Perturbation of retinoic acid (RA)-mediated limb development suggests a role for diminished RA signaling in the teratogenesis of ethanol§


  • Corey S. Johnson,

    1. The Department of Cell and Developmental Biology, and Bowles Center for Alcohol Studies, The University of North Carolina, Chapel Hill, North Carolina
    Search for more papers by this author
  • Robert M. Zucker,

    1. Reproductive Toxicology Branch, National Health and Environmental Effects Research Laboratory, Office of Research and Development, United States Environmental Protection Agency, Research Triangle Park, North Carolina
    Search for more papers by this author
  • Edward Sidney Hunter III,

    Corresponding author
    1. Reproductive Toxicology Branch, National Health and Environmental Effects Research Laboratory, Office of Research and Development, United States Environmental Protection Agency, Research Triangle Park, North Carolina
    • Reproductive Toxicology Branch, National Health and Environmental Effects Research Laboratory, Office of Research and Development, United States Environmental Protection Agency, Research Triangle Park, North Carolina
    Search for more papers by this author
  • Kathleen K. Sulik

    1. The Department of Cell and Developmental Biology, and Bowles Center for Alcohol Studies, The University of North Carolina, Chapel Hill, North Carolina
    Search for more papers by this author

  • This article is a US Government work and, as such, is in the public domain in the United States of America.

  • This work has been presented, in part, at the 62nd annual meeting of the Society for Developmental Biology, July 30–August 3, 2003 in Boston, MA and at the 44th annual meeting of the Teratology Society, June 26–July 1, 2004 in Vancouver, BC.

  • §

    Disclaimer: Although the research described in this article has been supported in part by the United States Environmental Protection Agency, it has not been subjected to Agency review and therefore does not necessarily reflect the views of the Agency and no official endorsement should be inferred. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.



A proposed mechanism for ethanol teratogenicity entails ethanol-mediated reductions in retinoic acid (RA). This premise was investigated utilizing a mouse model, with limb reduction defects as the teratogenic end point.


Ethanol, Disulfiram, or BMS-189453 was administered to C57BL/6J mice on the 9th day of pregnancy. Forelimb morphology was assessed on gestation day 18 using Alcian blue and Alizarin red staining. Nile blue sulfate or LysoTracker Red (LTR) vital staining identified cell death in the limb bud. The ability of RA to prevent ethanol-induced cell death was assessed by coadministration followed by laser scanning confocal microscopic examination of LTR-staining. In situ hybridization and qPCR were used to examine gene expression in treated limb buds.


Ethanol, Disulfiram, and BMS-189453 resulted in postaxial ectrodactyly, intermediate ectrodactyly, and other digital defects. Excessive Nile blue sulfate staining was evident in the presumptive AER following each of the three exposures. Ethanol-induced LTR staining was prevented by RA supplementation. Both in situ hybridization and qPCR illustrated decreases in Shh and Tbx5 in ethanol-exposed embryos as compared to control.


Contrary to studies of prolonged RA deficiency, acute exposure to functional antagonists of RA results in limb defects that are morphologically similar to those caused by ethanol. The rescue of ethanol-induced cell death by RA and similar changes in Shh transcription further suggest that RA contributes to ethanol-induced limb dysmorphology. Moreover, the repression of key mediators of limb development soon after ethanol exposure adds to the existing knowledge of the pathogenic effects of ethanol. Birth Defects Research (Part A), 2007. © 2007 Wiley-Liss, Inc.


Prenatal ethanol exposure results in a constellation of structural and functional abnormalities collectively known as fetal alcohol spectrum disorders (FASDs). Although less common than brain and craniofacial defects, prenatal ethanol exposure also causes limb defects such as radioulnar synostosis, shortened digits, camptodactyly, clinodactyly, and ectrodactyly (Spiegel et al., 1979; Herrmann et al., 1980; Cremin and Jaffer, 1981; Van Rensburg, 1981; Froster and Baird, 1992). Hypoplastic nails and abnormal palmar creases are also common ethanol-induced defects of the limb (Jones et al., 1973; Tillner and Majewski, 1978; Viljoen et al., 2005).

Limb defects have also been reported in an FASD mouse model (Kronick, 1976; Kotch et al., 1992). They include postaxial ectrodactyly, intermediate digit ectrodactyly, syndactyly, and abnormally large interdigital spaces. As reported by Kotch et al. (1992), the pathogenesis of ethanol-induced limb malformations involves localized cell death in the developing apical ectodermal ridge (AER), a region of specialized epithelium that promotes proximo-distal limb outgrowth and participates in the maintenance of the anterio-posterior signaling center, the zone of polarizing activity (ZPA).

