Pregnancy‐induced changes in β‐cell function: what are the key players?

Maternal metabolic adaptations during pregnancy ensure appropriate nutrient supply to the developing fetus. This is facilitated by reductions in maternal peripheral insulin sensitivity, which enables glucose to be available in the maternal circulation for transfer to the fetus for growth. To balance this process and avoid excessive hyperglycaemia and glucose intolerance in the mother during pregnancy, maternal pancreatic β‐cells undergo remarkable changes in their function including increasing their proliferation and glucose‐stimulated insulin secretion. In this review we examine how placental and maternal hormones work cooperatively to activate several signalling pathways, transcription factors and epigenetic regulators to drive adaptations in β‐cell function during pregnancy. We also explore how adverse maternal environmental conditions, including malnutrition, obesity, circadian rhythm disruption and environmental pollutants, may impact the endocrine and molecular mechanisms controlling β‐cell adaptations during pregnancy. The available data from human and experimental animal studies highlight the need to better understand how maternal β‐cells integrate the various environmental, metabolic and endocrine cues and thereby determine appropriate β‐cell adaptation during gestation. In doing so, these studies may identify targetable pathways that could be used to prevent not only the development of pregnancy complications like gestational diabetes that impact maternal and fetal wellbeing, but also more generally the pathogenesis of other metabolic conditions like type 2 diabetes.


Introduction
Multiple organ systems in the mother adapt their function during pregnancy to support the nutritional demands of the fetus for growth and development. Of particular importance are the alterations in maternal metabolism, including the acquisition of an insulin-resistant state, which increase glucose availability in the mother for transfer to the fetus. To appropriately balance the level of insulin resistance attained in the mother during pregnancy, pancreatic β-cells must functionally adapt and increase their production of insulin (Rieck & Kaestner, 2010). Metabolic adaptations begin early in pregnancy and are accompanied by changes in maternal production of hormones, including prolactin (PRL), oestrogen, progesterone and cortisol (Ryan & Enns, 1988;Grattan et al. 2008). The placental secretion of hormones, which commences immediately following implantation and continues throughout gestation, is essential for mediating maternal metabolic adaptations via indirectly modulating endocrine axes and directly inducing changes in metabolic tissues of the mother (Braunstein, 2011). Of note, under the influence of maternal and placental hormones, β-cell mass expands and the capacity of β-cells to produce and secrete insulin increases in gestation.
Akin to outside of pregnancy, β-cells in the mother are responsive to other physiological and environmental stimuli, such as alterations in dietary intake and stress. Therefore, maternal environmental conditions likely also play important modulatory roles in influencing β-cell adaptations during pregnancy with metabolic consequences (Rieck & Kaestner, 2010;Banerjee, 2018).
Indeed, a failure to appropriately adapt β-cell function very probably occurs during pregnancy in the presence, for example, of maternal obesity or excessive gestational weight, which are both risk factors for gestational diabetes mellitus (GDM). GDM is a metabolic disorder that arises from an inability of the mother to compensate for the insulin resistance induced during pregnancy. This results in maternal hyperglycaemia and glucose intolerance with serious health outcomes for mother and offspring, including the development of type 2 diabetes mellitus (T2DM) in the decades after birth (Buchanan et al. 2012;Moyce & Dolinsky, 2018;Christoforou & Sferruzzi-Perri, 2020).
The aim of this review is to examine the mechanisms controlling β-cell adaptation during pregnancy and illuminate pathways that could be targeted to combat β-cell maladaptation and hence the pathogenesis of GDM. We first describe the changes in maternal glucose and insulin dynamics during gestation. We then evaluate the role of maternal and placental hormones in the regulation of β-cell adaptation and explore their underlying molecular mechanisms mediating functional changes in the β-cells during pregnancy. Finally, we explore the influence of maternal environmental conditions and their interaction with the endocrine and molecular mechanisms controlling β-cell adaptation in gestation. In doing so, we hope to highlight valuable future research areas and identify targetable pathways that could be used to prevent not only the development of pregnancy complications like gestational diabetes, but also more generally the pathogenesis of other metabolic conditions like type 2 diabetes. Glucose homeostasis, insulin sensitivity and β-cell adaptation during pregnancy Glucose homeostasis. Glucose levels in the mother dynamically change during pregnancy. In early gestation, fasting glucose levels drop compared to the pre-gravid state, in part due to the haemodilution effect caused by an increased maternal blood volume. Maternal fasting glucose levels remain consistently low in the second trimester and are further decreased during the third trimester, largely as a result of enhanced utilization of glucose in the circulation by the fetal-placental unit (Catalano et al. 1992). However, maternal fasting hypoglycaemia during gestation is compensated by enhanced hepatic gluconeogenesis, which raises glucose levels and helps to maintain nutrient flow to the fetus (Catalano et al. 1995). On the contrary, postprandial glucose levels progressively elevate during pregnancy, relative to the pre-gravid state (Cousins et al. 1980;Butte, 2000). This elevation is related to impaired peripheral tissue insulin sensitivity and hence diminished postprandial glucose utilization by the mother (Di Cianni et al. 2003).
Insulin sensitivity. Insulin sensitivity changes in the mother during pregnancy. In early gestation, maternal insulin sensitivity is enhanced, which facilitates the growth of maternal organs, such as the adipose, as well as the storage of nutrients like glucose as lipids or glycogen. This is driven in part by the suppression of pituitary growth hormone production by the feto-placental unit during the first weeks of pregnancy. However, later in gestation insulin sensitivity of maternal peripheral tissues, including the white adipose tissue and skeletal muscle, markedly decreases and contributes to the elevated postprandial glucose levels mentioned previously (Angueira et al. 2015).
Diminished insulin sensitivity is induced by increased levels of placental and maternal hormones, including placental lactogen, growth hormone variant, progesterone, cortisol and PRL, which interfere with insulin receptor signalling particularly during the third trimester of pregnancy (Newbern & Freemark, 2011;Napso et al. 2018). Other mediators, including leptin, adiponectin and tumour necrosis factor α (TNF-α) produced by the placenta and maternal white adipose tissue also contribute to the decreased insulin sensitivity of peripheral tissues and do so in a cooperative fashion (Lain & Catalano, 2007;Romero et al. 2007;Napso et al. 2018). For instance, whilst leptin (which acts in nutrient storage) increases during late gestation, lactogens (such as PRL) induce central leptin resistance leading to increased feeding behaviour and a maintenance of body weight during late gestation (Newbern & Freemark, 2011). Although a reduction in insulin sensitivity (as well as leptin resistance) may appear pathological from the maternal perspective, this is vital for the diversion of glucose from maternal tissues to the fetal-placental unit for growth and development. Insulin sensitivity starts to return to pre-pregnancy baseline just prior to or after term, which is thought to help meet the metabolic demands of the mother during labour and lactation and ensure metabolic health post-partum (Kirwan et al. 2004;Friedman et al. 2008).
β-Cell adaptation. Insulin levels in the mother also alter during gestation. In particular, maternal circulating insulin increases in pregnancy, which is facilitated by an enlargement of pancreatic β-cell mass and enhanced insulin secretory capacity (Lain & Catalano, 2007). There is an increase in β-cell proliferation, hypertrophy and survival, enhanced insulin synthesis and storage, and a lowering of the threshold for glucose-stimulated insulin secretion (GSIS) in pregnancy (Sorenson & Brelje, 1997). Whether adaptations in pancreatic β-cells occur in response to, or are in anticipation of, the diminished insulin sensitivity of the mother in pregnancy is a matter of debate. However, the timing, magnitude and contribution of changes in pancreatic β-cells during pregnancy and their relationship to maternal insulin sensitivity have been most extensively studied in rodents, with some scant studies performed in humans (Rieck & Kaestner, 2010). Because of this, the remaining work reviewed herein will largely be from rodents and humans. β-Cell mass adaptation during rodent pregnancy. Studies in rodents have shown that β-cell adaptations in the mother occur prior to the onset of insulin resistance and thus they are not simply in response to increased insulin demand (Parsons, 1992;Sorenson & Brelje, 1997). In mice where pregnancy lasts around 20 days, the proliferation of pre-existing β-cells in pancreatic islets commences around day 9 of pregnancy, with a maximum rate around days 12-15, as indicated by Ki67 immunostaining or BrdU incorporation (Parsons, 1992;Xue et al. 2010;Szlapinski et al. 2019). There may also be protection from apoptosis/enhanced survival Karnik et al. 2007) and an increase in the size of β-cells in islets during pregnancy (Parsons et al. 1995;Huang et al. 2009) and thus enhanced viability, hyperplasia and hypertrophy contribute to the expansion of β-cell mass in gestation. However, as early as day 6 of pregnancy, proliferation of low Glut2 expressing β-cells located in extra-islet β-cell clusters has also been observed (Beamish et al. 2017). Lineage tracing studies suggest that the expansion of β-cell mass in the mother may also result from the differentiation of non-insulin expressing progenitors that acquire a β-cell phenotype in pregnancy (Abouna et al. 2010;Toselli et al. 2014), although the origin of this non-β-cell source and the J Physiol 600.5 contribution of trans-differentiation of mature cells into β-cells are unknown. Moreover, it is important to note that other studies have failed to detect neogenesis in the maternal pancreas during pregnancy in several mouse models (Parsons et al. 1995;Zhao, 2014). Whatever the mechanisms involved, these processes result an increased number of islets (Hakonen et al. 2014) and a 3-to 4-fold expansion of β-cell mass, with values peaking around days 14.5-16.5 of mouse gestation (Parsons et al. 1992;Huang et al. 2009;Beamish et al. 2017;Rieck et al. 2009). After delivery, the maternal β-cell mass regresses to non-pregnant levels and an increase in apoptosis of the β-cells is thought to contribute to this process (Scaglia et al. 1995;Kim et al. 2010;Nunes et al. 2014). β-Cell mass adaptation during human pregnancy. Studies of β-cell adaptation in human pregnancy have come from analyses of pancreatic tissue recovered from women who died during pregnancy. Compared to rodents, many mechanisms seem conserved but there are some differences in β-cell adaptation during human pregnancy. For instance, two studies have shown that β-cell mass is increased around 1.4-to 2.4-fold in pregnant women, which is less than the reported 3-to 4-fold increase in mice (Van Assche et al. 1978;Butler et al. 2010). One of those studies also found there was no increase in size, proliferation or decrease in apoptosis of β-cells in existing islets, suggesting that islet neogenesis could contribute to β-cell mass adaptation in human pregnancy (Butler et al. 2010). However, studies in human samples tend to be from a single gestational period, which may not capture when proliferation rates may be elevated. Further work is thus required to understand the process and temporal regulation of β-cell adaptations, relative to changes in insulin sensitivity, in women during pregnancy.

