Teratogenic Predisposition in Diabetic Rat Pregnancy

(1)

(2)

(3)

Everything in excess is opposed to nature Hippocrates (c. 400 BC)

(4)

Supervisor:

Ulf Eriksson, Professor

Department of Medical Cell Biology Uppsala University

Co-supervisors:

Parri Wentzel, Associate Professor Department of Medical Cell Biology Uppsala University

Niklas Nordquist, PhD Department of Neuroscience Uppsala University

External examiner:

Gernot Desoye, Professor

Department of Obstetrics & Gynaecology Medical University of Graz

Austria

Dissertation chairman

Peter Hansell, Professor, Uppsala University

Examination board:

Peter Bergsten, Professor, Uppsala University Lennart Dencker, Professor, Uppsala University Ted Ebendal, Professor, Uppsala University

Ulf Hanson, Associate Professor, Uppsala University Lisa Juntti-Berggren, Professor, Karolinska Institute

(5)

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Ejdesjö, A.*, Wentzel, P.*, Eriksson, U.J. (2011) Genetic and envi- ronmental influence on diabetic rat embryopathy. Am J Physiol En- docrinol Metab, 300(3):E454–67

II Ejdesjö, A., Wentzel, P., Eriksson, U.J. (2012) Influence of mater- nal metabolism and parental genetics on fetal maldevelopment in di- abetic rat pregnancy. Am J Physiol Endocrinol Metab, 302(10):E1198-209

III Ejdesjö, A., Wentzel, P., Eriksson, U.J. (2012) Alterations in the expression of tissue specific genes of the developing mandible and heart in rat diabetic embryopathy. Manuscript

IV Ejdesjö, A., Nawroth, P.P., Eriksson, U.J. (2012) Receptor for Ad- vanced Glycation End products (RAGE) knockout reduces fetal dysmorphogenesis in murine diabetic pregnancy. Manuscript

Reprints were made with permission from the respective publishers.

* Shared first authorship.

(6)

(7)

Abbreviations

3-DG 3-Deoxyglucosone

AGE Advanced Glycation End Products ANOVA Analysis of Variance

AR Aldose Reductase

Bmp4 Bone Morphogenetic Protein 4

Cat Catalase

CLP Cleft Lip and Palate COX-2 Cyclooxygenase 2

CRL Crown Rump Length

CuZnSOD Copper Zinc Superoxide Dismutase DAG Diacylglycerol

ECSOD Extracellular Superoxide Dismutase F1-hybrid First-generation Offspring of a Crossbreed

F2-hybrid Second-generation Offspring of a Mating of Two Identical F1-hybrids

G6pdh Glucose-6-Phosphate Dehydrogenase Gapdh Glyceraldehyde-3-Phosphate Dehydrogenase

GD Gestational Day

GDM Gestational Diabetes Mellitus

Gdnf Glial Cell Line-derived Neurotrophic Factor GLUT-1 or 2 Glucose Transporter 1 or 2

Gpx1 and 2 Glutathione Peroxidase 1 and 2

H Rat Outbred Sprague-Dawley Sub-strain, Malformation-resistant HMGB1 High-mobility Group Protein B1

i.p. Intraperitoneal i.v. Intravenous

L rat Inbred Sprague-Dawley Substrain, Malformation-prone LW Offspring of L Female and W Male

MD Manifestly Diabetic Condition

MD(+) MD Rat with 2% Glucose Added in Drinking Water MD(-) MD Rat given 35mg/kg STZ

MD-LW Manifestly Diabetic LW Rat

MD-LWLW Offspring of MD-LW Female Mated with N-LW Male MD-WL Manifestly Diabetic WL Rat

MD-WLWL Offspring of MD-WL Female Mated with N-WL Male MnSOD Manganese Superoxide Dismutase

(10)

Msx2 Msh Homeobox 2

Mthfr Methylene-Tetrahydrofolate-Reductase N Normal, Non-diabetic Condition N-LW Non-diabetic LW Rat

N-LWLW Offspring of N-LW Female and LW Male N-WL Non-diabetic WL Rat

N-WLWL Offspring of N-WL Female and WL Male NAC N-Acetylcysteine

NCC Neural Crest Cell

NO Nitric Oxide

NOD Non-obese Diabetic

Nrf2 Nuclear Factor Erythroid 2-like NTD Neural Tube Defect

P53 Tumor Protein p53

PARP Poly(ADP-ribose)Polymerase Pax3 Paired Box 3

PKC Protein Kinase C

PLSD Protected Least Significance Difference

RAGE Receptor for Advanced Glycation End Products Ret Ret Proto-oncogene

ROS Reactive Oxygen Species Shh Sonic Hedgehog Homologue SOD Superoxide Dismutase STZ Streptozotocin

U rat Outbred Sprague-Dawley Substrain, Malformation-prone W rat Wistar Furth Inbred Substrain, Malformation-resistant WL Offspring of W Female and L Male

WT Wildtype

(11)

Introduction

Diabetes and pregnancy are of concern in obstetrical and diabetic care. Since the discovery of insulin in the 1920s (10), morbidity and mortality have decreased in women with pre-gestational diabetes (1, 29, 49, 65, 84, 99, 131, 134). Despite optimization of diabetic care before and during pregnancy, diabetic gestation is still burdened with a two- to five-fold increase in the risk of congenital malformations (1, 29, 49, 65, 84, 99).

Diabetes Mellitus

The disease diabetes mellitus was first mentioned in the ancient Egyptian script Papyrus Ebers, written about 1500 B.C. Therein is a description of excessive urinary production (i.e. polyuria), believed to signify a diabetic state (11, 137). The sweetness of the urine in Diabetes Mellitus was first noted in old Indian texts dated to the 5^th century AD. These texts describe patients with excessive urinary production, thirst, and emaciation, and whose urine tasted sweet or honey-like (69, 137). The medical term diabetes was first introduced by Aretaeus of Cappadocia (1^st or 2^nd century AD). The term diabetes is derived from the Greek word siphon, meaning “pass or run through” (11, 91). In 1674, Thomas Willis (1621-1675) published his findings after evaporating and tasting urine from diabetic patients, and described the taste as sweet “as if imbued with honey (quasi melle) and sugar”. This finding made it possible to distinguish diabetes mellitus from other forms of polyuria. Willis hypothesized the sweetness originated from the blood and gave the disease its modern name, Diabetes Mellitus (11, 137). A century later, Mathew Dobson established the sweetness in both blood and urine of diabetic patients was due to an excess of saccharine (44). In 1815, Michael Eugene Chevreuil (1786-1889) identified grape sugar (glucose) as the excessive saccharine (11, 91).

