Dietary polyphenols

The MedDiet is characterised by a high intake of vegetable food⁽ Reference Trichopoulou, Bamia and Trichopoulos ^{31 )}, and polyphenols are the biggest class of plant-derived bioactive compounds⁽ Reference Bravo ^{32 –} Reference Neveu, Perez-Jimenez and Vos ^{34 )}. These phytochemicals are found in quite variable quantities in fruit, vegetables, cereals, nuts, tea, wine, chocolate, olives, extra-virgin olive oil and plant-derived foodstuffs, as well as spices and algae⁽ Reference Neveu, Perez-Jimenez and Vos ^{34 –} Reference Perez-Jimenez, Fezeu and Touvier ^{36 )}. Even more data link polyphenol intake with both health promotion and the prevention of non-communicable diseases such as CVD, stroke, T2D and some cancers⁽ Reference Xiao and Hogger ^{27 –} Reference Oh and Jun ^{29 ,} Reference Tsao ^{37 )}. Polyphenols are a complex class of compounds having a phenolic ring in their structure; they can be classified on the basis of the numbers of phenol rings they contain and the structural elements that bind these rings⁽ Reference Bravo ^{32 ,} Reference Manach, Scalbert and Morand ^{35 )}. The majority of polyphenols exist as glycosides, esters, polymers and in hydroxylated form⁽ Reference Manach, Scalbert and Morand ^{35 ,} Reference Tsao ^{37 )}. They are extensively modified throughout stomach, small intestine, colon and liver⁽ Reference Landete ^{38 )}; thus, a single polyphenol can generate several active metabolites⁽ Reference Del Rio, Costa and Lean ^{39 )}. Because of their chemical structures, dietary polyphenols exert multiple activities by interacting with several molecular pathways and cellular components including microRNA (miRNA)⁽ Reference Bahadoran, Mirmiran and Azizi ^{28 ,} Reference Tsao ^{37 ,} Reference Lancon, Michaille and Latruffe ^{40 ,} Reference Santangelo, Vari and Scazzocchio ^{41 )}. The main classes of polyphenols are flavonoids, phenolic acids, stilbenes and lignans⁽ Reference Bravo ^{32 ,} Reference Manach, Scalbert and Morand ^{35 )}. Flavonoids are the most abundant class of polyphenols in our diet and include different subclasses, that is, flavonols, flavones, flavanols, flavanones, anthocyanidins and isoflavones⁽ Reference Sebastian, Wilkinson Enns and Goldman ^{42 ,} Reference Zamora-Ros, Forouhi and Sharp ^{43 )} (Fig. 1). The evaluation of the actual contribution of dietary polyphenols to human health is challenging, as several factors come into play: food-related factors (i.e. amount consumed, food content and bioavailability) and host-related factors (i.e. genetics, obesity, pregnancy and gut microbiota)⁽ Reference D’Archivio, Filesi and Vari ^{44 )}. Moreover, chemical structure, synergistic effects among different polyphenols and interaction with cellular components (e.g. membrane, proteins, enzymes, receptors and transcription factors) render the evaluation quite hard⁽ Reference D’Archivio, Filesi and Vari ^{44 ,} Reference Fraga, Galleano and Verstraeten ^{45 )}. An accurate measurement of the exposure to specific compounds will make it possible to associate these factors with health status and disease outcomes. To this end, efforts have been made to assess the following: (i) the level of adherence to the MedDiet, by the Mediterranean Dietary Serving Score, which accounts for the consumption of foods and food groups per meal, day or week⁽ Reference Monteagudo, Mariscal-Arcas and Rivas ^{46 )}; (ii) the content of polyphenols in food and their metabolism⁽ Reference Neveu, Perez-Jimenez and Vos ^{34 )}, as well as the human intake of total and specific polyphenols, by the creation of online database (e.g. Phenol-Explorer and United States Department of Agriculture database)⁽ Reference Neveu, Perez-Jimenez and Vos ^{34 ,} Reference Sebastian, Wilkinson Enns and Goldman ^{42 ,} Reference Zamora-Ros, Forouhi and Sharp ^{43 ,} Reference Gonzalez, Fernandez and Cuervo ^{47 – 49 )}; and (iii) novel biomarkers for polyphenol ingestion in biological fluids, by measuring food metabolome⁽ Reference Wang, Tang and Wang ^{50 –} Reference Scalbert, Brennan and Manach ^{52 )}.

Fig. 1 Chemical structure and main dietary source of polyphenols discussed in this review regarding their capability to modulate glucose metabolism signalling.

Glucose homoeostasis regulation

GDM is defined as carbohydrate intolerance during pregnancy⁽ Reference Chen, Wang and Ji ^{1 )}. Pregnancy is characterised by a complex process of endocrine-metabolic changes, including physiological insulin resistance (IR), necessary to ensure the supply of nutrients to the fetus and to adequately prepare the maternal organism for childbirth and lactation⁽ Reference Sonagra, Biradar and K ^{53 )}. GDM develops when the pregnant woman is not able to produce an adequate insulin response to compensate for physiological IR⁽ Reference Catalano, Huston and Amini ^{54 )}. IR during pregnancy is uniquely associated with a decrease in insulin receptor substrate 1 (IRS-1) tyrosine phosphorylation, primarily because of a decreased expression of IRS-1 protein⁽ Reference Harlev and Wiznitzer ^{55 )}. Nevertheless, in GDM subjects, there is an additional decrease in tyrosine phosphorylation of the insulin receptor β subunit, associated with further decreases in glucose transport activity⁽ Reference Harlev and Wiznitzer ^{55 ,} Reference Catalano ^{56 )}. Lower levels of tyrosine kinase receptor activity (30–40 %), IRS-1 expression and activation, and GLUT4, in fact, have been detected in pregnancy with respect to pre-pregnancy status⁽ Reference Harlev and Wiznitzer ^{55 ,} Reference Catalano ^{56 )}. Specifically, IRS-1 tyrosine phosphorylation has been found to be significantly reduced in muscle from pregnant control and obese GDM subjects, as reflected by a 23 and 44 % reduction, respectively, compared with non-pregnant women⁽ Reference Friedman, Ishizuka and Shao ^{57 )}. AT, through a large arrays of secreted factors, participates in the induction and regulation of IR both in non-pregnant and pregnant subjects⁽ Reference Tersigni, Di Nicuolo and D’Ippolito ^{58 )}. Maternal pre-pregnancy obesity decreases insulin sensitivity and positively associates with the risk of GDM⁽ Reference Li, Liu and Guo ^{59 )}. A recent study has shown that AT inflammasome (caspase-1 and IL-1β) is involved in the development of IR in GDM-complicated pregnancies⁽ Reference Lappas ^{60 )}. Increased IL-1β expression has been observed in AT from GDM women; incubation of AT with IL-1β results in a significant attenuation of phosphorylated IRS-1 protein expression, GLUT4 mRNA and protein expression and glucose uptake, indicating the importance of inflammasome in the pathophysiology of GDM⁽ Reference Lappas ^{60 )}.