Role of maternal and placental hormones in β-cell adaptation during pregnancy
Hormones secreted by maternal tissues and the placenta are important regulators of β-cell adaptation and, therefore, the control of glucose homeostasis during pregnancy. They function either directly by interacting with receptors on the β-cells or indirectly by modulating the nerves, vasculature or blood flow to the pancreas and/or via changes in other hormones that signal to the β-cells. The circulating abundance of hormones, such as placental lactogens, oestrogen, progesterone and kisspeptin, undergoes significant changes over the course of pregnancy, and together with the expression of local regulators, like hepatocyte growth factor (HGF) and serotonin, is instrumental to the adaptive increase in β-cell mass and insulin secretory function (Rieck & Kaestner, 2010;Banerjee, 2018). In this section we will describe the main maternal and placental endocrine and local factors facilitating β-cell adaptations during pregnancy by describing their effects on β-cells, as well as the molecular pathways governing such effects (from receptor to downstream signalling pathways, Fig. 1). To explore the importance of endocrine and local factors in β-cell adaptations during gestation, researchers have modified input signal, receptor union or abundance. Such experiments have largely been performed using genetic manipulation in rodents using the Cre-LoxP system. However, it is important to mention that there may be potential limitations due to differences in recombination events, allele susceptibility, timing and efficiency of deletion, and background of strain used when using the Cre-LoxP system (Magnuson & Osipovich, 2013). Moreover, the RIP-Cre and Pdx1-Cre mouse lines, which have been used to conditionally manipulate β-cell gene expression, have been reported to develop glucose intolerance and/or show signs of β-cell dysfunction even when there are no floxed transgenes to target (Lee et al. 2006;Pomplun et al. 2007;Brouwers et al. 2014) due to the inclusion of a bovine growth hormone (bGH) minigene when the lines were created (Brouwers et al. 2014). Thus, interpretation of findings for certain studies using the RIP-Cre and Pdx1-Cre mouse lines when appropriate Cre controls have not been used is warranted and highlighted below.

Endocrine factors.
Placental lactogen, prolactin and growth hormone. Placental lactogens (PL) are key members of the PRL-growth hormone (GH) family of hormones secreted by the placenta during gestation (Napso et al. 2018). In mouse and rat, the placenta expresses all the PRL-GH family members except for PRL and GH (namely at least 23 PRL-related proteins, including placental lactogens 1 and 2). This is in contrast to the human placenta, which only expresses the GH and PL genes (Soares et al. 2007). The anterior pituitary also produces PRL and GH, but production is diminished by mid-pregnancy, when placental hormone production predominates (Bridges, 2015). Placental lactogens play a central role in the β-cell adaptations during pregnancy (Banerjee, 2018;Sferruzzi-Perri et al. 2020). Indeed, the rise in PL concentrations during gestation is correlated with increased β-cell proliferation and function during pregnancy (Parsons et al., 1992). Several studies have examined the role of PL and PRL on pancreatic islets and shown that lactogens increase β-cell insulin secretion, proliferation, survival and mass and lower the threshold for GSIS (Sorenson & Brelje, 2009).
In mice, transgenic overexpression of placental lactogen in β-cells caused hypoglycaemia, hyperinsulinemia, increased β-cell proliferation and a twofold increase in β-cell mass (Vasavada et al. 2000). Conversely, global Prlr knockout mice have glucose intolerance and reductions in β-cell mass, insulin gene expression and insulin content (Freemark et al. 2002). Prlr knockout mice have reproductive abnormalities, including an inability to maintain pregnancy beyond mid-gestation, which is not ideal for investigations of pregnancy physiology (Ormandy et al. 1997). However, the examination of heterozygous Prlr null females (Prlr +/-) during pregnancy showed that PRLR is required to maintain normal gestational glucose homeostasis, as pregnant Prlr +/mice developed glucose intolerance and showed attenuated β-cell proliferation and mass expansion (Huang et al. 2009). These defects were related to reduced phosphorylation of JAK2, insulin receptor substrate 2 (IRS2) and protein kinase B (AKT), as well as a failure to decrease levels of menin and the cell cycle inhibitors p18 and p27 in pancreatic islets of Prlr +/mice during pregnancy (Hughes & Huang, 2011). Moreover, using progesterone replacement to allow pregnancy studies in global Prlr knockout mice, studies have shown that expression of PRLR is required for the induction of many of the pregnancy-related mRNA changes in mouse islets (Goyvaerts et al. 2015). Studies using conditional deletion of Prlr in β-cells have further revealed that PRLR is critical for the induction of several critical pro-proliferative factors during mouse pregnancy, including tryptophan hydroxylase 1 (TPH1), forkhead box protein M1 (FOXM1) and cyclins A2, B1, B2 and D1 (Banerjee et al. 2016). Although these studies used the RIP-Cre mouse line to delete Prlr in β-cells (Banerjee et al. 2016), as the bGH minigene functions through PRLR, interpretation of those findings should not be confounded. Thus, PLs/PRLs act through several signalling pathways and transcription factors to promote β-cell mass expansion and insulin production during pregnancy.
Whilst GH is closely related to PL/PRLs and can bind to PRLR (in addition to GHR), its actions on pancreatic islets seems to be different (Somers et al. 1994). In addition, there are differences in the impact of GH on β-cells between species. For instance, unlike PL and PRL, GH does not significantly increase insulin secretion by adult human islets in vitro . There are also marked differences in the activation of STAT5 by PRL and GH in the insulinoma cell line Ins-1 and cultured rat islets; STAT5A and B was activated in a biphasic and prolonged fashion by PRL, but this activation was not biphasic and transient in nature with GH (Brelje et al. 2004). In non-pregnant mice, deletion of the Ghr gene specifically in β-cells results in reduced GSIS but normal β-cell mass and glucose tolerance on a chow diet (Wu et al. 2011). It is also linked to severely blunted first-phase GSIS, reduced β-cell mass and glucose intolerance when these conditional Ghr mutant mice are challenged with a high-fat diet (Wu et al. 2011). However, the contribution of bGH, which is present in the RIP-Cre used in this study to conditionally delete Ghr gene in the mice, requires elucidation (as floxed mice were used as controls). Unlike PRLR, GHR is expressed in both α-cells and β-cells of the rodent pancreas (Brelje et al. 2004). Recently, in mice it has been described that pancreatic α-cells undergo important morphological and functional changes, such as reduced glucagon secretion during pregnancy, that are likely regulated by pregnancy hormones, including PLs (Quesada-Candela et al. 2020). Taken together, the available data suggest that GH may be a relevant endocrine factor regulating maternal β-cell insulin secretion and could have additional roles in regulating pancreas endocrine function during pregnancy.
Oestrogens and progestogens. The function of progesterone and oestrogens on β-cells during pregnancy is not totally understood . Circulating levels of progesterone and oestrogens continuously rise during gestation (Costrini & Kalkhoff, 1971;Green et al. 1981;Nadal et al. 2009). In humans, these are secreted by the ovary initially and then largely by placenta during pregnancy. However, in rodents the corpus luteum continues to contribute to the circulating pool of steroid hormones during pregnancy (Napso et al. 2018). Direct effects of oestrogens on β-cells include protection against oxidative stress and apoptosis and regulation of GSIS and islet lipid homeostasis (Soo et al. 2005;Alonso-Magdalena et al. 2008;Wong et al. 2010;Tiano et al. 2011;Zhou et al. 2018). In the case of progesterone, early studies in rats have shown that progesterone increased β-cell proliferation and did not affect GSIS during gestation in vivo (Nieuwenhuizen et al. 1998). However, progesterone receptor-null female mice display increased β-cell proliferation, mass and insulin secretion (Picard et al. 2002). Moreover, progesterone and, to a lesser extent, oestrogen, reduced β-cell, proliferation, survival and insulin secretion when co-treated with prolactin in vitro Fujinaka et al. 2007). These data suggest that a main function of the rise in progesterone concentration during gestation is to oppose/control the stimulatory effects of lactogens on β-cells . Progesterone also exerts pro-apoptotic effects in a rat β-cell line (Nunes et al. 2014) and mass in the peri-partum period. However, overall, further work is required to understand the role and relationship between oestrogens and progesterone in mediating β-cell adaptations throughout pregnancy.

Other endocrine signals.
Kisspeptin. Kisspeptins are a family of peptides that are the endogenous ligands for the Gq-protein-coupled receptor, GPR54. GPR54 is highly expressed in the pancreatic β-cells (Hauge-Evans et al. 2006). Studies in rodent, porcine and human islets have reported that exogenous kisspeptin enhances GSIS in vitro (Bowe et al. 2012(Bowe et al. , 2013 and in rats and mice in vivo (Bowe et al. 2009). In humans, kisspeptin is released from the placenta into the maternal circulation increasing circulating levels several thousand-fold (Dhillo et al. 2006). Placental release of kisspeptin into the maternal circulation is closely correlated with the increased insulin secretory capacity of maternal β-cells (Horikoshi et al. 2003;Reynolds et al. 2009). Furthermore, low circulating levels of kisspeptin were observed in women with GDM (Bowe et al. 2019). In experimental animals, maternal circulating kisspeptin is unknown, although mRNA levels in the rodent placenta increase during gestation (Herreboudt et al. 2015). In non-pregnant mice, kisspeptin administration increases circulating insulin levels (Bowe et al. 2019) whereas pharmacological blockade or genetic ablation of β-cell GPR54 in mice during pregnancy results in impaired GSIS, reduced β-cell proliferation and glucose intolerance (Bowe et al. 2019). Thus, placental kisspeptin acting via β-cell GPR54 is important for normal glucose homeostasis during pregnancy. However, further work is required to know whether placenta-derived kisspeptin acting via GPR54 on β-cells is the relevant signal that amplifies the insulin secretory response to compensate for the insulin resistance attained during pregnancy.
