Fibroblast growth factor 21 (FGF21) can regulate glucose and lipid metabolism. The placenta actively synthesizes and secretes many hormones, but it is unknown whether this includes FGF21. This study aimed to analyze the placental expression of FGF21 in women with or without gestational diabetes mellitus (GDM).
FGF21 and peroxisome proliferator-activated receptor (PPAR)-α mRNA and protein expression were measured in the placentae of 20 women with and 18 without GDM. mRNA expression of PPARα, FGF receptors 1–4, the coreceptor β-klotho, and glucose transporter (GLUT)-1, -3, and -4 was investigated. Maternal and fetal circulating FGF21 levels were assessed in 10 mother-baby dyads per condition.
FGF21 was expressed in the placenta and its mRNA expression increased in women with GDM [10.75 (interquartile range 3.28–125.6 AU)] vs control [0.83 (0.22–4.78), P < .001], as is its protein expression [GDM 2.89 (1.44–5.10)] vs control [0.42 (0.05–1.98), P < .05]. PPARα mRNA but not protein expression was increased in GDM [2.94 (0.70–7.26)] vs control [0.99 (0.43–2.17), P < .05] and was positively correlated to FGF21 mRNA expression (ρ = 0.43, P < .01). Placental mRNA expression of FGF receptors and GLUT1 was unchanged, and β-klotho, GLUT3, and GLUT4 showed increased expression in GDM. Maternal circulating FGF21 levels were similar [GDM 323 (75–921) vs control 269 (49–731) pg/mL, P = .81]. FGF21 was undetected in fetal cord blood.
FGF21 is expressed in the placenta and its expression is increased in GDM. The absence of FGF21 in fetal cord blood suggests that neither placental FGF21 nor maternal circulating FGF21 is secreted into the fetal circulation. Placental FGF21 may be a regulator of placental metabolism.
Fibroblast growth factor (FGF)-21 is a hormone that is expressed mainly in the liver but also in other metabolically active tissues such as skeletal muscle, adipose tissue, and the pancreas (1–3). In animal studies including in rodents and monkeys, pharmacological doses of FGF21 lower circulating glucose and triglyceride levels and can prevent obesity through increasing energy expenditure (4, 5). In 3T3-L1 adipocytes, FGF21 induces the expression of glucose transporter (GLUT) 1, thereby lowering circulating glucose levels (5). FGF21 gene expression levels in 3T3-L1 adipocytes increase in the fed state and decrease in the fasted state (6). Exogenous FGF21 can cross the blood-brain barrier and exerts effects on food intake, energy expenditure, and hepatic insulin sensitivity in male obese rats (7).
FGF21 is detected in rodent and human pancreatic islets of Langerhans (4). In humans with type 2 diabetes mellitus, circulating FGF21 levels are increased (9). In contrast, in type 1 diabetes mellitus and latent autoimmune diabetes of adults, FGF21 levels are decreased (9). In humans, circulating FGF21 levels are markedly higher in obesity (10). In pregnancies affected by gestational diabetes, circulating FGF21 levels were reported to be unaltered when compared with controls matched for fasting insulin (11).
FGF21 is a member of the endocrine branch of the fibroblast growth factor family, and it binds to the general FGF receptors 1–4. Receptor activity is increased by the presence of the coreceptor β-klotho, which is mainly expressed in adipose tissue, the liver, brain, and pancreas in the rodent (12). In the liver, FGF21 expression is regulated by peroxisome proliferator-activated receptor (PPAR)-α (13), whereas in adipose tissue it is regulated by PPARγ (14). The PPARs are nuclear receptors and are important mediators of the metabolic condition of the organism. In humans, the pharmacological stimulation of PPARα leads to increased circulating FGF21 levels (15, 16).
