Contributions of dam and conceptus to differences in sensitivity to valproic acid among C57 black and SWV mice

To ascertain the relative contributions of genotypes of conceptus and dam to developmental toxicity occasioned by valproic acid (VPA), crosses were established between resistant C57BL/6JBk (C, C57) and susceptible SWV/Bk (S, SWV) strains of mice. These included matings of pure lines, reciprocal outcrosses, and reciprocal backcrosses with F1 hybrids. At 8 d:12 h +/- 5 h, for each mating, 0, 500, or 600 mg/kg aqueous VPA was injected ip. Fetuses were examined on gestation day (gd) 18 for exencephaly (the paradigmatic anomaly), other abnormalities, mortality, litter size, and fetal weight. At 600 mg/kg, sensitivity to exencephaly induction in all cases was that of the dam, regardless of sire. Thus exencephaly here seems to be largely a function of the uterine environment produced by the maternal genotype. This inference is confirmed in backcrosses where F1-dams x S-sires and F1-dams x C-sires produced-identical outcomes, and S-dams x F1-sires produced much higher frequencies of exencephaly than C-dams x F1-sires. For prenatal mortality, the genotypes of both dam and conceptus appear to be important determinants. Fetal contribution is inferred from the observations that S-dam x S-sire matings produced a much higher frequency of mortality than S-dams x C-sires, and C-dams x C-sires produced higher mortality than C-dams x S-sires. Therefore, heterozygosity of the conceptus was protective. Among backcrosses, fetal determination of sensitivity to mortality is also seen by the observation that F1-dams x C-sires produces the same fetal mortality as C-dams x F1-sires. The contribution of uterine environment is seen in the observation that matings of S-dams x C-sires resulted in higher fetal mortality than did those with C-dams x S-sires. Therefore, identical conceptuses in different dams showed different levels of fetal loss. Thus exencephaly response appears to be largely controlled by genes active in the dam, and mortality as a result of a multigenic outcome with contributing genes active in both conceptus and dam. The data also suggest that SWV pure-line dams make a contribution to prenatal mortality not seen in C57 or F1 dams. Mean litter size among VPA-exposed litters showed high variability in pure lines and outcrosses. In backcrosses, F1 dams produced larger litters than pure line dams, arguing for heterosis as a contributor to this parameter. Reduction in litter size occasioned by VPA exposure was great in pure line dams and nonexistent in F1 dams. The SWV dams crossed with F1 sires were the only group among the backcrosses to show reduction of litter size, providing further confirmation of the increased sensitivity of pure-line (i.e., homozygous) SWV dams to VPA exposure. Fetal weight seems to be a function of uterine environment because female SWV produced conceptuses with lower fetal weight in all crosses, and produced a greater reduction in fetal weight attributable to VPA exposure than C57 or F1 dams. Fetal weight did not correlate closely with litter size, suggesting that a lower fetal weight may be a strain characteristic, as are exencephaly induction and prenatal mortality in response to VPA. Differences in sensitivity to VPA insult are seen for all parameters investigated with SWV dams being the most sensitive, but mechanisms seem to differ for a number of the endpoints.

Teratogenic response to arsenite during neurulation: relative sensitivities of C57BL/6J and SWV/Fnn mice and impact of the splotch allele

Arsenic is an environmental contaminant that induces congenital malformations, primarily neural tube defects, in laboratory animals, and it may contribute to human birth defects. The acute doses of arsenicals required to elicit teratogenesis in outbred strains of mice, however, are orders of magnitude higher than those to which humans are exposed environmentally. In order to examine interactions between arsenite administration during neurulation and murine genotype, the present study compares two inbred mouse strains, establishes a teratogenic dose of arsenite, and evaluates the effect of the splotch mutation on arsenic-induced teratogenesis. SWV/Fnn or C57BL/6J females were injected intraperitoneally with sodium arsenite (10 mg/kg) on days 6.5, 7.0, 7.5, 8.0, 8.5, or 9.0 of gestation. A dose-response study was carried out in the C57BL/6J strain, and the effect of the splotch mutation, introduced via the male (C57BL/6J Sp/+), was assessed. Fetuses were examined for external, visceral, and skeletal malformations. Fetuses from crosses of C57BL/6J females with C57BL/6J Sp/+ males were genotyped by PCR. Ten-mg/kg sodium arsenite was teratogenic in nearly 50% of C57BL/6J fetuses, and the C57BL/6J strain was significantly more sensitive to arsenite-induced embryo-lethality and teratogenicity than the SWV/Fnn strain. The spectrum of malformations produced was dependent on the gestational time point of arsenite administration. Introduction of the splotch allele significantly increased neural tube defects and other specific malformations. This result demonstrates that a mutation in a single gene can increase sensitivity to arsenic-induced birth defects. This murine study examines the interaction between arsenite-induced teratogenicity and genotype.

