amounts than GLUT-1, in the rat (6, 36), mouse (15), sheep (12,17), and human (20, 31) placentas. Recently, the insulin-responsive GLUT-4 isoform was detected in the villous stromal cells of the human, mouse, and rat placentas (35), but its exact role in the placenta remains unknown.

Investigations in the chronically catheterized pregnant sheep model have demonstrated a gestational-dependent decline in placental GLUT-1 protein concentrations with a reciprocal increase in GLUT-3 protein amounts (12, 17). Thus peak amounts of GLUT-3 are present during late gestation (120–140 days) in the ovine placenta (12, 17). Previous studies in the chronically catheterized pregnant sheep have delineated the glucose kinetics across the placenta in response to experimentally controlled changes in both maternal and fetal glucose concentrations (9, 21, 22). However, the mechanism(s) underlying these alterations remain to be characterized. In other species, particularly the rat and mouse, the effect of a maternal diabetic state (natural or experimental) associated with chronic hyperglycemia on placental glucose transporter proteins has been examined, yielding differing results. In the streptozotocin-treated rat placenta, maternal hyperglycemia with hypoinsulinemia led to no change in placental GLUT-1 concentrations (6, 13) but caused an increase in GLUT-3 concentrations (6). In contrast, in the nonobese, genetically diabetic mouse, which presents with maternal hyperglycemia and hypoinsulinemia, no effect was uniformly observed on placental GLUT-1 or GLUT-3 concentrations (15). The cause behind the discrepancy in the placental GLUT-3 results could stem from a difference in the time of exposure to the perturbation in the glucose concentrations or from a difference in the models used. However, in all of these previous studies, only a single time point was examined during late gestation (6, 13, 15). To date, no studies examining the effect of maternal hypoglycemia on either rat or mouse placental GLUT-1 or GLUT-3 concentrations exist. Previously, we reported a decline in rat placental GLUT-1, when the placenta was exposed to uteroplacental insufficiency leading to fetal hypoxia and hypoglycemia, along with placental and fetal growth restriction (13). In separate experiments, we also demonstrated a time-dependent decline in sheep placental GLUT-1 concentrations in response to both chronic maternal hyperglycemia and hypoglycemia (13). However, other than the decline in the late-gestation placental GLUT-3 concentrations observed in diabetic humans compared with nondiabetic humans (31), which was not substantiated by others (25), no systematic experiments determining the timed and physiological effect of either maternal hyperglycemia or hypoglycemia on placental GLUT-3 concentrations exist. To address this deficiency, we undertook the present investigation in the chronically catheterized pregnant sheep model. We initially immunolocalized GLUT-3 in relation to GLUT-1 in the epitheliochorial ovine placenta, and then we determined the effect of acute and chronic maternal and fetal hyperglycemia and hypoglycemia on placental GLUT-3 protein concentrations in relation to the glucose flux from the uteroplacenta to the fetus.


