Novel Advanced In Vitro Methods for Longterm Toxicity Testing

The Report and Recommendations of ECVAM Workshop 45^1,2

Reprinted with minor amendments from ATLA 29: 393-426.

Walter Pfaller,³ Michael Balls,⁴ Richard Clothier,5 Sandra Coecke,⁴ Paul Dierickx,⁶ Björn Ekwall,⁷ Bryan A. Hanley,⁸ Thomas Hartung,⁹ Pilar Prieto,⁴ Michael P. Ryan,¹⁰ Gabriele Schmuck,¹¹ Dariusz Sladowski,⁴ Joan-Albert Vericat,¹² Albrecht Wendel,⁹ Armin Wolf¹³ and Jens Zimmer ¹⁴
³Institute of Physiology, University of Innsbruck, Austria; ⁴ECVAM, Institute for Health & Consumer Protection, European Commission Joint Research Centre, 21020 Ispra (VA), Italy; ⁵School of Biomedical Sciences, University of Nottingham, Nottingham NG7 2UH, UK; ⁶Institute of Public Health, Laboratory of Biochemical Toxicology, Brussels, Belgium; ⁷Pavals När, Stanga, Sweden; ⁸CSL Food Science Laboratory, Colney, Norwich, UK; ⁹Department of Biochemical Pharmacology, University of Konstanz, Germany; ¹⁰Department of Pharmacology, University College, Dublin, Republic of Ireland; ¹¹Bayer AG, Business Group Pharma, Ph-PD Toxicology, Wuppertal, Germany; ¹²SanofiSynthelabo Recherche, Porcheville, France; ¹³Novartis Pharma AG, Preclinical Safety, Toxicology/Pathology, Basel, Switzerland; ¹⁴Institute of Medical Biology, University of Odense, Denmark

¹ECVAM - The European Centre for the Validation of Alternative Methods. ²This document represents the agreed report of the participants as individual scientists.

Address for correspondence: Dr Walter Pfaller, Institute of Physiology, University of Innsbruck, Austria.

Address for reprints: ECVAM, TP 580, JRC Environment Institute, 21020 Ispra (VA), Italy

Preface:

This is the report of the forty-fifth of a series of workshops organised by the European Centre for the Validation of Alternative Methods (ECVAM). ECVAM's main goal, as defined in 1993 by its Scientific Advisory Committee, is to promote the scientific and regulatory acceptance of alternative methods which are of importance to the biosciences and which reduce, refine or replace the use of laboratory animals. One of the first priorities set by ECVAM was the implementation of procedures that would enable it to become well-informed about the state-of-the-art of nonanimal test development and validation, and the potential for the possible incorporation of alternative tests into regulatory procedures. It was decided that this would be best achieved by the organisation of ECVAM workshops on specific topics, at which small groups of invited experts would review the current status of in vitro tests and their potential uses, and make recommendations about the best ways forward. In addition, other topics relevant to the Three Rs concept of alternatives to animal experiments have been considered in several ECVAM workshops.

Description and Purpose

Every regulatory authority throughout the world requires assessments of the systemic effects in animals of chronic exposure to an agent or to a mixture of agents for which repeated or prolonged human exposure may be expected. As a result, there are some specific requirements for a study strategy that reflects regional sensitivities to the toxicological issues or different testing philosophies. Recently, the regulatory authorities of the United States, Japan and the European Union agreed to meet as part of an international harmonisation initiative to develop consensus guidelines for a range of nonclinical and clinical activities, including toxicology testing in animals (1).

Long-term, or repeated-dose, toxicity tests are usually defined as studies of longer than three months duration. For rodents, this represents 10% or more of their life-span. The studies are conducted in all classes of laboratory animals (mammals, birds, fish, etc.), including some economically important wild and domestic animals. Long-term toxicity encompasses the classic subchronic and chronic systemic toxicology studies, multigeneration reproduction studies, and carcinogenicity studies.

It is often stated that the principal reason for conducting a chronic toxicity study is to fulfil the regulatory requirements leading, it is hoped, to market approval. This so-called regulatory mandate not only implies that the type and the design of the study must meet expected criteria, but also that studies must be conducted and reported according to the highest possible standards, and that records must be maintained in a manner that ensures comprehensive "whole stock" independent review, i.e. to the standards of Good Laboratory Practice (GLP). International regulatory authorities, and national regulatory authorities, such as the Food and Drug Administration (FDA) and the Environmental Protection Agency (EPA) in the USA, require that preclinical toxicology studies are conducted in compliance with GLP guidelines. Many regulatory agencies have also entered into "memoranda of understanding" (MOUs) with the FDA that "provide for reciprocal recognition of each country's good laboratory practices programme and acceptance of the best data collected in either country for elevation of safety". Similar regulations ("good academic research practices in animal pharmacology") are being proposed for animal pharmacology studies, in support of product-development candidates. Taken as a whole, these regulatory initiatives for good scientific practices set the groundwork for undertaking studies appropriately.

Although long-term toxicity studies for animal models are defined to some extent at the regulatory authority level, no long-term alternative tests -- including those involving cell and tissue culture-based approaches -- are currently conducted at the regulatory level. However, many of these novel, advanced, in vitro approaches result in information which is often more relevant than animal studies for human hazard assessment, due to their use of human-derived proteins, cells and tissues.

The principal goals of nonclinical safety assessment, as formulated by Zbinden (2) for laboratory animal species, can be transferred directly to modern and advanced in vitro technologies, simply by expanding the goals formulated to include adequate in vitro systems (identified in italic typeface).

Identification of the spectrum of toxicity, detection of adverse effects in selected laboratory animal species or appropriate cell and tissue culture systems, and description of the dose-edect relationship over a broad range of doses.
Extrapolation and prediction of adverse effects in other species, particularly in man, or cultured cells or tissues obtained from both animals and humans.
Safety; prediction of safe levels of exposure in other species or cultured cellular systems derived from other species, specifically from man.

Chronic toxicology studies in laboratory animals, as well as in defined in vitro systems, can contribute to each of these goals. The identification of the spectrum of toxicity after repeated administration to man by using appropriate cell and tissue culture models was, of course, the primary objective of this workshop. This spectrum of toxicity (3) can describe primary target organs, lesion reversibility, sensitivity of clinical parameters, dose levels, and times to produce detectable changes. Results of chronic or long-term studies can also be used to assist in the selection of dose levels to be used in other studies in the same animal species (such as reproductive toxicology or carcinogenicity studies) or in vitro models. Understanding the comparative biology, differences in responsiveness across species, or cell systems derived from them, and exposure profiles of the test substance in each species or in vitro system, enables scientists to make educated assessments of likely adverse effects and safe exposure levels in other species, including humans. The practice of conducting long-term studies in multiple species (usually rodents and nonrodents), which was thought to decrease the uncertainty associated with interspecies extrapolation, and is an integral part of many regulations, has often been questioned (for example, 4).

In order to describe the toxicity associated with repeated administration or prolonged exposure to a test substance, it is generally expected that a chronic toxicity test should identify at least two extremes of response, i.e. toxic effects at one or more dose levels, and a no effect (NOEL) or a no adverse effect (NOAEL) at one or more lower dose levels.

It is hoped that any study will describe the spectrum of changes between these two extremes, such as dose levels that are not tolerated or are marginally tolerated by either an animal or an adequate in vitro system. Investigators often attempt to design chronic studies to describe a compound-related toxicity, and to identify a mechanism by which that toxicity occurs. In most situations, complexities and multiple variables inherent in such repeat-dosed animals preclude clear identification of the mechanism(s) by which toxicants act. A more realistic approach is to design chronic studies that will help to define (limit) variables to be evaluated in follow-up investigations of the mechanisms involved. Evaluation of different types of endpoints, such as enzyme induction, cell proliferation, immunohistochemical reactions, biomarkers, and differential gene or protein expression, in addition to routine examinations, may help in this regard.

Most of the parameters used to evaluate toxicity in animal test systems can be transferred one-by-one to relevant systems in vitro. An important prerequisite is that the cell or tissue culture systems utilised must match closely the in vivo expression of differentiated function. Another prerequisite is that the cultures used need to be maintained under stable and reproducible conditions over long periods of time (weeks or months).

If the maintenance of cultures can be combined with continuous replenishment of nutrient media, removal of waste products, repeated-dose toxicant administration, and long-term continuous administration of low doses of toxicant, a model can be achieved. Evaluation of some important endpoints, such as enzyme induction, modulation of cell proliferation, assessment of immunohistochemical reactions and changes in relevant biomarkers is more easily carried out (and with a larger number of replicated experiments) in cell-culture systems than in animal models. In addition, the mechanisms of action of the xenobiotics under examination are more readily identified, since the many variables due to whole-body homeostatic regulation mechanisms are absent or drastically reduced.

At present, there are no generally accepted in vitro models capable of replacing chronic testing in animals. Even the term "chronic in vitro toxicity testing" provokes debate. For this reason, we prefer the term "long-term testing". Although the term "acute toxicity testing" is already established, it would, by analogy, better be called "short-term testing". It was suggested by most of the participants in the workshop that "short-term" refers to testing periods of up to 72 hours, which reflects the maximum survival time of most conventional static culture systems without repeated renewal of the nutrient medium. Acute testing is usually related to administration of a single high dose of the test compound. Long-term testing is linked to exposure of appropriate in vitro systems over at least 5 days, and is related to either continuous exposure of the system to low concentrations of a test compound, or to the repeated administration of a test compound.

Information gained from long-term toxicity studies can be best utilised and understood when evaluated in the context of an integrated testing approach. The combined result of all toxicity studies in vitro and in vivo should suggest signs and symptoms of adverse reactions to look for in humans, following exposure. With the exception of idiosyncratic reactions and hypersensitivity, many of the systemic, organismic and cellular responses in humans are qualitatively predictable from the respective test system. However, extreme caution should be exercised in establishing quantitative expressions of human safety (or risk) based on either animal or in vitro data.

Current Status of Long-term In Vitro Toxicity Testing

Non-organ-specific long-term toxicity testing

Only very few attempts have been made to generate toxicological data from long-term exposure of cell-culture or tissueculture systems to toxic chemicals or drugs. Tests developed in this context have focused on non-organ-specific parameters such as proliferation and protein synthesis or cellular redox status (and thereby, mitochondrial function), the cytotoxicity endpoints used in acute testing. The difference is that compounds are administered at lower concentrations over long periods, either under conventional static cell-culture conditions or under constant perfusion conditions, i.e. the nutrient medium is continually replenished with fresh medium containing the test compound.

This requires cell types that can be maintained in culture over periods as long as several weeks without the need for subculturing. The cells used can be either proliferating or differentiated or both (non-quiescent).

