1

Retinoids and Developmental Neurotoxicity In Vivo and In 1

Vitro 2

Prepared for OECD by: 3

Joshua F. Robinson 4

University of California, San Francisco 5

513 Parnassus Ave, Box 0556 6

San Francisco, CA 94143 7

8

2

Abstract 9

Retinoids are critical signaling molecules involved in multiple aspects of mammalian 10

central nervous system (CNS) development. Retinoids bind to RAR/RXR receptors leading 11

to expression of molecules which control cellular processes, (e.g., morphogenesis, 12

patterning, differentiation and proliferation) critical for neurulation, neurogenesis, and 13

CNS maturation. Due to the importance of strict control in these processes, imbalances in 14

retinoid signaling (excessive or absence) can lead to teratogenicity and adverse CNS 15

neurodevelopmental and neurobehavioral outcomes. Studies in human and mammalian 16

models suggest retinoid effects to be highly dependent on several factors, including, 17

compound specificity, dose, exposure duration, timing of exposure, and species tested. 18

Many environmental compounds may disrupt retinoid signaling resulting in developmental 19

neurotoxicity (DNT). In response to the growing need to develop animal-free alternatives, 20

in vitro and in silico approaches have been established to evaluate chemicals in their ability 21

to cause DNT and several retinoids have been evaluated in this context. Here, we provide 22

a review of retinoid exposures and mammalian DNT, and the utilization of alternative 23

approaches to evaluate retinoid effects. 24

3

Table of Contents 25

Retinoids and Developmental Neurotoxicity In Vivo and In Vitro ................................................... 1 26

Abstract .................................................................................................................................................. 2 27

Abbreviations ......................................................................................................................................... 4 28

1. Retinoids and Mammalian Central Nervous System Development .............................................. 5 29

2. Retinoid Exposures and Adverse Outcomes of the CNS in Mammalian Models ........................ 9 30

3. Genetic Defects in RA Signaling Pathways and Adverse CNS Outcomes .................................. 11 31

4. Alternative Assays to Assess RA-induced Developmental Neurotoxicity ................................... 12 32

4.1 Rodent Whole Embryo Culture ................................................................................................... 12 33

4.2 Embryonic Stem Cells (Mouse and Human) ............................................................................... 15 34

4.3 Immortalized Cell Lines .............................................................................................................. 18 35

4.4 Zebrafish ...................................................................................................................................... 18 36

4.5 In Silico Modeling ....................................................................................................................... 19 37

4.6 Toxicogenomic Response to RA in Rodent and In Vitro Neurodevelopmental Models ............. 19 38

4.7 Environmental Chemicals Proposed to Alter RA-Signaling Pathways Linked with DNT .......... 20 39

4.7.1 Triazoles ................................................................................................................................ 20 40

4.7.2 Alcohol .................................................................................................................................. 20 41

4.7.3 Flame retardants .................................................................................................................... 21 42

4.7.4 Isoflurane ............................................................................................................................... 21 43

4.8 Definition of Retinoids and Literature Search ............................................................................. 21 44

5. References ........................................................................................................................................ 23 45

46

Tables 47

Table 1.1. Adverse CNS outcomes associated with genetic defects of RA metabolism and signaling 48

in mouse models. ............................................................................................................................. 7 49

Table 4.1. Toxicogenomic studies assessing effects of RA exposures. ................................................. 20 50

Table 4.2. Family of Retinoid Compounds ........................................................................................... 21 51

52

Figures 53

Figure 1.1. Retinoic acid metabolism. ..................................................................................................... 5 54

Figure 1.2. Retinoic acid and anterior-posterior axis formation. ............................................................. 7 55

Figure 2.1. Teratogenic CNS outcomes and RA exposures in utero in rodent. ...................................... 9 56

Figure 4.1. Rat postimplantation whole embryo culture. ...................................................................... 12 57

Figure 4.2. Retinoic acid induces CNS defects in rat whole embryo culture ........................................ 14 58

Figure 4.3. Human embryonic stem cell neural differentiation and response to retinoic acid. ............. 16 59

Figure 4.4. A proposed retinoic acid–neural tube/axial patterning adverse outcome pathway (RA–60

NTA AOP) framework .................................................................................................................. 19 61

62

4

Abbreviations 63

Tretinoin or all-trans-retinoic acid (RA) 64

Isotretinoin or 13-cis-retinoic acid (13-cis-RA) 65

Alitretinoin or 9-cis-retinoic acid (9-cis-RA) 66

Whole embryo culture (WEC) 67

Human embryonic stem cell (hESC) 68

Mouse embryonic stem cell (mESC) 69

Zebrafish (ZB) 70

Neural tube defect (NTD) 71

Homeobox (HOX) 72

Vitamin A (Retinol) 73

Central nervous system (CNS) 74

Retinoic X receptor (RXR) 75

Retinoic acid receptor (RAR) 76

77

5

1. Retinoids and Mammalian Central Nervous System Development 78

1. Retinoic acid (RA) also known as all-trans-retinoic acid is required for vertebrate 79

CNS development and is the active derivative of retinol (Vitamin A). Mammals are unable 80

to synthesize RA de novo and require intake of vitamin A or other retinoid precursors which 81

can be transformed into RA through food sources [1]. RA and metabolites act as 82

transcriptional activating ligands by binding to nuclear receptors known as retinoic acid 83

receptors (RARs) α, β, and γ [2-4] and retinoid X receptors (RXRs) α, β, and γ [5]. 84

RAR/RXRs are expressed in the embryo, in a regional- and age-dependent manner. In 85

general, RARα, RXRα, and RXRβ are ubiquitously expressed, while RARβ, RARγ, and 86

RXRγ are expressed in a more restricted profile [6]. The primary endogenous ligand for 87

RARs is RA, while 9-cis-RA readily binds to both RARs and RXRs. RARs form 88

heterodimer complexes with RXRs and bind to motifs known as RA response elements 89

(RARE). In contrast to RARs, RXRs are not specific to retinoids or binding of RARs and 90

form homodimers and heterodimers with other nuclear receptors. The activity and 91

specificity of retinoids is tightly regulated by a combination of RA metabolizing enzymes 92

(e.g., cytochrome p450 26 subfamily members, retinaldehyde dehydrogenases) and RA 93

binding proteins (CRAPB family) which facilitate RA transfer to the nucleus (Figure 1.1). 94

Figure 1.1. Retinoic acid metabolism. 95

96

6

Figure 1.1: The availability of retinoids is tightly controlled in the mammalian CNS. Mammals are unable to 97 synthesize retinoids de novo and require intake of vitamin A or other derivatives (β-carotenes) from food 98 sources. Vitamin A (retinol) is converted to retinaldehyde by two classes of oxidizing enzymes (alcohol 99 dehydrogenases and retinol dehydrogenases). Retinaldehyde is further oxidized to form all-trans-retinoic acid 100 (RA) by aldehyde dehydrogenases (e.g., RALDH2). The cytochrome p450 26 subfamily enzymes are critical 101 in regulating RA levels in the embryo and catalyze reactions to reduce RA bioavailability by converting RA to 102 4-OH-RA and further oxidized, less active metabolites. 103

2. The coordinated regulation of retinoids and molecular partners is necessary for 104

multiple aspects of mammalian central nervous system (CNS) development, including: 105

anterior/posterior axis formation, mesodermal segmentation, caudal elongation, 106

neurogenesis, and brain maturation [7]. As early as the gastrula stage of embryogenesis, 107

RA is biosynthesized and dispersed in a concentration gradient along the anterior-posterior 108

axis. RA abundance is regionally and tightly controlled by RA metabolizing (e.g., 109

CYP26A1, highly expressed in the anterior region) and biosynthesis (e.g., RALDH2, 110

expressed in the mesoderm) enzymes as well as signaling molecules (e.g., Fgf, Wnt, 111

expressed in the posterior region) which inhibit activity of CYP26 (Figure 1.2). During 112

neurulation, RA is synthesized in the paraxial mesoderm via RALDH2 and diffuses to 113

nearby tissues regulating expression of molecules critical for anterior-posterior patterning 114

(e.g., HOX genes), somitogenesis and neurogenesis (e.g., Cdx1, Dbx1, Ngn2, Hnf1b, 115

