Studies on the pharmacokinetics and metabolism of a gamma-secretase inhibitor BMS-299897, and exploratory investigation of CYP enzyme induction.

BMS-299897 is a gamma-secretase inhibitor that was effective in reducing amyloid beta-peptide (A beta) in transgenic mice and guinea pigs. Therefore, pharmacokinetic and drug metabolism studies were conducted in animals to support its clinical development. The compound appeared to have low to intermediate total body clearance and was orally bioavailable (24-100%). The oral absorption of BMS-299897 from solid dosage forms appeared to be dissolution rate-limited. BMS-299897 was distributed into extravascular space (V(ss) >or= 1.3 l kg(-1)), including brain (brain-to-plasma ratio = 0.13-0.50). BMS-299897 appeared to be a P-glycoprotein (P-gp) substrate as the brain-to-plasma ratio was two-fold higher in the mdr1a knockout mouse as compared with the wild-type. Apparent autoinduction by BMS-299897 was observed in murine and rat efficacy and toxicity studies. In vitro, BMS-299897 was a weaker inducer of cytochrome P450 3A4 (CYP3A4) and a weaker transactivator of human pregnane X receptor (hPXR) as compared with rifampicin. Induction of human UGT1A and UGT2B was evaluated in primary human hepatocytes, but the results were inconclusive. A low potential for autoinduction in humans was predicted at a clinical dose of 250 mg and the prediction was consistent with the findings from a clinical multiple-dose study with BMS-299897 in probable Alzheimer's patients.

Workshop overview: scientific and regulatory challenges for the reduction, refinement, and replacement of animals in toxicity testing.

Public concern for animal welfare has been expressed through legislative control of animal use for experimental purposes since the first legislation was introduced in 1876 in the United Kingdom. Legislative control of animal use has been introduced in virtually every developed country, with major initiatives in Europe (1986) and the United States (1966 and 1985). Advances in scientific thinking resulted in the development of the concept of the three Rs--refinement, reduction, and replacement--by Russell and Burch in 1959. The field has expanded substantially since, with specialist scientific journals dedicated to alternatives, World Congresses organized to discuss the scientific and philosophical issues, and European and U.S. validation organizations being launched. Current scientific attention is focused on validation of alternative methods. The underlying scientific principles of chemical toxicity are complicated and insufficiently understood for alternative methods for all toxicity endpoints of importance in protecting human health to be available. Important lessons have been learned about how to validate methods, including the need to have prediction models available before the validation is undertaken, the need to understand the variability of the animal-based data which is to be used as the validation standard, and the need to have well-managed validation programs. Future progress will depend on the development of novel methods, which can now be validated through international collaborative efforts.

Predicting In vivo Toxicity.

The objective of toxicity testing is to generate information that will help establish the safety of new products. Successful new in vitro tests will identify those endpoints and cell systems which provide evidence of pathologic change that can be used in determining safety. Chronic in vivo toxicity studies usually document adverse changes that are indicative of cell death in a target organ. For example, leakage of tissue specific enzymes is indicative of membrane damage and correlated with microscopic observation of cell death. Tissues, like the liver, can mount an effective regenerative response to moderate cell loss, but cell proliferation may be accompanied by the possibility of carcinogenesis. The balance between cell death and cell proliferation can make the difference between recovery, organ failure or tumour induction. Consequently, in vitro studies of toxicity have often focused on cell death as an endpoint. Molecular events preceding the irreversible commitment of a cell to die or proliferate are currently being investigated as additional potentially useful endpoints for in vitro toxicity studies. As molecular mechanisms become better understood, the number of potential endpoints for in vitro toxicity testing has greatly increased. However, at the earliest stages of interaction with the cell, it is difficult to predict whether changes induced by the xenobiotic are reversible (pharmacologic) or irreversible (pathologic) change. Changes in second messenger systems, binding to receptors, or early gene induction may explain the induction of toxicity by agents previously studied in vivo, but are potentially reversible changes that may not be useful for predicting the safety of untested products. Later changes that can be more easily identified as irreversible may provide more useful endpoints. One example is the activation of caspases in the cell as a precursor to apoptosis.

Inhibition of cholesterol synthesis by squalene synthase inhibitors does not induce myotoxicity in vitro.

