Immunohistochemical and microarray analyses of a mouse model for the smith-lemli-opitz syndrome.

The Smith-Lemli-Opitz syndrome is a mental retardation/malformation syndrome with behavioral components of autism. It is caused by a deficiency in 3beta-hydroxysteroid-Delta7-reductase (DHCR7), the enzyme required for the terminal enzymatic step of cholesterol biosynthesis. The availability of Smith-Lemli-Opitz syndrome mouse models has made it possible to investigate the genesis of the malformations associated with this syndrome. Dhcr7 gene modification (Dhcr7-/-) results in neonatal lethality and multiple organ system malformations. Pathology includes cleft palate, pulmonary hypoplasia, cyanosis, impaired cortical response to glutamate, and hypermorphic development of hindbrain serotonergic neurons. For the current study, hindbrain regions microdissected from gestational day 14 Dhcr7-/-, Dhcr7+/- and Dhcr7+/+ fetuses were processed for expression profiling analyses using Affymetrix oligonucleotide arrays and filtered using statistical significance (S-score) of change in gene expression. Of the 12,000 genes analyzed, 91 were upregulated and 98 were downregulated in the Dhcr7-/- hindbrains when compared to wild-type animals. Fewer affected genes, representing a reduced affect on these pathways, were identified in heterozygous animals. Hierarchical clustering identified altered expression of genes associated with cholesterol homeostasis, cell cycle control and apoptosis, neurodifferentiation and embryogenesis, transcription and translation, cellular transport, neurodegeneration, and neuronal cytoskeleton. Of particular interest, Dhcr7 gene modification elicited dynamic changes in genes involved in axonal guidance. In support of the microarray findings, immunohistochemical analyses of the netrin/deleted in colorectal cancer axon guidance pathway illustrated midline commissural deficiencies and hippocampal pathfinding errors in Dhcr7-/- mice. The results of these studies aid in providing insight into the genesis of human cholesterol-related birth defects and neurodevelopmental disorders and highlight specific areas for future investigation.

Abnormal serotonergic development in a mouse model for the Smith-Lemli-Opitz syndrome: implications for autism.

The Smith-Lemli-Opitz syndrome (SLOS) is a malformation/mental retardation syndrome resulting from an inborn error in 3beta-hydroxysteroid Delta7-reductase (DHCR7), the terminal enzyme required for cholesterol biosynthesis. Using a targeting strategy designed to virtually eliminate Dhcr7 activity, we have created a SLOS mouse model that exhibits commissural deficiencies, hippocampal abnormalities, and hypermorphic development of serotonin (5-HT) neurons. The latter is of particular interest with respect to current evidence that serotonin plays a significant role in autism spectrum disorders and the recent clinical observation that 50% of SLOS patients present with autistic behavior. Immunohistochemical analyses have revealed a 306% increase in the area of 5-HT immunoreactivity (5-HT IR) in the hindbrains of mutant (Dhcr7-/-) mice as compared to age-matched wild type animals. Amount of 5-HT IR was measured as total area of IR per histological section. Additionally, a regional increase as high as 15-fold was observed for the most lateral sagittal hindbrain sections. In Dhcr7-/- mice, an expansion of 5-HT IR into the ventricular zone and floor plate region was observed. In addition, the rostral and caudal raphe groups exhibited a radial expansion in Dhcr7-/- mice, with 5-HT IR cells present in locations not seen in wild type mice. This increase in 5-HT IR appears to represent an increase in total number of 5-HT neurons and fibers. These observations may help explain the behavioral phenotype seen in SLOS, and provide clues for future therapeutic interventions that utilize pharmacological modulation of the serotonergic system.

Selective vulnerability of embryonic cell populations to ethanol-induced apoptosis: implications for alcohol-related birth defects and neurodevelopmental disorder.