Animal Model

Late-gestation (125–140 days; term ∼145 days) singleton fetal sheep were studied (n = 53). In selected animals, maternal infusion and sampling catheters were placed at 90 days gestation to undertake longer term experiments (>25 days). These ewes were fasted for 1 day before surgery. Surgery was conducted with the ewes under ketamine anesthesia (12–15 mg/kg bolus induction followed by 0.3–0.5 mg · kg−1 · min−1 constant infusion), intramuscular diazepam (10 mg) for muscle relaxation and sedation, and 1% lidocaine local anesthesia at the incision site. A single maternal sampling catheter was placed into a femoral artery, and two maternal infusion catheters were placed into a femoral vein, all through a single groin incision. In these and all other animals, fetal catheters were placed at ∼115 days of gestation for studies of short to moderate duration (0–20 days). The fetal surgery was conducted under maternal spinal anesthesia (6 mg tetracaine HCl in 2.5 ml of 10% wt/vol dextrose in water) and pentobarbital sedation (5 mg/kg bolus plus intermittent 0.5-mg bolus infusions throughout the surgery). Each ewe and fetus also received 1% lidocaine local anesthesia into all wounds. The ewe's abdomen was opened through a midline laparotomy, and fetal hindlimbs were extracted through a uterine incision. Fetal sampling catheters were placed into the abdominal aorta via hindlimb pedal arteries, and fetal infusion catheters were placed into the femoral veins via hindlimb saphenous veins. After the uterine and abdominal wounds were closed, the catheters were tunneled subcutaneously through a skin incision and kept in a plastic pouch attached to the ewe's flank. Each ewe received intramuscular gentamicin (80 mg) and procaine penicillin G (600,000 U), and the fetus was treated with intra-amniotic ampicillin (500 mg) at the time of surgery alone. Postoperatively, each ewe was maintained in its own cart and allowed ad libitum access to water, alfalfa pellets, and a mineral block. The ewes were paired in the same room to decrease stress and were maintained under environmental conditions of 18 ± 3°C with 18 h of light and 6 h of darkness daily. Weekly intramuscular injections of a multivitamin preparation were given (B-complex Vitamins, Vedco, St. Joseph, MO). The catheters were flushed daily with 1.5 (fetal) or 3.0 ml (maternal) of a heparinized saline solution [150 U heparin/ml 0.9% (wt/vol) NaCl in H2O]. Each animal was allowed at least 4 days to recover from surgery before we started the chronic infusions. At the end of the studies, each ewe and fetus was killed with a rapid intravenous injection of euthanasia solution (Sleepaway, Fort Dodge, IA) (13, 14). Within 30 s of death, the ewe's abdomen was opened, the uterus was extracted, and five to seven placentomes from the uterine horn containing the fetus were cut free. Two to three placentomes were separated manually into fetal cotyledon and maternal caruncle portions. These and the other placentomes were cut into thin slices, snap-frozen in liquid nitrogen, and stored at −70°C until further analyses.

Glucose infusions.

The pregnant ewes were made chronically and markedly hyperglycemic by receiving a continuous 50% dextrose (wt/vol in H2O) intravenous infusion at a variable rate that was adjusted in response to twice daily measurements of plasma glucose concentrations (Yellow Springs model 2300 glucose analyzer, Yellow Springs Instrument, Yellow Springs, OH) to maintain maternal and fetal arterial plasma glucose concentrations at relatively high and constant levels (Δglucose above euglycemia = 2 mM) (n = 17) (13, 14).

Insulin infusions.

Insulin was infused into the ewes to produce acute and/or chronic hypoglycemia. Insulin (“regular” insulin, Iletin U-100, Eli Lilly, Indianapolis, IN) was made up in normal saline (0.9% wt/vol NaCl in water) and administered by constant infusion (∼20 ml/day) at 30–60 pmol (5–10 mU) · kg maternal weight−1 · min−1, as previously described (13, 14). The insulin infusion rate was adjusted daily in response to measurements of maternal arterial plasma glucose concentration, attempting to maintain maternal glucose concentrations that were ∼50% of normal (1.4–1.5 mM).

Sham controls.

Normal, euglycemic (∼2.8 mM) and normoinsulinemic pregnant ewes that were gestational age matched and catheterized (maternal and fetal) constituted the sham control group (n = 10). These animals were monitored and treated exactly as the experimental animals except for the continuous intravenous infusions (13, 14).

Glucose Flux Studies

Experimental methods.

Tritiated water (3H2O) was infused into the fetus to measure umbilical and uterine blood flows by the steady-state diffusion technique. At the start of this study, a3H2O bolus equal to 80 min of infusion was administered over 1 min through a catheter positioned in a fetal hindlimb vein, followed by a constant fetal infusion of3H2O at 20.8 μCi · h−1 · kg estimated fetal weight−1. The 3H2O infusate consisted of 20.8 μCi 3H2O in 1 ml of 0.9% NaCl in H2O; the actual concentration was confirmed at the start and at the end of the infusion by direct counting of the infusate. During the study sampling periods, blood samples were obtained at 15-min interval times four from the maternal femoral artery and uterine vein and from the fetal femoral artery and umbilical vein for analysis of plasma 3H2O concentrations.

Biochemical analysis.

For determination of 3H2O, 0.1-ml plasma samples were solubilized in 1.0 ml of Soluene-350 (quarternary ammonium hydroxide in toluene, Packard) and then mixed with 15 ml of Hionic Fluor (Packard). The 3H radioactivity was measured in a Packard Tri-Carb 460 C liquid scintillation counter.


Umbilical (PFumb) and uterine (Pfut) plasma flows (ml/min) were calculated from 3H2O samples using the steady-state transplacental diffusion method with3H2O as the flow indicator (27). Umbilical (BFumb) and uterine (BFut) blood flows were calculated as previously described (11). PFumb = (BFumb)/(1-fractional fetal hematocrit); PFUt = (BFUt)/(1-fractional maternal hematocrit).

