Diabetes mellitus is characterized by an insulin deficiency (180) and an insulin resistance (181). Islet consists of β cells, which are the only cell types that can secrete insulin to downregulate blood glucose. A follow-up cohort demonstrated that offspring of GDM and T1DM had reduced insulin sensitivity and insulin release at the age of 18–27 years (21). However, a clinical study reported that offspring of T1DM (O-T1DM) had a higher fasting serum insulin compared to offspring of GDM and heathy control at preschool age (22). No difference in fasting insulin profile was observed between the offspring of GDM (O-GDM) and control mothers in another clinical trial either (23). Due to anatomic position and ethical restrictions, scientists mimic intrauterine hyperglycemia by administration of streptozotocin (24, 115) or alloxan (111, 118) to specifically destroy the islet or feeding high fat high sucrose (HFHS) diet (112, 116) to induce PGDM or GDM in an animal model. Ding, et al. discovered that adult offspring of maternal diabetes presented impaired glucose tolerance and glucose-stimulated insulin secretion (GSIS) in mice (24). Research on rats came up with similar results: 15-week-old GDM offspring rats displayed an impairment in insulin secretion, glucose utilization, and oxidation in islets (25). Their GSIS capacity worsened when O-GDM was fed with a HFHS diet (26). Genes related to the islet development (24), metabolic enzymes (25), insulin secretion, and cell cycle (156) were demonstrated to be downregulated in the islet of Langerhans cells of O-GDM. By infusing glucose into maternal left uterine artery, hyperglycemic-exposed fetuses were observed to have higher levels of serum insulin and a larger β cell mass (27). The metabolomic analysis of fetal pancreas identified 219 biochemicals with significant changes. Among them, citrate acid and α-ketoglutarate were remarkably decreased in O-GDM, which suggested a metabolism-mediated mechanism (28).

Insulin secretion is regulated by pancreatic β cells according to blood glucose fluctuation, while target organs of insulin, including liver, adipose tissue, and skeletal muscle, bind with insulin through an insulin receptor, activate downstream signaling to uptake, and catabolize glucose (182).

As the main metabolic organ of glycogen synthesis, liver has been subjected to extensive research for years. A prospective longitudinal study showed no more intrahepatocellular lipid (IHCL) in infants of GDM mothers with a higher body mass index (BMI, mean BMI: 24.3 kg/m²; interquartile range: 21.7, 30.3) compared to control group (29). Another clinical study demonstrated that obese women (BMI>30 kg/m²) with GDM gave birth to neonates with more deposition of IHCL (30). In addition, adolescent O-T1DM may predispose to fatty liver (31). In animal models of mice (32) and rats (33), offspring livers of GDM or PGDM presented more detectable steatosis and hepatic triglyceride, accompanied by reduced lipid metabolic enzymes (32). Fat-laden hepatocytes could release hepatokines to induce pro-inflammatory signaling and hepatic insulin resistance (34). Overall, 20-week-old O-GDM mice were actually demonstrated to secrete more pro-inflammatory factors such as interleukin-1β (IL1β), IL6, and IL33 (35). Animal experiments suggested that, even though basal glucose utilization and insulin concentration were normal in O-GDM, glucose infusion into the liver was lower during a hyperinsulinemic clamp (38). This feature was consistent with a decreased level of insulin receptor and dephosphorylation of AKT in O-GDM livers in another study (36), which can be aggravated by postnatal HFHS diet (32). Forkhead Box O (FOXO) acts downstream of insulin signaling, which can be activated when insulin signaling is inhibited (182). In the offspring liver of maternal diabetes mellitus, phosphorylation of FOXO was repressed, its enzymatic activity was stimulated, and target genes were upregulated, which represented for a repressed insulin pathway (37).