Urocortin. Urocortin peptides UCN1, UCN2 and UCN3 are part of the corticotrophin releasing hormone (CRH) peptide family usually associated with hypothalamic neuroendocrine functions. However, recently, CRH peptides were described as novel placental signals involved in the control of β-cell function (Drynda et al. 2018;Simpson et al. 2020). Of note, placental Crh expression and islet Crhr1/2 are upregulated on day 12 of mouse pregnancy, when β-cell mass expansion is maximal (Drynda et al. 2018), and previous studies have demonstrated that CRH can modulate glucagon release (Moltz & Fawcett, 1985) and β-cell insulin synthesis and proliferation in vitro (Schmid et al. 2011). In mice, all members of the CRH family are expressed by the placenta, but only circulating levels of UCN2 are significantly increased during gestation (Simpson et al. 2020). Moreover, activation of CRHR2, which is responsive to UCN2, leads to amplified GSIS without concomitant alterations in β-cell mass or overt changes in glucose tolerance in pregnant mice (Simpson et al. 2020). Thus, urocortin/CRH peptides derived from the placenta may act via their cognate receptors on maternal β-cells to perfect β-cell adaptive responses during gestation.
Leptin and adiponectin. Hormones synthesized by maternal organs also act in an endocrine fashion to modulate β-cell adaptations during gestation. Leptin is a hormone produced by the white adipose tissue and is involved in regulating insulin production and insulin sensitivity. In humans and mice, concentrations of leptin rapidly rise throughout gestation, peaking towards term (Napso et al. 2018). In humans, leptin is released by the placental syncytiotrophoblast into the maternal circulation (Masuzaki et al. 1996). But, this is not the case for mice, as the murine placenta does not express leptin (Malik et al. 2005). Despite this, both the mouse and the human placenta can secrete a soluble form of the leptin receptor (Yamaguchi, 1998, Tsai, 2015, which may modulate the systemic influences of leptin on maternal β-cells. In vitro and in vivo, leptin inhibits insulin secretion from β-cells (Kulkarni, 1997). Leptin treatment also decreases fasting insulin, as well as GSIS in pregnant wild-type mice (Yamashita et al. 2001), whereas studies in mice have shown that a heterozygous deficiency of the leptin receptor (db/+) is linked to the development of glucose intolerance specifically in gestation, despite elevated rates of GSIS (Yamashita et al. 2001). Whether the development of glucose intolerance during gestation relates to changes in insulin sensitivity versus insulin production in mice with altered leptin abundance or signalling capacity needs further study. Non-pregnant mice with a β-cell-specific deletion of the leptin receptor (using the Pdx1-Cre line) display improved glucose tolerance and enhanced insulin secretion, β-cell hypertrophy, proliferation and mass (Morioka et al. 2007). These changes are associated with enhanced activation of the insulin signalling pathway in β-cells, with increased AKT-forkhead box protein O (FOXO) activity and reduced inhibitor of insulin signalling, SOCS3 (Morioka et al. 2007). However, pregnancy outcomes in response J Physiol 600.5 to β-cell leptin receptor loss have yet to be investigated. Moreover, studies using Pdx1-Cre mice as a control, instead of the leptin receptor floxed mice, will help to fully discern the role of leptin in β-cell functionality and whole body glucose handling (Brouwers et al. 2014).
Adiponectin is an adipocyte-secreted hormone that enhances insulin sensitivity and has direct effects on β-cell function. Adiponectin receptors are highly expressed on human and mouse β-cells. In non-pregnant mice, transgenic overexpression of adiponectin protects against pancreatic β-cell loss and high fat diet-induced insulin resistance (Holland et al. 2011). In pregnant mice, genetic ablation of the adiponectin gene leads to glucose intolerance due to insulin deficiency and failed β-cell mass expansion, but there is no impact on GSIS (Qiao et al. 2017). However, β-cell-specific knockdown of the adiponectin receptor genes AdipoR1 and AdipoR2 does not significantly impact β-cell mass or insulin secretion (Qiao et al. 2021). Instead, adiponectin is proposed to promote the production of placental PL/PRLs, which in turn mediate β-cell proliferation and islet expansion during mouse pregnancy (Qiao et al. 2021). The adiponectin gene and protein are reported to be expressed by the human and rat placenta (Caminos et al. 2005;Chen et al. 2006). These data suggest a crosstalk between maternal fat, the placenta and β-cells involving hormones like leptin and adiponectin, which may have relevance for the pathogenesis of GDM.

Exosomes. Exosomes are extracellular vesicles (EVs)
with key roles in cell to cell communication and conveying molecular signals to cells at distant locations. All cell types release exosomes, including adipose tissue, liver, pancreas, skeletal muscle and placenta during pregnancy (Jayabalan et al. 2017). Studies have identified that placenta-derived exosomes contain a wide variety of molecules, including 58 miRNAs encoded by a chromosome 19 cluster (Donker et al. 2012) and dipeptidyl peptidase IV, a biologically active molecule with the potential to regulate maternal insulin secretion (Kandzija et al. 2019). In addition, infusion of small EVs (sEVs) isolated from healthy pregnant women promoted GSIS, elevated insulin concentrations and skeletal muscle insulin resistance in non-pregnant mice (James-Allan et al. 2020). In contrast, sEVs from pregnant women with GDM failed to induce GSIS, did not increase insulin concentrations and resulted in exacerbated skeletal muscle insulin resistance and glucose intolerance in non-pregnant mice (James-Allan et al. 2020). However, more studies are required to understand the origin and role of circulating sEVs in regulating β-cells adaptations to pregnancy.
Local factors. In this section, involvement in the production of local factors in β-cell adaptation during pregnancy will be summarized, although, there may also be a role for nervous inputs into the pancreas in governing β-cell adaptations (Li et al. 2019). Further study is needed to explore whether this may have significance in the context of maternal β-cell function and glucose control during pregnancy.
Hepatocyte growth factor. Hepatocyte growth factor (HGF) levels are markedly increased during pregnancy. Both the placenta and amnion produce and secrete HGF, with the former likely contributing to maternal circulating concentrations (Horibe et al. 1995). However, during pregnancy HGF is also highly expressed within islet endothelial cells and HGF is thought to play an important local role in driving β-cell adaptations in the mother during gestation. In particular, islet endothelial cell proliferation and HGF expression are increased from early gestation, peaking on days 10-15 of pregnancy, prior to the enlargement of β-cell mass in rats in vivo (Johansson et al. 2006). Moreover, increased expression of both HGF and its receptor, c-Met (also called tyrosine-protein kinase Met or hepatocyte growth factor receptor), within islets is correlated with the peak of β-cells proliferation (Demirci et al. 2012). In vitro, vascular endothelial growth factor-A (VEGF-A) derived from pancreatic islet endothelial cells stimulates β-cell proliferation through secretion of HGF (Johansson et al. 2006). Taken together, the available data suggest a paracrine crosstalk between the islet vasculature and β-cells, which may be relevant during gestation and warrant examination.
Mice lacking c-Met in the pancreas (using Pdx1-Cre) have reduced β-cell proliferation, decreased GSIS and increased apoptosis during gestation, which manifests as reduced β-cell mass, glucose intolerance, reduced plasma insulin and hyperglycaemia (Demirci et al. 2012). Defects in β-cell adaptations in these null mice were related to a failure in PRLR signalling and reduced STAT5 nuclear localization. These data suggest that PRLR signalling in the β-cell requires intact c-Met activity. In addition, pancreas deficient c-Met mice fail to decrease levels of the cell cycle inhibitor p27, as well as upregulate GLUT2 in their islets during pregnancy (Demirci et al. 2012). However, further work is required to verify these studies as wild-type mice instead of Pdx1-Cre were used as controls when studying the effect of pancreatic c-Met deficiency on pregnancy physiology (Demirci et al. 2012;Brouwers et al. 2014). In addition, HGF from gestational tissues, like the placenta, may also be important for β-cell adaptations and requires study.
Serotonin. Serotonin is an indoleamine molecule derived from the amino acid tryptophan which contributes to β-cell adaptations to pregnancy (Kim et al. 2010;Pasek & Gannon, 2013). Serotonin principally acts locally, in either an autocrine or a paracrine fashion (Berger et al. 2009;Amireault et al. 2013). Evidence from mouse models has shown that serotonin expression in islets increases during pregnancy and drives β-cell mass expansion (Kim et al. 2010). Serotonin signalling also plays an important role in enhancing GSIS in the mother during pregnancy (Ohara-Imaizumi et al. 2013), thereby maintaining glucose homeostasis and sensitivity. Moreover, interfering with serotonin signalling by dietary tryptophan restriction or inhibiting TPH1, the rate-limiting enzyme responsible for the synthesis of serotonin, leads to impaired β-cell proliferation and insulin secretion during pregnancy (Kim et al. 2010;Schraenen et al. 2010b). Consistent with this, serotonin treatment of mouse islets in vitro induces β-cell proliferation (Kim et al. 2010).
The gene encoding TPH1 is one of the most highly upregulated genes in islets during gestation, and both PRL and PL signalling through PRLR in the β-cells activate Tph1 expression (Rieck et al. 2009;Kim et al. 2010;Schraenen et al. 2010b;Pepin et al. 2019). In particular, upon binding to PRLR, STAT5 is phosphorylated by JAK2 and migrates to the nucleus (Horvath, 2000) where it recognizes an interferon γ -activated motif in the Tph1 gene promoter to induce expression in β-cells (Iida et al. 2015). However, induction of Tph1 expression by PRL and PLs is thought to be more complex, as this process also requires the activation of extracellular signal-regulated kinase (ERK) and phosphoinositide 3-kinase (PI3K) signalling in β-cells (Iida et al. 2015). β-Cells contain all the machinery needed for serotonin synthesis, storage and secretion (Ohta et al. 2011). Thus, β-cells are thought to synthesize and co-secrete serotonin with insulin during pregnancy (Kim et al. 2010;Schraenen et al. 2010b;Goyvaerts et al. 2016).