BACKGROUND: Ethanol-induced cell death has been characterized in very few stages of embryogenesis. This investigation comprehensively maps patterns of both programmed and ethanol-induced cell death in the central nervous system and craniofacial region at 0.5-day intervals from gestational day (GD) 6.5 to 11 in mice. METHODS: A teratogenic dosage of ethanol (2.9 g/kg) or vehicle was administered via two intraperitoneal injections to pregnant C57BL/6J mice at various stages of gestation. Cell death patterns were characterized using Nile blue sulfate vital staining and histological analysis of plastic sections. Confocal laser scanning microscopy of LysoTracker Red-stained specimens allowed for three-dimensional visualization of areas of apoptosis and precise sectional imaging of mouse embryos. Apoptosis was also documented using a TUNEL technique on histological sections. RESULTS: Normal programmed cell death in control embryos was noted in the prechordal plate region at GD 8, the neuroepithelium of the fourth ventricle and anterior neuropore at GD 9, and within the ganglia of cranial nerves V, VII-VIII, IX, and X at GD 10. Acute maternal ethanol administration 12 hr before examination resulted in a dramatic increase in apoptosis within sites of programmed cell death in the embryo. Moreover, ethanol-exposed specimens exhibited stage-dependent excessive cell death in other distinct cell populations, particularly within the developing central nervous system. Ethanol-induced apoptosis was notable as follows: GD 7.5-neuroectoderm; GD 8-neural plate and primitive streak; GD 9-alar plate and presumptive neural crest of the rostral hindbrain, especially at the mesencephalon/rhombencephalon junction; GD 9.5-10-branchial arches and rhombomeres; and GD 11-diencephalon, basal ganglia, pons, and developing cerebellum. CONCLUSIONS: The results of this study revealed developmental stage-specific cell populations of the developing brain and craniofacial region that are vulnerable to ethanol-induced apoptosis and provide new insight relative to the genesis of alcohol-related birth defects.

Programmed cell death in extraocular muscle tendon/sclera precursors.

PURPOSE: This study was designed to examine the occurrence of natural cell death in the periocular mesenchyme of mouse embryos. METHODS: Vital staining with LysoTracker Red and Nile blue sulfate as well as terminal nick end labeling (TUNEL) were utilized to identify apoptotic cell death in whole and histologicaly sectioned gestational day 10.5 to 14 mouse embryos. Laser scanning confocal microscopy was used to provide a three dimensional representation of the cell death pattern. Immunohistochemical staining for neural crest and myoblast populations was utilized to indicate the cell population undergoing apoptosis. RESULTS: Programmed cell death was evident in the developing rectus muscle tendons/sclera on gestational days 11 through 12.5 (corresponding to the weeks 5-6 of human development). Although each of these peripheral periocular condensations has readily apparent amounts of apoptosis, the pattern of cell death varied among them. Cell death was most apparent in the superior rectus tendon primordium, while that for the lateral rectus had the least evidence of apoptosis. CONCLUSIONS: Although apoptosis was readily evident in the periocular mesenchyme in distinct regions located medial and distal to the developing rectus muscles, programmed cell death in these sites has not previously been reported. New imaging techniques coupled with stains that evidence apoptotic cell death have made it possible to define this tissue as a prominent region of programmed cell death. Although neuronal tissues, including particular regions of the developing eye, are well recognized as sites of programmed cell death, description of this phenomenon in the extraocular tendon/sclera precursors is novel.

Iron-mediated free radical injury in ethanol-exposed mouse neural crest cells.

Previous studies using cell and whole embryo cultures have shown that free radicals play an important role in the ethanol-induced death of mouse neural crest cells (NCCs; a significant cell type with respect to the genesis of alcohol-related birth defects). This investigation was spurred by reports of increased iron in ethanol-exposed fetuses and the knowledge that iron can initiate the production of reactive oxygen species. Initially, the ameliorative potential of two iron chelators, deferoxamine and phenanthroline, relative to ethanol-induced cell death was examined. Cotreatment of cultured NCCs with 100 mM ethanol and either 1 or 10 microM deferoxamine or 10, 50, or 250 microM phenanthroline significantly increased the percentage of viable cells as compared with exposure to 100 mM ethanol alone. These data indicate that iron is involved in the ethanol-induced cytotoxicity. To support this premise, the direct toxicity of iron to NCCs was also examined. As expected, loading the cells with Fe(II)/Fe(III) using 8-hydroxyquinoline as a carrier had an adverse effect on their viability as did treatment with a neurotoxin, 6-hydroxydopamine, that releases iron from ferritin storage. Cotreatment with an antioxidant, N-acetylcysteine, significantly diminished the toxicity of ethanol alone, that resulting from iron loading, as well as from the combination of ethanol exposure and iron loading. These results confirm the role of free radical-mediated damage in ethanol-induced cytotoxicity and highlight the potential role of iron relative to the genesis of alcohol-related birth defects.