A multinational cross-section study conducted in 12 countries showed a higher BMI, waist circumference, and body fat, increased general and central obesity in offspring at the age of 9–11 years once exposed to hyperglycemia in utero, but when adjusted for maternal normal weight, the said difference disappeared (39). A prospective cohort study of Swedish men displayed that the BMI of people who experienced intrauterine hyperglycemia was 0.89 kg/m² (95% CI, 0.31 to 1.47) greater than the BMI of siblings born before their mother acquired GDM (40). In animal models, the offspring of PGDM (O-PGDM) had a greater peripheral fat mass and a larger lipid diameter, especially epididymal fat pads. Epididymal adipocytes from O-PGDM were also more responsive to an insulin-stimulated glucose uptake with increased levels of insulin receptor, Acetyl-CoA carboxylase, and glucose transporter 4 (43). The adipose tissue can secrete adipokines, such as leptin and adiponectin, which play a major role in glucose homeostasis and insulin sensitivity. The subcutaneous adipose tissue of O-PGDM was shown to secrete less adiponectin (32). In fact, a Danish follow-up study concluded that O-GDM had a higher leptin, a lower adiponectin, and FGF21 (41). Another clinical observation suggested that the plasma level of leptin increased, while the gene expression of adiponectin in subcutaneous adipose tissue became reduced in the offspring of both GDM and T1DM patients (42).

The brown adipose tissue (BAT) is involved in non-shivering thermogenesis and metabolic homeostasis. However, intrauterine hyperglycemia impaired the BAT mitochondrial structure and restrained its energy expenditure (44).

The skeletal muscle accounts for almost half of the lean body mass in an adult and is responsible for the majority of issues related to postprandial and insulin-stimulated glucose disposal. The topic of how the skeletal muscle of an offspring reacts to intrauterine hyperglycemia has remained elusive yet. In insulin-resistant offspring of a T2DM parent or a grandparent, their insulin-stimulated rates of muscle glucose uptake and non-oxidative metabolism decreased as well as a deficit in muscle TCA cycle flux observed (45–47). As mitochondrion is the main organelle for catabolizing fatty acids, less mitochondria and impaired oxidative function can produce a lipotoxic deposit urging insulin insensitivity (183, 184). Insulin-resistant offspring with a family history of diabetes mellitus tend to generate a lower maximal oxygen (O₂) consumption because of mitochondrial dysfunction (46). Another research has revealed that only O-GDM rather than O-T1DM presented a decreased expression of peroxisome proliferator-activated receptor γ coactivator-1α (PPARGC1A) in a skeletal muscle (48). Ppargc1α is known to be the key transcription cofactor of mitochondrial biogenesis (185). Another microarray in the offspring of T2DM parents showed that lipoprotein lipase (LPL) was lower, which resulted in a less fatty acid flux into the myotube (49). However, most of the muscle samples collected in these studies were biopsies taken from offspring with a family history of diabetes mellitus rather than from offspring with specific maternal diabetes mellitus, the result of which may be ambiguous. Animal models demonstrated that fetal and neonatal skeletal muscles were insulin resistant after only 2-day continuous intrauterine hyperglycemia (50). Aberrant mitochondrial dynamics and structure were seen in an O-PGDM mice skeletal muscle from F1 to F3 via abnormal oocytes (51).