Pax6). The balance and tissue-specific timing of RA bioavailability is critical. Genetic 116

manipulation of key enzymes (CYP26A1, RALDH2) and specific RAR/RXRs in rodent 117

causes a variety of CNS defects (Table 1.1). While the mechanisms remain less understood, 118

RA also plays key roles in neuronal differentiation and specification, influencing postnatal 119

neurobehavioral outcomes. 120

7

Figure 1.2. Retinoic acid and anterior-posterior axis formation. 121

.122

123

Figure 1.2: During the early gastrula stage, retinoic acid (RA) is regionally restricted and induces posterior 124 embryonic growth. CYP26A1 is exclusively expressed in anterior tissues leading to the breakdown of present 125 RA and inactivation of RAR-signaling. Growth signaling factors (Fgf, Wnt) which promote a posterior 126 phenotype inhibit CYP26A1 activity and RALDH2 biosynthesis activity contribute to a gradient of RA 127 availability in the intermediate zone, enabling RAR/RXR activation of HOX genes and other molecules which 128 promote expansion of the posterior domain. Modified from [79, 80]. 129

Table 1.1. Adverse CNS outcomes associated with genetic defects of RA metabolism and 130 signaling in mouse models. 131

Gene CNS defects

Raldh2 Defects of posterior hindbrain (abnormal patterning) and

forebrain

Cyp26a1 Caudal region (severely truncated with spina bifida), anterior

hindbrain (abnormal patterning)

Rara; Rarb Hindbrain (abnormal patterning)

Rara; Rarg Hindbrain (abnormal patterning)

Rxra; Rxrb Caudal region (truncated), nasal region,

posterior

Raldh2; Raldh3 Forebrain, nasal region, optic vesicle

Cyp26a1; Cyp26c1 Forebrain, midbrain, branchial arches

(reduced), hindbrain (expanded), cranial

neural crest (deficient)

132

3. Despite its critical importance for embryonic growth and development, Vitamin A 133

(retinol) is a suspected human teratogen whose exposure in both deficiency and in excess 134

8

can result in fetal malformations and adverse neurodevelopmental outcomes. 135

Recommended daily allowance (RDA) guidelines for Vitamin A supplementation are 136

commonly provided as micrograms of retinol activity equivalents (RAE) to account for the 137

differing bioactivities of retinol and other Vitamin A precursors (e.g., β -carotenoids). 138

Women aged 19 years and older are recommended to consume at least 700 µg RAE/day, 139

750 µg RAE/day during pregnancy and 1,200 µg RAE/day during lactation [8]. Currently, 140

the recommended tolerable upper limit is set at 3,000 µg RAE (10,000 IU)/day. In general, 141

consumption of retinoids during pregnancy in line with these guidelines are believed to be 142

generally safe and not significantly associated with teratogenicity [9, 10]. However, at 143

higher exposure levels (>10,000 IU/day), the health risks remain undefined. In a case-144

control study, women who consumed >10,000 IU versus <5,000 IU of supplemented 145

vitamin A per day during pregnancy were found to have 4.8 times higher risk of having an 146

offspring with a cranial-neural crest defect [11]. Additionally, a significant association 147

between adverse pregnancy outcomes and Vitamin A concentrations in amniotic fluid 148

during mid-pregnancy (second trimester) was revealed in a prospective cohort study 149

(n=106; [12]). Women whose pregnancy resulted in a neural tube defect (NTD) had an 150

average concentration of vitamin A in amniotic fluid that was 1.5 times greater than 151

mothers whose pregnancies were normal. In contrast to these positive associations, the 152

prevalence of pregnancies resulting in offspring with NTDs was not significantly different 153

between women who consumed >8,000 or >10,000 IU per day of preformed Vitamin A 154

during the periconceptional period versus women who did not [9]. Furthermore, retinol 155

levels in maternal serum during early pregnancy (1st trimester) was not associated with 156

NTD risk in a case-control study examining NTD-affected (n=89) and normal pregnancies 157

(n=178; [13]). While cumulative evidence in humans suggests excessive Vitamin A 158

exposures are teratogenic and associated with CNS malformations, defining an upper limit 159

in vitamin A consumption during pregnancy remains unclear due inconsistencies in 160

reported risk-associations. Gaps in knowledge that remain are due to many factors, 161

including the analysis of small samples sizes, i.e., limited cases with CNS defects, and the 162

reliance of case-control study designs which commonly include recall bias due to the 163

retrospective nature of the study. 164

4. In addition to dietary supplementation, synthetic retinol derivatives such as 165

Isotretinoin and Tretinoin are widely-used for treatments against severe acne. In case-166

studies and retrospective and prospective cohorts, exposures (dermal or oral) to these 167

substances during early pregnancy are associated with congenital fetal anomalies, 168

including, CNS malformations such as microcephaly, mid/hindbrain malformations (e.g., 169

absent cerebellar vermis, cerebellar vermis hypoplasia or brainstem abnormalities) and 170

spontaneous abortion. An examination of 154 Isotretinoin-exposed pregnancies, (118 171

observed retrospectively and 36 observed prospectively) was associated with 21 malformed 172

infants (18 of which developed anomalies of the CNS). All malformations were associated 173

with mothers who were exposed to Isotretinoin on or before the 28th day of gestation [14], 174

suggesting a severe level of risk to retinoid exposures during vulnerable periods in early 175

embryogenesis and CNS development. 176

9

2. Retinoid Exposures and Adverse Outcomes of the CNS in Mammalian 177

Models 178

5. In animal models, including mouse [15], rat [16], hamster [17], and monkey [18], 179

excessive exposures to retinoids during prenatal life results in CNS malformations and 180

altered neurodevelopmental outcomes. Common defects associated with excess retinoid 181

exposures include: anterior (exencephaly or anencephaly) and posterior (spina bifida) 182

neural tube defects, microcephaly, irregular somtiogenesis, delayed or reduced caudal 183

elongation, vascular damage, notochord defects, and irregularities in the neural folds. On 184

the cellular level, morphological perturbations are a result of abnormal cellular changes in 185

proliferation, differentiation, viability (apoptosis), and migration driven by excessive RA 186

exposure, altered expression of RA metabolic factors, and the activation of RA-dependent 187

gene expression. Severity of outcomes may depend on several factors such as dose, time, 188

exposure route, timing of exposure, and genetic background (see example of RA-induced 189

teratogenicity in rodent in vivo studies, Figure 2.1). RA is the most active and potent 190

retinoid inducing neuro-teratogenicity; other retinoids such as retinol and 13-cis RA, at 191

excessive levels, also produce similar pathological phenotypes. In general, the most 192

sensitive windows to retinoid-induced gross CNS malformations appears when retinoid 193

exposures occur during neurulation and early organogenesis [19, 20]. Later periods in CNS 194

development may be vulnerable windows to excess or deficits in RA, resulting in altered 195

neurodevelopmental and neurobehavioral outcomes, which manifest later in life. For 196

example, RA-exposed juvenile rats display changes in basal serum corticosterone 197

concentrations, increased thickness of the adrenal cortex, and anxiety-like behaviors [21]; 198

while gestational deficiency in Vitamin A is linked with impaired learning and memory 199

[22] as well as autistic-like behaviors due to Rarb-CD38-oxytocin suppression in the 200

hypothalamus [23]. RA is critical for vertebrate development and imbalances during 201

specific windows of development are linked with CNS defects and more-subtle neural 202

alterations that underlie neurobehavioral adverse outcomes. 203

Figure 2.1. Teratogenic CNS outcomes and RA exposures in utero in rodent. 204

205

10

Excess RA induces developmental neurotoxicty in rats. Severity and specificity of effects is dependent on 206 administration of exposure (oral, subcutaneous, intraperitoneal), exposure duration (acute vs. chronic) and 207 gestational timing of exposure. Shading of circles representative of reported severity of toxicity induced by 208 exposure: severe (>80% anomalies); high (>50-80% anomalies); moderate (30-50% anomalies); low (<30% 209 anomalies); or not significant. Toxic RA exposures during neurulation/early rat organogenesis resulted in 210 anterior/posterior neural defects. In vivo studies are referenced [81-85] 211

