Diabetes and apoptosis: neural crest cells and neural tube

Abstract

Birth defects resulting from diabetic pregnancy are associated with apoptosis of a critical mass of progenitor cells early during the formation of the affected organ(s). Insufficient expression of genes that regulate viability of the progenitor cells is responsible for the apoptosis. In particular, maternal diabetes inhibits expression of a gene, Pax3, that encodes a transcription factor which is expressed in neural crest and neuroepithelial cells. As a result of insufficient Pax3, cardiac neural crest and neuroepithelial cells undergo apoptosis by a process dependent on the p53 tumor suppressor protein. This, then provides a cellular explanation for the cardiac outflow tract and neural tube and defects induced by diabetic pregnancy.

Introduction

The offspring of women with diabetes mellitus are at significantly increased risk of congenital malformation [1–9]. While improvements in diabetes control in recent decades have reduced the incidence of this diabetic complication, referred to as “diabetic embryopathy”, malformations are still up to five times more common in diabetic pregnancies than in nondiabetic pregnancy [6, 8, 10–13]. The malformations are induced early during organogenesis, therefore, they only occur in pregnancies in which maternal diabetes, either type 1 (T1DM) or type 2 (T2DM), is present at the onset of pregnancy. Gestational diabetes, which is associated with insulin resistance brought on in the latter half of pregnancy, does not induce structural malformations because it occurs after the period of organogenesis [14]. While some studies have observed an increase in malformations in pregnancies diagnosed with gestational diabetes, it is thought that these women actually had T2DM before pregnancy that had not been diagnosed [1, 13]. Maternal obesity prior to pregnancy also increases the incidence of malformations [15–18]. However, it is not clear whether obese women whose offspring developed malformations had undiagnosed T2DM [19], or whether other metabolic disturbances might be responsible. For example, adequate folic acid availability has been recognized for several years to reduce the risk for neural tube, and other, defects [20–22]. There is evidence that obese women are less likely to take folic acid supplements or to consume a diet rich in nutrients such as folate [23, 24]. As well, a study of postmenopausal women revealed that serum folate levels are inversely related to body fat composition [25].

Almost any organ system is susceptible to malformation during diabetic pregnancy [5, 26–29]. However, neural tube defects (NTD), including spina bifida and anencephaly, and cardiovascular defects, especially cardiac outflow tract defects (COTD), transposition of the great arteries, and atrial or ventricular septal defects, are the most common to occur [5, 6, 26]. The malformations are induced at the earliest stages of organogenesis, prior to the seventh week of gestation in humans [30].

Several investigators have employed rodent models to study the effects of diabetic pregnancy, or in vitro culture in high glucose-containing media, on postimplantation embryonic development [31–37]. In general, these models result in malformation of craniofacial structures and the neural tubes and growth retardation. There are several biochemical disturbances that have been proposed to play a role in the defective development, such as increased protein kinase C activity [38–40], decreased arachidonic acid uptake [41, 42], defective prostaglandin synthesis [43, 44], PPAR signaling [45–48], random DNA mutations [49, 50], and oxidative stress [37, 51–55]. However, none of these studies revealed how these biochemical disturbances could disturb morphogenesis.

Orderly cell death regulates normal processes in metazoan embryogenesis such as dorsoventral patterning of the neural tube, bone formation, fusion of the oral palate and neural tube, and craniofacial development [56–65]. Inappropriate cell death, on the other hand, has the potential to lead to defective development. Indeed, apoptosis has been associated with teratogens such as ethanol, retinoic acid, hypoxia, ionizing radiation, and hyperthermia [66], and, as described in this review, maternal diabetes. A critical question that needs to be answered is, how does apoptosis in response to maternal diabetes (or any teratogen) occur in a targeted fashion? Otherwise, if all cells of the embryo underwent cell death, the entire embryo would die; alternatively, if only a portion of the cells of all structures of the embryo underwent cell death, the embryo would be small for gestational age, but structurally normal. In this review, the mechanisms by which apoptosis in neural crest and neural tube progenitors occurs and causes malformation of the cardiac outflow tracts and neural tube will be discussed.

