Main

Congenital defects typically occur within sets or associations that have been described by statistical evaluation of epidemiologic data. One common association that has been validated by numerous epidemiologic and experimental studies is the relationship among craniofacial, CNS, and heart defects. The biologic basis of this relationship is straightforward: the mesenchyme of the face and the conotruncal septum of the heart are derived from cranial neural crest cells, the CNS is derived from the neural tube, and the neural crest is both geographically contiguous and functionally interchangeable with the neural tube in the early embryo(1). Thus a teratogenic agent that inhibits neural crest migration to produce facial or septal defects also would be expected to inhibit neural tube closure, and conversely.

Among the numerous factors that have been shown to be associated with congenital neural crest and neural tube defects are three that occur most frequently in the literature: folic acid deficiency, ingestion of ethanol, and exposure to environmental toxins. These three factors in aggregate may account for the majority of neural crest and neural tube defects that cannot be accounted for by the 22q11 deletion, Down's syndrome, or other known genetic factors(2). No common biologic basis for their common dysmorphogenetic effect has been reported for these three seemingly disparate teratologic agents. However, the unifying mechanism may be their shared ability to perturb the function of the NMDA family of glutamate receptors.

One of the most obvious and fully described effects of folic acid deficiency is a rise in the concentration of serum homocysteine, an amino acid that is produced in the metabolism of methionine(3, 4). Recently, we showed that homocysteine induces neural crest and tube defects in a time- and concentration-dependent fashion(5). At concentrations that correlate with the induction of neural crest and tube defects, homocysteine also acts as an NMDA receptor antagonist at the glycine site(6, 7). This ability to act as an NMDA receptor antagonist is shared by ethanol(8–11), and recent studies report that the teratogenic action of ethanol on neural tube derivatives is best understood when it is viewed as a consequence of this effect(12). Furthermore, aliphatic hydrocarbons that are teratogenic contaminants of water also induce craniofacial and neural tube closure defects(13–15). Chloroform and trichloroethylene are the most common of the halogenated hydrocarbons that occur as water pollutants and also are used as anesthetics. Recent evidence has shown that the anesthetic activity of these compounds is related to their ability to act as NMDA receptor antagonists(16–20).

From these data, it may be hypothesized that the ability to induce neural crest and neural tube defects is a general property of NMDA receptor antagonists. In this study we have examined the ability of different classes of NMDA receptor antagonists to cause teratogenic effects in chick embryos. NMDA receptors are dual gated ion channel receptors in that both glutamate and glycine binding are required for receptor activation, which opens a calcium/sodium/potassium-permeable channel. For this study we chose antagonists that act at the glycine or glutamate binding sites or antagonists that bind within the open ion channel (channel blockers). A teratogenic effect would have increasingly serious implications as the clinical use of NMDA agonists and antagonists increases. Of special concern is dextromethorphan, an orally active NMDA antagonist(21, 22), that is widely used as an over-the-counter antitussive medication. The results presented here show that, in avian embryos, NMDA receptor antagonists are potent inducers of neural crest and tube defects, acting in a dose-dependent fashion.

METHODS

Materials. Fertilized chicken eggs were purchased in bulk from HyLine International, Inc. (Dallas Center, IA) and kept refrigerated(14°C) until use. NMDA receptor antagonists were purchased from Tocris Cookson, Inc. (St. Louis, MO), Alexis Biochemicals (San Diego, CA), or else from Sigma Chemical Co. (St. Louis, MO). All stock solutions were made by dissolving drugs in 0.9% saline, and working dilutions for all drugs were prepared for these experiments using 0.9% saline in Falcon 15-mL tissue culture tubes and kept frozen (-20°C) until use.

Negative controls. The following groups were used as negative controls: untreated embryos; sham-treated embryos that were given only the 0.9% saline vehicle, and embryos treated with (-)MK-801.

Positive controls. Our experience has shown that avian embryos may demonstrate seasonal, genetic, or other variations in susceptibility to the experimental induction of developmental defects. To ensure that the susceptibility of these embryos to NMDA antagonists was comparable to that of earlier studies(5), and to provide a comparison defect rate, a set of embryos was treated with 5 μmol/d homocysteine given at the same times as the other NMDA receptor antagonists.

Experimental drugs. Antagonists. NMDA receptor antagonists selected for this study were: channel blockers,(+)MK-801, dextromethorphan, and memantine; glycine site antagonists, kynurenic acid, 7-chloro-kynurenic acid, and 7-chlorothiokynurenic acid; glutamate site antagonists, CPP, D-AP5, and the L-isomer of D-AP5 (L-AP5, an NMDA receptor antagonist known to have a much lower potency than D-AP5 in adult brain NMDA receptors).

Agonists. Selected NMDA receptor agonists included NMDA itself and HQ.

