Molecular multiple endpoint embryonic stem cell test--a possible approach to test for the teratogenic potential of compounds.

The embryonic stem cell test (EST) examines the cytotoxicity of chemical compounds on embryonic stem (ES) cells and 3T3.A31 fibroblasts. Additionally, the EST measures the ability of ES cells to differentiate into contracting cardiomyocytes following drug exposure. In this study, we introduce new endpoints to obtain a molecular multiple endpoint EST (mme-EST), enabling the identification of potential chemical effects on osteogenic, chondrogenic and neural differentiation in addition to the traditional endpoint of cardiomyocyte differentiation. Six compounds in three classes with known teratogenic in vivo potential were assayed with the mme-EST in a pilot study: penicillin G (non-teratogenic), 5-fluorouracil and retinoic acid (strongly teratogenic), diphenylhydantoin, valproic acid and thalidomide (moderately teratogenic). While the traditional EST measures a morphological endpoint, we included molecular markers of differentiation as endpoints. With the mme-EST, every compound could be classified correctly according to its known teratogenic potential in vivo. Penicillin G, 5-fluorouracil and diphenylhydantoin inhibited differentiation of all endpoints equally. Interestingly, valproic acid showed the strongest inhibition of neural differentiation, while thalidomide specifically inhibited osteogenic development. Retinoic acid, on the other hand, supported neural but inhibited chondrogenic and osteogenic differentiation concentration-dependently. Valproic acid and thalidomide, classified incorrectly with the established EST model, were classified correctly with the mme-EST according to their effects on specific endpoints. This pilot study indicates that the predictive value of the EST may be enhanced by including further differentiation endpoints.

Design and optimization of 20-O-linked camptothecin glycoconjugates as anticancer agents.

To improve the biological profile of 20(S)-camptothecin, a novel class of 20-O-linked camptothecin glycoconjugates has been designed for preferential cellular uptake into tumor cells by an active transport mechanism. Such conjugates have been optimized for enhanced solubility, stabilization of the camptothecin lactone ring, sufficient hydrolytic and proteolytic stability, and for an overall improvement in tumor selectivity. The constitution of the peptide spacer has a major impact on stability and biological activity of the conjugates both in vitro and in vivo. Glycoconjugates 17-22 with valine residues at the linkage position to camptothecin are sufficiently stable and show good antitumor activity in vitro against HT29 and other tumor cell lines. Fluorescence microscopy and flow cytometry experiments indicate that glycoconjugates such as 19 are taken up into lysosomal compartments of the tumor cell line HT29 by an active transport mechanism. The steric configuration of the particular amino acid residues linked to the camptothecin moiety has a major impact on the in vivo activity of the corresponding glycoconjugates in the breast cancer xenograft MX-1 model. Inhibiting tumor growth by >96%, the glycoconjugates 19 and 21 show the best activity in this particular model and have been investigated more extensively. The glycoconjugate 19 compares favorably to topotecan 4 and glycoconjugate 21 with respect to toxicity against hematopoietic stem cells and hepatocytes. Based on its profile, 19 has been selected for clinical trials.

Molecular markers in embryonic stem cells.

Embryonic stem cells are pluripotent cells derived from mouse blastocysts, which have the capacity to differentiate in vitro into a wide variety of cell types. Based on this potential the embryonic stem cell test (EST) has been developed, which represents an assay system for the classification of compounds for their teratogenic potential, based on the morphological evaluation of contracting myocard cells compared to the cytotoxic effects on undifferentiated stem cells and adult 3T3 fibroblasts. To expand the EST, the quantitative expression of the alpha- and beta-myosin heavy chain (MHC) genes under the influence of test compounds was studied employing real-time TaqMan PCR analysis. The molecular evaluation of the MHC genes allows a higher sensitivity for the classification of substances and the transfer of the EST to the molecular level allows to start experimental procedures at day 9 of culture. Thus, the modulated EST holds promise as a new easily quantifiable in vitro screening assay in teratology.

An in vitro model for peroxisome proliferation utilizing primary hepatocytes in sandwich culture.

Peroxisome proliferators comprise a group of structurally diverse chemicals which share as a common biologic effect the induction of peroxisomal fatty acid degrading enzymes. Concomitantly, the number and size of peroxisomes within hepatocytes increases. Following chronic administration some peroxisome proliferators act as non-genotoxic hepatocarcinogens in susceptible species such as rodents. To establish an in vitro model for the toxicological investigation of peroxisome proliferation, primary hepatocytes of rats, dogs and humans were cultivated in an organotypic cell culture model (sandwich model). By employing a panel of diverse compounds in this model a graded response was observed in the induction of carnitine acetyl transferase (CAT), the activity of which was determined as an endpoint. The following results were obtained in the order of decreasing inducing potential for rat hepatocytes: FOE 3798>nafenopin>fenofibrate (ciprofibrate>bezafibrate DEHP approximately ETYA>DEHA. Induction of CAT activity was generally higher than reported in earlier cell culture systems, probably reflecting the effect of the extracellular matrix provided by the collagen gel sandwich. In parallel, transcription of the rat CYP4A1 gene was induced by a similar order of magnitude as measured by TaqMan RT-PCR. In accordance with literature data, human and dog hepatocytes did not display such a strong and graded response but rather were not susceptible to this effect. In addition, (3)H-thymidine incorporation data demonstrated that nafenopin was able to induce DNA synthesis in rat hepatocytes whereas it did not in human hepatocytes.