Altered gene expression in embryos of diabetic mice leads to apoptosis and malformations

Altered expression of Pax3 in mouse embryos of diabetic mothers

The Pax gene family encodes transcription factors that are primarily expressed during development. The mammalian Pax genes were identified by the presence of a highly conserved 384 bp sequence encoding a DNA-binding paired box that was originally discovered in the Drosophila segmentation genes, paired (prd), gooseberry-proximal (gsb-p), and gooseberry-distal (gsb-d), and tissue-specific genes, Pox meso and Pox neuro [67–70]. There are nine known mammalian Pax genes. They are grouped into four classes based on their paired domain homology, the presence or absence of a complete or partial homeodomain, and the presence or absence of a conserved octapeptide [68, 71]. In contrast to the Hox genes, which are expressed in a region-specific anterior–posterior pattern, the Pax genes may be expressed along the entire anterior–posterior axis, or in a tissue-specific fashion. Paired box genes have been found in several animal phyla from nematodes to vertebrates [67, 70]. Pax3 is expressed in the vertebrate neural tube, neural crest, and somitic mesoderm; expression in mice begins in the dorsal region of the neural folds on day 8.5 [72]. Mutations in the DNA-binding domains of human PAX3 are associated with Waardenburg syndrome type I, which involves lateral displacement of the medial canthus of the eye, aniridia, white forelock, and congenital deafness [73–75]. The mouse Splotch alleles map to the Pax3 locus [73, 76–79]. Pax3^Sp/Sp embryos develop NTD (exencephaly, spina bifida, or both), lack skeletal musculature, and exhibit defects in neural crest-derived cell populations [80, 81]. Defective cardiac neural crest (CNC) development leads to COTD; embryos die of cardiac failure by day 14.5, probably due to impaired formation of conductive tissue of the heart, another process which is dependent on CNC [81–85]. Notably, NTD and COTD are among the most common defects induced by diabetic pregnancy in humans [6, 26]. Heterozygous Splotch (Pax3^+/Sp) embryos are viable and do not develop the structural defects that occur in Pax3^Sp/Sp embryos, although in pigmented strains, they develop white “splotches” of fur on the belly, due to deficiency of neural crest-derived melanocytes.

The morphological similarity between the NTD and COTD in embryos of diabetic mice, in which insulin-deficient diabetes had been induced by the pancreatic β cell toxin, streptozotocin (STZ), and Pax3^Sp/Sp embryos [86, 87], suggests that diabetic embryopathy phenocopies the effects of Pax3 mutation. This appears to be due to inhibition of Pax3 expression, as Pax3 expression is significantly reduced in embryos of diabetic mice as determined by in situ hybridization and reverse transcription-PCR (RT-PCR) of day 8.5 embryos [86]. Pax3 expression is also reduced in migrating neural crest and somites on days 9.5 and 10.5, as demonstrated by in situ hybridization [86], although, as described later, the decreased expression of Pax3 in neural crest is probably due to prior elimination of these cells by apoptosis [87]. These observations indicate that congenital malformations induced during diabetic pregnancy appear to be caused by misexpression of genes that control essential developmental processes. While there may be many genes whose expression could be affected by maternal diabetes, inasmuch as Pax3 deficiency is sufficient to cause malformation Pax3-expressing structures, simply inhibiting its expression below a critical threshold during diabetic pregnancy is sufficient to induce defects of Pax3-dependent structures.

Pax3 inhibits cell death

While Pax3 loss-of-function leads to deficient development of dependent structures, Pax3 gain-of-function is associated with immortalization or unregulated proliferation. For example, several of the Pax genes induce transformation when stably expressed in fibroblasts [88]. Many forms of pediatric alveolar rhabdomyosarcoma are associated with a rearrangement of the PAX3 gene, leading to expression of a fusion protein containing the DNA-binding domain of PAX3 and the trans-activation domain of Forkhead [89–92]. PAX3 is also expressed in many neural crest-derived tumors, such as neuroblastoma and melanoma, and is associated with an aggressive malignant phenotype and poor prognosis [93–98]. Notably, knocking down PAX3 (or PAX3-Forkhead) mRNA with antisense oligonucleotides induces cell death [98–100]. PAX3 also is crucial for proliferation of melanocytes in response to UV light [101]. These observations suggest that Pax3 may prevent cell cycle withdrawal, and/or cell death.