Although cell culture cannot adequately mimic whole-body responses to toxicant challenge, there are a number of advantages of such approaches, including ease of use, lower cost, the larger number of experiments that can be conducted, and the wider range of concentrations that can be tested (5). Compounds with very specific pharmacological activities can, in comparative tests, provide data that will facilitate the development of novel structures with improved or desired activities. The assay of activity can include ligand-binding determinations, reporter gene assays, effects on cell cyclerelated genes, etc. However, all of these approaches suffer from the fact that they are, at present, acute, i.e. high-dose (pharmacological), short-term assays, with cells being exposed for, at most, 2 or 3 days. Any extrapolation from this type of study to chronic toxicity is not generally thought to be valid. The Multicentre Evaluation of In Vitro Cytotoxicity (MEIC) study (6, 7), however, presented evidence that tests developed to detect acute toxicity (24 hours) may also be used to detect long-term (6-week) toxicity. Dierickx & Ekwall (6) have demonstrated that the so-called P150 value, the test compound concentration required to induce a 50% reduction in cellular protein content, could represent a promising parameter for the in vitro judgement of long-term toxicity. For a number of compounds tested in the MEIC programme, a good PI50 correlation could be demonstrated for the acute (24-hour) and the long-term (6-week) exposure. This was later confirmed for 27 of the MEIC chemicals, by using total cell numbers as the endpoint in Hep G2 cells treated and subcultured during 6 weeks. After this period, the cells appeared to be less sensitive to 7 of the 27 chemicals (8).

Another example of an attempt to assess long-term toxicity was the use of epithelial cells grown in hollow fibres and continuously perfused with culture medium containing low concentrations of a fungal neurotoxin, 3-nitropropanoic acid (3-NPA), an inhibitor of the mitochondrial respiratory-chain enzyme, succinic dehydrogenase (9). In this study, the effect on certain cell parameters (particularly mitochondrial activity) mirrored that observed in low-dose animal experiments, and would have helped to predict likely pathology in low-dose in vivo experiments.

Other than these approaches, few other models for long-term in vitro toxicity testing have been employed systematically. If long-term toxicity is assumed to be related to the development of chronic disease, which is always related to specific organs, it must be questioned whether it is biologically or pathologically relevant to try to assess non-organ-specific effects to predict the risk of development of persistent damage. The value of non-organ-specific endpoints in this context has still to be established, and must be proved with at least several relevant models. The workshop primarily focused on three of the most important organs for toxicity and detoxification: the liver, the kidney and the nervous system. The decision to select these organs was facilitated by the fact that the corresponding in vitro systems have been studied more intensively than others over the past two decades, and that the pathogenetic mechanisms involved in the development of chronic disease states are thought to be better understood for these organs than for other organ systems. Finally, socio-economic aspects also influenced the selection. The costs arising from treatment of chronic liver, kidney and brain diseases comprise a major proportion of the health-care expenses worldwide.

Organ-specific long-term testing

The term "chronic" describes the mode and "history" of development of a disease state, characterised by the usually slow progression of one or multiple dysfunctions in one or more organ systems of the human or animal body. In contrast to acute disease states, chronic disease states are always the consequence of persistent or progressively deteriorating dysfunctions of cells, organs or multiple organ systems. Although chronic disease may be triggered by acute effects linked to one single cell type, its development, maintenance and progression are related to the simultaneous dysfunction of several cell types, often located within the same organ. Development of dysfunction may be facilitated by sub-acute doses of toxic xenobiotics to which the target cells are exposed over prolonged intervals. During long-term exposure, some compounds can induce adaptive processes in the respective target cells. This may result in alterations in metabolism due to changes in gene expression at the transcriptional, translational or post-translational levels, through induction or inhibition of enzyme systems, and up-regulation or down-regulation of receptor-ligand interactions, thereby changing the functions of intracellular signal cascades. Therefore, prediction of the long-term toxicity of a new chemical entity (NCE) with regard to dysfunction or injury in key organs of the body is of the utmost importance.

Although there exist in vitro models to test for acute target organ toxicity, they have so far seldom been used for long-term toxicity testing.

Very recently, attempts have been undertaken to adapt epithelial liver-derived and kidney-derived monolayer cultures, primary cultures, and cell lines, to allow for the continuous low-dose, long-term administration of known hepatotoxic and nephrotoxic compounds(10).

Cell culture models in hepatotoxicity testing

Among the in vitro liver preparations recognised as useful experimental models in toxicology, intact cellular systems, such as the isolated perfused organ, tissue slices and isolated hepatocytes, offer various advantages over other systems, such as hepatocyte cell lines, subcellular hepatocyte organelles and genetically engineered cells. The isolated hepatocyte is the most widely used model (Table I).

Table I: Available in vitro liver models

Model	Integrity	Adverse effects	Mechanisms	Analytical methods	References
Isolated perfused liver	2 hours	Cholestasis Enzyme leakage	Bile flow, uptake and secretion ALT/AST	Gravimetry, radiolabelled bile acids, enzyme activity	11
Slices	7 days	Carcinogenesis	Cytochrome P450 induction	EROD/PROD, testosterone, Western blot fragmented and condensed nucleoli, TEM	12,131,132
Slices	7 days	Periportal or centrilobular necrosis or apoptosis	Inhibition of apoptosis		12,131,132
Hepatocyte couplets	3 hours	Cholestasis Hyperbilirubinaemia	Canalicular bile acid transporter, organic anion transporter, P-gp 170 kD protein	Fluorescent bile acids, GSSG-conjugates and P-gp 170 kD substrates	133, 134
Hepatocyte couplets	3 hours	Cholestasis Hyperbilirubinaemia	Filamentous actin	Immunohisto- chemistry	133, 134
Collagen sandwich cultures	15 days	Carcinogenesis	Peroxisome proliferation. Induction of cytochrome P450.	PCR	23, 24, 135
		Carcinogenesis	Inhibition of gap-junction mediated intracellular concentration	FRAP techniques
		Normal hepatic function	ALT/AST	Enzymatic activity
		Normal hepatic function	Albumin and acute-phase proteins	ELISA
Ito cells	96 hours	Fibrosis	Synthesis of extracellular matrix (collagen, fibronectin)	ELISA	26
Hepatocyte/ non-parenchymal co-cultures	48 hours	Inflammation	LPS-mediated processes Cytokine production	ELISA	136, 137
Primary cultures	24 hours	Lipidosis	Lipoprotein synthesis	Immunohisto- chemistry	11, 138, 139
		Lipidosis	Phospholipidosis Steatosis	Nile red staining
		Carcinogenesis	Unspecific cytotoxicity (Stimuli for proliferation)
Genetically engineered cell lines	unlimited	Metabolism-related specific effects	Various	Various
Cell lines	Unlimited	Carcinogenesis	Increased proliferation	MTT, Alamar Blue^TM (resazurin reduction) and BRDU assays
Subcellular fractions	1-3 hours	Impaired enzymic activities	Various	Various

ALT/AST = alanine aminotransferase/aspartate aminotransferase; EROD = 7-exthoxyresorufin-O-deethylase; FRAP = fluorescence recovery after photobleaching; GSSG = oxidised gluthathione; LDH = lactate dehydrogenase; LPS = lipopolysaccharide; PCR = polymerase chain reaction; PROD = 7-pentoxyresorufin-O-deethylase; TEM = transmission electron microscopy.

Additional enpoints: pharmaco-genomics -- gene expression using proteomics, PCR and DNA chip array techniques. Whenever possible should be extended to study in vitro systems of human origin.

BrdU = 5-bromo-2-deoxyuridine; LDH = lactate dehydrogenase.
Additional endpoints: pharmaco-genomics -- gene expression using proteomics, PCR and DNA chip array techniques.

Isolated perfused liver: Although the isolated perfused liver represents the closest in vitro model of the in vivo situation, its application to long-term toxicity is inhibited due to its short lifetime (2-3 hours). Its use is therefore limited predominantly to physiologically based toxicokinetic modelling, where intact bile flow must be maintained. In some cases, there seems to be a connection between the induction of a drug-induced side-effect after chronic in vivo administration and the general ability of a compound to induce a direct interaction with a cellular target after a single treatment. Such interactions, although normally reversible, can cause chronic toxicity when maintained over a long treatment period. Such effects were found, for instance, after perfusing isolated intact rat livers with two cyclosporins. The derivative SDZ IMM 125, which causes hepatic side-effects in man but not in animals, produced a stronger impairment of bile acid transport and a stronger leakage of hepatic aminotransferases than its non-toxic, structurally related congener, cyclosporin A (11).

Liver slices: The slice model has the advantage of retaining the in vivo tissue organisation, and has largely intact cell-cell and cell-matrix interactions. Although the lifetime of this model is limited for long-term in vitro testing, the model can be used to predict chronic side-effects occurring after long-term in vivo treatment. For instance, SDZ 249 665 is a capsaicin analogue that caused severe liver toxicity with periportal apoptosis-like hepatocellular necrosis/lysis after chronic treatment in dogs, but not in rats. By treatment of liver slices from dog, rat and man with SDZ 249 665, the in vivo situation was mimicked, and the results were used for risk assessment. The compound had nearly the same effect in dog and human slices, but not in slices from the rat. Since these results reflected perfectly the in vivo species-specific findings with SDZ 249 665, the potential risk for chronic liver damage in man was considered to be too high (12).

Isolated hepatocytes: The in vitro model most often used for long-term hepatotoxicity testing is the isolated hepatocyte. The hepatocyte 24-hour primary culture seems to be a suitable tool for studying toxic mechanisms involved in adverse sideeffects (13) on the liver. However, maintaining the culture for longer than 24 hours might generate misleading results. Major problems have been encountered with primary liver cultures because of their loss of liver-specific functions, such as phase I and phase II drug-metabolising enzyme activities (14-16). Cytochrome P450 concentrations drop by 50% during the first 24-48 hours in primary cultures (17-20), whereas the phase II detoxification enzyme glutathione S-transferase (7c isoform) is increased (21). Improvement was achieved by co-cultivating hepatocytes with either Kupffer cells or epithelial cells from the biliary system (14). The lifetimes of primary hepatocyte cultures were also prolonged when cultures were sandwiched between layers of collagen (22). Because of the maintenance of structural and functional integrity for over 15 days, the long-term hepatocyte sandwich model has clear advantages in comparison to the hepatocyte 24-hour culture (23). In addition to maintaining parts of the drug-metabolising capacity, the morphology of the hepatocyte sandwich culture is intact, with a normal cell shape, and the culture has an intact canalicular structure. The secretion of acute-phase proteins in the sandwich culture is well established (24), as are gap-junction mediated intracellular communication, ahd the secretion of bile acids via the canalicular bile acid transporter and the multi-organic anion transporter (23). Because the model is stable over a certain time interval, it has been used successfully to mimic chronic treatment in vivo. Hepatocytes in the collagen sandwich configuration clearly showed hepatic alanine amino transferase (ALAT) and aspartate amino transferase (ASAT) enzyme release into the cell-culture supernatant when given daily low doses of the hepatotoxin SDZ IMM 125 for 7 days. This effect was much greater than that produced by the less-hepatotoxic or non-hepatotoxic cyclosporin A derivatives, cyclosporin G or PSC 833 (25). Rat and human hepatocyte sandwich cultures are successfully used in industrial situations, especially when the cytotoxicity of a class of compounds can only be identified following a single short-term treatment at concentrations that are too high to be attainable in the plasma of chronically treated animals or humans.

