SEARCH

SEARCH BY CITATION

Keywords:

  • pregnancy;
  • Pax3;
  • oxidative stress

Abstract

  1. Top of page
  2. Abstract
  3. BACKGROUND AND HISTORICAL PERSPECTIVE
  4. MOLECULAR REGULATION OF NTD RESULTING FROM DIABETIC PREGNANCY
  5. SUMMARY
  6. REFERENCES

Maternal diabetes increases the risk for neural tube, and other, structural defects. The mother may have either type 1 or type 2 diabetes, but the diabetes must be existing at the earliest stages of pregnancy, during which organogenesis occurs. Abnormally high glucose levels in maternal blood, which leads to increased glucose transport to the embryo, is responsible for the teratogenic effects of maternal diabetes. Consequently, expression of genes that control essential developmental processes is disturbed. In this review, some of the biochemical pathways by which excess glucose metabolism disturbs neural tube formation are discussed. Research from the author's laboratory has shown that expression of Pax3, a gene required for neural tube closure, is significantly reduced by maternal diabetes, and this is associated with significantly increased neural tube defects (NTD). Pax3 encodes a transcription factor that has recently been shown to inhibit p53-dependent apoptosis. Evidence in support of this model, in which excess glucose metabolism inhibits expression of Pax3, thereby derepressing p53-dependent apoptosis of neuroepithelium and leading to NTD will be discussed. © 2005 Wiley-Liss, Inc.


BACKGROUND AND HISTORICAL PERSPECTIVE

  1. Top of page
  2. Abstract
  3. BACKGROUND AND HISTORICAL PERSPECTIVE
  4. MOLECULAR REGULATION OF NTD RESULTING FROM DIABETIC PREGNANCY
  5. SUMMARY
  6. REFERENCES

Diabetes mellitus is a disease that has affected humans, at least since the time of the ancient Greeks (who gave the disease the name, which roughly translates to, “copious production of honeyed-urine”). There are two major forms of diabetes, type 1 (formerly called, insulin-dependent diabetes or juvenile diabetes), in which the insulin-producing beta cells of the pancreas are destroyed by an autoimmune process, and type 2 (formerly called, non-insulin-dependent diabetes or adult-onset diabetes), in which insulin-dependent glucose uptake, primarily by fat and muscle, becomes impaired. In recent years, the terms, “insulin-dependent” and “non-insulin-dependent” have fallen out of use because even in type 2 diabetes, compensatory increased insulin production by pancreatic beta cells eventually causes insulin production to “exhaust,” necessitating insulin treatment. Similarly, the terms, “juvenile” and “adult-onset” diabetes are misleading, because both types of diabetes can occur in either the young or the old. Although type 1 diabetes has its onset primarily before the age of 20, it can occur in the 5th, or later, decade of life. More notably, type 2 diabetes, which, until recently, occurred primarily after the age of 40, has recently reached epidemic rates in children and young adults, owing to increased obesity and sedentary lifestyles [Mokdad et al., 2001; Kaufman, 2002].

The major long-term consequence of diabetes is that if post-prandial glucose fails to be transported and stored in fat and muscle, circulating glucose concentrations remain higher than normal, and chronic hyperglycemic exposure eventually leads to tissue pathology, including diabetic retinopathy, nephropathy, vasculopathy, and neuropathy. Until the availability of insulin in the 1920s, these sequelae were primarily relevant to type 2 diabetes, as individuals with type 1 diabetes, having almost complete loss of insulin-producing beta cells by the time the disease was diagnosed, usually died within a few years of onset of the disease. Once insulin became available, long-term survival of children and young adults with type 1 diabetes became possible, and pregnancy in women with pre-existing diabetes became increasingly more common. Maternal diabetes increased risk for miscarriage, toxemia, infant respiratory distress, and perinatal hypoglycemia, placing the lives of the infant, mother, or both at risk [White, 1937; Gabbe, 1993]. However, as management of these complications improved, congenital malformations remained the most significant adverse outcome of diabetic pregnancy [White, 1949]. The recognition that severity of pre-existing diabetic complications such as retinopathy and nephropathy were correlated with poor outcome of pregnancy, including congenital malformations, led to the development of the White's Classification Index of diabetic pregnancy [White, 1949]. As will be discussed further below, chronic poor control of diabetes as manifested in diabetic complications could be an indicator of the glycemic exposure of the embryo, and therefore its risk for maldevelopment; alternatively, there may be genetic modifiers which increase or decrease risk for long-term diabetic complications, which could also modify risk for diabetic teratogenesis.

There have been several clinical studies suggesting that the incidence and severity of diabetic pregnancy-induced malformations are correlated with poor glycemic control [Miller et al., 1981; Greene et al., 1989; Lucas et al., 1989; Langer and Conway, 2000; Schaefer-Graf et al., 2000].

There have been several clinical studies suggesting that the incidence and severity of diabetic pregnancy-induced malformations are correlated with poor glycemic control.

