Metformin is becoming a popular treatment before and during pregnancy, but current literature on in utero exposure to metformin lacks long-term clinical trials and mechanistic studies. Current literature on the effects of metformin on mature pancreatic β-cells highlights its dual, opposing, protective, or inhibitory effects, depending on metabolic environment. However, the impact of metformin on developing human pancreatic β-cells remains unknown. In this study, we investigated the potential effects of metformin exposure on human pancreatic β-cell development and function in vitro. In the absence of metabolic challenges such as high levels of glucose and fatty acids, metformin exposure impaired the development and function of pancreatic β-cells, with downregulation of pancreatic genes and dysfunctional mitochondrial respiration. It also affected the insulin secretion function of pancreatic β-cells. These findings call for further in-depth evaluation of the exposure of human embryonic and fetal tissue during pregnancy to metformin and its implications for long-term offspring health.

Maternal hyperglycemia is known to have a negative impact on conception, pregnancy progression, and embryonic and fetal development (1). Over the last 40 years, the antihyperglycemic agent metformin has been gaining popularity as a treatment before and during pregnancy for women with pregestational type 2 diabetes, gestational diabetes mellitus (GDM), or polycystic ovary syndrome because of its low cost, simple storage, lack of risk of inducing hypoglycemia, and easy administration compared with insulin injections (2). However, the safe use of metformin during pregnancy is still an underresearched issue, despite the fact that metformin readily crosses the placenta and circulates in the fetus (3,4). Previous reports have shown that metformin concentrations in umbilical cords can be at least half or nearly the same as the corresponding maternal plasma concentrations (3,5), with possibly higher local tissue concentrations where there is cellular drug accumulation (6). Unfortunately, current literature on the extent of metformin exposure in utero and its impact on human embryonic and fetal development is still limited. Although no short-term adverse effects on neonates have been reported, recent long-term studies have found that children exposed to in utero metformin treatment might have slightly higher BMI, larger subcutaneous fat deposits, and higher risk of childhood obesity (7,8), suggesting that metformin could have more subtle effects on development that may not be apparent at delivery.

To investigate the effects of prolonged metformin exposure on human embryonic and fetal development, we treated differentiating human pancreatic β-like cells and mature β-cells with therapeutic doses of metformin before evaluating their expression profiles and functionality. We found metformin to impair the development and function of human pancreatic β-like cells, with RNA sequencing (RNA-seq) analyses confirming a downregulation of genes involved in hormone production, transport and secretion. We also found that therapeutic doses of metformin actually impeded mitochondrial function in human pancreatic β-cells. Overall, we revealed that metformin elicited unexpected adverse effects on human pancreatic β-cell differentiation and function. Our findings suggest the need to further scrutinize the molecular effects of human embryonic and fetal exposure to metformin during pregnancy and their possible long-term health implications.

Human Embryonic Stem Cell Culture

H9 human embryonic stem cells (hESCs) (National Institutes of Health approval number NIHhESC-10–0062) were maintained at 37°C with 5% carbon dioxide and regularly confirmed to be mycoplasma free (MycoAlert Mycoplasma Detection Kit [catalog no. LT07-318]; Lonza Bioscience). Undifferentiated hESCs were maintained in either feeder-dependent or chemically defined TeSR-E8 medium. Briefly, for the feeder-dependent condition, hESCs were cultured on irradiated mouse embryonic fibroblast feeder cells in DMEM/F12 medium with GlutaMAX supplement (Gibco catalog number 11574466), Pyruvate (Invitrogen catalog number 10565042), 20% KnockOut Serum Replacement (Invitrogen catalog number 10828028), and 1× nonessential amino acids (Invitrogen catalog number 11140050), supplemented with 10 ng/mL fibroblast growth factor 2 (MACS catalog number 130-093-840). For the chemically defined condition, hESCs were cultured in TeSR-E8 medium (STEMCELL Technologies catalog number 05990) on plates coated with 0.1% gelatin (Sigma-Aldrich catalog number G1890), followed by high-glucose DMEM (HyClone catalog number SH30022FS) with 10% v/v FBS (HyClone catalog number 11521851). For both conditions, media was changed daily, and cells were passaged every ˜7 days using manual colony picking (feeder-dependent condition only) or chemical passaging by ReLeSR (1×) (STEMCELL Technologies catalog number 05873).

Pancreatic β-Cell Differentiation

H9 hESCs were differentiated to definitive endoderm, early pancreatic progenitors, pancreatic and duodenal homeobox 1–positive (PDX1+) NKX6.1+ pancreatic progenitors, and β-like cells following a previously published protocol (9). Using TrypLE Express Enzyme, 80–90% of confluent hESCs in TeSR-E8 condition were dissociated into single cells and replated at 106 cells/mL in TeSR-E8 medium supplemented with 10 μmol/L Y-27632 in suspension culture. The medium was changed after 24 h, and differentiation was initiated 48–72 h after splitting (designated as day 0) on a rotating platform. Metformin was added from an aqueous stock into differentiation media for every media change from day 0, 13, or 20 onward to the end of differentiation.

Pancreatic β-Cell Culture

EndoC-βH1 cells (fetal-derived β-cell line) (Univercell-Biosolutions S.A.S.) were cultured in low-glucose DMEM (1 g/L) with Pyruvate (Gibco catalog number 11885) with 2% BSA (Sigma-Aldrich catalog number A9418), 0.5 mL β-mercaptoethanol (1,000×) (Gibco catalog number 21985023), 10 mmol/L nicotinamide (100×) (Sigma-Aldrich catalog number N3376), 1% GlutaMAX supplement (Gibco), 6.7 ng/mL sodium selenite (Sigma-Aldrich catalog number 214485-100G) and 5.5 μg/mL transferrin (Sigma-Aldrich catalog number T8158-100MG). Cells were seeded onto 6-cm plates coated with 2 mL high-glucose DMEM (HyClone) with 4 μL fibronectin (Sigma-Aldrich catalog number F0895) and 20 μL ECM (Sigma-Aldrich catalog number E1270). The cells were split every ˜7 days using Trypsin-EDTA 0.25% (Gibco catalog number 25300054) at a density of 2–4 million cells/6-cm plate. MIN6 cells (mature mouse pancreatic insulinoma cell line) were cultured in high-glucose DMEM with 4.5 g/L glucose, 15% heat-inactivated GE Healthcare HyClone FBS, 2 mmol/L l-glutamine (or GlutaMAX supplement [Gibco]), and 50–55 μmol/L β-mercaptoethanol (1,000×) (Gibco). The medium was changed every 2–3 days, and cells were passaged every ˜7 days using Trypsin-EDTA 0.25%.

Pancreatic Islet Culture

The use of fully mature cadaveric human adult islets (Clinical Islet Laboratory, University of Alberta Hospital, Edmonton, Alberta, Canada) from healthy donors was approved by the National University of Singapore Institutional Review Board (number B-14–149; Singapore). Informed consent was obtained from next of kin of the donor by the Clinical Islet Laboratory. The islets were cultured in Miami media (Corning catalog number 98-021-CV) containing 2 mM glutathione overnight before use. The islets were then cultured in Miami media with metformin treatment for 72 h before glucose-stimulated insulin secretion (GSIS) assay.

Pancreatic β-Like Cell Clump Bright-Field Image Analyses

Diameters of cell clumps were measured in technical duplicate per clump (perpendicular axis of length and width performed twice) and averaged. To measure distance on bright-field images, the data were obtained using ImageJ and analyzed on Excel.

RNA Isolation and Quantitative RT-PCR

Total RNA was extracted and purified using the Nucleospin RNA Kit (Macherey-Nagel catalog number 740955.250). cDNA was generated using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems catalog number 4368814). Quantitative RT-PCR was performed in technical duplicate and biological triplicate using iTaq Universal SYBR Green Supermix (Bio-Rad Laboratories catalog number 1725125) on the CFX384 Touch Real-Time PCR Detection System (Bio-Rad Laboratories) using the following protocol: 3 min at 95°C, followed by 40 cycles of 5 s at 95°C and 30 s at 60°C. The signal was detected at 60°C. Expression level of the gene of interest was normalized within each sample against the expression of ACTIN (no change across differentiation) and calculated for relative expression values against the control samples using the 2−ΔΔCt method. All primers used for quantitative RT-PCR are listed in Supplementary Table 2.

RNA-Seq and Functional Annotation

RNA samples (collected as above) and library quality were assessed by bioanalyzer and quantified by quantitative PCR. RNA-seq was run in high-output mode on HiSEQ 2500 to a depth of at least 30 million reads. Single-end FASTQ reads were aligned to human genome build GRCh38 using hisat2 with default parameters (10). SAM files were processed and sorted to BAM format using sambamba (version 0.63) (11). Transcript counts were generated with htseq-count (version 0.5.4p3). Processed data were raw counts of transcripts as determined by htseq-count. The differentially expressed genes were determined using EdgeR based on both the P value (P < 0.05) and fold change (>1.5) (12). Transcript functional annotation was performed with DAVID, using the Gene Ontology (GO) biological process (BP) (13).

Flow Cytometry

Differentiated endocrine progenitors or β-like clumps were dispersed into single cells using TrypLE Express Enzyme (Thermo Fisher Scientific catalog number 12605010) and passed through a 40-μm filter to obtain a single-cell suspension. The cells were fixed with 4% paraformadelhyde for 1 h at room temperature, washed twice in PBS plus 5% FBS, and then permeabilized for 1 h at room temperature using FACS buffer (PBS plus 5% FBS plus 0.1% Triton X-100). Cells were then stained with primary antibodies (1:100 in FACS buffer) for 1 h at room temperature. Day 20 cells were stained for NKX6.1 (LifeSpan catalog number LS-C124275) and PDX1 (Abcam catalog number ab47308), while day 35 cells were stained for insulin (Abcam catalog number ab7842), glucagon (Santa Cruz Biotechnology catalog number sc-7780), somatostatin (Santa Cruz Biotechnology catalog number sc-7819), and chromogranin (Millipore catalog number ab9254). Cells were washed twice with FACS buffer and then incubated with secondary antibodies (1:2,000 in FACS buffer) for 1 h at room temperature. Cells were washed three times with FACS buffer and then resuspended in PBS and analyzed using the BD FACSymphony cell analyzer. Subsequent analysis was performed using FlowJo. Acquired events were plotted in a forward-scatter versus side-scatter plot and debris excluded based on size (Supplementary Fig. 1A).

Immunostaining

Day-35 β-like cell clumps were cryopreserved with optimal cutting temperature (14) compound. Samples were blocked and permeabilized for 1 h with 5% BSA and 0.1% Triton X-100 in PBS. Samples were then incubated overnight at 4°C with primary antibody (1:100) in 5% BSA and 0.1% Triton X-100 in PBS, incubated for 2 h at 4°C with secondary antibodies (1:500) in 5% BSA and 0.1% Triton X-100 in PBS, and stained with DAPI for 5 min. Images were acquired using the Olympus FV1000 confocal microscope and Olympus Fluoview (version 4.2) software.

GSIS Assay

GSIS was performed on the human fetal pancreatic β-cell line EndoC-βH1, adult human cadaveric islets (healthy donors), and mouse pancreatic insulinoma cell line MIN6. Briefly, ˜100 human islets or 500,000–750,000 single β-cells were picked/seeded into each well of a 24-well cell culture plate (Corning Costar catalog number CLS3527 or 3473) and rinsed three times with Krebs-Ringer Bicarbonate (KRB) buffer (129 mmol/L sodium chloride, 4.8 mmol/L potassium chloride, 2.5 mmol/L Calcium chloride, 1.2 mmol/L magnesium sulfate, 1.2 mmol/L monopotassium phosphate, 5 mmol/L sodium bicarbonate, 10 mmol/L HEPES, and 0.1% BSA in double distilled water, with a pH of 7.4 and sterile filtered) at 37°C. The cells were equilibrated in KRB buffer containing 2.8 mmol/L d-glucose (Sigma-Aldrich catalog number G8270) at 37°C for 1 h. Cells were then incubated in KRB buffer containing 2.8 mmol/L glucose at 37°C for 1 h, and supernatants were collected. Next, cells were rinsed three times with PBS and incubated in KRB buffer containing 16.7 mmol/L glucose at 37°C for 1 h, and supernatants were collected again. At the end of the experiment, the Human Insulin ELISA Kit (Mercodia catalog number 10-1113-10) was used to measure the insulin content in the media and in the cells following the manufacturer’s instructions.

Seahorse Assay

EndoC-βH1 cells were treated with metformin for 72 h before the assay. One day before the assay, the cells were seeded into 96-well Seahorse XF96 cell culture microplates (Agilent catalog number 101085–004). Oxygen consumption rate and extracellular acidification rate were measured using the Seahorse XF Cell Mito Stress Test Kit (Agilent Technologies catalog number 103015-100) and the Seahorse XF Glycolytic Rate Assay Kit (Agilent Technologies catalog number 103344-100), respectively, on the Seahorse XFe96 Analyzer. Analysis was performed using Seahorse Wave Desktop software.

Glucose Uptake Assay

EndoC-βH1 cells were seeded in appropriate cell culture plates 4 days before the assay. After the first 24 h, cells were treated with metformin for 72 h and then starved in DMEM medium with no glucose (Gibco catalog number 11966025) for 4 h. Cells were then incubated in DMEM medium containing 3 g/L d-glucose for a further 4 h before supernatants were collected. The remaining cells were lysed using M-PER protein extraction reagent and measured for protein content. The supernatants were then measured for residual glucose amount using the Glucose Uptake Colorimetric Assay Kit (Sigma-Aldrich catalog number MAK083). The level of glucose uptake was normalized to the protein level in each sample.

Cellular ADP-to-ATP Ratio Assay

EndoC-βH1 cells were seeded 4 days before the assay. After exposure to metformin for 72 h, the cells were washed with PBS and then lysed; cellular ADP-to-ATP ratio was measured using the ADP/ATP Bioluminescence Assay Kit (ApoSENSOR; BioVision Inc. catalog number K255) following the manufacturer’s protocol.

Cell Viability Assay

EndoC-βH1 cells were seeded onto 12-well plates at the same density and treated with metformin for 72 h. The cells were then detached from the plate using Trypsin-EDTA 0.25% (Gibco) and collected in 1 mL culture media per condition. The collected samples were subjected to a Trypan Blue dye exclusion test to determine cell viability. Viable cell populations were counted manually under a bright-field microscope, and the results were analyzed using GraphPad Prism software.

Quantification and Statistical Analysis

Quantification data were presented as mean ± SD. To directly compare two groups, a Student t test with two-tailed distribution and two-sample equal variance was used. To compare transcriptional expression, one-way ANOVA was used with GraphPad Prism and a Dunnett post hoc test.

Table 1

Key resources table

Reagent or resourceSourceIdentifier
Chemicals, peptides, and recombinant proteins   
 High-glucose DMEM HyClone SH30022FS 
 Low-glucose DMEM HyClone 11885 
 DMEM medium with no glucose Gibco 11966025 
 TeSR-E8 STEMCELL Technologies 05990 
 DMEM/F12 with GlutaMAX supplement, Pyruvate Invitrogen 10565042 
 MIAMI media Corning 98-021-CV 
 Glutathione Sigma-Aldrich G4251 
 GE Healthcare FBS HyClone 11521851 
 KnockOut Serum Replacement Invitrogen 10828028 
 BSA Sigma-Aldrich A9418 
 GlutaMAX supplement Gibco 11574466 
 Nonessential amino acids Invitrogen 11140050 
 β-mercaptoethanol (1,000×) Gibco 21985023 
 Trypsin-EDTA 0.25% Gibco 25300054 
 ReLeSR STEMCELL Technologies 05873 
 TrypLE Express Enzyme Thermo Fisher Scientific 12605010 
 D-glucose Sigma-Aldrich G8769 
 Metformin 1,1-dimethylbiguanide, hydrochloride Sigma-Aldrich CAS 1115-70-4 
 Gelatin Sigma-Aldrich G1890 
 Penicillin/streptomycin Gibco 15140122 
 Heparin Sigma-Aldrich H3149 
 ECM Sigma-Aldrich E1270 
 Fibronectin Sigma-Aldrich F0895 
 Transferrin Sigma-Aldrich T8158-100MG 
 Sodium selenite Sigma-Aldrich 214485-100G 
 Nicotinamide Sigma-Aldrich N3376 
 FGF2 MACS 130-093-840 
 Y27632 STEMCELL Technologies 72302 
Commercial assays   
 Nucleospin RNA Kit Macherey-Nagels 740955.250 
 MycoAlert Mycoplasma Detection Kit Lonza LT07-318 
 iTaq Universal SYBR Green Supermix Bio-Rad Laboratories 1725125 
 High-Capacity cDNA Reverse Transcription Kit Applied Biosystems 4368814 
 Glucose Uptake Colorimetric Assay Kit Sigma-Aldrich MAK083 
 ADP/ATP Bioluminescence Assay Kit (ApoSENSOR) BioVision Inc. K255 
 Seahorse XF Cell Mito Stress Test Kit Agilent Technologies 103015-100 
 Seahorse XF Glycolytic Rate Assay Kit Agilent Technologies 103344-100 
 Mercodia Insulin ELISA Mercodia Immunoassays and Services 10-1113-01 
Deposited data   
 RNA-seq This article GSE133087 
Experimental models: cell lines   
 H9 hESC line WiCell Research Institute NIHhESC-10-0062 
 EndoC-βH1 Univercell Biosolutions  
 Human islets University of Alberta, Canada  
 MIN6 (mouse insulinoma 6)  RRID: CVCL0431 
Software and algorithms   
 Prism (version 8) graphing and statistical software GraphPad Software https://www.graphpad.com/scientific-software/prism/ 
 Seahorse Wave Desktop Software Agilent Technologies https://www.agilent.com/en/products/cell-analysis/cell-analysis-software/data-analysis/wave-desktop-2-6 
 ImageJ National Institutes of Health https://imagej.nih.gov/ij/ 
Antibodies   
 Chromogranin A Millipore ab9254 
 Insulin Abcam ab7842 
 Glucagon Abcam ab82270 
 Glucagon Santa Cruz Biotechnology sc-7780 
 Cleaved caspase-3 Cell Signaling Technology 9661S 
 Ki-67 Abcam ab16667 
 NKX6.1 Lifespan LS-C124275 
 PDX1 Abcam ab47308 
 Somatostatin Santa Cruz Biotechnology sc7819 
 Alexa Fluor 594 goat anti–guinea pig Life Technology Corporation A11076 
 Alexa Fluor 488 donkey anti-mouse Life Technology Corporation A21202 
 Alexa Fluor 488 donkey anti-rabbit Life Technology Corporation A21206 
 Alexa Fluor 488 donkey anti-goat Life Technology Corporation A11055 
 Alexa Fluor 647 donkey anti-goat Life Technology Corporation A21447 
 Alexa Fluor 647 donkey anti-rabbit Life Technology Corporation A31573 
 Alexa Fluor 647 donkey anti–guinea pig Jackson ImmunoResearch Laboratories 706-605-148 
Reagent or resourceSourceIdentifier
Chemicals, peptides, and recombinant proteins   
 High-glucose DMEM HyClone SH30022FS 
 Low-glucose DMEM HyClone 11885 
 DMEM medium with no glucose Gibco 11966025 
 TeSR-E8 STEMCELL Technologies 05990 
 DMEM/F12 with GlutaMAX supplement, Pyruvate Invitrogen 10565042 
 MIAMI media Corning 98-021-CV 
 Glutathione Sigma-Aldrich G4251 
 GE Healthcare FBS HyClone 11521851 
 KnockOut Serum Replacement Invitrogen 10828028 
 BSA Sigma-Aldrich A9418 
 GlutaMAX supplement Gibco 11574466 
 Nonessential amino acids Invitrogen 11140050 
 β-mercaptoethanol (1,000×) Gibco 21985023 
 Trypsin-EDTA 0.25% Gibco 25300054 
 ReLeSR STEMCELL Technologies 05873 
 TrypLE Express Enzyme Thermo Fisher Scientific 12605010 
 D-glucose Sigma-Aldrich G8769 
 Metformin 1,1-dimethylbiguanide, hydrochloride Sigma-Aldrich CAS 1115-70-4 
 Gelatin Sigma-Aldrich G1890 
 Penicillin/streptomycin Gibco 15140122 
 Heparin Sigma-Aldrich H3149 
 ECM Sigma-Aldrich E1270 
 Fibronectin Sigma-Aldrich F0895 
 Transferrin Sigma-Aldrich T8158-100MG 
 Sodium selenite Sigma-Aldrich 214485-100G 
 Nicotinamide Sigma-Aldrich N3376 
 FGF2 MACS 130-093-840 
 Y27632 STEMCELL Technologies 72302 
Commercial assays   
 Nucleospin RNA Kit Macherey-Nagels 740955.250 
 MycoAlert Mycoplasma Detection Kit Lonza LT07-318 
 iTaq Universal SYBR Green Supermix Bio-Rad Laboratories 1725125 
 High-Capacity cDNA Reverse Transcription Kit Applied Biosystems 4368814 
 Glucose Uptake Colorimetric Assay Kit Sigma-Aldrich MAK083 
 ADP/ATP Bioluminescence Assay Kit (ApoSENSOR) BioVision Inc. K255 
 Seahorse XF Cell Mito Stress Test Kit Agilent Technologies 103015-100 
 Seahorse XF Glycolytic Rate Assay Kit Agilent Technologies 103344-100 
 Mercodia Insulin ELISA Mercodia Immunoassays and Services 10-1113-01 
Deposited data   
 RNA-seq This article GSE133087 
Experimental models: cell lines   
 H9 hESC line WiCell Research Institute NIHhESC-10-0062 
 EndoC-βH1 Univercell Biosolutions  
 Human islets University of Alberta, Canada  
 MIN6 (mouse insulinoma 6)  RRID: CVCL0431 
Software and algorithms   
 Prism (version 8) graphing and statistical software GraphPad Software https://www.graphpad.com/scientific-software/prism/ 
 Seahorse Wave Desktop Software Agilent Technologies https://www.agilent.com/en/products/cell-analysis/cell-analysis-software/data-analysis/wave-desktop-2-6 
 ImageJ National Institutes of Health https://imagej.nih.gov/ij/ 
Antibodies   
 Chromogranin A Millipore ab9254 
 Insulin Abcam ab7842 
 Glucagon Abcam ab82270 
 Glucagon Santa Cruz Biotechnology sc-7780 
 Cleaved caspase-3 Cell Signaling Technology 9661S 
 Ki-67 Abcam ab16667 
 NKX6.1 Lifespan LS-C124275 
 PDX1 Abcam ab47308 
 Somatostatin Santa Cruz Biotechnology sc7819 
 Alexa Fluor 594 goat anti–guinea pig Life Technology Corporation A11076 
 Alexa Fluor 488 donkey anti-mouse Life Technology Corporation A21202 
 Alexa Fluor 488 donkey anti-rabbit Life Technology Corporation A21206 
 Alexa Fluor 488 donkey anti-goat Life Technology Corporation A11055 
 Alexa Fluor 647 donkey anti-goat Life Technology Corporation A21447 
 Alexa Fluor 647 donkey anti-rabbit Life Technology Corporation A31573 
 Alexa Fluor 647 donkey anti–guinea pig Jackson ImmunoResearch Laboratories 706-605-148 

Data and Resource Availability

Key resources for cultures, assays, and analyses are listed in Table 1. All data generated or analyzed during this study are included in this article (and in the Supplementary Material). The Gene Expression Omnibus accession for the RNA-seq data reported in this article is GSE133087 with secure token wzebkiaghjmhpgh.

Metformin Impeded Pancreatic β-Cell Development From hESCs

We evaluated the impact of metformin on human pancreatic β-cell development using the differentiation of hESCs as a surrogate human in vitro model (9) (Fig. 1A). FACS analyses revealed that ∼70% of cells were PDX1+ (marker of pancreatic lineage) at day 20 of differentiation (Supplementary Fig. 1A), and ˜30% of cells were INS+ (marker of β-cells) at day 35, the end of differentiation (Supplementary Fig. 1B). These percentages are comparable to current literature (9), indicating efficient generation of human pancreatic β-like cells. Upon repeated β-cell differentiation experiments, we found that metformin exposure resulted in significantly smaller β-like cell clumps at the end of differentiation (Fig. 1B and C and Supplementary Fig. 2). Quantitative PCR analyses revealed that exposure to 100 μmol/L or 1 mmol/L of metformin from the start of β-cell differentiation significantly downregulated the expression of numerous pancreatic β-cell genes such as ISL1, PCSK2, CHGA, CHGB, INS, and GCG (Fig. 1D and E). Even when we mimicked clinical human embryonic/fetal exposure to metformin later during pregnancy by introducing metformin at day 13 or 20 of β-cell differentiation, we continued to observe a similar downregulation of pancreatic hormonal transcripts (Supplementary Fig. 3). Downregulation of INS, GCG, SST, and CHGA staining positivity by immunostaining and FACS analysis was also observed in β-like cells exposed to >100 μmol/L metformin through the course of differentiation, compared with cells not treated with metformin (Fig. 1F and G and Supplementary Figs. 4 and 5). However, we did not observe the same inhibitory effects at the protein level at lower concentrations of metformin exposure (Supplementary Figs. 4 and 5), although the transcript levels of PCSK2, CHGB, and GCG were reduced at lower concentrations of metformin (Supplementary Fig. 6). We did not observe any change in the protein expression of NKX6.1 or PDX1 at day 20 of β-cell differentiation (Supplementary Fig. 7). Together, these data indicate that exposure to these concentrations of metformin can perturb pancreatic β-cell differentiation from hESCs, without affecting the generation of NKX6.1+ and PDX1+ endocrine progenitors.

Figure 1

Metformin impedes pancreatic β-cell development from human embryonic stem cells. A: Schematic representation of the pancreatic β-cell differentiation protocol. B: Bright-field images of pancreatic organoids with/without 100 μmol/L metformin treatment in vitro. Scale bar = 200 μm. C: Diameters of pancreatic organoids with/without 100 μmol/L metformin treatment in vitro at the end of differentiation (day 35) (measurements taken from 47 untreated organoids and 98 metformin-treated organoids). Each dot represents the average readings of two diameter measurements from the same organoid. D and E: Expression of pancreatic markers ISL1, PCSK2, CHGA, CHGB, INS, and GCG during pancreatic β-cell differentiation with/without 100 μmol/L (D) or 1 mmol/L (E) metformin treatment (n = 3 biological replicates with their own technical triplicates; one representative experiment is shown in the figure). All error bars indicate SD of three replicates. F: Immunostaining of INS protein expression in day 35 pancreatic β-like cell clumps with/without 100 μmol/L metformin. Scale bar = 40 μm. G: FACS analysis of INS expression in day 35 pancreatic β-like cells with/without 100 μmol/L or 1 mmol/L metformin treatment. *P < 0.05 compared with untreated samples (Student t test) (C) or with untreated day 35 samples (one-way ANOVA) (D and E). DE, definitive endoderm; EP, endocrine progenitor; FSC, forward scatter; PGT, posterior gut tube; PP, pancreatic progenitor.

Figure 1

Metformin impedes pancreatic β-cell development from human embryonic stem cells. A: Schematic representation of the pancreatic β-cell differentiation protocol. B: Bright-field images of pancreatic organoids with/without 100 μmol/L metformin treatment in vitro. Scale bar = 200 μm. C: Diameters of pancreatic organoids with/without 100 μmol/L metformin treatment in vitro at the end of differentiation (day 35) (measurements taken from 47 untreated organoids and 98 metformin-treated organoids). Each dot represents the average readings of two diameter measurements from the same organoid. D and E: Expression of pancreatic markers ISL1, PCSK2, CHGA, CHGB, INS, and GCG during pancreatic β-cell differentiation with/without 100 μmol/L (D) or 1 mmol/L (E) metformin treatment (n = 3 biological replicates with their own technical triplicates; one representative experiment is shown in the figure). All error bars indicate SD of three replicates. F: Immunostaining of INS protein expression in day 35 pancreatic β-like cell clumps with/without 100 μmol/L metformin. Scale bar = 40 μm. G: FACS analysis of INS expression in day 35 pancreatic β-like cells with/without 100 μmol/L or 1 mmol/L metformin treatment. *P < 0.05 compared with untreated samples (Student t test) (C) or with untreated day 35 samples (one-way ANOVA) (D and E). DE, definitive endoderm; EP, endocrine progenitor; FSC, forward scatter; PGT, posterior gut tube; PP, pancreatic progenitor.

Genome-Wide RNA-Seq Analyses Revealed That Metformin Downregulates Genes Relating to Hormone Regulation, Transport, and Secretion in Pancreatic β-Cells

Next, to determine the impact of metformin treatment on global transcript changes during the differentiation of hESCs into pancreatic β-like cells, we performed RNA-seq analyses on day 0, 13, 20, and 35 cells (Supplementary Table 1), representing undifferentiated hESCs, pancreatic progenitors, pancreatic endocrine progenitors, and pancreatic β-like cells, respectively. Key pancreatic genes were appropriately expressed at the specific time points of pancreatic development (Fig. 2A). Uniform manifold approximation and projection analyses revealed that replicates at each time point clustered tightly (Fig. 2B), indicating that the β-like cells demonstrate distinctive transcriptional profiles at different time points with relatively small interexperimental variability. At day 35, a subset of genes associated specifically with pancreas development, islet biology, pancreatic β-cell function, and diabetes was significantly downregulated (Fig. 2C). GO BP analyses also supported this observation, with GO terms such as regulation of hormone levels, signal release, and peptide hormone secretion being downregulated in cells exposed to metformin (Fig. 2D). This suggests that metformin is likely to affect events during terminal differentiation.

Figure 2

Metformin exposure results in global transcriptional changes in pancreatic β-cell development from human embryonic stem cells. A: Heat map of expression pattern of genes involved in pancreatic development and function during pancreatic differentiation. B: Uniform manifold approximation and projection (UMAP) analyses of hESCs differentiated into pancreatic β-like cells at different time points of differentiation with/without 100 μmol/L metformin treatment. C: Heat map of a subset of pancreatic genes significantly up- and downregulated in pancreatic β-like cells treated with 100 μmol/L metformin. D and E: GO analyses of downregulated (P < 0.05; 1/fold change [FC] >1.5) (D) and upregulated (P < 0.05; FC >1.5) (E) genes in day 35 cells treated with 100 μmol/L metformin compared with untreated day 35 cells.

Figure 2

Metformin exposure results in global transcriptional changes in pancreatic β-cell development from human embryonic stem cells. A: Heat map of expression pattern of genes involved in pancreatic development and function during pancreatic differentiation. B: Uniform manifold approximation and projection (UMAP) analyses of hESCs differentiated into pancreatic β-like cells at different time points of differentiation with/without 100 μmol/L metformin treatment. C: Heat map of a subset of pancreatic genes significantly up- and downregulated in pancreatic β-like cells treated with 100 μmol/L metformin. D and E: GO analyses of downregulated (P < 0.05; 1/fold change [FC] >1.5) (D) and upregulated (P < 0.05; FC >1.5) (E) genes in day 35 cells treated with 100 μmol/L metformin compared with untreated day 35 cells.

Metformin Resulted in an Upregulation of Several Specific Mitochondrial Genes but a Net Decrease in Mitochondrial Respiration in Pancreatic β-Like Cells

On the contrary, GO BP terms such as oxidative phosphorylation and electron transport chain were upregulated (Fig. 2E). From our RNA-seq analyses, we determined that only a small subset of mitochondrial genes such as COX2, COX1, ND1, ATP6, CYTB, ND4L, ND4, ND5, ND2, and ND6 were uniquely upregulated at the end of differentiation in hESC-derived pancreatic β-like cells compared with other genes in the electron transport chain network (Fig. 3A). Metformin treatment from day 0 onward with 100 μmol/L metformin resulted in higher levels of many of these genes on day 35 of differentiation (Fig. 3B and Supplementary Fig. 8A).

Figure 3

Metformin perturbs mitochondrial gene expression and bioenergetics in pancreatic β-cells. A: Heat map of expression pattern of genes involved in oxidative phosphorylation during pancreatic differentiation in untreated pancreatic β-like cells. B: Heat map of a subset of electron transport chain genes significantly up- and downregulated in pancreatic β-like cells treated with 100 μmol/L metformin. C and D: Seahorse assay measuring mitochondrial stress in EndoC-βH1 cells treated with 0, 50, or 100 μmol/L or 1 mmol/L metformin (n = 4 biological replicates with 5 to 6 technical duplicates per condition). E: ATP-to-ADP ratio assay in EndoC-βH1 cells treated with 0, 25, 50, or 100 μmol/L or 1 mmol/L metformin (n = 4 biological replicates). All error bars indicate SD of three replicates. *P < 0.05 compared with untreated samples (Student t test). OCR, oxygen consumption rate.

Figure 3

Metformin perturbs mitochondrial gene expression and bioenergetics in pancreatic β-cells. A: Heat map of expression pattern of genes involved in oxidative phosphorylation during pancreatic differentiation in untreated pancreatic β-like cells. B: Heat map of a subset of electron transport chain genes significantly up- and downregulated in pancreatic β-like cells treated with 100 μmol/L metformin. C and D: Seahorse assay measuring mitochondrial stress in EndoC-βH1 cells treated with 0, 50, or 100 μmol/L or 1 mmol/L metformin (n = 4 biological replicates with 5 to 6 technical duplicates per condition). E: ATP-to-ADP ratio assay in EndoC-βH1 cells treated with 0, 25, 50, or 100 μmol/L or 1 mmol/L metformin (n = 4 biological replicates). All error bars indicate SD of three replicates. *P < 0.05 compared with untreated samples (Student t test). OCR, oxygen consumption rate.

To evaluate mitochondrial respiration and glycolysis processes in human β-cells in isolation (as opposed to heterogenous hESC-derived β-like cells), we performed metabolic assays using the Seahorse Analyzer on the human fetal β-cell line EndoC-βH1, which was exposed to a range of metformin concentrations (Fig. 3C). Increasing doses of metformin decreased basal respiration, ATP production, mitochondrial maximal respiration, spare respiratory capacity (Fig. 3D), and glycolytic function (Supplementary Fig. 9A), although there were no changes in transcript levels of mitochondrial ETC genes in these cells (Supplementary Fig. 8B). As further confirmation, we also assessed transcript levels of these genes in metformin-treated adult human pancreatic islets and obtained similar results (Supplementary Fig. 8C). Nevertheless, metformin led to significantly decreased ATP generation (Fig. 3E). To rule out impaired glucose uptake as a cause of impaired glycolysis and mitochondrial oxidative phosphorylation, we performed glucose uptake assays on these EndoC-βH1 cells and confirmed that glucose uptake was not significantly altered by metformin treatment (Supplementary Fig. 9B).

Metformin Perturbed Pancreatic β-Cell Insulin Secretory Function

Human β-cells start secreting insulin at ˜9–14 weeks. They then continue to mature and proliferate during the next two trimesters (15). At the point of birth, the β-cells of neonates are functional, like adult β-cells, which is not reflected in our pancreatic differentiation model. To more comprehensively represent pancreatic β-cells present during human fetal development, we exposed the immortalized fetal pancreatic β-cell line EndoC-βH1, human islets, and immortalized mouse insulinoma cell line MIN6 to increasing doses of metformin for 72 h in vitro before assessment by GSIS functional assays to investigate the possible impact of metformin on downstream pancreatic β-cell function. We observed that therapeutic levels of metformin at 50–100 μmol/L inhibited GSIS response in EndoC-βH1 cells (Fig. 4A and Supplementary Fig. 10A), with no effect on basal insulin secretion levels. Higher concentrations of metformin exposure (100 μmol/L concentration) were required to induce defective insulin secretion by human islets and MIN6 cells (Supplementary Fig. 10B). However, subsequent removal of metformin for 72 h, following 72 h metformin exposure, resulted in restoration of some GSIS functionality in EndoC-βH1 cells (Fig. 4B).

Figure 4

A and B: Metformin exposure perturbs human pancreatic β-cell function and viability. GSIS assay in human immortalized fetal pancreatic β-cell line EndoC-βH1 treated with 0, 25, 50, or 100 μmol/L or 1 mmol/L metformin for 72 h (n = 6–9 biological replicates) (A) or varying treatments up to 144 h (B). C: Cell viability assay using Trypan blue dye exclusion test in EndoC-βH1 cells treated with varying concentrations of metformin (n = 3 biological replicates). Error bars indicate SD. D: Summary figure depicting the effects of metformin on hESC differentiation into pancreatic β-like cells. *P < 0.05, **P < 0.01, ****P < 0.0001 compared with untreated samples (Student t test). ns, not significant.

Figure 4

A and B: Metformin exposure perturbs human pancreatic β-cell function and viability. GSIS assay in human immortalized fetal pancreatic β-cell line EndoC-βH1 treated with 0, 25, 50, or 100 μmol/L or 1 mmol/L metformin for 72 h (n = 6–9 biological replicates) (A) or varying treatments up to 144 h (B). C: Cell viability assay using Trypan blue dye exclusion test in EndoC-βH1 cells treated with varying concentrations of metformin (n = 3 biological replicates). Error bars indicate SD. D: Summary figure depicting the effects of metformin on hESC differentiation into pancreatic β-like cells. *P < 0.05, **P < 0.01, ****P < 0.0001 compared with untreated samples (Student t test). ns, not significant.

Reduced Viability of β-Cells With Increasing Concentration of Metformin

Finally, to assess if the reduced β-cell gene expression and function could be accounted for by reduced cell viability with metformin exposure, proliferation and apoptosis of EndoC-βH1 cells were evaluated after 72 h of metformin treatment. Metformin-treated EndoC-βH1 cells exhibited significantly reduced viability at concentrations of metformin from ≥50 μmol/L (Fig. 4C). Immunostaining of the proliferation marker Ki-67 and proapoptotic cleaved caspase-3 showed drastic reduction in the total number of cells at 1 mmol/L metformin (Supplementary Fig. 11A–C). However, there was no obvious difference in proportion of cells expressing Ki-67 at levels of metformin exposure under 1 mmol/L (Supplementary Fig. 11A and B). In contrast, there seemed to be a trend toward a decrease in the percentage of cells expressing cleaved caspase-3 (apoptosis) at higher concentrations of metformin exposure, although this was only significant at 50 μmol/L metformin. However, there was also a corresponding decrease in cell number at metformin concentrations >25 μmol/L, suggesting that cells remaining in culture are those that did not undergo apoptosis, thus explaining the decreased proportion of cells expressing cleaved caspase-3 at higher metformin concentrations (Supplementary Fig. 11C). Overall, our results indicated that prolonged direct exposure to metformin in an in vitro system is detrimental to pancreatic β-cell viability and function (Fig. 4D).

Together, data based on our human in vitro models suggest that metformin has inhibitory effects on human pancreatic β-cell development, mitochondrial function, glucose metabolism, and insulin secretion. Thus far, embryonic and fetal exposure to metformin in vivo does not seem to result in any obvious teratogenicity (16,17). A recent meta-analysis of clinical trials involving metformin treatment during pregnancy also concluded that there is no increased risk of neonatal malformation or perinatal morbidity or mortality (18). However, long-term human clinical trials studying the impact of metformin exposure in utero on offspring development have reported contrasting findings, with study populations being relatively small and lacking generalizability (19). Several randomized controlled trials in women with polycystic ovary syndrome and GDM reported increased body weight (7,20,21), fasting glucose levels (22), and fat deposits (21,23) in children exposed to metformin in utero, up to the age of 10 years. However, a trial in women with GDM showed no differences in these parameters between control and metformin-treated groups (24). A meta-analysis of 28 trials of metformin use in pregnancies with gestational diabetes including a total of 3,976 children indicated that children exposed to metformin in utero tended to have faster growth and higher BMI and body weight than their insulin-treated counterparts, despite weighing less at the point of birth (8).

Furthermore, studies at the molecular and cellular levels of the effect of metformin on mature and developing pancreatic β-cells have similarly reported contradicting effects. Several studies indicated that metformin can have protective effects on mouse pancreatic β-cells that are exposed to metabolic stresses (25,26). In contrast, in a nonmetabolically challenging environment, metformin inhibits proliferation and promotes apoptosis of rodent pancreatic β-cells in vitro (25,27,28). Primary human islets, mouse islets, and mouse and rat pancreatic β-cell lines have also shown inhibited insulin secretion under metformin treatment when in a nonmetabolically challenging environment (27,28), similar to what we observed in our pancreatic_EndoC-βH1 cell line. In developing fetal pancreatic cells, the effects of metformin are also unclear, with data in the published literature showing conflicting results regarding the effects of metformin on AMPK activation in fetal tissues (29,30), birth weight, and fat deposits in offspring (3133). Rodent studies also suggest a context-dependent effect of metformin. One study showed that offspring of healthy women exposed to metformin in utero had increased weight gain and impaired glucose tolerance when exposed to a high-fat diet, compared with controls not exposed to metformin in utero (31). However, a different study showed that fetuses of healthy women exposed to metformin in utero had increased numbers of pancreatic and endocrine progenitors during pancreas development (34), and adult offspring had improved glucose tolerance and insulin sensitivity (34). In contrast, offspring of female mice challenged with a high-fat diet before and during pregnancy had improved glucose tolerance when exposed to metformin in utero (32).

Overall, the current body of research points toward a dual function of metformin in pancreatic β-cells that is environment dependent. However, the lack of studies incorporating prospective long-term clinical follow-up of offspring in humans and animals, the lack of access to human embryonic/fetal tissues exposed to metformin in vivo, and the lack of information regarding local metformin concentrations in pancreatic tissue have given rise to a gap in understanding the impact of metformin on human embryonic and fetal tissues at the cellular and molecular levels.

In this study, we used hESC-derived pancreatic β-like cells as a model of human pancreatic development to study the impact of metformin exposure in vitro, because bona fide human fetal islets are not readily available for such studies. Considering that current therapeutic concentrations of metformin can be up to 1,800 mg/L (14 mmol/L), with an average of ˜90 mg/L (698 μmol/L) (35), and the fact that the fetal circulating concentration of metformin can range from half to the same level as that in maternal plasma (2,5,36), we chose experimental doses of 0, 10, 25, 50, and 100 μmol/L, up to 1 mmol/L (129 mg/L) metformin, to accommodate both the therapeutically relevant circulating concentrations of metformin administered clinically and the proposed concentrations of metformin accumulated in the mitochondria of cells (37,38).

In pancreatic β-like clusters, we observed that exposure to high concentrations of metformin (>50 μmol/L) consistently decreased the overall clump size. This is in part consistent with previous reports showing that metformin decreases and limits the formation of teratomas from ESCs by reducing cell proliferation (39) and increasing apoptosis (25,27). Although we did not observe a decrease in cell proliferation, possibly because of the lower concentration of metformin used in our study, our finding that 1 mmol/L metformin can cause excessive cell death is consistent with published reports that metformin can potentially promote apoptosis and thereby decrease organ size. It remains to be determined whether organs such as the pancreas are indeed smaller in an in vivo setting, when dams or women are treated with metformin around the time of conception.

When we differentiated hESCs into pancreatic β-like organoids in the presence of 100 μmol/L or 1 mmol/L metformin, we found numerous pancreatic β-cell genes, including INS, to be downregulated, demonstrating that metformin indeed affects human pancreatic β-cell development, at least in vitro. Besides β-cell development, we also found metformin to have detrimental effects on both mouse and human β-cell function. These results are surprising, given that metformin is an antidiabetic drug that exhibits protective effects on pancreatic β-cells under metabolic challenges in both animal and human studies (25,40). However, our data seem to be consistent with some limited reports that metformin can impair insulin secretion and β-cell proliferation and promote apoptosis in mouse and human islets in the absence of metabolic challenges, such as high concentrations of glucose and fatty acids (25,27,28).

To demonstrate the impact of metformin on mature and fully functional pancreatic β-cells, both in isolation and when present with other cells in the islet of Langerhans, we used an immortalized mouse insulinoma cell line (MIN6), an immortalized human fetal pancreatic β-cell line (EndoC-βH1), and primary adult human pancreatic islets. Although metformin impaired mitochondrial oxidative phosphorylation in pancreatic β-cells, consistent with its known role in inhibiting complex 1 of the mitochondrial respiratory chain in tissues such as the liver, there appeared to be a compensatory upregulation of some key mitochondrial genes in differentiating pancreatic β-cells (41). In contrast, we did not observe strong upregulation of mitochondrial ETC genes in the EndoC-βH1 cell line and adult human pancreatic islets, suggesting insufficient or no overall compensation against the inhibitory effects of metformin on mitochondrial oxidative phosphorylation (42). This may explain the failure of these cells to maintain cellular ATP levels and may explain the impaired β-cell GSIS function and increase in cell loss/death in vitro. This is consistent with the fact that EndoC-βH1s are functional pancreatic β-cells, which are likely to have different energy requirements to fetal pancreatic β-cells during early development. Differentiating pancreatic cells depend increasingly on oxidative phosphorylation for respiration (4345), which results in increasing expression of mitochondrial ETC genes as differentiation progresses. Thus, they may be more sensitive to metformin treatment, and a compensatory response might be triggered more readily.

Overall, our data are in agreement with the proposed dual characters of metformin; metformin can be beneficial when pancreatic β-cells are under metabolic stress (25,46) but is possibly harmful in healthy metabolic states (25). Even though our data suggest no significant impact at concentrations <50 μmol/L, it is worth noting that the intracellular concentration of metformin might not necessarily be the same as the circulating concentration and has been proposed to be much higher than the therapeutic concentration of ˜70 μmol/L (37,42,47). In this article, our studies also suggest harmful effects on developing and maturing human fetal β-cells. This postulation certainly requires further testing and validation, especially in humans in vivo, if ethically allowed.

While we have made unexpected observations on the negative effects of metformin on β-cells, we are mindful that our in vitro conditions do not fully mimic the conditions in utero. Exposure to metformin in an in vivo environment is certainly more dynamic, because metformin may first be taken up by the liver and also undergo active clearance by the maternal kidney, the clearance rate of which varies throughout pregnancy (5,36). The maternal plasma concentration of metformin can reach ˜70 μmol/L, and He and Wondisford (37) also reported the concentration of metformin in the liver circulation as 70 μmol/L (5). However, there is currently no original research clarifying the concentration of metformin in the adult pancreas and certainly not in the fetal pancreas. Interactions between different cell types within the pancreas and with other organs within the body could also cause additional differences between our in vitro model and the in vivo setting. It is also worth noting that metabolic diseases often manifest much later in life, while our in vitro organoid models only seek to represent the tissues of offspring during development. Therefore, we are planning for additional studies in rodent models with maternal hyperglycemia to investigate the impact of metformin on later stages of development.

Metformin treatment in pregnancy may result in appreciable amounts of metformin reaching the developing embryo and fetus. On the basis of our results showing adverse effects of metformin on developing pancreatic β-cells, we caution against excessive metformin dosing during pregnancy and advocate the need for more in-depth long-term studies to evaluate the effects of fetal exposure to metformin. In light of these findings, the risk-benefit balance of metformin use in the periconception and pregnancy periods will need to be reconsidered.

L.Y.L. and S.S.L.D. contributed equally to this work.

This article contains supplementary material online at https://doi.org/10.2337/figshare.14538108.

Acknowledgments. The authors thank members of the Teo laboratory for scientific support and discussions on this manuscript as well as Kong Kiat Whye (A*STAR) for help in processing the RNA-seq libraries.

Funding. L.N. is supported by the Singapore International Graduate Award from Agency for Science, Technology and Research (A*STAR). S.-Y.C. is supported by grants administered under the Singapore National Research Foundation and Singapore Ministry of Health National Medical Research Council Clinician Scientist Award (NMRC/CSA-INV/0010/2016), the National University of Singapore, the Singapore Institute for Clinical Sciences, and A*STAR. A.K.K.T. is supported by the Institute of Molecular and Cell Biology, A*STAR, an NMRC Open Fund-Young Individual Research Grant (OFYIRG16May014), A*STAR ETPL Gap Funding (ETPL/18-GAP005-R20H), a grant from the Lee Foundation (SHTX/LFG/002/2018), a grant from Skin Innovation (SIG18011), NMRC OF-LCG/DYNAMO, an FY2019 SingHealth Duke-NUS Surgery Academic Clinical Programme Research Support Programme Grant, a Precision Medicine and Personalised Therapeutics Joint Research Grant 2019, a grant from the Industry Alignment Fund–Industry Collaboration Project (I1901E0049), and the 2nd A*STAR-AMED Joint Grant Call (192B9002).

Sponsors had no involvement in the conduct of the research and/or preparation of the manuscript.

Duality of Interest. S.-Y.C. declares grants and nonfinancial support from industry funding received through the Epigen Academic Consortium, outside the submitted work. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.N. performed experiments, validation, formal analysis, and data curation and wrote the manuscript. L.Y.L. and S.S.L.D. performed experiments, validation, and formal analysis. N.S.A performed data analysis, edited data, and contributed to discussion. S.H. performed experiments and data analysis. S.-Y.C. and A.K.K.T. conceptualized and supervised the project and reviewed and edited the manuscript. A.K.K.T. acquired funding for this project. A.K.K.T. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the data and the accuracy of the data analysis.

1.
Schaefer-Graf
U
,
Napoli
A
,
Nolan
CJ
;
Diabetic Pregnancy Study Group
.
Diabetes in pregnancy: a new decade of challenges ahead
.
Diabetologia
2018
;
61
:
1012
1021
2.
Lindsay
RS
,
Loeken
MR
.
Metformin use in pregnancy: promises and uncertainties
.
Diabetologia
2017
;
60
:
1612
1619
3.
Vanky
E
,
Zahlsen
K
,
Spigset
O
,
Carlsen
SM
.
Placental passage of metformin in women with polycystic ovary syndrome
.
Fertil Steril
2005
;
83
:
1575
1578
4.
Nguyen
L
,
Chan
SY
,
Teo
AKK
.
Metformin from mother to unborn child - are there unwarranted effects?
EBioMedicine
2018
;
35
:
394
404
5.
Eyal
S
,
Easterling
TR
,
Carr
D
, et al
.
Pharmacokinetics of metformin during pregnancy
.
Drug Metab Dispos
2010
;
38
:
833
840
6.
Chaudhari
K
,
Wang
J
,
Xu
Y
, et al
.
Determination of metformin bio-distribution by LC-MS/MS in mice treated with a clinically relevant paradigm
.
PLoS One
2020
;
15
:
e0234571
7.
Hanem
LGE
,
Stridsklev
S
,
Júlíusson
PB
, et al
.
Metformin use in PCOS pregnancies increases the risk of offspring overweight at 4 years of age: follow-up of two RCTs
.
J Clin Endocrinol Metab
2018
;
103
:
1612
1621
8.
Tarry-Adkins
JL
,
Aiken
CE
,
Ozanne
SE
.
Neonatal, infant, and childhood growth following metformin versus insulin treatment for gestational diabetes: A systematic review and meta-analysis
.
PLoS Med
2019
;
16
:
e1002848
9.
Pagliuca
FW
,
Millman
JR
,
Gürtler
M
, et al
.
Generation of functional human pancreatic β cells in vitro
.
Cell
2014
;
159
:
428
439
10.
Kim
D
,
Paggi
JM
,
Park
C
,
Bennett
C
,
Salzberg
SL
.
Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype
.
Nat Biotechnol
2019
;
37
:
907
915
11.
Tarasov
A
,
Vilella
AJ
,
Cuppen
E
,
Nijman
IJ
,
Prins
P
.
Sambamba: fast processing of NGS alignment formats
.
Bioinformatics
2015
;
31
:
2032
2034
12.
Robinson
MD
,
McCarthy
DJ
,
Smyth
GK
.
edgeR: a Bioconductor package for differential expression analysis of digital gene expression data
.
Bioinformatics
2010
;
26
:
139
140
13.
Dennis
G
 Jr
,
Sherman
BT
,
Hosack
DA
, et al
.
DAVID: Database for Annotation, Visualization, and Integrated Discovery
.
Genome Biol
2003
;
4
:
P3
14.
Butler
AE
,
Matveyenko
AV
,
Kirakossian
D
,
Park
J
,
Gurlo
T
,
Butler
PC
.
Recovery of high-quality RNA from laser capture microdissected human and rodent pancreas
.
J Histotechnol
2016
;
39
(
2
):
59
-
65
15.
Pan
FC
,
Brissova
M
.
Pancreas development in humans
.
Curr Opin Endocrinol Diabetes Obes
2014
;
21
:
77
82
16.
Abu Hashim
H
.
Twenty years of ovulation induction with metformin for PCOS; what is the best available evidence?
Reprod Biomed Online
2016
;
32
:
44
53
17.
Nathan
N
,
Sullivan
SD
.
The utility of metformin therapy in reproductive-aged women with polycystic ovary syndrome (PCOS)
.
Curr Pharm Biotechnol
2014
;
15
:
70
83
18.
Butalia
S
,
Gutierrez
L
,
Lodha
A
,
Aitken
E
,
Zakariasen
A
,
Donovan
L
.
Short- and long-term outcomes of metformin compared with insulin alone in pregnancy: a systematic review and meta-analysis
.
Diabet Med
2017
;
34
:
27
36
19.
Barbour
LA
,
Feig
DS
.
Metformin for gestational diabetes mellitus: progeny, perspective, and a personalized approach
.
Diabetes Care
2019
;
42
:
396
399
20.
Hanem
LGE
,
Salvesen
Ø
,
Juliusson
PB
, et al
.
Intrauterine metformin exposure and offspring cardiometabolic risk factors (PedMet study): a 5-10 year follow-up of the PregMet randomised controlled trial
.
Lancet Child Adolesc Health
2019
;
3
:
166
174
21.
Rowan
JA
,
Rush
EC
,
Plank
LD
, et al
.
Metformin in gestational diabetes: the offspring follow-up (MiG TOFU): body composition and metabolic outcomes at 7-9 years of age
.
BMJ Open Diabetes Res Care
2018
;
6
:
e000456
22.
TB
,
Ludvigsen
HV
,
Carlsen
SM
,
Vanky
E
.
Growth, body composition and metabolic profile of 8-year-old children exposed to metformin in utero
.
Scand J Clin Lab Invest
2012
;
72
:
570
575
23.
Rowan
JA
,
Rush
EC
,
Obolonkin
V
,
Battin
M
,
Wouldes
T
,
Hague
WM
.
Metformin in gestational diabetes: the offspring follow-up (MiG TOFU): body composition at 2 years of age
.
Diabetes Care
2011
;
34
:
2279
2284
24.
Wouldes
TA
,
Battin
M
,
Coat
S
,
Rush
EC
,
Hague
WM
,
Rowan
JA
.
Neurodevelopmental outcome at 2 years in offspring of women randomised to metformin or insulin treatment for gestational diabetes
.
Arch Dis Child Fetal Neonatal Ed
2016
;
101
:
F488
F493
25.
Jiang
Y
,
Huang
W
,
Wang
J
, et al
.
Metformin plays a dual role in MIN6 pancreatic β cell function through AMPK-dependent autophagy
.
Int J Biol Sci
2014
;
10
:
268
277
26.
Masini
M
,
Anello
M
,
Bugliani
M
, et al
.
Prevention by metformin of alterations induced by chronic exposure to high glucose in human islet beta cells is associated with preserved ATP/ADP ratio
.
Diabetes Res Clin Pract
2014
;
104
:
163
170
27.
Kefas
BA
,
Cai
Y
,
Kerckhofs
K
, et al
.
Metformin-induced stimulation of AMP-activated protein kinase in beta-cells impairs their glucose responsiveness and can lead to apoptosis
.
Biochem Pharmacol
2004
;
68
:
409
416
28.
Leclerc
I
,
Woltersdorf
WW
,
da Silva Xavier
G
, et al
.
Metformin, but not leptin, regulates AMP-activated protein kinase in pancreatic islets: impact on glucose-stimulated insulin secretion
.
Am J Physiol Endocrinol Metab
2004
;
286
:
E1023
E1031
29.
Lee
HY
,
Wei
D
,
Loeken
MR
.
Lack of metformin effect on mouse embryo AMPK activity: implications for metformin treatment during pregnancy
.
Diabetes Metab Res Rev
2014
;
30
:
23
30
30.
Tong
JF
,
Yan
X
,
Zhao
JX
,
Zhu
MJ
,
Nathanielsz
PW
,
Du
M
.
Metformin mitigates the impaired development of skeletal muscle in the offspring of obese mice
.
Nutr Diabetes
2011
;
1
:
e7
31.
Salomäki
H
,
Vähätalo
LH
,
Laurila
K
, et al
.
Prenatal metformin exposure in mice programs the metabolic phenotype of the offspring during a high fat diet at adulthood
.
PLoS One
2013
;
8
:
e56594
32.
Salomäki
H
,
Heinäniemi
M
,
Vähätalo
LH
, et al
.
Prenatal metformin exposure in a maternal high fat diet mouse model alters the transcriptome and modifies the metabolic responses of the offspring
.
PLoS One
2014
;
9
:
e115778
33.
Gregg
B
,
Elghazi
L
,
Alejandro
EU
, et al
.
Exposure of mouse embryonic pancreas to metformin enhances the number of pancreatic progenitors
.
Diabetologia
2014
;
57
:
2566
2575
34.
Gregg
BE
,
Botezatu
N
,
Brill
JD
, et al
.
Gestational exposure to metformin programs improved glucose tolerance and insulin secretion in adult male mouse offspring
.
Sci Rep
2018
;
8
:
5745
35.
Kajbaf
F
,
De Broe
ME
,
Lalau
JD
.
Therapeutic concentrations of metformin: a systematic review
.
Clin Pharmacokinet
2016
;
55
:
439
459
36.
Charles
B
,
Norris
R
,
Xiao
X
,
Hague
W
.
Population pharmacokinetics of metformin in late pregnancy
.
Ther Drug Monit
2006
;
28
:
67
72
37.
He
L
,
Wondisford
FE
.
Metformin action: concentrations matter
.
Cell Metab
2015
;
21
:
159
162
38.
Bridges
HR
,
Jones
AJ
,
Pollak
MN
,
Hirst
J
.
Effects of metformin and other biguanides on oxidative phosphorylation in mitochondria
.
Biochem J
2014
;
462
:
475
487
39.
Vazquez-Martin
A
,
Cufi
S
,
Lopez-Bonet
E
, et al
.
Metformin limits the tumourigenicity of iPS cells without affecting their pluripotency
.
Sci Rep
2012
;
2
:
964
40.
Zhang
E
,
Mohammed Al-Amily
I
,
Mohammed
S
, et al
.
Preserving insulin secretion in diabetes by inhibiting VDAC1 overexpression and surface translocation in β cells
.
Cell Metab
2019
;
29
:
64
77.e6
41.
Andrzejewski
S
,
Gravel
SP
,
Pollak
M
,
St-Pierre
J
.
Metformin directly acts on mitochondria to alter cellular bioenergetics
.
Cancer Metab
2014
;
2
:
12
42.
Vial
G
,
Detaille
D
,
Guigas
B
.
Role of mitochondria in the mechanism(s) of action of metformin
.
Front Endocrinol (Lausanne)
2019
;
10
:
294
43.
Broche
B
,
Ben Fradj
S
,
Aguilar
E
, et al
.
Mitochondrial protein UCP2 controls pancreas development
.
Diabetes
2018
;
67
:
78
84
44.
Tsogtbaatar
E
,
Landin
C
,
Minter-Dykhouse
K
,
Folmes
CDL
.
Energy metabolism regulates stem cell pluripotency
.
Front Cell Dev Biol
2020
;
8
:
87
45.
Zhang
C
,
Skamagki
M
,
Liu
Z
, et al
.
Biological significance of the suppression of oxidative phosphorylation in induced pluripotent stem cells
.
Cell Rep
2017
;
21
:
2058
2065
46.
Simon-Szabó
L
,
Kokas
M
,
Mandl
J
,
Kéri
G
,
Csala
M
.
Metformin attenuates palmitate-induced endoplasmic reticulum stress, serine phosphorylation of IRS-1 and apoptosis in rat insulinoma cells
.
PLoS One
2014
;
9
:
e97868
47.
Fontaine
E
.
Metformin and respiratory chain complex I: the last piece of the puzzle?
Biochem J
2014
;
463
:
e3
e5
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/content/license.