In embryos of diabetic mice and Pax3^Sp/Sp embryos, neuroepithelial cells of the neural folds undergo apoptosis [86]. This indicates that elimination of the cells on the leading edges of neural folds halted neural tube closure. More recently, the depletion of migrating CNC cells in embryos of diabetic mice and Pax3^LacZ/LacZ embryos (carrying LacZ reporters transcribed under the control of Pax3 control elements, but which disrupt Pax3 3 coding sequences) was also shown to be associated with increased apoptosis [87, 102]. These observations indicate that during normal embryogenesis, Pax3 prevents neuroepithelial and neural crest apoptosis. Sulik noted that cells that would normally be rapidly proliferating were more susceptible to apoptosis during the pathogenesis of malformations [66]. Rapid proliferation is a common feature that drives the movement of the neural folds toward each other and the migration of neural crest toward their destinations [103, 104]. Thus, if Pax3 is necessary for proliferation of neural crest and neuroepithelial cells, insufficient expression of Pax3 could lead to apoptosis. This, then, explains how selective apoptosis, leading to congenital malformations, can occur.

Additional evidence for apoptosis in animal models of diabetic embryopathy has been obtained in recent years. For example, increased numbers of apoptotic cells have been observed within the fused neural tubes of day 11.5–13.5 mouse embryos [105–107]. However, these observations of apoptosis were scattered, and were observed after neural tube fusion, suggesting that they may be involved in further maturation of the CNS, rather than NTD. Maternal diabetes also increases the incidence of apoptosis in the mandible, a neural crest derivative, of day 11 rat embryos [108], and in the endocardial cushion of day 13.5 mouse embryos [109]. Abnormal temporal-spatial pattern of apoptosis during remodeling of pharyngeal arch arteries has also been noted in day 12.0–14.5 rat embryos of diabetic mothers [110]. This is likely to reflect the disturbances in cardiac outflow tract septation that is dependent on CNC cells, rather than a direct effect of maternal diabetes on apoptosis of progenitor cells. Pervasive apoptosis has also been observed in day 10 rat embryos of diabetic mothers [111]. However, these embryos were also growth retarded and developmentally delayed. Thus, there may be conditions, in which the intrauterine diabetic exposure induces global cell death. This is likely to result in fetal demise, rather than specific structural malformation.

Mechanisms by which maternal diabetes disrupts Pax3 gene expression

The current understanding of how diabetic pregnancy disrupts Pax3 gene expression is discussed briefly here. A more detailed description has been recently reported [112].

Excess glucose transported to the embryo is responsible for the adverse effects of diabetic pregnancy on embryogenesis

Many metabolic disturbances that occur during T1DM or T2DM, including hyperglycemia, β-hydroxybutyrate, somatomedin inhibitors, and hyper- or hypoinsulinemia, have been considered to be responsible for diabetic embryopathy [32–34, 113, 114]. However, hyperglycemia is the only common feature of T1DM and T2DM. The offspring of women with either T1DM or T2DM are prone to the same kinds of defects, and the risk for spontaneous abortion or congenital malformations is correlated with poor glycemic control [9, 11, 13, 115–117]. In the rat, glucose concentration in embryo neuroepithelium reaches equilibrium with glucose concentration in maternal serum [118]. Indeed, we have shown, both in vivo and in vitro, that excess glucose delivered to the embryo is necessary and sufficient to inhibit Pax3 expression, and induce neuroepithelial apoptosis and NTD [119]. Transient hyperglycemia before the onset of Pax3 expression is also sufficient to induce CNC apoptosis and COTD [87]. The early embryo expresses the high K_m Glut2 glucose transporter [120]. Thus, Glut2 would rapidly transport glucose into embryo cells when glucose concentrations are above its K_m (~15 mM), as can occur during episodes of hyperglycemia. Expression of Glut2 is essential for embryo susceptibility to hyperglycemia-induced NTD [121]. Thus, excess glucose delivery to the embryo appears to be responsible for the adverse effects of pregnancies affected by either T1DM or T2DM on embryo development.