Secreted serotonin acts in an autocrine fashion by binding to serotonin receptors on β-cells and regulates GSIS during pregnancy. Of note, 5-hydroxytryptamine (serotonin) receptor 3A (HTR3A) is an ionotropic receptor that functions as a serotonin-gated cationic ion channel on β-cells (Ohara-Imaizumi et al. 2013). When bound by serotonin, HTR3A allows the leak of extracellular Na + ions into the β-cell, mildly depolarizing the membrane and lowering the threshold for GSIS (Ohara-Imaizumi et al. 2013). In support of this, blocking HTR3A signalling reduces β-cell insulin secretion and impairs glucose tolerance in pregnant mice (Ohara-Imaizumi et al. 2013). Mice lacking another serotonin receptor, HTR2B, as well as mice treated with HTR2B antagonists also show defective β-cell proliferation, mass expansion and impaired glucose tolerance during pregnancy (Kim et al. 2010). There are also data which suggest that switching serotonin receptor subtype from HTR2B to HTR1D just prior to term may be an inhibitory signal that promotes postpartum regression of β-cell mass in mice (Kim et al. 2010). Taken collectively, these studies highlight the important role of serotonin signalling in the local regulation of β-cells and suggest that interfering with serotonin signalling through pharmacological agents and diets depleted in tryptophan may disrupt maternal β-cell expansion and increase the risk of developing GDM.
Vascular endothelial growth factor and placental growth factor. VEGF-A is a master regulator of endothelial-β-cell crosstalk during β-cell development and regeneration. As previously mentioned, VEGF-A originating from pancreatic endothelium stimulates HGF production with positive impacts on β-cell proliferation in vitro (Johansson et al. 2006). However, conditional reduction of VEGF-A signalling in β-cells reduces islet vessel maintenance and induces transient glucose intolerance without altering β-cell mass expansion in pregnant mice in vivo (Staels et al. 2017). Placental growth factor (PlGF) is a member of the VEGF family that is highly secreted by the placenta during gestation. PlGF is also expressed by β-cells, and local islet production is increased during pregnancy ). Studies in mice have shown that knock-down of PlGF in β-cells results in compromised β-cell proliferation, reduced β-cell mass expansion and impaired glucose tolerance during pregnancy . Moreover, in part these effects in the β-cell PlGF knockdown mice are mediated by reduced recruitment of macrophages to pancreatic islets during pregnancy, as macrophages highly express the receptor VEGFR1, which binds PlGF . In a mouse model showing gestational hypertension and significantly reduced serum PlGF levels, β-cell proliferation and mass expansion were impaired during pregnancy . Further work in this model showed that defects in β-cell adaptations were rescued by exogenous PlGF treatment to the mother during pregnancy . Collectively, these data highlight the importance of growth factors in the endocrine, paracrine and autocrine regulation of β-cell adaptations during pregnancy. Moreover, defects in such regulation during pregnancy could play a major role in the development of GDM. J Physiol 600.5 and recruit JAK2, which then phosphorylates STAT5 and STAT5 activation leads to the expression of key genes involved in β-cell growth and proliferation. However, these effects are amplified as lactogenic signalling leads to the auto-upregulation of PRLR expression via STAT5 (Galsgaard et al. 1999). This increase in PRLR is needed to maintain the prolonged activation of STAT5 by PLs in β-cells (Brelje et al. 2004). In addition to PLs, GH and PRL both acutely stimulate STAT5A and 5B isoform activity, but only PRL is responsible for STAT5B activation in rat islets and INS-1 cells. Whether this isoform preference is also observed in vivo and has relevance during gestation has not been established (Brelje et al. 2004). Interestingly, mice with β-cell-specific STAT5A/B deficiency (using RIP-Cre) do not show defects in islet development or function when they are young, but develop glucose intolerance as they get older and demonstrate slightly more glucose intolerance than normal during pregnancy (Lee et al. 2007). STAT5 regulates the expression of several genes that may be significant for β-cell adaptations during gestation. For instance, in mice during pregnancy, STAT5 activation in β-cells promotes the expression of cyclin-cyclin-dependent kinase (CDK) genes and upregulates phosphorylation of retinoblastome protein (Rb), which are both important for promoting cell cycle progression ). The STAT5 pathway may also be responsive to other local and endocrine factors important for β-cell adaptations. In particular, HGF is believed to execute its actions on β-cells through the STAT5 pathway, as pancreatic loss of c-Met signalling and failed maternal β-cell adaptation during gestation are accompanied by reduced nuclear localization of STAT5 (Demirci et al. 2012). Thus, through STAT5, PLs, PRLs, GH, HGF and possibly other factors command relevant cellular pathways involved in maternal β-cell adaptation during gestation. However, due to the intricacy and interplay between cellular pathways, failures in STAT5 activation may be compensated by other signalling transduction pathways in the β-cells. Moreover, additional work is required to further delineate the contribution of STAT5/B in β-cell adaptations and glucose handling as floxed mice rather than RIP-Cre mice were used as controls when studying the effect of β-cell-specific Stat5/B deficiency (Lee et al. 2006;Lee et al. 2007).
PI3K-AKT signalling. The PI3K-AKT cascade is also activated by PRLR signalling and plays a dominant role in the regulation of β-cell function in both non-pregnant (Jiang et al. 2018) and pregnant states (Banerjee et al. 2016). PI3K-AKT regulates β-cell proliferation through modulation of multiple proteins, including forkhead box protein O1 (FOXO1), glycogen synthase kinase-3 (GSK3) and mechanistic target of rapamycin (mTOR) (Jiang et al. 2018). Several studies in rodents have shown that pregnancy and lactogenic signalling activate IRS1, IRS2, PI3K, AKT, p70S6K and mTOR in β-cells (Amaral et al. 2004;Zahr et al. 2008;Hughes & Huang, 2011). Pregnant rats treated with antisense oligonucleotides to reduce PRLR have reduced AKT phosphorylation and p70S6K protein levels in pancreatic islets (Amaral et al. 2004). Treatment of mice with rapamycin, an inhibitor of mTOR signalling, impairs β-cell proliferation and expansion during pregnancy, but does not alter blood glucose levels or glucose tolerance (Zahr et al. 2008). In addition, PI3K activation seems to play a central role in in the HGF-induced mitogenic effect on β-cells (Gahr et al. 2002;Vasavada et al. 2007). Furthermore, activation of PI3K-AKT signalling is involved in the protective effect of HGF against pharmacologically induced cell death in pancreatic β-cells (García-Ocaña et al. 2003). However, the exact role of the PI3K-AKT signalling cascade in β-cell adaptations in response to local and endocrine factors has not been specifically addressed in the context of pregnancy. Interestingly, c-Met receptor loss in β-cells (using Pdx1-Cre) from pregnant mice reduced GSIS and expression of islet Glut2, Pdx1 and insulin mRNAs, but this was not observed in the pregnant mice with β-cell PRLR deficiency (mediated by RIP-Cre) (Demirci et al. 2012;Pepin et al. 2019). These data suggest that HGF and PLs/PRLs may mediate β-cell adaptations during pregnancy via divergent mechanisms downstream of PI3K-AKT. They may also highlight the difference between PRL/PL and GH in signalling via PRLR, as the bGH minigene is present in both the Cre transgenic lines used when studying c-Met and PRLR loss in β-cells, but appropriate Cre controls are needed to provide any firm conclusions (Lee et al. 2006;Demirci et al. 2012;Brouwers et al. 2014;Pepin et al. 2019).
Raf-MEK-ERK signalling. The Raf-mitogen-activated protein kinase kinase (MEK)-ERK signalling pathway is also activated downstream of PRLR in rodent β-cells. Activation of this pathway modulates cell growth, proliferation and differentiation via interaction with menin, a tumour suppressor protein (Chamberlain et al. 2014). In addition to PRLR, epidermal growth factor receptor (EGFR) also acts through the MEK-ERK signalling cascade. Expression of a dominant-negative EGFR in pregnant mice blocks placental lactogen-induced β-cell expansion and expression of survivin (Birc5), a key cell-cycle protein and anti-apoptotic factor (Hakonen et al. 2014). However, serotonin synthesizing enzymes and other key genes involved in β-cell adaptation were still upregulated in these mutant pregnant mice treated with placental lactogen (Hakonen et al. 2014). ERK5 activation is increased in islets coincidentally with β-cell expansion during mouse pregnancy (Chen et al. 2018). Pharmacological suppression of ERK5 activation during pregnancy reduced β-cell proliferation and cyclin D levels, increased blood glucose levels and impaired GSIS (Chen et al. 2018). Finally, ERK1/2 activation is required for platelet derived growth factor receptor (PDGFR) to promote β-cell expansion during development (Chen et al. 2011). Interestingly, enhancer of zeste homologue 2 (EZH2) expression was suppressed in pregnant mice with selective deletion of PRLR in β-cells (Pepin et al. 2019). Together, these data support the notion of a close interaction between PRLR and ERK signalling and the involvement of ERK signalling in mediating β-cell adaptations during pregnancy, more broadly.