Ito cell cultures and hepatocyte/Ito cell cocultures represent useful tools for studying liver fibrogenesis. Media of 5-day-old Ito cell primary cultures showed high levels of the major basement membrane components, collagen IV, laminin, and entactin/nidogen. When hepatocytes were added to 2-day-old Ito cell primary cultures, they established close contacts with Ito cells in less than 24 hours, and expressed the tight junction-associated protein, ZO-1, and abundant extracellular matrix containing laminin, fibronectin and collagens proIII and proIV, deposited over hepatocyte cords and between hepatocytes and Ito cells. Ito cell activation can be modulated, for instance, by TGF-P or retinoic acid (26). Most of the in vitro studies performed on cultured hepatocytes have been conducted with primary rodent cultures, and have focused predominately on drug-metabolising enzymes. There are major differences compared with human hepatocytes.

Long-term toxicity testing in hepatic in vitro models depends primarily on the availability of long-lived cultures with appropriate expression of differentiated hepatocyte functions, most importantly those involved in drug or xenobiotic metabolism. Currently, there are three categories of model: primary cultured hepatocytes; genetically engineered cells expressing single human or animal P450 enzymes; and cellular lines derived from human hepatomas (or from immortalised hepatocytes). Engineered cells are now widely used as tools to assess the involvement of certain enzymes in drug metabolism and the formation of metabolites. Primary cultured hepatocytes, specifically those of human origin, can produce a metabolite profile of a drug, very similar to that found in vivo, and therefore represent the model closest to the hepatocyte in vivo. Their availability is very limited. Primary cultures derived from experimental animals are employed in certain investigations, but are not necessarily able to deliver the insight required to uncover the mechanism of action of drugs or to predict side-effects of xenobiotics in humans. Consequently, the search for hepatic cell lines expressing the whole spectrum of human drug-metabolising and xenobiotic-metabolising enzymes, as an alternative to primary cultures, appears to be the strategy of choice for longterm testing. Most hepatoma cells show expression, activity and inducibility of cytochrome P(CYP)1A1 upon exposure to xenobiotics, but no or very low expression of other relevant CYPs. Transfection of these cells with genes encoding transcription factors controlling the expression of certain CYPs can provide re-expression of the key drug-metabolising enzymes (27). In addition, specific cell-culture media can help to provide the best possible hepatocyte differentiation over intervals of 2 weeks or even longer (28, 29). The use of conventional culture media in combination with continuous medium perfusion has been demonstrated with Hep G2 human hepatoma cells grown on microporous growth supports. This method has the possibility of being an attractive model for evaluating the effects of continuous low-dose long-term exposure (European Union, Standard, Measurement and Testing Programme, SMT4-CT96-2070).

Cell-culture models in nephrotoxicity testing

Cell-culture techniques, as a tool for in vitro nephrotoxicity studies, have gained in importance, due largely to improved methodologies for growing homogeneous cultures of renal cells. Two major strategies have been pursued: the use of primary cultures of glomerular mesangial and epithelial cells, and of renal epithelial cells from various sites along the nephron; and the use of permanent renal epithelial cell lines.

Depending on the retention of adequate renal cellular functions, which are known to interfere with xenobiotic or drug action (30), renal cell cultures have the advantage of providing an experimental model that is not influenced by higher-order regulatory systems. For successful applications of cell and tissue cultures to renal physiological and toxicological studies, the following requirements must be met.

Cells should have retained the polar architecture and junctional assembly of epithelia, and should express the proper polar distribution of membrane enzymes and transport systems (31).
Cells should exhibit vectorial transport of solutes and water, manifested by the formation of domes (32), and the generation of transepithelial electrophysiological characteristics, such as a spontaneous transepithelial potential difference resulting from a certain transepithelial resistance (33).
The cellular uptake of xenobiotics should occur from either the apical or the basolateral side, as observed in vivo (34).
Cells should retain nephron segment-specific characteristics, distinct metabolic and transport properties, and hormone responsiveness (35).

These requirements are best, though not completely, met by primary cultures of renal epithelial cells during the early period after culture initiation. None of the continuous renal epithelial cell lines used fully expresses all the necessary differentiated functions shown by their ancestor cell(s) in vivo and in situ (35).

Obtaining primary cultures from the kidney is somewhat hampered by the fact that the kidney and the nephron comprise at least 15 cell types (36). Therefore, one must ensure that homogeneous cultures have been obtained before any study can start (37).

Despite this difficulty, primary cultures of proximal nephron epithelium cells from various species have been developed, for example, mouse (38), rat (39, 40), rabbit (41) and pig (42). Although cells in primary culture tend to dedifferentiate within hours, the characteristics of those cells are usually closer to the in vivo situation than are cell lines, at least for a limited culture period.

The most sophisticated metabolic and endocrinological approach developed for specifically selecting certain cell types is the design of serum-free, hormone-supplemented culture media (43). Today, hormonally defined media are available for culturing cells of almost all nephron segments of the most widely used mammalian species (3 7, 38, 43-45), including man (46).

One of the disadvantages of establishing renal primary cultures is that there still exists a considerable gap in available information with respect to markers that may be used to identify the respective in situ ancestor cell unambiguously (35), and in particular those of glomerular origin after bulk isolation of cells by centrifugation. The situation is further complicated by the fact that cells tend to dedifferentiate quickly as they are maintained in culture.

Furthermore, not all nephron cells isolated so far can be passaged, due either to dedifferentiation or to the non-availability of specific culture media. For example, the growth of primary cultures of rat proximal tubules is extremely difficult (47). Some improvement has been achieved by using microporous growth supports (31), which ensure nutrient access from the apical, as well as from the basolateral side of cultured monolayers. In addition, coating the growth substrates with extracellular matrix molecules, such as collagen I, collagen IV, fibronectin, laminin, or commercially available mixtures of these compounds (for example, Matrigel®; 48) appears to improve adhesion, proliferation and differentiation of cells in culture. Despite these limitations, renal primary cultures represent a reliable, though not easy-to-handle, technique for studying basic renal cellular functions and their modulation by nephrotoxicants.

Primary cultures have been used successfully to study acute or short-term in vitro effects of cisplatin, gentamycin, cephalosporins, cysteine conjugates, butyl hydroperoxide, mercuric chloride and cadmium chloride (49-52). The majority of in vitro nephrotoxicity studies, however, have been performed on permanent or continuous renal epithelial cell lines, which have proven to be very powerful tools for the study of nephrotoxicity in vitro. A list of the most widely used lines and their sites of origin is given in Table II.

Table II: Most widely used continuous renal epithelial cell lines

Cell line	Animal derived from	Nephron segment of origin
LLC-PK1	Yorkshire pig	Proximal nephron
OK	North American opposum	Proximal nephron
JTC-12	Monkey	Proximal nephron
HK-2	Human	Proximal nephron
MDCK	Dog (Cocker Spaniel)	Collecting duct
A6	Xenopus laevis	Distal tubule/collecting duct

Besides their many advantages, such as an unlimited life-span and the lack of time-consuming isolation procedures, cell lines suffer from some drawbacks. Although they retain some of the differentiated functions of their in vivo ancestor cells (35), they have dedifferentiated in culture. Another drawback arises from transdifferentiation, where cells lose one phenotype and adopt another (53). As a result, cells that are claimed to be of proximal tubular origin may exhibit a combination of properties characteristic of different parts of the nephron. Two cell lines that are often used as a model for the proximal nephron are the porcine cell line, LLC-PK₁ and the OK cell line from the opossum kidney (Table II). LLC-PK₁ cells lack expression of the enzyme fructose-1,6-bisphosphatase, rendering them incapable of gluconeogenesis, a key metabolic pathway in proximal nephron cells (41). In addition, LLC-PK₁ cells are not responsive to parathyroid hormone, and lack a probenecid-sensitive organic anion transporter (48). OK cells, on the other hand, display very little y-glutamyl transpeptidase, and lack alkaline phosphatase. Both these enzymes are considered to be markers for the proximal nephron (35). This situation could be improved by the development of gluconeogenic strains of both cell lines by adaptation of cells to glucosefree media (54).

These results suggest that continuous cell lines could be re-differentiated by the use of appropriate culture conditions, including a well-defined extracellular matrix. Cell and molecular biology technologies, such as cell fusion and transfection techniques, could be used to provide the re-expression of lost functions, such as specific transporters or enzymes, and perhaps even specific receptors.

A combination of the advantages of a continuous cell line with improved differentiation of primary cultures can be obtained by immortalisation of the latter. Immortalisation of primary proximal tubular cell cultures has been achieved in several ways (55). Proximal nephron cell lines have been produced by targeted oncogenesis in transgenic mice by using a pyruvate kinase-SV-40(T) antigen hybrid gene (56). Transduction of the SV40 large T antigen has also been used to establish cell lines from rat primary proximal nephron cell cultures of the Wistar Kyoto rat (57) and the rabbit (58, 59). Immortal human proximal tubular cell lines have been generated by transduction with human papilloma virus (HPV 16) E6/E7 genes (58). Using a hybrid adeno 12-SV40 vector, successful immortalisation of rabbit (61) and human (62) proximal primary cultures was achieved. It remains to be established whether or not these newly developed lines have maintained all the desired characteristics of their in vivo precursor cells, and are able to retain their functional characteristics through multiple passages. The experience acquired with perfusion cultures and continuous toxicant administration led to the development of a new, more-refined perfusion system, EpiFlow, which not only provides an improved oxygenation of the monolayer cultures, but also allows the simultaneous cultivation of at least two cell types. Cultures of renal epithelial cell lines which, under conventional static culture conditions or even under simple perfusion conditions, display a predominantly glycolytic energy metabolism, redifferentiate under continuous medium supply in the presence of an elevated partial pressure of oxygen to a phenotype charactelrised by oxidative energy metabolism. They also change in morphological appearance to one more closely matching that of the in vivo parent cell type (EU biotechnology programme 1997-2000, BI04-CT97-2006).