Over the past years, the role of miRNA in gene regulation has been gaining attention; miRNA are implicated in regulating cholesterol biosynthesis, carbohydrate and lipid metabolism⁽ Reference Krutzfeldt and Stoffel ^{61 )}, as well as in insulin production, secretion and action⁽ Reference Mao, Mohan and Zhang ^{62 )} (i.e. in glucose homoeostasis)⁽ Reference Mao, Mohan and Zhang ^{62 –} Reference Chen, Lan and Roukos ^{65 )}, by interacting with specific mRNA targets⁽ Reference Dweep and Gretz ^{66 )}. Accordingly, miRNA have been explored in GDM with the aim of using them as early biomarkers of disease⁽ Reference Collares, Evangelista and Xavier ^{67 –} Reference Zhu, Tian and Li ^{70 )}, as they are present in the blood or other biological fluids⁽ Reference Collares, Evangelista and Xavier ^{67 ,} Reference Etheridge, Lee and Hood ^{71 )}. To date, few investigations have been performed; two studies conducted in China have found decreased levels of miR-29a, -132 -and -222⁽ Reference Zhao, Dong and Jiang ^{69 )}, and increased levels of miR-16-5p, miR-17-5p, miR-19a-3p, miR-19b-3p and miR-20a-5p, in gestational weeks 16–19 in the serum of pregnant women who later developed GDM in gestational weeks 25–28, with respect to women who did not develop GDM⁽ Reference Zhu, Tian and Li ^{70 )}. However, hyperglycemic non-pregnant rodents have shown opposite expression of mir-29 and mir-132 families. MiR-29 family is involved in the insulin-signalling pathway; muscle, fat and liver from hyperglycaemic Goto–Kakizaki rats exhibit up-regulation of miR-29a and miR-29b⁽ Reference He, Zhu and Gupta ^{72 )}. High glucose induces the up-regulation of miR29a associated⁽ Reference He, Zhu and Gupta ^{72 )} with the decrease in IRS-1, downstream kinase Akt and glycogen synthase kinase 3β, (GSK3β) proteins, in both rat myocytes⁽ Reference Yang, Jeong and Park ^{73 )} and human pancreatic β-cells⁽ Reference Bagge, Clausen and Larsen ^{74 )}. MiR-132 expression is up-regulated in pancreatic islets of pre-diabetic (6 weeks old) and diabetic db/db mice (14–20 weeks)⁽ Reference Nesca, Guay and Jacovetti ^{75 )}; this increase improved glucose-stimulated insulin release and increased cell proliferation, which suggests that the modification of miR-132 levels might contribute to compensatory β-cell mass expansion elicited in response to IR⁽ Reference Nesca, Guay and Jacovetti ^{75 )}. In GDM women, the up-regulation of miR-222 has been found to be associated with reduced protein levels of both oestrogen receptor (ER)-α and GLUT4 in omental AT, obtained at the time of caesarean delivery, and with increased serum estradiol levels⁽ Reference Shi, Zhao and Guo ^{76 )}. In consideration of these findings, we may assume that miRNA are expressed in a species-, tissue-specific and time-dependent manner.

Furthermore, miRNA profiles of peripheral blood mononuclear cells isolated from Brazilian GDM women, obtained by using microarray platforms, have identified ten miRNA that seemed to be specific for GDM, namely, miR-101, miR-1180, miR-1268, miR-181a, miR-181d, miR-26a, miR-29a, miR-29c, miR-30b and miR-595⁽ Reference Collares, Evangelista and Xavier ^{67 )}.The target genes of most of these miRNA are involved in insulin signalling, angiogenesis, IR and AT dysfunction⁽ Reference Dweep and Gretz ^{66 )}. It is noteworthy that miR-181a is increased in the serum of diabetic patients, as well as in IR cultured hepatocytes and liver⁽ Reference Zhou, Li and Qi ^{77 )}. Inhibition of miR-181a by antisense oligonucleotides restores insulin sensitivity in hepatocytes, thus providing evidence of a potential therapeutic strategy for treating IR and T2D⁽ Reference Zhou, Li and Qi ^{77 )}. On the other hand, miR-30 belonging to the miR-30 family that has a key role in angiogenesis⁽ Reference Arola-Arnal and Blade ^{78 )} might be involved in the hyper capillarisation of the placenta in women with mild hyperglycaemia⁽ Reference Bridge, Monteiro and Henderson ^{79 ,} Reference Pietro, Daher and Rudge ^{80 )}. Furthermore, miR-30 family members increase in abundance during the differentiation of pancreatic islet-derived mesenchymal cells into hormone-producing islet-like cell aggregates, thus indicating their participation in the regulatory signalling of the embryonic development of human pancreatic islets⁽ Reference Joglekar, Patil and Joglekar ^{81 )}. However, the miRNA modifications so far observed in GDM women result from a limited number of screened subjects (from six to twenty). In addition, people from different countries appear to have different miRNA profiles, except for miR-29, in the same metabolic condition. In conclusion, although the above data provide evidence that miRNA are promising, non-invasive biomarkers of GDM, further research with larger cohorts is warranted.