The placenta is the site of interaction between mother and baby, and it is a metabolically highly active organ. The placenta produces many hormones and growth factors with endocrine and paracrine effects, secreting into both maternal and fetal blood, depending on the properties of the hormone. It is unclear whether FGF21 is expressed in the placenta, and if so, whether the expression changes in pregnancies affected by gestational diabetes mellitus (GDM). Placental FGF21 expression could affect placental metabolism and nutrient transfer and thereby the growth of the fetus. Alternatively, FGF21 could be secreted into the fetal circulation and directly affect fetal metabolism. The placenta has been shown to express FGF receptors, and their expression is altered in type 1 diabetes mellitus (17) but it is not clear whether GDM is associated with changes in FGF receptor expression. In the current study, placental expression of FGF21, its (co)receptors, and the transcription factors PPARα and PPARγ were analyzed in both normoglycemic pregnancies and pregnancies affected by GDM. Placental expression of GLUT1, -3, and -4 was studied as possible downstream mediators for FGF21. We also measured FGF21 levels in the maternal and fetal circulation.
Materials and Methods
Study population
Twenty women with GDM and 18 normoglycemic (control) women matched for prepregnancy body mass index (BMI), maternal age, gestational age of delivery, sex of the baby, and birth centile were recruited for participation. The study was approved by the Human Research Ethics Committee of the Royal Brisbane and Women's Hospital. All participants gave written informed consent prior to enrollment. Prepregnancy BMI was calculated as prepregnancy weight (kilograms) divided by the squared height in meters. The customized birth centile was calculated with an online calculator (www.gestation.net). Screening for GDM was performed with a two-step procedure: all women underwent a 50-g glucose challenge test, and women with glucose values above the threshold had a 75-g oral glucose tolerance test to confirm a diagnosis of GDM according to the criteria of the Australasian Diabetes In Pregnancy guidelines (18). All patients were initially treated with diet, supplemented with metformin or insulin as required. Three patients were treated with insulin and one with metformin in this study. Maternal blood samples were collected late in the third trimester, at 38 ± 4 days of gestation for the women with GDM and at 38 ± 5 days for the normoglycemic women. Placenta and cord blood (of mixed arterial and venous origin) were collected at delivery and processed within 2 hours. One-cubic centimeter samples of placenta were excised from anatomically healthy parts of the placenta and immediately frozen in liquid nitrogen prior to storage at −80°C. In addition, 1-cubic centimeter samples were excised and fixed for 48 hours in 4% paraformaldehyde and kept in saturated sucrose solution until embedding for immunohistochemistry.
Gene expression
mRNA was isolated from placenta with the Allprep RNA/DNA extraction mini kit (QIAGEN). RNA was quantified by Nanodrop, and all samples had 260:280 ratios greater than 1.8 and 260:230 ratios greater than 1.7 for 33 samples and less than 260:230 less than 1.7 for five samples. These five samples were removed from the analysis. Seven hundred fifty nanaograms of mRNA were reverse transcribed to cDNA with the QuantiTect reverse transcription kit (QIAGEN) using a mixture of oligodeoxythymidine and random primers. Quantitative real-time PCR was performed on 18.75 ng of cDNA with 300 nM of primers and iTaq universal SYBR green mastermix (Bio-Rad Laboratories) on an iQ5 PCR machine (Bio-Rad Laboratories). The PCR protocol consisted of one cycle at 95°C for 10 minutes, 40 cycles of 95°C for 15 seconds, and 59°C for 1 minute followed by dissociation curve analysis. Primers unique for the target gene and covering exon-exon junctions were designed with primer BLAST. The primer sequences are presented in Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org. β-Actin (ACTB) and TATA-box binding protein (TBP) were used as endogenous controls and showed similar results; therefore, the results presented here are normalized to TBP because the expression levels of TBP were more in line with those of the target genes. To further adjust for the potential differences in the cellular composition of the placental samples, gene expression was normalized to the geometric mean of expression of TBP, cytokeratin 7 (CK7) as a marker for trophoblast cells, CD34 (CD34) for endothelial cells, and desmin (DES) for smooth muscle cells.
