Você está na página 1de 11

Mutation Research 375 1997.

183193

The comet assay: what can it really tell us?


Andrew R. Collins
a b

a,)

, Victoria L. Dobson a , Maria Dusinska b, Gayle Kennedy c , Rudolf Stetina d

Rowett Research Institute, Greenburn Road, Bucksburn, Aberdeen AB21, 9SB, UK Institute of Preentie and Clinical Medicine, Limboa 14, 83301 Bratislaa, Sloak Republic c Department of Biological and Biomedical Sciences, Uniersity of Ulster at Coleraine, Coleraine BT52 1SA, UK Institute of Experimental Medicine, Academy of Sciences of the Czech Republic, Laboratory of Deelopmental Toxicology, Olesnice O.h. 517 83, Czech Republic

Abstract A range of applications of the alkaline comet assay is covered, from investigations of the physicochemical behaviour of DNA, through studies of cellular responses to DNA damage, to biomonitoring of human populations. The underlying principles of this assay are discussed, and new evidence presented which supports the concept of relaxation of supercoiled loops, rather than alkaline unwinding, as the primary reason for comet tail formation. DNA-damaging agents that do not induce strand breaks directly can be detected when cellular repair processes convert lesions to transient strand breaks; an approach is outlined here which maximises this effect and thus widens the scope of the assay. Purified repair enzymes, applied to DNA during the course of the comet assay procedure, greatly increase the sensitivity and specificity of the assay; recent developments with formamidopyrimidine glycosylase recognising 8-OH-gua and other damaged purines. and uvrABC for bulky lesions. are presented. The kinetics of cellular repair after low doses of oxidative damage have been followed with this modified comet assay. Finally, the successful measurement of biomarkers of oxidative damage in human populations establishes the comet assay as a valuable tool in molecular epidemiology.
Keywords: SCGE; Comet assay; Repair enzyme; DNA repair inhibitor; Molecular epidemiology

1. Introduction The comet assay single cell gel electrophoresis or SCGE. is attractive for many reasons. Apart from the appeal of the images it produces, it is a quick, simple, sensitive, reliable and fairly inexpensive way of measuring DNA damage. But the most pleasing feature is the scope of its applications. It has made its name as a test for genotoxicity. However, it is also an invaluable tool for investigating fundamental
) Corresponding author. Tel.: q44 1224. 716-634; Fax: q44 1224. 716-687.

aspects of DNA damage and cellular responses to this damage. At the other end of the scale, it is being used successfully to monitor DNA damage in human populations. In this article, we will provide, mostly from our own work, examples of investigations that span this range from molecular to epidemiological studies. First, some thoughts about the behaviour of DNA in the comet assay. It should be obvious that DNA is not migrating as fragments, as in conventional electrophoresis, where the distance travelled is inversely related to fragment size. The alkaline comet assay resolves break frequencies up to a few thousand per

0027-5107r97r$17.00 Copyright q 1997 Elsevier Science B.V. All rights reserved. PII S 0 0 2 7 - 5 1 0 7 9 7 . 0 0 0 1 3 - 4