Serotonin signalling. In pregnant mice, serotonin synthesis is increased in at least 50% of β-cells within pancreatic islets (Goyvaerts et al. 2015). Transcriptomic analyses of islets in response to pregnancy have identified very strong upregulation of two paralogous genes encoding TPH enzymes, Tph1 and Tph2, which catalyse the formation of 5-hydroxytryptophan (Rieck et al. 2009;Kim et al. 2010, Pepin et al. 2019). The induction of these genes in β-cells is strongly dependent on the PRLR and JAK2/STAT5 signalling pathway during pregnancy (Schraenen et al. 2010b;Iida et al. 2015). Serotonin acts in an autocrine/paracrine fashion on β-cells expressing serotonin receptors to drive functional adaptations during pregnancy. For instance, serotonin can act locally on other β-cells through HTR2B and HTR3 receptors or may regulate blood flow to the islet, since serotonin has potent vasoconstricting effects. In addition, β-cell serotonin enters the blood circulation and mediates systemic effects (Goyvaerts et al. 2016). Mice which are deficient in Htr2b display impaired glucose tolerance and decreased β-cell proliferation rates in pregnancy (Kim et al. 2010). However, there are discrepancies about the relevance of HTR2B in mediating pregnancy-related changes in β-cells, as other studies have found only minor or weak expression of Htr2b in islets (Layden et al. 2010;Schraenen et al. 2010b). Interestingly, mice deficient in Htr3a do not show any alterations in the pregnancy-induced expansion of β-cell mass, but have insufficient GSIS and display glucose intolerance in gestation (Ohara-Imaizumi et al. 2013). Data suggest that serotonin acts through HTR3 to depolarize β-cells and lower the threshold for GSIS during pregnancy (Ohara-Imaizumi et al. 2013). Thus, serotonin is released from β-cells in response to lactogenic signalling and acts in an autocrine fashion to increase proliferation and insulin secretion via HTR2B an HTR3, respectively.
CISH and SOCS signalling. Cytokine-inducible SH2-containing protein (CISH) is a member of the suppressor of cytokine signalling (SOCS) family that negatively regulates PRLR signalling by blocking STAT5 activation and is strongly induced during gestation (Rieck et al. 2009;Kim et al. 2010;Layden et al. 2010;Pepin et al. 2019). Of note, the Cish and Socs2 (a SOCS family member) genes are highly expressed by pancreatic islets during pregnancy and are thought to play a critical role in limiting or controlling both the rate of proliferation and survival of β-cells (Rieck et al. 2009). However, ablation of Cish did not augment β-cell STAT5 activation, proliferation or expansion in mice during pregnancy (Jiao et al. 2013). In part, the lack of an effect of Cish loss on β-cells likely stemmed from compensatory upregulation of Socs2 in islets of mutant mice during pregnancy (Jiao et al. 2013).
Glucokinase. In β-cells, glucokinase is a recognized glucose sensor important for maintaining whole body and cellular glucose homeostasis (Matschinsky & Wilson, 2019). In response to pregnancy and lactogenic signalling, levels and activity of glucokinase increase in β-cells, resulting in a lower threshold for GSIS (Weinhaus et al. 1996). This is mediated via PRLRs on β-cells and through activation of the JAK2/STAT5 signalling pathway, which increases transcription of genes encoding glucokinase and insulin (Weinhaus et al. 2007). GLUT2 on β-cells is also induced via this process, which results in enhanced glucose utilization and oxidation that is important for enhanced glucose-stimulated insulin synthesis during pregnancy (Weinhaus et al. 2007).
Neogenic signalling. As previously mentioned, the formation of β-cells from non-β-cell progenitors (neogenesis) is one of the processes thought to be involved in β-cell expansion during murine pregnancy (Bonner-Weir et al. 2004;Abouna et al. 2010;Toselli et al. 2014). In part, this is thought to be related to the transcription factor, neurogenin-3 (NGN3), which is expressed by progenitor cells and becomes downregulated, secondary to the upregulation of its inhibitors, Pdx1, Sox9 and Hes1, in islets during pregnancy (Toselli et al. 2014). This process if thought to facilitate the differentiation of progenitor cells into hormone producing β-cells that can enter the cell cycle and proliferate in the mother. Indeed, the expression of many NGN3 target genes, including Tle3, NeuroD, Nkx2.2, IA1 and Rfx6, are also modified in maternal islets during pregnancy (Toselli et al. 2014).

Apoptosis signalling. A reduction in PRLR expression
does not affect β-cell apoptosis in pregnant mice, likely due to the low apoptosis rates naturally observed (Huang et al. 2009). However, pregnant mice lacking c-Met and rat insulinoma cells treated with PRLR siRNAs show increased β-cell apoptosis rates (Demirci et al. 2012;Arumugam et al. 2014). Moreover, transcriptomic analysis indicates that at least 60 genes involved in apoptosis processes, including Ngfr, Tnfrsf11b (Osteoprotegerin), Acvr1c, Nupr1 and Survivin, may be important in β-cell adaptations during pregnancy (Rieck et al. 2009). Interestingly, Survivin acts as an inhibitor of both the J Physiol 600.5 intrinsic and extrinsic apoptosis pathways by blocking the activity of several caspase proteins (Tamm et al. 1998) and Survivin expression is increased 5-fold in islets during pregnancy (Rieck et al. 2009), in part via EGFR signalling (Hakonen et al. 2014). Survivin is also induced by PRL signalling downstream of AKT, STAT5 and ERK and participates in PRL-mediated β-cell proliferation during pregnancy . Recently, overexpression of the transcription factor, paired box 8 (PAX8) was shown to reduce apoptosis rate in both murine and human pancreatic islets (Martin-Montalvo et al. 2019). In addition in mice, expression of PAX4 stimulated the proliferation of a β-cell subpopulation that expands during pregnancy and provides protection against endoplasmic reticulum stress-induced apoptosis (Lorenzo et al. 2015). Moreover, Lrrc55, a gene highly induced in islets during pregnancy, protects β-cells from glucolipotoxicity-induced apoptosis, through upregulation of pro-survival signals and attenuated calcium depletion (Makkar et al. 2019). Studies in rats have shown the presence of condensed chromatin, apoptotic bodies and programmed cell death in β-cells after delivery, which were related to TRPM2 and TGFβ1 expression (Scaglia et al. 1995). Thus, regulation of apoptotic pathways may be important in various aspects of β-cell adaptations in the mother both during and after pregnancy. FOXD3. In mice, pancreas-specific deletion of the transcription factor forkhead box D3 (FOXD3) (mediated by Pdx1-Cre) resulted in impaired glucose tolerance and decreased β-cell mass, proliferation and size in pregnancy (Plank et al. 2011). Defects in β-cell adaptations were related to the mis-expression of several genes known to regulate proliferation, including Foxm1, Skp2, Ezh2, Akt2 and Cdkn1a (Plank et al. 2011). Mice carrying floxed Foxd3 alleles were used as controls instead of Pxd1-Cre mice (Plank et al. 2011). Hence, further studies are needed to be sure of the involvement of FOXD3 in mediating decreased β-cell mass, proliferation and size in pregnancy (Brouwers et al. 2014) Indeed, this is relevant as Foxd3 gene expression is normally downregulated in islets at mid-late pregnancy (Plank et al. 2011) and while PL reduces islet Foxd3 expression in vitro, β-cell-specific PRLR deficiency does not impact Foxd3 expression in mid-late gestation in vivo (Banerjee et al. 2016). Thus, FOXD3 may be required to promote β-cell proliferation before PLs and PRLs rise in the maternal circulation and PRLR signalling appears to be required for Foxd3 downregulation during pregnancy.
MafB. MafB is expressed in immature β-cells during prenatal development and is then re-expressed in approximately 25% of β-cells in the mother during mouse pregnancy (Pechhold et al. 2009). This transient gestational expression of MafB requires lactogenic signalling and deletion of MafB in β-cells reduces both β-cell proliferation and mass during pregnancy, although the RIP-Cre mice were not used as controls for the in vivo observations so further study in this context is needed (Banerjee et al. 2016). MafB is the principal factor activating insulin expression during β-cell development (Artner et al. 2007). Furthermore, MafB is required for inducing MafA transcription in precursor β-cells and is a primary regulator of Pdx1, Nkx6.1 and Glut2 expression in mature β-cells (Artner et al. 2007). Thus, additional studies are warranted to explore whether MafB operates downstream of PLs/PRLs to help drive maternal β-cell adaptations in pregnancy.

FOXM1.
The cell-cycle-associated transcription factor FOXM1 is normally upregulated in maternal islets during late mouse gestation and decreases after parturition . Foxm1 deletion in mice results in a failure to increase β-cell proliferation during pregnancy and this defect is coupled to an increase in islet Menin and nuclear p27 staining . Islets treated with PL show increased Foxm1 expression in vitro, suggesting that FOXM1 may be a downstream mediator of PRLR signalling in β-cells Pepin et al. 2019). Insulin signalling in β-cells is required for FOXM1 activity, as signalling via (a) ERK facilitates FOXM1 localization to the nucleus, and (b) PI3K enhances FOXM1 mediated upregulation of CenpA and Plk1 expression, which promotes cell proliferation and viability (Shirakawa et al. 2017). In mice, CenpA deficiency in β-cells leads to impaired proliferation in response to pregnancy (Shirakawa et al. 2017). Specific inhibitors of STAT5, AKT and ERK pathways significantly decreased the protein levels of FOXM1 and Survivin in PRL-stimulated INS-1 cells . Thus, FOXM1 and Survivin are common downstream effectors of these pathways mediating β-cells adaptations during pregnancy with PRL .
HNF-4α. The orphan nuclear receptor hepatocyte nuclear factor-4α (HNF-4α) is required in β-cells for their proliferative response during pregnancy. Mice with β-cell HNF-4α deficiency fail to expand their β-cell mass during pregnancy (Gupta et al. 2007). Gene expression analysis of islets from these pregnant mice has shown that this failure is linked to the altered expression of numerous genes and pathways involved in metabolism, signal transduction and β-cell proliferation (Gupta et al. 2007). In addition, activation of Ras-ERK signalling during pregnancy was significantly reduced in HNF-4α-deficient islets, apparently in part by the downregulation of suppression of tumorigenicity 5 (ST5; Gupta et al. 2007). However, conditionally overexpressing ST5 in β-cells is not sufficient to enhance their proliferation during pregnancy (Ou et al. 2018). In part, the lack of an effect in ST5 transgenic animals may be due to increased islet activation of Rassf1a, a factor that inhibits cell cycle progression (Ou et al. 2018). Thus, there are complex interactions that determine the effect of HNF-4α in β-cell mass enhancement during gestation.