Polyphenols effects

The maintenance of glucose homoeostasis is extremely important for human physiology⁽ Reference Zhang, Tobias and Chavarro ^{82 )}. Growing evidence indicates that polyphenols, contained in fruits and vegetables, might influence glucose homoeostasis by several mechanisms such as by (i) inhibiting carbohydrate digestion and glucose absorption in the intestine, (ii) stimulating insulin secretion from pancreatic β-cells, (iii) modulating glucose release from liver and (iv) activating insulin receptors and glucose uptake in insulin-sensitive tissues⁽ Reference Hanhineva, Torronen and Bondia-Pons ^{83 )}. The bioactive food components can modify gene expression and regulate different signalling pathways, thus affecting muscle, liver, pancreatic β-cells, hypothalamus and AT functions, thereby regulating glucose homoeostasis⁽ Reference Berna, Oliveras-Lopez and Jurado-Ruiz ^{84 ,} Reference Babu, Liu and Gilbert ^{85 )}. This regulatory activity occurs also by modulating miRNA gene expression⁽ Reference Lancon, Michaille and Latruffe ^{40 ,} Reference Garcia-Segura, Perez-Andrade and Miranda-Rios ^{86 –} Reference Jimenez, Dorado and Hernandez-Perez ^{89 )} (Fig. 2). Isoflavones, anthocyanins, flavanols and catechins appear to enhance β-cell function and glucose tolerance in animal models and humans and to protect against diabetes⁽ Reference Oh and Jun ^{29 )}. Genistein and daidzein are the major dietary isoflavones present primarily in soya foods⁽ Reference Crozier, Jaganath and Clifford ^{90 )}. Long-term genistein exposure (1–10 µm for 48 h) improved glucose-stimulated insulin secretion (GSIS) in mouse and human pancreatic islets⁽ Reference Fu and Liu ^{91 )}; this effect appears to be mediated by cyclic AMP/protein kinase A signalling activation and increased intracellular Ca²⁺ levels⁽ Reference Fu and Liu ^{91 )} (Table 1).

Fig. 2 Crosstalk among signalling pathways in regulating glucose metabolism. All of the factors that appear in this scheme are potential points of action of polyphenols. , Activation; , inhibition; , modulation; JAK/STAT, Janus kinase/signal transducer and activator of transcription; AMPK, AMP-activated protein kinase; JNK, c-Jun N-terminal kinase; IRS1/2, insulin receptor substrate 1/2; MAPK, mitogen-activated protein kinases; PDK1, 3-phosphoinositide-dependent protein kinase-1; PI3K, phosphatidylinositol-3-kinase; Akt, protein kinase B; SIRT-1, sirtuin 1; NO, nitric oxide; eNOS, endothelial NO synthase; FOXO1, forkhead box protein O1; TF, transcription factors; miRNA, microRNA; FA, fatty acids.

Table 1 Effects of dietary polyphenols on molecular mechanisms associated with gestational diabetes mellitus

↑, Increases; ↓, decreases; NOD, non-obese diabetic; JNK, c-Jun N-terminal kinase; Foxo1, forkhead box protein O1; GSIS, glucose-stimulated insulin secretion; cAMP, cyclic AMP; PKA, protein kinase A; EGCG, epigallocatechin-3-gallate; IRS1/2, insulin receptor substrate 1/2; AMPK, AMP-activated protein kinase; C3G, cyanidin-3-glucoside; Sirt-1, sirtuin 1; STZ, streptozotocin; PDX1, pancreatic and duodenal homeobox-1; PCA, protocatechuic acid; T2D, type 2 diabetes; Akt, protein kinase B; INS-1, insulin 1; GK, glycerol kinase; p38 MAPK, p38 mitogen-activated protein kinase; HbA1c, glycated Hb; PI3K, phosphatidylinositol-3-kinase; ERK1/2, extracellular signal-regulated kinase; LPS, lipopolysaccharides; NO, nitric oxide; eNOS, endothelial NO synthase; BAEC, bovine artery endothelial cells; Tyr, tyrosol; OL, oleuropein; HT, hydroxytyrosol; Tax, taxifolin; VEGFR2, vascular endothelial growth factor receptor 2; VEGF, vascular endothelial growth factor; HUVEC, human umbilical vein endothelial cells.

The flavanol epigallocatechin gallate (EGCG), a main component of green tea extract, preserves the insulin secretory machinery; EGCG (10 µm for 48 h) stimulates the activation of the IRS-2 signalling in rat insulinoma pancreatic β-cells (RIN-m5F), also under chronic hyperglycaemia⁽ Reference Cai and Lin ^{92 )}. EGCG increases the activity of AMP-activated protein kinase (AMPK) through Thr¹⁷² phosphorylation, further strengthening the hypothesis that AMPK is involved in counteracting the glucolipotoxicity induced by high-glucose condition⁽ Reference Cai and Lin ^{92 )}. AMPK is an energy-sensing enzyme recognised as a master regulator of whole-body energy homoeostasis; its activation increases GLUT4 expression and membrane translocation in skeletal muscle, thereby improving glucose uptake⁽ Reference Yu, Qiu and Nan ^{93 )}. In this regard, AMPK has been considered as one of the targets of the insulin-sensitising drugs metformin and thiazolidinediones⁽ Reference Yu, Qiu and Nan ^{93 )}.