The cholesterol-lowering HMG CoA reductase inhibitors (HMGRI), pravastatin and lovastatin, have been associated with skeletal myopathy in humans and in rats. In a previous in vitro study, HMGRI-induced changes in neonatal rat skeletal muscle cells were characterized by reversible inhibition of protein synthesis and loss of differentiated myotubes at concentrations markedly lower than those inducing enzyme leakage. Myotoxicity was determined to be directly related to inhibition of HMG CoA reductase, since mevalonate, the immediate product of HMG CoA reductase metabolism, abrogated the drug-induced changes. Farnesol, geranylgeraniol, and squalene are metabolites of mevalonate. Squalene, formed from farnesol by squalene synthase, is the first metabolite solely committed to cholesterol synthesis. In contrast, geranylgeraniol, formed by the addition of an isoprene group to farnesol, is the first metabolite uncommitted to cholesterol synthesis. The objective of the present study was to determine the role of inhibition of cholesterol synthesis in HMGRI-induced in vitro myotoxicity. HMGRI-treated neonatal rat skeletal muscle cultures were supplemented with farnesol and geranylgeraniol, and in another study, muscle cultures were exposed to two squalene synthase inhibitors (SSI), BMS-187745 and its prodrug ester, BMS-188494. Endpoints evaluated for both studies included protein synthesis ([3H]leucine incorporation), total cellular protein (a measure of cell loss), intra- and extracellular lactate dehydrogenase activity (a measure of membrane integrity), cholesterol biosynthesis ([14C]acetate incorporation), and morphology. HMG CoA reductase inhibitor-induced morphologic changes and inhibition of protein synthesis were significantly ameliorated by supplementation with farnesol and geranylgeraniol. In contrast to HMGRI-induced in vitro myotoxicity, SSI induced an irreversible, minimal cytotoxicity at close to maximum soluble concentrations. These results indicate that depletion of metabolites of geranylgeranyl pyrophosphate, and not inhibition of cholesterol synthesis, is the primary cause of HMG CoA reductase-induced myotoxicity.

HMG CoA reductase inhibitor-induced myotoxicity: pravastatin and lovastatin inhibit the geranylgeranylation of low-molecular-weight proteins in neonatal rat muscle cell culture.

In previous studies, inhibition of cholesterol synthesis by HMG CoA reductase inhibitors (HMGRI) was associated with myotoxicity in cultures of neonatal rat skeletal myotubes, and rhabdomyolysis in rats, rabbits, and humans in vivo. In vitro myotoxicity was directly related to HMGRI-induced depletion of mevalonate, farnesol, and geranylgeraniol, since supplementation with these intermediate metabolites abrogated the toxicity. Both farnesol and geranylgeraniol are required for the posttranslational modification, or isoprenylation, of essential regulatory proteins in mammalian cells. The objective of the present study was to measure changes in protein isoprenylation in cultured neonatal rat skeletal muscle cells exposed for 24 hr to increasing concentrations of pravastatin or lovastatin. Proteins were labeled with [3H]mevalonate, [3H]farnesyl pyrophosphate (FPP), or [3H]geranylgeranyl pyrophosphate (GGPP), and then separated by SDS-PAGE and quantitated by scintillation counting and densitometry of autoradiographs. Mevalonate and FPP labeling of the majority of proteins increased in a concentration-dependent manner, even at concentrations greater than 2 microM lovastatin and 25 microM pravastatin that completely inhibited cholesterol synthesis. In contrast, mevalonate and FPP labeling of three protein bands with molecular weights of 26.6, 27.7, and 28.9 kDa was markedly inhibited at concentrations higher than 1 microM lovastatin and 400 microM pravastatin, which inhibited protein synthesis and disrupted myotube morphology after longer exposures in a previous study. In contrast, these proteins were equally well labeled by GGPP at all HMGRI concentrations tested, suggesting that isoprenylation of the 26.9-, 27.8-, and 28.9-kDa proteins requires geranylgeraniol. The results of this study indicate that HMGRI-induced myotoxicity is most likely related to reduced posttranslational modification of specific regulatory proteins by geranylgeraniol.

Analysis of the steady-state dynamics of organelle motion in cultured neurites: putative indicator of neurotoxic effect.