Net glucose uptake rates were calculated according to the Fick principle as the following: net uterine glucose uptake rate = (PFUt) × [ΔG] V − A; net umbilical glucose uptake rate = (PFumb) × [ΔG] v − a; net uteroplacental glucose uptake rate = net uterine glucose uptake rate − net umbilical glucose uptake rate, where [ΔG] is the plasma glucose concentration difference in the uterine venous (V) and maternal femoral artery (A) and the difference in the umbilical venous (v) and fetal femoral arterial (a) vessels.

Immunohistochemical Analysis

Placental cotyledons were excised into the maternal and fetal halves of the tissue. Cryostat (8 μm) sections of the whole placentome, maternal caruncle, and fetal cotyledon were subjected to immunohistochemical analysis as previously described (26). The primary antibody consisted of a peptide affinity-purified IgG, which was generated in rabbits, against the keyhole limpet hemocyanin-linked 10 COOH-terminal amino acids of the ovine GLUT-3 (dilutions ranged from 1:5 to 1:10) (17) with a cysteine residue added at the NH2-terminal. An affinity-purified IgG that was raised in rabbits against the COOH-terminus peptide of the rat GLUT-1 (dilutions ranged from 1:100 to 1:500) was used for comparison (13, 14). Appropriate controls included omission of the primary antibody, normal rabbit serum, and the preimmune serum. The primary antibody incubations were performed overnight at 4°C. This was followed by three 10-min washes in phosphate-buffered saline at 23°C. Subsequently, incubation for an hour at 23°C with the rhodamine-labeled goat anti-rabbit IgG (dilution 1:100) (Sigma, St. Louis, MO) was undertaken in the dark. The sections were washed three times in the dark as described above, coverslipped, and visualized using an Olympus microscope with an epifluorescence attachment using the appropriate filter. Adjacent sections fixed in 4% paraformaldehyde were subjected to hematoxylin and eosin morphological staining to determine the histological localization of GLUT-3 and GLUT-1 proteins.

Western Blot Analysis

Thoroughly washed placentomes were homogenized using a Tekmar tissuemizer (Cincinnati, OH) and then sonicated (60 sonic, Dismembrator, Fisher Scientific, Pittsburgh, PA) using two 50-s cycles of 5–7 W to ensure adequate homogenization of tissue. Protein content was determined by the Bio-Rad protein dye-binding assay (Bio-Rad, Richmond, CA). One hundred micrograms of placental homogenates were subjected to discontinuous 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by electroblot transfer to nitrocellulose (Nytran, Schleicher & Scheull, Keene, NH). Equality of loading and efficiency of transfer were assessed by Coomassie blue staining of the gel and by the transfer of prestained markers. The nitrocellulose filters were incubated overnight at 4°C in 5% nonfat dry milk in phosphate-buffered saline to decrease nonspecific binding of the antibody. This was followed by incubation for 16 h at 4°C with the affinity-purified rabbit anti-ovine GLUT-3 antibody (5 μg/ml) that was extensively characterized previously (17). Subsequently, the peroxidase-linked, goat anti-rabbit IgG-treated filters (1 h at room temperature) were extensively washed and exposed to a chemiluminescence reagent (Amersham Life Science, Little Chalfont, Buckinghamshire, UK). The chemiluminescence was captured by autoradiography over an optimal period of time (5–15 min). GLUT-3 concentrations were assessed by densitometry of the autoradiographs (14). The optical density was corrected for interlane loading by using an internal protein band (∼45 kDa, β-actin), and the data were expressed as a percentage of the mean of the respective euglycemic and normoinsulinemic gestational age-matched sham control group.

Data Analysis

All results are expressed as means ± SE. Differences when comparing more than two groups or time points were determined by one-way analysis of variance followed by the post hoc Newman-Keuls test. When two groups were compared, the Wilcoxon's rank test (nonparametric) and t-test were used.