Most of these studies are based on increased non-enzymatic glycation of hemoglobin (either HbA1 or HbA1C) as an index of poor glycemic control. Non-enzymatic glycation is a slow, mass action-dependent process [Furth, 1997]. Because of the non-enzymatic nature of the reaction, the accumulation of glycated proteins is directly related to the substrate (glucose) availability, and the stability of the protein. Thus, glycation of hemoglobin, a relatively long-lived circulating protein, serves as a marker for glycemic exposure during the approximately 60 preceding days, although it does not indicate when excursions from euglycemia occurred, nor the relative peaks and nadirs of glycemia. However, diabetes can cause malformation of organ systems that are induced at different times within the first several weeks of gestation [Lucas et al., 1989; Becerra et al., 1990; Martinez-Frias, 1994], indicating that precise windows of susceptibility to specific organ defects may be difficult to identify using glycated hemoglobin.

One large, prospectively-designed, multi-center study found that malformations were increased 2.5-fold in the offspring of diabetic women whose pregnancies were identified within 21 days of conception, whereas they were increased 4.5-fold if pregnancies were identified later than 21 days following conception [Mills et al., 1988]. In the early entry group, there was no significant increase in glycated hemoglobin in diabetic pregnancies, leading to the conclusion that diabetes per se, and not glycemic control, increased the risk for malformation. However, this conclusion was met with some skepticism, as 93% of diabetic subjects had HbA1 values as much as 7 standard deviations above the non-diabetic mean, which was not statistically significant in this analysis. It should also be noted that glycated hemoglobin may not be an appropriate indicator of glycemic control and risk for diabetic pregnancy-induced defect, not just because it is an indicator of cumulative glycemic exposure, rather than acute exposure at critical stages of organogenesis, but also because of the increased blood volume and erythropoiesis during pregnancy progressively dilutes previously synthesized hemoglobin. Thus, during pregnancy, glycated hemoglobin may under-reflect prior glycemic exposure.

Perhaps the best evidence in support of the role of poor glycemic control and risk for malformation came following the Diabetes Complications Control Trial (DCCT), in which 1,441 subjects with type 1 diabetes enrolled in the study at 29 sites throughout the US and Canada between 1983 and 1993. Subjects were randomly assigned to conventional (i.e., without sampling of blood glucose and insulin injection at frequent intervals to attain euglycemia) or intensive (i.e., with four or more daily samplings of blood glucose and adjustment of insulin delivery to attain euglycemia) diabetes therapy. This study showed that intensive therapy lowered glycated hemoglobin levels and reduced the incidence and rate of progression of the traditional diabetic complications such as retinopathy, nephropathy, amputations, and cardiovascular disease [The DCCT Research Group, 1995a,b,c]. During the course of the study, 180 women either sought to become, or became, pregnant. Because of the suspected contribution of poor glycemic control to congenital malformations, women who were originally assigned to the conventional therapy group were reassigned to the intensive therapy group [The DCCT Research Group, 1996]. Glycated hemoglobin was significantly higher at conception in women previously assigned to the conventional control group compared to the intensive control group (8.1% vs. 7.4%), but did not differ during gestation (6.6% in both groups). Notably, nine congenital malformations were identified, eight of which occurred in the pregnancies previously assigned to conventional therapy. The conclusions from this study were that rigorous glycemic control during organogenesis can reduce the risk for congenital malformations, but that aggressive control should be initiated prior to conception.

Nevertheless, even in the 21st century, there continue to be reports that birth defects, particularly those affecting the neural tube and the heart, are significantly increased (up to 5-fold) in diabetic pregnancies [Langer and Conway, 2000; Suhonen et al., 2000; Aberg et al., 2001; Loffredo et al., 2001; Sheffield et al., 2002; Wren et al., 2003]. The interpretation from these studies is generally that availability of pre-pregnancy counseling on the importance of glycemic control, prevention of unplanned pregnancies by diabetic women, and patient compliance with intensive insulin therapy, should be improved. However, another consideration is that altered production of hormones that regulate glucose homeostasis upon pregnancy necessitates modification of insulin administration, diet, and exercise in order to maintain appropriate glycemic control. It should also be recognized that even intensive insulin therapy does not bring the rate of any diabetic complication to the non-diabetic rate, and so it may be necessary to develop additional interventions, or perhaps, more physiologic methods of controlling diabetes, before congenital malformations in diabetic pregnancy are no greater than in non-diabetic pregnancy.

It should also be recognized that even intensive insulin therapy does not bring the rate of any diabetic complication to the non-diabetic rate, and so it may be necessary to develop additional interventions, or perhaps, more physiologic methods of controlling diabetes, before congenital malformations in diabetic pregnancy are no greater than in non-diabetic pregnancy.

Excess glucose transported to the embryo is responsible for the adverse effects of maternal diabetes to cause malformations.