Cell culture models in neurotoxicity testing

Whereas in vitro systems for long-term testing of hepatotoxicity and nephrotoxicity still need to be improved, the brain as the central part of the nervous system represents an exception, in that there already exist tissueculture methods for long-term testing of toxic xenobiotics. In vitro neural systems can be predictive for neurotoxicity, except for xenobiotics that act on the central nervous system primarily by affecting the function or integrity of the blood-brain barrier. Primary neuronal and glial cultures from cerebral tissue of embryonal rodents have been examined extensively in the last two decades (63-65). They have been used as models for investigating neurotransmitter pathways (66) and electrophysiological properties (67), and as cellular correlates of neuronal diseases (Alzheimer's disease, Parkinson's disease, Huntington's disease and epilepsy). Receptor binding tests have been established to design new drugs and to identify the target proteins. In pharmacology, neuronal cell cultures have been used for screening purposes or for investigation of the modes of action of new pharmaceuticals (68). Until now, little effort has been made to establish toxicological tests based on neuronal cell cultures. As in classical toxicology, chemicals have to be examined in organ specific cell cultures to elucidate target-organ specificity. Neurotoxins and neuroactive pharmaceuticals affect neuronal cells and/or interneuronal communication. Information transport is not restricted to neurons, but also occurs between neurons and astrocytes and among astrocytes. Both cell types reveal neurotransmitter- sensitive receptors (69) which regulate ion-sensitive channels. Neurotransmitter metabolism and neurotransmitter uptake and storage have been characterised for both cell types. The metabolic-activating capacity was comparable to that of the nervous system in vivo. It was shown that various types of organophosphates, such as tri-o-cresyl phosphate, are metabolised (70).

However, the need for target organ specificity cannot be satisfied by permanent cell lines or by avian cell cultures. The standard of neuronal cell cultures for long-term toxicity testing -- although at a high level -- requires further optimisation. The use in toxicology is based on three different cell cultures: neurons; astrocytes or oligodendrocytes; and microglia.

Neurons: Primary neuronal cell cultures from various brain regions containing both mature differentiated neurons, together with minor amounts of astrocytes, may be considered as morphologically, as well as functionally, comparable to the native neuronal tissue. This in vitro model permits evaluation of fundamental neurotoxic mechanisms. Neurodegeneration, excitotoxicity, oxidative stress, receptor-regulated functions and neurotransmitter imbalance are some of the most prominent features that could be investigated in vitro.

Chemically induced neurodegeneration is characterised by different patterns of neuronal death, swollen or destroyed axons, or destruction of the myelin sheath. As biochemical changes precede these morphological endpoints, not only cell viability, but also the integrity of the cytoskeleton and the neuronal energy state should be examined (71). Various neurotoxins, whose modes of neurotoxic action have been reasonably well characterised (Table III), have been used to characterise the suitability of primary neuronal cell cultures derived from fetal rats, as models for neuronspecific toxicity. The effects of these model neurotoxins on primary neuronal cell cultures were examined through the endpoints of cytotoxicity, cytoskeleton integrity, and cellular energy status (72, 73). This test system correctly identifies neurodegenerative compounds based on long-term (7-14 day) exposure of the cell cultures. Long-term neuronal cell cultures have been used in industry for some time to screen and identify mechanistic effects of pharmaceuticals and agrochemicals. This is also true for the oxidative stress, an interesting endpoint for pharmaceutical research as well as for toxicology, where recently developed fluorescent dyes are very helpful (74).

Table III: Overview of neurotoxins successfully screened in vitro

Compound	Cellular target	Effect	References
Mipafox	Cytoskeleton/NTE/AchE	OPIDP	76, 140
Paraoxon	AchE	Cholingergic symptoms	140
Acrylamide	Cytoskeleton, glycolysis	Distal axonopathy	141, 142
2,5-Hexanedione	Cytoskeleton, glycolysis	Distal axonopathy	143, 144
Carbon disulfide	Cytoskeleton	Central and peripheral axonopathy	145, 146
Disulfiram	Cytoskeleton	Central and peripheral axonopathy	147
IDPN	Cytoskeleton	Proximal axonopathy	147
Paraquat	Redox cycling	Parkinsonism, ganglionic degeneration	149, 150
KCN	Oxidase Cytochrome C	Delayed Parkinsonism	151
3-NP	Succinate dehydrogenase Complex II and Krebs cycle	Huntington's disease-like symptoms	152, 153
NMDA	NMDA channel	Delayed neuronal cell death	68

Excitotoxicity caused by a cellular overload of Ca²⁺ in the neurons by activation of the N-methyl-D-aspartate (NMDA) receptor is another important mechanism for neurotoxicological investigation. In the EU Fourth Framework BIOTECHNOLOGY programme, a demonstration project was funded, which was based on the work of Griffith et al. (75). This project focused on the expression of an early gene, c-fos, in cerebellar granular cells of the mouse as a sensitive and early marker for excitotoxicity, which may lead to a delayed cell death. The cell-culture system used was developed by Schousboe et al. (65), and is now widely used in this field.

Receptor regulation and transmitter imbalance have been widely investigated in various systems. Patch clamp investigations or other electrophysiological applications in primary cell cultures or transfected cell lines have been most prominent in this field in pharmacology and toxicology. Receptors, as well as neurotransmitters, can be detected by various biochemical, immunological or physical methods intracellularly and extracellulary. Additional methods include reporter gene assays, which are used mainly in high-throughput screening.

The replacement of neuronal cell cultures by permanent cell lines is only partly possible at present. The only long-term screening system based on permanent cell lines was developed for the detection of delayed organophosphate neurotoxicity (76).

Astrocytes or oligodendrocytes: Astrocytes are another prominent type of glial cell in the brain. They can be divided into subtypes, depending on their maturation and or function. They are important during the brain's development, in its metabolic capacities, and in brain injury The process of reactive astrocytosis has been described as gliosis, and the final lesion is often referred to as a glial scar. An activation of resting astrocytes occurs after neuronal cell death, through the action of the toxicants themselves. Model compounds used in toxicity studies in vitro have included paraquat, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), kainic acid, trimethyltin and heavy metals; the endpoints for toxicity testing were most commonly the expression of glial fibrillary acidic protein (GFAP) or mRNA. Morphology and proliferation were also sensitive endpoints. Reactive astrocytes were comparable to microglia in their immune responsivity, which could be determined by cytokine production and the expression of class I and class II MHC molecules or adhesion proteins. Astrocytes dedifferentiate during their activation to a more fetal stage by expressing vimentin, MAP-2, NSE or GABA, all of which can be quantitatively determined by using specific monoclonal antibodies (77). The culture of various types of astrocytes is now highly developed, and has been described in the literature over recent decades. Testing for toxicity requires well-differentiated cultures, long-term culture and, in some cases, such as paraquat, long-term treatment (73).

Oligodendrocytes, the cells involved in myelin synthesis, organisation and maintenance, have unique characteristics. These cells can synthesise a large amount of membrane. They are also able to respond to numerous environmental signals thought to underlie "oligodendroglial plasticity". Mechanical separation from the monolayer of astrocytes was described over 20 years ago (78). In vitro models for dysmyelination and demyelination have been established for hypoxic injury, ethanol or other toxicants, oxidative stress, inflammatory processes and infections. These effects were characterised by morphology and by specific antibodies against galactocerebroside, A2B2 or myelin-associated protein (79). Oligodendrocytes are derived from postnatal dogs and need long-term conditions for maturation and/or toxicity testing.

For both astrocytes and oligodendrocytes, specific surface markers are changed following toxic insult, and these can be assessed using modern techniques such as cell imaging or fluorescence activated cell sorter (FACS) scan analysis.

Microglia: In recent years, microglial cells have become recognised as the "brain's representatives of the immune system". The ability to respond quickly to brain lesions bears many similarities to inflammatory reactions and, accordingly, some researchers postulate that microglial products are potentially neurotoxic. Microglia can respond rapidly to neuronal injury, whether induced by a neurotoxic agent, by physical trauma or by some other means. The response of microglia to neuron damage is characterised by a number of different events, which include mitosis/proliferation, changes in morphology and phenotype, and altered secretory activity. All these features have to be mimicked in vitro when testing toxic compounds in dissociated cell cultures (80) or organotypic brain slices (81). The endpoints used in testing are, in addition to cytotoxicity, cytokine expression and production, morphology, phagocytosis and proliferation (82). The activation of reactive microglia after treatment with toxicants, such as ethanol, kainic acid or trimethyltin, occurs within hours, but the production of mature, well-characterised microglia requires 1-2 weeks. Therefore, a long-term culture system is necessary cell-culture models and endpoint measurements have been successfully developed, including the expression of markers on reactive or non-reactive microglia, that allow the effective screening of cell populations in the FACS scan.

In summary, the use of primary glia and neuronal cell cultures for toxicological investigations has been adopted widely in academia and in industry. These cultures are used for screening purposes, as well as for mechanistic investigations following unexpected findings in animals. Results from these in vitro studies may also become accepted by regulatory authorities, once they have been scientifically validated and published in peer-reviewed journals. Of the most frequently used cell systems (Table IV), efforts should be focused on simplifying test parameters by means such as electrophysiological applications, the use of specific dyes instead of time-consuming extractions, or toxicogenomics.

Table IV: The most widely used cell types and culture conditions in neurotoxicology

Cell type	Preparation	Reference
Neurons: cortex, hippocampus	Fetal rat; adherent cultures on lamine-coated plates: OPTI MEM/B27 medium Separation: mechanically	73
Neurons: cerebellum	Postnatal mouse; adherent cultures on poly-D-lysine plates: DMEM/10% FCS Separation:enzymatically	75
Neurons: various brain areas	Reaggregation cultures Fetal rat; roller bottles DMEM + supplements: serum-free Separation: mechanically	83
Neurons: various brain areas	Organotypic brain slices Roller tube	154
Glia: microglia	Fetal rat; adherent cultures isolated from mixed astrocyte cultures; DMEM (4.5 g/l glucose)/10% FCS Separation: shaking	155
Glia: microglia	Reaggregation cultures Fetal rat; roller bottles Serum-free	81
Glia: astrocytes	Fetal rat; adherent cultures DMEM (1 g/l glucose)/10% FCS Separation: mechanically	82

Mixed reaggregation cultures from neurons and astrocytes have been developed for long-term screening in toxicology, with the advantage of rebuilding the cytoarchitecture of the nervous system, which is not possible in static cultures (83). However, this benefit is limited by the complexity of these models, which limits the value of this technique for routine screening.

Brain slices: Isolated brain slices from the hippocampus are frequently used in toxicology for the detection of excitotoxic or convulsive properties of drugs. The neuroactive potentials of such compounds are identified in high-dose treatments in animals, such as dogs or monkeys, or late in clinical studies. Therefore, a sensitive screening system in vitro would be helpful, because these effects are barely detectable in rodents. Fluoroquinolones, as a pharmaceutical example, and various pesticides, can be tested in this model (84). The other important endpoint is the detection of learning and memory deficits. For some agrochemicals, complex in vivo experiments were necessary to detect these effects in rodents. Long-term potentiation of the hippocampus slice is a well-accepted alternative method for this endpoint (85). This technique is also used for agrochemical testing in vitro (84).

The Need for Long Term In Vitro Toxicity Testing

Methods and techniques

In order to develop reliable and practicable procedures that, in the near future, will allow for in vitro long-term toxicity studies, the few currently existing methodologies need to be improved and novel methods need to be developed. This will require a multistep strategy with concerted interaction between scientists having expertise in several fields of cell culture and tissue culture technology, cell biology, cell physiology, molecular biology, biochemistry and analytical chemistry.