Since the dysmorphogenic effects of ethanol upon the embryo were first described (Lemoine et al., 1968; Jones and Smith, 1973), many experiments have probed the biochemical, physiological, and developmental processes impacted by ethanol exposure. Investigators have observed that morphological similarities are present between humans with FASD and experimental animals exposed to a vitamin A (retinol) deficient diet. Pullarkat (1991) and Duester (1991) noted that retinoic acid (RA) synthesis depends on oxidation of retinol and retinal by enzymes that also metabolize ethanol: alcohol dehydrogenase and aldehyde dehydrogenase (ALDH). Pullarkat and Duester hypothesized that the structural abnormalities induced by ethanol result from diminished RA concentrations, which, in turn, occur as a result of the competitive inhibition of RA synthetic enzymes by ethanol. Indeed, it has been recognized for some time that ethanol reduces RA concentrations in adult tissue (Van Thiel et al., 1974), a phenomenon that has more recently been demonstrated in the mammalian embryo (Deltour et al., 1996). Another possible explanation for the similarities seen between ethanol-exposed embryos and those exposed to vitamin A (retinol) deficient diets is that ethanol interacts with components of RA signaling. Zachman and Grummer have shown in their investigations (reviewed in Zachman and Grummer, 1998) that RA receptors (RARs) and cellular retinol binding protein are disregulated by ethanol in several developmental systems.

It is well established that RA is essential for normal limb development (Helms et al., 1994, 1996; Mic et al., 2004). During the initiation phase of limb bud development, the ectoderm expresses Fgf8. The mesenchyme proliferates in response to Fgf8, and produces Fgf10, which in turn maintains the ectoderm. As limb initiation concludes, the domain of Fgf8-expressing ectoderm narrows as a result of dorsal and ventral ectoderm signals, forming a discrete domain of specialized ectoderm at the ventro-distal margin of the limb bud. With the outgrowth of the limb bud, the influence of Fgf8 is restricted to a distal, mitotically active subpopulation of mesenchyme beneath the developing AER, the subridge mesenchyme (SRM). Fgf8 continues to maintain expression of Fgf10 in the mesoderm (Mahmood et al., 1995; Vogel et al., 1996; Ohuchi et al., 1997; Moon and Capecchi, 2000), which is necessary for SRM proliferation and AER maintenance. The SRM appears to require the RA-dependent factor, Tbx5, to maintain Fgf10 expression; failure of Tbx5 expression is associated with failed AER maintenance and formation.

RA also has a role in the function and maintenance of the ZPA. It was initially described as the signal from the ZPA that induced ectopic digits when grafted to the anterior margin of a host limb. The proximate polarizing signal was later identified as Shh, a RA-inducible factor. Shh is stimulated by and dependent upon RA (Helms et al., 1994; Stratford et al., 1996). There is some evidence that the RA dependency of Shh may be mediated by dHand, a posteriorly restricted mesenchymal transcription factor. Localization of Shh depends on dHand in the posterior mesenchyme, and also on Fgf8 secreted from the posterior AER (Lewandoski et al., 2000; Moon and Capecchi, 2000). Alterations in the ZPA and the AER have been investigated regarding their role in ethanol-induced limb malformations. In addition to the work of Kotch et al. (1992) showing excessive cell death in the AER, Chrisman et al. (2004) demonstrated that perturbations in the AER and ZPA are accompanied by the loss of their respective molecular markers, Fgf8 and Shh.

Other studies also suggest that perturbations in Shh signaling may be important to ethanol-induced malformations. In addition to the limb, Shh is important in the development of other structures, including the CNS and craniofacial region (Wilson and Maden, 2005; Motoyama, 2006). Ahlgren et al. (2002) showed that Shh is decreased in craniofacial regions following exposure to ethanol, and rescue of the ethanol-induced phenotype was possible with application of Shh protein. Given the similarity between the limbs of Shh null mutants and those exposed to ethanol, and the ability of Shh to diminish the detrimental effects of ethanol in other embryonic structures, a Shh-mediated pathogenesis of ethanol-induced limb malformations seems likely.

As an upstream regulator of Shh, perturbation of RA may underlie ethanol-induced limb malformations. However, because of the dissimilarities between limbs exposed to ethanol and those permanently lacking normal RA signaling, this has not been previously investigated. Limb reduction defects related to RA signaling failure as described in RAR-α and RAR-γ double knockout mice (Lohnes et al., 1994) include malformations of the scapulae, radius, carpals, and digits. Mice lacking the major RA synthetic enzyme in the limb, RALDH2, that have been given limited dietary supplementation or gavage administration of RA exhibit a similar range of limb defects (Niederreither et al., 2002).