The functional aspects of the pancreas (the name derived from pan=all and creas=flesh) were first elucidated by Johann Conrad Brunner (1653- 1727) through the removal of the gland from dogs that subsequently developed polyuria and polydipsia (11). At that time, the pancreas was considered a non-vital exocrine organ involved in the conversion of fatty matter and to convert starch into sugar (52). In 1889, Minkowski and Mering demonstrated pancreatectomized dogs developed diabetes mellitus, and in 1892, they

(12)

established this morbidity could be reversed when pancreatectomized dogs were treated with subcutaneous implants of fragments of their own pancreas (137). In 1869, Paul Langerhans (1849-1888) described the morphological features of the pancreatic islands. In 1900, Opie suggested diabetes was a disease caused by degeneration of the islets of Langerhans and that these islets had an endocrine function. Further experiments by Massaglia in Italy in 1912 illustrated that destruction of the pancreatic acini did not cause gly- cosuria, whereas, destruction of the islets of Langerhans did (11). In 1909, Jean de Meyer named the presumptive substance secreted from the islets of Langerhans as insulin (137). The search for insulin finally ended when in 1922, Frederick Banting (1891-1941) and Charles Best (1899-1978) published their experiments on the treatment of pancreatectomized dogs with exogenous insulin produced from fetal calf pancreas (10) that demonstrated a regression of diabetic symptoms.

Diabetes Mellitus is considered as diverse diseases with different etiolo- gies that share one common symptom, hyperglycemia, i.e. excess glucose in the blood. There are two main types of diabetes mellitus: type I and type II.

The etiology behind type I diabetes is an autoimmune destruction of the β- cells in the islets of Langerhans. Type II diabetes emerges from disturbed β- cell function and insulin secretion, often in combination with enhanced resistance to insulin in body tissues. These disturbances finally lead to the destruction of the pancreatic cells (8).

As an epidemic disease, diabetes mellitus affected 171 million people globally in 2000. The prevalence of 2.8% in 2000 is projected to increase to 4.4% in 2030, and affecting 366 million people worldwide (186).

Diabetic Complications

Diabetes mellitus is accompanied by several complications. Acute symptoms such as thirst, polyuria, and weight loss are cardinal symptoms along with more severe morbidities such as ketoacidotic coma and iatrogenic hypoglycemia (17, 131). Long-term, diabetes leads to pathological changes in the vascular system that causes both micro- and macroangiopathies (17, 131). In the final stages, diabetic complications result in ischemic heart disease, stroke, nephropathy and end-stage renal disease, retinopathy, neuropathies, venous ulcer, and bacterial infections (17, 131).

Diabetic Pregnancy

In the pre-insulin era, diabetic women of childbearing age had decreased fertility, but in the case of a conception, morbidity was high for both the

(13)

diabetic pregnancies were hampered by high perinatal mortality (35), and it was not until the 1970s that the outcome for diabetic pregnancies improved (99). Even in the 2010s, diabetic pregnancy still has a 2-5 times higher rate of malformations than a non-diabetic pregnancy (1, 29, 49, 65): the frequency of complications during pregnancy is similar for both Type I and Type II diabetes (14, 109). However, the prevalence of pre-gestational diabetes in pregnancy has increased due to an increase in Type II diabetes (109). Alt- hough approximately 0.5% of all pregnancies are complicated by pre- existing diabetes mellitus, there is wide geographical and ethnic variation in prevalence (14, 171).

Despite optimal glycemic control and good pre-gestational care, it has been difficult to normalize the malformation rate, which has been a goal since the 1989 St. Vincent declaration (1, 29, 65, 134). Infants born to mothers with pre-existing diabetes have increased risk for cardiovascular and spine malformations, esophageal-gut-anal atresia, limb reduction defects, polydactylia, orofacial clefts and hypospadias (1).

Besides increased malformation rates, diabetic pregnancy is associated with an increased risk of macrosomia, prematurity, dystocia, preeclampsia, and increased frequency of caesarian sections (16, 65, 134). Even in well- controlled diabetic pregnancies, fetal macrosomia occurs: this phenomenon is postulated (43) to be due to disturbed buffering of glucose in the placenta.

Neonatal morbidity as high as 80% is reported in insulin treated Type I pregnancies, with the newborn child frequently having low Apgar scores and increased risk of hypoglycemia, respiratory disorders, and hyperbiliru- binemia (65).

The best predictor for adverse fetal outcome in human diabetic pregnancy is increased maternal HbA1c concentration (84, 112, 161). The most effec- tive clinical prevention of malformations appears to be pre-conceptional care that aims for optimal control of the diabetic state (65, 132, 139, 157).

Gestational diabetes mellitus (GDM) is a condition in which impaired glucose tolerance and insulin resistance develop during pregnancy, most frequently in the third trimester (6). The placenta and its endocrine function are partly involved in insulin resistance during pregnancy (42). The prevalence of GDM is approximately 1% in Sweden (66), whereas, in the United States, the prevalence is much higher, ranging from 2% to 7%, depending on area and study (93), and in some populations, such as native North Ameri- cans in Canada, the prevalence is as high as 12.8% (93). GDM is associated with increased risk of fetal loss, fetal macrosomia, neonatal hypoglycemia and maternal hypertension (6), and is considered a predisposition for type II diabetes unmasked by pregnancy, and women with GDM have increased risk of developing type II diabetes after pregnancy (2, 6, 7, 20).

