Obesity and metabolic syndrome are associated with systemic inflammation and increased perinatal morbidity. Metformin improves metabolic and inflammatory biomarkers in nonpregnant adults. Using in vivo and in vitro models, we examined the effect of metformin on maternal and fetal inflammation.
Female Wistar rats (6-7 weeks old) were fed a normal diet (NORM) or a high-fat/high-sugar diet (HCAL) for 5-6 weeks to induce obesity/metabolic syndrome. After mating with NORM-fed male rats, one-half of the HCAL-fed female rats received metformin (300 mg/kg, by mouth daily). All dams continued their respective diets until gestational day 19, at which time maternal and fetal outcomes were assessed. Maternal and fetal plasma and placentas were analyzed for metabolic and inflammatory markers. Cultured human placental JAR cells were pretreated with vehicle or metformin (10 μmol/L-2.5 mmol/L) before tumor necrosis factor α (TNF-α; 50 ng/mL), and supernatants were assayed for interleukin-6 (IL-6).
HCAL rats gained more prepregnancy weight than NORM rats ( P = .03), had higher levels of plasma insulin and leptin, and exhibited dyslipidemia ( P < .05). Fetuses that were exposed to the HCAL diet had elevated plasma IL-6, TNF-α, and chemokine (C-C motif) ligand 2 levels ( P < .05) and enhanced placental TNF-α levels ( P < .05). Maternal metformin did not impact maternal markers but significantly decreased diet-induced TNF-α and chemokine (C-C motif) ligand 2 in the fetal plasma. Finally, metformin dose-dependently reduced TNF-α–induced IL-6 and IκBα levels in cultured placental JAR cells.
Diet induced-obesity/metabolic syndrome during pregnancy significantly enhanced fetal and placental cytokine production; maternal metformin reduced fetal cytokine levels. Similarly, metformin treatment of a placental cell line suppressed TNF-α–induced IL-6 levels by NFκB inhibitor.
Metabolic syndrome is defined as a cluster of symptoms that are associated with the development of serious health conditions, such as type 2 diabetes mellitus and cardiovascular disease. The National Heart, Lung, and Blood Institute in collaboration with the American Heart Association has identified components of metabolic syndrome: obesity (primarily abdominal), insulin resistance, atherogenic dyslipidemia (elevated triglycerides, reduced high-density lipoprotein [HDL] cholesterol), hypertension, and a proinflammatory/prothrombotic state. The prevalence of metabolic syndrome in the United States is 16% among women who are <40 years old. Overweight and obese women are approximately 5 and 17 times, respectively, more likely than normal weight women to exhibit metabolic syndrome.
Obesity during pregnancy has been implicated in increased susceptibility to gestational diabetes mellitus, preeclampsia, stillbirth, cesarean deliveries, and poor maternal wound healing. Consumption of high-calorie diets by humans and animals can lead to the development of obesity and metabolic dysfunction. In pregnancy, high-calorie diets promote numerous factors that are associated with increased maternal and fetal morbidity and death, which include obesity, inflammation, aberrant lipid metabolism, and insulin resistance.
In nonpregnant obese individuals, antidiabetic drugs may reduce the risks of metabolic syndrome and insulin resistance. Specifically, metformin is one drug that can decrease the adverse consequences of obesity-related insulin resistance by improving hepatic insulin sensitivity. It has been given to pregnant women outside of the United States since the 1970s and is now increasing in acceptance as an alternative treatment of infertility and diabetes mellitus in pregnancy in the United States. Finally, metformin has been shown to exert antiinflammatory activity. Thus, the advantage of metformin for treating insulin resistance is its potential to moderate metabolic dysfunction and inflammation.
Because metformin modifies both insulin resistance and inflammation, we investigated whether metformin modulated maternal and fetal metabolic dysfunction in an acute diet-induced obesity model in pregnant rats.
Materials and Methods
The Institutional Animal Care and Use Committee (IACUC) approved all animal studies before animal experimentation (IACUC #2010-031). Female Wistar rats (6-7 weeks old; Taconic Farms, Germantown, NY) were acclimatized initially with free access to standard rat chow and water for at least 72 hours. Rats were assigned randomly to 1 of 2 ad libitum diets: (1) control rat diet or normal-fed (NORM; n = 10) or (2) high-calorie diet (HCAL; n = 20) for 5 weeks. The HCAL diet consisted of 33% ground commercial rat diet, 33% full fat sweetened condensed milk, 7% sucrose, and 27% water, as previously described. After 5-6 weeks on their diets, lean and acutely obese female rats (maintained on their respective diets) were mated with NORM-fed male Wistar rats. On gestation day 1, one-half of the dams that were fed the HCAL diet received metformin (300 mg/kg, by mouth daily). All dams continued their respective diets (with/without metformin) throughout gestation. Rats were weighed immediately before diet, before mating, and then on gestational days 1, 5, 12, and 19. Maternal plasma (nonfasting) was collected before mating and on gestational day 19 by retroorbital bleeding. On gestational day 19, dams were euthanized by CO 2 inhalation followed by exsanguination by cardiac puncture with heparinized needles/syringes. Fetuses that were delivered by cesarean section were euthanized by decapitation, and blood (nonfasting) was collected into heparinized capillary tubes. Maternal weight gain, placental weight, and fetal outcomes (fetal weights and litter sizes) were assessed. Maternal and fetal blood (pooled from each dam) was centrifuged, and plasma was isolated. Maternal and fetal plasma and placentas (free of decidua) were flash frozen in liquid nitrogen and stored at –80°C.