EGCG (20, 40 μm for 24 h) improved the insulin-stimulated glucose uptake also in dexamethasone-treated skeletal muscle cells L6, associated with an increased AMPK phosphorylation and GLUT4 membrane translocation⁽ Reference Zhang, Li and Liang ^{94 )}. Furthermore, the dexamethasone-stimulated inactivation of IRS-1 because of increased Ser³⁰⁷ phosphorylation has been significantly reversed by EGCG treatment⁽ Reference Zhang, Li and Liang ^{94 )}. Results from the miRNA profiling suggest that EGCG may exert its biological functions through up- or down-regulating multiple miRNA as well⁽ Reference Tsang and Kwok ^{95 )}. In particular, EGCG (50 mg/l for 5 h), grape seed (100 mg/l for 5 h) and cocoa proanthocyanidin extracts (100 mg/l for 5 h) down-regulated miR-30b expression in human hepatocellular carcinoma HepG2 cells⁽ Reference Arola-Arnal and Blade ^{78 )}. A longer EGCG (100 μm for 24 h) treatment reduced miR-181d and miR-222 expression as well, in HepG2 cells⁽ Reference Tsang and Kwok ^{95 )}.

Anthocyanins are widely distributed in the human diet through berries, fruits, vegetables and red wine⁽ Reference Takikawa, Inoue and Horio ^{96 )}. Pretreatment with bayberry fruit extracts (0·5 μmol/l cyanidin-3-glucoside (C3G) for 12 h) up-regulated pancreatic duodenal homeobox-1 gene expression, which is associated with increased levels of insulin-like growth factor-II gene transcript and insulin, in rat insulinoma cell line INS-1β-cells⁽ Reference Sun, Zhang and Zhang ^{97 )}.

We have shown that C3G (10·50 µm for 18 h) and its metabolite protocatechuic acid (PCA; 100 µm for 18 h) increased glucose uptake and GLUT4 membrane translocation in murine adipocytes and visceral human adipocytes⁽ Reference Scazzocchio, Vari and Filesi ^{98 –} Reference Scazzocchio, Vari and Filesi ^{100 )}. Specifically, PCA appeared to exert insulin-like effects by stimulating IRS-1 tyrosine phosphorylation and Akt activation⁽ Reference Scazzocchio, Vari and Filesi ^{98 –} Reference Scazzocchio, Vari and Filesi ^{100 )}.

Caffeic acid, naringenin and quercetin (10⁻⁶ µm for 72 h each), present in many plants, enhance GSIS and glucose sensitivity in rat pancreatic INS-1E cells⁽ Reference Bhattacharya, Oksbjerg and Young ^{101 )}. These compounds differently modulate gene expression profiles to improve β-cell survival and function during glucotoxicity (e.g. naringenin and quercetin up-regulated GLUT2, glucokinase, AKT1, whereas caffeic acid increased AKT2 and all the three compounds increased INS-1, mRNA expression)⁽ Reference Bhattacharya, Oksbjerg and Young ^{101 )}. Similarly, curcumin (5–15 µm for 24 h), a yellow pigment isolated from Curcuma Longa, increases insulin gene expression and GSIS in a dose-dependent manner in INS-1 cells⁽ Reference Song, Wang and Zhu ^{102 )}. In doing so, curcumin up-regulates the expression of GLUT2, the phosphorylation of IRS-1, phosphatidylinositol-3-kinase (PI3K) and Akt, suggesting that curcumin prevents the high-glucose-induced reduction of insulin expression and secretion by activating the PI3K/Akt/GLUT2 pathway in INS-1 cells⁽ Reference Song, Wang and Zhu ^{102 )}. Curcumin treatment modulates the expression profile of several miRNA, including up-regulation of miRNA-26a in A549 human lung adenocarcinoma cell line (curcumin: 15 µm for 48 h)⁽ Reference Zhang, Du and Wu ^{103 )} and miR-181a in BxPC-3 human pancreatic carcinoma cell line (curcumin: 10 µm for 72 h)⁽ Reference Sun, Estrov and Ji ^{104 )}.

Consistent with in vitro studies, animal intervention studies showed that supplementation with genistein (0·2 g/kg diet) or daidzein (0·2 g/kg diet) for 9 weeks preserved β-cells and reduced blood glucose levels by inducing insulin secretion, in non-obese diabetic mice⁽ Reference Choi, Jung and Yeo ^{105 )}. Similarly, administration of 0·1 ml of bayberry fruit extracts (containing 150 μg of C3G)/10 g of body weight twice per day for 30 d significantly reduced blood glucose and improved glucose tolerance in streptozotocin-induced diabetic mice⁽ Reference Sun, Zhang and Zhang ^{97 )}.

Changes in the expression of a large number of different miRNA have been observed after each polyphenol supplementation⁽ Reference Milenkovic, Jude and Morand ^{88 )}. MiRNA profile expression was evaluated in the livers of wild-type (C57B6/J) mice or apoE–/– mice fed diets supplemented with different polyphenols (including quercetin, hesperidin, naringenin, proanthocyanidin, caffeic acid, curcumin)⁽ Reference Milenkovic, Jude and Morand ^{88 ,} Reference Milenkovic, Deval and Gouranton ^{106 )}. MiR-29a and miR-30b expression were down-regulated by almost all the tested polyphenols; miR-222 was down-regulated by caffeic acid (300 mg/d for 2 weeks) and hesperidin (30 mg/d for 2 weeks); miR-181a was down-regulated by curcumin (30 mg/d for 2 weeks) and hesperidin and up-regulated by naringin (30 mg/d for 2 weeks), quercetin (30 mg/d for 2 weeks) and proanthocyanidin (300 mg/d for 2 weeks); miR-132 was up-regulated by naringin⁽ Reference Milenkovic, Jude and Morand ^{88 )}. This study highlights the importance of miRNA as modulators of dietary polyphenols activities, in vivo ⁽ Reference Milenkovic, Jude and Morand ^{88 )}.