GDM is heterogeneous regarding congenital malformations. In most studies, there is no difference in malformation rate between GDM pregnancy and non-diabetic pregnancy (1, 133); although there appears to be a subgroup of

(14)

GDM pregnancies with increased malformation rate where a considerable fraction of unrecognized pre-gestational diabetes may be involved (67, 133).

Experimental Animal Models

The rat (Rattus norvegicus) is an appropriate and well-investigated animal model for experimental diabetic pregnancy. The gestational period in the rat is 22 days, and the developmental stages of rat pregnancy are outlined in Table 1. Organogenesis begins on gestational day (GD) 9. At this time, the embryo undergoes major development and all the main structures of the fetal body and organs are defined (5, 101). The teratogenic window in diabetic rat pregnancy occurs between GD6 to GD10 (54): GD10 is the optimal time for blood sampling of maternal parameters during organogenesis (160). Gesta- tional outcome is generally evaluated at GD20, which roughly corresponds to human gestational week 30 (5).

Table 1. Developmental stages of rat in utero. Modified from Growth (5).

CRL=Crown Rump Length

Age (GD) CRL (mm) Stage Development

1 – 5 Cleaveage & blastula From 1 cell stage to

free blastocyst in uterus

6 – 7.75 Gastrula Implantation

8.5 Primitive streak Primitive streak

9 – 11 1.0 – 3.3 Neurula Neural plate, organo-

genesis begins. So- mite stages 1 - 25 11.5 – 12.375 3.8 – 6 Tail bud embryo Somite stages 26 - 40

12.5 6.2 Complete embryo Somite stages 41- 42

12.75 – 16 7 – 15.5 Metamorphosing

Embryo Somite stages 43 – 65

17 – 22 16 – 40 Fetus Structures complete,

growth of fetus

A common procedure for inducing a metabolic state in experimental ani- mals, including the rat, is the intravenous (i.v.) or interaperitoneal (i.p). ad- ministration of streptozotocin (STZ), a β -cell toxic agent. STZ is taken up via GLUT-2 in β-cells and induces DNA alkylation and nitrous oxide (NO) mediated DNA damage (165). The DNA modifications activate Poly(ADP- ribose)Polymerase (PARP) to consume NAD⁺, which depletes cellular ATP:

the concomitant NAD⁺ and ATP deficiency further leads to β -cell necrosis (147, 165).

(15)

Alloxan, another agent used to induce a diabetes-like state in animals, is taken up by β-cells, and in its reduction by hexokinase induces a redox cycle that increases the production of reactive oxygen species (ROS) and leads to β-cell necrosis (165).

A third procedure, used preferentially in mice, is to inject glucose subcu- taneously into the pregnant animal, this renders them diabetic for a specific period (68).

Comparative studies of inbred rat strains are suitable, e.g. for linkage analysis, for studying genetic susceptibility for adverse fetal outcome. In a linkage study (122) two inbred rat strains, W (Wistar-Furth) with low teratogenic susceptibility, and L (Sprague-Dawley), with high teratogenic susceptibility, were investigated. The W strain is resistant to diabetes-induced skeletal malformations and has a resorption frequency of 13% in diabetic pregnancy and 7% in non-diabetic pregnancy (122). Offspring of diabetic L mothers present with agnathia and micrognathia at a combined frequency of 20% but there are no such malformations when the L mother is non-diabetic.

The resorption rate is 30% in diabetic L rats and 8% in non-diabetic L rats:

malformation types are illustrated in Figure 1.

Figure 1. Photographs of GD20 normal, malformed (micrognathia, agnathia and cleft lip & palate), and resorbed rat fetuses.

NORMAL MICROGNATHIA AGNATHIA

CLEFT LIP & PALATE RESORPTION

(16)

The L rat strain originates from a locally outbred strain of Sprague-Dawley rats, denoted U rat, where the offspring have an elevated risk of 15-20% for congenital malformations in diabetic pregnancy but less than 1% in non- diabetic pregnancy (56, 62). On exposure to a diabetic milieu in utero, the U strain is susceptible to an increased rate of resorptions, decreased fetal weight, increased placental weight and delayed skeletal maturity (56, 57, 62).

In studies of diabetic embryopathy, other animal models are also used. Non- obese diabetic (NOD) mice (126) spontaneously develop diabetes, and the offspring of diabetic NOD dams present with neural tube defects (NTD) and kinky or waved vertebral columns. ICR (Institute for Cancer Research) mice, made diabetic with streptozotocin, present with NTD and craniofacial defects (127, 135). However, there is currently little information on malformation susceptibility in diabetic mice from the C57Bl/6 strain. Pregnant transgenic (CuZnSOD) diabetic mice have decreased malformations and fetal loss, whereas, offspring of CuZnSOD knockout dams have high total morbidity (80, 198). Furthermore, the Bio-Breeding/Edinburgh (BB/E) rat develops an autoimmune mediated destruction of β-cells, resembling human type I diabetes, and offspring have increased fetal resorption rates and neonatal mortality, decreased fetal weight, disturbed skeletal development and increased rate of congenital malformations (55). A model for diet-induced type II diabetes, the Cohen rat, has been used to investigate embryonic and fetal outcome: the malformation rate is higher in diabetic Cohen rats than in STZ-induced diabetic rats (200).

Teratogenic Mechanism

Maternal diabetes alters several metabolic pathways in the pregnant mother.