According to retrospective studies, PGDM was reported to be highly associated with congenital heart diseases (CHDs) (52–56) and fetal diastolic dysfunction (58). Even in non-diabetic pregnant women, an elevated glucose level in the first trimester was directly correlated with an increased risk of CHD in the offspring (57). In the animal model, isolated neonatal cardiomyocytes showed disrupted mitochondrial bioenergetics once exposed to late GDM (66), which caused mitochondria-mediated cell death under metabolic stress in aged male offspring (67). PGDM neonatal cardiomyocytes were reported to have cardiac dysfunction and hypertrophy with an inefficient ATP production and an increased lipid peroxidation (68). The fetal heart of PGDM mice exhibited increased reactive oxygen species (ROS) levels, impaired epicardial epithelial-to-mesenchymal transition (EMT), and decreased coronary artery volume (69). Actually, endothelial cells from human umbilical vein endothelial cells (HUVECs) also displayed damaged proliferation and migration and intensive apoptosis in hyperglycemic fetuses (59). A comparative analysis of gene expression showed poor endothelial tube formation and compromised angiogenic capabilities in isolated GDM HUVECs and mesenchymal stem cells from GDM neonates (60). The proteomic analysis of human fetal endothelial cells revealed an alteration in antioxidant signaling in GDM groups. When ROS attacked, nuclear factor erythroid-related factor 2 increased its expression, traveled into nucleus, and bound to an antioxidant response element sequence for restoration of vascular redox homeostasis (147). Endothelial colony-forming cells (ECFCs) play a key role in repairing damaged endothelium and forming new blood vessels. Researchers found enhanced senescence and worsened angiogenesis in GDM neonate-derived ECFCs (61). Another article found that a reduction in circulating progenitor cells (CPCs) in GDM neonates was correlated with loss of angiogenesis (62). Vascular dysfunction could lead to hypertension and heart insufficiency in an adult. It was reported that maternal hyperglycemia in pregnancy was associated with at least one higher blood pressure (BP) value in girls (63), but the difference in BP between groups became insignificant after adjusting for offspring BMI (186). A Danish follow-up study showed that maternal diabetes mellitus was more likely to develop a hypertensive disease, heart failure, deep vein thrombosis, and pulmonary embolism as well as an increased risk of early onset in offspring (64). Furthermore, during the Trier social stress test (TSST), there was an increase in systolic BP, cardiac output, and stroke volume in O-GDM, even after adjusting for various demographic factors such as sex, age, socioeconomic status, and BMI (65). Animal experiments bear witness to the clinical findings mentioned above. Echocardiography demonstrated a significant diastolic dysfunction in O-GDM male rats. In isolated hearts, the baseline cardiac function and left ventricular compliance were significantly diminished (70). Diabetic offspring was also shown to have a higher blood pressure (71), a greater vasoconstriction under electrical stimulation (72), an impaired vascular nitric oxid (NO)–ROS signaling (148), and endothelial dysfunction (73) in adulthood. HFD after birth induced cardiomyocyte hypertrophy, pro-inflammation status, and cardiac trauma in O-GDM (143). Ischemia-induced cardiac infarction size was bigger in O-GDM. Meanwhile, the ROS leveled up, and the heart function was significantly inhibited (74). The impairment of arterial response and a decrease in the flow in renal peripheral vessels were also shown in maternal diabetic offspring (75), which indicated that kidney was another target of intrauterine hyperglycemia.

As the kidney plays an important role in the pathogenesis of hypertension and diabetic complications, the influence of maternal diabetes mellitus on kidney development in offspring has sprouted out. Epidemiologic studies suggested that O-PGDM was associated with strong teratogenic effects on the kidney (76) and the urinary tract (77). However, another study concluded that the maternal glucose level was not associated with volume and dimensions of fetal kidney (78). When children aged 5–18 years were affected, urinary calcium and magnesium excretion was lower in O-T1DM (79). In animal models, widespread interstitial fibrosis (80) and severe glomerulosclerosis (71, 81) with functional glomeruli loss (82, 83) were present at different periods in maternal diabetic offspring, leading to a decreased creatinine clearance (87), an increased glomerular filtration rate (88), and microalbuminuria (80). Urinary calcium and magnesium output was less (84, 85), similar to the excretion of basal sodium and urine volume (86). Apart from this fact, activation of the intrarenal renin–angiotensin system (RAS), together with intrarenal dopamine receptor phosphorylation (86) and increased transforming growth factor-β1 (Tgfβ1) (80), was evident in the cortex of hypertensive offspring of diabetic mothers (80, 82, 88, 89).