Table 1 demonstrates the maternal and fetal glucose and fetal insulin concentrations in the timed hyperglycemia and hypoglycemia experiments compared with the euglycemic controls. The maternal plasma glucose concentrations by experimental design were about twofold higher than the age-matched sham control euglycemic group in the hyperglycemia study group. This was paralleled by approximately twofold higher fetal plasma glucose concentrations. Fetal insulin concentrations in the hyperglycemic group were higher than those in the euglycemic group; the values were fourfold higher at 2–24 h and declined subsequently to 50% higher than control levels at 13 to 20 days. Fetal oxygen saturations and contents, similar to previous investigations, were not different in the hyperglycemic group; there were no significant changes in pH or in catecholamine levels (7, 8). Maternal and fetal glucose concentrations were ∼40–50% lower in the hypoglycemic group compared with those in the euglycemic controls. Fetal insulin levels were 40% lower in this group as well.

View this table:
Table 1.

Maternal glucose and fetal glucose and insulin concentrations

As shown in Table 2, net glucose uptake rates by the uterus, fetus, and uteroplacenta were related to maternal glucose concentration. Not all animals had all such studies completed (arbitrary choice), explaining the difference in “n” in this set of data from the total group of animals studied. Of note, with chronic hyperglycemia the mean fetal glucose uptake rate, representing net maternal-to-fetal transplacental glucose flux, was not different from the mean control value, whereas the net uteroplacental glucose uptake rate was significantly increased (P < 0.01). With hypoglycemia the lowest uterine, fetal, and uteroplacental glucose uptake rates occurred with the longest duration of hypoglycemia. Also, the fetal-to-uteroplacental ratio of net glucose uptake rates was significantly lower in the chronically hypoglycemic group (0.45 ± 0.05) than in the control group (0.60 ± 0.06) (P< 0.05). The maternal-to-fetal glucose concentration ratios during acute hyperglycemia and acute and chronic hypoglycemia were not significantly less than those during control conditions, whereas the maternal-to-fetal glucose concentration ratio during chronic hyperglycemia was significantly less (P < 0.05) than during control conditions.

View this table:
Table 2.

Net glucose flux to the fetus from the uteroplacenta in relation to maternal and fetal arterial plasma glucose concentrations

Figure 1 demonstrates the immunolocalization results of GLUT-3 (Fig. 1, B and F) and GLUT-1 (Fig. 1, C and G) along with the hematoxylin and eosin staining of placental sections (Fig. 1,A and E). GLUT-3 immunoreactivity was observed on the apical surface of the trophoblast cell layer (trophoectoderm) covering the fetal vasculature in the fetal cotyledonary villi (Fig. 1,B and D). No such immunostaining was noted in the basolateral surface of this cell layer (Fig. 1, B andD). In contrast, GLUT-1 immunoreactivity was observed in the basolateral and apical surfaces of the trophoectoderm cells in the fetal cotyledon (Fig. 1, C and H). In addition, GLUT-1 was observed in the plasma membrane of the chorionic binucleate cells within the maternal uterine epithelium (Fig. 1 G). A closer look at the maternal uterine epithelium revealed GLUT-1 immunoreactivity in the basolateral and apical surfaces of the cells lining the uterine glands (Fig. 1 G). In contrast, GLUT-3 was observed in minimal amounts, if any, on the apical aspect of the epithelium lining the uterine glands (Fig. 1 F). This localization pattern did not change with hyperglycemic or hypoglycemic exposure.

Fig. 1.

Immunolocalization experiments. Photomicrographs of ovine placental sections treated with hematoxylin and eosin stains to demonstrate the histology (A and E), rabbit anti-ovine GLUT-3 IgG (1:8 dilution) (B and F), and rabbit anti-rat GLUT-1 IgG (1:200 dilution) (C and G) in fetal (A-C) and maternal (E-G) aspects of the placental cotyledon. D and Hdemonstrate a single trophoblast cell layer shown in B(GLUT-3) and C (GLUT-1) at a higher magnification. Negative controls (not shown) did not show any immunolabeling. M, maternal epithelium; F, fetal vascular space; UG, uterine gland. Arrows, basolateral surface of the trophoectodermal epithelial cell layer. Arrow heads, basolateral surface of the epithelial cells lining the uterine gland.