Besides hyperglycemia, the diabetic state increases serum levels of triglycerides, cholesterol, β -hydroxybutyrate and branched chain amino acids and causes derangements of free amino acid levels (50, 51, 160, 180). In vitro experiments on embryos highlight a teratogenic effect of D-Glucose, but not L-Glucose (34), suggesting the metabolism of D-Glucose is a major teratogenic process. However, the multifactorial etiology of diabetic embryopathy is further strengthened by repeated findings of malformations in embryos cultured in media containing other diabetes-related metabolic disturbances than just excess glucose, such as increased β -hydroxybutyrate levels (26, 92), excess keto-isocaproic acid (58), and non-hyperglycemic diabetic serum (180). The teratogenic process of diabetic pregnancy is linked to several cellular disturbances in the embryo. The diabetic milieu increases oxidative stress (58, 59, 194), enhances lipid peroxidation (30, 184), and decreases

(17)

teratogenic mechanism is associated with decreased catalase activity (31), disturbed inositol metabolism (85, 89, 162, 140, 185), disturbed arachidonic acid metabolism (77), lowered prostaglandin E2 levels (184) and diminished cyclooxygenase-2 (COX-2) gene expression (184), increased apoptosis and profound disturbances in cell signaling through alterations in protein kinase C (PKC) (72, 185). Although there is an association between increased sorbitol accumulation and dysmorphogenesis in the embryo (63, 85, 89, 162), attempts to diminish this accumulation with inhibitors of aldose reductase (AR) has not prevented diabetes-induced maldevelopment in the offspring (63, 89).

Another key enzyme in glycolysis, Glyceraldehyde-3-phosphate dehydro- genase (Gapdh), is inhibited in rat embryos subjected to diabetes in vivo and high glucose in vitro (176). Gapdh inhibition may be the result of the poly(ADP-)ribosylation of Gapdh by activated poly(ADP-ribose) polymerase (PARP) (46) in an attempt to repair diabetes-induced DNA damage (102, 103). The decreased glycolytic flux proximal to Gapdh (46) and the presence of increased ambient glucose levels subsequently yields an enhanced flux in the sorbitol (63, 89) and hexosamine pathways (90). The increased availabil- ity of proximal glycolytic intermediaries increases diacylglycerol (DAG) production and activates several PKC isoforms (72, 88) and enhances the flux in the Advanced Glycation End product (AGE) pathway (37, 64). Thus, several consequences of inhibited Gapdh activity may contribute to different facets of the teratogenic outcome in diabetic pregnancy.

Another component of diabetic embryopathy may be a disruption in NCC migration, proliferation, and differentiation. The fate and programming of NCC is determined before migration occurs. This has been proved experi- mentally through transplanting neural crest cell populations into different migratory regions, which results in ectopic tissue expressing differentiation specific for the origin of the transplant (121).

The mandible develops from the first pharyngeal arch and this development is disturbed in offspring of diabetic U and L rats (78, 121, 152). The pharyngeal arch is composed of an inner layer of endoderm, an outer layer of ectoderm, and a core of mesoderm (78). The mesodermal core of the first pharyngeal arch is populated with NCC (40, 108, 121) arising from the cau- dal midbrain and rhombomeres 1 and 2. These NCCs subsequently form the connective and skeletal tissue of the first pharyngeal arch (108, 121).

Prevention

The clinically used method of preventing congenital malformations in diabetic pregnancy is by diminishing the severity of the diabetic state by insulin treatment of the mother (29, 49, 65, 84, 131, 132, 134). Since this treatment regimen is not able to completely block the increased malformation rate (1,

(18)

29, 49, 65, 84, 131, 132, 134, 161), other possible anti-teratogenic agents have been explored. The teratogenic effects of maternal diabetes in vivo and hyperglycemia in vitro are diminished by the addition of antioxidants N- acetylcystein (179, 183, 184), vitamin E (151, 172, 195), vitamin C (150), SOD (183), catalase (59), glutathione peroxidase (59), lipoic acid (188) and folic acid (127, 182). Furthermore, the addition of glutathione (145) in vivo and arachidonic acid (179), PGE2 (179) and inositol (85, 140, 185) in vitro prevents the generation of malformations in embryos subjected to a hyperglycemic milieu.

Genetic Predisposition

Case reports from the pre-insulin era indicate a clustering of morbidities in certain diabetic mothers, whereas, other women have repeated normal pregnancies, despite being diabetic (187). Diabetic twin pregnancies are uncom- mon, especially with monozygotic twins. In both monozygotic and dizygotic twin diabetic pregnancies, one fetus can be seriously malformed, whereas, the other is not malformed (9, 199). It is also evident that for several terato- gens, both maternal and fetal genomic factors are known to influence the susceptibility for malformations (189).

Although the importance of embryonic genotype and maternal environment for dysmorphogenesis has not yet been fully elucidated, diverse susceptibility for ethanol-induced malformations in different mice strains pro- vides evidence for genomic contribution from both mother and fetus. Mice strains sensitive to ethanol-induced birth defects cross-mated with resistant strains and further reciprocal backcrosses demonstrate that gestational outcome is dependent on both maternal environment and embryonic genotype (74), although the exact factors, i.e. differences in intrauterine environment, maternal metabolism or fetal sensitivity, have not been investigated (45, 74- 76). Blastocyst transfer between non-diabetic ICR mice and diabetic NOD mice highlights the importance of both diabetes in the mother and genetic predisposition in the fetus, as ICR embryos developed in NOD uteri and NOD embryos developed in ICR uteri have lower implantation rates and viability and higher malformation rates than ICR embryos transferred to ICR mice (126).

Through comparing the outcome between two outbred sub strains of Sprague-Dawley rats (U-strain and H-strain) and their corresponding hybrids, early work by Eriksson (57) established that the difference in malformation susceptibility is due to the genetic disposition of both the mother and the embryo. Diabetic mothers of the malformation-prone U strain deliver approximately 17% malformations when the fetus carries the genotype U/U, whereas, diabetic mothers of the non-malformation-prone H strain deliver

(19)

The backcrossing of F1 hybrids of U and H (57) indicates increased malformations and resorptions occur when the diabetic mother is at least 50% U.

Although an increase in malformation rate does not occur in mothers with 100% U genome when fetuses still are 50% U. Furthermore, when the fetal genome contains 75-100% U genome, the malformation rate is further increased. In an attempt to identify the genes responsible for skeletal malformations in the L strain, a linkage study between the malformation resistant W and the malformation prone L strains revealed a strong genetic association to seven regions on chromosomes 4, 10, 14, 18, and 19, and a weaker association to 14 other loci on several other chromosomes, thereby, indicating a polygenic teratogenic mechanism for agnathia and micrognathia in this rat model (122).