Human intervention trials and cohort studies have reported that dietary supplementation with 22·5 g of whole blueberry twice daily (1462 mg of total phenolics, 668 mg of anthocyanins) for 6 weeks resulted in blood glucose concentration reduction and in insulin sensitivity improvement in obese, non-diabetic, IR subjects⁽ Reference Stull, Cash and Johnson ^{107 )}. Data from three prospective cohort studies involving 200 000 US men and women showed that a higher consumption of anthocyanin-rich fruit is associated with a lower risk of T2D⁽ Reference Wedick, Pan and Cassidy ^{108 )}. Collectively, anthocyanins and its glycosides, alone or in combination, improve glucose homoeostasis by influencing β-cell mass and function, insulin sensitivity, glucose uptake and insulin signalling⁽ Reference Babu, Liu and Gilbert ^{85 )}.

There is no doubt that dietary polyphenols have beneficial effects on β-cell function and glucose homoeostasis taking into account that most of them are capable of modulating almost all the players, including several miRNA, in the glucose uptake machinery. Although promising data have been obtained so far, no information on the effects of these bioactive compounds in GDM setting are available as yet.

Adipokines and gestational diabetes

It is well known that adipocyte-derived cytokines (adipokines) are involved in the modulation of a wide range of physiological processes including lipid metabolism, atherosclerosis, angiogenesis and insulin sensitivity⁽ Reference Lappas ^{109 ,} Reference D’Ippolito, Tersigni and Scambia ^{110 )}. The human gestational tissues, as well as AT, produce and release a large array of pro-inflammatory cytokines; the activation of some inflammatory pathways is necessary to induce maternal IR, which is required for the progression of normal gestation⁽ Reference Santangelo, Vari and Scazzocchio ^{9 ,} Reference Hauguel-de Mouzon and Guerre-Millo ^{111 ,} Reference Friis, Paasche Roland and Godang ^{112 )}. GDM associates with the amplification of the low-grade inflammation already existing in normal pregnancy⁽ Reference Barbour, McCurdy and Hernandez ^{113 ,} Reference Li, Chen and Li ^{114 )}. Indeed, serum TNFα, IL-1β and IL-6 levels are higher in GDM women as compared with control pregnant women⁽ Reference Xu, Jin and Sun ^{115 )}. Notably, pre-pregnant obesity in GMD women worsens this inflammatory picture by up-regulating TNFα, IL-1β and/or leptin, and by down-regulating genes involved in AT lipid metabolism such as PPARα, PPARδ, PPARγ, retinoid X receptor-α (RXRα) and sterol regulatory element-binding protein-1c (SREBP1c), in AT⁽ Reference Lappas ^{109 )}.

Over the past years, various adipokines such as adiponectin and leptin have been shown to play a role in normal pregnancy, as well as in complications of pregnancy including GDM⁽ Reference Santangelo, Vari and Scazzocchio ^{9 ,} Reference D’Ippolito, Tersigni and Scambia ^{110 ,} Reference Miehle, Stepan and Fasshauer ^{116 ,} Reference Retnakaran, Hanley and Raif ^{117 )}. Low adiponectin serum levels appear to be linked with T2D and IR⁽ Reference Rasouli and Kern ^{118 )}. A study conducted on 445 pregnant women demonstrated that lower levels of adiponectin during the first trimester of pregnancy are associated with increased risk of developing GDM during the second trimester, independently of other early pregnancy major risk factors⁽ Reference Lacroix, Battista and Doyon ^{119 )}. Further, a nested case–control study showed that lower adiponectin levels measured 6 years before pregnancy were associated with a 5-fold increased risk of developing GDM⁽ Reference Hedderson, Darbinian and Havel ^{120 )}. Thus, an early low adiponectin expression, which most likely reflects pre-existing IR, could be considered a useful biomarker to identify women at high risk for GDM and allow for early therapeutic interventions⁽ Reference Hedderson, Darbinian and Havel ^{120 )}. Adiponectin acts as an insulin-sensitising agent, and its action is complex and incompletely defined, but the collected evidence suggests that, by binding with its receptors, adiponectin activates three key signalling pathways in muscle and liver: AMPK, p38 mitogen-activated protein kinase (p38 MAPK) and PPAR⁽ Reference Yamauchi, Kamon and Minokoshi ^{121 )}. The activation of these pathways results in FA oxidation (FAO) and glucose uptake in skeletal muscle, and in the inhibition of gluconeogenesis in the liver⁽ Reference Yamauchi, Kamon and Minokoshi ^{121 )}. Reduced adiponectin could be held accountable for the lower FAO observed in placentas of GDM women compared with the control group⁽ Reference Visiedo, Bugatto and Sanchez ^{122 )}. Adiponectin gene expression, processing and secretion is up-regulated by PPARγ agonist; PPARγ activation results in increased insulin sensitivity, in skeletal muscle and liver, by inducing the release of insulin-sensitising adipokines including adiponectin⁽ Reference Astapova and Leff ^{123 )}. Administration of adiponectin to diabetic mice⁽ Reference Berg and Scherer ^{124 )} and obese pregnant mice enhances insulin activity⁽ Reference Aye, Rosario and Powell ^{125 )}, prevents fetal overgrowth⁽ Reference Aye, Rosario and Powell ^{125 )}, normalises placental insulin/mammalian target of rapamycin complex 1 (mTORC1), PPARα signalling and placental nutrient transport⁽ Reference Aye, Rosario and Powell ^{125 )}. Adiponectin also mediates the crosstalk between AT and the vessel wall; thus, in virtue of its pleiotropic activities on multiple targets, it seems to have a protective role against vascular dysfunction induced by obesity and diabetes⁽ Reference Li, Cheng and Lam ^{126 )}.

TNFα and leptin have been suggested as the strongest predictors of pregnancy-related IR⁽ Reference Lepercq, Cauzac and Lahlou ^{127 ,} Reference Lappas ^{128 )}. Hyperleptinaemia in early pregnancy (13 weeks) appears to be a good predictor of GDM risk, independent of maternal adiposity⁽ Reference Qiu, Williams and Vadachkoria ^{129 ,} Reference Al-Badri, Zantout and Azar ^{130 )}. In week 30 of pregnancy, leptin levels have been found to be similar in both GDM and no-GDM subjects, but they remained elevated after delivery in the GDM group⁽ Reference Saucedo, Zarate and Basurto ^{131 )}.