184

A.R. Collins et al.r Mutation Research 375 (1997) 183193

cell, so the distances between breaks are of the order of 10 9 Da, definitely well beyond the range of fragment size for which conventional electrophoresis is suitable. The length of such a fragment is about 1 mm; the length of the tail of a comet is a few hundredths of this. So, are we perhaps seeing just the free, broken ends of DNA being pulled out into the tail? An analogy would be a tangled ball of string where a few judicious cuts can make pulling the ball apart much easier. But nuclear DNA is not a tangle of string. Even after treatment with detergent and strong salt solution, as in the standard SCGE procedure, the nucleus or nucleoid. has a structure; the DNA is organised as loops, which retain the supercoils that were formerly contained in the nucleosomes. The presence of supercoiled loops was deduced by Cook et al. w1x, and they observed that, when DNA was broken by irradiation, supercoiling was relaxed and loops spilled out into a halo around the nucleoid core. By analogy, we can assume that the comet tail is made up of relaxed loops, and that the number of loops in the tail or, simply, the relative tail intensity. indicates the number of DNA breaks. This model certainly fits with the observation that, with increasing amount of damage, the tail intensity rather than length increases, and tail length is presumably determined primarily by the length of the loops. Whether the tail consists of relaxed DNA loops or of free ends could perhaps be decided by looking at the distribution of ends using an end-labelling technique to incorporate tagged nucleotides with subsequent immunofluorescent detection, since differing patterns would be predicted by the two models. Meanwhile, some evidence is presented in this paper which supports the loop model. The alkaline comet assay detects single and double. DNA strand breaks. But single-strand breaks SSBs. are not the most interesting of lesions. They are quickly repaired, and are not regarded as a significant lethal or mutagenic lesion. Many genotoxic agents do not induce strand breaks directly. They may create AP sites, which are alkali-labile and are probably converted to breaks while DNA is in the electrophoresis solution at high pH. Furthermore, breaks will be transiently present when cells repair lesions via base excision or nucleotide excision, and so a high level of breaks in the comet assay may

indicate either high damage or efficient repair. In fact, much useful information can be obtained by exploiting cellular repair to produce DNA breaks and thus to reveal or amplify the effects of genotoxins that otherwise would not show positive effects in the comet assay, as we will describe. We have pioneered a modification of the comet assay which enormously increases its range, sensitivity and specificity. The DNA in the gel, following lysis, is digested with a lesion-specific repair endonuclease, which introduces breaks at sites of damage. In principle, any lesion for which a repair endonuclease exists can be detected in this way. We first showed this effect with endonuclease III, whose substrate is oxidised pyrimidines. Analogous trials with formamidopyrimidine glycosylase FPG, active on 8-OH-gua. and uvrABC recognising bulky lesions. have recently been carried out. The ability to detect specific lesions opens up a new area of investigation cellular DNA repair. It is generally true that the only unambiguous and quantitative indicator of cellular repair is the removal of damage from DNA. We have compared the kinetics of repair of oxidative damage in normal human lymphocytes and transformed epithelial HeLa. cells. Repair processes can be studied in subcellular systems, particularly useful in fractionation and reconstitution studies, with the aim of purifying repair proteins, and in characterising repair-defective mutant cells. Here again, the comet assay provides a simple assay system with the advantage of allowing examination of the sequential steps of incisionrexcision, and resynthesisrligation. Finally, at the other end of the scale, we have applied the modified comet assay, making use of lesion-specific enzymes, to the monitoring of human populations for oxidative DNA damage. The comet assay has proved to be a valuable tool in molecular epidemiology.

2. Materials and methods 2.1. Cells: culture and treatment with DNA-damaging agents HeLa transformed human epithelial. cells were routinely cultured as monolayers in Glasgow-mod-