As predisposing genetic factors, in both the mother and fetus are important, an attractive working hypothesis would be that the genes on the maternal side of the placenta determine the degree of metabolic derangements and that the fetal genome defines the threshold for a particular malformation or anomaly to occur.

Developmental Genes

Ret and Gdnf

Ret Proto-oncogene (Ret) is a tyrosine kinase receptor that regulates cell proliferation, migration, differentiation and tropism (136) and is expressed in the developing embryo in all lineages of the peripheral nervous system and cranial and vagal neural crest cells (136). Ret signals through the activation of Ras/Erk, Pi3k/Akt, p38Mapk and Jnk pathways (96), and disturbed Ret signaling in the developing embryo, i.e. the neural crest cells, is one of the mechanisms behind neurocristopathies such as Hirschprung’s disease (110, 136). Ret is the receptor for the glial cell derived neurotrophic factor (Gdnf) family of ligands (GFL). Gdnf is essential for neural crest cell migration and colonization, as Gdnf is a chemo attractant produced in the trajectory and target tissue of migrating neural crest cells (86, 196). Gdnf is expressed in several sites in the developing central nervous system (123) and Gdnf stimu- lates nerve fiber outgrowth and neuronal survival in the peripheral nervous system (47). Beside its action in the nervous system, Gdnf is expressed in many different tissues in the developing embryo (123).

Shh

The morphogen sonic hedgehog (Shh) belongs to the hedgehog family of genes and was first isolated as the signal in the zone of polarizing activity

(20)

(ZPA) (141). Shh activates the receptor Patched1 (Ptch1) to release its inhibition on the protein Smoothened (Smo), affecting downstream up- and down-regulation of target genes via the Gli family of transcription factors (142). Outgrowth of the first pharyngeal arch is dependent on NCC stimuli from Shh expressed in the foregut endoderm, and deprivation of Shh signaling inhibits the development of the first pharyngeal arch and hampers jaw development (22, 192). In chicken embryos where the foregut endoderm is deleted before the 7th somite stage, the mandible does not develop due to massive apoptosis of neural crest cells populating the first pharyngeal arch (22). Furthermore, in 5-8 somite stage chicken embryos, i.e. before NCC migration, the ectopic source of Shh near the presumptive level of the first pharyngeal arch induces supernumerary lower jaws through the induction of Bmp4 and Fgf8 gene expression (23). In addition, ectopic mandibular development is not induced in chicken embryos exposed exogenous Shh in other regions near the first pharyngeal arch, suggesting each cephalic segment is already primed to a specific response to Shh (23). In the chicken embryo, foregut endoderm is the source of Shh that is critical (22) for NCC survival in the first pharyngeal arch.

Bmp4

Bone morphogenetic protein 4 (Bmp4), a member of the transforming growth factor-beta (Tgf-β) superfamily, is involved in several developmental processes, such as gastrulation and odontogenesis, and was first identified for its ability to induce osteogenesis. Bmp4 is important for the segregation of the rhomboencephalic neural crest into three streams through inducing apoptosis in neural crest cells in rhombomeres 3 and 5. Neural crest cells from the first, trigeminal stream constitute the first pharyngeal arch (79, 154). As an inhibitor of mandibular outgrowth, Bmp4 down-regulates Fgf-8, which is essential for proliferation and cell survival in the developing mandible (158).

Msx2

Msh Homeobox 2 (Msx2), an important morphogen in craniofacial development, is expressed in the developing mandible and at several tissue sites in the developing embryo where epithelial-mesenchymal interactions occur (3).

Bmp4 signaling induces Msx2 gene expression, and this Bmp4/Msx2 inter- action is involved in selective apoptosis of NCC in rhombomeres 3 and 5 and the regulation of osteogenic activity in cranial sutures (3). Msx2 repress- es osteogenesis in the development of avian mandibles (113).

(21)

Pax3 and p53

Transcription factor Paired Box 3 (Pax3) is involved in diabetes-induced NTD (68) and dysmorphology in the Splotch mouse model (53). Embryos exposed to high glucose levels have decreased Pax3 gene expression in the neural tube region (68, 135), which is associated with increased neuroepithe- lial apoptosis (68). Pax3 suppresses Tumor Protein p53 (p53), and thereby apoptosis, through stimulating p53 degradation and inactivation in neural tube closure and cardiac neural crest cell dependent cardiac septation (116, 174). Suppression of p53 in Pax3 deficient mice prevents NTD (128) and the maldevelopment of cardiac outflow tract (116).

Antioxidant Genes

Scavenging enzymes, especially superoxide dismutase (SOD), catalase and glutathione peroxidase (61), are part of cellular defense against oxidative stress, i.e. reactive oxygen species (ROS). The addition of these scavengers protects rat embryos from glucose-induced dysmorphogenesis in vitro (59).

Superoxide dismutases (SOD) are composed of three different enzymes, cytosolic CuZnSOD, mitochondrial MnSOD, and extracellularly localized ECSOD, and convert superoxide radicals to the dismutation product H2O2

through the reaction (81):

2O₂·- —> H₂O₂ + O₂

Catalase (Cat) and glutathione peroxidase (Gpx) further convert H2O2 to H2O through the reaction (81):

2H2O2 —> 2H2O + O2

In rat embryos subjected to a diabetic environment in vivo, the gene expres- sion of CuZnSOD, MnSOD, ECSOD, Cat and Gpx1 is disturbed with decreased mRNA levels (197). Furthermore, in a hyperglycemic environment, gene expression of the CuZnSOD, MnSOD, ECSOD, Cat and Gpx1 decreas- es in rat cranial NCC but not in truncal NCC (178), and maternal diabetes in vivo decreases fetal mandible and cardiac gene expression of these scaven- gers (50). In rat embryos subjected to maternal diabetes both Cat activity (31) and the levels of cardiac Gpx1 decrease (181), suggesting the disturb- ance in ROS scavenging is an important part of the teratogenic process. Of note, however under normal circumstances, free radicals and ROS have an important roles in normal embryonic development (41).