1. The objective of this study was to develop a physiologically based method to evaluate the neurotoxic potential of drug candidates in vitro. Rat embryo midbrain cells were grown in micromass culture, and the movement of mitochondria labelled with the fluorescent dye rhodamine 123 was quantified in fasciculated neurites, using a laser cytometer. 2. The rhodamine 123 signal in a defined region of fascicle was quantified and photobleached with the laser. A series of post-photobleach scans revealed the movement of fluorescent-labelled mitochondria into the bleached region from adjacent unbleached regions. Recovery of fluorescence is a measure of the size of the mobile pool of mitochondria relative to the total (moving plus stationary) pool. 3. The steady-state levels of fluorescence recovery was dependent on intracellular calcium and magnesium concentrations, energy status (ATP), and microtubule integrity (post taxol or vinblastine treatment). 4. This technique may be a useful indicator of neurotoxic effect.

In vitro myotoxicity of the 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors, pravastatin, lovastatin, and simvastatin, using neonatal rat skeletal myocytes.

Pravastatin, lovastatin, and simvastatin, drugs which lower cholesterol by inhibiting 3-hydroxy-3-methylglutaryl coenzyme A (HMG CoA) reductase, have been linked to skeletal myopathies in humans and rats. The myotoxicity of these three drugs was compared, after 48 hr exposure, in cultures of primary neonatal rat skeletal myotubes. Measurements included HMG CoA reductase activity ([14C]acetate incorporation into cholesterol), indicators of membrane damage (CPK, LDH, and AST), cell viability (mitochondrial dehydrogenase metabolism of MTT), protein synthesis ([3H]leucine incorporation), and energy status (ATP). All three drugs inhibited cholesterol synthesis to the same extent in rat hepatocytes (IC50s approximately 0.07 microM). Lovastatin- and simvastatin-induced inhibition of cholesterol synthesis in myotubes was unchanged compared to that of hepatocytes, but pravastatin was 85-fold less potent (IC50 = 5.9 microM). Protein synthesis and ATP levels were the most sensitive indicators of toxicity. Pravastatin (IC50 = 759 microM) was > 100-fold less inhibitory of protein synthesis than lovastatin (IC50 = 5.4 microM) or simvastatin (IC50 = 1.9 microM). Addition of mevalonic acid (the immediate product of the HMG CoA reductase reaction), as 100 microM mevalonic acid lactone, reversed the toxicity of all three drugs. Removal of serum for 24-72 hr did not alter the toxicity of any of the drugs compared to cultures containing 10% serum, suggesting that differences in protein binding did not account for the differences in toxicity of the drugs. These results indicate that pravastatin is less myotoxic than lovastatin or simvastatin in this in vitro system using neonatal rat skeletal muscle cells, and this differential toxicity is correlated with the selective decrease in inhibition of HMG CoA reductase by pravastatin in nonhepatic tissues.

In vitro studies of the toxicity of nucleoside analogues used in the treatment of HIV infection.

The nucleoside analogues, 3'-azido-3'-deoxythymidine (AZT, zidovudine), 2',3'-dideoxyinosine (ddI, didanosine) and 2',3'-dideoxycytidine (ddC, zalcitabine), used in the treatment of human immunodeficiency virus (HIV) infection, have been associated with a number of dose-limiting toxicities in clinical studies. These include myelotoxicity (AZT), myopathy (AZT), peripheral neuropathy (ddC, ddI) and pancreatitis (ddI). Myopathy, peripheral neuropathy and pancreatitis are also observed in HIV-infected patients who have not been treated with the nucleoside analogues. Thus, nucleoside analogue toxicity can be confused with the adverse effects of HIV infection. Animal models exist for some, but not all, aspects of nucleoside analogue-related toxicity. In vitro studies have been used extensively to elucidate the mechanisms of nucleoside analogue toxicity. Cellular purine and pyrimidine metabolizing enzymes phosphorylate the analogues, which can then interact with DNA polymerases. Inhibition of one of these, HIV reverse transcriptase, is responsible for the antiviral activity of the nucleoside analogues. Toxicity is caused by inhibition of nuclear or mitochondrial DNA polymerases (or both) and by chain termination of replicating DNA at the point of insertion of the nucleoside analogue. The different toxicities observed in the case of each nucleoside analogue are most likely explained by different affinities for each of the cellular DNA polymerases.

In vitro tests for teratogens: desirable endpoints, test batteries and current status of the micromass teratogen test.