The previous discussion indicated that as standards of diabetic treatment have improved, the incidence and severity of congenital malformations have been reduced. This would argue that the cause of malformations in the embryos of diabetic mothers is not inherently linked to the disease (e.g., it is not genetic or autoimmune in origin) but is linked to the metabolic disruption of normal fuel metabolism associated with the disease. In considering the metabolic disturbances of diabetes, which could be responsible for congenital malformations, excess circulating glucose, insulin (or insulin insufficiency), or other metabolic disturbances which are secondary to defective glucose storage, such as ketosis, altered circulating levels of somatomedin-binding proteins, or serum lipids, have been proposed. Circulating levels of ketones, somatomedin-binding proteins, and triglycerides differ between type 1 and type 2 diabetics, and yet susceptibility to the same kinds of congenital malformations is the same in pregnancies affected by either type 1 or type 2 diabetes [Towner et al., 1995; Schaefer-Graf et al., 2000; Aberg et al., 2001], although elevated ketones may exacerbate the effects of elevated glucose [Buchanan et al., 1994]. Insulin effects on the embryo (or effects of insulin insufficiency) is not likely to be responsible because maternal insulin is not transported to the conceptus [Moore and Persaud, 1993], and the period of embryogenesis in which malformations occur precedes the onset of fetal insulin synthesis [Slack, 1995]. Circulating glucose, on the other hand, equilibrates with the embryo [Sussman and Matschinsky, 1988]. Despite other metabolic differences, hyperglycemia is a common characteristic of both type 1 or type 2 diabetes, supporting elevated glucose as primarily responsible for malformations. Finally, post-implantation culture of rat embryos in elevated glucose-containing media induces more numerous and more severe malformations than when cultured in low glucose-containing media [Eriksson et al., 1982; Eriksson, 1991]. However, the concentrations of glucose required to significantly increase malformations was far in excess of those, which occur during diabetes and increase risk for malformations, suggesting that glucose is not primarily responsible for malformations, or that the rat embryo culture is not a sensitive model for glucose teratogenicity.

It is important to establish whether glucose, or some other metabolic disturbances, are responsible for the adverse effects of diabetic pregnancy on embryogenesis in order to investigate the biochemical pathways involved.

It is important to establish whether glucose, or some other metabolic disturbances, are responsible for the adverse effects of diabetic pregnancy on embryogenesis in order to investigate the biochemical pathways involved.

As will be explained further below, we developed a mouse model of diabetic pregnancy and showed that increased neural tube defects (NTD) were correlated with reduced expression of Pax3, a gene required for neural tube closure [Phelan et al., 1997]. Using three approaches, we showed that excess glucose was responsible for decreased Pax3 expression and increased NTD [Fine et al., 1999]. First, embryo tissue fragments were cultured in media containing 5 mM glucose (∼90 mg/dl) or 15 mM glucose (∼270 mg/dl). High glucose inhibited Pax3 expression compared to culture in media containing “physiological” glucose; this was not due to an osmotic effect of high glucose, because excess l-glucose failed to inhibit Pax3 expression. Second, inducing hyperglycemia (≥250 mg/dl) in pregnant mice by glucose injection on the day prior to the onset of Pax3 expression significantly inhibited its expression and increased NTD. Third, administration of phlorizin, a drug which inhibits renal tubular reabsorption of glucose, reduced blood glucose levels of pregnant diabetic mice, and the rate of NTD. These results indicated that excess glucose delivered to the embryo, as a consequence of maternal hyperglycemia, is necessary and sufficient to cause malformations. Furthermore, the fact that high glucose inhibited Pax3 expression in embryo tissue fragments suggested that the effects of excess glucose were due to direct effects on the embryo, and were not due to, for example, a hostile uterine environment, or even effects on extraembryonic structures.

The recognition that excess glucose, even episodic and transient, can be teratogenic, may explain the increase in malformations, especially NTD, in pregnancies of obese women [Shaw et al., 1996; Werler et al., 1996; Moore et al., 2000; Mikhail et al., 2002; Watkins et al., 2003]. Obese individuals may be insulin resistant, and so, circulating glucose concentrations may fluctuate above those of non-diabetic, insulin sensitive individuals, or such individuals may have undiagnosed type 2 diabetes. With the epidemic of obesity among young adults and adolescents (1; 2), the risk for birth defects and the role of increased glucose exposure is an important area of future investigation.

MOLECULAR REGULATION OF NTD RESULTING FROM DIABETIC PREGNANCY

  1. Top of page
  2. Abstract
  3. BACKGROUND AND HISTORICAL PERSPECTIVE
  4. MOLECULAR REGULATION OF NTD RESULTING FROM DIABETIC PREGNANCY
  5. SUMMARY
  6. REFERENCES

Virtually any organ system can be affected by diabetic embryopathy, but defects of the neural tube and the heart, especially outflow tract defects, are among the most frequently observed in clinical studies [Becerra et al., 1990; Martinez-Frias, 1994]. The NTD are generally closure defects, especially spina bifida, although non-closure defects, such as holoprosencephaly or syntelencephaly are also observed [Barr et al., 1983; Robin et al., 1996]. In rat embryos from diabetic dams or cultured in high glucose-containing media, externally-detectable malformations, including exencephaly, spina bifida, micrognathia, torsional defects, and growth and developmental retardation, have been observed [Eriksson et al., 1982, 1991]. In mouse embryos of diabetic or glucose-injected dams, open NTD, especially exencephaly, are significantly increased [Phelan et al., 1997; Fine et al., 1999], although closed defects resembling holoprosencephaly may also occur [Phelan et al., 1997; Liao et al., 2004]. Unlike the rat, there is no growth or developmental delay observed in mouse embryos of diabetic dams [Phelan et al., 1997]. The morphological similarity and significant increase in NTD in both rodent and human embryos by diabetic pregnancy supports the validity of rodent models to study the fundamental mechanisms by which these defects occur.