Recent systematic reviews have highlighted higher and lower levels of plasma leptin and adiponectin, respectively, in the first or second trimester of pregnancy⁽ Reference Bao, Baecker and Song ^{132 )}, and this condition persisted, associated also with high TNF-α levels, in the late second or third trimester of pregnancy⁽ Reference Xu, Zhao and Chen ^{133 )}, in GDM patients compared with normal pregnancies. These findings corroborate the role exerted by the altered balance pro-inflammatory/anti-inflammatory factors in the impaired glucose homoeostasis in GDM.

Increased levels of leptin have been found in visceral AT of obese and GDM women⁽ Reference Lappas ^{109 )}. Leptin, predominantly produced by adipocytes, is considered an essential factor in maintaining energy homoeostasis, as it regulates food intake and energy expenditure via specific receptors in the hypothalamus⁽ Reference Hegyi, Fulop and Kovacs ^{134 )}. Leptin acts by activating Janus kinase/signal transducer and activator of transcription signalling, although other important intracellular signalling including PI3K, AMPK and MAPK appear to be involved as well⁽ Reference Hegyi, Fulop and Kovacs ^{134 )}.

In GDM women, TNFα and leptin levels are elevated; the increased TNFα expression causes a chronic inflammatory environment with enhanced leptin production, which, in turn, increases TNFα and IL-6 production by monocytes⁽ Reference Santos-Alvarez, Goberna and Sanchez-Margalet ^{135 )}, generating a vicious circle that amplifies the inflammatory situation and worsens the metabolic dysfunction in GDM. Although the significance of leptin and adiponectin is relatively well characterised as for their effects on glucose and lipid metabolism in diabetes, the molecular mechanisms by which these adipokines exert their effects on insulin action are not completely defined. Much more research is needed especially in the light of the growing list of molecules identified as adipokines, for example, visfatin, apelin, retinol-binding protein 4, vaspin, omentin and adiposity FA-binding protein 4 (FABP4), which renders the comprehension of the existing network and the interactions among them more complicated ⁽ Reference Santangelo, Vari and Scazzocchio ^{9 ,} Reference Wojcik, Chmielewska-Kassassir and Grzywnowicz ^{136 ,} Reference Li, Xiao and Li ^{137 )}.

The health effects of PUFA

A substantial amount of FA is required as additional source of energy to support fetal cellular growth, and gestational hyperlipidaemia normally occurring during late pregnancy enhances placental access to FA⁽ Reference Rebholz, Burke and Yang ^{158 ,} Reference Jones, Mark and Waddell ^{159 )}. GDM associates with an altered maternal lipid profile and affects the quantity and/or quality of lipids transferred to the fetus⁽ Reference Herrera and Ortega-Senovilla ^{160 )}. In GDM, a positive correlation between maternal TAG and NEFA levels, and fetal growth and fat mass, has been found even in diabetic mothers with appropriate glycaemic control⁽ Reference Herrera and Ortega-Senovilla ^{160 )}. Fetus development depends on the maternal supply of essential PUFA such as LA (18 : 2 n-6), α-linolenic acid (ALA, 18 : 3 n-3) and long-chain (LC)-PUFA, EPA (20 : 5 n-3), DHA (22 : 6 n-3) and arachidonic acid (AA, 20 : 4 n-6)⁽ Reference Haggarty ^{161 )}. The placenta regulates adequate LC-PUFA delivery to the fetus in a directional, preferential and timely manner⁽ Reference Haggarty ^{161 )}, preferentially transferring n-3 LC-PUFA⁽ Reference Jones, Mark and Waddell ^{159 )}. In GDM women, the placental transfer of LC-PUFA appears to be altered, with an impaired maternal-fetal transfer of DHA, which might affect neurodevelopment programming in the offspring⁽ Reference Gil-Sanchez, Demmelmair and Parrilla ^{162 )}. A reduced concentration of LC-PUFA in maternal, placental and fetal compartments in GDM women compared with healthy pregnant women has been highlighted⁽ Reference Pagan, Prieto-Sanchez and Blanco-Carnero ^{163 ,} Reference Araujo, Correia-Branco and Ramalho ^{164 )}.

However, increased, unchanged or lower hyperlipidaemia in diabetic pregnant women v. non-diabetic pregnant women has been observed too⁽ Reference Herrera and Ortega-Senovilla ^{160 )}. In this regard, it has been hypothesised that the degree of metabolic control and sex hormonal dysfunction may influence the different degree of dyslipidaemia observed in diabetic pregnant women⁽ Reference Herrera and Ortega-Senovilla ^{160 )}. NEFA are taken up by the placenta and FA are released from maternal lipoproteins by endothelial lipase (EL) and lipoprotein lipase (LPL), which are present in the maternal-facing microvillus membrane of the syncytiotrophoblast⁽ Reference Gil-Sanchez, Demmelmair and Parrilla ^{162 )}. Increased EL mRNA has been found in placenta from obese women with GDM compared with lean GDM women or normoglycemic pregnant women⁽ Reference Gauster, Hiden and van Poppel ^{165 )}. In contrast, a recent study has shown that no difference in the expression of LPL and EL exists between GDM and normoglycaemic placentas, suggesting that these lipases are not involved in the increased newborn adiposity⁽ Reference Barrett, Kubala and Scholz Romero ^{166 )}. NEFA may enter the cell by passive diffusion or by several plasma membrane-located transport/binding proteins including FA translocase (FAT/CD36), plasma membrane FABP(pm), FA transport proteins (FATP1–6) and intracellular FABP⁽ Reference Duttaroy ^{167 )}. DHA and EPA are endogenous ligands of several transcription factors such as PPAR, RXR and SREBP1⁽ Reference Jawerbaum and Capobianco ^{168 )}. A decreased expression of genes involved in FA uptake, intracellular transport, storage and synthesis (e.g. LPL, FATP2, FATP6, FABPpm, acyl-CoA synthetase long-chain family member 1), and of transcription factors involved in lipid metabolism regulation (e.g. liver X receptor (LXRα), PPARα, PPARδ, PPARγ, RXRα, SREBP1c), has been observed in AT obtained from obese pregnant women and women with GDM⁽ Reference Lappas ^{109 )}. These results indicate that pre-existing maternal obesity and GDM are associated with abnormal AT lipid metabolism, which may play a role in the pathogenesis of both diseases⁽ Reference Lappas ^{109 )}.