Protein expression
Placentae were lysed with a radioimmunoprecipitation assay buffer consisting of 50 mM Tris, 1% Triton X-100, 0.1% sodium dodecyl sulfate, 0.5% sodium deoxycholate, 150 mM NaCl, and protease inhibitor cocktail (Roche, Applied Science). Tissue was disrupted by violent shaking with a 5-mm stainless steel bead in a TissueLyser (QIAGEN). After lysis, the sample was centrifuged for 10 minutes at 4°C, and the protein content in the supernatant was determined by a bicinchoninic acid assay (Sigma-Aldrich). Thirty micrograms of protein were loaded onto a 4%–12% gradient NuPAGE Bis-Tris gel (Life Technologies), transferred onto a polyvinylidene difluoride membrane (Millipore), and blocked for 1 hour with 5% nonfat dry milk in PBS. The primary antibody for rabbit anti-FGF21 (ab54857; Abcam) at 1 μg/mL, rabbit anti-PPARα at 1:150 (H-98; Santa Cruz Biotechnology), or rabbit anti-KLB at 1:1000 (ab106794; Abcam) and mouse anti-β-Actin (A5316; Sigma Aldrich) at 1:20 000 in 5% nonfat dry milk in PBS-Tween 20 were incubated overnight at 4°C with agitation. Secondary LI-COR antibodies, goat antirabbit 800CW (dilution 1:10 000; 926-32211, LI-COR), and donkey antimouse 680LT (dilution 1:15 000; 926-68022, LI-COR) were incubated for 1 hour at room temperature, and protein was detected by the Odyssey infrared imaging system (LI-COR). Protein expression was analyzed by densitometry with FGF21 protein levels normalized to β-actin levels as protein loading control.
FGF21 immunohistochemistry
Paraffin-embedded tissue sections were baked and hydrated, and antigen retrieval was performed by boiling for 30 minutes in 10 mM sodium citrate and 0.05% Tween 20 at pH 6.0. The slides were blocked for 15 minutes with Biocare Background Sniper (MACH2; Biocare Medical). The 1:300 primary antibody (ab54857; Abcam) was incubated overnight at 4°C. After washing, rabbit horseradish peroxidase-polymer was incubated for 60 minutes at room temperature followed by diaminobenzidine incubation for 1 minute. Slides were counterstained with Harris' hematoxylin (HHS16; Sigma Aldrich) and mounted with coverslips.
FGF21 ELISA
Twenty-five microliters of serum from 10 mother-baby pairs with GDM and 10 mother-baby pairs without GDM were analyzed in duplicate by FGF21 ELISA according to the manufacturer's recommendations (ab125966; Abcam). The detection range of the assay is from 31 to 2000 pg/mL, the intraassay coefficient of variation is 4.8%, and the interassay coefficient of variation is 7.4%.
Statistical analysis
Differences between the clinical characteristics of the study participants were analyzed by parametric t tests. Nonparametric, two-sided Mann-Whitney U tests were used in the analysis of experimental data, which were not normally distributed. Correlation analysis was performed with Spearman's rho testing, including all women irrespective of glucose status into the analysis. The results are presented as median (interquartile range). Values of P < .05 were considered statistically significant.
Results
Study participants
Twenty women with GDM and 18 normoglycemic (control) women were selected for this study. The women were matched for maternal BMI, gestational age at delivery, sex of the baby, and birth weight. There were no significant differences between the groups in other clinical characteristics except for blood glucose values in response to the 50-g glucose challenge test (Table 1).