Information from in vitro tests can be usefully used as a component of the risk/hazard assessment process. In vivo studies will be required to confirm the in vitro data. If the in vitro test system is designed around endpoints that reflect changes following in vivo toxic insult then it may be possible to modify the in vitro system to account for some of the discrepancies observed between in vivo and in vitro outcomes. When the discrepancy can be accounted for by low bioavailability in vivo, pharmacokinetic studies may be required to determine the relevance of the in vitro toxic concentrations. Reproductive hazard, especially teratogenicity, has been the subject of intensive in vitro test development. The observation of teratogenicity may affect the development of new products more significantly than any other type or category of reproductive toxicity. The micromass test, involving culture of differentiating rat embryo limb and midbrain cells exposed to test agents, may be useful as part of a battery of in vitro tests for teratogens. The most recent protocol for the micromass test is described, followed by a summary of validation and mechanistic studies confirming its usefulness. The test is robust in its transfer to new laboratories. Interlaboratory variability is small.

Characterization of rabbit primary proximal tubule kidney cell cultures grown on Millicell-HA membrane filters.

With the aim of developing a kidney cell culture system that can be used to assess renal toxicity in vivo, freshly isolated rabbit proximal tubules were plated on Millipore cellulose filters mounted in plastic inserts (Millicell-HA). DNA synthesis peaked on day 6 of culture and cells reached confluency by days 12-14. The integrity of the monolayer was confirmed by exclusion of [(14)C]inulin and cell viability demonstrated by linearity of protein synthesis over a 24-hr period. In confluent cultures, the organic anion, [(14)C]p-aminohippuric acid (PAH) and cation [(14)C]tetraethylammonium bromide (TEA) were shown to be transported from the basolateral to the apical side at a rate 5-6 times greater than that from the apical to basolateral side during the first 60 min of exposure. Probenecid decreased PAH transport by 60% and N-methylnicotinamide and quinine inhibited TEA transport by 40 and 56%, respectively. Uptake of [(14)C]alpha-methylglucopyranoside into the cells was three times greater when label was added to the apical side than when label was added to the basolateral side. Apical uptake of glucose was sodium dependent and inhibited by more than 90% with phloridzin. Thus, kidney proximal tubule cells in the filter insert culture system display functional polarity which appears to mimic function in vivo and may be useful for examining mechanisms of nephrotoxicity.

Teratogenicity of phenylhydantoins in an in vitro system: molecular orbital-generated quantitative structure-toxicity relationships.

1. The ability of 20 mono- and di-phenylhydantoin derivatives to inhibit differentiation of rat embryo mid-brain and limb bud cells in culture has been used as an index of the teratogenic hazard represented by these compounds. 2. Molecular orbital calculations on these compounds, using the MINDO-3 (modified intermediate neglect of differential overlap) and CNDO-2 (complete neglect of differential overlap) methods, were combined with indices of teratogenicity in the two cell types, to generate a coherent structure-toxicity relationship. 3. Teratogenicity correlated with frontier orbital electron density of the N1 hydantoin ring atom (HOMO-N1) in a sub-series of 12 monophenylhydantoins, whereas the corresponding toxicity for both mono- and di-phenylhydantoins related more to the molecular polarizability (alpha mol) of the molecule. 4. Furthermore the same structural parameter (alpha mol) exhibited a parallelism with log P values of these 20 compounds, indicating the importance of lipophilicity in the toxicity of these compounds. 5. Overall, the data emphasize the ability of electronic structural calculations to identify chemical descriptors of toxicity.

Localization and identification of cytochrome P-450 in differentiating rat embryo cells in vitro and in foetal tissue in vivo by immunocytochemistry.

The presence of the isoenzymes b, e and c of cytochrome P-450 in foetal rat limb-bud and mid-brain tissue has been investigated in vivo and in micromass cell cultures of limb-bud and mid-brain cells derived from rat embryos by a sensitive immunocytochemical technique. The cytochromes could not be detected by antibody staining at the start of the culture nor in 13-day-old embryos from which cultures were prepared. Two different antibodies directed against cytochrome P-450 revealed the ontogenic profile of the phenobarbitone-inducible b and e forms, which appeared at an earlier stage of development, day 1 of culture (equivalent to day 14 of gestation), than did the 3-methylcholanthrene-inducible c form, which appeared on day 3 of culture (equivalent to day 16 of gestation). These isoenzymes were not tissue specific. Comparison of the localization and intensity of staining of cells cultured in vitro for 5 days with tissue from the equivalent foetal developmental stage (day 18) in vivo revealed the presence of cytochrome P-450 in corresponding areas. In day 18 limb sections, cytochrome P-450 was localized in the perichondrial and myogenic tissue, which corresponded to the cells in the periphery of the chondrogenic foci in vitro. In mid-brain whole tissue, the enzyme was located in connective tissue and neurofibrils, corresponding to cells in the periphery of the foci of neurones in vitro. The correlation between in vitro and in vivo observations from time course, location and quantitative aspects, illustrated that the micromass culture technique is a valid model for metabolism studies with these specific isoenzymes.