Assessment of fetal rat plasma and placental cytokines
After a brief centrifugation step, fetal plasma samples were analyzed for cytokines with the use of customized rat cytokine 5-plex kits (chemokine [C-X-C motif] ligand 1 [CXCL1]; chemokine [C-C motif] ligand 2 [CCL2]; interleukin – 6 [ IL-6]; interleukin-1 beta [IL-1β]; and tumor necrosis factor alpha [TNF-α]; Meso Scale Discovery, Rockville, MD), according to the manufacturer’s directions. This particular group of cytokines/chemokines is representative of the proinflammatory state that is associated with obesity and metabolic syndrome and could be assessed with multiplex formats. After reading on the Sector Imager 2400 plate reader (Meso Scale Discovery), raw data were measured as electrochemiluminescence signals that were detected by photo-detectors and analyzed with the use of the Discovery Workbench software (version 3.0; Meso Scale Discovery). A 4-parameter logistic fit curve that was generated for each cytokine with these standards was used to determine the concentration of the individual cytokines in each sample (sensitivity, <3 pg/mL for each cytokine). Frozen rat placentas were homogenized with freshly prepared lysis buffer (150 mmol/L NaCl, 20 mmol/L Tris [pH 7.5], 0.25 % Tx-100, 10 mmol/L NaF, and phosphatase/protease inhibitor cocktail [Thermo Scientific, Waltham, MA]). After centrifuging was completed, cell-free homogenates were analyzed for CXCL1, CCL2, IL-1β, IL-6, and TNF-α, with the Meso Scale Discovery technology described earlier. Placental cytokine data were normalized for protein concentrations, as determined by the Bradford method (Bio-Rad Laboratories, Hercules, CA).
Assessment of maternal plasma cytokines, adipokines, lipids, and metabolic hormones
Maternal nonfasting plasma samples were analyzed for CXCL1, IL-1β, IL-6, insulin, leptin, and TNF-α levels with the use of Luminex XMAP technology (Millipore, St. Louis, MO). In addition, maternal nonfasting plasma samples were assayed for HDL cholesterol, low-density lipoprotein cholesterol, and triglyceride levels by the Core Laboratories of the North Shore-LIJ Health System.
In vitro studies with the use of a human placental cell line
Human placental choriocarcinoma JAR cells (ATCC, Manassas VA) were cultured in 96-well plates (1 × 10 5 cells/well) in Roswell Park Memorial Institute medium that contained 10% fetal bovine serum and 1% penicillin-streptomycin-glutamine. JAR cells were pretreated with either vehicle or metformin (0.01-2.5 mmol/L; Calbiochem, San Diego, CA) before stimulation with recombinant human TNF-α (0-50 ng/mL).
After overnight stimulation, IL-6 levels in cell-free culture supernatants were measured by enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN). Cell viability was determined after the JAR cells were treated with vehicle or metformin for 24 hours with the sulforhodamine B cytotoxicity assay method. To evaluate mechanisms of inhibition, JAR cells were pretreated with an NFκB inhibitor, Bay 11-7082 (20 μmol/L; Calbiochem) for 15 minutes before TNF-α stimulation (50 ng/mL).
After overnight stimulation, IL-6 levels were assessed as described earlier. Western blotting was performed with JAR cell lysates that were prepared after treatment with TNF-α (in the presence and absence of metformin) using an antibody that is specific for IκBα (Cell Signaling Technology, Danvers, MA), which is the inhibitor of NFκB nuclear localization and activation.
