Since the liver is the main organ involved in the metabolism and the toxicity of xenobiotics, isolated rat hepatocytes have been increasingly used in recent years as a model to identify pharmacological and toxicological responses of drugs. However, it is generally recognised that isolated hepatocytes retain most of their functions only for a short period. For this reason, numerous models and techniques have been developed to study and improve the metabolic capacity of hepatocytes in vitro over an extended time period and in application for drug metabolism studies. In the present study, we compared four different cell culture models to fulfill these requirements and have therefore harvested hepatocytes and cultured them in different culture systems over two weeks. In order to prove certain advantages or disadvantages of each model, we compared the metabolic capacity, albumin secretion, the release of cytosolic and mitochondrial enzymes, as well as the capacity to metabolise diclofenac (DF).
We found that rat hepatocytes in all studied culture models (except the Unisyn Bioreactor) were able to metabolise DF to the same extent as found in vivo. However, the concentration of metabolites was found to decrease with culture time using the monolayer although the DF metabolite level in the collagen Sandwich culture was higher than that of the monolayer culture. The 3D-membrane bioreactor preserved the metabolic capacity for a prolonged period of time. The concentrations of DF metabolites in the Unisyn hollow fiber bioreactor were below the detection limit, which corresponded to other parameters such as albumin secretion and cytochrome P450 activity, disqualifying this culture system clearly for the use of in vitro primary hepatocyte cultures. The other three systems all have their place in drug metabolism with different advantages. However, our studies clearly showed that hepatocytes cultured within a collagen sandwich or in the 3D-membrane bioreactor qualify to study various aspects of drug metabolisms over a long time period. Further studies are needed to prove if the later two culture models may really help to reduce animal testing.

Excised human skin has so far been considered to be one of the most suitable in vitro methods to evaluate the penetration of dermatologically applied substances. The limited supply and the relatively high donor variability stimulated many research groups to use animal skin as a substitute for human skin. Since nowadays reconstructed skin equivalents are commercially available, we examined these cultures for their suitability as a percutaneous absorption model for different pharmaceutical formulations. One such equivalent is EpiDermTM (EPI-606, MatTek corporation, Ashland Massachusetts) which was investigated using the lipophilic model drug flufenamic acid. Permeation studies with the Franz diffusion cell were undertaken to evaluate the model for the establishment of a new in vitro method to study the percutaneous absorption of different dosage forms. The drug was applied in two pharmaceutical formulations to the intact surface of the skin disk: dissolved in wool alcohol ointment (0.1125 %), and dissolved in Soerensen phosphate buffer pH 7.4 (0.1125% solution). HPLC was used for the analysis of drug content. It was shown that the model forms a barrier towards diffusion by comparing the permeation across the tissue-free inserts to the equivalents. Flux values were calculated and the permeation across the skin equivalent from the solution was noted to be almost forty times higher than from the ointment. Two different batches of the skin equivalent showed no statistically significant difference. Finally the permeability of the reconstructed skin was compared to human epidermis, and a five times higher flux value was found for the skin equivalent model.
Our results suggest that reconstructed skin equivalents based on human keratinocytes have potential as a pharmaceutical test system to study dermal drug transport from topical formulations.

In the Pharmeuropa Vol. 10, No. 4, December 1998, the European Pharmacopoeia Commission published an appeal to replace the pyrogen test by the limulus test for a total of 10 biological substances. These substances included the active ingredients aprotinin and urokinase, contained in our sales products AntagosanÃ’ and ActosolvÃ’. After successful validation of the limulus test, we calculated endotoxin limit concentrations for these and submitted these to the European Department for Quality of Medicines (EDQM). We expect - not least on the basis of the data submitted by ourselves - that the pyrogen test will be replaced in future as a quality control test by the limulus test in testing aprotinin and urokinase compounds for the presence of pyrogenic substances. This leads to a reduction in the number of rabbits used as opposed to the current pyrogen test for testing the stated compounds.

It is our objective to establish a virtual laboratory on the Internet to allow for an in silico estimation of harmful effects triggered by drugs, chemicals and their metabolites. Presently, our database includes validated models for five biological targets - the Aryl hydrocarbon, serotonin 5HT-2A, cannabinoid, GABA (gamma-amino butter acid), and steroid receptors. It shall be continuously extended to include surrogates for any bioregulator known or presumed to mediate harmful effects. Free access to this virtual laboratory shall allow any interested party to estimate the harmful potential of a given substance prior to its synthesis. This is achieved by generating the three-dimensional structure of the compound and its possible metabolites in the computer, followed by calculating their binding affinity towards each receptor surrogate in the database. Only compounds/metabolites passing through this surrogate battery without displaying a significant affinity towards any member may be cleared for synthesis and preclinical studies. This way, potentially harmful compounds can be withdrawn from the evaluation pipeline before in vivo test are conducted, hence contributing to the reduction of animal testing in chemical and pharmaceutical research and development.