Treatment of embryos. Embryos in ovo were divided into groups of 18 and preincubated in a rocking humidified incubator set at 39°C for 4 h. Embryos in ovo then were removed from the incubator, and the first dose (50 μL in 0.9% saline) was administered to each treatment group, which is a typical dose size for experiments wherein drugs or micronutrients are delivered to avian embryos(5). Immediately after treatment of embryos, the eggs were resealed using a small amount of melted paraffin wax and returned to the incubator. Similar treatments were given at 24 and 48 h after this time, or after 28 and 52 h of total incubation time, respectively. At 24 h after the last treatment (76 h total incubation time), the embryos were were stained in ovo with 0.02% neutral red stain in 0.9% saline. Five minutes later eggs were visualized under a dissection microscope and examined for developmental abnormalities. Some embryos showing exemplary defects were prepared for observation with scanning electron microscopy.

Statistical analysis. At least three separate experiments were performed with each concentration of a drug, with 18 embryos in each treatment group. Developmental defects were assigned to the following categories:“spinal” including spinal dysraphism and caudal dysgenesis;“craniofacial” including dysraphism, partial or complete absence of facial development, small or missing eyes, or bilateral facial asymmetry;“multiple” including any combination of craniofacial and spinal defects; and “other” or any obvious defect not included in the neural crest or neural tube categories (e.g. very small or very large tubular heart). The total number of defects in each category was normalized to the total number of surviving embryos. Similarly, dead embryos were counted from each data group. All data were then entered into an electronic database using Excel 6.0 (Microsoft, Inc.) and Prism 2.0 (GraphPad, Inc.) and normalized to numbers of defects or deaths per 100 embryos. Data shown are the average ± SEM of values determined for each drug and dosage level among all experiments. Where comparisons of means were performed, determinations of significance were performed using a single-factor ANOVA followed by Fisher post hoc analysis, using Statview 2.0 (Abacus Concepts, Inc.).

RESULTS

Spontaneous developmental defects occurred at a low rate in the untreated embryos, with an average of 0.97 ± 0.41 total observable defects per 100 sham-treated embryos. The number of spontaneous deaths in vehicle-treated groups was also low, 1.46% ± 0.52%. Differences in defect and mortality rates between sham-treated and untreated control embryos were negligible (data not shown). Treatment of embryos with 5 μmol/d DL-homocysteine thiolactone resulted in a total neural tube defect rate of 27.7 ± 3.4 neural tube defects/100 embryos, and a mortality of 46.75 ± 2.7%, confirming a previous study(5).

Channel blockers. Exposure to NMDA receptor channel blockers(+)MK-801, dextromethorphan, or memantine all induced a significant increase in the incidence of craniofacial and spinal developmental lesions(Fig. 1, top). The (-)-stereoisomer of MK-801, a drug known to posses at least an order of magnitude lower potency at blockade of adult brain NMDA receptors, was inactive up to 500 nmol/embryo/d, consistent with the hypothesis that these drugs act to induce neural tube defects by inhibition of NMDA receptor function.

Figure 1
figure 1

Effect of channel blocker NMDA receptor antagonists on incidence of developmental defects and mortality in developing chicken embryos. During the first 2 d of development chicken embryos (groups of 18) were exposed to 0.5-500 nmol/embryo/d of various NMDA receptor channel blockers as described in “Methods.” After 76-h development, embryos were examined for gross developmental abnormalities (top) or death(bottom). Data represent means ± SEM for defect and mortality rates in at least three separate experiments. Asterisks denote significance(*p < 0.05, **p < 0.01, *** p < 0.001vs vehicle control group). DXM, dextromethorphan.

Full size image

Only the channel blockers increased the mortality rate to a significant degree (Fig. 1, bottom). At 500 nmol/embryo/d, dextromethorphan and memantine treatments each resulted in a significant increase in mortality. In particular, dextromethorphan treatment resulted in 56.7 ± 10.6 deaths/100 embryos (p < 0.001 versus control embryos). Dextromethorphan was unique among all the drugs examined in this study in that it caused a significant increase in mortality when it was given in doses as low as 50 nmol/embryo/d (14.1% ± 10.7% embryos dead,p < 0.05 versus control).

The majority of defects caused by NMDA receptor channel blockers were of structures derived from the neural tube and neural crest(Table 1). The drugs differed somewhat in the kinds of defect that they induced. For example, half of the defects induced by(+)MK-801 were spinal, whereas dextromethorphan induced spinal defects in only 3% of affected embryos; on the other hand, 40% of the defects induced by dextromethorphan were craniofacial. An example of cranial dysraphism induced by dextromethorphan is shown in the scanning electron photomicrographs of Figure 2.