Menin. Menin, the protein product of MEN1 gene, is mutated in multiple endocrine neoplasia type 1, a disease characterized by tumours in endocrine glands, including pancreatic islets (Thakker, 2014). In mice, islet levels of Menin decrease in pregnancy and then increase to pre-gestational levels 1 week after birth. In addition, the transgenic induction of Menin expression in β-cells prevents islet expansion, induces hyperglycaemia and impairs glucose tolerance during mouse pregnancy (Karnik et al. 2007). In mice, PRL infusion reduces islet Menin expression in association with a decrease in the expression of the CDK inhibitors p27 and p18, which prevent β-cell proliferation. Attenuation of Menin expression in islets by PRL during mouse pregnancy involves an intact PRLR, AKT signalling and STAT5 mediated expression of Bcl6, a transcriptional repressor of the Menin1 gene (Karnik et al. 2007;Hughes & Huang, 2011). Thus, the downregulation of Menin via lactogenic signalling is important for the β-cells to functionally adapt during pregnancy.
Aryl-hydrocarbon receptor nuclear translocator. Aryl-hydrocarbon receptor nuclear translocator (ARNT) is a transcription factor that is decreased in islets from individuals with type 2 diabetes and is upregulated two-fold in maternal islets during pregnancy in mice (Lau et al. 2013). In mice, β-cell-specific deletion of ARNT (mediated by RIP-Cre) impairs glucose tolerance, insulin secretion and β-cell proliferation during pregnancy in alliance with decreases in the expression of Irs2, G6pi and cyclin D2 in islets (Lau et al. 2013). Although RIP-Cre mice were not used as controls for this study, the investigators reported that there are no deviations in glucose tolerance when compared to either wild-type or Arnt floxed mice, but there was no mention of studying this with respect to pregnancy (Lau et al. 2013). In the non-pregnant state, mutant mice also exhibit defects in GSIS and glucose intolerance and this is related to reduced HNF-4α and insulin and AKT signalling components in pancreatic islets (Gunton et al. 2005). However, further work is required to fully understand the significance, and upstream regulators and signalling pathways governing ARNT action in β-cells in the mother during pregnancy. Epigenetic mechanisms. There are data that suggest that epigenetic processes, namely non-coding RNAs, chromatin remodelling/histone modifications and DNA methylation, are involved in mediating β-cell adaptations during pregnancy. MicroRNAs (miRNAs) have important roles in modulating β-cell gene expression with impacts on β-cell differentiation, proliferation, function and survival (Filios & Shalev, 2015). Transcriptomic analysis of islets from rats identified that β-cell expansion during pregnancy was associated with changes in the expression of five miRNAs (Jacovetti et al. 2012). Of note, miR-218, miR-338-3p and miR-874 were downregulated, whereas miR-144 and miR-451 were upregulated and both sets of changes were strongly correlated with the peak of β-cell proliferation around gestational day 15 (Jacovetti et al. 2012). Reducing miR-338-3p in either rat or human islets in vitro did not alter insulin secretion but lead to enhanced Survivin, Foxm1, Ccnd, Igf1r, Irs2, Bcl2 and Bcl2l1 expression, which are important for β-cell proliferation and survival during pregnancy in vivo (Jacovetti et al. 2012). Oestrogens potentially regulate the expression of miRNAs in β-cells during gestation, as exposure of both rat and human β-cells to oestradiol potently reduced the expression of miR-338-3p in vitro (Jacovetti et al. 2012). The expression of miR-338-3p and miR-451 (but not the other miRs altered by pregnancy) is also coupled to the improved resistance of β-cells to pro-apoptotic stimuli and compensatory β-cell mass expansion in response to developmental and environmental cues, including nutritional challenges in rodents (Jacovetti et al. 2012). In women with poor glycaemic control or GDM during pregnancy, circulating miR-330-3p levels are elevated and correlated to low insulin concentrations and pregnancy outcome (Sebastiani et al. 2017;Pfeiffer et al. 2020). Moreover, in humans, several studies have identified miRNAs that are associated with abnormal weight gain and the development of GDM during pregnancy (Vasu et al. 2019). Thus, many miRNAs may participate in β-cell adaptations with important implications for glucose control in the mother during pregnancy. Further work is still required to explore the upstream and downstream mediators of miRNA actions in β-cells.
Molecular profiling of islets from pregnant mice with β-cell PRLR deletion has revealed aberrant expression of several epigenetic modifiers that regulate chromatin remodelling (Ezh2 and Suz12) and DNA methylation (Gadd45g and Dnmt3b) (Pepin et al. 2019). Moreover, other work has shown that PRL downregulates pancreatic islet expression of Menin, which functions as part of a histone methyltransferase complex to promote tri-methylation of histone 3 lysine 4 and expression of cell cycle inhibitors p27 and p18 (Karnik et al. 2007). These findings suggest that downstream of PRLR, many genetic and epigenetic regulators coordinate β-cell adaptations during pregnancy (Pepin et al. 2019). Platelet derived J Physiol 600.5 growth factor receptor (PDGFR) also upregulates the chromatin remodelling protein EZH2, a protein required for β-cell expansion during development (Chen et al. 2011) that was suppressed in pregnant mice with β-cell PRLR loss (Pepin et al. 2019). This suggests there may be overlap in the epigenetic pathways contributing to β-cell adaptations during gestation. Finally, insulin expression is regulated by DNA methylation and increased methylation of CpG sites in the insulin promoter is associated with elevated occupancy of methyl CpG binding protein 2 (MeCP2) and reduced binding of the transcription factors ATF2 and CREB in islets (Kuroda et al. 2009). However, work is required to examine the contribution of DNA methylation in the regulation of insulin production and in β-cell adaptations during pregnancy.

Maternal conditions that affect glucose metabolism and β-cell adaptation during pregnancy
Conditions that impact maternal physiology have the capacity to disrupt β-cell adaptation and thus the development of GDM during pregnancy. In this section, the impact of different environmental conditions that alter maternal body weight, adiposity and insulin-glucose homeostasis will be described in the context of β-cell function during pregnancy (Table 1). In addition, the effects of maternal conditions on the placenta will be mentioned because of the role of placenta in transferring nutrients to the fetus and in the endocrine regulation of maternal insulin sensitivity and β-cell adaptations.

Maternal diet or nutritional status.
Obesity and incidence of GDM in pregnant women. Abnormal nutrient status, indicated by maternal obesity, over-nutrition and altered gestational weight gain, as well as certain genetic predispositions, can lead to failed β-cell adaptation, hyperglycaemia and GDM during pregnancy. However, the mechanisms underlying failed or aberrant β-cell adaptation particularly under conditions of abnormal maternal nutritional status are not clear (Moyce & Dolinsky, 2018). For instance, there are some studies that suggest that while obese women have increased insulin resistance, they may have either an increased or a decreased insulin response compared to normal weight women during pregnancy (Catalano, 2010;Fan et al. 2020). Examinations of islets from women during pregnancy also vary widely with respect to maternal body mass index (BMI) and gestational age and do not include samples from mothers with GDM (Van Assche et al. 1978;Butler et al. 2010). However, studies relating dietary intake (reported through food frequency questionnaires) to the development of glucose abnormalities during pregnancy, like GDM have provided some insights. For instance, increased fat and  low carbohydrate intakes during the second trimester of pregnancy were associated with an increased risk of impaired glucose tolerance and GDM in women (Saldana et al. 2004). In another prospective study, pre-pregnancy BMI and elevated intake of n−3 fatty acids in the first trimester were associated with the development of GDM (Radesky et al. 2008). However, studies in pregnant rodents subjected to dietary manipulations have provided further information of the changes in maternal glucose homeostasis and the contributions of regulatory pathways governing β-cell adaptations during pregnancy. These findings are summarized below.
High sugar diet. Rats fed a high fructose and/or sucrose diet (HSD) from the start of pregnancy show normal glucose tolerance but increased plasma insulin concentration, likely as a result of compensatory β-cell mass expansion (Jen et al. 1991;Alzamendi et al. 2012). A maternal HSD during pregnancy was also associated with disrupted placental development (Jen et al. 1991;Alzamendi et al. 2012) suggesting that the placental production of endocrine signals important for adapting β-cell function may be altered and contribute to the maternal hyperinsulinaemia observed. While the impacts of maternal HSD intake from prior to pregnancy on β-cell adaptations have not been studied, β-cells can become overwhelmed by prolonged exposure to high glucose leading to glucotoxicity, oxidative stress, apoptosis and insulin insufficiency (Bensellam et al. 2012). Thus, a maternal HSD, both acute or chronic, may lead to β-cell dysfunction and increased susceptibility of the mother to develop GDM.
High fat diet. The impact of a high fat diet (HFD) on β-cell adaptations and glucose homeostasis during pregnancy depends on the duration of feeding, the percentage of dietary fat and the gestational day studied. In rodents, exposure to a HFD from prior to and/or during pregnancy increases maternal body weight/adiposity Liang et al. 2010;Carter & Tenlep, 2015) and impairs glucose tolerance during gestation (Carter & Tenlep, 2015;Wu et al. 2015;Huang et al. 2017Huang et al. , 2019Chung et al. 2019). In addition, concentrations of glucose, insulin and leptin are increased in HFD fed dams during pregnancy (Holness et al. 2007;Liang et al. 2010;Li et al. 2013;Aye et al. 2015;Huang et al. 2017). Maternal lipid profile is also altered in HFD-fed dams of most studies Heerwagen et al. 2013;Carter & Tenlep, 2015;Chung et al. 2019) together with an increase in the levels of inflammatory markers in the blood (Li et al. 2013). Mice fed a HFD for >3 months prior to pregnancy show greater β-cell mass but higher rates of apoptosis and pro-inflammatory markers in their β-cells (Li et al. 2013). Mice fed a HFD just during pregnancy also show greater β-cell mass . In rats fed a HFD during gestation, GSIS is enhanced in vivo, but insulin secretion response to glucose in islet perifusions is impaired (Holness et al. 2007). These latter findings are thought to reflect a partial compensation of β-cells for peripheral insulin resistance that is limited by an intrinsic defect in insulin secretion in the pregnant mother in response to the HFD. In support of this, female rats fed a HFD from around the time of mating show decreased expression of β-cell functional genes, namely Pax6, Pdx1, MafA and Glut2, in their pancreas . In part, interference of maternal β-cells with HFD feeding was rescued by a pharmacological induction of peroxisome proliferator-activated receptor α activation (Holness et al. 2007). Transcriptomic analysis of islets from HFD fed mice compared to β-cell-specific Prlr knockout mice suggests that dietary and lactogenic signalling operate via different mechanisms to modulate maternal β-cell functions during pregnancy (Pepin et al. 2019). Indeed indirect influences from other endocrine signals disrupted by exposure to a HFD could also impact β-cell adaptations during gestation (Moyce & Dolinsky, 2018). However, there could also be some common mechanisms involved, as miR-338-3p has been described as a mediator of compensatory β-cell mass expansion with both pregnancy and HFD-induced obesity in rodents (Jacovetti et al. 2012).