Participant Characteristics
| Control | GDM | P Value | |
|---|---|---|---|
| n | 18 | 20 | ND |
| Maternal age, y | 32.7 ± 4.5 | 32.6 ± 3.3 | .96 |
| Prepregnancy BMI, kg/m2 | 25.0 ± 5.7 | 28.7 ± 8.7 | .14 |
| Systolic BP, mm Hg | 115.2 ± 11.9 | 116.6 ± 15.7 | .75 |
| Diastolic BP, mm Hg | 72.4 ± 9.6 | 69.47 ± 12.6 | .44 |
| Gestational age at delivery, d | 271.7 ± 4.6 | 274.6 ± 7.3 | .17 |
| Baby birth weight, g | 3466 ± 334 | 3502 ± 440 | .78 |
| Baby customized birth centile | 57.3 ± 28.8 | 52.7 ± 27.9 | .64 |
| Sex of the baby (males/females) | 8/10 | 8/12 | ND |
| Control | GDM | P Value | |
|---|---|---|---|
| n | 18 | 20 | ND |
| Maternal age, y | 32.7 ± 4.5 | 32.6 ± 3.3 | .96 |
| Prepregnancy BMI, kg/m2 | 25.0 ± 5.7 | 28.7 ± 8.7 | .14 |
| Systolic BP, mm Hg | 115.2 ± 11.9 | 116.6 ± 15.7 | .75 |
| Diastolic BP, mm Hg | 72.4 ± 9.6 | 69.47 ± 12.6 | .44 |
| Gestational age at delivery, d | 271.7 ± 4.6 | 274.6 ± 7.3 | .17 |
| Baby birth weight, g | 3466 ± 334 | 3502 ± 440 | .78 |
| Baby customized birth centile | 57.3 ± 28.8 | 52.7 ± 27.9 | .64 |
| Sex of the baby (males/females) | 8/10 | 8/12 | ND |
Abbreviations: BP blood pressure; ND, not determined. Data are presented as mean ± SD, and differences are analyzed by unpaired t test.
Participant Characteristics
| Control | GDM | P Value | |
|---|---|---|---|
| n | 18 | 20 | ND |
| Maternal age, y | 32.7 ± 4.5 | 32.6 ± 3.3 | .96 |
| Prepregnancy BMI, kg/m2 | 25.0 ± 5.7 | 28.7 ± 8.7 | .14 |
| Systolic BP, mm Hg | 115.2 ± 11.9 | 116.6 ± 15.7 | .75 |
| Diastolic BP, mm Hg | 72.4 ± 9.6 | 69.47 ± 12.6 | .44 |
| Gestational age at delivery, d | 271.7 ± 4.6 | 274.6 ± 7.3 | .17 |
| Baby birth weight, g | 3466 ± 334 | 3502 ± 440 | .78 |
| Baby customized birth centile | 57.3 ± 28.8 | 52.7 ± 27.9 | .64 |
| Sex of the baby (males/females) | 8/10 | 8/12 | ND |
| Control | GDM | P Value | |
|---|---|---|---|
| n | 18 | 20 | ND |
| Maternal age, y | 32.7 ± 4.5 | 32.6 ± 3.3 | .96 |
| Prepregnancy BMI, kg/m2 | 25.0 ± 5.7 | 28.7 ± 8.7 | .14 |
| Systolic BP, mm Hg | 115.2 ± 11.9 | 116.6 ± 15.7 | .75 |
| Diastolic BP, mm Hg | 72.4 ± 9.6 | 69.47 ± 12.6 | .44 |
| Gestational age at delivery, d | 271.7 ± 4.6 | 274.6 ± 7.3 | .17 |
| Baby birth weight, g | 3466 ± 334 | 3502 ± 440 | .78 |
| Baby customized birth centile | 57.3 ± 28.8 | 52.7 ± 27.9 | .64 |
| Sex of the baby (males/females) | 8/10 | 8/12 | ND |
Abbreviations: BP blood pressure; ND, not determined. Data are presented as mean ± SD, and differences are analyzed by unpaired t test.
Gene expression
Three RNA samples from the GDM group and two from the control group were removed due to insufficient RNA quality, leaving 18 women with and 15 women without GDM for gene expression analysis. Placental mRNA expression of FGF21 had high interindividual variability, which was more pronounced in the GDM group (range 0.21–5944 AU) than in the control group (range 0.06–12.13 AU). Women with GDM had a significantly higher median expression of FGF21 in placenta [8.15 (interquartile range 2.61–76.75)] than control women [0.75 (0.22–3.87), P < .001] (Figure 1A). FGF21 mRNA expression levels were not correlated to maternal prepregnancy BMI (r = −0.005, P = .98).