Table 1 Distribution of defect types among embryos treated with NMDA receptor antagonists
Full size table
Figure 2
figure 2

Scanning electron photomicrographs of embryos harvested at 76 h. In each embryo the heart (H) has been broken away from the inferior aspect of the face for improved observation. In each case, the bar = 0.1 mm. (A) This embryo was treated with saline vehicle only and is normal. The epidermis is smooth and completely covers the dorsal aspect of the embryo (arrow). The eye (E) and the nasal placode (N) are placed normally. (B) This embryo was given 50 nmol/d (3 doses total) of dextromethorphan as described in “Methods.” A large neural tube dysraphism includes the region of the brain (B) and the spinal cord(S). The thin epidermis over the dysraphic neural tube is coarsely corrugated; this membrane is transparent for light microscopy, permitting visualization of the dysraphic neural tube. The small eye (E) is displaced caudally in relation to the nasal placode (N). This embryo was placed in the category, “multiple” defects (see Table 1). (C) This embryo was given 500 nmol/d dextromethorphan (3 doses total) as described in “Methods.” The thin epidermis was removed from the craniofacial region before the specimen was prepared for scanning electron microscopy to demonstrate the complete brain dysraphism (large arrow).

Full size image

Glycine site antagonists. As a class, NMDA receptor antagonists that act at the glycine co-agonist site were less potent and efficacious in the induction of developmental defects using our model system(Fig. 3, top). However, 7-chlorokynurenic acid (500 nmol/embryo/d) elevated the incidence of neural tube lesions significantly(12.2 ± 3.1 defects/100 embryos, p < 0.01 versus control). Of the defects induced by 7-chlorokynurenic acid, 31% were spinal, 38% were craniofacial, and the remaining 31% were multiple site defects. The kynurenic acid derivatives increased mortality rates somewhat but the increase was not statistically significant (Fig. 3, bottom).

Figure 3
figure 3

Effect of glycine site NMDA receptor antagonists on incidence of developmental defects and mortality in developing chicken embryos. During the first 2 d of development chicken embryos (groups of 18) were exposed to 5-500 nmol/embryo/d of various NMDA receptor glycine site antagonists as described in “Methods.” After 76-h development, embryos were examined for gross developmental abnormalities (top) or death(bottom). Data represent means ± SEM for defect and mortality rates in at least three separate experiments. Asterisks denote significance(**p < 0.01, vs vehicle control group). 7-Cl-Kyn, 7-chlorokynurenic acid; 7-Cl-Thiokyn, 7-chlorothiokynurenic acid.

Full size image

Glutamate site antagonists. Glutamate site antagonists CPP and D-AP5 induced only a small increase in the occurrence of developmental defects(Fig. 4, top) and no increase in the mortality rate(Fig. 4, bottom). The L-isomer of D-AP5 (L-AP5, an NMDA receptor antagonist known to be less potent than D-AP5 in adult brain NMDA receptors) showed little activity.

Figure 4
figure 4

Effect of glutamate site NMDA receptor antagonists on incidence of developmental defects and mortality in developing chicken embryos. During the first 2 d of development chicken embryos (groups of 18) were exposed to 5-500 nmol/embryo/d of various NMDA receptor channel blockers as described in “Methods.” After 76-h development, embryos were examined for gross developmental abnormalities (top) or death (bottom). Data represent means ± SEM for defect and mortality rates in at least three separate experiments.

Full size image

Glutamate site agonists. Neither NMDA per se nor the highly potent NMDA receptor agonist HQ induced early developmental disorders at the doses tested (Fig. 4).

DISCUSSION

The results of this study demonstrate that exposure of early avian embryos to NMDA receptor antagonists, particularly NMDA receptor channel blockers, can result in the induction of neural crest/neural tube defects and can increase significantly the incidence of mortality in the early avian embryo. These data suggest a pivotal role for NMDA receptors, or perhaps another closely related receptor that is blocked by NMDA receptor antagonists, in neural tube development, neural crest cell migration, or differentiation. NMDA receptors mediate synaptic transmission and neural plasticity in the CNS(23–27). NMDA receptors are also known to regulate neuronal development by stabilizing converging synapses(28) by stimulating cerebellar granule cell migration, growth, and differentiation(29–33), and by modulating apoptosis(34–38).

NMDA receptors are comprised of at least two families of receptor subunits, NMDA-R1(39, 40) and NMDA-R2(41–43), both of which are required for a functional receptor. NMDA-R2 subunits represent four different gene products which impart most of the pharmacologic and functional diversity of NMDA receptors(42–44). Recent evidence has also shown that NMDA receptor subunits are differentially expressed during development(45, 46). The specific NMDA receptor subunits that are responsible for the teratogenic effects reported here are not shown by these experiments.