11

3. Genetic Defects in RA Signaling Pathways and Adverse CNS Outcomes 212

6. Due to the strict control of RA and related signaling pathways during embryonic 213

development, genetic modulation of RA metabolic enzymes, RA carrier proteins, and RA 214

receptors are linked with perturbations in patterning, neural tube closure, and CNS 215

morphology. Adverse CNS outcomes are dependent on the normal expression profile of 216

RA-signaling mediators. For example, CYP26A1 is predominately expressed in the 217

anterior region of the embryo and is responsible for inhibiting RA-activity. CYP26A1 -/- 218

embryos exhibit increased exencephaly and spina bifida [24, 25]. Raldh2 -/- mouse 219

embryos display reduced capability to synthesize RA and also exhibit increased posterior 220

NTDs [7]. The CYP26A1 -/- phenotype is rescued by also eliminating embryonic 221

expression of RARγ, which is normally expressed within the folds of the neural tube prior 222

to closure [26]. RARγ is expressed prior to closure of the neural tube, while RARβ and 223

RARα are expressed as the neural tube folds adjoin [27-29]. RARα/γ double null mouse 224

mutants [30] have significantly higher levels of exencephaly, while RARγ–/– are less 225

susceptible to anterior or posterior NTDs [31, 32]. 226

7. In humans, case-control studies have discovered genetic variants in enzymes 227

controlling RA synthesis and degradation as correlates of NTDs. Deak et al. [33] identified 228

a significant association between genetic polymorphisms in ALDH1A2 and spina bifida. 229

Rat et al. (2006) found a functional genetic deletion in CYP26A1 to be linked with spina 230

bifida in an Italian/Caucasian population. Li et al. [34] discovered variants in exonic and 231

upstream regions of RA binding motifs, i.e., RA receptor elements (RARE) of CYP26A1, 232

CRABP1, ALDH1A2, to be associated with NTDs. In summary, these studies suggest 233

phenotypes observed in rodents associated with genetic deletion enzymes controlling RA 234

metabolism are also relevant for human risk to CNS defects. 235

12

4. Alternative Assays to Assess RA-induced Developmental Neurotoxicity 236

8. Thousands of industrial chemicals registered for use have not been thoroughly 237

examined for their potential to cause developmental neurotoxicity (DNT), and many of 238

these compounds are found in humans during vulnerable periods of development (e.g., 239

gestation and infancy). Traditionally, chemicals must be evaluated for DNT using 240

standardized whole animal rodent toxicological tests. While informative, these assessments 241

are costly and time-consuming. Furthermore, efforts to use animal-free approaches are 242

underway, thus, the training of scientists in the areas of anatomy, teratology, and traditional 243

animal toxicology is becoming more and more scarce. In vitro and in silico models are 244

quickly emerging as viable replacements for traditional in vivo animal studies for the 245

purposes of assessing DNT. Sponsored research by the United States Environmental 246

Protection Agency, National Institutes of Environmental Sciences (e.g., Tox21/ToxCast 247

[35]) and the European Commission [36] has promoted efforts towards rapid development 248

of a potential animal-free testing framework to screen industrial chemicals for toxicological 249

assessments. Due to the biological complexity of human neurodevelopment, multiple 250

models are needed to thoroughly predict DNT. Here, we overview models currently being 251

utilized to evaluate retinoids in the context of causing DNT. 252

4.1 Rodent Whole Embryo Culture 253

9. The rodent whole embryo culture (WEC) model is an established tool to study the 254

influence of both environmental and genetic factors, on early molecular and cellular 255

changes of the neuroepithelium and surrounding embryonic tissues (e.g. paraxial 256

mesoderm, notochord, and surface ectoderm) which support formation and closure of the 257

neural tube. As compared to traditional rodent toxicological studies, WEC involves a 258

reduction in animals (~50% less) and cost, shortened experimental time, and control over 259

dosage and compound specificity. Due to these attributes, the WEC has been proposed as 260

a predictive assay to determine teratogenic compounds of concern for toxicological testing. 261

In brief, rat or mouse embryos are isolated and cultured during the initial stages of neural 262

tube formation (gestational day (GD) 8, mouse; GD 10, rat) and successfully progress 263

through neurulation and early organogenesis, parallel to the transitions which occur in vivo 264

(Figure 4.1; [37]). Morphological development of the neural tube (e.g., cranial or caudal 265

NT closure; primitive fore-, mid-, hindbrain, and spinal cord formation) can be evaluated 266

for abnormalities or growth delays in the context of compound exposure. 267

Figure 4.1. Rat postimplantation whole embryo culture. 268

269

13

(A) Cultured embryos undergo neural tube closure (arrows) and the initial stages of CNS morphogenesis in a 270 similar manner to in vivo embryogenesis. 271

272

273

(B) Neurogenesis gene expression patterning in WEC as compared to rat (in vivo) and human (in vivo) embryos 274 during neurulation and early organogenesis (Adapted from [37]). 275

10. The WEC has proven to be a valuable model to evaluate teratogenic potency of 276

retinoids, timing of exposure, and mechanisms underlying teratogenicity. Morris and Steele 277

[38] demonstrated the use of the WEC model to profile the teratogenic effects of excess 278

retinol and RA exposures during early organogenesis. These initial studies validated in vivo 279

observations that elevated exposures to retinoids during neurulation results in anterior and 280

posterior defects of the neural tube as well as malformations of the palate and limb. 281

Subsequent studies by Steele et al. [39] and Klug et al. [40] using the WEC demonstrated 282

the metabolite, 13-cis RA, to also cause CNS defects and impair embryonic growth. These 283

studies demonstrate the ability of the WEC to evaluate retinoid exposures and relevant 284

teratogenic endpoints to the in vivo condition. 285

11. Recent studies examining RA effects in WEC have incorporated transcriptomic 286

approaches to identify global changes in gene expression which may underlie and precede 287

morphological alterations. Luijten et al. [41] utilized transcriptomic profiling to identify 288

early responses to teratogenic levels of RA (0.5ug/mL) in single cultured rat embryos 289

initially exposed at 2-4 somites. After 4h exposure, RA affected 260 genes after a 4h 290

exposure duration, including genes involved in embryonic patterning and CNS 291

development such as Gbx2 and Otx2 as well as genes linked with RA metabolism 292

(Cyp26a1, Cyp26b1, Dhrs3) and RA gene activation (Rarb). Gene expression changes due 293

to RA were similar across embryos of 2-4 somites or between embryos cultured in rat serum 294

vs. a rat-bovine serum mixture. In a study by Robinson et al. [42], RA effects on 295

morphological and transcriptomic levels were compared between the rat postimplantation 296

WEC model and embryos in vivo undergoing neurulation (Figure 4.2). In both models, RA 297

(0.5 μg/ml or 50mg/kg BW oral gavage) caused teratogenicity, including growth reduction 298

(size, somites) and defects of the neural tube. On the transcriptomic level, RA response was 299

evaluated across six timepoints (GD10 + 2-48h) in both models. In total, RA perturbed 300

expression of 845 genes in a time-dependent manner. Consistent with Luitjen et al. [41] 301

perturbations in expression were enriched for genes involved in CNS and embryonic 302

development as well as neuronal differentiation. More than 50% of identified DE genes in 303

either model overlapped with the other system. Furthermore, similar patterns in gene 304

expression due to RA exposure was observed in terms of directionality and significance 305

between embryos in vivo and WEC. Differences in the timing of perturbations to select 306

genes related to RA metabolism (e.g., Cyp26a1, Cyp26b1, Dhrs3) and embryonic 307

patterning (e.g., Pbx1, Pbx2, Gli1, Dlx1, Hoxa9, Hoxa10, Hoxc10, Hoxd10) associated 308

with significant differences in the propensity of NT defects to be anterior (WEC) vs. 309

14

posterior (in vivo), suggesting toxicokinetics, and developmental timing of RA exposure to 310

be critical factors in teratogenic outcomes and RA response. 311

Figure 4.2. Retinoic acid induces CNS defects in rat whole embryo culture 312

313 (A) Examples of CNS anterior or posterior defects caused by excessive retinoic acid (RA) exposure in WEC and in utero. 314