Differential sensitivity of mouse neural crest cells to ethanol-induced toxicity.

Neural crest cells (NCCs) have been identified as an important target population relative to ethanol-induced teratogenicity in both mouse and avian models. Additionally, whole embryo culture mouse models have shown strain-related differences in sensitivity to ethanol-induced damage following acute exposure during early NCC development. That differential sensitivity of NCCs may contribute to these strain differences has been unexplored. For this purpose, cultured NCCs from an inbred mouse strain (C57BL/6J; C57) that is more sensitive to ethanol-induced teratogenicity than an outbred strain (ICR) were compared. This study showed that the incidence of cell death was significantly higher for the C57 NCCs than those from the ICR strain at all ethanol concentrations tested, and as early as 12 hours after initial exposure to 100 mM ethanol. The lateral mobility of the membrane lipids was faster and the membrane GM1 content was lower in C57 cells than ICR cells both under control conditions and at all doses and times tested. Ethanol exposure resulted in significant increases in the membrane lipid lateral mobility, and decreases in the membrane GM1 content that occurred in a dose and time-dependent manner in the NCCs from both strains. A significant correlation was found between the GM1 content and lateral mobility of the membrane lipids, the lateral mobility of membrane lipids and cell viability, as well as the GM1 content and cell viability in the NCCs from both strains. These results suggest that different strain sensitivities to ethanol-induced teratogencity may lie, at least in part, in the interstrain differential response of the NCC population and that the vulnerability of the NCCs to ethanol-induced death may be related to their endogenous membrane GM1 content.

CYP2E1 is not involved in early alcohol-induced liver injury.

The continuous intragastric enteral feeding protocol in the rat was a major development in alcohol-induced liver injury (ALI) research. Much of what has been learned to date involves inhibitors or nutritional manipulations that may not be specific. Knockout technology avoids these potential problems. Therefore, we used long-term intragastric cannulation in mice to study early ALI. Reactive oxygen species are involved in mechanisms of early ALI; however, their key source remains unclear. Cytochrome P-450 (CYP)2E1 is induced predominantly in hepatocytes by ethanol and could be one source of reactive oxygen species leading to liver injury. We aimed to determine if CYP2E1 was involved in ALI by adapting the enteral alcohol (EA) feeding model to CYP2E1 knockout (-/-) mice. Female CYP2E1 wild-type (+/+) or -/- mice were given a high-fat liquid diet with either ethanol or isocaloric maltose-dextrin as control continuously for 4 wk. All mice gained weight steadily over 4 wk, and there were no significant differences between groups. There were also no differences in ethanol elimination rates between CYP2E1 +/+ and -/- mice after acute ethanol administration to naive mice or mice receiving EA for 4 wk. However, EA stimulated rates 1.4-fold in both groups. EA elevated serum aspartate aminotransferase levels threefold to similar levels over control in both CYP2E1 +/+ and -/- mice. Liver histology was normal in control groups. In contrast, mice given ethanol developed mild steatosis, slight inflammation, and necrosis; however, there were no differences between the CYP2E1 +/+ and -/- groups. Chronic EA induced other CYP families (CYP3A, CYP2A12, CYP1A, and CYP2B) to the same extent in CYP2E1 +/+ and -/- mice. Furthermore, POBN radical adducts were also similar in both groups. Data presented here are consistent with the hypothesis that oxidants from CYP2E1 play only a small role in mechanisms of early ALI in mice. Moreover, this new mouse model illustrates the utility of knockout technology in ALI research.

Widespread neuronal ectopia associated with secondary defects in cerebrocortical chondroitin sulfate proteoglycans and basal lamina in MARCKS-deficient mice.