As previously noted, defects induced by acute ethanol exposure at early stages of limb bud outgrowth typically include postaxial ectrodactyly and other reduction defects. For the current study, we sought to determine whether temporary abrogation of RA signaling results in limb defects comparable to those resulting from ethanol exposure. To accomplish this, a pan-RAR antagonist, BMS-189453 (Schulze et al., 2001), and Disulfiram, an ALDH inhibitor (Vallari and Pietruszko, 1982), were utilized (Stratford et al., 1996, 1997; Xavier-Neto et al., 1999). Because excessive cell death in the AER is detectable within hours of ethanol exposure, the presence of abnormal cell death in the limbs of ethanol-, Disulfiram-, or BMS-189453-exposed embryos was compared in order to ascertain whether a common pathogenesis exists. Subsequently, the hypothesis was tested that RA could restore normal limb development to embryos exposed to ethanol. To establish a link between ethanol exposure and abrogation of RA signaling, the RA-dependent genes Tbx5, dHand, and Shh were analyzed using in situ hybridization and real-time PCR (qPCR). Transcripts were examined within hours of ethanol exposure in order to minimize ambiguity regarding the cellular targets of ethanol. These experiments provide evidence that RA-mediated limb development is, indeed, impacted by ethanol exposure.



C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and housed in a temperature- and humidity-controlled vivarium on a 14 h light, 10 h dark cycle. Mice were mated at the beginning of the light cycle at 8 am and inspected for a vaginal plug at 10 am. The presence of a plug was designated as gestational day (GD) 0:0.

In Vivo Exposure to Ethanol

At day 9, hours 6 (9:6) and 9:10, pregnant mice were given intraperitoneal (ip) injections of 25% (v/v) ethanol in phosphate buffered saline (PBS). The ethanol solution was administered at 0.015 mL/g of maternal body weight, resulting in a teratogenic dose of 2.9 g/kg (Sulik et al., 1981). Control animals were treated with two injections of PBS at 0.015 mL/g on GD 9:6 and 9:10. At selected times the embryos or fetuses were removed from the uteri and placed in PBS for subsequent analyses.

In Vivo Exposure to BMS-189453

BMS-189453, a gift from Bristol-Myers Squibb (Wallingford, CT), has been verified as a pan-RAR antagonist (Chen et al., 1995; Yang et al., 1999), effective in embryonic limb bud cells (Ali-Khan and Hales, 2006). Mice were given a single ip injection of 50 mg/kg BMS-189453 in DMSO; ip injection (0.005 mL/g of maternal body weight) was performed on day 9:10, based on pilot studies of time-dependent effects (data not shown). Control animals were treated with DMSO at 0.005 mL/g at day 9:10. At the time selected for observation, the embryos or fetuses were removed from the uteri and placed in cold PBS.

In Vivo Exposure to Disulfiram

Mice were given a single 75 mg/kg ip injection of Disulfiram in DMSO on day 9:6 of pregnancy. Injection volumes were 0.005 mL/g of maternal body weight. This dose was determined based on pilot studies of dose-dependent effects (data not shown). Control animals were treated with DMSO at 0.005 mL/g at day 9:6. At the time selected for observation, the embryos or fetuses were removed from the uteri and placed in cold PBS.

Skeletal Morphology

GD 18 fetuses dissected from extraembryonic membranes were transferred to, and maintained in 95% (v/v) ethanol for at least 3 days. When ready to stain, the head, skin, and viscera were removed from the trunk. The trunks were transferred to a 0.015% Alcian blue solution in 75% (v/v) ethanol and 20% (v/v) glacial acetic acid for 3–4 days of staining. After clearing in a 1% (w/v) KOH solution for 1–2 days, they were stained for 1 day in 0.025% (w/v) Alizerin Red in 1% KOH. Finally, stained and cleared skeletons were rinsed once in water and transferred to a 1:1 solution of glycerin and 70% ethanol (v/v) for photography and storage. Six litters for each of the PBS and DMSO control groups were examined; 10 litters were examined for each experimental group. The numbers of resorptions, live fetuses, and defects were noted for each litter, and the type and severity of limb malformation(s) were noted for each fetus. Percentages of live fetuses at GD 18 and live fetuses with defects were calculated; the percentages of postaxial ectrodactyly, intermediate ectrodactyly, and other defects were determined.

Nile Blue Sulfate (NBS) Staining

NBS is a vital stain that is sequestered into apoptotic bodies and phagolysosomes found in cells neighboring apoptotic cells (Allen et al., 1997). At GD 9:14 or GD 10:0 embryos were dissected from extraembryonic membranes and transferred to a 1:50,000 solution of NBS (Kotch et al., 1992) in lactated Ringers solution at 37°C. Embryos were then incubated for 30 min with agitation every 10 min. Following staining, embryos were rinsed in cold Ringers solution and the right forelimbs photographed using a Nikon D70 camera mounted on a Leica DMRB microscope. At least three embryos from each of three litters were stained and photographed for visual comparison.