A.R. Collins et al.r Mutation Research 375 (1997) 183193

185

ified Eagles Minimal Essential Medium ICN Flow. with 5% foetal calf serum, 5% newborn calf serum, penicillin and streptomycin, at 378C in a 5% CO 2 atmosphere. Oxidative damage was introduced by replacing the medium with phosphate-buffered saline PBS. containing H 2 O 2 at the required concentration; treatment was for 5 min on ice. Cells were treated with methylmethane sulphonate MMS. in medium for 30 min at 378C. UV irradiation of HeLa cells, after removing medium, was carried out with a germicidal lamp emitting predominantly at 254 nm, at a dose rate of 0.1 Jmy2 sy1 . After treatment with UV or H 2 O 2 , cells were either processed at once for SCGE or incubated in medium to allow repair. Lymphocytes were isolated from 30 ml aliquots of human blood finger-prick samples., mixed with 1 ml of RPMI 1640 medium ICN Flow. with 10% foetal calf serum, cooled on ice for 30 min, and centrifuged at 200 = g for 3 min at 48C over a layer of 100 ml of Histopaque 1077 Sigma.. Cells from just above the Histopaque layer were mixed with 1 ml of PBS and treated with H 2 O 2 or with UV in suspension, but otherwise as described for HeLa cells. 2.2. Single cell gel electrophoresis The comet assay, based on the method of Singh et al. w2x, was generally carried out as described w3,4x, with the inclusion of a digestion with enzyme endonuclease III, FPG or uvrABC. after lysis of cells in the gel. The enzyme was added to the gel in 50 ml of appropriate buffer see below., covered with a coverslip, and incubated at 378C for times indicated in figure legends. After electrophoresis, neutralisation and staining with DAPI, comets were analysed by fluorescence microscopy with visual inspection, giving each comet a value of 04 according to the degree of damage w5x. The overall score for each slide 100 comets analysed. was therefore between 0 and 400. Although these units are arbitrary, they can be related to the relative tail intensity which itself is a function of break frequency w6x. The relationship is discussed elsewhere w5x; it deviates from linearity, but a score of 300, for instance, corresponds to about 50% of DNA in the tail or 1.6 breaks per 10 9 Da of DNA see also w7x.. In some experiments, images

were analysed by computer using Komet 3.0 software Kinetic Imaging Ltd.. The % of DNA in the tail and tail moment were recorded. 2.3. Neutral electrophoresis Exceptionally Fig. 1., we used a neutral version of SCGE. Following lysis, slides were placed in either Trisacetate TA; 40 mM Trisacetic acid, 1 mM EDTA, pH 8.0. or Trisborate TB; 90 mM Trisboric acid, 2 mM EDTA, pH 8.0. buffer, left for 40 min at 48C and then electrophoresed for 30 min. 2.4. Acridine orange staining In one experiment, comets were stained with 0.02 mgrml of acridine orange in 10 mM sodium phosphate at pH 7.4, and rinsed 3 times in this buffer. 2.5. Lesion-specific enzymes Endonuclease III was isolated from an overproducing bacterial strain kindly provided by Dr. R. Cunningham, Department of Biological Sciences, State University of New York, NY, USA. Reactions were carried out in 0.1 M KCl, 0.5 mM EDTA, 40 mM HEPESKOH, 0.2 mgrml bovine serum albumin, pH 8.0. FPG was provided by Dr. S. Boiteux, Institut GustaveRoussy, Villejuif, France, and by Dr. G.P. Margison, Paterson Institute for Cancer Research, Manchester, UK. FPG reactions were carried out in 0.1 M KCl, 2 mM Na 2 EDTA, 5% glycerol, 50 mM HEPESKOH, pH 7.6. In recent experiments, we have employed the same buffer that described above for endonuclease III for reactions with both enzymes.. UvrA, uvrB and uvrC were obtained from Dr. P. van de Putte, Department of Molecular Genetics, Leiden, The Netherlands. The uvrABC reaction was carried out in the buffer used for FPG with the addition of 6.7 mM MgCl 2 and 1 mM ATP, pH 7.6. The uvrA, uvrB and uvrC subunits were each present at 20 nM in trials, double this concentration gave identical results..

186

A.R. Collins et al.r Mutation Research 375 (1997) 183193

Fig. 1. Comparisons of neutral and alkaline SCGE. a: visual analysis of comets from lymphocytes treated with 30 mM H 2 O 2 , compared with untreated controls, in neutral TA or TB buffer or in alkaline solution. One hundred comets per slide were sorted visually into classes 04 representing increasing amounts of damage. b: image analysis of comets from lymphocytes treated with H 2 O 2 . Top four panels: neutral TA. and alkaline electrophoresis, 0 and 30 mM H 2 O 2 . Results from 50 comets are shown in terms of tail moment. Lower two panels: dose response of cells to H 2 O 2 , assessed in TA or alkaline buffer. c: image analysis of comets from HeLa cells treated with MMS. Top four panels: neutral TB. and alkaline electrophoresis, 0 and 0.25 mM MMS. Lower two panels: dose response of cells to MMS, assessed in TB or alkaline buffer, expressed as tail moment.