Teratology of retinoids

Either an excess or a deficiency of vitamin A and related compounds (retinoids) causes abnormal morphological development (teratogenesis). Potential retinoid sources come from dietary intake, nutritional supplements, and some therapeutic drugs. Therefore, understanding the mechanisms of retinoid teratogenesis is important. This review first gives an overview of the principles of teratology as they apply to retinoid-induced malformations. It then describes relevant aspects of the biochemical pathway and signal transduction of retinoids. The teratogenic activity of various retinoid compounds, the role of the retinoid receptors, and important toxicokinetic parameters in teratogenesis are reviewed.

Methimazole embryopathy: delineation of the phenotype

We report on a further case of congenital anomalies in a child exposed to methimazole during the first trimester of pregnancy (from first to seventh gestational week), and define a specific malformation pattern related to prenatal methimazole exposure and consisting of choanal and esophageal atresia, scalp defects, minor facial anomalies and psychomotor delay.

A role for DNA methylation in gastrulation and somite patterning

DNA methylation constitutes an important epigenetic factor in the control of genetic information. In this study, we analyzed expression of the DNA methyltransferase gene and examined DNA methylation patterns during early development of the zebrafish. Maternal transcripts of the zebrafish DNA methyltransferase gene (MTase) are ubiquitously present at high levels in early embryos with overall levels decreasing after the blastula stage. At 24 h, methyltransferase mRNA is predominantly found in the brain, neural tube, eyes, and differentiating somites. Expression of MTase in the somites is highest in the anterior cells of the somites. Despite the high levels of MTase mRNA in blastula-stage embryos, we observe DNA hypomethylation at the blastula and gastrula stages compared to sperm or older embryos. Zebrafish embryos treated with 5-azacytidine (5-azaC) and 5-aza-2-deoxycytidine (5-azadC), nucleotide analogs known to induce cellular differentiation and DNA hypomethylation in mammalian cells, exhibit DNA hypomethylation and developmental perturbations. These defects are specifically observed in embryos treated at the beginning of the blastula period, just prior to midblastula transition. The most common phenotype is the loss of tail and abnormal patterning of somites. Head development is also affected in some embryos. Histological and in situ hybridization analyses reveal whole or partial loss of a differentiated notochord and midline muscle in treated embryos. When examined during gastrulation, 5-azaC-treated embryos have a shortened and thickened axial mesoderm. We propose that DNA methylation is required for normal gastrulation and subsequent patterning of the dorsal mesoderm.

Cytoplasmic factors do not contribute to a maternal effect on ethanol teratogenesis

Both maternal and fetal genetic factors influence variations in response to prenatal ethanol exposure. To assess the effect of maternal genotype on the incidence of ethanol teratogenesis, a reciprocal cross study was conducted in an animal mode using the relatively susceptible C57BL/6J (B6) and the relatively resistant DBA/2J (D2) inbred mice. This mating pattern produced four embryonic genotypes: true-bred B6B6 and D2D2 litters and hybrid B6D2 and D2B6 litters. To examine the role of maternal egg cytoplasm as the source of variation that could account for a maternal effect, B6D2 and D2B6 F1 females were mated back to B6 males, which produced two additional embryonic genotypes: B6D2.B6 and D2B6.B6. Dams were intubated with either 5.8 g/kg of ethanol or an isocaloric amount of maltose-dextrin on day 9 of pregnancy. On day 18 of pregnancy, dams were sacrificed, fetuses were removed, weighed, sexed, and examined for gross morphological malformations. Every other fetus within a litter was prepared for either skeletal or soft tissue analysis. Results showed a higher rate of teratogenesis in the B6D2 group compared to the genetically similar D2B6 group, which indicates an influence of maternal genotype on susceptibility to ethanol teratogenesis. The percentage of affected male and female fetuses did not differ, which suggests that sex-linked factors are not responsible for the maternal effect. The backcross B6D2.B6 and D2B6.B6 litters did not differ significantly for any measure of teratogenesis, suggesting that differences in maternally transmitted cytoplasmic material are not the cause of the maternal effect. Factors that could account for the maternal effect are differences in the maternal uterine environment and genomic imprinting. Separating maternal from fetal-mediated mechanisms responsible for susceptibility to ethanol teratogenesis is needed for identifying mothers and infants at risk.