Whole-Mount TUNEL Staining

Following treatments, embryos at GD 9:14 were dissected from extraembryonic membranes and transferred to 4% paraformaldehyde and kept overnight at 4°C. Embryos were rinsed three times in PBS and then washed three times for 30 min in PBX (PBS + 1% Triton X). Embryos were preincubated 30 min at 37°C in TdT labeling buffer. The labeling reaction was conducted using the reaction mixture provided by Trevigen (Gaithersburg, MD) for 3 h at 37°C. After a series of PBX washes, a 1:500 dilution of Streptavidin-Fluorescein (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) solution was used to label the TdT enzyme. Embryos were washed and visualized on a TE300 Nikon fluorescence microscope and photographed with a RT Slider Spot Camera. Three embryos from each of three litters were stained and photographed for visual comparison.

LysoTracker Red (LTR) Staining

To examine the ability of RA administration to prevent ethanol-induced cell death, pregnant mice were treated on day 9:6 with ethanol alone (as described above) or with ethanol plus 2.5 mg/mL RA in corn oil. RA was administered by gavage 30 min following the first ethanol injection. Control groups included a vehicle control (corn oil and PBS) or a dose of 25 mg/kg RA and 0.015 mL/g PBS (a subteratogenic dose in the 6J strain, unpublished observations). Three embryos from three litters of each control or experimental group were stained with LTR. For visualization of cell death patterns, a procedure previously published by Zucker et al. (2000), Zucker and Jeffay (2006), and Zucker (2006) was used for confocal microscopic imaging of LTR staining. Stained specimens were cleared in a solution of 1:2 (v/v) benzyl alcohol and benzyl benzoate and then sealed in specially made aluminum slides. The right forelimbs of embryos were initially visualized using a fluorescent microscope and representative specimens were chosen for subsequent visualization by confocal microscopy. The specimens were imaged using a Leica laser scanning confocal microscope (TCS-SP) with a 10X objective. The LTR dye was excited using the 568 nm laser line; the emission fluorescence was observed between 580–630 nm. Specimens were approximately 1.2 mm thick and were analyzed at 20 μm intervals. Using Leica software, data were prepared for presentation as a maximum projection.

Probe Preparation and Whole-Mount In Situ Hybridization

At GD 9:12 the embryos were removed from the uteri and placed in cold PBS. Three stage-matched embryos from each of three litters were utilized to provide a reliable sample of embryos. Antisense RNA probes were hybridized to embryos as described by Correia and Conlon (2001). The probes for dHand, Shh, and Tbx5 were kindly provided by E. Olsen, E. Michaud, and V. Papaioannou, respectively. Three embryos from each of three litters were stained and photographed for visual comparison. Representative embryos were chosen for illustration.

RNA Extraction, Reverse Transcription, and Real-Time PCR

Right forelimb buds from a single litter were obtained for quantitative real-time PCR (qPCR) at 6 or 18 h following treatment or vehicle ip injection. Limb buds were put into a microtube, and immediately placed on dry ice before transferring to −80°C storage. Five samples (one litter per sample) were used for each treatment or control group. Total RNA was extracted from each sample using the RNeasy Mini Kit (Qiagen) according to the provided protocol. Traces of genomic DNA were removed using the RNase-free DNase Set (Qiagen). RNA was eluted in 40 μL of RNase-free water. For reverse transcription, 20 μL of reaction mixture from the High Capacity cDNA Archive Kit (Applied Biosystems) was mixed with 400 ng of sample RNA in 20 μL of water. Reverse transcription reactions were run according to the suggested protocol.

TaqMan Universal PCR MasterMix and probes (Applied Biosystems, Foster City, CA) were used for the PCR step: Cyp26a1 - Mm00514486_m1; dHand - Mm00439247_m1; Fgf8 - Mm00438921_m1; Shh - Mm00436527_m1; Tbx5 - Mm00803521_m1. Amplification and detection were performed using the ABI PRISM 7700 Sequence Detection System (Applied Biosystems) with the following profile: one cycle at 94° for 10 min, and 40 cycles each at 95° for 15 s and 60° for 1 min.

The threshold cycle (Ct) was determined for each sample and primer combination. The relative expression of each mRNA was calculated by the comparative Ct method, using the value obtained by subtracting the average Ct value of GAPDH mRNA from the Ct value of each mRNA: the ΔCt. ΔCt calculations were based on mean GAPDH Ct values for each treatment or control group. ΔΔCt values were obtained by subtracting the ΔCt value of a 0 h group from that of each treatment or control group. Data are represented as fold change: 2−(ΔΔCt). t test comparisons were conducted on ΔCt values. p values <.05 were considered statistically significant.