315 316

317 (B) Venn diagram comparison of genes identified to be differentially expressed due to due to RA exposure in WEC vs. rat (in 318

vivo) across six timepoints. (C) Cross-scatter plot displaying correlation of RA-response at 4h in WEC vs. rat (in vivo) 319

15

320 (D) Relative expression of 845 genes significantly altered by RA exposure in WEC and/or rat (in vivo) across six timepoints 321

[42]. Comparisons are shown with previous transcriptomic datasets evaluating RA-response at 4h in WEC [41] 322

12. Overall, the WEC has proven to be an acceptable model to examine RA-induced 323

defects during neurulation and early CNS organogenesis. Excessive RA exposure leads to 324

molecular changes that can be evaluated in the context of embryo morphogenesis. Despite 325

the usage of animals, the WEC may be considered as a valid alternative to test retinoids 326

and other environmental compounds in their effects on RA signaling in the context of early 327

CNS development. 328

4.2 Embryonic Stem Cells (Mouse and Human) 329

13. ESCs derived from the inner cell mass of the mammalian blastocyst possess an 330

unlimited potential for self-renewal and the capacity to differentiate into multiple somatic 331

cell types, including neurons, oligodendrocytes, glial cells and astrocytes. Utilizing mouse 332

or human ESCs, investigators have demonstrated the unique characteristics of these models 333

to capture early in vivo developmental events in vitro, including biological processes 334

critical for neurogenesis and neurulation (e.g., self-renewal, proliferation, differentiation, 335

morphological patterning; see example, Figure 4.3). ESC model systems have been used 336

to screen compounds for their potential to induce DNT. While both systems may be 337

applicable for toxicological screening, hESC models may provide advantages over the 338

murine system due to interspecies extrapolations [43]. While the predictive value of ESCs 339

for DNT screening has yet to be defined, early investigations suggest that these model 340

systems can be used to study a wide-range of risk factors on multiple aspects of 341

neurogenesis [44, 45], including retinoids and other environmental chemicals that cause 342

neurotoxicity by perturbing RA-signaling pathways. 343

16

Figure 4.3. Human embryonic stem cell neural differentiation and response to retinoic acid. 344

345

(A) Example schematic of transitions of human neurogenesis and hESC neurodevelopment ([86]). 346

347

B) Genes commonly differentially expressed during hESC neuronal differentiation and human embryos 348 undergoing neurulation, categorized as the NeuroDevelopmental Biomarker (NDB) gene set. 349

350

(C) Over 50% of NDB genes were significantly altered by RA in a hESC-neural rosette model. 351

352

17

353

(D) Heatmap displaying change in expression with concentrations of RA. (E) Absolute average fold change in expression of 354 NDB genes and significant/non-significant subsets with RA exposure in hESCs [51]. 355

14. Due to its well-established role in promoting neuronal differentiation [46, 47], RA 356

is commonly added to ESC cultures to induce neural differentiation. Utilizing a 357

transcriptomic approach at four time points during mESC-embryoid body differentiation, 358

Akanuma et al. (2012; [48]) demonstrated the stage-dependent sensitivity to RA at two 359

concentrations (0.01 or 0.1uM). These analyses provided information regarding gene 360

networks influenced by RA exposure at 0, 2, 8, and 36 days of mESC differentiation in 361

parallel with cell morphological changes (increased growth and neuronal differentiation), 362

and identified particular targets (e.g., Gfap, Gbx2) that may amplify RA-signaling and 363

neuronal development. In a protocol developed by Theunissen et al. [49], mESCs 364

differentiated towards a neuronal fate with the addition of RA, leading to a significant 365

downregulation of Cyp26a1 and upregulation of Aldh1a2 as well as dramatic changes in 366

expression of patterning molecules, Hoxa1, Hoxa2, and a master transcription factor for 367

neuroectodermal differentiation, Pax6. 368

15. The influence of excessive RA or differential potencies of RA compounds has also 369

been examined in hESC neurodevelopmental models. For example, the hESC-based neural 370

rosette (NR) model has been proposed as a system to evaluate the ability of chemicals to 371

disrupt early CNS morphogenesis and specifically the apical neuroepithelial organization 372

of the neural tube. Colleoni et al. [50] demonstrated the use of the hESC-NR model to 373

assess DNT by examining morphological and molecular features. In a concentration-374

dependent manner, RA altered morphology, e.g., disorganized NR patterning, induced 375

cytotoxicity, and perturbed expression of molecules critical for CNS patterning (HOXA1, 376

HOXA3, HOXB1, HOXB4) and neural differentiation (FOXA2, FOXC1, OTX2, PAX7). 377

These results suggest that altered features due to RA (in vivo) may be evaluated utilizing 378

the hESC-NR model. 379

16. Aiming to integrate datasets of hESC neural differentiation and human 380

neurodevelopment (in vivo) to define neurodevelopmental biomarkers for neurotoxicity 381

studies, Robinson et al. [51], performed a meta-analysis of publicly available 382

transcriptomic datasets (Figure 4.3 B-E). First, genes differentially expressed commonly 383

across five independent studies examining hESC neural differentiation were identified. 384

These analyses suggested a core-group of genes to be commonly changing as cells 385

transformed from pluripotent hESCs to neurons. Secondly, genes identified in vitro were 386

matched with human embryo transcriptomic datasets evaluating three unique 387

developmental time periods: blastocyst formation, neurulation, and organogenesis. Based 388

on expression profiles, hESC model of neurodevelopment significantly correlated with 389

human embryos undergoing neurulation between the 3rd and 4th week of pregnancy. Genes 390

in common between the two systems were termed the NeuroDevelopmental Biomarker 391

18

(NDB) gene set which contained 304 genes. NDB genes were significantly enriched for 392

biological processes involved in neurogenesis and CNS development. As a proof-of-393

principle to demonstrate the utility of the NDB biomarker set for toxicological studies, RA 394

was shown to significantly alter over 50% of the NDB set in a concentration-dependent 395

manner in a hESC neurodevelopmental model. 396

17. ESCs, isolated from either human or rodent, may provide a capable screening 397

system for retinoids and other environmental chemicals. RA is commonly used as a 398

supplement to induce ESC neural differentiation, and excessive amounts lead to 399

perturbations in neuronal development, including alterations in molecules which regulate 400

neurogenesis and neuronal development and patterning in mammals. 401

4.3 Immortalized Cell Lines 402

18. Immortalized mammalian cell lines may be important model systems to evaluate 403

retinoid-induced effects and DNT. The P19 mouse embryonic carcinoma stem cell line 404

originally derived from an teratoembryocarcinoma can be used to assess chemical effects 405

and mechanisms of action related to neurogenesis and gliogenesis, including, morphology, 406

proliferation, differentiation, neuronal specification as well as the functionality of ion 407

channels and metabotropic receptors [52]. Like ESCs, retinoic acid is commonly used to 408

induce neuroectodermal differentiation and P19 cells have been utilized to cross-evaluate 409

the sensitivity of retinoids, RA, 13-cis-RA and 9-cis RA, in inducing neuronal 410

differentiation [53]. The NE-4C cell line was established from the cerebral vesicles of 9-411

day-old mouse embryos lacking the functional p53 genes. In NE-4C RA and RA-412

precursors, retinal and retinyl acetate, induce neural differentiation and alter expression of 413

genes responsible for RA metabolism, synthesis, and signaling [54]. 414

4.4 Zebrafish 415

19. The zebrafish (ZB) has also been proposed as a high-throughput screening model 416

to assess chemical-induced DNT [55, 56]. The ZB embryo develops rapidly and the effects 417

of compounds on CNS morphology can be externally evaluated within a 72h period 418

(Figure to be added). Initial reports suggest that the ZB model is predictive in identifying 419

chemical toxicants. Similar to mammals, RA signaling is critical for patterning of the 420

anterior-posterior axis, and development of the hindbrain and spinal cord [57-59], and the 421

ZB is also highly sensitive to RA-induced teratogenicity [60]. Although, the ZB may 422

diverge from mammals in the need of RA for body axis extension [61]. 423

20. Holder and Hill [62] demonstrated RA to cause teratogenic effects on CNS and tail 424

development in a concentration-dependent manner. Follow up studies by Herrmann et al 425