Mice deficient in MARCKS, a prominent neural substrate for protein kinase C (PKC), die before or shortly after birth. They exhibit high frequencies of exencephaly, universal agenesis of forebrain commissures, and abnormalities of cerebral cortical and retinal lamination. We show here that these mice have wide-spread and severe neuronal ectopia in the outer layers of the developing forebrain, manifested by the migration of clusters of developing neuroblasts through the basal lamina and often through the pial membrane and into the subarachnoid space. This abnormality became apparent by Embryonic Day (E) 13 or 14, shortly after the formation of the early marginal zone. MARCKS deficiency was associated with decreased staining for marginal zone chondroitin sulfate proteoglycans; this decrease was detectable earlier in development than the neuronal ectopia. Later in development, there was also marked disruption of the basal lamina at the pial-glial interface, as evidenced by gross abnormalities in laminin and reticulin staining; however, the basal lamina appeared normal at E9.5. These data indicate that MARCKS is required for the prevention of neuronal ectopia during development. Potential mechanisms responsible for the neuronal ectopia in the MARCKS-deficient mice include decreased expression or increased proteolytic destruction of basal lamina proteins and marginal zone chondroitin sulfate proteoglycans in the developing brain.

Pathogenesis of malformations in a rodent model for Smith-Lemli-Opitz syndrome.

The fact that Smith-Lemli-Opitz syndrome (SLOS), a syndrome comprising major malformations involving a number of organ systems, results from an abnormality in cholesterol biosynthesis, was discovered only recently. Utilizing a drug (BM 15.766) to inhibit the same step in cholesterol biosynthesis as is abnormal in those affected with SLOS, we have developed a rat model that presents with abnormalities observed as early as gestational day 12 that appear to be consistent with some of those subsequent malformations that comprise the human syndrome. Abnormalities of the brain and face include deficiency in the midline region of the upper face, narrowing of the forebrain hemispheres and of the cerebral aqueduct, and deficiency in the developing lower jaw. Associated pathogenesis, as observed on gestational day 11 in histological sections and with scanning electron microscopy, involves abnormal cell populations at the rim of the developing forebrain and in the alar plate of the lower midbrain and hind-brain. The affected cells appear abnormally rounded up, having apparently lost their normal cell contacts. The potential basis for the selective vulnerability of this cell population and the impact of its vulnerability relative to subsequent dysmorphogenesis is discussed.

Limb, genital, CNS, and facial malformations result from gene/environment-induced cholesterol deficiency: further evidence for a link to sonic hedgehog.

Low cholesterol levels produced by treating cholesterol deficient mutant mice with a cholesterol synthesis inhibitor (BM 15.766) between days 4 to 7 of pregnancy resulted in malformations consistent with those in the Smith-Lemli-Opitz syndrome (SLOS). Facial anomalies in mildly affected gestational day 12 mouse embryos included a small nose and long upper lip; in more severely affected embryos, the facial and forebrain anomalies are representative of holoprosencephaly. Additionally, abnormalities of the mid- and hind-brain were observed and included stenosis of the cerebral aqueduct at the level of the isthmus and apparent absence of the organ progenitor for the cerebellar vermis. Although not previously directly linked to cholesterol deficiency in experimental animals, limb and external genital defects were a notable outcome in this multifactorially-based cholesterol deficiency model. The results of this study provide new evidence supporting an important role for cholesterol in early embryonic development, provide additional support for the hypothesis that this role may involve the function of specific gene products, such as sonic hedgehog (shh) signaling protein, and provide a description of the pathogenesis of some of the characteristic malformations in SLOS.

The membrane disordering effect of ethanol on neural crest cells in vitro and the protective role of GM1 ganglioside.

The teratogenic effect of ethanol appears to be related to excessive cell death in selected cell populations including craniofacial neural crest. Because there is a large body of evidence suggesting that a primary site of action of ethanol is at the membrane level, the current study was designed to examine and attempt to ameliorate ethanol-induced neural crest cell membrane changes that proceed cell death. To this end, neural crest cells were grown as primary cultures from mouse cranial neural tube be explants. In these cultured cells, the relationships between changes in membrane lipid lateral mobility (a measure of membrane fluidity) as determined using the technique of fluorescence recovery after photobleaching (FRAP), ethanol-induced cell death, and the protective role of GM1 ganglioside were examined. A dose-response study showed that treatment with 50, 100, 150, or 200 mM ethanol respectively, for 24 h was positively correlated with membrane lipid lateral mobility and negatively correlated with cell viability. Pre- or co-treatment of the cells with GM1 ganglioside diminished the ethanol-induced increases in membrane fluidity and decreases in cell viability. The results of this study suggest that change in membrane fluidity can account, in part, for ethanol-induced neural crest cell death and that the protection conferred by GM1 ganglioside may result from membrane stabilization and subsequent preservation of the biophysical properties and biological function of the ethanol-exposed cell membranes.