Skeletal Morphology

As evident in skeletons stained with Alcian blue and Alizarin red (Fig. 1), each experimental exposure resulted in limb reduction defects. Ethanol administration on GD 9:6 and 9:10 resulted in 67% of fetuses with forelimb defects per litter. Of the affected limbs, 71% exhibited postaxial ectrodactyly, 21% had an intermediate ectrodactyly, and 8% had other limb defects (Table 1). GD 9:6 Disulfiram exposure resulted in 30% of fetuses per litter with a forelimb defect. Postaxial ectrodactyly was found in 68% of the affected limbs. Intermediate ectrodactyly was observed in 21% of forelimbs with defects, and the remaining defects were seen in 11% of the affected limbs. Preliminary studies indicated that BMS-189453 administration on GD 9:6 resulted in no limb defects. However, exposure on GD 9:10 resulted in 72% of fetuses per litter with a limb defect. Of the affected limbs, 41% had postaxial ectrodactyly, 13% had intermediate ectrodactyly, and 47% had other remaining limb defects. These remaining limb defects mainly consisted of large spacing between adjacent digits, a type of defect that was present in all three treatment groups. Control fetuses exposed to PBS or DMSO did not exhibit any limb defects. Ulnar agenesis or distal deficiencies accompanied postaxial loss of three or more digits. In these limbs, there was also a noticeable decrease in the overall size of the limb. Defects of radius, humerus, and scapula were not found in any treatment group. In summary, ethanol, Disulfiram, and BMS-189453 treatments each produced an array of forelimb malformations common to all treatment groups.

Figure 1.

Skeletal staining reveals abnormalities in the forelimbs of GD 18 C57BL/6J mice. Postaxial defects are common in mice exposed to ethanol (b), the RAR antagonist BMS-189452 (c), or the ALDH inhibitor Disulfiram (d) during the 9th day of development. Defects are not found in vehicle control groups (a).

Table 1. Effects of Chemical Exposure on Forelimbs of GD 18 Fetuses
Treatment groupLitters examinedLive fetuses examinedMean per liter
ImplantationsLive fetusesLive with defects
  1. As described in the Materials and Methods, ethanol was administered at GD 9:6 and 9:10. BMS-189453 was administered at GD 9:10 and Disulfiram at GD 9:6. Plus/minus sign indicates the standard deviation.

Ethanol10707.0 (±2.38)7.0 (±2.71)4.7 (±3.23)
Control – ethanol vehicle5307.2 (±0.84)6.0 (±1.23)0.0 (±0.00)
BMS-189453 vehicle10567.8 (±1.14)5.6 (±2.50)2.4 (±2.12)
Control – BMS-189435 vehicle5348.2 (±0.45)6.8 (±1.30)0.0 (±0.00)
Disulfiram10446.8 (±1.23)4.4 (±2.72)1.3 (±1.83)
Control – disulfiram vehicle5419.2 (±0.84)8.2 (±1.48)0.0 (±0.00)

Cell Death

Limbs of control embryos exhibited very little NBS staining in the developing AER at GD 9:14. However, excessive NBS staining was obvious in GD 9:14 embryos exposed to ethanol. At 18 h after the onset of ethanol exposure (GD 10:0) a high level of staining was evident relative to control groups (Fig. 2b). Staining was found at the distal apex of the AER near the center of the preaxial-postaxial axis. A minority of specimens had a slight postaxial shift of this staining pattern.

Figure 2.

NBS staining of GD 10 embryos demonstrates excessive cell death (arrowheads) in the presumptive AER of the forelimb following exposure to ethanol (b), Disulfiram (c), or BMS-189453 (d). A small amount of staining is present in vehicle-exposed embryos (a). Ethanol was administered on GD 9:6 and 9:10; Disulfiram was administered on GD 9:6; BMS-189453 was administered on GD 9:10. The left side of images is the cranial aspect of the embryos; the right forelimb is shown.

BMS-189453-treated embryos exhibited a similar localization of NBS to the AER at GD 9:14 and GD 10:0 (Fig. 2c). While ethanol and BMS-189453 both produced similar NBS localization, ethanol-treated embryos were more intensely stained than those treated with BMS-189453. Disulfiram treatment resulted in a cell death pattern similar to BMS-189453 and ethanol treatment by GD 10:0 (Fig. 2d), but this pattern was not distinguishable from that in control limb buds at GD 9:14.

Prevention of Ethanol-Induced Cell Death by RA

As with NBS, LTR staining was evident within 8 h of exposure to ethanol (Fig. 3c). Staining of the AER was intense in ethanol-exposed embryos, contrary to that in vehicle- and RA-treated embryos (Fig. 3a,b). LTR staining was present in the mesenchyme of limb buds of ethanol-treated embryos; a pattern not observed using whole-mount TUNEL or NBS staining techniques. Control and RA-treated limb buds had very little LTR staining in the AER or mesenchyme. Embryos exposed to both ethanol and RA exhibited a greatly reduced LTR-positive domain, compared to those embryos exposed to ethanol alone. However, the staining of the proximal limb mesenchyme observed in ethanol-exposed embryos was not prevented with RA cotreatment. Whole-mount TUNEL and NBS staining techniques provided identical findings with respect to the AER of control and treatment group limbs.