Whereas most diabetes in humans is a naturally-occurring disease resulting from a combination of genetic susceptibility and environment, most rodent studies of diabetic pregnancy have not been performed using strains that naturally develop diabetes, but with strains in which insulin-deficient diabetes is induced with streptozotocin or alloxan, both of which have primary effects to destroy the β cells of the pancreas. There are rodent models of type 2 diabetes, the ob/ob and db/db mouse and the fa/fa rat, all of which have defects in production or signaling of the anorexigenic hormone, leptin [Beck, 2000]. However, since leptin is also required for reproductive function in females [Schubring et al., 2000], homozygous animals are sterile and cannot be used for the study of diabetic pregnancy. Perhaps the best studied rodent model of type 1 diabetes, the non-obese diabetic (NOD) mouse, develops diabetes with high frequency (approximately 80% occurrence among females in the Joslin colony), however, reproductive performance diminishes markedly once animals become manifestly diabetic. (Nevertheless, severe NTD, including exencephaly and craniorachischisis, was observed in the fetuses of a diabetic NOD mouse, which had successfully delivered normal sized, non-malformed litters prior to developing diabetes [Phelan et al., 1997], suggesting that increased malformations may occur as a result of maternal diabetes in NOD mice, and contribute to diminished reproductive performance.)

New mouse models of diabetes, with modifications of immune regulation or insulin signaling are being generated by transgenic and knockout technology and may be useful for the study of diabetic pregnancy-induced malformations. Nevertheless, having demonstrated that maternal hyperglycemia is responsible for the adverse effects of diabetic pregnancy on embryonic development, it may be perfectly appropriate to employ chemically induced models of insulin insufficiency. In addition, a non-diabetic animal can be used to perturb pathways, which mediate the effects of high glucose on embryogenesis to elucidate their involvement in this process.

We developed a mouse model of diabetic pregnancy in which diabetes is induced about 3 weeks prior to pregnancy with streptozotocin. Streptozotocin has a half-life of 15 min [Gilman et al., 1980], and its effects on embryogenesis can be prevented by sufficient normalization of glucose in diabetic animals [Fine et al., 1999], suggesting that streptozotocin is not teratogenic per se, when administered weeks prior to organogenesis, but its adverse effects are due to its induction of diabetic hyperglycemia. Within a week of streptozotocin administration, mice become hyperglycemic (blood glucose levels >300 mg/dl, compared to normal blood glucose 110–150 mg/dl). Diabetes is treated with subcutaneously implanted insulin pellets that constitutively produce insulin. The insulin pellets control the diabetes well enough that the mice gain weight and ovulate normally. However, once mice become pregnant, increased glucocorticoid production and gluconeogenesis induce hyperglycemia (mean blood glucose >300 mg/dl) beginning on day 4.5 of pregnancy. Age-matched non-diabetic mice, however, have normal beta cell function and maintain euglycemia during pregnancy [Phelan et al., 1997]. Notably, in embryos examined on day 10.5 or 11.5, NTD, primarily exencephaly, were increased 3-fold compared to embryos of non-diabetic mice [Phelan et al., 1997]. The exencephaly was morphologically similar to those caused by homozygous Splotch (Sp) alleles, which result from null mutation of the Pax3 gene [Auerbach, 1954; Epstein et al., 1993; Chalepakis et al., 1994]. We showed that Pax3 expression is significantly inhibited in embryos of diabetic mice beginning on day 8.5 of development [Phelan et al., 1997]. This is not due to universal inhibition of gene expression or developmental delay, as expression of control genes such as fibronectin, 36B4, and GAPDH was not affected, and embryo growth and development on day 8.5 (as determined by crown/tail length, number of somites, initiation of neural tube fusion) was morphologically normal. This suggested that deficient expression of Pax3 in embryos of diabetic mothers phenocopied the effect of Pax3 loss of function.

Undoubtedly, there are other developmental control genes whose expression, depending on the timing and severity of the diabetic state, could be inhibited by maternal diabetes and lead to embryonic defects. However, since there are no redundant pathways to compensate for Pax3 deficiency on neural tube closure, simply reducing Pax3 expression below a critical threshold will lead to a NTD with certainty. Therefore, investigating how maternal hyperglycemia disturbs the normal expression of this gene, and how Pax3 deficiency leads to NTD, is important to understand and prevent serious maldevelopment during diabetic pregnancy.

Molecular Regulation of Pax3 by Excess Glucose in the Embryo

There is much evidence that reactive oxygen species (ROS) are increased in tissues exposed to hyperglycemia, and that oxidative stress plays a critical role in the etiology of several diabetic complications [Giugliano and Ceriello, 1996; Feldman et al., 1997; Ruggiero et al., 1997; McDonagh and Hokama, 2000; Vinik et al., 2000; Kowluru and Kennedy, 2001].

There is much evidence that reactive oxygen species (ROS) are increased in tissues exposed to hyperglycemia, and that oxidative stress plays a critical role in the etiology of several diabetic complications. In rodent embryos, many studies have shown that ROS are increased by maternal diabetes or high glucose culture, and that administration of antioxidants, or transgenic over expression of Cu++/Zn++ superoxide dismutase (SOD), can prevent hyperglycemia-induced developmental defects.