[63] demonstrated the utilization of the ZB model to evaluate cross-potency of nine retinoid 426

compounds. Similar to differential potencies described in rodents, RA was identified to be 427

most potent followed by 9-cis RA, 13-cis RA, tretinoin, 3,4-didehydro and 4-oxo, in 428

causing malformations, including defects of the CNS (anencephaly, microcephaly, 429

anophyalmy). Furthermore, much like mammals, conserved elements involved in 430

regulating RA embryonic distribution, including the RA-metabolizing enzymes (e.g., 431

CYP26A1 ([64, 65]), RALDH2 [66], DHRS3A [67], and RA binding proteins (CRABP2A; 432

[68]) are suspected to control RA signaling as well as prospective HOX signaling and the 433

prospective organization of the hindbrain/spinal cord in ZB. 434

21. In ZB, excessive RA leads to disturbances in pathways previously identified in 435

mammals, including alterations in CYP metabolizing enzymes; RAR/RXR-mediated 436

19

transcription, and downstream patterning signaling pathways. Global transciptomic 437

analyses suggest hundreds of genes to be disrupted in the zebrafish due to exposures during 438

early stages of CNS formation. Genetic manipulations of RA metabolism and signaling 439

pathway members (e.g., aldh1a2, cyp26a1, cyp261, raldh2; as reviewed [69]) are also 440

associated with CNS malformations, providing additional evidence of the relevancy of RA 441

signaling in the ZB embryo. 442

22. While high concentrations to RAs during early periods in brain formation cause 443

gross malformations, lower exposures to RA during embryogenesis are associated with 444

neurobehavioral changes (e.g., hyperactivity) in juveniles [70]. 445

23. In general, these studies suggest that the zebrafish may be used as a surrogate to 446

rodent studies and be an appropriate model to evaluate RA-effects on CNS development, 447

providing advantages over cellular models due to the ability to assess the CNS 448

morphogenesis and conserved neurodevelopmental pathways across multiple life-stages. 449

4.5 In Silico Modeling 450

[Text for this section pending] 451

Figure 4.4. A proposed retinoic acid–neural tube/axial patterning adverse outcome pathway 452 (RA–NTA AOP) framework 453

454

(A) Adverse outcome pathway framework for neural tube and axial defects due to altered retinoic acid levels 455 during embryogenesis [71]. 456

Prediction of in vivo developmental toxicity of all-trans-retinoic acid based on in 457

vitro toxicity data and in silico physiologically based kinetic modeling [72] 458

4.6 Toxicogenomic Response to RA in Rodent and In Vitro Neurodevelopmental 459

Models 460

24. The recent use of genomic approaches, including transcriptomic, proteomic, and 461

methylomic profiling, has greatly increased our ability to identify environmental targets in 462

association with DNT. To date, the effects of retinoids have been profiled in diverse 463

20

relevant models, including rat, WEC (rat), mESC, hESC, and zebrafish (see examples, 464

Table 4.1). In general, these studies have demonstrated that excess RA disrupts expression 465

of genes involved in RA metabolism, embryonic patterning and neurodevelopmental 466

signaling pathways. Complementary to studies utilizing genetic modulation (e.g., 467

transgenic, lentiviral, CRISPR/CAS9) in animal model systems, proposed mechanisms of 468

RA and biomarkers of response are increasingly being defined. 469

Table 4.1. Toxicogenomic studies assessing effects of RA exposures. 470

NCBI/NCI (Geodata) and EMBL/EBI (Arraytrack) data repositories were searched for relevant datasets 471 assessing retinoic acid exposure in vitro or in vivo.. 472

. 473

Geodata Species Model Design

GSE104173 Homo sapiens hESC (H1, H9, HUES1);

neural differentiation

model

Time-dependent response

EMEXP3577 Homo sapiens hESC (HUES1); neural

rosette model

Concentration-dependent response (4

total)

GSE1596 Mus musculus Neuroblastoma (Neuro2a) Time-dependent response

GSE4075 Mus musculus mESC; embryoid body Time-dependent response in embryoid

bodies

GSE28834 Mus musculus in vivo Spinal cord of GD18 fetuses

GSE33195 Rattus norvegicus WEC, in Vivo RA exposed embryos (GD10); Time-

dependent response to teratogenic dose of

RA (6 timepoints)

GSE135781 Danio rerio Zb Vagus motor neurons at 38 hours post

fertilization

474

4.7 Environmental Chemicals Proposed to Alter RA-Signaling Pathways Linked with 475

DNT 476

4.7.1 Triazoles 477

25. Exposures to triazole‒antifungal agents‒cause CNS malformation similar to RA in 478

rodent WEC and are suspected to alter retinoic acid signaling pathways. In a concentration- 479

and time-dependent manner, flusilazole induces anterior/posterior defects of the neural 480

tube, compromises CNS morphology, and alters expression of RA metabolic enzymes 481

(Cyp26a1, Cyp26b1, Dhrs3) and signaling molecules critical for hindbrain/spinal cord 482

development (e.g,. Gbx2, Cdx1; [73]). Comparisons across flusilazole, cyproconazole and 483

triadimefon suggest that these agent may impact RA-signaling in a similar manner, and 484

specific altered targets, i.e., Cyp26a1, Dhrs3, are conserved across WEC, mESC, and Zb, 485

model systems [74]. 486

[**could add figure here comparing gene expression changes between triazoles and RA in 487

rat WEC] 488

4.7.2 Alcohol 489

26. In Zb, alcohol disrupts RA signaling leading to defects of the mid/hind brain 490

boundary and endogenous RA can rescue these effects [75]. 491

21

4.7.3 Flame retardants 492

27. In Zb embryos, acute exposure to DE-71, a common flame retardant mixture of 493

polybrominated diphenyl ethers (PBDEs), reduces retinol and RA levels, and upregulates 494

transcription of raldh2, crabp1a, crabp2a, and raraa, suggesting a disruption of RA 495

homeostasis. Alterations are in association with defects in eye development of fish larvae 496

[76] 497

4.7.4 Isoflurane 498

28. Liu et al. [77] observed isoflurane, a common anesthetic, to impair mouse fetal 499

growth and development (in vivo, exposed pre-neurulation) and alter mESC self-renewal 500

(in vitro) and RA-signaling pathways. In vitro, Vitamin A supplementation was observed 501

to attenuate effects of isofluorane, suggesting toxicity to occurring through RAR-mediated 502

pathways. 503

4.8 Definition of Retinoids and Literature Search 504

29. In this review, we compiled literature related to retinoid compounds and 505

environmental chemicals suspected to disrupt retinoid acid signaling in the context of 506

developmental neurotoxicity (DNT). We defined members of the retinoid family, which 507

included first, second, and third generation retinoid molecules (Table 4.2). We identified 508

human epidemiological and in vivo experimental studies which examined the effects of 509

retinoids on pre- and/or postnatal CNS development using the NCBI PubMed database 510

(July of 2019). We applied the following search terms (see below) in the context of human, 511

monkey, mouse, rat, rabbit, hamster, in vitro, or in silico model systems. 512

(“Retinoid*” OR “Retinoic Acid” OR "Retinol” OR “Vitamin A” OR “Tretinoin” 513

OR “all-trans-retinoic acid” OR “isotretinoin” OR “13-cis-retinoic acid” OR 514

“alitretinoin” OR “9-cis-retinoic acid” OR “Etretinate” OR “Acitretin” OR 515

“Adapalene” OR “bexarotene” OR “Tazarotene") 516

AND (“*brain*” OR “central nervous system” OR “CNS” OR “neural tube” OR 517

“spinal cord” OR “spina bifida”) 518

AND (“gestation” OR “pregnancy” OR “infant” OR “infancy” OR “postnatal” OR 519

“neonate” OR "*natal" OR “perinatal” OR “postnatal” OR “prenatal” OR 520

“neonatal” OR “neurulation”) 521

Table 4.2. Family of Retinoid Compounds 522

Name Generation

Retinol (Vitamin A) 1st

Tretinoin (all-trans-retinoic acid) 1st

Isotretinoin (13-cis-retinoic acid) 1st

Alitretinoin (9-cis-retinoic acid) 1st

Etretinate 2nd

Acitretin 2nd

Adapalene 3rd

Bexarotene 3rd

Tazarotene 3rd

523

22

30. We conducted single-screening of title and abstracts followed by full-text review 524

of all articles retrieved from PubMed. We reviewed and identified relevant human and in 525

vivo studies reporting original data that examined retinoid compound exposures in the 526

context of CNS developmental outcomes. 527

23

5. References 528

1. Marceau, G., D. Gallot, D. Lemery, and V. Sapin, Metabolism of retinol during 529

mammalian placental and embryonic development. Vitam Horm, 2007. 75: p. 97-530

115. 531

2. Brand, N., M. Petkovich, A. Krust, P. Chambon, H. de The, A. Marchio, P. Tiollais, 532

and A. Dejean, Identification of a second human retinoic acid receptor. Nature, 533