Oxidative stress mediates the effects of high glucose to inhibit Pax3 expression, thereby inducing apoptosis

Excess glucose metabolism increases the production of superoxide by the electron transport chain; if these free radicals are not scavenged, oxidative stress will result [122]. There is ample experimental support that oxidative stress contributes to malformations in animal models exposed to a high glucose milieu [37, 51–55]. In addition to increased superoxide production from increased oxidative metabolism, the pathways responsible for oxidative stress are complex; they include decreased expression of γ-glutamyl-cysteine synthetase, which would decrease production of the antioxidant, reduced glutathione (GSH) [123, 124], and increased activity of the hexosamine flux pathway and diacylglycerol/protein kinase C signaling [40, 125]. Both of these latter pathways would decrease production of NADPH, which is needed to produce GSH from oxidized glutathione (GSSG). In addition, the increased oxygen consumption caused by increased glucose metabolism creates a relative state of hypoxia in embryos, which, in turn, increases superoxide production [126].

Oxidative stress may lead to oxidation of proteins, lipids, and DNA and can induce apoptosis [127]. However, if oxidative stress resulting from excess glucose metabolism was directly responsible for cell death, all cells of the embryo would be equally susceptible. Rather than inducing limited defects in organogenesis, the entire embryo would be affected. However, oxidative stress inhibits expression of Pax3 in vivo and in vitro, and activation of pathways that contribute to oxidative stress, such as hexosamine flux and hypoxia, also inhibit Pax3 expression [125, 126, 128]. Importantly, oxidative stress which is sufficient to inhibit Pax3 expression is not sufficient to induce DNA strand breaks [128]. Thus, the oxidative stress induced by diabetic pregnancy appears to induce apoptosis in a cell-selective fashion, leading to specific malformations, by inhibiting expression of genes which regulate cell viability in a cell-selective fashion. The biochemical and molecular mechanisms by which oxidative stress inhibits expression of Pax3 still remain to be elucidated. However, the window of susceptibility in the mouse embryo is approximately 24 h before the onset of Pax3 expression [87, 102, 126, 128].

Could inhibition of Pax3 expression or Pax3-dependent processes be a common cause of NTD?

Ionizing radiation and certain anticonvulsant drugs increase risk for NTD and increase oxidative stress [129, 130]. Thus, it is possible that inhibition of Pax3 expression is a common mechanism leading to NTD in response to various sources of oxidative stress.

Folic acid might also affect Pax3 expression, or complement Pax3-dependent processes. Folic acid increases the conversion of homocysteine, which is an oxidant, to methionine, and folic acid is reported to have antioxidant properties independent of lowering homocysteine levels [131–133]. Thus, folic acid may prevent the generation of oxidative stress, as well as contribute to suppression of endogenous oxidative stress. Folic acid has been reported to reduce the incidence of malformations in rodent models of diabetic pregnancy [134, 135]. This could be explained by the effects of folic acid to reduce oxidative stress. However, it has also been reported that folic acid can suppress NTD in Pax3^Sp2H/Sp2H embryos [136]. This would indicate that a folic acid-dependent process(es) might be regulated by Pax3, or that Pax3 and folic acid both regulate a common developmental process. On the other hand, folic acid was not found to suppress NTD in Pax3^Sp/Sp embryos [137]. The differences between the compensatory effects of folic acid in Pax3^Sp2H/Sp2H and Pax3^Sp/Sp embryos could be explained by differential penetrance of the Splotch^2H and Splotch mutations on NTD. However, it should be noted that the Pax3^Sp/Sp embryos whose mothers had been treated with folic acid were examined on day 18.5 [137]. Since Pax3^Sp/Sp embryos are not normally viable past day 14.5 due to cardiac insufficiency [81], this suggests that folic acid might have compensated for Pax3 deficiency with regard to CNC development. It still needs to be resolved whether folic acid could be involved in pathways distal to Pax3 during neural tube closure.