Inducibility and functionality of rat embryonic/foetal cytochrome P-450: A study of differentiating limb-bud and mid-brain cells in vitro.

The presence of constitutive levels of cytochrome P-450 isoenzymes in cultures derived from rat embryo limb-bud (LB) and mid-brain (CNS) cells was demonstrated immunocytochemically by staining with specific monoclonal and polyclonal antibodies of cytochrome P-450. The b and e forms of cytochrome P-450 were found to be non-inducible by either in vitro co-incubation for 5 days or by transplacental maternal induction with phenobarbitone (PB), 3-methylcholanthrene (3MC) or beta-naphthoflavone (betaNF) in either cell type. Consistent with this lack of response was the observation that both in vitro and in vivo inducer treatment did not alter the toxicity of the teratogens diphenylhydantoin (DPH) or cyclophosphamide (CPA). In contrast, 3MC induction was achieved by both in vitro and transplacental regimens as gauged by the increased intensity of peroxidase staining using a monoclonal antibody to cytochrome t-450 c, in both cell types. There was also a concomitant increase in DPH toxicity (>20% gauged by a decrease in IC(50) values) in LB cells by both induction regimens but the CNS cells were refractory. betaNF induction of cytochrome P-450 was observed following in vitro and in vivo exposures in both cell types. There was no modulation of DPH or CPA toxicity after in vitro exposure to the inducers, but in vivo induction caused a strong staining reaction in both cell types, commensurate with a 30% increase in DPH toxicity in LB cells and activation of the pro-teratogen CPA. The b and e forms of cytochrome P-450 were non-inducible but it is highly likely that the c form was both inducible (by 3MC and betaNF) and functional, the latter being assessed by modulation of DPH toxicity and CPA activation. It may be possible to induce cytochrome P-450 in cells derived from embryos. The system used may be suitable for detailed investigations of the types of metabolizing systems involved in the mechanisms underlying toxicity/teratogenicity.

In vitro toxicology: a commercial proposition?

1. The commercial value of any test system must be equated with its scientific value. In vitro toxicity tests only have scientific value when they are accurate models of toxicity in vivo. 2. Toxicity is a complex event so that the simple tests of cell viability are unlikely to be useful models as these tests often use cell lines which are functionally remote from cells in tissues of the whole animal. 3. Primary cultures retain their differentiated function for hours (hepatocytes) or days (nerve cells) and may acquire differentiated functions in vitro (embryo cells) and are the models of choice for predicting toxic hazard. 4. When the in vitro test is a satisfactory predictive model of toxicity there are sound commercial reasons for its use. It saves time and effort and can be used at an early stage in compound development to sort out the toxic from the non-toxic candidate compounds. Large numbers of structurally similar compounds can be tested (QSAR studies) and a greater understanding of the mechanism of toxic action can often be achieved than in vivo because of the numbers of animals required and the interplay of different organ systems in the toxic events that make interpretation difficult. Thus the in vitro test can not only save animals but can also save compounds.

Prediction of phototoxic potential using human A431 cells and mouse 3T3 cells.

An assay using established cell lines, human A431 epidermal cells and mouse 3T3 fibroblasts, has been developed to predict the phototoxic potential of compounds. The test determines the viability of the two cell lines in response to UV light in both the presence and absence of the test compound. The end point for cytotoxicity is determined from the mitochondrial dehydrogenase conversion of a tetrazolium salt (MTT) to a colored formazan product. The cytotoxicity of the test compound is established prior to UV exposure, and the highest no-effect concentration observed is then applied to cells that are subsequently exposed to different periods of UVA and UVA plus UVB light. A phototoxic effect is considered to have occurred when a biologically significant enhancement of toxicity is shown for the UV light with the compound present when compared to that of UV light alone. The test system has been validated with 30 compounds classified as strong, idiosyncratic, and negative based on the frequency of reported adverse reactions in humans. The in vitro phototoxicity assay was able to highlight the potential for phototoxicity in the strong category of phototoxic compounds and several of the idiosyncratic compounds. Only one of the negative compounds produced any activity in the assay in terms of enhancing UV toxicity. Some of the compounds were shown to protect the cells from the toxic effects of UV exposure.