Phenytoin-induced alterations in craniofacial gene expression

In utero exposure to the anticonvulsant drug phenytoin has been shown to alter normal embryonic development, leading to a pattern of dysmorphogenesis known as the Fetal Hydantoin Syndrome. This embryopathy is characterized by growth retardation, microcephaly, mental deficiency, and craniofacial malformations, although the precise mechanism(s) by which phenytoin alters normal developmental pathways remains unknown. To better understand the molecular events involved in the pathogenesis of phenytoin-induced congenital defects, alterations in gene expression were examined during critical periods of craniofacial development. Pregnant SWV mice were administered phenytoin (60 mg/kg/day) from gestational day 6.5 until they were sacrificed at selected developmental time points. Tissue from the craniofacial region of control and exposed embryos was isolated, and samples were subjected to in situ transcription, antisense RNA amplification, and hybridization on reverse Northern blots to quantitatively assess expression of 36 candidate genes. Chronic phenytoin exposure significantly altered expression of several genes at distinct times during morphogenesis. Results of these studies show that expression of the retinoic acid receptors (RAR) alpha, beta, and gamma were significantly increased by phenytoin exposure. Elevations in gene expression of laminin beta 1, and the growth factors IGF-2, TGF alpha, and TGF beta 1, were also demonstrated in the craniofacial region of phenytoin-exposed embryos. As several of these genes are transcriptionally regulated by retinoic-acid-responsive elements in their promoter regions, phenytoin-induced alterations in expression of the RAR isoforms may have severe downstream consequences in the regulation of events necessary for normal craniofacial development. Such alterations occurring coordinately at critical times during craniofacial development may account for the dysmorphogenesis often associated with phenytoin exposure.

Does thalidomide cause second generation birth defects?

The proposed association between thalidomide and second generation birth defects is an improbable hypotheses which lacks, so far, any credible scientific foundation. However, the media have chosen to give it extensive coverage. So much so that even the hard-headed scientist may start wondering if there is anything in it. However, there is no reason to suppose that people with birth defects caused by exposure to thalidomide during embryonic life have any greater or lesser chance of producing children with birth defects. This appears to be the case in practice. The question could be reworded to, 'Can thalidomide be responsible for identical, or similar, birth defects in 2 generations of the same family?' For such a phenomenon to be possible, a mechanism must be proposed and there appear to be only 2 possible candidates. The first is that the defects in the parent, originating during embryonic life, have somehow been transmitted to the next generation. The second is that thalidomide is a mutagen as well as a teratogen. The first mechanism can be excluded, since Lamarckism has long since been abandoned by scientists. The hypothesis that thalidomide is a mutagen and might be responsible for birth defects in the children of thalidomide-damaged people is without any scientific foundation. Birth defects appear to be no more common amongst the children of thalidomide-affected parents than in the general population. It is important that thalidomide-affected adults are firmly reassured on this point. Most of them have now completed their own families, but they may still worry about their grandchildren. Therefore, unless and until further supportive evidence is reported by a separate and independent source, the answer to the question, 'Can thalidomide cause second generation defects?' is a very definite 'No.'

Industrial Genotoxicology Group (IGG): Aneuploidy, White House Hotel, London, UK, 16 May 1997

Aneuploidy has important consequences for human health. There is a requirement to detect aneugens that may not be fully satisfied by the current test batter. New cytogenetic techniques look promising.

Genetic, carcinogenic and teratogenic effects of radiofrequency fields

This paper reviews the literature data on the genetic toxicology of radiofrequency (RF) radiation. Whereas in the past most studies were devoted to microwave ovens and radar equipment, it is now mobile telecommunication that attracts most attention. Therefore we focus on mobile telephone frequencies where possible. According to a great majority of the papers, radiofrequency fields, and mobile telephone frequencies in particular, are not genotoxic: they do not induce genetic effects in vitro and in vivo, at least under non-thermal exposure conditions, and do not seem to be teratogenic or to induce cancer. Yet, some investigations gave rather alarming results that should be confirmed and completed by further experiments. Among them the investigation of synergistic effects and of possible mechanisms of action should be emphasised.