Prestudy power analysis estimated the need for approximately 6 rats in each group for comparison. The negative control group (NORM) was designated to be those dams that received the normal diet. In addition, only lean dams that gained <1 SD above the overall mean body weight percent increase for this group were included in the analyses. This allowed us to maintain an optimal control group for analysis by excluding animals with a weight gain that approached the mean weight gain (within 1 SD) of the HCAL-fed group. Maternal and fetal outcome data (weight gain, fetal weight, placental weight, litter size, and maternal lipids) were analyzed, and the medians were compared with the use of nonparametric tests. Maternal and fetal cytokine, insulin, and leptin data were normalized by a log transformation; the geometric means were compared with the use of the Student t test, and significance was determined to be a probability value of < .05. Two separate comparisons were performed in parallel. First, we analyzed the difference between the NORM- vs HCAL-fed groups. Preconception data (weight gain, insulin, and leptin) were combined for all dams that received either NORM or HCAL diets because, at this point, there were no additional differences in treatment (ie, metformin was not started). Then, we examined the effect of treatment on exposure to the HCAL diet (HCAL + metformin vs HCAL alone).
For the JAR cell data, IL-6 levels among the various TNF-α and metformin concentrations vs vehicle alone were analyzed with analysis of variance. Pairwise comparisons of various concentrations of TNF-α or metformin to vehicle-treated cells (no metformin) were performed; significance was determined to be a probability value of < .05.
HCAL diet promotes weight gain and metabolic dysregulation
HCAL-fed rats gained significantly more weight during the prepregnancy period than NORM diet-fed rats (61.2% ± 15.7% vs 47% ± 10.4%; P = .03; Figure 1 ). Although HCAL-fed dams gained more weight than NORM-fed dams during pregnancy, pregnancy weight gain (unadjusted and adjusted for litter size) for the 2 groups was not significantly different ( Figure 1 ). Consistent with the observed weight increase that is associated with the HCAL diet, nonfasting plasma leptin levels were significantly higher in HCAL-fed animals after 5-6 weeks of diet (preconception; 2795 ± 980 pg/mL vs 9687 ± 4092 pg/mL; P < .0001; Table 1 ). In addition, at gestational day 19, HCAL-fed dams had significantly higher plasma triglycerides, lower HDL concentrations, and elevated cholesterol:HDL ratios ( P < .05; Table 1 ). The HCAL diet also significantly increased plasma insulin levels when measured on gestational day 19 ( P < .05; Table 1 ). Surprisingly, on gestational day 19, maternal plasma cytokine levels were not different in HCAL dams compared with control dams ( Table 1 ). Rats that were fed the HCAL diet had elevated plasma leptin levels compared with NORM-fed rats when assessed before pregnancy; as predicted, pregnancy was associated with a significant increase in circulating leptin levels in NORM-fed dams ( Table 1 ). By contrast, leptin levels in the HCAL-fed rats did not significantly increase over gestation ( Table 1 ). Maternal metformin administration to HCAL-fed dams during pregnancy did not significantly affect maternal weight gain, lipid profiles, plasma insulin, leptin levels, or circulating cytokine levels ( Table 1 ).
|Analyte||Diet||P value a|
|Normal (control)||High-fat/high-sugar||High-fat/high-sugar + metformin|
|Preconception (gestational day 0)|
|Insulin, pg/mL b||1101.2 ± 647 (10)||1700.8 ± 1212.6 (19)||—||A: .13|
|Leptin, pg/mL b||2795 ± 980.41 (10)||9687.8 ± 4092.4 (19)||—||A: .0001|
|Gestation day 19|
|High-density lipoprotein, mg/dL b||34.6 ± 13 (8)||22.5 ± 11.5 (10)||19.3 ± 9.1 (9)||A: .03/B: .55|
|Triglycerides, mg/dL b||379 ± 152 (8)||597 ± 208 (10)||689 ± 265 (9)||A: .007/B: .39|
|Cholesterol/high-density lipoprotein ratio b||2 ± 0.7 (8)||3.8 ± 2.1 (10)||4.5 ± 2.6 (9)||A: .03/B: .62|
|Insulin, pg/mL b||895.2 ± 444.3 (6)||1715.4 ± 838.8 (6)||1621.3 ± 719.9 (6)||A: .026/B: .82|
|Leptin, pg/mL b||4070.2 ± 1274.5 (6)||8825.2 ± 1372.5 (6)||10690.1 ± 4017.9 (6)||A: .0004/B: .37|
|Interleukin-6, pg/mL b||ND||ND||ND|
|Tumor necrosis factor α, pg/mL b||ND||ND||ND|
|Chemokine (C-C motif) ligand 2, pg/mL b||119.9 ± 29.7 (6)||96.3 ± 28.5 (6)||89.9 ± 44.2 (6)||A: .23/B: .58|
|Interleukin-1β, pg/mL b||74 ± 26.7 (6)||106.4 ± 51.6 (6)||199.3 ± 179.4 (6)||A: .68/B: .10|
HCAL diet promotes fetal and placental inflammation that is reduced by maternal metformin treatment
Although fetal weights on gestational day 19 were similar among the HCAL and NORM groups, HCAL-fed dams had slightly smaller placental weights ( P = .07; Table 2 ). The average fetal weight of animals that were exposed to the HCAL + metformin diet was 1.38 ± 0.35 g], which was not significantly different from those that were exposed to HCAL alone (1.37 ± 0.29 g). Placental weight was higher on average among HCAL + metformin dams (0.39 ± 0.06 g) when compared with HCAL-alone dams (0.35 ± 0.08 g), but this was not significant ( Table 2 ). Few fetal resorptions/fetal deaths were observed in all groups.