Figure 3.

Three-dimensional reconstruction of serial optical sections through mouse embryos stained with LysoTracker Red (LTR), obtained by confocal laser scanning microscopy. LTR staining of GD 9:14 embryos demonstrates the prevention of ethanol-induced cell death by RA in the presumptive AER. Vehicle- (a) and RA-treated controls (b) have a small amount of LTR staining (orange). GD 9:6 and 9:10 ethanol treatment results in intense staining along the presumptive AER (c). Embryos exposed to both ethanol and RA (d) show reduced staining in the same region. Arrowheads indicate localized LTR staining in the presumptive AER. Green autofluorescence results from gluteraldehyde fixation to allow visualization of the embryo. The right forelimb is shown, with the cranial aspect of embryos towards the top of the page.

In Situ Hybridization and Real-Time PCR

In the forelimb buds of GD 9:12 embryos, dHand was expressed in the postaxial mesenchyme. It was also present in the lateral plate mesoderm proximal and caudal to the limb bud. In most GD 9:12 embryos exposed to ethanol for 6 h, the dHand expression domain was noticeably smaller and less intense, relative to embryos exposed to PBS (Fig. 4a,b). A minority of treated embryos were not obviously different from controls. qPCR indicated that dHand is not significantly decreased by ethanol relative to controls at GD 9:12. At GD 10:0, there appeared to be a slight reduction in transcripts (p = .084; Fig. 5). Coadministration of ethanol and RA resulted in a significant reduction of dHand transcripts relative to ethanol alone at GD 10:0 (p = .040). BMS-189453 did not have an effect on dHand transcription at either time.

Figure 4.

In situ hybridization of vehicle- and ethanol-exposed embryos. dHand expression is shown in embryos exposed to PBS (a) or ethanol (b) for 6 h. Tbx5 expression is shown in embryos exposed to PBS or ethanol for 6 (c,d) and 18 h (e,f). Shh expression is shown in embryos exposed to PBS (g) or ethanol (h) for 18 h. Arrowhead indicates faint Shh expression in ethanol-exposed embryo.

Figure 5.

Quantitative real-time PCR analysis of transcripts found in limb buds exposed to ethanol or BMS-189453 alone, or in conjunction with RA. Data demonstrate the similar effects of ethanol and BMS-189453 on RA-dependent genes, important to limb development. Coadministration with RA has variable effects relative to ethanol or BMS-189453 exposure: dHand was significantly lower (p < .05) in RA and ethanol cotreatments than ethanol alone; Fgf8 and Shh were significantly lower (p < .05) in RA and BMS-189453 cotreatments than BMS-189453 alone. Data were converted to fold change and normalized to a no-treatment control group at GD 9:6. *p < .05.

Shh expression was undetectable in the limb buds of GD 9:12 control embryos by in situ hybridization. However, by GD 10:0 a strong signal was present on the postaxial margin of the limb bud (Fig. 4g). Most embryos exposed to ethanol exhibited little or no observable Shh expression at GD 10:0. In the few instances in which Shh staining was observed in ethanol-exposed embryos a weaker and smaller expression domain was present than in the limbs of comparably staged control embryos. qPCR demonstrated that expression of Shh is extremely low at GD 9:6, the time of initial ethanol exposure (Ct = 32 cycles). In control embryos, 6 h later (GD 9:12), expression increased fivefold before climbing an additional 12.8-fold by 14 h (GD 10). Ethanol decreased Shh expression substantially at 6 h of exposure, relative to control embryos (p = .006). By 18 h, the difference was not statistically significant (Fig. 5b). BMS-189453 had no effect on Shh expression after 2 h of exposure (GD 9:12), but by GD 10:0, expression was substantially decreased (p = .009).

Tbx5 expression was present throughout the limb buds of GD 9:12 control embryos and those exposed to ethanol. Ethanol had no discernable effect on the level or distribution of Tbx5 expression in the limb at this time (Fig. 4c,d). However, by GD 10:0, Tbx5 expression was substantially higher in the limb buds of control embryos, compared to the expression of Tbx5 in the limbs of ethanol-exposed embryos (Fig. 4e,f). qPCR confirmed these findings; no substantial change was evident at GD 9:12 following ethanol exposure, but by GD 10:0, Tbx5 transcripts were decreased in the ethanol-exposed limbs (p = .021; Fig. 5c). BMS-189453 had no effect on Tbx5 expression at any time examined.