In rodent embryos, many studies have shown that ROS are increased by maternal diabetes or high glucose culture, and that administration of antioxidants, or transgenic over expression of Cu++/Zn++ superoxide dismutase (SOD), can prevent hyperglycemia-induced developmental defects [Eriksson and Borg, 1991; Hagay et al., 1995; Eriksson and Siman, 1996; Sivan et al., 1996, 1997; Wentzel et al., 1997; Reece et al., 1998]. However, how oxidative stress disturbs embryogenesis has, until recently, been an unanswered question.

The distinct anatomical and developmental characteristics of the early post-implantation embryo confer particular vulnerability to oxidative stress. First of all, at this stage of development, embryos express the high Km Glut2 glucose transporter (as well as the low Km transporters, Glut1, Glut3, and possibly, Glut8 [Hogan et al., 1991; Takao et al., 1993; Carayannopoulos et al., 2000], which may not be of consequence during non-diabetic pregnancy, but could explain why intracellular glucose concentrations of the embryo are at equilibrium with maternal serum, regardless of maternal glycemia [Sussman and Matschinsky, 1988]. Thus, glucose flux into the embryo could be 2–3 (or more) times greater than during non-diabetic pregnancy. Second, during normal development, the early post-implantation embryo develops in a relatively hypoxic environment (2%–8% O2, compared to 20% in maternal arterial circulation [Rodesch et al., 1992; Fischer and Bavister, 1993]. This is due to the absence of vascular tissue in the embryo proper, and the several tissue layers across which maternal O2 must diffuse before reaching the embryo. Since there is very little O2 available, only about 8% of glucose is metabolized oxidatively before establishment of the embryo's circulatory system [Akazawa et al., 1994], and so, very little Omath image would be produced by the embryo during normal development. Thus, it makes physiologic sense that the embryo does not express free radical scavenging enzymes at high levels until shortly before birth [Frank and Groseclose, 1984; el-Hage and Singh, 1990]. However, this being the case, even modest increases in Omath image production could profoundly increase oxidative stress.

We hypothesized that pathways activated by high glucose uptake by the embryo would induce oxidative stress, and that oxidative stress inhibits Pax3 expression. To test this, the diets of diabetic and non-diabetic mice were supplemented with vitamin E succinate, raising the dietary intake of vitamin E approximately 60-fold, and Pax3 mRNA and NTD were assayed [Chang et al., 2003]. Vitamin E succinate prevented the increase in malondialdehyde (a marker of lipid peroxidation, used as an indicator of oxidative stress) caused by maternal diabetes, without affecting blood glucose concentrations; vitamin E prevented the approximately 8-fold decrease in Pax3 mRNA in embryos of diabetic mice, as well as the significant increase in NTD.

Conversely, to test whether increased oxidative stress inhibits Pax3 expression, pregnant mice were injected with antimycin A (AA), an inhibitor of complex III of electron transport, to increase Omath image production [Turrens et al., 1985; Garcia-Ruiz et al., 1995]. AA (which had no effect on blood glucose levels) significantly inhibited Pax3 expression and increased NTD, showing that oxidative stress is sufficient to elicit these effects of maternal diabetes.

As an additional approach, the effects of antioxidants to block the effects of high glucose-containing media on Pax3 expression was tested. Vitamin E, and also the membrane-permeable analog of the important cellular antioxidant, reduced glutathione (GSH), GSH ethyl ester, prevented the inhibition of Pax3 expression, and AA replicated the effects of high glucose, demonstrating that the adverse effects of glucose to increase oxidative stress occurred within the embryo. The effect of oxidative stress to inhibit Pax3 expression was not due to direct cytotoxicity resulting in death of Pax3-expressing cells, as AA which is a much more potent oxidant than glucose metabolism, did not induce DNA strand breaks at concentrations that were sufficient to inhibit Pax3 expression.

Potential Mechanisms by Which Oxidative Stress Inhibits Pax3 Expression

While it is not known how oxidative stress could affect embryo gene expression, there are examples from other systems. For example, in the NF-κB pathway, oxidative stress can activate I-κB kinase (IKK) which phosphorylates IκB, which thereby dissociates from NF-κB, allowing NF-κB to translocate to the nucleus and regulate gene expression [Karin and Ben-Neriah, 2000]. There are many other examples in which transcriptional activities or DNA binding of factors, such as p53, AP-1, and Sp1, are regulated by redox status [Cox et al., 1995; Wu et al., 1996; Jayaraman et al., 1997; Marshall et al., 2000; Moran et al., 2001; Webster et al., 2001].