Dietary effects

PUFA are fundamental for human beings, and particularly important is the balance of n-6:n-3 FA in maintaining homoeostasis, normal development and mental health throughout the lifecycle⁽ Reference Haggarty ^{161 )}. Different studies have led to the conclusion that humans currently need a diet containing n-3 and n-6 PUFA in a ratio of about 1:5; on the contrary, in Western diets, this ratio is 1:10–20, indicating a deficiency in n-3 FA and/or an excess of n-6 FA⁽ Reference Simopoulos ^{169 )}. The essential FA, LA and ALA cannot be synthesised by humans and therefore must be supplied through the diet; a series of sequential desaturation and elongation reactions acting in concert transform LA and ALA into their unsaturated derivatives, namely AA from LA, EPA and DHA from ALA⁽ Reference Calder ^{170 )}. EPA and DHA are abundantly present in specific fishes, for example, fresh tuna, salmon and mackerel, as well as in fish oil and nuts⁽ Reference Russo ^{171 ,} Reference Zinati, Zamansani and Hossein KayvanJoo ^{172 )}. Nuts, vegetables, vegetable oils, for example, soyabean, flaxseed and rapeseed oil, and cereals are also food sources of ALA and LA^{( 49 ,} Reference Calder ^{170 )}. Walnuts are widely consumed in Mediterranean regions and are rich in ALA (2·95 g/28·4 g)⁽ Reference Burns-Whitmore, Haddad and Sabate ^{173 )}. In the PREDIMED (PREvención con DIeta MEDiterránea) trial that involved subjects at high risk for CVD, a diet rich in fats of vegetable origin, olive oil and mixed nuts (30 g/d, as 15 g walnuts, 7·5 g hazelnuts and 7·5 g almonds) has been shown to be beneficial in the management of the metabolic syndrome⁽ Reference Mayneris-Perxachs, Sala-Vila and Chisaguano ^{174 )}. The fat content of vegetables is low, but a high consumption of vegetables will result in a substantial intake of ALA. The content of ALA in some edible plants growing in the Mediterranean regions is remarkable, as in the case with purslane (300–400 mg ALA/100 g)⁽ Reference Galli and Marangoni ^{175 ,} Reference Uddin, Juraimi and Hossain ^{176 )}.

Different cohort studies have shown a positive association between the adherence to MedDiet style, and plasma DHA and EPA content, indicating that the MedDiet ensures higher n-3 PUFA bioavailability than Western diets⁽ Reference Panagiotakos, Kalogeropoulos and Pitsavos ^{177 –} Reference Scoditti, Capurso and Capurso ^{180 )}.

The European Commission and the International Society for the Study of Fatty Acids and Lipids specifically recommend that pregnant and lactating women consume a minimum of 200 mg DHA/d⁽ Reference Koletzko, Lien and Agostoni ^{181 )}. Several researches indicate that maternal dietary supplementation with n-3 PUFA during pregnancy can reduce the risk of pregnancy complications by limiting placental inflammation and oxidative stress⁽ Reference Jones, Mark and Waddell ^{159 ,} Reference Pietrantoni, Del Chierico and Rigon ^{182 ,} Reference Carvajal ^{183 )}, with positive action on insulin function, improvement of glucose tolerance and lipid profiles⁽ Reference Mohammadi, Rafraf and Farzadi ^{184 ,} Reference Lu, Borthwick and Hassanali ^{185 )} (Table 2).

Table 2 Maternal n-3 fatty acids (FA) supplementation

↑, Increases; ↓, decreases.

A daily dose of 600 mg of DHA, starting from the first trimester until delivery, can ameliorate red cell membrane anomaly in pregnant women with T2D and in neonates, and it prevents the decline of maternal DHA during pregnancy⁽ Reference Min, Djahanbakhch and Hutchinson ^{186 )}. These results suggest that DHA supplementation should be started in the antenatal care of pregnant women with T2D. An Italian pilot study revealed that daily administration of 200 mg of DHA, from the 8th week until delivery, reduced AA levels in erythrocytes and plasma during gestational time⁽ Reference Pietrantoni, Del Chierico and Rigon ^{182 )}. Thus, the increased DHA levels in the maternal plasma and the consequent reduction of AA-derived eicosanoids decrease inflammation, with a beneficial effect on fetal development and growth, and IR of amniochorial membranes as well⁽ Reference Pietrantoni, Del Chierico and Rigon ^{182 )}. New identified PUFA-derived mediators (specialised pro-resolving lipid mediators), namely resolvins, protectins, maresins and lipoxins, are involved in inflammatory resolution⁽ Reference Recchiuti and Serhan ^{187 )}. It has been observed that maternal dietary supplementation with n-3 PUFA might help placenta in resolving inflammation by increasing the levels of these pro-resolving mediators, in rat⁽ Reference Jones, Mark and Keelan ^{188 )} and in human placenta⁽ Reference Keelan, Mas and D’Vaz ^{189 )}. GDM women who received 1000 mg/d n-3 FA supplementation (containing 180 mg of EPA and 120 mg of DHA) for 6 weeks, starting in weeks 24–28 of gestation, showed a significant decrease in serum insulin levels and in HOMA index (homoeostasis model of assessment–IR)⁽ Reference Samimi, Jamilian and Asemi ^{24 )}, high-sensitivity C-reactive protein (CRP) and better newborn outcome⁽ Reference Jamilian, Samimi and Kolahdooz ^{190 )}, compared with placebo groups.