It was suggested in the ICCVAM workshop that the Register of Cytotoxicity (RC), using in vitro cytotoxicity data to predict the in vivo starting doses, should be implemented into acute toxicity testing as soon as possible. The validity of the in vitro cytotoxicity data to establish appropriate starting doses for acute toxicity testing will be assessed experimentally. Secondly, in order to replace the use of animals in acute lethality testing a formal validation will be conducted in which the ability to predict rodent LD50 values and toxicity classes from cytotoxicity data will be evaluated.

Phototoxicity is an acknowledged property of some UV and/or visible light absorbing substances some of which are used as pharmaceuticals or in cosmetic preparations. In recent years attention has been called upon the fact that toxic intermediates that are generated upon photoactivation of a substance can also lead to DNA damage. Such damage may lead to mutated/ initiated skin cells which in turn can contribute to an elevated skin cancer risk. The method of choice to test for photo-related skin carcinogenesis is a 1-year study in genetically hairless mice in which the formation of skin papilloma and their latency time are assessed. Here, in vitro test approaches to test for photogenotoxicity can be used in a tiered assessment approach asking the use of in vitro genotoxicity tests for prediction of rodent/human carcinogenicity. In the past few years some effort has been put into the evaluation for such systems. In particular standard test protocols have been generated for the in vitro photo-micronucleus test and the in vitro photo-comet assay with Chinese hamster V79 cells. The data that have been produced so far show promising results regarding the implementation of these systems in a tiered approach for photocarcinogenicity assessment of UV- and/or visible light absorbing substances but the systems will have to be validated in further collaborative studies.

Therapy with the antiepileptic drug valproic acid (2-propylpentanoic acid, VPA) during early pregnancy can cause similar teratogenic effects (neural tube defects) in human and mice. In this study a new molecular bioassay is presented using following endpoints: differentiation of F9 teratocarcinoma cells, altered cell morphology, induction of possible targed genes, and the induction of viral RSV-promoter. The induction of a transiently transfected viral (RSV) promoter driven luciferase gene by VPA was used to screen a set of VPA-derivatives. Structure-activity investigations showed: the longer the aliphatic side chain the more the induction of the RSV-reporter gene. The specific induction was stereoseletive. The teratogenic enantiomer S-4-yn-VPA (2-propyl-4-pentynoic acid) induced the RSV-driven reporter gene while the non teratogenic R-4-yn-VPA does not. Heptyl-4-yn-VPA was the most potent teratogen in vitro and in vivo. Non teratogenic VPA-derivatives like R-4-yn-VPA and 2-en-VPA (2-propyl-2-pentenoic acid) were ineffective in this system. Thus, the teratogenic effect of VPA and VPA-derivatives in the mouse correlated with the specific induction of the viral RSV-promoter controlled reporter in F9-cells. Acid compounds such as fatty acids are known to interact with peroxisome proliferator-activated receptors (PPARs). To test structure-acitivity relationships by VPA or its derivatives we used CHO cells stably expressing hybrid proteins of the ligand-binding domain of either of the PPARs. The teratogen VPA and the teratogenic derivatives of VPA activated the PPAR-d construct in a very specific structure- and stereoselective way which correlated well with the activities in the reporter gene assay (bioassay) and those in vivo. No such correlation was found with respect to activation of PPAR-a or PPAR-g. These structure-activity relationships indicate that PPAR-d may be a potential mediator of VPA-induced differentiation of F9 cells and may possibly be involved in the mechanism of teratogenicity of VPA in vivo. Furthermore two bioassays were designed with clearly defined endoints, amenable to automation and screening of great number of compounds. The test system allows to replace animal experiments in the preclinical development of new antiepileptics drugs with reduced teratogenic risk. Supported by BgVV-ZEBET (Berlin).

Germ cell mutagenesis is required by the 7th amendment of the directive 67/548 EEC into the national regulations on existing chemicals. Officially accepted in vivo test systems for stage specific mutagenicity are the dominant lethal (DL) test and the specific locus test (SLT) in mice. An acceptable in vitro alternative designed to address germ cell mutagenesis and discriminate between male and female specific effects is not available at present. In order to offer a sensitive and predictive in vitro method to assess the genotoxic potential of chemical agents on male and female reproduction, we established primordial germ (PG) cell-derived permanent embryonic germ (EG) cell lines of the mouse (strain BALB/cJ). The differences in developmental sensitivity of the EG3 cell line and differentiated fibroblast cells 3T3 were comparatively tested with cytotoxicity assay (MTT test ) and genotoxic studies (SCE-assay) under identical test conditions. The concentration-response curves reflected the female cell line EG3 to be extremely sensitive concerning cytotoxic and genotoxic endpoints. Therefore this cell line was used to classify in vivo genotoxic and non-genotoxic test substances with different potential endpoints. Applying linear discriminant analysis three endpoints were identified for the correct classification (100%) of all test chemicals, namely the SCE200 value (increase of 200% in the mean number of SCEs per metaphase spread) for EG3 (3hrs and 24hrs assay) and the IC50 value for EG3 after 3hrs of exposure to test chemicals.