Knockout experiments with individual NMDA receptors failed to demonstrate a role for these receptors in neural tube development(47–50), suggesting the possibility that a closely related receptor that also is responsive to NMDA receptor antagonists [e.g. see Yuzaki et al.(51)] may be partly responsible for NMDA receptor teratogenicity, alternatively that receptor loss is somehow compensated for in the knockout mouse. There are other examples of such functional compensation: embryos exposed to neural cell adhesion molecule antibodies develop severe neural tube defects(52), whereas embryos that are homozygous-deficient for neural cell adhesion molecule do not(53).

Teratogens that induce neural tube developmental abnormalities include homocysteine(5), ethanol(12), and aliphatic hydrocarbons(13–15). Many of these compounds also have known NMDA receptor antagonist activity. We report here that exposure of avian embryos to NMDA receptor antagonists results in increases in neural tube developmental defects and embryonic mortality rates. The results of this study do not show why the NMDA receptor channel blocker class of compounds are the most potent and efficacious teratogens among the NMDA receptor antagonists; however, this efficacy may be due to differences in drug bioavailability. When 48-h incubated eggs were each given a single 50-nmol dose of (+)MK-801, dextromethorphan, homoquinolinic acid, or 5,7-dichlorokynurenic acid, each of which had been labeled with 3H, we observed concentrations of 24 ± 14, 67 ± 30, 2 ± 0.4, and 4 ± 2 μM, respectively (our unpublished observations, n = 3 embryos per drug). Hence, it appears that channel blocker antagonists MK-801 and dextromethorphan may concentrate to a much higher degree in the embryo than agonists or agonists of the glycine site (5,7-dichlorokynurenic acid) or glutamate site (homoquinolinic acid). This is consistent with the fact that channel blocker NMDA receptor antagonists tend to have chemical structures which lend to higher lipophilicity than the highly charged kynurenic acid or amino acid analogs.

The possibility that bioavailability per se may be a key feature in teratogenicity of the NMDA receptor antagonists is supported by our previous report on the teratogenicity of homocysteine(5). Although homocysteine may act as a glycine site antagonist of the NMDA receptor(6, 7), it is aggressively teratogenic(5). The resolution of this apparent ambiguity may be availability: like the calcium channel blockers but unlike the glycine site blockers used in the present study, homocysteine becomes highly concentrated in the embryos(5).

Another or alternative explanation for the difference in potency between the channel blocker class and glutamate site or glycine site antagonists may be in the physiology of NMDA receptor activation during development. If the developmental events that require activation of an NMDA receptor naturally involve a massive release of glutamate and glycine onto the active site, then the presence of relatively small amounts of competitive glutamate or glycine site antagonists would not be sufficient to overcome competition with these endogenous neurotransmitters. In contrast, a noncompetitive antagonist, by definition, would act independently of agonist concentration, and a noncompetitive or uncompetitive antagonist such as a channel blocker would be more effective.

Although these results were not derived directly from human or other mammalian embryos, a substantial amount of our present knowledge about heart and neural tube development has been acquired from studies of the avian embryo, and this knowledge has been shown to apply to mammals as well(54–56). Furthermore, modern techniques of molecular biology have shown clearly that the genes regulating early embryonic developmental events (e.g. neural tube closure and neural crest migration) are highly conserved. These include the genes whose perturbations are likely to be involved in the congenital defects reported here(e.g. PAX-3, sonic hedgehog, or brachyury)(57, 58). There are Drosophila, avian, murine, and human orthologs for each of these, therefore data derived from experiments with early avian embryos may perhaps be applied broadly to the expected results for other species when they are exposed to the same teratogens, including the early human embryo.

Thus, these results are alarming in light of the fact that the clinical use of various NMDA receptor antagonists is increasing. As a result of NMDA receptor involvement in neurodegenerative diseases, antagonists of NMDA receptors have recently emerged in clinical research(59) for the potential treatment of stroke, CNS trauma(60), epilepsy(61, 62), pain(63), Huntington's disease(64), AIDS dementia(65, 66); with possible application in Alzheimer's(67) and Parkinson's diseases(68, 69) as well. However, the most widespread use of an NMDA receptor antagonist, and perhaps the greatest potential threat to human embryos, is the nonprescription cough suppressant dextromethorphan, which is shown by the present data using this model to be a powerful inducer of both congenital defects and embryonic death when present in micromolar concentrations: a dose of 50 nmol/embryo/d caused significant embryo mortality in our experimental system. After a single 30-mg therapeutic dose, dextromethorphan in humans reaches micromolar concentrations in plasma(70), with a plasma half-life of 1-7 h(71). Furthermore, because there is a known polymorphic distribution in cytochrome P-450-dependent dextromethorphan metabolism(71), slow metabolizers of dextromethorphan would be expected to have more highly elevated and prolonged plasma levels after dextromethorphan ingestion. The results of the present study suggest that further studies of the teratogenic effect of dextromethorphan are warranted at this time.