Aims and purpose

The development of long-term in vitro toxicity testing methods will depend largely on the availability of more-advanced cell-culture techniques. These should guarantee organ-level culture systems that stably express as many functional properties of their in vivo counterparts as possible, and which can be maintained over prolonged periods of time under organotypic conditions. Furthermore, these culture systems should ensure that the administration of test compounds can be performed in a way comparable to that which occurs under in vivo conditions, in a continuous or continual manner.

A reductionist approach involving the establishment of homogeneous cultures of a single cell type in adequate quantities will not fulfil the needs of long-term toxicity testing. The maintenance of highly differentiated cell phenotypes matching the in vivo situation, with dependence on interaction with neighbouring cells through paracrine signals, will need advanced co-culture systems, particularly if long-term in vitro testing is aimed at mimicking the conditions that develop in an organ in a human or in the corresponding in vivo animal models during long-term exposure.

Because the development of chronic disease states is orchestrated by signal mechanisms, for example, inflammatory processes causing triggering of a disturbance in the interaction between epithelial, mesenchymal and neuronal cells with glial, vascular endothelial and immune cells, the design of appropriate co-culture systems will be necessary. In order to establish organotypic culture conditions, the use of conventional static cultures or co-culture technologies will not suffice, and will have to be replaced by systems that allow for the permanent renewal of the culture medium, or a part of it, at variable rates.

Biological Systems: General Considerations

Cell cultures

The currently available cell-culture techniques focus on the use of either primary cultures or continuous/immortal cell lines, usually grown on solid substrates (glass or plastic surfaces). Some of the cell lines have developed spontaneously from primary cultures. Others have gained immortality after transfection with defined viral genes, for example, SV40 large T antigen.

Primary cultures are still considered by many investigators to reflect most closely the differentiated morphology and function of their in vivo ancestors. Unfortunately, this does not apply in the majority of cases. Primary cells in culture quickly dedifferentiate or transdifferentiate, i.e. they lose or change a number of key metabolic functions, such as the phase I and II detoxification systems or the gluconeogenic pathway. In addition, cell types with a well-developed oxidative metabolism quickly change to glycolytic energy production, since the culture conditions do not provide sufficient oxygen to the cells. These changes develop progressively, and therefore do not provide stable cell functions over prolonged periods of time, which would be necessary for long-term testing. Quite often, depending on the cell type, their lifespan is restricted to only a few days or passages.

Cell lines, compared with primary cultures, bear the advantage of virtually unlimited life-spans. The phenotypes are stable over many passages, providing that the culture conditions, especially the type of culture medium, remain unchanged. Due to their prolonged maintenance under non-organotypic or non-physiological conditions, cell lines may also, however, have lost several of the differentiated functions of their in vivo counterparts. They may have started to display functions not normally expressed, due to the fact that the vast majority of cell lines represent a heterogeneous population of cells. A good illustration is the ability to select or adapt clones with different metabolic capabilities or transport characteristics from the renal proximal tubular LLC-PKl wild-type cell line (54).

Because cell cultures were originally developed for the propagation of viruses needed for vaccine development, almost all of the established cell-culture methodologies, especially with respect to culture nutrients, have been, and still are being, designed for the selection of proliferating cells. This remains true for cell cultures used by biochemists and physiologists to study membrane traffic, gene expression, transport, membrane electrophysiological properties, and intracellular signal cascades.

The efforts undertaken to create culture media for the generation of non-proliferating, but differentiated, cells have been limited to the development of media supplemented with certain hormones, growth factors or chemicals (86). In most of the trials, no changes with respect to the basal composition of the media have been made. Very few investigations defining the physical conditions of medium application, static cultures, culture medium volumes, growth support for cells, etc., exist at present.

It has recently been shown for renal epithelial cells in static cultures that medium volumes markedly affect cell metabolism (87). Similarly, distinct changes in medium composition can produce new cell phenotypes. For example, the transient omission of glucose, but maintenance of carbon sources for nucleoside synthesis (for example, uridine or pyruvate) lead to a selection of gluconeogenic phenotypes in the renal LLC-PK1 and OK cell lines (87). Regrowth in the presence of glucose demonstrated constitutive expression of the formerly missing gluconeogenic key enzymes, fructose-1,6-bisphosphatase and phosphenol pyruvate carboxykinase (PEPCK). With respect to liver cells, it has been demonstrated that high concentrations of pyruvate in the medium can markedly improve the life-span of isolated hepatocytes in culture (29). Growth of epithelial cells on microporous substrates, with nutrients supplied from both the apical and basolateral sides, could induce the expression of vasopressin receptors (88).

Similar observations have been reported when rabbit primary cultures of renal cortical collecting ducts were maintained on microporous supports under constant medium perfusion (89).

There is general agreement that all culture systems used for long-term testing should preferably be of human origin. Although this could be achieved by the use of primary cultures, their establishment is constantly hampered by the restriction of the availability of samples, usually from surgical sources, and the requirements for ethical permission for use of tissue from patients, by the fact that life-span of cultures is limited, and because such cultures have a limited capability to be passaged.

Furthermore, it has to be kept in mind that cell phenotypes can change rapidly, depending on the culture conditions. For this reason, surrogate cells have to be used. These can be cell lines that have formed spontaneously (non-tumorigenic cell lines), cell lines that have been created by immortalisation of primary cultures (transformed), or cell lines that have been derived from certain differentiated turnours. The vast majority of these cell lines are, unfortunately, not well characterised with respect to their functional characteristics (for example, metabolism, transport), even if they have been derived from transgenic animals. Thus, the only reasonable alternative at present is to use functionally well-characterised cell lines of animal or human origin. The cell lines available for the liver and kidney are listed in Tables V and VI.

Table V: Relevant biological in vitro systems for nephrotoxicity testing with potential for long-term studies

Cell type	Species and organ of origin	Primary culture	Cell line	Trans- fected	Spon- taneous	Compounds tested (acute toxicity)	Specific characteristics
Monoculture
LLC-PK1	Pig kidney		+	-	+	antibiotics, detergents (SDS), CsA, paracetamol, cisplatin, cytokines, heavy metals	Proximal tubular epithelial cells
LLC-PK1-F+	Pig kidney					antibiotics, detergents (SDS), CsA, paracetamol, cisplatin	Proximal tubular epithelial cells, gluconeogenic
MDCK	Dog kidney		+			detergents	Collecting duct (intercalated cell?), epithelial cell
HK-2	Human kidney		+	+			Proximal tubular epithelial cells, papilloma virus transfected
RPTEC	Human kidney	+					Proximal tubular epithelial cells, commercially available from Clonetics^TM
HPT	Human kidney	+				antibiotics, detergents (SDS), CsA, paracetamol, cisplatin	Proximal tubular epithelial cells, derived from surgical samples
RPT	Rabbit kidney	+				antibiotics, detergents (SDS), CsA, paracetamol, cisplatin	Proximal tubular epithelial cells
HGMEC	Human kidney	+				cytokines	Glomerular endothelial cells, derived from surgical samples
HGMC	Human kidney	+				cytokines	Glomerular mesangial cells, derived from glomerular explant cultures
T-MEC	Human kidney		+	+		cytokines	Glomerular mesangial cells
5A32	Human kidney		+			cytokines	Dermal microvascular endothelial cell mesangial cell
Monoculture
HPT- HGMEC	Human Human	+ +				cytokines	Co-cultures of primary renal epithelial and microvascular endothelial cells
HPT- 5A32	Human Human	+	+	+		cytokines	Co-cultures of primary renal epithelial cells and a human dermal microvascular endothelial cell line
Hk-2- HGMEC	Human Human	+	+	+		cytokines	Co-cultures of renal epithelial and dermal microvascular endothelial cell lines
HGMC- HGMEC	Human Human	+ +				cytokines	Co-cultures of primary glomerular mesangial and endothelial cells
HGMC- 5A32	Human Human	+	+	+		cytokines	Co-cultures of primary glomerular mesangial cells and a dermal microvascular endothelial cell line

MDCK = Madin Darby canine kidney; SDS = sodium dodecyl sulphate

Table VI: Relevant biological in vitro systems for hepatotoxicity testing with potential for long-term studies

Cell type	Species and organ of origin	Primary culture	Cell line	Trans- fected	Compounds tested (acute toxicity)	Specific characteristics
Monoculture
PHC	Rodents or human	+			numerous chemicals and drugs	Primary hepatocyte cultures derived from rodents, other species or human
Hep G2	Human		+		numerous chemicals and drugs	Hepatoma cell line
Kupffer cells	Rodents or human	+			LPS cytokines	Primary hepatic monocytic cells derived from rodents, other species or human, with characterisitics of macrophages
HEC	Rodents or human	+				Primary hepatic endothelial cells (sinusoidal cells) derived from rodents, other species or human
THP-1	Human monocytic cell		+		cytokines	Monocytic tumour cell, which can be differentiated to macrophage-like cells
BDC	Rodents or human	+				Primary cultures of biliary duct epithelium derived from rodents, other species or human
3T3	Mouse fibroblasts		+		numerous chemicals and drugs	Fibroblasts exhibiting contact inhibition derived from Swiss albino mouse
Co-cultures
PHC- Kupffer cells	Rodents or humans	+ +			LPS cytokines	Primary hepatocyte cultures
HPC- BDC	Rodents Human	+ +			numerous chemicals and drugs	Co-cultures or primary hepatocytes and biliary duct epithelial cells
HPC- HEC	Rodents or human				numerous chemicals and drugs	Co-culture of primary hepatocytes and hepatic endothelial (sinusoidal) cells
PHC- 3T3	Human and rodent				numerous chemicals and drugs	Co-culture of primary hepatocytes and a fibroblast cell line
HEP G2- THP-1	Human		+ +		LPS cytokines	Co-culture of primary heptaocytes and a human macrophage cell line

BDC = biliary duct cells; LPS = lipopolysaccharide; PHC = primary hepatocytes.

In the case of neurotoxicity testing, tissue culture and, more specifically, slices of certain areas of the brain (of human origin, if available), can be maintained over extremely long periods (months) in culture and can thus offer an excellent possibility for studying the long-term effects of new chemical entities.

In contrast to the other culture systems listed above, brain slices represent a prototypical non-proliferative type of cell culture. In the future, the controlled differentiation of stem cells may provide a source for certain tissues, for example, neural, cardiomyocyte and haematopoetic tissues.

Cell culture systems

Static cultures: Static cell cultures represent the classical approach in cell-culture technology. Primary cells or cell lines are seeded onto solid surfaces or growth supports made of a variety of plastic materials (polystyrene, polycarbonate, polytetrafluoroethene [PTFE; TEFLON®], PTX polyester, or PVC) or glass. The surfaces are usually hydrophobic and can be manipulated with regard to their surface charges by specific pretreatments. The surfaces can also be coated with extracellular matrix materials such as collagen, fibronectin or laminin or the matrix material, Matrigel®, which influence epithelial cell orientation and phenotypic expression due to the interaction of receptor sites on the cell surface with specific sites in the matrix. Complex matrix materials, such as Matrigel, can contain varying amounts of impurities in the form of growth factors or undefined autocrine or paracrine signal substances, which could affect functional differentiation (90).