Occupational exposure to pesticides and congenital malformations: a review of mechanisms, methods, and results

Pesticides are chemicals that are widely used all over the world. Human beings can be exposed through environmental contamination and/or occupational use of pesticides. Although there is substantial information on the acute toxicity of many of these chemicals, available knowledge on delayed effects is much more limited. This paper reviews epidemiological studies on occupational exposure to pesticides, mainly in agricultural workers, and risk of congenital malformations. The discussion includes postulated mechanisms for birth defects from paternal or maternal exposure, a detailed review of method of the studies carried out so far on the relationship between occupational exposure to pesticides and congenital malformations, and a summary of main results obtained. From available information, it seems reasonable to conclude that, to date, there is inadequate evidence for either establishing a relationship between pesticides exposure in human beings and birth defects or for rejecting it. Avoiding the main limitations of previous studies, some of the key elements for future research are presented.

Antisense approaches for investigating mechanisms of abnormal development

Although the use of antisense oligodeoxynucleotides in cell culture systems has been beneficial for examining functional roles of genes in biological processes, very few studies have adapted this valuable approach to developmental studies. This oversight may be due to the fact that many scientists are not familiar with the numerous in vitro approaches available for use as developmental system models. Increased knowledge concerning the mechanisms by which oligonucleotides are incorporated into the cell and how these molecules disrupt targeted gene expression has resulted in improved oligonucleotide design and better controls for these studies. The combined use of developmental in vitro approaches, with improved antisense oligodeoxynucleotide strategies presents valuable experimental models for examining functions and interactions of genes in embryogenesis. This review focuses on a comprehensive description of the characterized properties of oligodeoxynucleotides, control design, and various developmental in vitro approaches for accomplishing these studies in embryonic systems.

Alcohol dehydrogenase-2*3 allele protects against alcohol-related birth defects among African Americans

Considerable variation in offspring outcome is observed after intrauterine alcohol exposure. The underlying mechanism may include genetic diversity in the enzymes responsible for alcohol metabolism. Of the known genetic polymorphisms, differences at the alcohol dehydrogenase-2 locus (ADH2) are likely most critical because the resulting enzymes are >30-fold different in their kinetic constants. To test whether differences in maternal or offspring ADH2 genotype are determinants of risk for alcohol-related birth defects, maternal-infant pairs (n = 243) were enrolled on the basis of maternal alcohol intake during pregnancy and maternal ADH2 genotype. Infant outcome was measured using the Bayley Scales of Infant Development Mental Index (MDI) at 12 months of age. Drinking during pregnancy was associated with lower MDI scores but only in the offspring of mothers without an ADH2*3 allele (P < .01, analysis of variance, post hoc). The offspring of drinking women with at least one ADH2*3 allele had MDI scores similar to those of nondrinking women of either ADH2 genotype. Lower MDI scores were associated with the three-way interaction among increasing alcohol intake and maternal and offspring absence of the ADH2*3 allele (P < .01, multiple linear regression). We suggest that the protection afforded by this allele is secondary to its encoding of the high-Km/high-Vmax ADH beta3 isoenzyme, which would provide more efficient alcohol metabolism at high blood alcohol concentrations. These observations are supportive of alcohol, rather than acetaldehyde, being the more important proximate teratogen and are the first observations of a specific genetic explanation for susceptibility differences to alcohol-related birth defects.

Genome scan for teratogen-induced clefting susceptibility loci in the mouse: evidence of both allelic and locus heterogeneity distinguishing cleft lip and cleft palate