In addition to mitochondrial production of Omath image, there are several metabolic pathways that regulate redox status and may have transcriptional effects. For example, increased glycolytic flux stimulates the hexosamine biosynthetic pathway (HBP), in which the enzyme, glutamine:fructose-6-phosphate amidotransferase (GFAT), catalyzes the amido transfer from glutamine to the glycolytic intermediate, fructose-6-phosphate, yielding glucosamine-6-phosphate [Marshall et al., 1991]. The HBP provides the substrates for O-linked glycosylation of proteins therefore, increased HBP flux could increase glycosylation, rather than phosphorylation, of cytoplasmic or nuclear proteins [Wells et al., 2001]. Glucosamine-6-phosphate also inhibits glucose-6-phosphate dehydrogenase (G6PD), the rate-limiting enzyme of the pentose shunt pathway [Kanji et al., 1976]. G6PD activity is coupled to reduction of NADP+ to NADPH. NADPH is an important regulator of redox status by serving as a co-factor for glutathione reductase, which converts oxidized glutathione (GSSG) to reduced glutathione (GSH), and by enhancing the activity of catalase, which converts H2O2 to H2O and O2. The ratios of oxidized and reduced NAD co-factors have been shown to regulate transcription factor binding and expression of circadian clock genes [Rutter et al., 2002]. In addition, cell cycle-dependent activation of histone H2B transcription requires the glycolytic enzyme, glyceraldehyde-3-PO4 (GAPDH), as a co-activator, and that association of GAPDH with a DNA-binding complex is enhanced by NAD and inhibited by NADH [Zheng et al., 2003]. GAPDH activity may be inhibited by mitochondrial production of Omath image [Nishikawa et al., 2000a,b] which would increase NAD:NADH. Inhibition of GAPDH activity is linked to malformations in rat embryos [Wentzel et al., 2003], although it is not known whether this is due to altered ratios of NAD co-factors or transcriptional effects of GAPDH, or an indirect effect of inhibition of GAPDH activity. Inhibition of GAPDH may increase accumulation of diacylglycerol (DAG) and stimulation of protein kinase C (PKC) [Nishikawa et al., 2000a,b]. We have observed that DAG and PKC are stimulated in embryos of diabetic and glucose-injected mice [Hiramatsu et al., 2002], although we do not yet know whether increased PKC activity is mechanistically involved in altered gene expression and malformations. While activation of PKC could directly regulate transcription, PKC also stimulates NADPH oxidase activity [Hua et al., 2003], which would contribute to decreased production of GSH. Clearly, there are multiple interacting biochemical processes by which increased glucose metabolism in the embryo can increase oxidant production and inhibit scavenging. Precisely how oxidative stress interferes with developmental control of transcription in the embryo will depend upon identification of the transcription factors involved, and understanding how their activation during development is disturbed by increased oxidant status.

Susceptibility to Inhibition of Pax3 Expression by Maternal Diabetes may be Modified by Genetic Background

In humans, there is evidence that genetic background and ethnicity influence susceptibility to diabetic complications such as nephropathy and cardiovascular disease, independent of glycemic control [Doria et al., 1995; Cappuccio, 1997; Weijers et al., 1997; Hosey et al., 1998; Martins et al., 2002]. While it is known that ethnicity is a factor in the development of gestational diabetes [Homko et al., 1995],

While it is known that ethnicity is a factor in the development of gestational diabetes, which increases risk for subsequent pre-gestational type 2 diabetes, the contribution of genetic background to diabetic embryopathy in humans has not been carefully examined. In animal models, penetrance or expressivity of a phenotype associated with a genetic mutation is well known to be influenced by strain background, although the specific modifier gene(s) responsible are seldom characterized.

which increases risk for subsequent pre-gestational type 2 diabetes [Schaefer-Graf et al., 2002] the contribution of genetic background to diabetic embryopathy in humans has not been carefully examined. In animal models, penetrance or expressivity of a phenotype associated with a genetic mutation is well known to be influenced by strain background [Linder, 2001], although the specific modifier gene(s) responsible are seldom characterized. In the rat, the effects of maternal diabetes to induce malformations is strain dependent [Eriksson, 1988]. Embryos from a malformation-resistant rat strain express increased levels of Mn++-SOD and catalase mRNA, suggesting that higher levels of free radical scavenging enzymes protect these embryos from oxidative stress caused by maternal diabetes [Cederberg et al., 2000].

We have found one mouse strain, C57Bl/6J, in which diabetic pregnancy fails to increase NTD, whereas diabetic pregnancy generally increases NTD three or more fold in several other strains (FVB, ICR, 129/Sv) [Pani et al., 2002a]. We hypothesized that in C57Bl/6J embryos, maternal diabetes fails to inhibit Pax3 expression below a critical threshold. If this hypothesis is correct, then investigating the genetic basis for resistance to this effect of maternal diabetes could point to essential biochemical pathways involved. Furthermore, if susceptibility to diabetic pregnancy-induced NTD is polymorphic in humans as well as in mice, it may be possible to screen individuals for risk for this cause of NTD.

Consistent with the hypothesis, maternal diabetes significantly inhibits Pax3 expression in a strain that is susceptible to diabetes-induced NTD, FVB, but not in C57Bl/6J embryos. Resistance to NTD caused by diabetic pregnancy is determined by the genotype of the embryo and is a dominant trait, as heterozygous C57-FVB embryos are resistant whether the mother or the father is of the resistant strain. There is no significant difference between strains in expression of catalase, Cu++/Zn++ SOD, Mn++-SOD, γ-glutamcystein synthetase, and glutathione peroxidase, suggesting that expression of genes that regulated free radical scavenging does not account for the genetic differences. However, differences in enzymatic activity, rate of free radical generation, or other glucose-regulated pathways are some possible explanations for the difference between susceptible and resistant strains.