It also matters whether prenatal exposure to high glucose affects neuropsychiatric development. According to epidemiology, GDM led to adverse neurocognitive and behavioral outcomes in offspring, such as intellectual disability (90, 91) and psychiatric disorders (92, 93). A stratified analysis undertaken on diabetic mothers displayed a higher hazard ratio of autism spectrum disorder (ASD) in their offspring after adjustment (95). A retrospective study suggested that when maternal diabetes required medication, it was positively associated with the incidence of attention deficit/hyperactivity disorder in the offspring (92). Comparing GDM mothers treated with lifestyle or antidiabetic agents with their non-diabetic counterparts, the linear regression analysis showed that O-GDM mothers developed their neuropsychiatric conditions at a younger age (94). Exposure to diabetes mellitus in utero also led to eating disorders in the offspring at adolescence (96, 97). The hypothalamic blood flow in response to glucose in children of GDM mothers was obviously higher than control, combined with hypothalamic sensitivity to glucose stimulation (98). Based on experimental research, we arrived at conclusions as follows. First, anxiety-like behaviors were reduced in O-GDM adult mice as seen from the fact that they spent more time in center and open fields than control mice (99, 100). Male mice of GDM were reported to be more alarmed by the fear recall (105); second, impairment of recognition memory was demonstrated by the Novel Object Recognition (NOR) test as the diabetic offspring were not capable of distinguishing familiar from novel objects (101) or after they were fed HFD (102). Additionally, in the radical Y-maze test, O-GDM adult mice made more re-entry mistakes, which indicated a deficiency in short-term memory (100, 103, 104); third, the cell cycle was altered to a stage where apoptosis-related genes were promoted in GDM neonate-derived cells (107). Hippocampi derived from the offspring of diabetic mothers were characterized by cell death and susceptibility to LPS stimulation (145); fourth, hyperglycemia led to malformation of the brain (105), especially disturbed neocortical lamination (106), neural tube malformation (107), and hippocampal synaptic derangement (101, 108). Difference in the number of neurons in the ventromedial nucleus of the hypothalamus vanished after an islet transplantation in the diabetic group (187); fifth, fetal hippocampal formation was exposed to a neuroinflammatory environment because of hyperglycemia, so increased activated microglia and cytokines persisted into young adulthood (101), and O-GDM became more sensitive to an inflammatory response (145). However, another article suggested that microglia were not affected in GDM neonatal hippocampi (104); finally, both hippocampus and cortex were demonstrated to have a higher level of ROS and an enhanced oxidative metabolism together with inhibited superoxide dismutase (SOD) activity (100, 107). Importantly, overexpressing SOD2 in 6-week-old O-GDM can partly rescue its autism-like behavior in 8 weeks (149). There was also some disagreement among researchers on topics such as neural differentiation, where one researcher tested out that neuronal differentiation markers were less in PGDM fetal brains (107); while another stated that neural stem cells isolated from a 7-day-old fetus exited cell cycle in advance and entered premature differentiation after exposed to maternal hyperglycemia (106). In addition, hippocampal excitability was tested in cultured hippocampal neurons: action potentials were larger and the decay time was shorter in hyperglycemic-exposed offspring; and the resting potential was more hyperpolarized because outwards potassium channel density was larger (99).

As mentioned above, the influence of intrauterine adverse environment could inherit to the second generation (24, 103) through fetal primordial germ cells (169), thus it is reasonable for us to consider that maternal diabetes mellitus leaves negative effects on the reproductive system. Actually, a Danish national cohort study reported that female O-GDM had an earlier onset of puberty (109). In animal models, female offspring of diabetic mothers showed a smaller ovarian section (110). The number and diameter of primary follicles were also decreased (110, 111). O-GDM were more inclined to ovulate fewer oocytes after HFD, thus causing subfertility in adult life (112). Researchers attributed these phenomena to increases in apoptosis (110) and oxidative stress (111). Proteomic analysis showed that differentially expressed proteins identified in ovaries were involved in hydrogen peroxide catabolic process (151). A number of scientists have suggested that, because of lipidomic metabolites (113) and hyperactivity of insulin signaling (114), high glucose increased their ovarian weight and estrogen levels in diabetic female offspring. As for male, serum testosterone levels and sperm count were decreased in GDM male offspring (113, 115–117), possibly due to ROS (117) and apoptosis (116). Researchers once discovered that GDM inhibited glycogen synthase kinase-3β (GSK3β) signaling and delayed the differentiation of stem Leydig cells in fetal testis. Thus, the number of Leydig cells decreased and the anogenital distance was shortened (115). However, in another animal model, scientists suggested that the anogenital distance index was significantly increased, testes descent and preputial separation were started earlier when exposed to maternal hyperglycemia (118).