(22)

Glucose Metabolizing Enzymes

Gapdh

Glyceraldehyde-3-phosphate dehydrogenase (Gapdh) is a glycolytic enzyme that converts glyceraldehyde-3-phosphate into 1,3-bisphosphoglycerate and is involved in several cell functions, such as cell cycle regulation (28), trans- lational control, proliferation and apoptosis (12). In rat diabetic embryopathy, Gapdh activity and expression decreases in embryos exposed hypergly- cemic environments both in vivo and in vitro (176). Furthermore, in cultured embryos, iodoacetate inhibits Gapdh which generates the same pattern of dysmorphogenesis as high glucose, indicating a important role of Gapdh inhibition in diabetic embryopathy (176). Gapdh is inhibited through both poly(ADP)-ribosylation by PARP in response to ROS-induced DNA damage (46) and superoxide radical mediated reaction, linking NADH with thiol on cys-149 (143). The subsequent reduced flow through glycolysis forces ex- cessive glucose to enter other metabolic pathways i.e. the polyol, hex- osamine and AGE pathways (24).

AR

Aldose reductase (AR) reduces glucose to sorbitol. In normoglycemia, the activity of AR is low, whereas, in a hyperglycemic environment, the flux in the polyol pathway is much greater (191). AR metabolism of glucose is involved in diabetic retinopathy, cataract, glomerulosclerosis, and neuropa- thies (191). Although several AR inhibitors have been developed (i.e. Sorb- inil), they are not yet in clinical practice (191). In the context of diabetic embryopathy, the fetus of pregnant diabetic rats and rat embryos cultured in high glucose have increased sorbitol accumulation, however, AR inhibitors administered to embryos exposed to a diabetic/hyperglycemic milieu have failed to block adverse outcomes (60, 63).

8-iso-PGF

2α

Lipid peroxidation of arachidonic acid by free radicals produces prostaglan- din-like products (F2-isoprostanes). In vivo, the production of F2- isoprostanes by ROS favors the formation of 8-iso-PGF2α (118). 8-iso-PGF2α

levels are an indicator for oxidative stress (30, 184). The offspring of diabet- ic rats have higher 8-iso-PGF2α levels than non-diabetic offspring (177).

Furthermore, the addition of 8-iso-PGF2α to cultured rat embryos in vitro generates similar malformations as high glucose, a toxicity that is reversible by adding antioxidants such as N-acetylcystein or SOD (177).

(23)

Advanced Glycation End Products (AGE)

Glycation (i.e. non-enzymatic glycosylation) of proteins and lipids (13) and the accumulation of Advanced Glycation End Products (AGE) (13, 148) both increase in diabetes. The classic formation of AGE proceeds through the Maillard reaction, which starts when a carbonyl group on a reducing sugar reacts with the N-terminal amino acid residue in a protein; this produces a Schiff-Base and results in an Amadori product. The first steps in the Maillard reaction are reversible, but further rearrangement lead to irreversi- ble cross-linking of the Amadori products and finally the formation of an AGE compound (18). Several highly reactive AGE precursors, such as glyoxal, methylglyoxal and 3-deoxyglucosone (3-DG) are produced under hyperglycemic conditions, either through spontaneous degradation of glucose from the Schiff’s base or through intermediary metabolism of nutrients (167). The different pathways involved in AGE formation are illustrated in Figure 2. Methylglyoxal is detoxified via the glyoxalase system, but under hyperglycemic conditions, the amount of methylglyoxal increases resulting in increased AGE formation (167, 173). Increased accumulation of AGE is involved in the pathogenesis of several diabetic complications, such as cataract (164), retinopathy (82), atherosclerosis (156), neuropathy (107, 169), and nephropathy (166).

Figure 2. Different pathways for AGE formation by hyperglycemia. SDH=Sorbitol Dehydrogenase

In rodent embryos cultured in high glucose concentrations, embryonic tissues have increased levels of the AGE precursor 3-DG (64), and malfor-

Glucose Glucose-6-P

Fructose-6-P Fructose 1,6-bis-P Glyceraldehyde-3-P Dihydroxyacetone-P

bisphosphoglycerate 1,3- GAPDH

Pyruvate Electron transport

O2·-

-

AGE

Methylglyoxal Glyoxal 3-deoxyglucosone

Protein

Schiff’s Base

Amadori Product Protein

Sorbitol Fructose

Fructose-3-phosphate

AR SDH

Lipid peroxidation

(24)

mations are induced if 3-DG is added to a culture medium containing physi- ologic concentration of glucose: this effect is reversible through adding SOD (64).

RAGE

AGE interact with the receptor for advanced glycation end products (RAGE). RAGE is a pattern recognition multiligand receptor belonging to the immunoglobulin superfamily of cell surface molecules. Ligand stimula- tion of RAGE initiates a signaling cascade via phosphatidylinositol-3 kinase, Ki-Ras, and MAPK kinase Erk1 and Erk2 (Figure 3). This induces intracel- lular activation of NF-κB (19, 148) and activation of NADPH-oxidases.

Finally, this pathway yields increased intracellular oxidative stress (107, 173). The RAGE receptor also binds other ligands involved in inflammation, such as S100/calgranulins, high mobility group box 1 (HMGB1), Amyloid- beta (138), and MAC-1 (124).

The inhibition or knockout of RAGE attenuates the detrimental effects of hyperglycemia in neuropathy (169) and nephropathy (166), and the addition of soluble RAGE (sRAGE) in vivo suppresses diabetes-induced atheroscle- rosis in aopE-null mice (129).

(25)

Figure 3. RAGE receptor ligands and their pathway to ROS overproduction.