With regard to diabetic pregnancy, one study found that folic acid deficiency or aberrant metabolism does not explain increased defects in human pregnancies affected by pre-gestational T1DM or T2DM [138]. However, further research is needed to investigate whether folic acid regulates Pax3 expression or Pax3 downstream effects, and whether folic acid requirements might be greater in diabetic pregnancies.

The p53 tumor suppressor protein mediates effects of Pax3 deficiency in neuroepithelium and neural crest

Inactivating p53 prevents apoptosis, NTD, and COTD in Pax3-deficient embryos

The observation that apoptosis was increased in neuro-epithelium of embryos of diabetic mice or Pax3^Sp/Sp embryos suggested either that Pax3 inhibits apoptosis, or else, that if a Pax3-dependent development fails, cells undergo apoptosis. Evidence in support of the former mechanism, and specifically, that Pax3 inhibits apoptosis that is regulated by the p53 tumor suppressor protein, was obtained by inactivating p53 by either genetic or chemical means [139]. By crossing mice with a germline p53 mutation [140] with Pax3^+/Sp mice to generate Pax3^Sp/Sp p53^−/− embryos, or treating pregnant Pax3^+/Sp mice with pifithrin-α, an inhibitor of p53 nuclear and mitochondrial translocation [141–143], it was demonstrated that apoptosis and NTD failed to occur in Pax3^Sp/Sp embryos in which p53 was inactivated [139]. This indicated that Pax3 is not required to direct expression of genes that control neural tube closure, but it is required to block p53-dependent apoptosis.

A role of Pax3 to inhibit p53-dependent apoptosis during CNC migration and cardiac outflow tract septation has recently been reported [102]. Pax3^Sp/Sp embryos exhibit defects in CNC migration and outflow tract septation [81]. CNC cells arise from neuroepithelium between the midotic placode and the third somite, make an epithelial to mesenchymal transformation, and migrate through pharyngeal arches 3, 4, and 6 to the primitive, single outflow tract of the heart [144, 145]. CNC are required for septation of the cardiac outflow tract into the aorta and pulmonary arteries [146–148]. Mouse embryos with mutant Splotch alleles develop COTD, but it has been disputed whether the onset of CNC migration was delayed, whether either the numbers of migrating cells or rate of migration was reduced, or whether migration was normal, but CNC inductive properties on the vascular was defective [83–85, 144, 149]. Using Pax3^LacZ/LacZ mouse embryos, it was recently shown that the onset of CNC migration is normal, but that during day 9.0–9.5, there are progressively fewer CNC that can be detected. Instead, there are progressively more apoptotic cells where CNC would normally be detected [102]. This indicated that Pax3 is not required for the initiation of CNC migration, but it is required for CNC viability while they are migrating toward the heart. Notably, both CNC apoptosis and COTD failed to occur in Pax3^LacZ/LacZ embryos that were p53-null, or whose mothers had been treated with a p53 inhibitor [102]. This indicated that, as in the neural tube, Pax3 is not needed to induce genes which regulate CNC migration or outflow tract septation, but it is needed to prevent p53-mediated CNC apoptosis during CNC migration. It should be noted that, although development of Pax3-expressing somitic derivatives was defective, this was not associated with increased apoptosis, and defective skeletal muscle development was not prevented by inactivating p53. In fact, Pax3^LacZ/LacZ fetuses that were p53-null, or whose mothers had been treated with a p53 inhibitor, survived and displayed normal heart development when examined on day 18.5, the day before normal term delivery, however, they are not viable, probably due failure of skeletal muscles required for respiration to develop [102]. This suggests that Pax3 regulates other processes instead of, or in addition to, inhibition of p53 in somites.