Essential role of beta-adrenergic receptor kinase 1 in cardiac development and function.

The beta-adrenergic receptor kinase 1 (beta ARK1) is a member of the G protein-coupled receptor kinase (GRK) family that mediates the agonist-dependent phosphorylation and desensitization of G protein-coupled receptors. We have cloned and disrupted the beta ARK1 gene in mice by homologous recombination. No homozygote beta ARK1-/- embryos survive beyond gestational day 15.5. Prior to gestational day 15.5, beta ARK1-/- embryos display pronounced hypoplasia of the ventricular myocardium essentially identical to the "thin myocardium syndrome" observed upon gene inactivation of several transcription factors (RXR alpha, N-myc, TEF-1, WT-1). Lethality in beta ARK1-/- embryos is likely due to heart failure as they exhibit a > 70% decrease in cardiac ejection fraction determined by direct in utero intravital microscopy. These results along with the virtual absence of endogenous GRK activity in beta ARK1-/- embryos demonstrate that beta ARK1 appears to be the predominant GRK in early embryogenesis and that it plays a fundamental role in cardiac development.

Developmental expression of MARCKS and protein kinase C in mice in relation to the exencephaly resulting from MARCKS deficiency.

The roles of protein kinase C and its substrates in development are poorly understood. Recently, we disrupted the mouse gene for a major cellular substrate for protein kinase C, the MARCKS protein (Proc. Natl. Acad. Sci. USA, 92, 944-948, 1995). The resulting phenotype consisted of universal perinatal lethality, agenesis of the corpus callosum and other forebrain commissures, and neuronal ectopia and other cortical and retinal lamination disturbances. These mice also had high frequencies of exencephaly (25% overall, 35% in females). In the present study, we have examined the normal expression of MARCKS and the various isozymes of protein kinase C at the time of cranial neural tube closure, in an attempt to correlate MARCKS expression in time and anatomical location with the exencephaly characteristic of MARCKS deficiency. Failure of neural tube closure occurred at various sites in the cranial neural tube, suggesting a cellular functional defect that was not limited to a specific location. Non-exencephalic MARCKS-deficient embryos appeared to be anatomically normal on embryonic day (E) 8.5-9.5. MARCKS and PKC alpha were expressed at the plasma membrane of the neuroepithelial cells comprising the future neural tube, as well as in the surface ectoderm and underlying mesenchyme. Endogenous protein kinase C species, comprising either or both alpha and delta, were capable of phosphorylating MARCKS in intact E8.5 embryos. Thus, MARCKS is expressed at the plasma membranes of the specific cell types involved in cranial neurulation; its deficiency presumably results in a still-to-be-elucidated functional defect in these cells that leads to exencephaly in a high proportion of cases.

Free radicals and ethanol-induced cytotoxicity in neural crest cells.

Associations between ethanol-induced cranial neural crest cell (NCC) damage in mammalian embryos and subsequent malformations as observed in human fetal alcohol syndrome have previously been illustrated. The vulnerability of NCCs to this teratogen may result, at least in part, from their sensitivity to free radical damage. To examine relationships between free radical generation and NCC cytotoxicity, primary culture of mouse NCCs was used. NCC viability was determined in both dose- and time-response studies involving ethanol exposure. After 48 hr of culture, cell viability was significantly diminished at all doses tested (i.e., 50, 100, 150, and 200 mM ethanol). At 100 mM ethanol (a dosage that is teratogenic in vivo and in whole embryo culture), cell viability decreased to approximately 50% of control values over the first 12 hr of culture, and decreased further, to approximately 20% by 48 hr. Using nitroblue tetrazolium as a probe, it was observed that exposure of NCCs to ethanol stimulated the production of superoxide anion radicals. Co-treatment of the ethanol-exposed NCCs with free radical scavengers including 300 units/ml of superoxide dismutase, catalase (500 units/ml), or alpha-tocopherol (300 microM) significantly improved NCC viability. These results suggest that the ethanol-induced NCC injury is mediated, at least in part, through the generation of free radicals. To test this hypothesis further, NCCs were exposed in culture to xanthine/xanthine oxidase. Exogenous free radicals generated by the xanthine/xanthine oxidase system resulted in reduced NCC viability, the severity of which increased in a time and enzyme concentration-related manner. Superoxide dismutase (300 units/ml) and catalase (500 units/ml) significantly reduced the effects of the xanthine/xanthine oxidase-generated free radicals on NCC viability. The similarity between the susceptibility of NCCs to ethanol and their susceptibility to exogenous free radicals in concert with the free radical scavenger-mediated amelioration of ethanol and exogenous free radical-induced NCC death strongly suggest that free radicals play a significant role in ethanol-induced NCC death.