AGES100/Calgranulins HMGB1

beta-amyloid

V1

C1

C2

NADPH NADP⁺

2 O2 2 O₂^·- + H⁺ NADPH oxidase

H2O2

CuZnSOD Intracellular

domain of RAGE

Regulation of transcription

m

Activation of NF B MAP kinase

(Erk 1/2) PI-3-K Ki-Ras

2 O2 2 O₂

H2

ROS

p21 RAS

m

ROS Production RAGE

(26)

Aims

Diabetic pregnancies are hampered with increased risk of congenital malformations. Genetic susceptibility for malformations is observed in experimental diabetic pregnancies with different rodent strains. Therefore, the aims of this work were:

Aim 1. To compare fetal genetics and maternal metabolism between two different inbred rat strains W and L, with low (W) and high (L) occurrences of malformations in diabetic pregnancy (Papers I and II).

Aim 2. To examine tissue-specific gene expression alterations of ROS scav- engers and key developmental genes in target organs, i.e. developing mandi- ble and heart anlage in embryos exposed to a diabetic environment in utero (Paper III).

Aim 3. To investigate the importance of the RAGE receptor in the cellular pathways leading to malformations in diabetic pregnancy (Paper IV).

(27)

Materials and Methods

Animals

The ‘Principles of Laboratory Animal Care’ (NIH publication no. 85–23, revised 1985; http://grants.nih.gov/grants/olaw/references/phspol.htm) were followed and the Ethical Committee on Animal Experiments in Uppsala, Sweden, approved the research protocol including all experimental procedures involving animals. All animals were maintained at an ambient room temperature of 22°C with a 12 hour light/dark cycle and fed a commercial pellet diet (R36, Analycen AB, Linköping, Sweden) with free access to both food and tap water.

Rats (Papers I-II)

The animals were from either a Wistar Furth strain (denoted W, purchased from B and K, Sollentuna, Sweden), or from a locally housed Sprague- Dawley-derived inbred strain, denoted L, which has increased incidence of mandibular and cardiac malformations in diabetic pregnancy (56). Both W and L rats were used either as pure breed or as F1 generation. The animal crossbreeding scheme is displayed in Figure 4.

Manifest diabetes (MD) was induced in W, L, WL and LW rats (denoted MDW, MDL, MDWL, MDLW) through injection of 40 mg/kg streptozotocin (Sigma-Aldrich Stockholm, Stockholm, Sweden) into the tail vein; in a subset of L rats, MD was induced with 35 mg/kg streptozotocin (denoted MD(-)L). MD was confirmed within 1 week after the injection (Freestyle Mini; Abbot Laboratories, Chicago, IL). A blood glucose value ≥20 mmol/l denoted MD. The control group of non-diabetic (N) animals was not injected. In a subset of MDW rats, denoted MD(+)W, 2% glucose was added to the drinking water.

N and MD female rats were caged with N males according to the cross- breeding scheme presented in Figure 4. Conception was verified the next morning through the presence of sperm in a vaginal smear: this was desig- nated gestational day 0 (GD 0). On GD 10, venous blood was drawn from the tail vein of pregnant N and MD rats. The blood was centrifuged and the serum used for analysis of glucose/lipid compounds, free amino acids, and 8- iso-PGF2α.

(28)

On GD 20, the animals were killed by cervical dislocation after mild ether anesthesia, and the uterine horns were quickly dissected. From each horn, the fetuses were dissected from their surrounding membranes and umbilical cord. Fetuses were subsequently weighed, visually inspected for external malformations and sex determination, and then decapitated. The placentas were dissected from adherent membranes and wiped on a filter paper to re- move excess fluid before weighing. Resorptions were counted, weighed, and morphologically evaluated. External fetal malformations were recorded;

these were mainly alterations of the facial skeleton, i.e. micrognathia, agna- thia and cleft lip and palate. The fetal livers were collected for 8-iso-PGF_2α evaluation. The heart tissue and mandible bone with cartilage were dissected from surrounding soft tissue and both were divided into two portions: one part from both the heart and the mandible was submerged in lysis buffer (Buffer RLT, QiaGEN GmbH, Hilden, Germany), and one part was snap- frozen in liquid nitrogen for later determination of enzyme activities.

Figure 4. Crossbreeding schematics for W and L matings in Papers I and II.

Rats (Paper III)

Animals were from a locally outbred Sprauge-Dawley substrain, denoted U, which displays an increased incidence of mandibular and cardiac malformations in diabetic pregnancy (56). MD was induced in the U rats by injection of 40 mg/kg streptozotocin (Sigma-Aldrich Stockholm, Stockholm, Sweden) into the tail vein, and was confirmed within one week after the injection (Freestyle Mini; Abbot Laboratories, Chicago, IL): a blood glucose

Manifestly Diabetic (MD)Non-Diabetic (N)

WW LL WL LW WLWL LWLW

N LWLW N LW LW

MD LWLW MD LW LW N WLWL

N WL WL

MD WLWL MD WL WL N WW

N W W

MD WW MD W W

N LL N L L

MD LL MD L L

N WL

N W L

MD WL MD W L

MD LW MD L W N LW N L W

MD LW MD L W

P0 Generation F1 Generation F2 Generation

Crossbreeding Scheme

(29)

value ≥20 mmol/l denoted MD. The control group of non-diabetic (N) animals was not injected.

Non-diabetic control (N) and manifestly diabetic female U rats were caged over night with N male U rats. Conception was verified the next morning through the presence of sperm in a vaginal smear: this was desig- nated gestational day 0 (GD 0). On GD11 or GD13, venous blood samples were taken from the tail vein for determination of blood-glucose levels (Freestyle Mini, Abbot). The pregnant rats were weighed and subsequently killed by cervical dislocation after mild ether anesthesia. From each uterine horn, embryos with intact membranes were excised and counted. Embryos were then further dissected from surrounding membranes with watchmaker’s forceps under a stereomicroscope. For each embryo, crown-rump length (CRL), somite number (not counted at GD 13), and visual morphology were assessed. The malformation score of the embryos was determined by assign- ing 0, 1, 5, or 10 points, respectively, to no, minor, less severe, or severe malformation (183). A malformation score of 0 indicated a completely normal embryo, fully rotated with a closed neural tube. Embryos given a score of 1 showed a single minor deviation from this pattern, mainly an open pos- terior end of the neural tube. A score of 5 signified one major malformation, most often an open neural tube in the rhombencephalon area or a slight tail twist, and a score of 10 indicated an embryo with multiple major malformations such as open neural tube, rotational defects, and/or heart enlarge- ment. A malformation score ≥1 denoted a malformed embryo. In addition, we determined the crown-rump length and somite number of each embryo.