The major disadvantage of static cultures is that the medium composition is continuously modified by the cells, and needs replenishment at defined intervals. Furthermore, the access of medium to the growth-support side of the cell is not, the same as that in vivo, and may therefore affect the cell phenotype (88). In addition, gaseous exchange (C0₂, 0₂) is affected. The latter can be improved by using oxygen-permeable and/or carbon dioxide-permeable, thin plastic supports (such as Permanox).

The nutrient access to cells is improved by using microporous growth supports, for example, in the form of filter wells. This technique also allows the co-culture of different cell types on each side of the support, as well as the administration of test compounds from different sides, which may be of importance for epithelial and endothelial cell barriers. Long-term and repeated dosing of a test compound is difficult with this system, since the nutrient medium has to be replaced frequently. The changes in nutrient consumption and metabolite production result in a medium composition that is continuously changing, and this influences the metabolic state of the cultured cells. Microscopic examination of cultures may sometimes be problematic, since not all growth supports are readily transparent. Appropriate, but not always easy to perform, fluorescent microscopy techniques must be applied in these situations.

Non-static cultures: The most widely used non-static culture systems are roller and micro-carrier cultures. In roller cultures, if cells are adhesive, they will gradually attach to the inner surfaces of culture bottles and grow to form a monolayer. This system has three major advantages over static monolayer culture: a greater surface area; a constant, but gentle agitation of the medium; and an increased ratio of medium surface area to volume, allowing enhanced gaseous exchange to take place through the thin film of medium over the cells not actually submerged. The disadvantages are the difficulty of examining cultures microscopically, and the access of the medium to the cells from only one surface.

In rocking or shaking cultures, the culture dishes are tilted slightly at a defined frequency. This exerts some mechanical stress on the monolayers and improves the gaseous exchange in a way similar to that which occurs with roller cultures (91).

In micro-carrier culture, monolayer cells are grown on plastic microbeads of approximately 100 µm diameter and made of polystyrene, sephadex, polyacrylamide, collagen, or gelatin (92). This culture system increases the maximum ratio of surface area to medium volume, and has the additional advantage that cells can be treated in suspension. An advantage of this technique would be the possibility of renewal of the medium at a constant rate.

In the perfusion culture technique, because the supply of medium and gaseous exchange become limiting at higher cell densities (93), a combination of microporous growth support and continuous replenishment of the medium was developed, originally based on the use of hollow microporous plastic fibres, maintained in a container permanently perfused with culture medium. A continuous-flow bioreactor, in which cells grow around hollow fibres through which the culture medium flows, was evaluated as a long-term exposure system, with 3-NPA, CdCl₂ and paraquat as model compounds (EU Contract No. 13467-97-11 FIEI ISP GB). The results obtained showed that it is possible to detect toxic effects below a NOEL set by shortterm conventional culture methods. In addition, a large cell population is available to carry out a wide range of assays, including flow cytometry. However, this technology needs further refinement, since it is not easy to handle, very large amounts of cells and reagents are needed, and it is not convenient for medium-throughput screening purposes. The disadvantages of these systems are an imperfect gaseous exchange, the difficulty of using bicarbonate -buffered media, and the complicated procedures required to monitor cell growth by microscopic methods.

For this reason, several modifications have been developed (89, 94, 95). The Minucell^TM system, for example, combines the advantage of filter inserts with constant replenishment of the medium. Multiple filter inserts can be placed in a perfusion chamber. The inserts can even be supplied with different cell types (co-cultures), if the cultures can be maintained in the same culture medium. A modification of this system allowed, for the first time, the use of "gradient cultures" (89). Here, barrier-forming, confluent monolayers, grown on a special filter insert sealed against the perfusion chamber by O-rings, are perfused with media of various compositions on the apical and basolateral side of the epithelial monolayers. Disadvantages of the system are the large dead space in the perfusion chambers and the imperfect gas exchange, as well as the fairly small growth surface area for cells, which makes the assessment of biochemical and molecular biology parameters difficult, or sometimes impossible.

Most of these problems have been overcome by a new perfusion culture system called EpiFlow (95). This system is also based on microporous growth supports. In contrast to the above system, each growth support is larger and is attached to a gas compartment, which ensures constant equilibration of the perfused medium with 0₂ and C0₂, permitting the use of bicarbonate-buffered medium, even outside a C0₂ incubator. The disadvantage is that, under these conditions, microscopic observation of the cultures is impossible. Where microscopic examination is necessary, a variant of the system, in which gas compartments are replaced by glass covers, is available. This set-up, however, demands extra (external) gassing of the supply medium, which needs to be connected to the perfusion chamber by gas-tight tubing.

The main advantages of all perfusion culture systems are that minimal fluctuations in culture medium composition are achieved and that, if appropriately designed systems are used, the gas exchange can be optimised. With most perfusion systems, cell phenotypes that more closely match their in vivo counterparts are produced. Perfusion cultures often become quiescent and lose their potential to proliferate, i.e. they change from proliferative to non-proliferative cultures (89). By using such a culture apparatus, LLC-PK1 cells and renal primary cells have been maintained in culture over periods of more than a month without morphological changes: LLC-PK1 cells could be kept over these extended periods in serum-free media (Walter Pfaller, unpublished observation; Paul Jennings, personal communication). An additional advantage of perfusion culture is that test compounds can be administered continuously at extremely low concentrations. The concentration and the production of degradation products or, if known, metabolites, can be monitored and analysed in the outflowing medium.

Perfusion culture under optimised conditions more closely approaches organotypic conditions. If used in combination with coculture, simplified types of organ level cultures may be achieved.

Such culture systems have been utilised to study the cross-talk by means of human cytokines between renal endothelial and epithelial cells, and between hepatocytes and monocytic cells in a liver co-culture model.

A further advantage of perfusion in combination with constant supply of respiratory gases is that the gas composition can be changed. Hence, hypoxia settings can be altered by changing the oxygen partial pressure (0₂), and respiratory acidosis or alkalosis can be initiated by changing the carbon dioxide partial pressure (pC0₂), or toxic gases can be applied.

Finally, with both the Minuth gradient culture system (89) and the Epiflow system, medium perfusion on one side of a monolayer culture can be replaced by a gas phase, which, for the first time, allows reliable gas-liquid interphase culture (96). Accelerated medium flow may allow the exertion of shear stress, a prerequisite for organotypic cultures that have to include endothelial cells. Last, but not least, non-adherent cells can be included in the perfusion medium to help model the interaction of circulating cells with epithelial or endothelial monolayers. It has recently been shown that perfusion and gassing markedly influence cell differentiation for the renal LLC-PK1 cell line, which under these conditions changed from a glycolytic to an oxidative phenotype, and developed a morphological appearance very close to the proximal tubular ancestor cell in vivo. This transformation was achieved within 5 days of perfusion culture (97).

A slightly modified perfusion system was described by Minuth (89) and consists of a stack of six microporous supports (polycarbonate filter supports) upon which can be grown human hepatoma (Hep G2) and renal proximal HK-2 cells for long-term testing of hepatotoxins or nephrotoxins, respectively. A comparison of static and perfusion cultures of the human Hep G2 hepatocyte cell line exposed to various compounds (DMSO, menadione, paracetamol, Triton X-100, methimazole and iodoacetate) for 2 days under static culture conditions, and for 7 days under conditions of continuous medium perfusion, was performed. The concentration-response curves from Alamar Blue^TM (resazurin reduction to resorufin [981) or cellular gluthathione S-transferase (GST) activity indicators showed that cell survival was higher in statiic cultures, which would be in line with the assumption that accumulated reactive metabolites are responsible for this effect. It has been shown that Alamar Blue reduction takes place more rapidly in dividing cells (99), and can be up-regulated in cells responding to a low-level toxic insult. Most interestingly, however, the hepatotoxic compound paracetamol was found to be about ten times more toxic in cultures perfused over 7 days, than in static cultures exposed for 48 hours. A similar result was achieved with renal HK-2 cells exposed to cyclosporin A under static and perfusion conditions, when lactate dehydrogenase (LDH) and y-glutamyltransferase (GGT) activities were measured in either the supernatant medium or the perfusate (100).

Culture conditions

Requirements for specific components in the culture medium during long-term testing cannot be readily met at present. From several studies, it is known that the composition of the medium determines the cell phenotype obtained. Defined media are currently used for a number of cell-culture applications. These media vary in complexity from Eagle's minimum essential medium (MEM; 101), containing essential amino acids, vitamins and salts, to complex media, such as RPMI 1640 (102), 199 (103), F12 (104), and a wide range of serum-free formulations (86), such as media for keratinocytes. Complex media are often supplemented with extra metabolites (nucleosides, citrate cycle intermediates and lipids). Many cultures, in addition to glucose, contain glutamine as a major energy and carbon source. The concentration of nutrients may be low (F12 medium) or high (Dulbecco's modified Eagle's medium; 105). Most of these media only contain vitamins of the B complex. They should be supplemented with other vitamins when low serum concentrations are used. This is only a short outline; more details can be obtained from any one of the many handbooks on cell-culture techniques. Whenever serum, the most frequent additive to culture media, is present, the corresponding batch should be checked with respect to its effects on the growth or differentiation characteristic of the cell type cultivated. The constituents of serum, which is usually obtained from calves, are difficult to standardise. In general, they consist of all kinds of proteins, which may not all be required for the cell cultures of interest. Serum contains numerous polypeptides, such as PDGF, FGF, EGF, HGF, IGF-1, IGF-2, HGF and NGF, and hormones such as insulin, glucagon, glucocorticoids and mineralocorticoids, and thyroxin, which act as growth factors or as mitogenic compounds, influencing cell metabolism or the expression of certain receptors. Consequently, they upregulate or down-regulate various signalling cascades, and thereby influence cell growth and differentiation. Some may even represent inhibitors of cell proliferation and/or differentiation, such as TGF-β (106).

Serum-free formulations are of particular interest in the context of long-term toxicity testing, since they would permit a reasonably well-defined standardisation, as compared with serum-containing formulations. Numerous serum-free formulations have been developed and studied (86). These media take advantage of the identified essential constituents required by the cells, and by their mechanisms of action, in order to achieve either promotion of growth or expression of defined differentiated functions. The disadvantage of serum-free media is that they are commercially available for only a limited number of cell types (107). New serum substitutes have been brought onto the market in recent years, but are not yet sufficiently well defined with respect to their usefulness for the wide variety of systems used in vitro toxicity testing. With the problems related to bovine spongiform encephalitis (BSE), and the increasing ethical concerns related to obtaining fetal or neonatal calf serum, this area is likely to gain in importance in the near future.

The development and use of long-term cultures will certainly need more and better-defined culture media that will ensure the maintenance of the differentiation status of the cells over long periods.