Pathogenesis of caudal dysgenesis/sirenomelia induced by ochratoxin A in chick embryos.

Caudal dysgenesis/sirenomelia is a malformation complex for which the pathogenesis is controversial. This report describes the particular vulnerability of specific caudal structures to Ochratoxin A (OA), a fungal toxin, as the basis for caudal dysgenesis in an avian model. The experimental procedure involved injection of 1 microgram of OA into the air sac of eggs that had been incubated for 48 hours prior to treatment (i.e., embryos that had reached Hamburger and Hamilton stage 9-10 (6-10 somite pairs) [Hamburger and Hamilton (1951) Dev. Dyn. 195:231-272] by the time of treatment). Six to twelve hours following OA injection, excessive cell death, as shown by vital staining and routine histology, was evident in selected cell populations, including cells of the caudal-most mesoderm (the mesoderm that apparently forms the external genitalia and median infraumbilical region), the tail bud, and the neural tube caudal to the wing buds (corresponding to the level of the presomitic mesoderm). The notochord was not severely affected, although there were degenerative changes in the presomitic mesoderm. Except for positional abnormalities, development of the lateral plate mesoderm from which the leg buds are derived appeared relatively normal in most of the treated embryos. Six days post-treatment, varying degrees of caudal dysgenesis, presenting in severely affected specimens as sirenomelia, were observed in approximately 30% of the surviving treated embryos. The potential basis for the differential vulnerability of the affected cell populations and, therefore, the cellular basis for the genesis of caudal dysgenesis/sirenomelia in this model are discussed.

Brca1 deficiency results in early embryonic lethality characterized by neuroepithelial abnormalities.

The breast and ovarian cancer susceptibility gene, BRCA1, has been cloned and shown to encode a zinc-finger protein of unknown function. Mutations in BRCA1 account for at least 80% of families with both breast and ovarian cancer, as well as some non-familial sporadic ovarian cancers. The loss of wild-type BRCA1 in tumours of individuals carrying one nonfunctional BRCA1 allele suggests that BRCA1 encodes a tumour suppressor that may inhibit the proliferation of mammary epithelial cells. To examine the role of BRCA1 in normal tissue growth and differentiation, and to generate a potential model for the cancer susceptibility associated with loss of BRCA1 function, we have created a mouse line carrying a mutation in one Brca1 allele. Analysis of mice homozygous for the mutant allele indicate that Brca1 is critical for normal development, as these mice died in utero between 10 and 13 days of gestation (E10-E13). Abnormalities in Brca1-deficient embryos were most evident in the neural tube, with 40% of the embryos presenting with varying degrees of spina bifida and anencephaly. In addition, the neuroepithelium in Brca1-deficient embryos appeared disorganized, with signs of both rapid proliferation and excessive cell death.

Ethanol-induced teratogenesis: free radical damage as a possible mechanism.

To investigate the possibility of a free radical mechanism for ethanol-induced teratogenesis, gestational day 8 mouse embryos were exposed for 6 hr in whole embryo culture to a teratogenic dosage of ethanol alone (500 mg%) or in conjunction with an antioxidant, superoxide dismutase (SOD; 300 U/ml). For subsequent analysis, some embryos were examined at the end of this 6-hr period, while others were removed to control medium and cultured for an additional time period. Ethanol exposure resulted in increased superoxide anion generation and increased lipid peroxidation (as noted 6 hr after initial ethanol exposure) and in excessive cell death (as noted 12 hr after initial exposure) in the embryos. Following a total of 36 hr in culture, a high incidence of malformation, including failure of the anterior neural tube to close in 63% of the ethanol-exposed embryos, was noted. The ethanol-induced superoxide anion generation, lipid peroxidation, excessive cell death, and dysmorphogenesis were diminished in embryos co-treated with SOD, suggesting that the teratogenicity of ethanol is mediated, at least in part, by free radical damage.