Diabetic mothers were believed to elevate the number of bone marrow myeloid progenitors and total cellularity in male offspring after a 10-month high fat diet (123). Hematopoiesis gives rise to circulating and tissue-resident immune cells (188). In fact, researchers also found that immune cells in offspring failed to function after exposed to maternal hyperglycemia at different periods. Splenocytes were induced to secrete more IL1β after a persistent HFD diet in O-PGDM (124). In the offspring of diabetic mothers, the neutrophil count dropped (123), while the lymphocyte count increased (119, 120). The percentage of activating T cells also decreased (119, 121), so did the capacity of proliferation after antigen stimulation (125, 126). T1DM mothers increased the suppressor T cells in offspring (120, 122). The proliferative ability of B cells was also found to be restrained in diabetic offspring (125). Non-obese diabetic (NOD) mouse colonies can be divided into two groups, NOD^low indicates a lower incidence for T1DM development and NOD^high indicates a higher incidence for T1DM development (189). B cell activation was demonstrated to be encouraged in mesenteric lymph nodes of NOD^high and co-housed NOD^low offspring (127). The disturbance of immune system led to higher pro-inflammatory cytokines, such as IL1β, IL6, and tumor necrosis factor-α (TNF-α), and lower anti-inflammatory cytokine such as IL2 (125, 126).

In addition to the organs and systems mentioned above, other organs that are affected by intrauterine hyperglycemia have been researched. Newborns and 3-week-old offspring of diabetic mothers had smaller lungs (128) and delayed maturation (129). Male O-GDM had increased lung compliance and reduced lung resistance at 10 weeks old, which may evolve into a chronic obstructive pulmonary disease (130). A high glucose environment was also demonstrated to activate the apoptosis of primary dental papilla cells and dental epithelial stem cells via nuclear factor κB (NF-κB) signaling pathway, resulting in impaired odontogenesis in maternal diabetic offspring mice (131).

Recently, gut microbiota of mothers was thought to be associated with PGDM and GDM (190–192). Remarkably, the microbiota composition of neonates was associated with GDM pregnancy. One research study stated that GDM was related to increase in microbes that are involved in suppressing an early immune cell function in neonates (132); whereas the other research study suggested that a 7-day-old O-GDM showed a higher relative abundance of pro-inflammatory taxa (133). The fecal microbiota of infants aged 7 days and 9 months showed consistent difference between O-GDM and control, but some taxa did not (134). A cross-section study showed that genus Anaerotruncus was increased in 5-year-old children of GDM through gut microbiome. Mothers and their children had a more similar microbiome composition when compared with others (135). The mice model also demonstrated that gut microbiota could be transmitted via delivery and perinatal nursing. Multi-omics analysis supported that both glucometabolic deficits and microbial composition in PGDM mothers were subsequently transmitted to their offspring. By Cesarean delivery and cross-fostering, the microbiota vertical transmission and insulin resistance were blocked (136). There was yet another interesting phenomenon. Exposing NOD^low mice to NOD^high after weaning cannot increase the incidence of T1D. However, if co-housing NOD^low and NOD^high mothers in advance, the offspring of NOD^low acquired increased the incidence of diabetes, which indicated that maternal microbiota and lactation environment were two important transmissible diabetes-promoting factors (127).