Studies in rodents have also demonstrated that the placenta is susceptible to HFD, whereby a HFD is associated with elevated oxidative stress and endothelial cellular damage (Liang et al. 2010). Placental size and lipid content are also increased (Heerwagen et al. 2013), insulin and mTORC1 signalling is greater  and transport of glucose and amino acids is elevated in vivo in mice fed a HFD Aye et al. 2015). As a consequence of the abnormal maternal metabolic profile and increased placental nutrient transport in HFD fed dams, the fetus is exposed to hyperglycaemia and displays increased fetal weight Aye et al. 2015). However, whether abnormalities in placental hormone production contribute to the altered maternal metabolic profile with a HFD during gestation or not is unknown. These studies show that consumption of a HFD by the mother impacts β-cell adaptations, glucose homeostasis and the placenta with consequences for maternal metabolic health and offspring development.
High fat and sugar, obesogenic and cafeteria diet. Intake of a diet high in both sugar and fat (HSHF) during pregnancy also impacts maternal β-cell adaptations, but the specific effect depends on the diet composition and dietary protocol used. Nevertheless, pregnant mice fed a HSHF diet show increased adiposity in association with an altered lipid profile, increased levels of leptin and defects in tissue insulin sensitivity (Holemans et al. 2004;Sferruzzi-Perri et al. 2013;Fernandez-Twinn et al. 2017;Musial et al. 2017Musial et al. , 2019J Physiol 600.5 Martins Terra et al. 2019). In addition, a HSHF diet during gestation induces hyperinsulinaemia, hyperglycaemia and glucose intolerance in the mother (Sferruzzi-Perri et al. 2013;Fernandez-Twinn et al. 2017;Musial et al. 2017;Martins Terra et al. 2019). Rats fed a HSHF diet also present with increased maternal adiposity and impaired glucose tolerance in gestation (Holemans et al. 2004;Vanzela et al. 2010). Moreover, consumption of a HSHF diet for 1 week prior to pregnancy is sufficient to cause glucose intolerance, impair β-cell proliferation and reduce fasting insulin in mid-pregnancy (Pennington et al. 2017). Thus, a HSHF diet, even if consumed acutely or just prior to pregnancy, is able to alter glucose homeostasis and β-cell adaptations, increasing the risk of the mother to develop GDM. Obesity is associated with increased production of pro-inflammatory cytokines like interleukin (IL)-1β, IL-6 and TNF-α from the placenta and/or maternal adipose during pregnancy (Pantham et al. 2015;Moyce & Dolinsky, 2018). Recent work has shown that such pro-inflammatory cytokines induce β-cell dedifferentiation and diminish insulin secretion capacity in vitro (Nordmann et al. 2017). Thus, further work is warranted to explore how a pregnancy-induced inflammatory profile which is aggregated by obesity may modify β-cell adaptation and the development of GDM.
In most cases studied, placental function was disturbed by a HSHF diet and there were alterations in placental glucose metabolism, nutrient transport and/or growth (Sferruzzi-Perri et al. 2013;Rosario et al. 2015;Musial et al. 2017). Of note, placentas from HSHF diet fed dams also show reduced expression of PRL-and PL-related genes, which may have consequences for β-cell proliferation and secretion of insulin in response to the hyperglycaemia induced by the dietary manipulation (Huang et al. 2009;Banerjee et al. 2016;Musial et al. 2017;Sferruzzi-Perri et al. 2020). Further work is required to examine the interaction of changes in maternal β-cell adaptations, glucose handling and the placenta in pregnancies when the mother is consuming a HSHF diet.
Low protein diet. Studies have evaluated the importance of dietary protein intake during pregnancy on maternal metabolism. This has involved using diets that contain anywhere between 6% casein and 23% casein as control. In rodents, intake of low protein diets containing 6-9% casein during pregnancy does not change the weight of the pregnant dam (Rutland et al. 2007;Salvatierra et al. 2015). However, a more minor reduction of dietary protein content (18% casein) in pregnant mice is associated with an increase in maternal hysterectomized body weight in late gestation, although whether this reflects an expansion of select tissues, lean and/or fat mass is unknown (Coan et al. 2011). In addition, a 6% low protein diet fed to rats during pregnancy does not impact maternal glucose tolerance but is linked to reduced pancreatic islet abundance of proteins important for intercellular communication and coupled release of insulin by β-cells, including connexin 36 and β-catenin (Milanski et al. 2005;Marçal-Pessoa et al. 2015). The same diet during pregnancy also reduced maternal islet area in association with reduced PI3K signalling, but no change in insulin content was observed . Interestingly, pregnant female mice exposed to chronic protein restriction show reduced body weight, glucose intolerance, impaired β-cell GSIS and attenuation of the phospholipase C-protein kinase C signalling pathway that is involved in the insulin secretory process (Ignácio-Souza et al. 2013;Szlapinski et al. 2019).
In part, alterations in β-cell adaptations and glucose handling during pregnancy with maternal dietary protein restriction are thought to be mediated via changes in local serotonin production (Baeyens et al. 2016). This is because tryptophan is an essential amino acid found in dietary proteins that is required for the synthesis of serotonin (Kim et al. 2010;Nongonierma & Fitzgerald, 2015). Indeed, a reduction in dietary tryptophan intake is linked to reduced β-cell mass expansion and impaired glucose tolerance in mice during pregnancy (Kim et al. 2010). Changes in maternal β-cell adaptations and glucose homeostasis could also stem from alterations in the placenta in response to a low protein diet. Decreased protein intake during pregnancy alters the development and nutrient transport capacity of the placenta (Rutland et al. 2007;Coan et al. 2011). Maternal dietary protein deprivation is linked to altered formation of placental endocrine cells and expression of the PRL-related genes in the mouse placenta (Gonzalez et al. 2016). Future work should examine the impact of poor maternal dietary protein intake on the relationships between changes in maternal β-cells, local islet signalling and the placenta.
Although there are some differences in β-cell adaptations between rodent and human pregnancies, the available data using rodent models have been instrumental in showing the importance of maternal diet and nutritional status, including body adiposity, in the capacity of the mother to undergo β-cell adaptations and appropriately control blood glucose concentrations during pregnancy.
Hypoxia. Hypoxia generated by living at high altitude (>2500 m) is a frequent cause of maternal physiological maladaptation during pregnancy (Moore et al. 2004), but prevalence of GDM is not increased in high altitude populations (Euser et al. 2018). Compared to women at sea level, pregnant women at high altitude present with decreased fasting plasma glucose concentrations and higher insulin sensitivity, although calculated β-cell function is similar and fasting insulin concentration is lower (Krampl et al. 2001a,b). Studies in rodents have evaluated the effect of different levels of inhalation hypoxia on maternal metabolic support of fetal growth. Pregnant mice maintained under 15% hypoxia from the start of pregnancy show reduced maternal weight gain and adipocyte area during gestation (Määttä et al. 2018). They also show greater glucose tolerance, fail to acquire insulin resistance and have elevated serum glucagon levels in late pregnancy (Määttä et al. 2018). Pregnant mice maintained under 10% hypoxia, but not 13% hypoxia, in the last third of gestation also have reduced maternal body weight although they have normal circulating insulin and glucose concentrations (Higgins et al. 2016). Exposure to 13% hypoxia from mid gestation, however, reduced maternal glucose concentration without a change in circulating insulin abundance (Higgins et al. 2016). Together the available data indicate that, under hypoxic conditions, maternal glucose homeostasis adapts differentially depending on the timing, duration and severity of the oxygen reduction. Indeed, there are severity-and timing-dependent effects of hypoxia on the transport function as well as the size and metabolism of the endocrine region in the placenta in mice (Higgins et al. 2016;Sferruzzi-Perri et al. 2019). However, it is not known if exposure to hypoxia during gestation alters the production of hormones that impact maternal metabolic adaptations, namely β-cell adaptations during pregnancy.
Although the maternal pancreas is a probable target of hypoxia-induced changes in glucose handling, there are no studies exploring if, and how, hypoxia may directly impact β-cell adaptations during pregnancy. ARNT, which helps to mediate the cellular effects of hypoxia, is implicated in the regulation of β-cell function and is significantly decreased in humans with T2DM (Gunton et al. 2005). As previously mentioned, ARNT expression is increased in β-cells during pregnancy and may be required for the normal augmentation of insulin secretion and β-cell proliferation although further examination is needed (Lau et al. 2013). Therefore, pancreatic islets could be particularly sensitive to changes in oxygen levels/hypoxia during pregnancy.

Circadian disruption during pregnancy.