Figure 2 A depicts a representative Western blot demonstrating 50- and 65-kDa GLUT-3 protein bands in the placenta obtained at different times in the maternal hyperglycemia studies. Figure 2 B demonstrates the quantitative data of these two bands and the total GLUT-3 protein concentrations. Whereas no difference in GLUT-3 levels was observed at 2–24 h, 10 days, and 13–15 days of maternal and fetal hyperglycemia, a 60% decline in placental GLUT-3 concentrations was noted at 17–20 days. A representative Western blot depicting GLUT-3 protein bands in the hypoglycemic experiments is shown in Fig.3 A. Quantification of GLUT-3 protein, shown in Fig. 3 B, demonstrated no change in placental GLUT-3 protein concentrations at 4–48 h, 5–15 days, or 25–41 days of hypoglycemia.

Fig. 2.

Hyperglycemia studies and Western blot analysis.A: representative autoradiograph of a Western blot demonstrating two GLUT-3 immunoreactive protein bands (65 kDa and 50 kDa) in the euglycemic, normoinsulinemic sham-operated control (C) and the hyperglycemic experimental group where the placenta was exposed to varying durations of hyperglycemia: 2–24 h, 10 days, 13–15 days, and 17–20 days. B: quantitative analysis of the GLUT-3 protein bands (open bars, 65 kDa; solid bars, 50 kDa; gray bars, composite of 65 kDa + 50 kDa) in the ovine placenta in response to maternal and fetal hyperglycemia of varying durations (2–24 h,n = 7; 10 days, n = 3; 13–15 days,n = 4; 17–20 days, n = 3). The results are expressed as means ± SE. The glucose transporter band amount is expressed as a percentage of the respective mean of the euglycemic, normoinsulinemic, age-matched sham control values denoted as a horizontal dashed line (n = 10). * P< 0.05, # P < 0.01 vs. the euglycemic, normoinsulinemic sham-operated control.

Fig. 3.

Hypoglycemia studies and Western blot analysis.A: representative autoradiograph of a Western blot demonstrating two GLUT-3 immunoreactive protein bands (65 kDa and 50 kDa) in the euglycemic, normoinsulinemic sham-operated control (C) and the hypoglycemic experimental group where the placenta was exposed to varying durations of hypoglycemia: 4–48 h, 5–15 days, and 25–41 days. B: quantitative analysis of the GLUT-3 protein bands (open bars, 65 kDa; solid bars, 50 kDa; and gray bars, composite of 65 kDa + 50 kDa) in the ovine placenta in response to maternal and fetal hypoglycemia of varying durations (4–48 h,n = 4; 5–15 days, n = 4; 25–41 days, n = 7). The results are expressed as means ± SE. The glucose transporter band amount is expressed as a percentage of the respective mean of the euglycemic, normoinsulinemic, age-matched sham control values denoted as a horizontal dashed line (n = 10).


We have immunolocalized both GLUT-3 and GLUT-1 in the epitheliochorial ovine placenta for the first time. GLUT-3 revealed an asymmetrical localization in the trophoblast cells that constitute the trophoectoderm layer between the maternal epithelial tissue and the fetal epithelial layer and vasculature. The presence of GLUT-3 on the apical maternal-facing surface of these cells suggests a polarized localization of this protein. This localization pattern contrasts with that of GLUT-1, which was observed in trophoblast cells at both the basolateral and apical membrane surfaces in the fetal cotyledon. This distribution pattern of GLUT-3 is not specific to the placenta alone but has been noted in other polarized cells such as blastocysts (29), neurons (26), and colonic epithelial cells (19). The immunolocalization pattern of ovine GLUT-3 is similar to that described in the trichorial placenta of the rat, where GLUT-3 was primarily noted on the maternal-facing surface of the syncytiotrophoblast layer 1, whereas GLUT-1 was noted on the maternal- and fetal-facing surfaces of syncytiotrophoblast layer I and the maternal-facing surface of the syncytiotrophoblast layer II (32). In addition, studies involving murine morulae and blastocysts revealed GLUT-3 immunoreactivity first appearing on plasma membranes, and then concentrating in the apical surface of the polarizing outer cells of the compact morula, GLUT-3 later was limited only to the apical surface of the trophoectoderm of blastocysts in this mouse model. In contrast, GLUT-1 was limited to the basolateral membranes of the blastocysts and morula (29). Thus this early establishment of GLUT-3 polarity within the trophoectoderm persists into late gestation in a trichorial placenta (15,32).