Another potential mechanism is inflammation. As we summarized above in the section “Target on hemopoietic and immune system”, immune cells were reprogrammed under intrauterine hyperglycemia. Not only in fetal or maternal cord blood, but also in children's peripheral blood, innate immunocytes such as leukocytes (120, 123, 137) and acquired immunocytes such as lymphocytes (119–122) were all dysfunctional, forming a status of imbalance of increased pro-inflammatory cytokines and decreased anti-inflammatory cytokines (124–126, 139). In periconception diabetic mice, the intrauterine pro-inflammatory factors, such as IL1α, TNFα, and interferon-γ (IFNγ), were increased. Thus, the blastocyst development was impaired, the embryo implantation was inhibited, and the growth and development of the fetus during middle and late gestational periods were delayed (140). Systematic pro-inflammatory cytokines or local immune cells acted on islets (27, 141), liver (35), peripheral adiposity (142), heart (75, 143, 144), kidney (80), brain (145, 146), reproductive tract (140), lung (130), and microbiota (133), resulting in damage to organ function.

Excess glucose in blood and enhanced intracellular glucose oxidation lead to a mitochondrial overproduction of superoxide (193). The ROS acts as an intracellular toxic substance (194) and amplifies mitochondrial damage of hyperglycemia and lipotoxicity. The role of ROS in the pathogenesis of metabolic diseases (195), including atherosclerosis (196) and neurodegenerative diseases (197), has been highlighted. Diabetes-exposed neonatal cardiomyocytes showed a decreased mitochondrial copy number and an impaired palmitate oxidation. Lipotoxicity (66) and ROS accumulation (69) in the heart resulted in cardiac dysfunction. Impaired NO–ROS signaling and increased superoxide production in GDM-exposed endothelial cells (147) and arteries led to hypertension (148). Furthermore, treatment with antioxidant or mitochondrial transfer in diabetic offspring could rescue deterioration in heart ischemic injury (74) and cardiomyocyte bioenergetics (68). Activation of the oxidative stress pathway was also demonstrated in renal vessels among diabetic offspring (75). SOD2, an enzyme that converts superoxide to less reactive hydrogen peroxide and diffuse freely, is strongly involved in the progression of neurodegenerative diseases (198). The transcription of Sod2 was found to be repressed through epigenetic mechanisms in maternal diabetes-induced autistic brain (100, 107), including neurons (149) and hematopoietic stem cells (150). In addition, the imbalance in oxidant and antioxidant was suggested to be an underlying factor resulting in testicular malformation (117) and altered oocyte–granulosa interaction (111) in diabetic offspring.

Telomere was thought to be related to cellular lifetime. Higher percentage of trophoblast cells exhibited telomere capture in GDM pregnancies (152). Telomerase activity was higher in cord blood from T1DM and GDM, but not T2DM pregnancies (153). Telomere length was observed to be shortened in GDM-born girls' peripheral blood at 9–16 years of age when compared to the control female. Telomere length showed significant association with maternal GDM status positively and insulin levels or HOMA-IR negatively (154). The absence of telomere shortening and oxidative DNA damage were observed in 16- to 23-year-old adult O-T1DM (155).

Cell cycle was another factor associated with cell viability. Cyclin-dependent kinase (Cdk), a cell cycle marker, was transcriptionally inhibited in diabetes-exposed islets (156). Inflammation (145) and ROS (107) in diabetes-exposed brains both could modulate cell cycle, contributing to apoptosis and neurodegeneration. In addition, cell cycle in reproduction plays an important role in gametogenesis. Intrauterine hyperglycemia impaired neonatal folliculogenesis (114) and initiation of meiosis (110) via promotion of apoptosis and inhibition of proliferation. In vivo and in vitro experiments also demonstrated that hyperglycemia suppressed proliferation and enhanced apoptosis in tooth germs via activation of NF-κB signaling pathway (131).

Gestational complications are thought to influence the offspring by means of epigenetic changes accordingly. Epigenetics mainly consists of DNA methylation, histone modification, and microRNAs without alterations in DNA sequence.