Placental gene expression of FGF21 (A), its receptors and cofactor (B), the transcription factors PPARA and PPARG (C), and the glucose transporters GLUT1, -3, and -4 (D). mRNA expression is normalized to the expression of the endogenous control TBP and adjusted for cellular composition. Boxes represent median with interquartile range, whiskers at 2.5th and 97.5th percentiles. Gray boxes, Control, normoglycemic pregnancies (n = 19); white boxes, GDM pregnancies (n = 19). **, P < .01.
Placental mRNA expression for receptors to FGF21
FGF receptor (FGFR)-1, FGFR2, FGFR3, and FGFR4 were detected in both groups (Figure 1B). There were no significant differences in the expression of any of the four receptor isoforms studied. The median mRNA expression in control placentae was 0.91 (0.52–1.39) for FGFR1, 0.84 (0.39–2.43) for FGFR2, 1.02 (0.33–1.70) for FGFR3, and 0.92 (0.54–2.47) for FGFR4. In GDM placentae, the median mRNA expression values were 1.43 (0.64–3.64) (P = .25); 1.72 (1.05–3.43) (P = .10); 0.97 (0.49–3.95) (P = .54); and 1.17 (0.61–2.67) (P = .52) for FGFR1, FGFR2, FGFR3, and FGFR4, respectively.
FGFR function is enhanced by binding to the cofactor β-klotho. Placental mRNA expression of β-klotho (KLB) was detected with large interindividual variability. The overall KLB mRNA expression was increased in the placenta of GDM women [2.45 (0.65–27.41)] compared with normoglycemic women [0.59 (0.23–12.93), P = .05] (Figure 1B).
Because FGF21 gene expression is regulated by PPARα in liver and PPARγ in adipose tissue, the mRNA expression levels of these transcription factors were studied in placenta. Placental expression of PPARα was higher in GDM women [2.94 (0.74–6.76)] than in control women [0.69 (0.41–2.14), P < .05]. Placental mRNA expression of PPARγ was not different between GDM or control pregnancies: 1.85 (0.93–4.28) vs 0.94 (0.42–1.87) (P = .09 for GDM vs control pregnancies, respectively) (Figure 1C). Moreover, there was a significant correlation between FGF21 and PPARα gene expression: Spearman's rho = 0.43 (P < .01). Because both diabetes in pregnancy and FGF21 have been shown to alter the expression of glucose transporters in the placenta and adipocytes, respectively (19, 20), we analyzed placental mRNA expression of GLUT1, -3, and -4 in our samples (Figure 1D). GLUT1 mRNA expression was not significantly different in GDM and control pregnancies: 2.79 (0.19–7.71) vs 1.28 (0.39–6.77, P = .67); GLUT3 mRNA was increased in GDM placentas: 2.74 (1.17–5.21) vs 0.60 (0.43–2.03) (P < .05); and GLUT4 mRNA was significantly increased in GDM placentas: 3.20 (1.11–4.85) vs 0.66 (0.57–1.64) (P < .01). FGF21 and GLUT4 mRNA expression levels were significantly correlated: Spearman's rho = 0.36 (P < .05).
Protein expression
Protein expression of FGF21 also showed a high interindividual variability in placenta (see Figure 2A). FGF21 protein expression in placenta is higher in women with GDM [2.89 (1.44–5.10)] than in control women [0.42 (0.05–1.98), P < .05] (Figure 2A). FGF21 protein expression was localized to syncytiotrophoblasts and endothelial cells but also to stromal cells including Hofbauer cells (Figure 2, B and C). There was no clear difference in the expression or localization between placentae from control women (Figure 2B) or those with GDM (Figure 2C). PPARα protein showed a trend for increased expression placenta from women with [1.52 (0.58–1.74)] vs without [0.90 (0.26–1.75)] GDM (P > .05) (Figure 2D). Placental β-klotho protein expression was not different between women with [0.97 (0.72–2.05)] vs without [1.37 (1.29–1.68)] GDM (P > .05) (Figure 2E).