Time course of developmental effects of maternal diabetes to induce apoptosis, NTD, and COTD

Use of chemical agents to induce or inhibit oxidative stress, or inhibit p53-dependent apoptosis, at specific times of gestation, has elucidated the approximate stages of development which are susceptible to perturbations in glucose metabolism or Pax3 insufficiency:

• Day 7.5: the primitive streak forms. Transient hyperglycemia (or oxidative stress or hypoxia) inhibits processes necessary to induce expression of Pax3 [119, 126, 128].

• Day 8.5: Pax3 expression, neural tube closure, and neural crest migration begins [72]. Pax3 expression is significantly reduced in pregnancies in which hyperglycemia, or oxidative or hypoxic stress was induced on day 7.5, even though blood glucose or markers of oxidative stress have returned to normal [119, 126, 128]. Nuclear and/or mitochondrial translocation of p53 must be inhibited within approximately the first 4 h of Pax3 expression (approximately noon-4:00 p.m. on day 8.5) [102].

• Day 9.0–9.5: defective CNC migration and apoptosis are detectable in embryos whose Pax3 expression was inhibited or are expressing null Pax3 alleles; these defects do not occur if p53-dependent processes were inhibited on day 8.5 [87, 102].

• Day 10.5: defective neural tube closure and apoptosis are detectable in embryos whose Pax3 expression was inhibited or are expressing null Pax3 alleles; these defects do not occur if p53-dependent processes were inhibited on day 8.5 [119, 150].

• Days 9.5–12.5 CNC cells migrate to the heart and participate in septation of the single outflow tract into the aorta and pulmonary arteries. Outflow tract septation fails in embryos whose CNC had undergone apoptosis prior to reaching the heart, and fetuses die by day 14.5. Inhibiting p53-dependent processes on day 8.5 rescues cardiac outflow tract septation and fetal viability in Pax3-deficient fetuses [102].

These results indicate that all of the effects of maternal diabetes on apoptosis that lead to COTD and NTD can be attributed to effects of oxidative stress, which occur approximately 24 h before the onset of Pax3 expression, to inhibit induction of Pax3; that the initiation of neural tube folding and CNC migration occurs independent of Pax3; that Pax3 must inhibit p53 synthesis or stability within the first few hours of its expression, although elimination of Pax3-deficient cells may not be completed until 24–48 h later; and that preventing p53-dependent apoptosis of neuroepithelium and CNC allows the neural tube to close and cardiac outflow tracts to septate. A diagram describing the pathways by which these processes occur in control and diabetic pregnancies is shown in Fig. 1.

Fig. 1.

Flow chart of developmental events during control pregnancies, in which embryos are exposed to normal glucose levels (euglycemia), and diabetic pregnancies, in which embryos are exposed to elevated glucose levels (hyperglycemia)

Regulation of p53 by Pax3

The prevention of apoptosis and malformations in Pax3-deficient embryos by inactivating p53 suggests that p53 might be regulated by Pax3 at the level of transcription or translation. While there were no differences in p53 mRNA levels between Pax3^+/+, Pax3^+/Sp, and Pax3^Sp/Sp embryos, p53 protein levels were increased in Pax3^Sp/Sp embryos [150]. This suggests that Pax3 inhibits p53-dependent apoptosis, not by inhibiting p53 gene expression, but by inhibiting p53 translation or stability. It should be noted that Pax5, -2, and -8 (which fall into the same Pax paralogous group as each other, but a different paralogous group from Pax3) can inhibit p53 expression at the level of transcription [151]. This suggests that down regulation of p53 may occur in response to several of the Pax proteins, but at different levels of control.

Since Pax3 is a transcription factor, it might control expression of genes which regulate p53 translation [152, 153] or turnover [154]. While it has not been determined how Pax3 regulates p53 steady state levels in normal embryo cells, transfection of PAX3 and p53 expression vectors into p53-null human lung carcinoma cells has suggested that stability of p53 may be reduced in the presence of PAX3 [155].