A.R. Collins et al.r Mutation Research 375 (1997) 183193

187

3. Results and discussion 3.1. DNA loops and alkaline unwinding The comet assay, as now generally practised, includes incubation of DNA at high pH before and during electrophoresis, although the original work of Ostling and Johanson w8x employed near-neutral pH. The idea has developed that SCGE is related to methods for DNA break analysis, such as alkaline elution and alkaline unwinding. In these procedures, the separation of DNA strands at high pH is facilitated by the presence of strand breaks, and the extent or rate of separation is the index of DNA breakage. It has been suggested w2x that the detection of single-strand breaks in the comet assay, similarly, requires alkaline treatment. In theory, relaxation of supercoiled DNA loops by SSBs will occur at neutral as well as at alkaline pH. In practice, Ostling and Johanson w8x, with their neutral procedure, achieved a level of sensitivity to SSBs very similar to that seen with alkaline SCGE see w3x for comparison.. It is unnecessary and, we believe, unhelpful to consider SCGE as an alkaline unwinding method. As there has not, to our knowledge, been a direct investigation of neutral and alkaline electrophoresis performed in parallel, we have now examined the sensitivity of two variants of the assay in detecting DNA breaks induced by oxidative or alkylation damage. Fig. 1 shows examples from various experiments, with comets analysed in different ways: by visual scoring; or by image analysis expressing results in terms of tail moment essentially the product of tail length and relative DNA content.. The neutral and alkaline methods can both detect low levels of DNA damage. The images seen after neutral and alkaline electrophoresis are different Fig. 2; compare with Fig. 3 or Fig. 5.; the tail is more clearly resolved from the head under alkaline conditions, and this may account for the apparent failure of neutral comets to reveal effects at higher doses of damage Fig. 1.. The maximum amount of DNA released into the tail is about 25% in the neutral system, whereas it can reach 90% or more under alkaline conditions. It is clear, however, that in terms of sensitivity, i.e., the detection of low levels of strand breakage, neutral and alkaline versions of

the assay are similar not what would be expected if alkaline unwinding were an essential step in revealing lesions. Alkaline electrophoresis is followed by neutralisation. While the DNA in the tail of the comet remains single-stranded, the head DNA is able to reanneal and appears as double-stranded when stained with acridine orange Fig. 3.. This reannealing is to be expected if the head DNA consists of intact supercoiled loops; the strands of relaxed and broken tail loops are, in contrast, free to separate in alkali so that reassociation is unlikely to occur. 3.2. Manipulating cellular repair UV-C irradiation produces photolesions cyclobutane pyrimidine dimers and 6-4. photoproducts in DNA, but no detectable strand breaks. However, if human lymphocytes are incubated after UV irradiation, DNA breaks are seen Fig. 4. in the comet assay w5,9x as in other assays w10x. These breaks represent transient repair intermediates. After UV endonuclease has carried out the initial incision step, a finite time elapses until repair synthesis and ligation are completed and the gap is sealed. Depending on cell type, the gaps may be long- or short-lived. In the case of lymphocytes, the rate-limiting step of excision repair appears to be synthesis of the repair patch probably because the lymphocyte, as a non-dividing cell, is very poorly supplied with DNA precursors w11x. Fig. 4 clearly shows that, if DNA precursors are supplied exogenously, in the form of deoxyribonucleosides, the synthetic stage of repair is encouraged to the extent that no DNA breaks are seen above background. There is also no additional effect of adding DNA synthesis inhibitors. In contrast, no DNA breaks are detectable by the comet assay in HeLa cells unless aphidicolin, or other repair synthesis inhibitors, are added Fig. 5. reflecting the actively proliferating status of the HeLa cell. The principle of accumulating incomplete repair sites i.e., DNA breaks. has been widely applied to amplify the DNA-damaging effects of a variety of genotoxins, with detection of DNA breaks by alkaline unwinding or alkaline elution techniques e.g., w12x.. It can be used to similar effect with the comet assay.