Placental protein expression of FGF21. Representative Western blot shows the interindividual variability of FGF21 in green and the loading control B-actin in red (A). Densitometry analysis of FGF21 expression normalized to B-actin in placentae from normoglycemic women (n = 10) and women with GDM (n = 12). *, P < .05. Representative immunohistochemistry of FGF21 in placenta from normoglycemic women (B) and women with GDM (C) and a negative control placental section (F) is shown. Representative Western blot showing PPARα protein (D) and B-klotho (KLB) (E) expression in green and the loading control B-actin in red. Densitometry analysis of PPARα and B-klotho expression normalized to B-actin in placentae from normoglycemic women (n = 8) and women with GDM (n = 10) is shown. Boxes represent median with interquartile range, whiskers at 2.5th and 97.5th percentiles. Gray boxes, Control, normoglycemic pregnancies; white boxes, GDM pregnancies.
Serum levels
FGF21 levels in maternal serum at the end of the third trimester were measured. Serum FGF21 levels were similar in women with GDM [323 (75–921) pg/mL] and control women [269 (49–731) pg/mL, P = .81] (Figure 3A). Maternal serum concentrations were correlated to placental β-klotho mRNA levels (Spearman's rho = 0.49, P < .05) (Figure 3B) but not placental FGF21 mRNA expression levels, BMI, or infant birth weight.
Maternal FGF21 levels in serum. Gray box, Control, serum levels in normoglycemic women (n = 10); white box, GDM, serum levels in women with GDM (n = 10) (A). Correlation between maternal circulating FGF21 levels and placental β-klotho mRNA expression (B) is shown.
FGF21 was not detected in cord blood of any of the babies, irrespective of the glucose status of the mother.
Discussion
In this study, we established for the first time, the presence of FGF21 mRNA and protein in the placenta. FGF21 mRNA and protein were significantly increased in GDM placental tissue. However, there was no difference in maternal circulating FGF21 levels compared with normoglycemic pregnancies. These results are consistent with a previous report in a larger group of women who had samples assessed approximately 28 weeks' gestation (shortly after the diagnosis of GDM) and matched for fasting insulin levels (11). The circulating FGF21 levels were highly variable in this study in accordance with previous reports (11, 21, 22). There was large interindividual variation in FGF21 expression levels in placenta, which may be due the nature of sampling. In the present study, a sample obtained from the maternal side of the placenta was analyzed. Due to the large heterogeneity in placental anatomy, there could be variation in the tissue composition of the biopsies and thereby variation in FGF21 expression. However, the large interindividual variation remained after adjusting for the cellular composition of the biopsy. This indicates that the variation may result from other factors.
The FGF21 expression levels correlated with those of PPARα but not PPARγ. This suggests that FGF21 expression in the placenta is influenced by PPARα rather than PPARγ. Although we found a significant increase in PPARα mRNA levels and a trend for increased PPARα protein levels, a previous study has reported reduced protein but no change in mRNA levels of PPARα in GDM (23). That study involved seven women in each group with a lower mean BMI (control 20.6 and GDM 23.0 kg/m2) than the women involved in the current study (control 25.0 and GDM 28.7 kg/m2). In a recently published study in which women with GDM were subdivided according to normal BMI (21.1 kg/m2, n = 6) or overweight/high BMI (32.8 kg/m2, n = 6), a tendency toward higher PPARα mRNA was noted in women with GDM compared with normal-weight controls (21.9 kg/m2, n = 6), which was more pronounced in the overweight/obese group (24). Furthermore, protein levels of PPARα were significantly higher in the overweight/obese women with GDM in that study (24). Therefore, our findings of increased PPARα mRNA in women with GDM may contrast with other studies due to differences in BMI of women studied. The combined results of the latter study and the current study indicate that both BMI and GDM contribute to the up-regulation of PPARα in the placenta.