Insufficient Expression of Pax3 Leads to NTD by De-Repressing p53-Dependent Apoptosis

Pax3 encodes a DNA-binding transcription factor [Goulding et al., 1991], and several groups have found that expression of a number of genes (those encoding N-CAM, myelin basic protein, the c-Met receptor, MyoD and Myf-5, Pax7, and Msx2) is up- or down-regulated by Pax3 [Laborda, 1991; Moase and Trasler, 1991; Kioussi et al., 1995; Epstein et al., 1996; Maroto et al., 1997; Tajbakhsh et al., 1997; Borycki et al., 1999; Kwang et al., 2002]. In addition, we identified two additional genes, Dep-1, and cdc46, whose expression is increased and decreased, respectively, by Pax3 [Cai et al., 1998; Hill et al., 1998]. Therefore, it has generally been held that Pax3 participates in the regulation of genes that direct tissue-specific morphogenesis. However, of the genes for which transcription control elements have been characterized, only MyoD contains an element with optimal binding sites for both the paired and homeodomains of Pax3 which we identified to be essential for high affinity binding [Phelan and Loeken, 1998]. This raises the question of whether altered expression of at least some of these genes may be indirect, or secondary to some other Pax3-dependent process. Expression of Pax3, as well as other members of the Pax family, had been shown to cause oncogenic transformation of cell lines and tumor formation in nude mice [Maulbecker and Gruss, 1993], and translocation of a part of the human PAX3 gene encoding the DNA binding domains is associated with pediatric rhabdomyosarcoma [Barr et al., 1993]. Furthermore, down regulation of the protein produced by this translocation with PAX3 antisense oligonucleotides activates cell death of rhabdomyosarcoma cells in culture [Bernasconi et al., 1996]. This prompted us to investigate whether Pax3 might be necessary for cell viability during neural tube formation. We found that Pax3 deficiency, caused either by reduced expression from maternal diabetes or the Sp/Sp genotype, leads to apoptosis. This, then, provided a cellular explanation for NTD resulting from Pax3 deficiency [Phelan et al., 1997]. Our studies were limited to the neural tube, but others showed that Pax3 is necessary for migration of neural crest cells and limb muscle precursors [Daston et al., 1996; Yang et al., 1996; Conway et al., 1997] and that apoptosis is prevalent in newly formed somites of Sp/Sp embryo [Borycki et al., 1999]. Therefore, loss of Pax3 expressing cells by apoptosis appears to be a general feature of abnormal development in Pax3 deficient embryos. However, it was not clear if Pax3 inhibits a cell death program, or if apoptosis resulted from failure of a Pax3-dependent developmental program.

To examine the regulation of apoptosis by Pax3 more closely, and specifically, to test whether this apoptosis occurs by a p53-dependent mechanism, we crossed Sp/+ and p53+/− mice to generate double heterozygotes [Pani et al., 2002b]. Sp/+ p53+/− mice were crossed, and embryos were obtained on day 10.5. As an additional approach, a p53 inhibitor, pifithrin-α, was administered during neural tube formation (days 8.5 and 9.5) to pregnant Sp/+ mice. As expected, all Sp/Sp embryos that were w.t. at the p53 locus developed spina bifida ± exencephaly with 100% penetrance, and TUNEL-positive cells were abundant at sites of NTD. Remarkably, homozygous p53 deficiency rescued not only apoptosis in Sp/Sp embryos, but also neural tube closure, as none of the Sp/Sp p53−/− embryos displayed NTD. p53 heterozygosity or inhibition with pifithrin-α prevented NTD in about half of the Sp/Sp embryos. These results profoundly alter our understanding of how Pax3 directs neural tube closure. Since neural tube closure was completely normal in Sp/Sp embryos as long as they were p53-deficient, this suggests that neural tube closure is a Pax3-independent process, and the sole required function of Pax3 in neural tube closure is to inhibit p53-dependent apoptosis.

Since Pax3 is a transcription factor, and there is evidence that another member of the Pax family, Pax5, inhibits p53 expression at the transcriptional level [Stuart et al., 1995], we considered that Pax3 could inhibit p53 gene expression. On the other hand, acute regulation of p53 protein levels is primarily post-translational, due to dissociation from mdm2, a ubiquitin ligase that stimulates proteosome-dependent degradation [Colman et al., 2000]. Phosphorylation or acetylation of p53 in response to stresses, such as radiation, can inhibit association with mdm2, and enhance p53 stability [Craig et al., 1999]. Activation of PKB/Akt increases phosphorylation of mdm2, which enhances it association with p53, thereby decreasing p53 stability, but the PTEN tumor suppressor protein inhibits activation of Akt, thereby enhancing p53 stability [Mayo and Donner, 2002]. P14(Arf) binds to mdm2 and sequesters it away from p53, thereby enhancing p53 stability [Kamijo et al., 1998]. There is also evidence that p53 protein levels may regulated at the level of translation, as there are sequences that suppress translation of p53 in response to UV or γ irradiation or growth arrest that are located in both the 5′ and 3′ UTR's [Mosner et al., 1995; Fu et al., 1999]. A 40 kDa COO-terminal fragment of p53 may derepress the 5′ UTR [Mokdad-Gargouri et al., 2001], while the RNA binding protein, HuR, specifically binds to the 3′ UTR of p53 mRNA and enhances p53 translation in response to UV irradiation [Mazan-Mamczarz et al., 2003].