Nonsyndromic clefting of the lip and palate in humans has a highly complex etiology, with both multiple genetic loci and exposure to teratogens influencing susceptibility. Previous studies using mouse models have examined only very small portions of the genome. Here we report the findings of a genome-wide search for susceptibility genes for teratogen-induced clefting in the AXB and BXA set of recombinant inbred mouse strains. We compare results obtained using phenytoin (which induces cleft lip) and 6-aminonicotinamide (which induces cleft palate). We use a new statistical approach based on logistic regression suitable for these categorical data to identify several chromosomal regions as possible locations of clefting susceptibility loci, and we review candidate genes located within each region. Because cleft lip and cleft palate do not frequently co-aggregate in human families and because these structures arise semi-independently during development, these disorders are usually considered to be distinct in etiology. Our data, however, implicate several of the same chromosomal regions for both forms of clefting when teratogen-induced. Furthermore, different parental strain alleles are usually associated with clefting of the lip versus that of the palate (i.e., allelic heterogeneity). Because several other chromosomal regions are associated with only one form of clefting, locus heterogeneity also appears to be involved. Our findings in this mouse model suggest several priority areas for evaluation in human epidemiological studies.

Effect of multifactorial genetic liability to exencephaly on the teratogenic effect of valproic acid in mice

The present study shows that the multifactorial genetic liability to spontaneous exencephaly in the SELH/Bc mouse strain (10-20% of embryos) also confers an elevated risk of exencephaly induced by valproic acid. Treatment of pregnant dams (600 mg/kg sodium valproate in distilled water, i.p.) during the critical period on day 8 (D8) of gestation resulted in D14 exencephaly frequencies of 69% in SELH/Bc contrasted with 39% in each of the SWV/Bc and ICR/Bc strains. Analysis of these data under the assumptions of the threshold model indicated that the valproic acid-induced-shift in mean liability was similar for all three strains, and therefore the effects of genotype and teratogen were additive, not synergistic. A similar exencephaly response pattern for the same three strains was observed previously with retinoic acid [Tom et al. (1991) Teratology 43:27-40], a pattern that, combined with the data of Finnell et al. [(1988) Teratology 38:313-320], argues that strain differences in exencephaly response are not due to strain differences in teratogen metabolism. SWV/Bc and ICR/Bc embryos differ in location of the Closure 2 initiation site of cranial neural tube closure [Juriloff et al. (1991) Teratology 44:225-233], but the observation that they do not differ in risk of exencephaly produced by either valproic acid or retinoic acid contradicts the hypothesis that this particular morphological difference underlies strain differences in exencephaly risk. The high exencephaly response of SELH/Bc to two teratogens predicts that human conceptuses with a genetically determined elevated risk for neural tube defects could be easily tipped into high risk by mild teratogens.

Phenytoin-induced teratogenesis: a molecular basis for the observed developmental delay during neurulation

PURPOSE

We wished to determine whether chronic phenytoin (PHT) exposure could impair neural development and if any morphological alterations could be linked to changes in gene expression.

METHODS

Pregnant SWV mice were chronically administered PHT 40 mg/kg/day from gestational day (GD) 0:12 (day:h) until they were killed at various timepoints throughout neural tube closure (NTC). At each timepoint, embryos from both treated and control dams were collected and scored for their progression through NTC. The neural tubes were then isolated and subjected to in situ transcription (IST) and antisense RNA amplification procedures. Using these techniques, we examined the expression of 10 genes: N-cadherin (Ncad), collagen type IV (col-IV), bcl-2, c-jun, PAX-3, collular retinol binding protein-2 (CRBP-2), retinoic acid receptor alpha (RAR alpha), transforming growth factor(beta2) (TGF(beta2)), wee-1, and EMX-2.

RESULTS

Chronic PHT exposure not only caused a delay in NTC whereby exposed embryos lagged behind the controls at each collection timepoint, but also significantly altered the expression of specific genes at distinct times during NTC. Early in NTC, PHT induced a significant reduction in the expression of N-cad, col-IV, and c-jun in exposed embryos as compared with controls. In contrast, during the midstages of NTC, the only significant molecular alterations observed in the PHT-exposed embryos was the continued decreased expression of col-IV and an increase in CRBP-2 expression. Finally, in the latter stages of NTC, PHT caused a significant reduction in the expression of bcl-2, RAR alpha, TGF(beta2), EMX-2, and PAX-3.

CONCLUSIONS

These results show that although the effects of PHT are morphologically subtle, causing a delay in the development of the neural tube, this delay is accompanied by alterations in critical genes at crucial times of neural development that may account for the observed neurological deficits often associated with PHT exposure.