Fgf8 expression was assayed by qPCR. Ethanol-exposed limb buds contained approximately 26% fewer transcripts than found in control limb buds at GD 9:12, but this finding lacked statistical significance (p = .070). By GD 10:0, expression levels were comparable between ethanol-exposed and control limb buds (Fig. 5d). As with dHand and Tbx5 transcription, Fgf8 was unaffected by BMS-189453 within 14 h (GD 10:0). At GD 10:0, both Fgf8 and Shh were significantly reduced in BMS-189453 and RA cotreatment groups relative to BMS-189453 alone (p = .010 and .038, respectively).

To confirm the RAR antagonist activity of BMS-189453, Cyp26a1 expression was examined, because this gene is sensitive to changes in RA (White et al., 1997). In the presence of BMS-189453, Cyp26a1 expression was decreased relative to the control group at GD 10:0 (p = .002; Fig. 5e). BMS-189453 and exogenous RA cotreatment resulted in Cyp26a1 expression levels intermediate to control and BMS-189453 treatment groups.

Ethanol had no significant effect on Cyp26a1 expression (GD 9:6, p = .2731; GD 10:0, p = .3871), but in combination with RA had a robust effect, increasing expression 18-fold.


This study supports the hypothesis that ethanol perturbs RA-mediated limb development. Unlike previous investigations involving a long-term attenuation of the RA signal in the limb bud, this study demonstrates that a single exposure to a RAR antagonist or ALDH inhibitor results in a narrow range of digital defects. The induction of these defects is associated with dramatically increased cell death in the developing AER. Additionally, the induction of excessive cell death by a RAR antagonist or ALDH inhibitor demonstrates a role for RA in AER maintenance.

Coadministration of RA prevents ethanol-induced excessive cell death, demonstrating that the induction of cell death is either upstream of RA function or mediated by perturbed RA signaling. In addition, the nascent ZPA, as identified by the RA-inducible genes dHand and Shh, is negatively impacted by ethanol exposure. The common changes in Shh expression in the developing limb combined with common pathogenic changes and abnormal phenotype produced by ethanol and a RA antagonist support the hypothesis that one facet of ethanol teratogenesis is, at least in part, mediated by disruption of RA function during development. Further studies are needed in order to better define the extent to which such a perturbation contributes to dysmorphogenesis; it is possible that parallel mechanisms of teratogenesis contribute to or enhance the effects of ethanol on RA function. An examination of the early transcriptional changes following ethanol exposure (prior to 6 h) may reveal whether RA or other mediators play a central role in the mechanism of ethanol teratogenesis.

The observed distribution of cell death in the developing AER is typical of previous descriptions of teratogenesis following ethanol administration on the 9th day of gestation in mice (Kotch et al., 1992). However, the mechanism(s) whereby ethanol impacts cellular and developmental targets resulting in changes essential to produce malformations is unknown. The excessive cell death in the AER that occurs following ethanol exposure may result from the direct toxicity of ethanol or from perturbation of supporting cell populations, the ZPA and SRM. In favor of the latter hypothesis, Disulfiram and BMS-189453 induce excessive cell death in the AER, presumably as a result of its requirement for RA-dependent supporting cell populations. Because of the close similarity between the pathogenesis of ethanol, BMS-189453, and Disulfiram exposures, cell death occurring by a RA-dependent mechanism is favored for ethanol, rather than a direct toxicity. Prevention of ethanol-induced excessive cell death in the AER by exogenous RA further supports this hypothesis.

Though recognized as a component in the pathogenesis of ethanol-induced limb defects, there has been no study to date investigating the secondary effects of excessive AER cell death on other limb tissues. Because removal of the AER results in apoptosis in the SRM (Sun et al., 2002) and a great number of AER cells die following ethanol administration, it is surprising that cell death is not also prevalent in the SRM. Gene expression is down-regulated in the mesenchyme following ethanol exposure (Tbx5, dHand, Shh). Only Shh depends on signals from the AER for its transcription. Detailed studies over a number of developmental stages will be necessary to examine the mesenchyme in regions of overlying AER cell death.

Given the intensity of ethanol-induced cell death in the AER at GD 9:12, a reduction of Fgf8 expression would be expected. However, as determined by qPCR, this was not found. This may be accounted for by the fact that the AER represents only a small proportion of cells in the limb buds used for qPCR, and that the technique requires the pooling of limb buds of several embryos, all of which are not equally affected by ethanol exposure. Indeed, in situ hybridizations, like those conducted by Chrisman et al. (2004) at later stages following ethanol exposure, did demonstrate the loss of Fgf8 expression in regions where the AER is deficient.