A previous larger study has found a correlation between maternal circulating FGF21 levels and markers of insulin resistance (increased homeostasis model assessment index of insulin resistance and decreased adiponectin) and dyslipidemia [higher triglycerides and lower high density lipoprotein (HDL) cholesterol] (11). There was no correlation between maternal BMI and placental FGF21 levels in the present study. Both plasma triglycerides and HDL cholesterol are regulated by hepatic PPARα, with plasma triglycerides decreasing and HDL cholesterol increasing (25). This may be partially mediated through FGF21 (13, 26). If this regulation is similar in the placenta, the increase in PPARα and FGF21 in GDM may reduce triglyceride but increase HDL cholesterol transfer to the fetus.
A recent study showed that there appears to be a circadian rhythm in circulating free fatty acids (FFAs) and FGF21, in which the peak in FFAs precedes the FGF21 peak by 3–4 hours (27). The placenta expresses many of the clock genes driving circadian rhythm, although their functionality has not been demonstrated in great detail in humans (28). Our samples were collected at the time of delivery (throughout the day and night). If placental FGF21 expression is regulated by circadian rhythm, the variation in the time of delivery may also contribute to the large interindividual variation seen in FGF21 expression. Detailed analysis of placental CLOCK gene expression and FGF21 expression in a large group of women could shed light on this. A valuable area of future research would involve the inclusion of a group of women with GDM with large-for-gestational-age babies to investigate whether placental FGF21 can prevent excessive fetal weight gain. Additionally, cell culture experiments exposing primary trophoblasts to FFAs, triglycerides, and HDL cholesterol could provide functional and mechanistic data on the regulation of FGF21 in trophoblasts.
Placental mRNA expression of the glucose transporters GLUT1, -3, and -4 were studied. The significant increase in placental GLUT3 and -4 mRNA expression in placentas from women with GDM as well as the positive correlation of placental GLUT4 mRNA with FGF21 mRNA indicate a potential mechanism by which changes in FGF21 can result in altered placental metabolism. In the literature, FGF21 has been noted to induce the expression of GLUT1 but not GLUT4 in adipocytes (20, 29, 30), whereas it did not change GLUT1 expression in hepatocytes (30), and the effects on GLUT3 have not been studied. The effect of FGF21 administration on GLUT expression in the placenta is unknown. In human placenta, GLUT3 is located on the trophoblast and is more efficient in transporting glucose to the fetus than GLUT1 due to its higher affinity for glucose (31). In rat placenta, GLUT3 mRNA expression increases over the course of pregnancy, which is compatible with the need for increased glucose transport to the growing fetus (32). Increased expression of GLUT3 in GDM could therefore lead to increased glucose supply to the fetus, which could contribute to increased growth and fat mass. Increased placental GLUT4 expression by prenatal stress has previously been reported in rats (33). In human placenta, GLUT4 has been shown to be present in syncytiotrophoblasts and stromal cells by some (34) but not all studies (35). Stimulation of healthy term placental explants with supraphysiological concentrations of insulin did not increase glucose uptake, which perhaps reflects the relatively low expression of GLUT4 receptors (36). Whether increased GLUT4 expression in GDM placentas increases the insulin sensitivity or affects glucose uptake needs to be studied in more detail.
There was an absence of detectable FGF21 in fetal cord blood in any of the mother-baby dyads studied, independent of maternal circulating FGF21 levels, placental FGF21 expression levels, or maternal glucose status. It has previously been shown that FGF21 levels are very low in mouse fetal plasma (13), indicating a species difference between humans and mice. The neonatal mouse levels rose quickly after birth with a peak between postnatal days 2 and 6, and thereafter the FGF21 levels returned to adult levels. This study showed that food intake, and specifically the fatty acids present in high amounts in breast milk, is necessary for the induction of hepatic FGF21 mRNA expression at birth in mice. The signal transduction pathway is through activation of PPARα. Fetal and newborn FFA levels in plasma are very low in rodents but increase rapidly after suckling. The absence of FGF21 in cord blood suggests that circulating FGF21 does not regulate fetal metabolism. It is not clear whether fetal tissues express FGF21 and whether this expression is affected by feeding postnatally.