Methanol causes posteriorization of cervical vertebrae in mice

Inhalation of methanol by pregnant mice before gestation day nine (gd 9) produces fetal skeletal alterations, principally in the cervical region. The appearance of these defects suggests homeotic shifts in segment identity, patterning, or both. To explore this possibility, detailed morphological analyses of the effects of methanol on fetal skeletal development were done. Pregnant mice were gavaged with 0, 4.0, or 5.0 g/kg methanol (MeOH) split in two doses on gd 7, the most sensitive day for induction of skeletal alterations with methanol. Dams were killed on gd 18 and the fetuses were counted, weighed, and examined externally. Fetuses were double stained with alcian blue and alizarin red for examination of cartilaginous and ossified vertebral and rib characteristics, and in selected fetuses cervical vertebrae were disarticulated for more detailed analysis. Observations indicative of methanol-induced homeotic transformations were as follows: [tabular data: see abstract volume] Examination of disarticulated vertebrae revealed foramina and other distinguishing characteristics on vertebrae anterior to those on which they normally appear. These results demonstrate that maternal methanol exposure can alter segment patterning in the developing mouse embryo, producing posteriorization of cervical vertebrae.

Valproic acid-induced somite teratogenesis in the chick embryo: relationship with Pax-1 gene expression

The repeated pattern of the axial skeleton results from the segmentation and re-segmentation of the mesodermally derived somites. During these early events of somite development, the vertebrate embryonic axial skeleton is most susceptible to the teratogenic effects of a variety of pharmaceutical and environmental agents. One example is the anticonvulsant drug valproic acid (VPA), which has been shown to cause craniofacial and minor and major skeletal defects in human and animal embryos. We hypothesize that a candidate set of molecular targets of teratogens are the Pax family of pattern-forming genes, specifically Pax-1, which has been previously demonstrated to be an important regulator of axial skeletal patterning at the somite level. In this study, early developmental stage chick embryos were treated with VPA and dose-dependent malformations in somite development were observed. Two classes of anomalies were evident: class I included discrete sites of somitic fusions or mis-segmentation, and Class II included large areas of disorganized somite patterning. Northern blot analysis revealed a decreased level of Pax-1 expression in VPA-treated embryos. Whole mount in situ hybridization analysis showed that somite anomalies correlate spatially with regions of decreased Pax-1 expression. Finally, comparison of the VPA-induced somitic anomalies with those caused by gene-specific perturbation of Pax-1 gene expression through the use of an antisense oligonucleotide revealed significant similarities. Taken together, these results support the hypothesis that Pax-1 is a molecular target in VPA axial skeletal teratogenicity.

Expression of Fgf-3 in relation to hindbrain segmentation, otic pit position and pharyngeal arch morphology in normal and retinoic acid-exposed mouse embryos

The gene Fgf-3 is expressed in rhombomeres 5 and 6 of the hindbrain and has been functionally implicated in otic development. We describe new sites of expression of this gene in mouse embryos in the forebrain, the midbrain-hindbrain junction region, rhombomere boundaries, a cranial surface ectodermal domain that includes the otic placode, and in the most recently formed somite. In the early hindbrain, high levels of Fgf-3 transcripts are present in rhombomere 4. The surface ectodermal domain at first (day 8 1/2) extends laterally from rhombomeres 4 and 5 (prorhombomere B), in which neuroepithelial levels of expression are highest, to the second pharyngeal arch ventrally; at day 9, when the region of highest level of neuroepithelial Fgf-3 expression is in rhombomeres 5 and 6, the dorsal origin of the surface ectodermal domain is also at this level, extending obliquely to the otic placode and the second arch. The initially high level of Fgf-3 transcripts in the otic placode is downregulated as the placode invaginates to form the otic pit. Fgf-3 is a good marker for the epithelium of pharyngeal arches 2 and 3, and our in situ hybridization results confirm the dual identity of the apparently fused first and second arches in some retinoic acid-exposed embryos, and the fusion of the first arch with the maxillary region in others. Correlation between Fgf-3 expression and morphological pattern in craniofacial tissues of normal and retinoic acid-exposed embryos indicates that prorhombomere B, the second arch and the otic ectoderm represent a cranial segment whose structural integrity is maintained when hindbrain morphology and pharyngeal arch morphology are altered. Comparison of normal Fgf-3 expression domains with those of Fgf-4 and with the phenotype of Fgf-3-deficient mutant embryos suggests that there is some functional redundancy between Fgf-3 and Fgf-4 in otic induction and second arch development.