However, our observations in the sheep are in distinct contrast to the hemochorial placenta of the human. GLUT-1 in the human placenta has been observed in greater amounts on the apical microvillous surface rather than the basal surface, although it is present on both surfaces (33). Until recently, considerable controversy existed with respect to the presence of the GLUT-3 protein in human placenta (1, 3, 24). However, a recent study (20) demonstrated its presence only in endothelial cells lining the fetal vasculature of the intermediate mature villi with no expression in trophoblast cells. These studies collectively attest to a species-specific placental distribution of GLUT-3 and GLUT-1, which is not unlike what has been described in the brain. Whereas in the mouse or rat brain, GLUT-3 was limited to neuronal cells in situ, canine and human brain revealed the presence of GLUT-3 in neurons and endothelial cells lining the microvasculature (18, 26).

The sheep placental localization pattern that we have observed is suggestive of a collaboration between GLUT-3 and GLUT-1 at the maternal-fetal tissue barrier to ensure an adequate glucose supply and milieu for the developing conceptus. The presence of GLUT-3 at the apical surface of the trophoblast helps trap circulating maternal glucose into the trophoblast cells surrounding the fetal vasculature. Once in these cells, the exit of glucose is mediated via GLUT-1 noted on the basolateral surface of these same cells. GLUT-3, by way of its lower K m, might more efficiently direct maternal glucose into trophoblast cells under maternal hypoglycemic conditions. This ensures glucose supply for placental glucose metabolism when maternal glucose supply is decreased. Under these conditions brought about by very low maternal glucose concentrations, glucose supply to the fetus actually is limited relative to that of the placenta. This relatively higher placental-to-fetal net glucose uptake ratio might be caused in part by the decreased glucose flux at the basolateral membrane.

Whereas very little GLUT-3 exists in the maternal portion of the placental cotyledon, higher amounts of GLUT-1 were observed in the epithelial cells lining the uterine glands and perhaps play a role in the peptide/hormonal secretory function of the placenta during pregnancy. Again, this distribution mimics that previously reported in the rat and murine placentas, where only GLUT-1 was observed in the junctional region as opposed to the labrynthine region of exchange that expresses both GLUT-1 and GLUT-3 (15, 32).

We also demonstrated a time-dependent effect of maternal and fetal hyperglycemia on ovine placental GLUT-3 protein levels. Western blot analysis demonstrated two distinct bands, one of ∼50 kDa and the other of ∼65 kDa. These bands are unlike the single ∼49-kDa band described in the rat and human placentas (6, 20) and do not resemble the compact 42- and 90-kDa actin bands (20). However, the ovine placental GLUT-3 protein band described previously was a diffuse broad band that spanned 45 to 70 kDa (17). Our present results are in close agreement with this previous report, except that optimal autoradiographic exposure was able to resolve the diffuse band into two subtypes. Whereas differing glycoslated forms of GLUT-3 have been described in the placenta (6, 20), unlike GLUT-1 (2), GLUT-3 has not been reported to dimerize and aggregate resulting in a higher molecular weight species.

We previously observed that placental GLUT-1 concentrations acutely increase at 48 h, subsequently decline by 10–13 days, and remain low through 14–20 days in response to maternal and fetal hyperglycemia (13). Complementary to our previous placental GLUT-1 observations, whereas no change was observed from 2–24 h to 13–15 days of hyperglycemia, a significant decline in placental GLUT-3 concentrations was observed at 17–20 days in our present investigation. Thus it appears that placental GLUT-1 levels decrease earlier than GLUT-3 concentrations in response to maternal and fetal hyperglycemia. Furthermore, our present results in the ovine model are similar to previous observations in the nonobese diabetic mouse model (15) but do not reflect the increase in placental GLUT-3 amounts reported in a streptozotocin diabetic rat model (6). This difference may stem from the disease state of a chemically induced diabetic condition that includes perturbations in various substrates other than circulating glucose. Furthermore, in this chemically induced severe diabetic model with a higher degree of hyperglycemia (30), an increase in placental GLUT-3 might lead to enhanced placental and fetal complications due to increased maternal-fetal glucose transport (6, 30), including significant fetal hyperglycemia and growth restriction noted in this model (6, 10). Our present study focused on characterizing the change in placental GLUT-3 only in response to physiological hyperglycemia in the absence of a disease state. Under these conditions, both placental GLUT-1 (13) and GLUT-3 concentrations declined, leading to a diminution of the total placental glucose transporter pool. The decrease in GLUT-1 concentrations may represent a mechanism to protect the developing fetus, because excess glucose could lead to fetal hypoxia, tissue acidosis, and perhaps other tissue-specific forms of glucose toxicity (14, 30). Under these chronic but relatively milder hyperglycemic conditions, there remains an absolute and relative increase in the uteroplacental-to-fetal net glucose uptake rate ratio. This and the observation that uterine glucose uptake rate was just as high as it was during acute maternal hyperglycemia indicate that the decrease in GLUT-3 alone or along with a decline in GLUT-1 does not adequately decrease uterine glucose uptake or uteroplacental glucose uptake. It also is possible that the decrease in GLUT-1 does not adequately reduce fetal glucose uptake under these conditions, contrary to our speculation based on previous in vitro investigations (13), because the fetal-to-maternal glucose concentration ratio under these conditions was relatively higher than during acute hyperglycemia (perhaps due to relatively decreased fetal glucose utilization rate). Thus a decline in net glucose transport to the fetus may rely completely on a reduction in the maternal-fetal glucose concentration gradient and not solely on any diminution of either GLUT-1 or GLUT-3. Alternatively, despite the hyperglycemia-induced protective decline in placental GLUT-1 and GLUT-3 concentrations, the maternal-fetal transfer of glucose remained high perhaps due to an increase in transporter intrinsic activity. Thus the degree of placental transporter concentration decline noted in this study may prove to be inadequate to completely and absolutely protect the fetus from hyperglycemia and its potential toxicity.