1988. 332(6167): p. 850-3. 534

3. Krust, A., P. Kastner, M. Petkovich, A. Zelent, and P. Chambon, A third human 535

retinoic acid receptor, hRAR-gamma. Proc Natl Acad Sci U S A, 1989. 86(14): p. 536

5310-4. PMC297611 537

4. Petkovich, M., N.J. Brand, A. Krust, and P. Chambon, A human retinoic acid 538

receptor which belongs to the family of nuclear receptors. Nature, 1987. 330(6147): 539

p. 444-50. 540

5. Mangelsdorf, D.J., Vitamin A receptors. Nutr Rev, 1994. 52(2 Pt 2): p. S32-44. 541

6. Dolle, P., Developmental expression of retinoic acid receptors (RARs). Nucl Recept 542

Signal, 2009. 7: p. e006. PMC2686085 543

7. Niederreither, K., V. Subbarayan, P. Dolle, and P. Chambon, Embryonic retinoic 544

acid synthesis is essential for early mouse post-implantation development. Nat 545

Genet, 1999. 21(4): p. 444-8. 546

8. Medicine, I.o., Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, 547

Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, 548

Silicon, Vanadium, and Zinc. 2001, The National Academies Press: Washington, 549

DC. 550

9. Mills, J.L., J.L. Simpson, G.C. Cunningham, M.R. Conley, and G.G. Rhoads, 551

Vitamin A and birth defects. Am J Obstet Gynecol, 1997. 177(1): p. 31-6. 552

10. Shaw, G.M., E.M. Velie, D. Schaffer, and E.J. Lammer, Periconceptional intake of 553

vitamin A among women and risk of neural tube defect-affected pregnancies. 554

Teratology, 1997. 55(2): p. 132-3. 555

11. Rothman, K.J., L.L. Moore, M.R. Singer, U.S. Nguyen, S. Mannino, and A. 556

Milunsky, Teratogenicity of high vitamin A intake. N Engl J Med, 1995. 333(21): 557

p. 1369-73. 558

12. Parkinson, C.E., J.C. Tan, and I. Gal, Vitamin A concentration in amniotic fluid 559

and maternal serum related to neural-tube defects. Br J Obstet Gynaecol, 1982. 560

89(11): p. 935-9. 561

13. Mills, J.L., J. Tuomilehto, K.F. Yu, N. Colman, W.S. Blaner, P. Koskela, W.E. 562

Rundle, M. Forman, L. Toivanen, and G.G. Rhoads, Maternal vitamin levels during 563

pregnancies producing infants with neural tube defects. J Pediatr, 1992. 120(6): p. 564

863-71. 565

14. Lammer, E.J., D.T. Chen, R.M. Hoar, N.D. Agnish, P.J. Benke, J.T. Braun, C.J. 566

Curry, P.M. Fernhoff, A.W. Grix, Jr., I.T. Lott, and et al., Retinoic acid 567

embryopathy. N Engl J Med, 1985. 313(14): p. 837-41. 568

24

15. Tibbles, L. and M.J. Wiley, A comparative study of the effects of retinoic acid given 569

during the critical period for inducing spina bifida in mice and hamsters. 570

Teratology, 1988. 37(2): p. 113-25. 571

16. Colakoglu, N. and A. Kukner, Teratogenicity of retinoic acid and its effects on 572

TGF-beta2 expression in the developing cerebral cortex of the rat. J Mol Histol, 573

2004. 35(8-9): p. 823-7. 574

17. Shenefelt, R.E., Morphogenesis of malformations in hamsters caused by retinoic 575

acid: relation to dose and stage at treatment. Teratology, 1972. 5(1): p. 103-18. 576

18. Hendrickx, A.G., G. Tzimas, R. Korte, and H. Hummler, Retinoid teratogenicity in 577

the macaque: verification of dosing regimen. J Med Primatol, 1998. 27(6): p. 310-578

8. 579

19. Adams, J., Structure-activity and dose-response relationships in the neural and 580

behavioral teratogenesis of retinoids. Neurotoxicol Teratol, 1993. 15(3): p. 193-581

202. 582

20. Cunningham, M.L., A. Mac Auley, and P.E. Mirkes, From gastrulation to 583

neurulation: transition in retinoic acid sensitivity identifies distinct stages of neural 584

patterning in the rat. Dev Dyn, 1994. 200(3): p. 227-41. 585

21. Cai, L., X.B. Yan, X.N. Chen, Q.Y. Meng, and J.N. Zhou, Chronic all-trans retinoic 586

acid administration induced hyperactivity of HPA axis and behavioral changes in 587

young rats. Eur Neuropsychopharmacol, 2010. 20(12): p. 839-47. 588

22. Zhang, X., X. Yuan, L. Chen, H. Wei, J. Chen, and T. Li, The change in retinoic 589

acid receptor signaling induced by prenatal marginal vitamin A deficiency and its 590

effects on learning and memory. J Nutr Biochem, 2017. 47: p. 75-85. 591

23. Lai, X., X. Wu, N. Hou, S. Liu, Q. Li, T. Yang, J. Miao, Z. Dong, J. Chen, and T. 592

Li, Vitamin A Deficiency Induces Autistic-Like Behaviors in Rats by Regulating 593

the RARbeta-CD38-Oxytocin Axis in the Hypothalamus. Mol Nutr Food Res, 594

2018. 62(5). 595

24. Abu-Abed, S., P. Dolle, D. Metzger, B. Beckett, P. Chambon, and M. Petkovich, 596

The retinoic acid-metabolizing enzyme, CYP26A1, is essential for normal 597

hindbrain patterning, vertebral identity, and development of posterior structures. 598

Genes Dev, 2001. 15(2): p. 226-40. PMC312609 599

25. Sakai, Y., C. Meno, H. Fujii, J. Nishino, H. Shiratori, Y. Saijoh, J. Rossant, and H. 600

Hamada, The retinoic acid-inactivating enzyme CYP26 is essential for establishing 601

an uneven distribution of retinoic acid along the anterio-posterior axis within the 602

mouse embryo. Genes Dev, 2001. 15(2): p. 213-25. PMC312617 603

26. Abu-Abed, S., P. Dolle, D. Metzger, C. Wood, G. MacLean, P. Chambon, and M. 604

Petkovich, Developing with lethal RA levels: genetic ablation of Rarg can restore 605

the viability of mice lacking Cyp26a1. Development, 2003. 130(7): p. 1449-59. 606

27. Smith, S.M. and G. Eichele, Temporal and regional differences in the expression 607

pattern of distinct retinoic acid receptor-beta transcripts in the chick embryo. 608

Development, 1991. 111(1): p. 245-52. 609

28. Chen, W.H., G.M. Morriss-Kay, and A.J. Copp, Genesis and prevention of spinal 610

neural tube defects in the curly tail mutant mouse: involvement of retinoic acid and 611

25

its nuclear receptors RAR-beta and RAR-gamma. Development, 1995. 121(3): p. 612

681-91. 613

29. Ruberte, E., P. Dolle, P. Chambon, and G. Morriss-Kay, Retinoic acid receptors 614

and cellular retinoid binding proteins. II. Their differential pattern of transcription 615

during early morphogenesis in mouse embryos. Development, 1991. 111(1): p. 45-616

