In vitro toxicity screening as pre-selection tool

Drug discovery relies on massive screening of compound libraries with in vitro cell-based target assays. These pharmacological screens have been well accepted. For in vitro toxicological screening, this privilege has only been obtained for the Ames, chromosomal aberration and eye irritation tests. At the moment, a number of cellular assays for cytotoxicity, genotoxicity, embryotoxicity, cellular metabolic activation processes and endocrine disruption await general acceptance. From that point onwards, tools will become available to identify unwanted pharmacological or toxicological effects at a much earlier stage in the drug development process.

One of the challenges of the pharmaceutical industry is to bring more compounds to the market in a shorter space of time. The introduction of combinatorial chemistry and high throughput screening technologies not only resulted in an increased number of compounds, but also a greater diversity of potential drug candidates. Therefore, new, fast and predictive tests are needed to select the best drug candidates.

However, the current success rate of newly developed drugs is too low, as only one out of twenty compounds reaches the market. A superior pre-screening strategy with improved selection criteria would undoubtedly lead to a better choice of drug candidates; thereby proving to be the most appropriate way in which to reduce high costs in clinical development. Therefore, the failures made in the past relating to negative pharmacology and toxicology should be taken into account to improve the future quality of the medicines. Throughout 1990 and 2007, the reasons to stop the development of new drugs changed1.

In 1991, 40% of the compounds failed due to bad pharmacokinetics and bioavailability. In 2000, these problems were reduced to only 10% as a result of more reliable preclinical in vitro and in vivo assays. In 2000, clinical effectiveness, toxicological and pharmacological safety combined to 40% of the reasons for failures in drug development. Unfortunately, these numbers remain unchanged in 2007; this emphasises the need for better predictive tests in these areas. The use of new screening technologies has resulted in the prepared amounts of compound also becoming smaller: in vitro assay development should have the highest priority.

An additional beneficial effect of these in vitro tests is the reduction in animal studies that are needed in early preclinical development for compounds with a low success rate.

Strategy

Reanalysis of the Organon databases showed that 50% of the candidate drugs were ceased due to toxicological problems in (pre)clinical development.

Liver, cardiovascular, skin and neuronal toxicity were most frequently identified with the compounds of Roche. Due to another composition of its portfolio, 40% of the toxic events could be attributed to reproductive toxicity and genotoxicity at Organon. The (sub)acute, subchronic and chronic toxicity could be divided into four categories of similar size, being 12% each for liver, heart, kidney and other organs.

When trying to lower the attrition rate, a focus on these aspects during lead optimisation would of course be most beneficial. However, few causes of drug failure can easily be solved with a ‘yes/no’ approach, such as genotoxicity and embryotoxicity, whereas the other types of toxicity might, at best, only lead to a ranking of compounds and a prioritisation.

In vitro cytotoxicity

Interference of normal cell function can be elicited by impairment of basic cell functions such as energy metabolism, cytoskeletal organisation, membrane integrity or by disturbance of cell-specific functions such as protein production and excretion, uptake and secretion of metabolic waste products, glycogen storage and electrical conductivity. Whereas general cytotoxicity can be determined in almost any cell-type, tissue specific functions may need studies in the cognate relevant cell lines, primary cells or tissue slices.

For studying general cytotoxicity, various laboratories have developed assays measuring easy-to-interpret endpoints, such as membrane integrity, DNA content or endpoints giving clues for the mechanism of toxicity such as glutathione content, intracellular calcium concentration and mitochondrial activity.

As there are many pathways leading to the death of the cell, a combination of assays measuring cell death as an endpoint and tests evaluating mechanistic parameters should ideally be used.

The current situation within Organon is that assays were validated for mitochondrial cellular activity (ATP-Lite, Cyto-Lyte, Alamar Blue), for DNA proliferation (Hoechst 33342), production of reactive oxygen species (DCF), calcein uptake (Calcein-AM) and gluthathione depletion (Monochlorobimane). These luminometric and fluorometric assays were validated with four different cell lines (HepG2, ECC-1, HeLa and CHO)2,3. In a collaboration with Débiton (Inserm, Clermond-Ferrand, France) these assays were also performed with mouse L929 and/or primary human fibroblast cells. The results, with 110 reference compounds being toxic in rats or humans, show that there is a very high similarity between the different cell lines used. The data for ATP-Lite, Cyto-Lite, Alamar Blue and Hoechst 33342 assays were very similar in dose response curves (Figure1 on page 40). The glutathione depletion and calcein uptake measurements had a large overlap with the former four assays. The ROS assay, on the other hand, were less sensitive, with only very few compounds being identified as toxic.

Currently, 70% of the compounds can be characterised as being toxic in these cytotoxicity assays, which is in line with MEIC studies4.

The lack of cytotoxicity for 30% of the compounds might be due to incompetent metabolism in the cells used. Thus, possible reactive metabolites of the compound may not have been formed. Inducing metabolism in the cells with a sort of food cocktail mix might lead to a higher predictivity. Another way to increase the predictivity of cytotoxicity assays is to screen with bioimaging techniques, like Cellulomics, in which several parameters can be examined in one single HepG2 cell with different fluorophores5. In the latter approach, the in vitro prediction of clinically identified toxic compounds was 90%, which is even higher than in vivo rat studies. These assays also have the advantage of a medium to high throughput screening potential.

Besides cell death, cellular proliferation can have serious consequences in vivo. Proliferative potencies of compounds can be tested in conjunction with cytotoxicity assays, measuring the DNA content. Depending on the chemical and pharmacological class of the compound, tailor-made in vitro systems can be developed to study the most frequently encountered toxicity or adverse pharmacological side effects of these drugs. Examples include the development of tests to rank compounds as being specific for one nuclear or G-protein receptor type, one ion channel or as modulators of potassium channels (QT prolongation), or hepatic haem production.

Metabolic activation and/or inactivation

As mentioned above; nuclear receptors are heavily involved in the up-regulation of Phase I and II enzymes and therefore in drug disposition and metabolism. The most relevant nuclear receptors that are involved in the induction of CYP, conjugation and antioxidant enzymes are:

• Pregnane X receptors, which activate CYP3A4, Glutathion-S-transferase (GST), UDP-glucuronosyltransferase (UGT) and sulfonyltransferase (SULT)
• Constitutive androstane receptors (CAR), which constitutively inhibit CYP3A4 and activate CYP2B6
• Farnesoid X receptors, which activate the CYP7A enzymes
• Aromatic hydrocarbon receptors (AhR), which induce CYP1A1, CYP1A2, DT-diaphorase, GST, UGT and aldehyde dehydrogenase
• Peroxisomal proliferator activated receptors, which regulate the transcription of fatty-acyl-CoA oxidase, fatty acid binding protein and CYP4A1
• Liver X receptors, which regulate cholesterol metabolism via CYP7A

All these nuclear receptors induce specific CYP enzymes and particular Phase II enzymes. Moreover, all these receptors form heterodimers with the retinoic acid X receptor α, with exception of AhR and all use the same coactivators and corepressors. Thus, cross-talk between these receptors is inevitable.

Genotoxicity

DNA damage induced by a compound can be measured in several ways; the Ames test is a mutagenicity test that uses different salmonella typhimurium strains that have mutations in genes involved in histidine synthesis and therefore require histidine for growth. After compound incubation, bacteria are grown in a histidine free medium, in which only revertants will survive, the number of these revertants is a measure for mutagenicity.

Another more innovative test with S. typhimurium TA104 cells (Vitotox, Thermo, Finland) shows DNA damage more directly. The bacteria contain a luciferase gene under transcriptional control of the recN promotor. Normally, the recN promoter is strongly repressed, but in the presence of a DNA damaging genotoxic compound the RecA regulator protein recognises the resultant free ends or mismatches in DNA. This results in a cascade of reactions leading to derepression of the strong recN promoter. A second S. typhimurium TA104 strain constitutively expressing luciferase is used for measuring cytotoxicity. Compared to the Ames test, the assay format is much faster and corresponds with high throughput screening procedures to a greater degree. Moreover, for metabolic activation S9 liver fractions of aroclor pretreated rats can still be used, as with the Ames test.

Since bacterial mutagenicity tests can still under predict genotoxicity in eukaryotes, also a yeast strain (GreenScreen, Gentronix, UK) was examined. In this test genotoxic compounds activate a RAD54 promoter, which induces the expression of green fluorescence protein (GFP) within 24hours. A drawback of this system is that S9 liver fractions can not be used and that some of the tested compounds gave autofluorescence at the wave length of GFP. With the set of Organon compounds, this resulted in lower predictive values for Ames positive compounds. The Vitototox and Greenscreen test were able to detect 90.2 % and 44.7 % of the Ames mutagens, respectively (Table 1).