Determination of Apoptosis in Primary Rat Hepatocytes by Real-time Quantitative PCR (TaqMan PCR).

Primary hepatocytes in collagen gel sandwich culture are a well suited model for studying effects like xenobiotic metabolism, hepatotoxicity, apoptosis, and others. Apoptosis is programmed cell death. It plays an essential role in embryonal development as well as in adult tissue for the elimination of undesired cells. The physiologic balance between growth and death can be disturbed by xenobiotics which may induce apoptosis. For the characterization of apoptosis, the expression of two genes, namely bax and p53 was analysed by quantitative real-time PCR. Following incubation of rat primary hepatocytes with camptothecin (CPT), a time- and concentration-dependent increase of mRNA expression was measured for bax at approximately 350% and p53 at approximately 600%. Further, camptothecin and topotecan (TPT) were compared for their effect on bax mRNA expression. Whereas camptothecin induced a continuous increase in bax transcription from 0.1 mum to 10 mum, only the highest concentration of topotecan (10 mum) led to a comparable increase of bax transcription. Our results demonstrate that apoptosis can be rapidly and simply quantitated in hepatocytes by TaqMan PCR.

3-D coculture of hepatic sinusoidal cells with primary hepatocytes-design of an organotypical model.

Models for cocultures of parenchymal (PC) and nonparenchymal cells (NPC) of the liver relied on mixing the cells in a two-dimensional configuration or on establishing spheroidal aggregates. In vivo hepatic nonparenchymal cells, such as endothelial cells and Kupffer cells, are separated from parenchymal cells by extracellular matrix (ECM). Due to their location outside of the space of Disse they can form a barrier toward the sinusoid. Hepatocytes are attached to ECM of the space of Disse via two opposing sinusoidal surfaces. No three-dimensional coculture model reflecting this specific microenvironment of the liver cell plates in vivo has been available to date. We designed a three-dimensional model by positioning NPC on top of PC enclosed as a monolayer within a collagen sandwich. A gas-permeable membrane support can be used to allow the supply of oxygen to the resulting cell plate also from underneath the cell layers. Morphological analysis was performed by inverse and cross-sectional studies by light microscopy, scanning, and transmission electron microscopy of the coculture model. Cuboidal hepatocytes formed confluent layers below the NPC layer. They regularly expressed bile canaliculi at intercellular contact zones. Both sinusoidal surfaces expressed microprojections. Characteristic NPC including endothelial cells, Kupffer cells, and Ito cells completely covered the second matrix layer within a week. Kupffer cells were located on top of endothelial cells. Ito cells were intermingled and could be identified by their intracytoplasmic lipid droplets. LPS stimulation of cocultures resulted in a depression of albumin secretion. Phase I and phase II metabolites of the cytochrome P-450 1A1 substrate ethoxyresorufin were generated independently from the presence of cocultured NPC. This study describes the development of a novel three-dimensional coculture model, which intends to mimic more closely the microenvironment of the hepatic sinusoid by respecting the specific plate structure of the liver parenchyma. The model could serve as a complex tool to study potential collaborations between PC and NPC of the liver.

Effects of fluoroquinolones and glucocorticoids on cultivated tendon cells in vitro.

Achilles tendon rupture is known to occur after administration of various drugs to patients. It is also reported to occur after treatment with fluoroquinolone and glucocorticoids. To study potential cytotoxic effects of fluoroquinolones (ciprofloxacin, pefloxacin, sparfloxacin) and triamcinolonacetonide on tendons an in vitro model of cultivated tendon cells (human, dog, mini-pig, rat, marmoset) was established. The cells were characterized by their morphological appearance and by the synthesis of proteoglycans and collagen type I. The cytotoxicity of the tested drugs was determined by measurement of mitochondrial dehydrogenase activity (MTT assay) and other parameters (viability, proteoglycan content, proliferation). These investigations revealed the following: (1) no marked differences were found between various species after ciprofloxacin treatment, only rat tendon cells reacted slightly less sensitively at the highest concentration (100 mug/ml); (2) no age-dependent effects of ciprofloxacin and triamcinolonacetonide were found in cultivated human tendon cells from patients of different ages; (3) the simultaneous administration of fluoroquinolones and triamcinolonacetonide resulted in a significantly greater reduction of cell viability than when they were administered alone. This could indicate that the combined administration of fluoroquinolones and glucocorticoids can favour the occurrence of tendon rupture.

Chondrotoxicity of quinolones in vivo and in vitro.

Chondrotoxicity is a rare toxicological finding which is observed in dogs after administration of quinolone antibacterials. To study this effect chondrocytes from articular cartilage of dogs were isolated, and incubated with quinolone derivatives. The effects on cell viability, mitochondrial dehydrogenase, and proteoglycan synthesis were determined. These results were compared with in vivo findings in dogs treated with these quinolones. It was concluded that inhibition of mitochondrial dehydrogenase activity and of proteoglycan synthesis are major reasons for cartilage damage. Therefore this in vitro model is capable of identifying strongly arthropathogenic quinolones without the need of performing animal studies.

A membrane-associated form of C-reactive protein is the galactose-specific particle receptor on rat liver macrophages.

Rat liver macrophages express a galactose-specific receptor which mediates endocytosis of particles or neuraminidase-treated blood cells. From rat serum we now have isolated and purified a galactose-specific lectin by affinity chromatography. Comparative analysis of this serum galactose-binding protein with the galactose-particle receptor protein purified from rat liver macrophages and with C-reactive protein (CRP) reveals close relation or identity of these proteins. An apparent m.w. of 30,000 was determined for all three proteins by SDS-PAGE under reducing conditions and m.w. of about 130,000 by native PAGE. All three proteins exhibit the same pentameric, ring-shaped structure in electron microscopy after negative staining. Antibodies raised against the serum galactose-binding protein or against the macrophage receptor cross-react. A mAb specific for rat neo-CRP labels liver macrophages but not hepatocytes and reacts with the isolated protein in a Western blot assay. Furthermore, the galactose-particle receptor can be functionally replaced by purified CRP: the binding capacity for neuraminidase-treated E of receptor-depleted liver macrophages can be restored by preincubation with purified rat CRP. We therefore conclude that CRP occurs as a membrane-associated protein constitutively expressed on liver macrophages functioning as a receptor mediating galactose-specific binding of particulate ligands.

Binding, uptake, and transcytosis of ligands for mannose-specific receptors in rat liver: an electron microscopic study.

We have investigated the initial distribution of mannose-specific binding sites in rat liver as well as the uptake and transcytosis pathways of ligands for this receptor in in situ and in vivo experiments. As ligands we used mannan adsorbed onto colloidal gold particles of sizes 5, 17, and 35 nm (Man-Au5, Man-Au17, or Man-Au35). The in situ binding pattern of Man-Au5 in the prefixed liver is identical to the one described earlier for galactose-exposing ligands in the same organ. With the exception of the binding by hepatocytes, where only scarce binding of Man-Au5 was observed, ligands were found adhering in a preclustered pattern all over the cell surface of liver macrophages and binding in aggregates over the coated pits of endothelial cells. In double-labeling experiments different particle sizes were used for glycoproteins with terminal mannosyl or galactosyl residues. This simultaneous localization of the two binding activities revealed that on endothelial cells the two activities are always found to be present in the same coated pit. On liver macrophages the clustered binding occurred at different membrane areas. Uptake and transcytosis of Man-Au5, 17, 35 were studied after their injection into the tail vein. Three and fifteen minutes after injection most of the Man-Au5 and all of Man-Au17 or Man-Au35 was found in sinusoidal liver cells, i.e., macrophages and endothelial cells. One hour after injection, endocytosed ligand is redistributed from large--presumably lysosomal--vacuoles to small noncoated vesicles that are localized predominantly near the space of Dissé. Between 1 and 40 h after injection, ligands of all sizes are transcytosed and found in the hepatocytes. No ligand accumulation is observed in hepatocytes as an indirect indication for secretion into bile. With this investigation we give evidence for transcytotic activity not only of liver endothelium but also of the resident liver macrophages.

Galactose-specific receptors on liver cells. I. Hepatocyte and liver macrophage receptors differ in their membrane anchorage.

Colloidal gold particles coated with asialoglycoproteins are bound by hepatocytes as well as by liver macrophages. Binding by both cell types is inhibited by N-acetylgalactosamine and related saccharides and is dependent on the presence of Ca2+. We have now performed an electron microscopic study on receptor anchorage in the plasma membranes. Cells with prebound ligand were treated with 20 mM EDTA at 4 degrees C, washed free of chelator and tested for residual galactose-specific receptor activity. Whereas hepatocytes preserve binding activity (73% of untreated control), liver macrophages lose galactose-specific receptor activity (12% of untreated control). Liver macrophages regain binding activity after a 2 min incubation at 37 degrees C allowing for receptor recycling. If the macrophages were fixed with low glutaraldehyde concentration prior to EDTA treatment they fully retained their receptor activity (74% of control). Ligands were also removed from both cell types by incubation with 80 mM N-acetylgalactosamine. After washing the cells free of the competing monosaccharide, both the hepatocytes as well as the macrophages show full binding activity (120% and 85% of untreated controls). Therefore, membrane anchorage sites of the macrophage receptors are not identical to ligand-binding sites. These results suggest a Ca2+-Mg2+-dependent receptor anchorage on the macrophage plasma membrane. As shown in the accompanying paper (Roos, P.H., Hartmann, H.J., Schlepper-Schäfer, J., Kolb, H. and Kolb-Bachofen, V. (1985) Biochim. Biophys. Acta 847, 115-121), EDTA-induced dissociation from the membrane can be used for isolation of the galactose-specific receptors of liver macrophages.