Because the AER positions and maintains the ZPA, cell death in the posterior regions of the AER would be expected to reduce Shh expression and cell proliferation in the ZPA. Indeed, as shown in this study, Shh expression is reduced within 6 h of ethanol administration, and Kotch et al. (1992) demonstrated that posterior mesenchyme fails to form in GD 12 limb buds. Despite these observations, excessive cell death is infrequently observed in the ectoderm overlying the nascent ZPA. This suggests that the ZPA may be a target of ethanol, independent of the effects on the AER. Further precisely timed experiments will be necessary to discern the subpopulation that is the proximate target of ethanol.

As previously noted, the present study reveals that ethanol down-regulates Shh expression shortly after the onset of exposure. Whether ethanol-induced reduction of Shh transcripts results from the attenuation of Shh expression or from reduced numbers of Shh-expressing cells remains to be determined. Others have suggested that limb defects arising from GD 9 teratogen exposure disrupt the AER-ZPA epithelial-mesenchymal interaction without directly interfering with the transcription of Shh (Bell et al., 1999, 2005; Scott et al., 2005). Only when posterior mesenchymal cell loss becomes evident following acetazolamide or cadmium exposure is Shh transcription lowered. A previous investigation regarding the effects of ethanol demonstrated a decrease in Shh expression, at 24–48 h after initial ethanol exposure (Chrisman et al., 2004), at a time when postaxial tissue loss would be evident. The results presented herein indicate that ethanol decreases Shh expression well before the loss of posterior tissue occurs and suggest that ethanol-induced dysmorphogenesis involves interference with a process that drives production of the postaxial phenotype. In this light, it is notable that the proximal mesenchyme is both a location of ethanol-induced cell death and RA synthesis (Swindell et al., 1999; Mic et al., 2004). Excessive cell death in this particular subpopulation of cells represents one possible route to decreased RA availability in the limb.

As expected (White et al., 1997), RA increased expression of the RA catabolic enzyme, Cyp26a1. Predictably, BMS-189453 blocked this effect. Exogenous RA, on the other hand, decreased expression of Shh and Fgf8 in the presence of the RAR antagonist, suggesting that the RA-mediated down-regulation or toxicity of these transcripts occurs by non-RAR mediated pathways.

Given that exogenous RA prevents the cell death induced by ethanol in the distal limb bud, it is perhaps paradoxical that RA diminished dHand gene expression in ethanol-exposed limb buds. One possible explanation is that Cyp26a1 is differentially expressed throughout the limb bud. Because the effects of RA are concentration dependent, it is likely that the distal limb contains a means to eliminate toxic levels of RA (MacLean et al., 2001), while the proximal limb does not. The proximal region, where Cyp26a1 is not usually expressed, encompasses the domain of dHand expression, while its distal domain encompasses the developing AER.

The available number of RA-dependent genes known to play a role in limb development has been a limiting factor in the current study. While Shh was decreased by BMS-189453 exposure, other genes that are reported to be RA-dependent, such as dHand and Tbx5, were not regulated by RAR antagonist treatment as observed within 14 h. Frequently time points of 2 or more days post-treatment are employed in examination of changes in gene expression resulting from experimental manipulations. While observations at these later time periods result in dramatic changes in gene expression, they do little to confirm the nature of the relationship between developmental regulators. The difference in the effects of the RAR antagonist and ethanol on Tbx5 and dHand expression may reflect the difference in the timing, distribution, or relative effectiveness of the two agents. Alternatively, these genes are not RA-dependent within the time frame examined.

While others have demonstrated the role of RA in activating ZPA gene expression (Riddle et al., 1993; Fernandez-Teran et al., 2000), this investigation has demonstrated the necessity of RA for AER maintenance. That at least two important regions of the limb bud, the developing AER and dHand and Shh-expressing posterior mesenchyme, are affected within hours of ethanol exposure suggests the possibility of a common, proximal pathogenesis. As dictated by minor, but significant differences in developmental stage, or other variations among embryos, ethanol may preferentially impact one cell population over another, or the same population at different stages, thus giving rise to a variety of defects of the digits.

Finally, Disulfiram, known as the prescription drug Antabuse, causes limb defects in this mouse model and cell death in the AER by GD 10:0. The teratogenicity of Disulfiram has been noted previously by Webster et al. (1983), although no sensitive region of the embryo was identified. Commonly prescribed to alcoholics to discourage alcohol consumption, Disulfiram prevents alcohol metabolism by inhibiting ALDH, resulting in acetaldehyde accumulation and nausea. Although a link between Disulfiram exposure and human limb defects has been previously suggested (Nora et al., 1977), this finding has received little clinical attention. Because alcohol and Disulfiram are concurrently present in humans, the teratogenesis of the latter is difficult to establish. Results of the present study strongly suggest that Disulfiram is, indeed, teratogenic and should be avoided during pregnancy.


The authors would like to thank Dr. George Breese for critical reading of the manuscript and Carmen Wood for her assistance with the qPCR.