A positive genotoxic score is seen as a strong negative factor for a compound, which usually leads to de-selection of that compound. At present, the Ames II and Vitotox tests are used in the pre-selection phase within Organon. Of each compound, 25-40mg is required and a throughput of two to three compounds per week for Ames II and 8 to 40 for Vitotox assays is feasible. Introducing a genotoxicity assay in the early phase of LO has an important advantage in early de-selection of the lead or in chemically adapting the lead.

Clastogenicity (chromosomal damage) is another factor that leads to the rejection of a compound. In the late development phase, clastogenicity is measured by means of an in vitro chromosomal aberration test and an in vitro or in vivo micronucleus test. Both tests are time-consuming and unsuitable for incorporation in the early research phase. Efforts are being made to develop tests for the early research phase.

Embryotoxicity

Embryotoxicity is one of the most important issues in drug development. Costly (1500-2000 man-hours/test) in vivo tests in rabbits and rats are regulatory requirements for every potential drug intended for use by women with child-bearing potential. In addition, these pivotal tests are scheduled quite late on in development, so high investments have already been made in (pre)clinical studies before the decision to drop the compound on basis of embryotoxicity is made.

Many research groups have put effort in developing easy and cheap in vitro screening assays to predict embryotoxicity. Initially, the goal was to develop assays that could replace the classical in vivo tests. However, embryonic development is too complex to catch all facets in one or several in vitro models. On the other hand, for the purpose of screening, the predictive value does not need to be 95-100%, but a predictive value of 70-80% would suffice.

Recently, under coordination of ECVAM, the Institute for Health and Consumer Protection, an extensive inter-laboratory validation of the three mostly used in vitro embryotoxicity tests have been performed6. The three evaluated assays are:

• The whole embryo culture
• The micromass assay
• The embryonic stem cell culture

A report on this validation is released and the variety of papers published on these systems demonstrates that the systems have a reasonable to good predictive value, sensitivity, specificity and selectivity. In particular, the micromass assay and the embryonic stem cell culture have the potency to be transferred into a high to medium throughput format, respectively.

Conclusion

Although a battery of in vitro assays for toxicology is becoming available, implementation of these tests in regulatory toxicology is extremely low.

The main reason for this is that the methods which are currently used are still recognised as the most effective methods available. This acceptance is based on a wide experience with these tests and the availability of huge databases. Alternative methods will be able to effectively replace the current regulatory required studies only when they will be able to model the in vivo biological complexity. Nevertheless, in vitro methods can play an important role in (pre)clinical development when specific mechanistic issues are involved in the toxicity of a class of compounds, or to aid in understanding a toxicological finding in one of the tox-species and to interpret the human relevance of this finding.

The function of in vitro toxicology in drug development will play an important role in both the reduction of the attrition rate and at a minimum, the refinement of animal experimentation. Nevertheless, in vitro toxicologists should keep in mind a comment made by Fry and George in ATLA (2001)7, which is as follows:

“The question that needs to be answered for the future of (in vitro) toxicology, is not what should we do with our new tools, but where do we have unmet needs and what tools do we need to employ to meet them?”

References

1. I. Kola and J. Landis: Can the pharmaceutical industry reduce attrition rates? Nature Reviews Drug Discovery, issue 3, pp. 711-715, (2004).
2. W. Schoonen et al: Cytotoxic effects of 110 reference compounds on HepG2 cells and for 60 compounds on HeLa, ECC-1 and CHO cells. II mechanistic assays on NAD(P)H, ATP and DNA contents. Toxicology In Vitro, issue 19, pp. 491-503, (2005a).
3. W. Schoonen, W. Westerink, J. de Roos and E. Debiton: Cytotoxic effects of 100 reference compounds on Hep G2 and HeLa cells and of 60 compounds on ECC-1 and CHO cells: I mechanistic assays on ROS, glutathione depletion and calcein uptake. Toxicololgy In Vitro, issue 19, pp. 505-516, (2005b).
4. C. Clemeds and B. Ekwall: Overview of the MEIC results: I the in vitro – in vitro evaluation. Toxicology In Vitro, issue 13, pp. 657-663, (1999).
5. P. O’Brien et al: High concordance of drug-induced human hepatotoxicity with in vitro cytotoxicity measured in a novel cell-based model using high content screening. Archieves Toxicology, issue 80, pp. 580-604, (2006).
6. E. Genschow et al: The ECVAM international validation study on in vitro embrytoxicity tests: Results of the definitive phase and evaluation of prediction models. ATLA, issue 30, pp. 151-176, (2002).
7. J. Fry and E. George: In vitro toxicology in the light of new technologies. ATLA, issue 29, pp. 745-748, (2001).