60. 617

30. Lohnes, D., M. Mark, C. Mendelsohn, P. Dolle, A. Dierich, P. Gorry, A. 618

Gansmuller, and P. Chambon, Function of the retinoic acid receptors (RARs) 619

during development (I). Craniofacial and skeletal abnormalities in RAR double 620

mutants. Development, 1994. 120(10): p. 2723-48. 621

31. Lohnes, D., P. Kastner, A. Dierich, M. Mark, M. LeMeur, and P. Chambon, 622

Function of retinoic acid receptor gamma in the mouse. Cell, 1993. 73(4): p. 643-623

58. 624

32. Iulianella, A. and D. Lohnes, Contribution of retinoic acid receptor gamma to 625

retinoid-induced craniofacial and axial defects. Dev Dyn, 1997. 209(1): p. 92-104. 626

33. Deak, K.L., et al., Analysis of ALDH1A2, CYP26A1, CYP26B1, CRABP1, and 627

CRABP2 in human neural tube defects suggests a possible association with alleles 628

in ALDH1A2. Birth Defects Res A Clin Mol Teratol, 2005. 73(11): p. 868-75. 629

34. Li, H., J. Zhang, S. Chen, F. Wang, T. Zhang, and L. Niswander, Genetic 630

contribution of retinoid-related genes to neural tube defects. Hum Mutat, 2018. 631

39(4): p. 550-562. PMC5839987 632

35. Richard, A.M., et al., ToxCast Chemical Landscape: Paving the Road to 21st 633

Century Toxicology. Chem Res Toxicol, 2016. 29(8): p. 1225-51. 634

36. Worth, A.P. and G. Patlewicz, Integrated Approaches to Testing and Assessment. 635

Adv Exp Med Biol, 2016. 856: p. 317-342. 636

37. Robinson, J.F., A. Verhoef, and A.H. Piersma, Transcriptomic analysis of 637

neurulation and early organogenesis in rat embryos: an in vivo and ex vivo 638

comparison. Toxicol Sci, 2012. 126(1): p. 255-66. 639

38. Morriss, G.M. and C.E. Steele, Comparison of the effects of retinol and retinoic 640

acid on postimplantation rat embryos in vitro. Teratology, 1977. 15(1): p. 109-19. 641

39. Steele, C.E., R. Marlow, J. Turton, and R.M. Hicks, In-vitro teratogenicity of 642

retinoids. Br J Exp Pathol, 1987. 68(2): p. 215-23. PMC2013012 643

40. Klug, S., C. Lewandowski, L. Wildi, and D. Neubert, All-trans retinoic acid and 644

13-cis-retinoic acid in the rat whole-embryo culture: abnormal development due to 645

the all-trans isomer. Arch Toxicol, 1989. 63(6): p. 440-4. 646

41. Luijten, M., V.A. van Beelen, A. Verhoef, M.F. Renkens, M.H. van Herwijnen, A. 647

Westerman, F.J. van Schooten, J.L. Pennings, and A.H. Piersma, Transcriptomics 648

analysis of retinoic acid embryotoxicity in rat postimplantation whole embryo 649

culture. Reprod Toxicol, 2010. 30(2): p. 333-40. 650

42. Robinson, J.F., A. Verhoef, J.L. Pennings, T.E. Pronk, and A.H. Piersma, A 651

comparison of gene expression responses in rat whole embryo culture and in vivo: 652

time-dependent retinoic acid-induced teratogenic response. Toxicol Sci, 2012. 653

126(1): p. 242-54. 654

26

43. Ginis, I., Y. Luo, T. Miura, S. Thies, R. Brandenberger, S. Gerecht-Nir, M. Amit, 655

A. Hoke, M.K. Carpenter, J. Itskovitz-Eldor, and M.S. Rao, Differences between 656

human and mouse embryonic stem cells. Dev Biol, 2004. 269(2): p. 360-80. 657

44. Fritsche, E., M. Barenys, J. Klose, S. Masjosthusmann, L. Nimtz, M. Schmuck, S. 658

Wuttke, and J. Tigges, Current Availability of Stem Cell-Based In Vitro Methods 659

for Developmental Neurotoxicity (DNT) Testing. Toxicol Sci, 2018. 165(1): p. 21-660

30. 661

45. Singh, S., A. Srivastava, V. Kumar, A. Pandey, D. Kumar, C.S. Rajpurohit, V.K. 662

Khanna, S. Yadav, and A.B. Pant, Stem Cells in Neurotoxicology/Developmental 663

Neurotoxicology: Current Scenario and Future Prospects. Mol Neurobiol, 2016. 664

53(10): p. 6938-6949. 665

46. Maden, M., Role and distribution of retinoic acid during CNS development. Int Rev 666

Cytol, 2001. 209: p. 1-77. 667

47. Janesick, A., S.C. Wu, and B. Blumberg, Retinoic acid signaling and neuronal 668

differentiation. Cell Mol Life Sci, 2015. 72(8): p. 1559-76. 669

48. Akanuma, H., X.Y. Qin, R. Nagano, T.T. Win-Shwe, S. Imanishi, H. Zaha, J. 670

Yoshinaga, T. Fukuda, S. Ohsako, and H. Sone, Identification of Stage-Specific 671

Gene Expression Signatures in Response to Retinoic Acid during the Neural 672

Differentiation of Mouse Embryonic Stem Cells. Front Genet, 2012. 3: p. 141. 673

PMC3413097 674

49. Schulpen, S.H., P.T. Theunissen, J.L. Pennings, and A.H. Piersma, Comparison of 675

gene expression regulation in mouse- and human embryonic stem cell assays during 676

neural differentiation and in response to valproic acid exposure. Reprod Toxicol, 677

2015. 56: p. 77-86. 678

50. 5Colleoni, S., C. Galli, J.A. Gaspar, K. Meganathan, S. Jagtap, J. Hescheler, A. 679

Sachinidis, and G. Lazzari, Development of a neural teratogenicity test based on 680

human embryonic stem cells: response to retinoic acid exposure. Toxicol Sci, 2011. 681

124(2): p. 370-7. 682

51. Robinson, J.F., M.J. Gormley, and S.J. Fisher, A genomics-based framework for 683

identifying biomarkers of human neurodevelopmental toxicity. Reprod Toxicol, 684

2016. 60: p. 1-10. PMC4867143 685

52. Negraes, P.D., T.T. Schwindt, C.A. Trujillo, and H. Ulrich, Neural differentiation 686

of P19 carcinoma cells and primary neurospheres: cell morphology, proliferation, 687

viability, and functionality. Curr Protoc Stem Cell Biol, 2012. Chapter 2: p. Unit 688

2D 9. 689

53. Adler, S., M. Paparella, C. Pellizzer, T. Hartung, and S. Bremer, The detection of 690

differentiation-inducing chemicals by using green fluorescent protein expression in 691

genetically engineered teratocarcinoma cells. Altern Lab Anim, 2005. 33(2): p. 91-692

103. 693

54. 5Orsolits, B., A. Borsy, E. Madarasz, Z. Meszaros, T. Kohidi, K. Marko, M. Jelitai, 694

E. Welker, and Z. Kornyei, Retinoid machinery in distinct neural stem cell 695

populations with different retinoid responsiveness. Stem Cells Dev, 2013. 22(20): 696

p. 2777-93. PMC3787404 697

27

55. Hermsen, S.A., E.J. van den Brandhof, L.T. van der Ven, and A.H. Piersma, 698

Relative embryotoxicity of two classes of chemicals in a modified zebrafish 699

embryotoxicity test and comparison with their in vivo potencies. Toxicol In Vitro, 700

2011. 25(3): p. 745-53. 701

56. Selderslaghs, I.W., A.R. Van Rompay, W. De Coen, and H.E. Witters, 702

Development of a screening assay to identify teratogenic and embryotoxic 703

chemicals using the zebrafish embryo. Reprod Toxicol, 2009. 28(3): p. 308-20. 704

57. Costaridis, P., C. Horton, J. Zeitlinger, N. Holder, and M. Maden, Endogenous 705

retinoids in the zebrafish embryo and adult. Dev Dyn, 1996. 205(1): p. 41-51. 706