Regulation of neuroepithelial/neural crest cell viability by Pax3 suppression of p53

Currently, it is not known how p53 regulates apoptosis of neural crest or neuroepithelial cells in Pax3-deficient embryos. However, it may be secondary to cell cycle withdrawal. For example, the Warburg effect (in which the anaerobic:aerobic metabolism ratio is high even when O₂ is not limiting) is characteristic of embryonic and cancer cells, and provides anabolic substrates for rapidly growing cells [156, 157]. p53 suppresses the Warburg effect, in part by transcriptional activation of genes that regulate respiration, and by inhibiting activity of glycolytic enzymes that prevent replicative senescence [158, 159]. Thus, Pax3, by inhibiting p53, may maintain CNC and neuroepithelium in a rapidly proliferating mode.

In hepatoma cells, stimulating oxidative pathways inhibits cell cycling, whereas stimulating glycolysis promotes cell cycling. However, folate, tetrahydrofolate, or purines can override the effects of oxidative pathways to inhibit cell cycling [160]. Thus, it is interesting to speculate that, in the early embryo, folic acid might function distal to Pax3 to promote purine synthesis and cell cycling. This might explain the effects of folic acid in embryos of diabetic rats and Pax3^Sp2H/Sp2H embryos to reduce NTD [134–136] despite the presence of w.t. p53.

Effects of maternal diabetes on preimplantation embryos, apoptosis, pregnancy loss, and malformations

The focus of this review has been the apoptosis of neural crest and neuroepithelium in the pathogenesis of diabetic embryopathy. However, there is also evidence that maternal diabetes can induce apoptosis of the preovulatory oocyte or preimplantation embryo in animal models. This can be a cause of diminished fertility, although, recently, there has been evidence that it can be a cause of embryonic malformation. This topic has been discussed in more detail recently [161, 162] but is addressed briefly here.

Using STZ-induced diabetic mice, it has been shown that germinal vesicle breakdown of preovulatory follicles is delayed and preimplantation embryo development is abnormal; these effects are prevented by insulin administration [163, 164]. There is also increased apoptosis of granulosa and cumulus cells [165]. Culture of preimplantation embryos from diabetic mice in media containing physiological glucose does not reverse the effects of diabetes [163, 166]. These results suggest that diabetes impairs oocyte and follicular development that leads to irreversible adverse effects on the preimplantation embryo. However, culture of embryos from nondiabetic mice in high glucose media from the 2-cell to blastocyst stage causes the same abnormalities as does maternal diabetes, indicating that the early preimplantation embryo is also vulnerable to the adverse effects of maternal diabetes [167, 168]. By the blastocyst stage, there are significantly reduced numbers of embryo cells, and this is associated with increased apoptosis [169–171]. The apoptosis is due to increased expression of the pro-apoptotic protein, Bax, which occurs in response to p53 [172, 173].

Paradoxically, exposure of preimplantation embryos to elevated glucose does not increase glucose uptake, as in the post-implantation embryo, but it decreases glucose uptake due to decreased expression of Glut-1, -2, and -3 [174]. A novel insulin-responsive glucose transporter, Glut-8, is expressed by blastocysts [175]. Thus, during diabetic pregnancy, insulin-stimulated glucose uptake via Glut-8 might be reduced. Taken together, the adverse effects of maternal diabetes on preimplantation embryo cells appear to be mediated by glucose deprivation, whereas effects on post-implantation embryo cells appear to be mediated by glucose excess. There are also reports that markers of oxidative stress are increased in blastocysts cultured in high glucose [171], although, it still remains to be determined how decreased glucose availability would generate oxidative stress.