An in vitro test for teratogens: its practical application.

An in vitro test for teratogens has been used successfully for more than 3 years. The method involves exposing undifferentiated rat embryo midbrain and limb cells to test compounds and observing the effect on subsequent cell differentiation. Experience of using the test has confirmed the accuracy of prediction (greater than 90%) suggested by a blind trial. The test has been used at the early stages of pharmacological evaluation in the selection of non-teratogenic pharmaceuticals and 250 compounds are tested on average each year. Maternal metabolism is modelled by the inclusion of Aroclor 1254-induced rat liver homogenate plus cofactors (S-9 mix). The concentrations of S-9 mix (50-100 microliters/ml culture medium) conventionally used in the Ames bacterial mutagenicity test are toxic to rat embryo cells, but greatly reduced concentrations (3-5 microliters/ml) are not toxic but are still able to activate pro-teratogens such as cyclophosphamide. However, most potentially teratogenic compounds tested are toxic in the absence of active preparations of drug-metabolizing enzymes. The conclusion that most teratogens are direct acting may be premature, since evidence has been found for drug-metabolizing activity in the embryo cells themselves.

In vitro metabolism of teratogens by differentiating rat embryo cells.

Rapid and accurate prediction of teratogenic hazard had been achieved using cultures of differentiating limb mesenchyme (LB) and midbrain (CNS) cells from 13-day-old rat embryos. In this study we have used these cultures to examine the role of metabolism in the in vitro teratogenic activity of diphenylhydantoin (DPH) and cyclophosphamide (CPA). Two approaches were used. The first involved modulation of cytochrome P-450 activity by co-incubation in vitro with a variety of inhibitors at concentrations that were non-cytotoxic to the cells. This enhanced the toxicity of DPH by 13-82% in LB and by 3-52% in CNS cells. Benzimidazole and ellipticine caused the greatest enhancement and SKF 525A the least. DPH appears to be the proximate teratogen and there appear to be embryo-tissue cytochrome P-450s that assist in its detoxification. Following prior transplacental induction, CPA was toxic in vitro to LB cells from beta-naphthoflavone-pretreated mothers. CPA was non-toxic in cells of control, phenobarbitone- or 3-methylcholanthrene-treated embryos. Thus there appear to be inducible levels of cytochrome P-448 in embryo cells. In the second approach, positive immunocytochemical staining of the cells with both monoclonal and polyclonal P-450 antibodies identified phenobarbitone, beta-naphthoflavone- and 3-methylcholanthrene-inducible cytochrome P-450s at a constitutive level. Cytochromes P-448 (beta-naphthoflavone type) and P-450 (phenobarbitone type, PB3 fraction) were inducible, confirming that cytochrome P-450s are in fact present in the embryo cells.

An in vitro assay for teratogens with cultures of rat embryo midbrain and limb bud cells.

A short-term in vitro assay for teratogens has been evaluated and shown to have a high predictability (greater than 90%). Cultured cells derived from midbrain (CNS) and limb buds (LB) of 34 to 36 somite rat embryos were exposed to 46 compounds (27 teratogens, 19 nonteratogens) in a blind trial. Rat liver post mitochondrial supernatant fraction plus cofactors were included in the cultures to provide metabolizing enzymes. Differentiation of neurons from CNS cells and chondrocytes from LB cells was measured after 5 days of culture. Inhibition of differentiation (assessed by reduction of number of foci) was the indicator of potential teratogenicity. Variation between experiments was limited. In repeat experiments with two direct-acting teratogens, aldrin and diphenylhyantoin, interexperiment variability was low (coefficient of variation; range 7 to 24%). Of 27 teratogens 25 (93%) and only 2 of 19 nonteratogens (11%) inhibited differentiation (CNS or LB). Inhibition of differentiation in one cell type alone was less predictive (CNS: 85%, LB: 82%).