In contrast, maternal and fetal hypoglycemia did not alter placental GLUT-3 concentrations in this ovine model. This is similar to studies in the human placenta obtained from pregnancies bearing intrauterine growth-restricted infants (25) and our preliminary studies in the maternal uterine artery-ligated rat model of uteroplacental insufficiency (13). In our previous studies in pregnant sheep, we noted a decline in placental GLUT-1 levels from 24–48 h through 33–41 days due to maternal and fetal hypoglycemia (13). Thus, in the presence of hypoglycemia, the relative contribution of placental GLUT-3 to the total glucose transporter pool increases. This relative increase in GLUT-3 may serve to maintain glucose supply to the trophoblast or placental portion of the fetal cotyledon of the placentome. Thus GLUT-3, with its higher affinity for glucose and its unique localization in the ovine placenta, might play a role of maintaining placental glucose consumption relative to the fetus under conditions of limited glucose availability, such as with maternal hypoglycemia. Before a change in GLUT-3 concentrations, it would be necessary to saturate the placental glucose transporters with the available glucose. From these conditions, we speculate that prolonged hypoglycemia might be associated with an unsaturated state of this transporter isoform due to low amounts of available glucose, negating a need for further perturbations in the transporter concentration when the available complement of placental glucose transporters is not being fully utilized.

It is interesting to note that the only condition in which the maternal-to-fetal glucose concentration ratio was different from the control ratio was in the group of animals studied under chronic hyperglycemic conditions. In this case, the reduced ratio was due to a relatively higher fetal glucose concentration. This could come about by a reduction in glucose utilization rate capacity in the fetus. One explanation for such a condition is our previous observation in fetal sheep that chronic, marked hyperglycemia reduces fetal plasma insulin concentrations to normal or lower values, which would have the effect of decreasing fetal glucose clearance (7). In the present study, although we did not observe such insulin levels, a steady time-dependent decline almost close to normal values was noted. This would produce relatively higher fetal plasma glucose concentrations even at normal rates of fetal net glucose uptake, as occurred in the present studies. This and the other data defining direct relationships between the maternal glucose concentration and/or the maternal-fetal glucose concentration gradient indicate that transplacental glucose flux was predominantly under the control of glucose concentration relationships, as previously described (23). Whether the temporal, direct relationships between transplacental glucose flux and the amounts of GLUT-3 and GLUT-1 also are causal will require further experiments of transplacental glucose transport capacity. Models for such studies have not been developed which, by their nature of producing different steady-state concentrations of glucose and insulin in the maternal and fetal plasma, do not alter the very expression and activity of the GLUT-3 and GLUT-1 transporters that were experimentally produced for investigation.