These effects of maternal hyperglycemia on ovarian follicle and preimplantation embryo development could be partially responsible for the decreased fertility of diabetic women. However, until a recent report from Wyman et al. [176], there was no evidence that maternal diabetes prior to organogenesis could lead to structural malformation of the post-implantation embryo. The Moley group transferred zygotes or blastocysts from STZ-induced diabetic mice to pseudopregnant nondiabetic recipients. When the pregnancies were examined on day 14.5, there was a higher rate of malformations (affecting the cranium, abdominal wall, and limbs) in embryos derived from zygotes or blastocysts transferred from diabetic mothers than in embryos derived from transfer from nondiabetic mothers. One interpretation of this result is that exposure of the preovulatory oocyte to a diabetic environment may influence expression of genes involved in organogenesis well before these genes would be induced. This could be explained by epigenetic modifications, such as chromatin acetylation or methylation, leading to suboptimal expression of affected genes later in development, or, effects on oocyte mitochondria so that they are unable to meet all of the energetic demands of rapidly growing embryonic structures during organogenesis. However, another interpretation is that STZ, which a nitrosourea compound, causes alkylation of DNA and carbamoylation of proteins [177], rather than maternal diabetes, was responsible for defective oogenesis. Since the ability of insulin treatment to prevent the effects of STZ-induced diabetes was not examined, this possibility can not be excluded at this time. There were also increased reabsorptions and growth retardation of embryos derived from zygotes and blastocysts transferred from diabetic mothers. The reabsorptions could be due to structural malformation of an essential organ (such as the heart or placenta), or misexpression of (embryonic and/or extraembryonic) genes that are essential during post-implantation development, leading to embryo demise. The growth retardation could be due to overall decreased proliferation or increased apoptosis. The molecular mechanisms responsible for these various outcomes (malformation, embryo demise, growth retardation) could be the same, with only the severity of the effect on the oocyte or zygote explaining the different outcomes.

Using a different experimental design, embryos were obtained from nondiabetic mothers and then cultured in vitro from the two-cell to blastocyst stage in media containing elevated D-glucose, or elevated L-glucose as an osmotic control [176]. When these embryos were transferred to pseudopregnant nondiabetic recipients and then examined on day 14.5, no malformations were observed in either treatment group. There were, however, fewer implantations and more reabsorptions, as well as growth retardation, among the embryos that had been cultured in elevated D-glucose. This result suggests that apoptosis of a limited number of cells of the blastocyst due to elevated glucose leads to growth retardation, but elimination of a critical mass of cells of the blastocyst will lead to implantation failure or demise of the entire embryo. However, since the fates of cells of the inner cell mass are not yet determined, specific organ systems of individual embryos are not affected. The failure of high glucose exposure only during the two cell to blastocyst stage to cause malformations suggests that, unlike the embryo at this stage of development is not susceptible to the effects of diabetes to cause malformations.

Summary

Pax3 is a developmental control gene whose expression is inhibited by oxidative stress in response to excess glucose metabolism. The oxidative stress must occur approximately 24 h before Pax3 expression begins, suggesting that there is a sequence of developmental steps that must be set in motion to induce Pax3 expression. Since Pax3 expression is sensitive to cellular redox status, it may function as a developmental energy sensor that functions to preserve the “stem-ness” of neuroepithelial and neural crest cells. It may do so by suppressing the activity of p53 to promote withdrawal from the cell cycle and terminal differentiation. However, if Pax3 levels are insufficient, p53 steady state levels rise, ultimately leading to apoptosis of neuroepithelium or neural crest. With the loss of the progenitor cells of the neural tube and the CNC which participate in cardiac outflow tract remodeling, NTD and COTD result. It is likely that there are other genes, expressed in other organ systems, that are sensitive to cellular redox status. Thus, misexpression of genes which regulate progenitor cell viability at critical stages of organogenesis may be a general mechanism responsible for malformations in diabetic pregnancy.

Abbreviations

T1DM

Type 1 diabetes mellitus

T2DM

Type 2 diabetes mellitus

STZ

Streptozotocin

NTD

Neural tube defect(s)

COTD

Cardiac outflow tract defect(s)

CNC

Cardiac neural crest Glut Glucose transporter(s)

GSH

Reduced glutathione

GSSG

Oxidized glutathione