Monitoring of culture systems

The maintenance of in vitro systems for long-term testing requires permanent control of their status. Such quality monitoring must be reasonable and practicable, and should deliver information on the viability of cells, their rate of proliferation, and their functionality related to their differentiated status. This can be performed by either invasive or noninvasive procedures. The latter are preferred, especially if they are based on parameters that can be assessed by analytical methods applied to the culture perfusate. The parameters monitored can be concentrations of molecules constitutively released by the various cell types kept in culture, metabolic conversion of certain nutrients or markers, consumption of oxygen, or production and release of certain metabolites. These metabolites and their parent compounds should have low toxicity, or they will themselves affect the parameters monitored. Another possibility for monitoring the status of a culture is the examination of its morphological appearance by microscopical techniques. However, this can be a problem, in that cells can retain their morphology, but lose specific receptors over time; for example, HMEC-1 cells can respond to histamine (108), but this responsiveness is lost over time in culture. In addition, and in analogy to histopathology in animal testing, some use of invasive methods will be unavoidable.

Figure 1: Non-invasive monitoring parameters

Figure 2: Invasive monitoring parameters

Exposure to test items

The design of long-term toxicity studies for small molecules should ensure that their toxic effects, if they exist, should be demonstrated in functionally and biologically relevant in vitro systems. The design should be robust enough to ensure that the results are reliable, and the numbers of false-negative results are reduced to an absolute minimum. The study design should include the scientific rationale and a comprehensive written protocol that describes the objectives of the study, and the methods for conducting it. This protocol should provide sufficient details to enable technical personnel to conduct the study as defined, and should allow scientists to obtain study data as planned, in accordance with GLP (109). It should be developed and evaluated according to the scheme recommended for prevalidation studies (110).

Test materials

Because long-term exposure of the tissue can be required, additional considerations may be necessary that might not be relevant in short-term acute toxicity studies in vitro. The quality and stability of the test material must be carefully determined before starting the study.

The test material employed in a long-term test should be produced in essentially the same way, and at approximately the same scale, as the final product, so that it will contain similar impurities at approximately the same concentrations as the final product. If the test article is intended to be administered as a special dosing form, it must be stable in the carrier. If the test article is a formulation, such as an injectable antibiotic, the components of the tested formulation should be identified and be similar, if not identical, to the final product at formulation.

Data on the stability of the test substance in the dosing form must cover the range of concentrations to be used in the long-term toxicity study, and the longest time anticipated between formulation and final utilisation.

Information must be available on the physicochemical properties of all components of all fractions of the test compound with respect to interactions with materials of the cell-culture equipment, particularly tubing and cell-culture containers. Quite often, solvents and drugs dissolve in silicone rubber and Tygon tubings, or adsorb onto polycarbonate, polyester and polystyrene. Less adsorption or interference occurs with Viton, glass, polyethylene and PTFE.

Much of the supporting information must be generated from pilot studies or from special analytical studies completed before the long-term toxicology study is initiated. The data generated in the studies can only be as reliable as the sampling procedures, analytical methods, and the statistical methods used to analyse the data. Therefore, the existence of optimised validated methods for measurement of the test article alone or in complex matrices is a prerequisite for longterm toxicity testing. It is equally important that the results of the studies should be rigorously documented and included with documentation for the long-term studies. A correct study can only be performed when appropriate supporting characterisation and analytical information relating to the test article are available.

Endpoints in long-term toxicity testing

Since no generally accepted model for long-term in vitro toxicity testing exists, it is very difficult to recommend defined endpoints. Based on experience from acute toxicity studies and investigations, it is, however, possible to focus on mechanistic aspects of toxicity testing, and to consider a battery of reliable and relevant endpoints for the tissues chosen as targets.

As stated in an ECVAM workshop report on acute toxicity testing in vitro and the classification and labelling of chemicals (111), three main types of chemically induced toxic effects can be defined at the cellular level: general (basal) cytotoxicity; toxicity altering specific cell function; and selective cytotoxicity. Chronic repeated exposure to chemicals can result in the induction of certain metabolic or enzymic pathways that are not relevant in the case of acute cytotoxicity. For example, chronic exposure of the upper respiratory tract in humans to irritant chemicals results in squamous metaplasia of the respiratory epithelium (112). This illustrates an important 4ifference that can occur between acute and chronic toxicity. Although both can result in cell death through either necrosis or apoptosis, chronic or repeat exposure can result in functional, structural, biochemical and physiological changes that do not necessarily result in necrosis or apoptosis.

Whereas there is evidence from the MEIC project that human chronic toxicity can be predicted by using acute assay results or general cytotoxicity endpoints (7), it remains to be established whether this also applies to cell types with highly differentiated functions. There are also sets of cellular events that through their up-regulation or downregulation will signal that the cell targets have been adversely affected. These can be divided into two groups. The first group consists of the so-called "housekeeping" activities, i.e. events that occur in all or most cells from various tissues (Table IV), for example, the effects of chemicals on squamous differentiation of keratinocytes as a model for respiratory metaplasia (113). The second category of events includes those that reveal adverse reactions that become cumulative on chronic or repeated exposure. It is well established, for example, that with Hepatitis B there is a steady increase in the activity of ALAT, whereas with Hepatitis C, the change is temporary with a subsequent return to normal. These cumulative effects can also be temporary; so the timing of the assays and the ability to re-assay at appropriate time intervals is important.

There are certain effects that accumulate over time. For example, redox changes can lead to changes in hydroxynonenal through lipid peroxidation. Lipid peroxidation endproducts can be revealed by accumulation of lipofuscin. There are various lipid peroxidation products, and 4-hydroxynonenal is known to be particularly cytotoxic to neural cells, causing cell dysfunction (114). Similarly, the sensitivity of T cells to hydroxynonenal has been demonstrated for Jurkat T cells, and has been linked to the level of glutathione reductase, with a reduction in GSH levels (measured by using monobromobimane as the thiol probe) occurring on exposure to hydroxynonenal (115). The capacity of cells to maintain the glutathione pool is vital for their protection from cytotoxicity in both the acute and chronic phases of a response (74).

A number of assay systems have been suggested when considering oxidative stress. DT-diaphorase plays a key role, and has been quantified by using dichlorophenol-indophenol as a substrate (50). It has been shown that the Alamar Blue reduction assay involves DT-diaphorase (116). GSH and total thiol levels in cells have been measured by using monobromobimane (117). Whereas these commutative effects on the cells' capacity to resist the adverse effects of oxidative stress are important in terms of cell death, other effects are more subtle.

Changes in the ability of cells to respond to chemical challenge have been observed, such as when fibroblasts exposed to certain of the MEIC chemicals at low, normally subacute, toxic levels, became less or more sensitive to the same chemical or to other related and non-related materials (118). Similarly, changes in the ability of epithelial cells to restore tight-junctional integrity has been demonstrated following exposure to surfactants (119). When considering suitable specific endpoints for long-term toxicity, the relevance of this response is important for particular organs, as outlined in the ECVAM acute toxicity workshop report (111). In the absence of data relating to the precise effects on human target organs of long-term exposure to toxicants, the majority of toxic effects will use endpoints relevant to the "housekeeping" functions.

It is also necessary when considering cumulative-effect endpoints to include metabolism-mediated changes, which can play a significant role, and to take into account the fact that metabolism in one organ can result in toxic changes in another. With certain materials, it should be possible, through HPLC analysis, to detect the in vitro production and stability of metabolites produced, for example through liver metabolism. Co-culture systems can be useful in determining whether such stable metabolites cause adverse effects in other organs. Examples of such co-culture systems, for example, hepatocytes with 3T3 cells, have been published (120-122).

When considering the target organs for which long-term culture techniques have been, or are rapidly being, developed, namely the liver, kidney, lung and immune system, specific relevant endpoints can be chosen.

In the case of the liver, the medium from the long-term culture systems can be monitored by using the basic automated assay systems that are used clinically to monitor liver dysfunction. These include the standard enzymes, GPT, GOT, LDH and cytochromes P450, GSH conjugates, and the production of albumin and heat-shock proteins. With certain long-term methods that retain the 3-dimensional architecture of the liver, with hepatocytes retaining the biliary system, the level of bile acid secretion can also act as a useful indicator of toxic damage. Where possible, the tissue cultures should be monitored for morphological changes as, in vivo, the location of the dysfunctional hepatocytes is an important indicator of the type of toxic compound that the liver has been exposed to.

The kidney has a more complex architecture, in which part of the functional activity relies on vascular integrity and blood pressure. At present, this aspect of the kidney's function cannot be modelled, and hence it is not possible to employ relevant endpoints pertinent to such functions in vivo. However, as understanding of chemicals that cause changes in blood pressure grows, it may be possible to devise suitable endpoints. Representative cell systems that retain certain aspects of the capsule, proximal or distal tubules, loop of Henle, and collecting ducts are available.

For both the proximal and distal tubules, the ability to transport ions across the cells selectively can be monitored through changes in electrical resistance (123, 124). The integrity of cellular adhesion systems, including tight junctions, can be monitored by inhibition of the passage of fluorescent dyes, such as sodium fluorescein (119, 125, 126), or fluoreseently labelled non-toxic marker molecules of particular molecular sizes, such as the dextrans or inulin (127). In addition, in the proximal tubules, the presence of a brush-border with high levels of enzymes such as alkaline phosphatase and y-glutamyltransferase will be suitable markers for impaired brush-border activity (128). The loop of Henle cells and collecting duct models can be monitored by measuring specific distal tubular enzymes, such as hexokinase, a glycolytic enzyme (129).

The lung is an important organ, but although at present there are limited models for generating data comparable with those obtained in vivo, in vitro systems are being developed to examine the squamous keratinisation of respiratory epithelium in the trachea. This stratification is a chronic response to irritation, and results in a degree of keratinisation that can be modelled by using a fluorescent probe for transglutaminase-mediated cross-linking of glycine and lysine residues (113). New methods of cultivating Type I and Type II alveolar pneumocytes, including air-liquid interface cultures of these cells, are under development. These have been used to develop a new procedure for measuring surfactant release in pneumocyte type II function. This assay, in combination with perfusion techniques and video imaging, may represent a promising strategy for monitoring long-term effects on the alveolar epithelium (130).

For the immune system, where receptor-mediated control of functionality is crucial, the release of cytokines can be measured by specific ELISA assays. It is also important to determine the possible overstimulation and immunosuppression of T and B cell functions. In addition, cytokines released will act as signals to the cell itself or to other cells, either as growth factors or mitogens, which may result in the alteration of transcriptional and translational activities of the cells affected. Thus, long-term exposure to such molecules may markedly alter the state of cellular differentiation.