The need of higher PUFA intake in pregnancy is highlighted by data collected from 600 women participating in the Alberta Pregnancy Outcomes and Nutrition study⁽ Reference Jia, Pakseresht and Wattar ^{191 )}. Through the development of a dietary database for n-3 PUFA, the authors evaluated PUFA intake each trimester of pregnancy and 3 months postpartum: results indicate that the majority of women did not meet the European Union recommendation for DHA during pregnancy⁽ Reference Jia, Pakseresht and Wattar ^{191 )}.

The mechanism through which n-3 FA intake might influence insulin metabolism is not well known. It has been hypothesised that n-3 PUFA supplementation might affect insulin metabolism and lipid profiles by activating AMPK⁽ Reference Higuchi, Shirai and Saito ^{192 )}. In agreement with this finding, a recent study showed that RSV supplementation, before, and throughout, pregnancy increased placental DHA uptake capacity in pregnant non-human primates fed a high-fat diet (i.e. 36 % fat supplemented with 0·37 % RSV)⁽ Reference O’Tierney-Ginn, Roberts and Gillingham ^{193 )}. This effect was associated with increased AMPK activity and FATP-4, CD36 and FABPpm mRNA expression⁽ Reference O’Tierney-Ginn, Roberts and Gillingham ^{193 )}. In addition, n-3 PUFA might modulate gene expression through a variety of transcription factors with tissue-specific effects; n-3 PUFA improve insulin metabolism by inhibiting pro-inflammatory cytokine release and NF-κB protein expression⁽ Reference Bellenger, Bellenger and Bataille ^{194 )}, and by activating their G-protein-coupled receptor 120⁽ Reference Oh, Talukdar and Bae ^{195 )}.

A recent meta-analysis showed that long-term supplementation of marine-derived n-3 PUFA results in a significant reduction of IL-6, TNFα and CRP plasma levels, in both healthy and non-healthy subjects⁽ Reference Li, Huang and Zheng ^{196 )}.

Moreover, the increasing vegetarian and vegan population requires alternative sources of LC-PUFA to guarantee them the appropriate intake of n-3 FA⁽ Reference Saunders, Davis and Garg ^{30 ,} Reference Harris ^{197 ,} Reference Rajaram ^{198 )}. In addition to eating a larger quantity of vegetables, with a higher content of essential n-3 FA, a promising source of n-3 PUFA is represented by different species of microalgae⁽ Reference Adarme-Vega, Lim and Timmins ^{199 ,} Reference Lenihan-Geels, Bishop and Ferguson ^{200 )}. To this purpose, the European Commission has authorised the use in certain foods of oil derived from microalgae Schizochytrium sp. rich in DHA and EPA (EU decision no. 2015/545 and no. 2015/546),

Altogether, these data provide useful scientific information for planning future investigations aimed at exploring PUFA effects when considering a nutritional therapy of inflammatory diseases and GDM.

Conclusions and future research

GDM is increasing worldwide; obesity worsens this condition with increased risk of developing metabolic disorders in both mother and offspring later in life. The adoption of healthy lifestyle, with the adherence to a healthy dietary pattern, has positive effects on the prevention and management of diabetes. The Mediterranean diet is considered the model of healthy diet, and even more studies demonstrate its power in reducing the burden of several disorders in human being.

The few studies investigating the impact of healthy diet on GDM indicate that adherence to a healthy dietary pattern before pregnancy can reduce the risk of developing GDM; very few data about the association between dietary exposure and GDM outcomes are available.

Accordingly, this descriptive review tried to provide a picture on what we know about the impact of specific dietary biocompounds (i.e. polyphenols and PUFA) on the molecular mechanisms involved in glucose homoeostasis. Literature data, mostly derived from in vitro or animals studies, indicate that almost all subclasses of flavonoids, as well as the stilbene RSV and some olive oil phenolic compounds, interact and modulate several molecular pathways regulating insulin sensitivity in pancreatic β-cell, adipocyte, liver and muscle. Polyphenols drive activation and/ or silencing of transcription factors and consequently influence genes expression. Moreover, the fact that miRNA are the target of polyphenols action is to be taken into account in the future studies aimed at improving the understanding of the biological effects of polyphenols. However, how these compounds might work in GDM setting is a matter of speculation. Although increasing data indicate the beneficial effect of PUFA on pregnancy, a high-level evidence has not yet been achieved to make recommendations for GDM.

Many gaps, in the nutritional field and in the knowledge of pathophysiological processes, have to be filled in to progress in GDM prevention and management. GDM results from a complex interaction between the genetic background and environment. Identifying a number of modifiable biomarkers (including adipokines and miRNA) and the nutritional factors (e.g. polyphenols and their metabolites) able to target them can be useful in both early diagnosis and disease management.

It will be necessary to carry out dietary intervention studies involving large cohorts, and take into account genetic and dietary differences among races and countries. The ‘omics’ techniques will be fundamental tools to monitor dietary changes in populations and to identify the association between dietary exposure and disease outcomes. The integration of data from an individual’s genetic predisposition, endogenous metabolome, food metabolome and mirRNome profiles by systems biology approaches will provide a holistic view (systems biology approaches) of the relations between diet and disease, and will make it possible to translate these data in nutritional recommendations and diseases management. In particular, differences in metabolic profiles will be useful to identify novel biomarkers of dietary intake associated with disease. In this regard, the recent creation of a database to estimate intake and dietary seafood source of LC-PUFA in pregnancy provides a useful tool and represents an important step for future intervention studies.

Sharing this database along with others providing information on dietary polyphenols and polyphenol metabolites will help researchers to better understand the mechanisms governing interaction between dietary active compounds and the human organism.