To test whether the inhibition of p53-dependent apoptosis by Pax3 might be due to either inhibition of p53 mRNA or protein accumulation, we assayed p53 mRNA levels in w.t., Sp/+, and Sp/Sp embryos by RT-PCR, and p53 protein levels by immunoblot. There was no effect of Pax3 on p53 mRNA. However, in whole embryos, there was about 2-fold less p53 protein in w.t. compared to Sp/Sp embryos. Because these assays were performed using whole embryos, and Pax3 is expressed at only limited sites in the embryo, the magnitude of the effect of Pax3 on p53 in Pax3-expressing cells is probably much more than 2-fold. Understanding exactly how Pax3 regulates p53 either translationally or post-translationally will require additional investigation.

A related question is why Pax3 is required to inhibit p53-dependent apoptosis. Noting that transformation of fibroblasts with Pax-expressing plasmids induces focus formation in vitro and tumors in nude mice [Maulbecker and Gruss, 1993], and that PAX3 is expressed in several human neuroepithelial, neural crest, or myoblast-derived malignancies such a melanoma, neuroblastoma, pediatric rhabdomyosarcoma, and Ewing's sarcoma [Galili et al., 1993; Shapiro et al., 1993; Schulte et al., 1997; Barr et al., 1999; Vachtenheim and Novotna, 1999; Scholl et al., 2001; Harris et al., 2002], it is possible that p53 activates cell cycle withdrawal. Repression of this could be essential for the growth and migration of neural crest and neuroepithelium as well as proper dorsal ventral patterning of the neural tube. However, when PAX3 is inappropriately expressed in tumors, repression of p53-dependent cell cycle withdrawal may contribute to tumor growth. Alternatively, noting that oxidative stress can activate p53 and lead to cell death [Webster and Perkins, 1999; Trinei et al., 2002; Menendez et al., 2003], and that the sudden increase in delivery in oxygen to the embryo upon the establishment of its own cardiovascular system and blood supply, which occurs more or less coincident with the beginning of neural tube closure, could increase Omath image and activation cell death, it may be necessary to activate expression of genes which would override p53-induced cell death. Thus, during diabetic pregnancy, the increased neuroepithelial apoptosis leading to NTD may be due to a combination of oxidative stress-induced stabilization of p53, along with de-repression of p53 due to insufficient expression of Pax3.

SUMMARY

  1. Top of page
  2. Abstract
  3. BACKGROUND AND HISTORICAL PERSPECTIVE
  4. MOLECULAR REGULATION OF NTD RESULTING FROM DIABETIC PREGNANCY
  5. SUMMARY
  6. REFERENCES

A general scheme to explain how diabetic pregnancy causes NTD can be summarized as follows:

  • 1.
    Maternal diabetes increases glucose concentrations in maternal circulation, including in the uterine artery.
  • 2.
    Glucose, originating from the maternal circulation, is freely transported to the embryo and transported into embryo cells.
  • 3.
    Increased glucose taken up by embryo cells, through a combination of increased oxidative metabolism at a stage of development in which free radical scavenging is immature, and altered activities of biochemical pathways that contribute to redox homeostasis, causes oxidative stress in the embryo.
  • 4.
    Oxidative stress leads to decreased expression of Pax3, a gene required for neuroepithelial and neural crest development. Expression of other genes that participate in the formation of the neural tube, or other organs, may also be affected, but since there are no redundant pathways, which compensate for Pax3 deficiency, simply reducing the expression of Pax3 below a critical threshold at vulnerable stages in neural tube formation may be sufficient to induce a NTD.
  • 5.
    As a result of decreased production of Pax3, synthesis or stability of p53 protein increases, and this activates cell death, which aborts the process of neural tube closure.

It should be noted that deficient Pax3 expression leading to derepression of p53-dependent apoptosis may be involved in NTD resulting from causes other than diabetic pregnancy. For example, ionizing radiation and certain anticonvulsant drugs may induce oxidative stress [Nicol et al., 2000; Agrawal et al., 2001]. While the mechanisms by which increased folic acid administration reduces NTD in genetically susceptible individuals is not well understood, folate increases the production of methionine from homocysteine, thereby preventing the accumulation of homocysteine, which is an oxidant [Rosenquist et al., 1996; Chern et al., 2001]. In addition, folic acid may have antioxidant properties of its own [Nakano et al., 2001]. Thus, while it remains to be experimentally demonstrated, deficient Pax3 expression may result from multiple different sources of increased oxidant status in the embryo at critical stages of neural tube closure.

REFERENCES

  1. Top of page
  2. Abstract
  3. BACKGROUND AND HISTORICAL PERSPECTIVE
  4. MOLECULAR REGULATION OF NTD RESULTING FROM DIABETIC PREGNANCY
  5. SUMMARY
  6. REFERENCES