Conclusions and Recommendations

The current reliance on chronic tests in laboratory animals is of limited value, because of species differences, general economic and logistical considerations, and the high cost of mechanistic studies.
The development and validation of procedures for long-term testing with in vitro systems represents a very demanding challenge. Nevertheless, there are numerous reasons why there should be concerted attempts to meet these challenges.
In vitro systems can be developed involving human cells (or surrogates, i.e. cell lines of human origin) and tissues. Mechanistic studies focused at the cell and molecular levels are feasible. The use of human in vitro systems is of particular relevance for the prediction of human safety. Also, the assessment of harmful effects induced by biotechnology-derived human therapeutic proteins and peptides is only possible when using human cell-based test systems.
The identification of potential hazards resulting from long-term, repeated exposure to chemicals, mixtures of chemicals and products of various kinds (for example, medical devices) is an important aspect of toxicity testing for the protection of human beings, other species, and the environment in general. These activities could take place in the light of parallel developments in relation to biokinetic modelling, the use of biomarkers, and toxicogenomics and proteomics. It should be noted that the prediction of outcomes through the measurement of intermediate, predictive biomarkers is not generally possible unless there is a clear pathway of causality, i.e. the underlying mechanisms should be known. Genomics and proteomics may be of value but, as with all large collections of data, it is essential to ask the relevant question, rather than indulge in a fishing expedition.
One of the major problems with in vitro toxicity testing is the lack of a precise definition of "long-term", as compared with in vivo approaches. "Long-term" in this context should be defined as meaning a minimum of 5 days, but the development of systems usable over several months is necessary and feasible.
Relatively stable cultures are a prerequisite for the accomplishment of "long-term" in vitro toxicity studies. Presently, such systems as brain-tissue slices, static cell cultures, transformed or immortalised cell lines, and genetically engineered cell lines meet this criterion.
The long-term maintenance of in vitro systems will require the use of new methodological approaches. Refined perfusion culture systems will be needed. Some of the promising systems that are already available should be further developed, with an emphasis on miniaturisation (low perfusate volumes) and practicability, to permit the use of a sufficient number of parallel samples and of relatively high-throughput approaches.
Particular attention needs to be paid to the definition, control and monitoring of the culture conditions (with an emphasis on organ-specific requirements), to the determination of biologically relevant doses, and to establishing meaningful dosing and sampling regimes.
The endpoints used need to be carefully selected for their relevance to general cytotoxic mechanisms on the one hand, and cell-type-specific mechanisms of toxicity or expression of toxic effects, on the other. There is also a need to further develop non-invasive measurements for monitoring effects, which could be applied to the same preparation over long periods.
Systems should be developed for predicting and studying toxicotolerance and multi-drug resistance.
The experimental approach for long-term toxicity studies needs to be redefined. Although IC50/EC50 data collection is suitable for acute toxicity, it is not necessarily relevant for long-term studies. The assumption can be made that a compound that is toxic over a short period will also be toxic over a long period, and thus longterm toxicity testing should be conducted with acute NOEL concentrations of compounds. Such an approach would greatly increase the output of data collected from long-term studies.
It is important that the measurements made in an in vitro model can be validated in a human system. Since it is usually not possible to measure effects directly in an organ in vivo (at least not non-invasively, and not for subtle changes), methods to confirm effects in organs through measurements of relevant criteria in surrogate tissues (generally blood, urine or faeces), or by non-invasive imaging techniques should be developed.
The initial focus should be on identifying compounds with well-established effects for use as reference standards, and on tackling specific problems relevant to particular kinds of tissues and test items (ideally based on a knowledge of effects in humans), rather than on trying to replace chronic testing in vivo as a whole. There are already a number of examples relating to hepatotoxicology, nephrotoxicology and neurotoxicology that provide encouragement for further effort and for identifying and overcoming some of the major problems to be faced in developing relevant and reliable procedures for long-term testing in vitro.
As opposed to acute toxicity test systems, the development of long-term in vitro assays should not be focused predominantly on their ability to perform high-throughput studies.
The pharmaceutical industry requires information on long-term effects as much as it requires acute-toxicity data. The further complications of systems for long-term studies should not preclude their industrial application, provided that such systems can be evaluated as a predictive indicator of long-term toxicity testing. Our goal is not to provide simple toxicity-testing models, but more-relevant models. The nature of long-term toxicity and the maintenance of in vitro systems with physiological characteristics close to the in vivo conditions require more-complicated systems than exist at present.

Acknowledgements

The workshop participants gratefully acknowledge the generous support of this workshop by the Austrian Federal Ministry of Science, Education and Culture, in addition to the support provided by the European Commission, via ECVAM.

Dr Björn Ekwall, who was a participant of this workshop has sadly passed away. His critical contributions to this workshop and his engagement and enthusiasm for the world of in vitro toxicity testing, which culminated in the creation of the MEIC programme, will always be remembered.

References

D'Arcy, P.F. & Harron, D.W.G. eds (1992). Proceedings of the First International Conference of Harmonization, Belfast, UK: Queen's University.
Zbinden, G. (1991). Predictive value of animal studies in toxicology. Regulatory Toxicology and Pharmacology 14: 167-177.
Gralla, E.J. (1981). Protocol preparation: design and objectives. In Scientific Considerations in Monitoring and Evaluating Toxicological Research (ed. E.J. Gralla), pp. 1-26. Washington, DC, USA: Hemisphere.
Zbinden, G. (1993). The concept of multispecies testing in industrial toxicology. Regulatory Toxicology and Pharmacology 17: 85-94.
Soto, A.M., Lin, T.M., Justica, H., Silvia, R.M. & Sonnenschein, C. (1998). An "in culture" bioassay to assess the estrogenicity of xenobiotica. Journal of Clean Technology, Environmental Technology and Occupational Medicine 7: 331-343.
Dierickx, P.J. & Ekwall, B. (1992). Long-term cytotoxicity testing of the first twenty MEIC chemicals by the determination of protein content in human embryonic lung cells. ATLA 20: 285-289.
Ekwall, B., Clemedson, C., Ekwall, C., Ring, P & Romert, L. (1999). EDIT: a new international multicentre programme to develop and evaluate batteries of in vitro tests for acute and chronic systemic toxicity. ATLA 27: 339-349.
Scheers, E.M., Ekwall, B.A. & Dierickx, P.J. (2001). In vitro long-term cytotoxicity testing of 27 MEIC chemicals on Hep G2 cells and comparison with acute human toxicity data. Toxicology In Vitro 15: 153-161.
Hanley, A.B., McBridge, J., Oehlschlager, S. & Opra, E. (1999). Use of a flow bioreactor as a chronic toxicity model system. Toxicology In Vitro 13: 847-851.
Hartung, T. (2001). Summary: EU Standards Measurements and Testing Project PL95-3407. Development of a Standardised In Vitro Methodology for Hepatic and Renal Toxicity Testing. ATLA 29: 493-495.
Wolf, A., Schramm, U., Fahr. A., Aicher, L., Cordie, A., Trommer, W.E. & Fricker, G. (1998). In vitro evaluation of hepatocellular effects of Cylosporin A and its derivative SDZ IMM 125. Journal of Pharmacology and Experimental Therapeutics 284: 817-825.
Schramm, U., Lorez, C. & Wolf, A. (1999). Species-specific induction of apoptosis in rat and dog liver slices by SDZ 249 665. Toxicological Sciences 48: 9-15.
Wolf, A., Trendelenburg, C.F., Fernandez-Diez, C., Prieto, P., Houy, S., Trommer, W.E. & Cordier, A. (1997). Cyclosporine A-induced oxidative stress in rat hepatocytes. Journal of Pharmacology and Experimental Therapeutics 280: 1328-1334.
Guguen-Guillouzo, C. & Guillouzo, A. (1983). Modulation of functional activities in cultured rat hepatocytes. Molecular Cell Biochemistry 53/54: 35-56
Schuetz, E.G., Omiecinski, J., Muller-Eberhard, U., Kleinmann, H.K., Elswick, B. & Guzelian, P.S. (1993). Regulation of gene expression in adult rat hepatocytes cultured on a basement membrane matrix. Journal of Cell Physiology 134: 309-329.
Sirica, A.E. & Pitot, H.C. (1980). Drug metabolism and effects of carcinogens in cultured hepatic cells. Pharmacological Reviews 31: 205-228.
Bégué, J.M., Guguen-Guillouzo, C., Pasdeloup, N. & Guillouzo, A. (1984). Prolonged maintenance of active cytochrome P450 in adult rat hepatocytes co-cultured with another liver cell type. Hepatology 4: 839-842.
Fahl, W.E., Michalopulos, G., Sattler, G.L., Jefcoate, C.R. & Pitot, H.C. (1979). Characteristics of microsomal enzyme controls in primary cultures of hepatocytes. Archives of Biochemistry and Biophysics 192: 61-72.
Guzelian, P.S., Bissell, D.M. & Meyer, U.A. (1977). Drug metabolism in adult rat hepatocytes in primary monolayer culture. Gastroenterology 72: 1232-1239.
Morel, F., Beaune, P.H., Ratanasavanh, D., Flinois, J.P., Yang, C.S., Guengerich, F.P. & Guillouzo, A. (1990). Expression of cytochrome P450 enzymes in cultured human hepatocytes. European Journal of Biochemistry 191: 437-544.
Asakura, I., Sugimoto, M., Watanabe, M. & Ito, K. (1993). Glutathione S-transferases in primary cultured rat hepatocytes. Gastroenterologia Japonica 28: 34-45.
Liu, X., LeCluyse, E.L., Brouwer, K.R., Lightfoot, R.M., Lee, J.I. & Brouwer, K.L. (1999). Use of Ca²⁺ modulation to evaluate biliary excretion in sandwich-cultured rat hepatocytes. Journal of Pharmacology and Experimental Therapeutics 289: 1592-1599.
Houy, S (1998). Primary rat hepatocytes in collagen-sandtvich configuration. Characterization and first application as a toxicological in vitro tool. PhD thesis, Department of Chemistry, University of Kaiserslautern, Germany.
Bader, A., Rinkes, I.H., Closs, E.I., Ryan, C.M., Toner, M., Cunningham, J.M., Tompkins, R.G. & Yarmush, M.L. (1992). A stable long-term hepatocyte culture system for studies of physiologic processes: cytokine stimulation of the acute phase in rat and human hepatocyte. Biotechnology Progress 8: 219-225.
Lorez, C., Trommer, W.E., Cordier, A. & Wolf, A. (1996). Cyclosporin toxicity in the hepatocyte collagen sandwich culture. Naunyn Schmiedeberg's Archives of Pharmacology (Supplement 1): 354, R25.
Loreal, 0., Levavasseur, F., Fromaget, C., Gros, D., Guillouzo, A. & Clement, B. (1993). Cooperation of Ito cells and hepatocytes in the deposition of an extracellular matrix in vitro. American Journal of Pathology 143: 538-544.
Coecke, S., Rogiers, V, Bayliss, M., Castell, J., Doehmer. J., Fabre, G., Fry, J., Kern, A. & Westmoreland, C. (1999). T

Novel Advanced In Vitro Methods for Longterm Toxicity Testing

The Report and Recommendations of ECVAM Workshop 451,2

The Report and Recommendations of ECVAM Workshop 45^1,2