We conclude that in the ovine placenta, GLUT-3 is immunolocalized in a polarized fashion to the apical maternal-facing surface of the trophoectoderm that surrounds the fetal vasculature, whereas GLUT-1 is in the basolateral and apical membranes of this cell layer along with the epithelium lining the uterine glands found on the maternal aspect of the placentome. Maternal and fetal hyperglycemia imposes a time-dependent decline in GLUT-3 concentrations. This decline, along with the previously reported decline in GLUT-1 levels (13), might serve to relatively but not completely protect the trophoectoderm-fetal unit against potentially deleterious effects of maternal hyperglycemia, because these changes in GLUT-3 and GLUT-1 concentration were associated with increased uterine and uteroplacental with a return to normal of fetal net glucose uptake rates. Thus these placental transporter changes may protect the fetus, but not the uteroplacental unit, from potential glucose toxicity. In contrast to hyperglycemia, maternal insulin infusions causing maternal hypoglycemia had no effect on placental GLUT-3 concentrations. Despite no change in GLUT-3 levels, GLUT-1 concentrations declined with chronic hypoglycemia (13), producing a decrease in uterine, uteroplacental, and fetal net glucose uptake rates. The relative increase in the GLUT-3 pool perhaps serves to ensure an adequate glucose supply to the trophoectoderm at the expense of the fetus and uterine tissues. This potential effect of the higher GLUT-3-to-GLUT-1 ratio is consistent with the lower fetal-to-uteroplacental ratio of the net glucose uptake rate in the chronically hypoglycemic group (0.45) versus the normal control group (0.60). Finally, it is reasonable to consider that factors other than changes in glucose transporter expression might develop during chronic hyperglycemic and hypoglycemic conditions that also could regulate placental glucose uptake and transport to the fetus. Given the high metabolic rate of glucose by the placenta, it might be possible, for example, that with chronic hyperglycemia or hypoglycemia, the greater glucose consumption rate by the placenta relative to the rate of glucose transport to the fetus could be due to a relative increase in the metabolic rate of glucose by trophoblast cells, and this could be caused by regulatory steps in glucose metabolism beyond GLUT-mediated cell glucose uptake. Future studies will be needed to define transport kinetic parameters in relation to GLUT-3 and GLUT-1 expression and location using transplacental glucose transport kinetic measurements. Such studies also should address factors other than glucose transporter expression that might account for changes in glucose uptake, metabolism, and transport by the placenta in response to acute and chronic changes in maternal and fetal glucose concentrations.


Although we have shown temporal changes between absolute and relative GLUT-3 and GLUT-1 concentrations and placental transport of glucose to the fetus, there are several reasons to be cautious about interpreting a direct causal relationship between these factors. First, the measured transplacental glucose flux rates do not define placental glucose transport capacity (V max) or sensitivity to the transplacental glucose concentration gradient (K m), because these parameters would require performing studies (22) that would markedly alter the conditions (e.g., glucose and perhaps insulin concentrations, as well as transporter protein expressions and activities) at the time of the study. Second, whereas changes in transporter concentrations (and activity) might affect glucose flux, transplacental glucose flux still is dependent on the maternal glucose concentration, the transplacental glucose concentration gradient, and the rate of glucose consumption by the uteroplacenta. Third, placental glucose transporters normally operate well below saturation over the range of glucose concentrations at which measurements of glucose flux and transporter concentrations were assessed in the present study. Fourth, the different transporters and their different locations, transport parameters (K m, V max), and relative concentrations all could vary independently, precluding assumptions of their specific roles in regulating flux from global measurements of transporter concentration, transporter location, and transplacental glucose flux alone. Fifth, in the hyperglycemic studies, glucose flux was greatest early when GLUT-3 [and GLUT-1, (13)] was highest, and lowest later when GLUT-3 (and GLUT-1) was lowest. Normal flux rates were observed with chronic hypoglycemia when GLUT-3 was unchanged, but GLUT-1 was reduced (13). These results reinforce but do not prove our conclusion that the absolute and relative amounts of GLUT-3 and GLUT-1 are related to transplacental glucose flux.


We thank Alan W. Bell, Cornell University, Ithaca, NY for the use of the anti-ovine GLUT-3 antibody, which was developed in his laboratory.


  • This work was supported by National Institutes of Health (NIH) Grants HD-33997 and HD-25024 to S. U. Devaskar, and Grants HD-28794 and HD-20761 to W. W. Hay, Jr.

  • Address for reprint requests and other correspondence: S. U. Devaskar, Dept. of Pediatrics, 10833, Le Conte Ave., MDCC B-375, Los Angeles, CA 90095-1752 (E-mail: sdevaskar{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract