Screening of toxic compounds in mammalian cell cultures.

The cytotoxic concentrations of about 100 randomly selected drugs and chemicals, tested on HeLa cells in the MIT-24 system and/or in primary cultures of fetal chicken cells, were compared with the lethal doses and/or concentrations of the agents in the mouse and in man. Most agents (80%) had a similar toxicity in vitro and in vivo, suggesting a lethal interference in man with basal functions common to all specialized human tissues as well as cultured cells, i.e., basal cytotoxicity. This high frequency of basally cytotoxic agents opens possibilities for screening chemicals for toxicity and for studying cytotoxic mechanisms with a standard battery of a few appropriate cell tests. This battery may be used in three ways: (1) to study cytotoxic mechanisms of all chemicals, and apply the resulting knowledge to understanding toxicity in man of basally cytotoxic agents; (2) to supplement conventional animal tests in acute toxicity test programs; (3) to screen chemicals and extracts for their potential basal cytotoxicity. To validate these ideas and to select suitable tests for the battery, results from cytotoxicity tests on a wide variety of chemicals in several in vitro systems must be compared with one another and with the toxicity of the agents in animals and man.

Screening of toxic compounds in tissue culture.

To screen toxicity of chemicals most often easily manageable cultures of less differentiated cells have been used. This work includes 3 fields: (i) Screening of chemicals and fermentation broths for their cytoinhibitory effect, to predict antineoplastic activity. A related practical approach is to achieve optimal antitumour drug therapy by testing drugs on cultures of tumour cells from the patient. (ii) Screening of metal and plastic materials used in medicine, surgery and dentistry for their cytoinhibitory effect to predict local irritation. (iii) Screening of the mutagenicity or transformation capacity of chemicals in tissue culture, to predict their carcinogenicity. In addition, organ-specific cultures of most specialized cells (hepatocytes, ova, nerve cells, heart cells, skin cells, respiratory mucosa, and macrophages) have also been used to predict drug action on corresponding targets in the body. The author's group has focused on 2 new uses of standard cells for screening chemical toxicity: (i) Comparisons of in vitro cytotoxicity with in vivo toxicity of 85 randomly selected drugs indicated that for most drugs a systemic lethal action was brought about by cytotoxicity. A screening model is advocated by which results of cytotoxicity tests are compared with systemic toxicity in vivo to evaluate the systemic cytotoxicity of chemicals. (ii) Combinations of compounds with a cytotoxic lethal action in man indicated by the previous method have been screened in vitro for their combined systemic toxicity. By systematic comparison of results from standardized in vitro tests with in vivo toxicity, steps have been taken to resolve the question of the relevance of screening in tissue culture and to contribute to the development of an emerging subdiscipline to toxicology -- in vitro cytotoxicology.

Toxicity to HeLa cell of 205 drugs as determined by the metabolic inhibition test supplemented by microscopy.

The toxicity of 205 drugs to HeLa cells was evaluated by the Metabolic Inhibition Test supplemented by microscopy. Two end points of cyto-inhibition were estimated; total and partial inhibition after 24 h, based on absence or scarcity of spindle-shaped cells, respectively, and total and partial inhibition after 7 days, based on different degrees of basic pH change of the phenol red included in the cell medium. Direct drug-induced pH changes and precipitates in the cultures were also recorded. Many drugs were found to induce a culture zone with a particularly low pH after 7 days of incubation, at concentrations below the cyto-inhibitory concentration. Forty-three drugs regularly caused this hyperacid reaction, while 44 drugs caused the reaction irregularly. Since the reaction was always continuous with the cyto-inhibitory zone, it was provisionally judged to represent excitatory cell injury. Since many of the drugs which regularly induced the reaction are also known to induce proliferation of the endoplasmic reticulum in various cells, the reaction may be related to this and allied effects. Many drugs that are known to accumulate in cells displayed a high inhibitory toxicity, which might have been due to the paucity of cells in the test system. However, the high 7-day inhibitory toxicity shown by antineoplastic and some anti-inflammatory drugs, including triamterene and disulfiram, may constitute a genuine antimetabolic drug action. Seven grand mal antiepileptics were only very slightly toxic as compared with their precipitating tendency, which may be of significance in their therapeutic action.

Toxicity of 29 plasticizers to HeLa cells in the MIT-24 system.

The toxicity to HeLa cells of 29 plasticizers was determined in the MIT-24 test system. The 7-day IC50 for HeLa cells varied from 260 to 1.5 g/l. Phthalates, adipates, sebacates, azelates and phosphates with long carbon chain alcohols were very non-toxic to the cells, probably due to insolubility in water of the compounds, while the citrates, some phosphates and the 2 polymer plasticizers had a higher toxicity to the cells. A comparison of the HeLa cytotoxicity with the toxicity in vitro to other cells for 7 plasticizers showed a similarity of the cytotoxicity to all cell types. A comparison of the HeLa cytotoxicity for 20 plasticizers with i.p. lethal dosage in rodents demonstrated a rough similarity of values, suggesting a toxicity in rodents of the compounds by toxic interference of the agents with basal functions and structures of tissues (basal cytotoxicity). Tissue culture studies of the cytotoxic mechanisms of the plasticizers therefore could reveal modes of toxic action in vivo.

Preliminary studies on the validity of in vitro measurement of drug toxicity using HeLa cells. IV. Therapeutic effects and side effects of 50 drugs related to the HeLa toxicity of the therapeutic concentrations.

As a measure of the relevance to human drug effects of drug cytotoxicity in vitro, calculated and actual therapeutic blood concentrations of 50 drugs were divided by their 50% inhibitory concentrations for HeLa cells in vitro to make up three types of cytotoxic quotients (CQTv, CQTd, and CQTr). When the various quotients were compared with known cytotoxic effects of the drugs the parameters correlated well. While most drugs at single dosage reached concentrations with low quotients (CQTv and CQTd) corresponding to few reported cytotoxic effects and side effects, a substantial number of drugs at repeated dosage reached concentrations with high quotients (CQTr) corresponding to known or suspected cytotoxic action in man.

Preliminary studies on the validity of in vitro measurement of drug toxicity using HeLa cells. II. Drug toxicity in the MIT-24 system compared with mouse and human lethal dosage of 52 drugs.

By a comparison of the 50% inhibitory concentration to HeLa cells in the microtitre Metabolic Inhibition Test supplemented by microscopy of cells after 24 h incubation (the MIT-24 test), of 52 drugs with their mouse i.v. LD50 and available approximate human lethal dosage, 7 were found to have a human LD considerably lower than HeLa IC50, indicating a lethal action to specialized functions of the human body not found in vitro, while 39 were found to have a gross similarity between the human LD and HeLa IC50.

Preliminary studies on the validity of in vitro measurement of drug toxicity using HeLa cells. III. Lethal action to man of 43 drugs related to the HeLa cell toxicity of the lethal drug concentrations.

The human lethal plasma concentrations of 46 drugs were divided by their IC50 for HeLa cells in vitro to make up a series of cytotoxic quotients (CQLv). CQLv was then compared with the recorded lethal action to man of 43 of the drugs. While the 7 drugs with the lowest CQLv values produce a non-cytotoxic interference with neuro-transmission, most of the remaining 36 drugs have a known local or systemic cytotoxicity to man. A majority of the 36 drugs induces a non-specific central nervous system (CNS)-depression at lethal dosage, intermingled with function loss from organs outside CNS in proportion to decreasing drug accumulation in CNS cells and increasing CQLv. The remaining drugs which do not penetrate CNS cells and at lethal dosage induce a widespread injury and function loss of tissues outside the CNS, have a CQLv near unity. Non-specific CNS-depression may thus be the primary human reaction to lethal systemic drug cytotoxicity, while widespread drug injury to various tissues outside CNS--conventionally considered to be cytotoxic in origin--may be the obligatory human reaction to drugs that do not penetrate cells well. The present findings indicate a relevance to human toxicity of the HeLa toxicity for most drugs.

Preliminary studies on the validity of in vitro measurement of drug toxicity using HeLa cells. I. Comparative in vitro cytotoxicity of 27 drugs.

Combined drug toxicity to HeLa cells was studied in vitro with use of the microtitre MIT-24 test system. Whether drug toxicity to HeLa cells is representative of drug toxicity to other cultivated cells was investigated by a comparison of the MIT-24 toxicity of 27 drugs to HeLa cells with their toxicity to various permanent cell lines and more differentiated primary cell cultures as reported in the literature, together with original recordings of the MIT-24 toxicity of 9 of the drugs to human fetal kidney cells. A similarity of drug toxicity to all cell types was found. Thus the MIT-24 recordings may be representative of a basal drug cytotoxicity, probably corresponding to local drug irritation and to causal systemic drug toxicity.

Overview of the Final MEIC Results: II. The In Vitro--In Vivo Evaluation, Including the Selection of a Practical Battery of Cell Tests for Prediction of Acute Lethal Blood Concentrations in Humans.

In MEIC, all 50 reference chemicals were tested in 61 in vitro assays. To provide a background to the in vitro/in vivo evaluation, mouse LD(50) values were compared with human lethal doses, resulting in a good correlation (R(2) 0.65). To study the relevance of in vitro results, IC(50) values were compared with human lethal blood concentrations (LCs) by linear regression. An average IC(50) for the ten 24-hour human cell line tests predicted peak LCs better (R(2) 0.74) than other groups of tests. When IC(50) values for 32 chemicals which rapidly enter brain were divided by a factor of 3.2 and 48-hour IC(50) values were compared with 48-hour human LCs for 10 slow-acting chemicals, the prediction improved considerably. Human toxicity was clearly underpredicted for only four chemicals, namely digoxin, malathion, nicotine and atropine, indicating a high relevance of the human cell line toxicity. All chemicals entering the brain induced a CNS depression, explaining this syndrome as a cytotoxic effect. Multivariate analysis was used to select an optimal combination of assays, resulting in a battery of three 24-hour human cell line tests (endpoints: protein, ATP and morphology/pH) with good direct prediction of human peak LCs (R(2) 0.77).

Overview of the Final MEIC Results: I. The In Vitro--In Vitro Evaluation.

In the MEIC study, the first 30 reference chemicals were tested in 82 in vitro toxicity assays while the last 20 chemicals were tested in 67 assays. To increase understanding of the performance of in vitro toxicity tests, these two subsets of results were compared by principal components analyses (PCA) combined with a "random probe" analysis of five key methodological factors, that is, the results from all pairs of methods which were similar in all other respects than the analysed factor were systematically compared by linear regression. This paper is an overview of these published comparisons, and also includes a new "random probe" analysis of another segment of the same MEIC results, namely tests of all 50 reference chemicals by 61 of the methods. A PCA indicated high general similarity (around 80%) of all the results from the 61 methods. According to the new "random probe" analysis, this similarity must depend on the high correlation of results from assays with different cell types (mean R(2) 0.81) and/or different viability endpoints (mean R(2) 0.85). Main factors contributing to the 20% dissimilarity of results were different exposure times and the use of phylogenetically distant test objects in the non-analogous ecotoxicological assays. As expected, the new analysis of the 61 methods gave roughly similar results as the previous "random probe" analyses of the other two segments of the data. Findings support the basal cytotoxicity concept and will improve future in vitro toxicity testing. This testing will probably need assays with varied exposure times. As judged from the similarity of the results, simple assays with cell lines may replace complicated primary culture systems for many testing purposes (e.g. screening).

Principles for the validation of in vitro toxicology test methods.

This review addresses key aspects of validation of in vitro toxicity test methods as they have been presented by different authors during the last decade. The MEIC study is presented as a practical case in which several of these aspects have been considered.

In vitro long-term cytotoxicity testing of 27 MEIC chemicals on Hep G2 cells and comparison with acute human toxicity data.

Within the framework of the EDIT (Evaluation guided Development of In vitro Toxicity and toxicokinetic tests) programme, the long-term cytotoxicity of 27 chemicals was investigated on Hep G2 cells. The first step in the experiments was to determine the PI50(24h) of the chemicals. This is the concentration of compound needed to reduce the total protein content by 50% after 24 h of treatment. In the long-term experiments the chemicals were tested in six different concentrations, using the PI50(24h) as maximum concentration. The cells were treated twice a week with the same concentration of test compound and were trypsinised and counted once a week (dynamic culture). The number of cells was compared to the number of cells of the control. Three major long-term cytotoxicity patterns could be distinguished. After 6 weeks, the EC50(6w)s were determined. This is the concentration of compound needed to reduce the number of cells by 50% after 6 weeks of treatment. These values were compared with the PI50(24h). A good correlation was found for the 27 chemicals (r(2)=0.860). After 6 weeks, the concentration of test compound needed to reduce the total cell protein content by 50% after 24 h after 6 weeks of pretreatment of the cells with a particular concentration of test compound was measured: the PI50(24h-6w). For the majority of compounds there is no difference between the PI50(24h) and the PI50(24h-6w). For ethanol, arsenic (III) oxide, verapamil hydrochloride and orphenadrine, the PI50(24h-6w) increased in comparison to the PI50(24h). For some compounds a doseresponse was observed, indicating that the cells have become more resistant or more sensitive. Linear regression analysis revealed a good correlation (r(2)=0.709) between the EC50(6w) and the human acute toxicity. All these data indicate that a good alternative test may be found for predicting the long-term human toxicity.

Preliminary results from the Scandinavian multicentre evaluation of in vitro cytotoxicity (MEIC).

The multicentre evaluation study of in vitro cytotoxicity tests (MEIC) is organized by the Scandinavian Society of Cell Toxicology. All interested laboratories are invited to test a published list of 50 reference chemicals in their various in vitro assays with a bearing on general toxicity. Submitted results will be centrally evaluated for their relevance to human toxicity, including a comparison with the efficiency of conventional animal tests. This brief communication presents the very first preliminary results of the study, that is, prediction of human acute lethal toxicity for the first 10 MEIC chemicals by all the results submitted to date, that is, five in vitro cytotoxicity assays. As a baseline for judging the efficiency of the cytotoxicity tests, rat and mouse LD(50) values were compared with human acute lethal dosage of the chemicals. Rat LD(50) prediction was relatively poor, but mouse LD(50) values correctly predicted the human lethal dose for six out of the 10 substances. A multivariate method of comparison including all cytotoxicity test results, predicted human lethal blood concentrations as well as the mouse LD(50) prediction of dosage. Since the blood concentrations used in the comparison were derived from human lethal dosage with the help of two simple pharmacokinetic factors (absorbed fraction in the intestine and distribution volume of chemicals), the cytotoxicity assays were found also to be able to predict human dosage, as well as did the mouse LD(50) prediction.

EDIT: A New International Multicentre Programme to Develop and Evaluate Batteries of In Vitro Tests for Acute and Chronic Systemic Toxicity.

The Multicenter Evaluation of In Vitro Cytotoxicity (MEIC) programme provided a battery of three basal cytotoxicity tests with a good (R2 = 0.77) prediction of human acute lethal blood concentrations. The predictive power of this battery would be considerably improved by the addition of new supplementary in vitro tests. The development of these new tests will be facilitated by a close coupling of test development to evaluation. The Cytotoxicology Laboratory, Uppsala (CTLU), is therefore inviting all interested in vitro toxicologists to take part in the Evaluation-guided Development of In Vitro Toxicity and Toxicokinetic Tests (EDIT). All EDIT activities (subprojects) will be designed on a case-by-case basis, but will follow a common pattern. The CTLU will use the accumulated MEIC/EDIT data, and its experience from the previous MEIC evaluation, to suggest priority areas, i.e. the need for certain in vitro toxicity data/tests as supplements to existing in vitro models/batteries on human systemic toxicity. Detailed research programmes corresponding to these areas will be published on the Internet. The CTLU will also try to raise funds for some projects and will coordinate multilaboratory studies. Interested laboratories developing or already using priority tests are encouraged to join the subprojects and to test specific sets of substances (usually sets of MEIC reference chemicals) in their new assays. The CTLU will provide adequate human reference data and will also evaluate results as single components of complex models, together with the laboratory conducting the test. At present, ten priority areas have been identified: a) repeat dose toxicity in vitro; b) urgent mechanistic information from in vitro studies of protein denaturation, morphology of cell injury, differential toxicity between various rapidly measured endpoints (10-60 minutes) and 24-hour cytotoxicity, toxicity to aerobic cells, and discrimination between rapid and slow cytotoxic mechanisms; c) in vitro tests on vitally important, specific receptor toxicity in humans; d) excitatory cytotoxicity; e) reversibility of cell toxicity; f) in vitro tests on passage across the blood-brain barrier; g) in vitro tests on absorption in the gut; h) protein binding in vitro; i) in vitro tests on distribution volumes (Vd); and j) in vitro tests on biotransformation to more-toxic metabolites (hepatocytes plus target cells). This paper gives a short presentation of the rationale for each subproject and reports on ongoing activities.

Protein precipitation In Vitro as a measure of chemical-induced cytotoxicity: an EDIT sub-programme.

As a priority area of the Evaluation-Guided Development of In Vitro Toxicity and Toxicokinetic Tests (EDIT) programme, an in vitro protein precipitation (PP) assay was used on the 50 reference chemicals of the Multicentre Evaluation of In Vitro Cytotoxicity (MEIC) project, to confirm and extend the MEIC results. Dose-response curves were generated for only 30 of the chemicals, and the concentrations causing 10% (EC10) and 50% (EC50) protein precipitation versus the positive control were chosen as endpoints. The number of chemicals with a positive response increased to 46 when a new endpoint, the minimum effect concentration (MEC) that induces protein precipitation with respect to the negative control, was used. When the results were correlated with in vitro cytotoxicity in human cell lines, a similarly good correlation was found between the various endpoints of the PP assay at 5 hours and the 24-hour IC50 average cytotoxicity in human cell lines, even though the number of chemicals included in the correlation was larger for the MEC. Using the prediction error, the endpoint that gave the best correlation between the PP assay and human cell cytotoxicity was once more found to be the 5-hour MEC, and this was chosen for the PP assay. The sensitivity of the PP assay is lower than that of the in vitro cell-line cytotoxicity assay, possibly due to its shorter exposure period and because precipitation is the ultimate event in the sequence of a protein disturbance. It is expected that earlier denaturation steps would give better sensitivity. However, this simple, inexpensive and rapid assay could be useful in the early stages of testing chemicals.

In vitro comparative toxicity of selected drugs and chemicals in HeLa cells, Chang liver cells, and rat hepatocytes.

The cytotoxicity of 16 agents was compared in four different in vitro cellular systems: HeLa cells, Chang liver cells, freshly isolated rat hepatocytes in suspension, and primary monolayer cultures of postnatal rat hepatocytes. The compared cytotoxicity values had been obtained by different techniques: inhibition of spreading of HeLa cells by 50% after 24 hr incubation and leakage of lactate dehydrogenase (LDH) by 50% in Chang liver cells, isolated rat hepatocytes in suspension, and primary cultures of rat hepatocytes after 4, 1, and 3 or 24 hr incubation with the chemicals. Carbon tetrachloride, halothane, acetaminophen, and tetracycline were more toxic to the hepatocytes compared with the two cell lines. Alcohol, sodium salicylate, caffeine, papaverine, nitrofurantoin, and five tricyclic antidepressants showed similar toxicity in all compared cellular systems. Thus, the agents with a known or suspected metabolism-mediated liver toxicity indeed had a high differential hepatotoxicity in vitro.

Screening for eye irritancy using cultured HeLa cells.

To investigate whether toxicity tests on HeLa cells were predictive of eye irritancy, 18 compounds of known eye irritancy and in vitro cytotoxicity were tested on HeLa cells in the MIT-24 system. The results correlated well with eye irritancy as determined by the Draize test in rabbits for 16 of the test substances, but failed to detect the high eye irritancy of 1-heptanol and allyl alcohol, both of which were cytotoxic in other cellular systems.