58. Grandel, H., K. Lun, G.J. Rauch, M. Rhinn, T. Piotrowski, C. Houart, P. Sordino, 707

A.M. Kuchler, S. Schulte-Merker, R. Geisler, N. Holder, S.W. Wilson, and M. 708

Brand, Retinoic acid signalling in the zebrafish embryo is necessary during pre-709

segmentation stages to pattern the anterior-posterior axis of the CNS and to induce 710

a pectoral fin bud. Development, 2002. 129(12): p. 2851-65. 711

59. Lee, K. and I. Skromne, Retinoic acid regulates size, pattern and alignment of 712

tissues at the head-trunk transition. Development, 2014. 141(22): p. 4375-84. 713

60. Papalopulu, N., J.D. Clarke, L. Bradley, D. Wilkinson, R. Krumlauf, and N. Holder, 714

Retinoic acid causes abnormal development and segmental patterning of the 715

anterior hindbrain in Xenopus embryos. Development, 1991. 113(4): p. 1145-58. 716

61. Berenguer, M., J.J. Lancman, T.J. Cunningham, P.D.S. Dong, and G. Duester, 717

Mouse but not zebrafish requires retinoic acid for control of neuromesodermal 718

progenitors and body axis extension. Dev Biol, 2018. 441(1): p. 127-131. 719

PMC6064660 720

62. Holder, N. and J. Hill, Retinoic acid modifies development of the midbrain-721

hindbrain border and affects cranial ganglion formation in zebrafish embryos. 722

Development, 1991. 113(4): p. 1159-70. 723

63. Herrmann, K., Teratogenic effects of retinoic acid and related substances on the 724

early development of the zebrafish (Brachydanio rerio) as assessed by a novel 725

scoring system. Toxicol In Vitro, 1995. 9(3): p. 267-83. 726

64. White, R.J., Q. Nie, A.D. Lander, and T.F. Schilling, Complex regulation of 727

cyp26a1 creates a robust retinoic acid gradient in the zebrafish embryo. PLoS Biol, 728

2007. 5(11): p. e304. PMC2080651 729

65. Emoto, Y., H. Wada, H. Okamoto, A. Kudo, and Y. Imai, Retinoic acid-730

metabolizing enzyme Cyp26a1 is essential for determining territories of hindbrain 731

and spinal cord in zebrafish. Dev Biol, 2005. 278(2): p. 415-27. 732

66. Begemann, G., T.F. Schilling, G.J. Rauch, R. Geisler, and P.W. Ingham, The 733

zebrafish neckless mutation reveals a requirement for raldh2 in mesodermal 734

signals that pattern the hindbrain. Development, 2001. 128(16): p. 3081-94. 735

67. Feng, L., R.E. Hernandez, J.S. Waxman, D. Yelon, and C.B. Moens, Dhrs3a 736

regulates retinoic acid biosynthesis through a feedback inhibition mechanism. Dev 737

Biol, 2010. 338(1): p. 1-14. PMC2858591 738

68. Cai, A.Q., K. Radtke, A. Linville, A.D. Lander, Q. Nie, and T.F. Schilling, Cellular 739

retinoic acid-binding proteins are essential for hindbrain patterning and signal 740

robustness in zebrafish. Development, 2012. 139(12): p. 2150-5. PMC3357909 741

28

69. Samarut, E., D. Fraher, V. Laudet, and Y. Gibert, ZebRA: An overview of retinoic 742

acid signaling during zebrafish development. Biochim Biophys Acta, 2015. 743

1849(2): p. 73-83. 744

70. Wang, Y., J. Chen, C. Du, C. Li, C. Huang, and Q. Dong, Characterization of 745

retinoic acid-induced neurobehavioral effects in developing zebrafish. Environ 746

Toxicol Chem, 2014. 33(2): p. 431-7. 747

71. Tonk, E.C., J.L. Pennings, and A.H. Piersma, An adverse outcome pathway 748

framework for neural tube and axial defects mediated by modulation of retinoic 749

acid homeostasis. Reprod Toxicol, 2015. 55: p. 104-13. 750

72. Louisse, J., S. Bosgra, B.J. Blaauboer, I.M. Rietjens, and M. Verwei, Prediction of 751

in vivo developmental toxicity of all-trans-retinoic acid based on in vitro toxicity 752

data and in silico physiologically based kinetic modeling. Arch Toxicol, 2015. 753

89(7): p. 1135-48. 754

73. Dimopoulou, M., A. Verhoef, B. van Ravenzwaay, I.M. Rietjens, and A.H. 755

Piersma, Flusilazole induces spatio-temporal expression patterns of retinoic acid-, 756

differentiation- and sterol biosynthesis-related genes in the rat Whole Embryo 757

Culture. Reprod Toxicol, 2016. 64: p. 77-85. 758

74. Robinson, J.F., E.C. Tonk, A. Verhoef, and A.H. Piersma, Triazole induced 759

concentration-related gene signatures in rat whole embryo culture. Reprod 760

Toxicol, 2012. 34(2): p. 275-83. 761

75. Zhang, C., A. Anderson, and G.J. Cole, Analysis of crosstalk between retinoic acid 762

and sonic hedgehog pathways following ethanol exposure in embryonic zebrafish. 763

Birth Defects Res A Clin Mol Teratol, 2015. 103(12): p. 1046-57. PMC4697881 764

76. Xu, T., L. Chen, C. Hu, and B. Zhou, Effects of acute exposure to polybrominated 765

diphenyl ethers on retinoid signaling in zebrafish larvae. Environ Toxicol 766

Pharmacol, 2013. 35(1): p. 13-20. 767

77. Liu, S., L. Zhang, Y. Liu, X. Shen, and L. Yang, Isoflurane inhibits embryonic stem 768

cell self-renewal through retinoic acid receptor. Biomed Pharmacother, 2015. 74: 769

p. 111-6. 770

78. Baker, N., T. Knudsen, and A. Williams, Abstract Sifter: a comprehensive front-771

end system to PubMed. F1000Res, 2017. 6. PMC5801564 772

79. Bayha, E., M.C. Jorgensen, P. Serup, and A. Grapin-Botton, Retinoic acid signaling 773

organizes endodermal organ specification along the entire antero-posterior axis. 774

PLoS One, 2009. 4(6): p. e5845. PMC2690404 775

80. Cunningham, T.J. and G. Duester, Mechanisms of retinoic acid signalling and its 776

roles in organ and limb development. Nat Rev Mol Cell Biol, 2015. 16(2): p. 110-777

23. PMC4636111 778

81. Tembe, E.A., R. Honeywell, N.E. Buss, and A.G. Renwick, All-trans-retinoic acid 779

in maternal plasma and teratogenicity in rats and rabbits. Toxicol Appl Pharmacol, 780

1996. 141(2): p. 456-72. 781

82. Cai, W., H. Zhao, J. Guo, Y. Li, Z. Yuan, and W. Wang, Retinoic acid-induced 782

lumbosacral neural tube defects: myeloschisis and hamartoma. Childs Nerv Syst, 783

2007. 23(5): p. 549-54. 784

29

83. Nolen, G.A., The effects of various levels of dietary protein on retinoic acid-785

induced teratogenicity in rats. Teratology, 1972. 5(2): p. 143-51. 786

84. Danzer, E., U. Schwarz, S. Wehrli, A. Radu, N.S. Adzick, and A.W. Flake, Retinoic 787

acid induced myelomeningocele in fetal rats: characterization by histopathological 788

analysis and magnetic resonance imaging. Exp Neurol, 2005. 194(2): p. 467-75. 789

85. Turton, J.A., G.B. Willars, J.N. Haselden, S.J. Ward, C.E. Steele, and R.M. Hicks, 790

Comparative teratogenicity of nine retinoids in the rat. Int J Exp Pathol, 1992. 791

73(5): p. 551-63. PMC2002014 792

86. Chen, H., H. Seifikar, N. Larocque, Y. Kim, I. Khatib, C.J. Fernandez, N. Abello, 793

and J.F. Robinson, Using a Multi-Stage hESC Model to Characterize BDE-47 794

Toxicity during Neurogenesis. Toxicol Sci, 2019. PMC6736394 795