In accordance with previous results in human placentae from uncomplicated pregnancies, we detected mRNA for FGFR1, -2, and -3 (37) as well as mRNA and protein for the coreceptor β-klotho. mRNA for FGFR4 was also detected by Anteby et al (37). FGFR1 mRNA was localized by in situ hybridization to syncytiotrophoblasts in term placenta (17). Protein for FGFR1 was detected by immunohistochemistry in vascular endothelial cells and stromal cells, whereas FGFR4 protein was present in Hofbauer cells. FGFR2 and FGFR3 protein was not detectable (17, 37). The FGF receptors are not specific for FGF21 but bind a variety of FGF isoforms such as FGF2 (17), FGF7 (38), and FGF10 (37). The FGF pathway in placenta stimulates the proliferation of fibroblasts, the vascular endothelium, smooth muscle, and neural cells (17). FGF2 mRNA and protein are increased in the placentae of women with type 1 diabetes and GDM (17, 39). FGFR1 mRNA and protein are also increased in the placentae of women with type 1 diabetes (17), but in the placentae of women with GDM, there was no increase in FGFR1 mRNA levels in the present study. The localization of FGFR1 on the vascular endothelium and its presence in many fetal tissues (40) indicates that FGF family members do affect the fetus. A recent study indicated the preferential binding of FGF21 to FGFR1-KLB complexes in adipose tissue (8). Binding to an FGFR1-KLB complex in placental vascular endothelium could be a potential way for FGF21 to indirectly influence fetal growth and metabolism despite the lack of FGF21 protein in the fetal cord blood.
The maternal serum samples were obtained in late pregnancy, from 36 weeks' gestation onward. Although two of the GDM samples were in the fasting state, this could not be ascertained for the other samples. Removal of the two GDM samples, however, did not alter the results of the analysis. The range of detection for serum FGF21 is from 30 to 2000 pg/mL. It could therefore be that the cord blood levels of FGF21 are below 30 pg/mL.
In conclusion, we have shown that FGF21 mRNA and protein are expressed with high interindividual variability in human placenta. FGF21 is localized to syncytiotrophoblasts, endothelial cells, and stromal cells. FGF21 mRNA and protein levels are increased in placentae from women with GDM compared with normoglycemic women. PPARα mRNA levels are also increased in women with GDM and PPARα mRNA and FGF21 mRNA are correlated. Serum levels of FGF21 are not different between normoglycemic women and women with GDM. FGF21 could not be detected in cord blood of infants independent of maternal glucose status. These results indicate that PPARα may regulate FGF21 expression in the placenta. There is no evidence for FGF21 secretion from the placenta into the fetal circulation. However, the expression of FGF21 and its receptors in the placenta suggest that it could affect placental metabolism. This may contribute to downstream regulation of fetal growth and metabolism.
Acknowledgments
This work was supported by the National Health and Medical Research Council (Grant 569693) and the Australasian Diabetes in Pregnancy Society Novo Nordisk Research grant (to M.D.N.). M.D.N. was supported by a Patricia Dukes Fellowship from the Royal Brisbane and Women's Hospital Foundation.
Disclosure Summary: The authors have nothing to disclose.
Abbreviations
- BMI
body mass index
- FFA
free fatty acid
- FGF
fibroblast growth factor
- FGFR
FGF receptor
- GDM
gestational diabetes mellitus
- GLUT
glucose transporter
- HDL
high-density lipoprotein
- KLB
β-klotho
- PPAR
peroxisome proliferator-activated receptor
- TBP
TATA-box binding protein.
Reference