|Characteristic||Diet||P values a|
|Normal (control)||High-fat/high-sugar||High-fat/high-sugar + metformin|
|Fetal weight, g b||1.45 ± 0.21 (8)||1.37 ± 0.29 (10)||1.38 ± 0.35 (9)||A: 0.19/B: 0.78|
|Fetuses/litter, n b||12.4 ± 1.9 (8)||11.6 ± 2.9 (10)||11.85 ± 2.7 (9)||A: 0.65/B: 0.96|
|Placental weight, g b||0.41 ± 0.04 (8)||0.35 ± 0.08 (10)||0.39 ± 0.06 (9)||A: 0.07/B: 0.16|
|Dams with ≥1 fetal resorptions, n||1 (8)||3 (10)||2 (9)||A: NS /B: NS|
When compared with the fetal pups from the NORM-diet group, the fetal pups from the HCAL-diet group had elevated plasma CCL2, IL-6, and TNF-α levels ( P < .05; Figure 2 ). There was no significant effect of HCAL diet exposure on fetal plasma CXCL1 or IL-1β levels ( Figure 2 ). Maternal metformin treatment significantly decreased HCAL-induced CCL2 and TNF-α levels in the fetal plasma ( Figure 2 ). Likewise, HCAL-induced fetal plasma IL-6 levels were reduced by maternal metformin administration; however, this effect was not significant ( Figure 2 ). Placental CCL2, CXCL1, IL-1β, and IL-6 levels were not significantly different between the HCAL-diet and NORM-diet groups ( Table 3 ). However, TNF-α levels were significantly elevated ( P < .05) in the placentas that were obtained from the HCAL-diet group when compared with NORM-fed group ( Table 3 ). Although metformin treatment reduced HCAL-induced placental TNF-α levels, the effect was not quite significant ( P = .06; Table 3 ).
|Analytes||Diet a||P value b|
|Normal (control)||High-fat/high-sugar||High-fat/high-sugar + metformin|
|Chemokine (C-C motif) ligand 2, pg/mg||15.0 ± 2.6 (4)||17.4 ± 3.1 (5)||21.5 ± 13.4 (7)||A: .26/B: .81|
|Chemokine (C-X-C motif) ligand 1, pg/mg||34.3 ± 16.7 (4)||32.1 ± 15.9(5)||33.5 ± 14.0 (7)||A: 1.0/B: .73|
|Interleukin-1β, pg/mg||12.0 ± 2.2 (4)||14.5 ± 2.5 (5)||13.0 ± 3.0 (7)||A: .19/B: 1.0|
|Interleukin-6, pg/mg||32.1 ± 7.2 (4)||40.5 ± 9.8(5)||31.8 ± 7.2 (7)||A: .16/B: .16|
|Tumor necrosis factor α, pg/mg||6.2 ± 1.3 (4)||9.1 ± 1.3 (5)||6.8 ± 1.5 (7)||A: .0317/B: .0635|
Metformin reduces TNF-α–induced IL-6 production by JAR cells through the NFκB pathway
Using the placental JAR cell line, we observed that TNF-α treatment (0.1-50 ng/mL) enhanced IL-6 production in a dose-dependent manner ( Figure 3 , A). Metformin treatment (≤2.5 mmol/L for 18 hours) did not affect JAR cells viability ( Figure 3 , B). When JAR cells were pretreated with metformin at concentrations of 0.01-2.5 mmol/L, TNF-α–induced IL-6 production was significantly reduced in a dose-dependent manner. With 500 μmol/L metformin, TNF-α–induced IL-6 production was reduced by 41.9% ( P < .05; Figure 3 , C); with higher doses of metformin (1 and 2.5 mmol/L), IL-6 production was reduced by 53.6% and 76.4%, respectively ( P < .001; Figure 3 , C). Similar to the suppression observed with metformin, treatment of JAR cells with Bay 11, an NFκB inhibitor, significantly reduced TNF-α–induced IL-6 production ( Figure 3 , D). To confirm that metformin reduced TNF-α–induced IL-6 production through the NFκB pathway, we examined the effect of metformin on the degradation of IκBα (the inhibitor of NFκB activation) using JAR cells. We found that metformin (2.5 mmol/L) significantly reduced TNF-α–induced IκBα degradation ( Figure 3 , E).