From the embryos, the first pharyngeal arch bud and developing heart an- lagen were carefully excised and then placed separately in 350 µl lysis buffer (Total RNA Lysis Solution, Bio-Rad Laboratories AB, Sundbyberg, Swe- den) for subsequent gene expression analysis. In addition, the remaining embryonic portion (hereafter denoted whole embryo) was submerged in 350 µl lysis buffer (Total RNA Lysis Solution, Bio-Rad) for gene expression analysis.

Mice (Paper IV)

RAGE knockout mice (RAGE^-/-) were obtained from the University of Hei- delberg, Germany, by courtesy of Professors Bierhaus and Nawroth. Knock- out of the RAGE receptor was initially performed on 129/B6 mice with a LoxP flanked (floxed) cassette, which was further excised with Cre recom- binase. The F1 Cre/LoxP hybrids with deleted RAGE receptor were then backcrossed to the C57Bl/6 strain (36, 149). The wildtype (WT) mice were of the C57Bl/6 strain (Charles River, Sulzfeldt, Germany). Diabetes was induced in WT and RAGE^-/- according to the multi streptozotocin model with 50 mg/kg Streptozotocin (Sigma-Aldrich) i.p. for five consecutive days.

(30)

Five days after the last injection, blood samples were taken from the tail vein and blood glucose was measured (Freestyle Lite, Abbot Laboratories): a blood glucose value above 18 mmol/L denoted manifest diabetes (MD). The control group of non-diabetic (N) animals was not injected. N, MD WT and N, MD RAGE^-/- females were mated with N males that were WT and RAGE^-/-, respectively. The presence of a vaginal plug denoted gestational day (GD) 0.

At GD18, the pregnant mice were weighed, and blood glucose in tail vein was measured (Freestyle Lite, Abbot); thereafter, the mice were anesthetized by i.p. administration of 600 µl 2.5% Avertin (27). Blood samples were drawn by heart puncture; 5µl 0.5M EDTA was added to the blood and the samples were centrifuged at 3000 rpm for 10 min. The plasma supernatant was collected and stored at -80°C until analysis. After the heart puncture, the animals were killed by cervical dislocation. The abdominal wall was incised and opened and a part of the maternal liver was removed and snap-frozen for 8-iso-PGF_2α analysis. Viable fetuses, resorptions and the corresponding placentas were dissected from each uterine horn and freed from surrounding membranes; then, the fetuses were inspected for external malformations and weighed. The sex of each fetus was determined by visual inspection of the genital area.

Laboratory Procedures

Maternal Metabolite Analysis (Papers I, II and IV)

The concentrations of D-glucose, fructosamine, triglycerides, and cholesterol were measured with a Konelab 30 analyzer (ThermoFisher Scientific, Van- taa, Finland). Standard reagent kits were used for all analyses: for D-glucose (83) and cholesterol (4) analyses the kits were from ThermoFischer Scien- tific (Vantaa, Finland), and for fructosamine (98) the kit was from Horiba ABX (Montpellier, France). The serum concentrations of β-hydroxybutyrate were determined with an enzymatic colorimetric test and reagent kit Liqui- Color Procedure No. 2440 (Stanbio Laboratories, Boerne, Texas) (83) on a Cobas MIRA Multichannel analyzer (Roche Diagnostica, Basel, Switzer- land).

Maternal Serum Amino Acid Analysis (Papers I and II)

After deproteinization of 100 µl serum with 200 µl sulphosalicylic acid, serum concentrations of free amino acids were determined chromatograph- ically with a Biochrom 20 (Biochrom Ltd, Cambridge, UK) and a 4.6x200 mm high resolution PEEK column with Ultrapac 8 resin (Biochrom Ltd,

(31)

8-iso-PGF2α Analysis (Papers I, II and IV)

8-iso-PGF2α concentrations in fetal liver and maternal serum were estimated with the 8-Isoprostane EIA Kit (Cayman Chemical Co, Ann Arbor, MI, USA), according to the manufacturer’s instructions and as described in pre- vious work (119, 184). The methods of either Lowry (106) or Bradford (21) were used to estimate the protein content of the liver samples, with bovine serum albumin as a standard.

Enzyme Activity Analysis (Papers I and II)

For enzyme activity analysis, heart and mandible tissue were homogenized on ice by ultrasound disruption (20 kHz, 60 W for 5s; Vibra Cell; Sonics &

Materials, Danbury, CT, USA) in 140 µl 100 mM Triethanolamine buffer (pH 7.6) and centrifuged at 4°C for 40 min at 13000 rpm.

AR activity was measured with a modified version of the method de- scribed by Wu et al (190): all chemicals were purchased from Sigma-Aldrich Sweden AB (Stockholm, Sweden). The supernatant (80 µl of the heart and 50 µl of the mandible) was added to a reaction mixture, containing 0.4 M ammonium sulfate and 0.2 mM NADPH in a 5 mM sodium phosphate buffer (pH 6.3), to a final volume of 500 µl: this was preincubated at 37°C for 20 min. The reaction was started by adding 50 µl 100 mM DL-glyceraldehyde, followed by 5 min in a spectrophotometer (UVMini 1240, Shimadzu, Kyoto, Japan) at 340 nm. The difference in absorbance over time in the linear part of the reaction (∆A) was used for determining enzyme activity, which was calculated with the equation in Figure 5.

Figure 5. Formula for calculating the enzyme activity of AR.

Gapdh activity was measured with a modified method by Bergmeyer (15), which has been described previously (176). All chemicals were purchased from Sigma-Aldrich. For the assay, 20 µl of the heart supernatant and 50 µl of the mandible supernatant were used and enzyme activity was calculated with the equation in Figure 6.

A

_AR

[C]

Protein in sample