Simulated shift work. Circadian disruption, also known as chronodisruption, refers to a breakdown between the phasing of internal biological systems relative to external environmental changes. It can be induced by exposure to shift work, which leads to a disturbance or desynchronization of the organization of physiological functions that normally have a rhythm over a 24 h period (Erren & Reiter, 2009). Epidemiological studies suggest that shift work is associated with an increased risk of chronic diseases, including diabetes and obesity (Gan et al. 2015;Proper et al. 2016). For instance, 4 days of simulated night shift work is sufficient to reduce insulin sensitivity and glucose tolerance in healthy non-shift workers (Bescos et al. 2018). Research in humans has revealed associations between chronodisruption and problems with fertility and pregnancy, particularly miscarriage . However, no studies have assessed a direct relationship between shift work before and/or during pregnancy and changes in glucose homeostasis associated with β-cell dysfunction in the mother. Studies in rats have revealed that exposure to chronic phase shifts during pregnancy, simulating a shift work schedule, is associated with altered maternal daily rhythms of several key endocrine signals, including melatonin, insulin, glucose, leptin and corticosterone (Varcoe et al. 2013;Mendez et al. 2016). However, maternal weight is only transiently decreased in early gestation and there are no significant differences in glucose and insulin tolerance in these exposed pregnant rats (Varcoe et al. 2013;Mendez et al. 2016). Exposure to chronic phase shifts during pregnancy reduced weight of the rat placenta, but the impact on the production of placental hormones with systemic metabolic effects in the mother is not known (Mendez et al. 2016). Studies in pregnant sheep with circadian disruption due to a simulated shift work protocol show reduced pregnancy weight gain, disrupted daily profiles of melatonin, glucose and insulin, impaired glucose tolerance and increased GSIS (Gatford et al. 2019). Thus, there appear to be species-specific differences in the effect of chronodisruption on maternal β-cell adaptations and glucose homeostasis in pregnancy.
A common feature of shift workers is the exposure to artificial light at night leading to the suppression of plasma melatonin concentration. Melatonin has a fundamental role in regulating the timing of several physiological functions, including glucose homeostasis, insulin secretion and energy metabolism (Owino et al. 2019). In addition, melatonin stimulates glucagon release from pancreatic α-cells (Gottlieb et al. 2005;Donga et al. 2010). It is perhaps not surprising then that genetic variants of melatonin receptors have been associated with a number of metabolic disorders, such as T2DM, GDM and obesity (Bouatia-Naji et al. 2009;Lyssenko et al. 2009). For instance, the melatonin receptor 1B variant rs10830963 is associated with GDM in a European Cohort (Rosta et al. 2017). This same variant is also correlated with the development of GDM and the absolute level of insulin secretion and β-cell compensation in a cohort of Mexican Americans (Ren et al. 2014). These data highlight that circadian rhythm disruption is able to negatively impact maternal β-cell adaptations and glucose homeostasis during pregnancy likely through the melatonin signalling pathway.
Sleep duration and quality. Studies have shown that short sleep duration and frequent snoring are positively J Physiol 600.5 correlated with impaired glucose tolerance and high circulating glucose concentrations in women during pregnancy, particularly when the woman is obese (Facco et al. 2010;Qiu et al. 2010). Indeed, each hour of reduced sleep time is associated with a 4% increase in glycaemia during gestation and pregnant women with increased sleep-disordered breathing, frequent snoring and sleep duration of 7 h per night have increased risk of developing GDM (Reutrakul et al. 2011). Other studies have found a U-shaped association of sleep duration and GDM, with less or more than 7-9 h of sleep being associated with an increased risk of GDM (Rawal et al. 2017;. Endocrine disrupting chemicals. Endocrine disrupting chemicals (EDCs) are structurally similar to many hormones and are capable of mimicking and interfering with the synthesis, secretion, transport, activity and/or elimination of natural hormones. Because of this, exposure to EDCs may cause a wide range of adverse health effects, including T2DM and obesity. Moreover, exposure to EDCs during pregnancy may disrupt maternal β-cell adaptations, glucose homeostasis and increase the risk of GDM (Ehrlich et al. 2016). However, whether impacts of EDCs on maternal metabolic status are brought about via changes in placental hormone output warrants investigation (Strakovsky & Schantz, 2018;Filardi et al. 2020).
Bisphenol A. Bisphenol-A (BPA) is a synthetic oestrogen used as the base compound in the manufacture of polycarbonate plastic and as an additive in others. BPA can leach from polycarbonate containers and heat, acidic or basic conditions can accelerate the release of BPA, with resultant impacts on the levels of human exposure. Epidemiological studies have investigated the association between BPA and clinically diagnosed GDM risk, but no strong associations have been found (Robledo et al. 2013;Shapiro et al. 2015), although other studies have shown BPA exposure during the second trimester of pregnancy is positively associated with blood glucose concentration in sub-fertile women (Chiu et al. 2017). In addition, urinary concentrations of BPA are associated with higher glucose levels among overweight/obese women (Bellavia et al. 2018). Conversely, in a cohort of Chinese pregnant women, high urinary BPA concentrations were associated with reduced risk of GDM, an effect that was maintained even after adjustment for pre-pregnancy BMI and other covariates .
Research in rodents strongly supports BPA as a disruptor of maternal glucose homeostasis, and apparently this effect is mediated by the activation of oestrogen receptor β (Alonso-Magdalena et al. 2015a). During pregnancy, oestrogens are relevant signals involved in β-cell mass expansion with their function partly mediated through repression of miR-338-3p (Jacovetti et al. 2012). Interestingly, in vitro studies have shown that the expression of miR-338 is altered in BPA-exposed islets (Wei et al. 2017). Moreover, exposure of β-cells to BPA for 48 h in vitro increases insulin secretion capacity (Makaji et al. 2011), likely via its impacts on both oestrogen receptor-α and oestrogen receptor-β (Nadal, 2019). More prolonged BPA exposure, however, reduces β-cell viability in vitro (Weldingh et al. 2017). Pregnant mice treated with BPA during gestation develop a GDM-like metabolic profile with impaired glucose tolerance, hyperinsulinaemia and increased triglyceride and leptin concentrations . The metabolic effects of BPA exposure on the mother were found to persist until 4-6 months post-partum, with increased body weight, insulin, leptin and triglyceride concentrations, as well as glucose intolerance and decreased insulin sensitivity observed , 2015b. A final assessment of mice 7 months after exposure to BPA during pregnancy revealed that dams had decreased pancreatic β-cell mass and reduced insulin secretion both in vivo and in vitro (Alonso-Magdalena et al. 2015b). Interestingly, expression of cyclin D2 and cyclin-dependent kinase-4 were diminished and apoptosis and expression of the cell cycle inhibitors p16 and p53 were increased in the β-cells of exposed post-partum dams (Alonso-Magdalena et al. 2015a). Altogether, these data suggest that BPA exposure during pregnancy impacts maternal β-cell adaptations with important consequences for the development of GDM, as well as the subsequent progression to T2DM in later life.
Phthalates. Phthalates are EDCs widely used in the manufacture of plastic. Similar to BPA, phthalates have been found to exert specific oestrogenic activity with important biological consequences (Harris et al. 1997). However, phthalate metabolites are less potent than BPA in inducing reduced viability and increased insulin release from the pancreatic β-cells in vitro (Weldingh et al. 2017). In pregnant rats injected with streptozocin to induce GDM, di-n-butyl phthalate further exaggerated maternal hyperglycaemia and impaired glucose handling in vivo (Chen et al. 2020). In addition, di-n-butyl phthalate impaired STAT1 signalling and led to inhibition of FoxM1 and reduced viability of β-cells in vitro (Chen et al. 2020). In a Mexican GDM cohort of women, a significant association was found between specific phthalate metabolites in urine and the circulating abundance of miRNAs implicated in metabolic disease, namely miR-9-5p, miR-29a-3p and miR-330-3p (Martínez-Ibarra et al. 2019). In addition, there were specific associations between distinct urinary phthalate metabolite concentrations and maternal body mass during pregnancy (James-Todd et al. 2016). Of note, mono-ethyl phthalate was associated with impaired glucose tolerance and excessive gestational weight gain in the second trimester, whereas a high di-2-ethylhexyl phthalate was associated with a lower risk of impaired glucose tolerance (James-Todd et al. 2016). In addition, phthalate metabolites are also found to be linked to glucose intolerance during pregnancy, with possible stronger associations in certain racial/ethnic subgroups (Shaffer et al. 2019). Whether the impacts of phthalates on glucose handling during pregnancy relate to alterations in maternal β-cell adaptations requires further study.

Summary
Maternal β-cell adaptations are critical for appropriate handling of glucose during pregnancy. This process is governed by a variety of environmental, hormonal, cellular and molecular factors (Fig. 2). For instance, maternal β-cell adaptations are mediated via hormones secreted by the placenta and organs like the adipose tissue, which promote β-cell proliferation, growth, neogenesis, enhanced insulin secretion and protection from apoptosis. Such hormones operate through several signalling cascades, transcription factors and epigenetic regulators that modify receptor abundance, cell cycle-related genes and the threshold for glucose-stimulated insulin secretion. The environment of the mother is also a key factor in determining β-cell adaptations during pregnancy, as dietary variations, exposure to hypoxia, circadian disruption and endocrine disrupting chemicals are able to interfere with β-cell proliferation, survival, insulin secretion and/or associated molecular pathways. In part, environmentally induced changes in β-cells seem to link to alterations in placental endocrine function and maternal body weight, adipose mass and insulin sensitivity. However, the precise mechanisms through which β-cells integrate the various environmental, metabolic and endocrine cues and thereby determine appropriate β-cell adaptation during gestation requires further research. Combining β-cell-specific knock-out models, placental hormone-specific models, environmental manipulations and high throughput molecular analysis of isolated β-cells will provide powerful tools for understanding pathways and relevant factors controlling β-cell adaptations and the pathways leading to the development of GDM. These studies may also have broader implications in paving the way to understanding the mechanistic underpinnings of metabolic dysfunctions including T2DM.

Figure 2. Summary description
An abnormal diet, hypoxia, circadian disruption or endocrine disrupting chemicals directly modify the endocrine maternal environment impacting normal β-cell adaptation. In addition, endocrine and metabolic signals coming from the cross-talk between placenta and maternal tissues tightly regulate β-cell function promoting proliferation, growth/expansion, neogenesis, GSIS, and apoptosis. Disruption of the coordinated process described will alter insulin sensitivity, glucose homeostasis and handling in the pregnant mother.