188

A.R. Collins et al.r Mutation Research 375 (1997) 183193

A.R. Collins et al.r Mutation Research 375 (1997) 183193

189

3.3. Applying exogenous repair enzymes Rather than making use of the cells own repair enzymes to reveal damage, we can achieve greater specificity and higher sensitivity by treating the DNA with purified repair enzymes which will convert particular lesions into breaks. The first application of this idea to the comet assay w3x used the UV dimerspecific endonuclease V. Cells were permeabilised with saponin, incubated with this enzyme, and then processed for SCGE. UV doses as low as 0.02 Jmy2 were detectable, and it was estimated, by carrying out a parallel assay based on alkaline unwinding and hydroxyapatite analysis, that 30% of the dimers were recognised by the enzyme. Permeable cells are not ideal; the chromatin is still intact, and a fraction of the DNA remains inaccessible to enzymes. A far better substrate is the DNA of nucleoids, after detergent and high salt treatment of cells, since the DNA is more or less free of other macromolecular associations. Therefore we investigated the possibility of introducing an enzyme digestion step into the comet assay protocol. With an interest in quantitating low levels of oxidative DNA damage, we tested endonuclease III, which recognises oxidised pyrimidines w13x. In initial experiments, we incubated H 2 O 2-treated HeLa cells for 1 h to allow them to rejoin the SSBs that are the predominant lesion caused by this agent. The cells were then embedded in agarose on slides, lysed, washed with appropriate buffer, and incubated with endonuclease III before electrophoresis. The enzyme did, indeed, reveal additional damage, the number of extra breaks being related to the dose of H 2 O 2 Fig. 6.. Endonuclease III is capable of detecting oxidised

pyrimidines in human lymphocytes which have not been treated with any genotoxic agent see below.. The occurrence of endogenous DNA oxidation, via reactive oxygen species released during normal metabolism, has been suggested w14x as a possible cause of cancer. A more common biomarker of oxidative DNA damage is 8-OH-dG, often measured by HPLC. The enzyme FPG, in spite of its name, probably has 8-OH-gua as its main substrate in living cells w15x. Like endonuclease III, it combines the actions of glycosylase and AP endonuclease on a single macromolecule. We carried out experiments with HeLa cells treated with various doses of H 2 O 2 and embedded in agarose either immediately or after incubation to allow them to rejoin SSBs. Fig. 7a shows the relationship of SSBs to H 2 O 2 concentration, and the further breaks that appear on incubation in gel. with FPG. Because of the preponderance of SSBs, the estimation of extra enzyme sites is only approximate. However, when cells are incubated after H 2 O 2 treatment, strand breaks return to background level, and oxidised purines are then easily detected Fig. 7b.. An important observation from this figure is that, for different H 2 O 2 concentrations, the yield of FPG-sensitive sites Fig. 7b. is similar to the yield of SSBs Fig. 7a.. It is known that H 2 O 2 induces base oxidation and SSBs at comparable frequencies w16x, and so we are confident that the enzyme is working efficiently under these conditions. Quantitative measurement of 8-OH-dG in DNA is thus possible. We used the uvrABC enzyme complex to detect UV damage. To maintain low background levels of damage in untreated cells, it was necessary to wash the agarose-embedded nucleoids after lysis. 3 times

Fig. 2. Typical comets seen after neutral SCGE of untreated lymphocytes a. and lymphocytes treated with 30 mM H 2 O 2 b.. Fig. 3. Comets after alkaline electrophoresis. stained with acridine orange. Human lymphocytes were treated with 100 mM H 2 O 2 , inducing varying degrees of damage. Relatively undamaged cells give comets consisting of an orange head, indicating double-stranded DNA. Moderately damaged comets have a distinct orange head with a red single-stranded. tail. The arrow indicates a severely damaged comet with almost all DNA in the tail, stained red. Fig. 4. DNA breaks in human lymphocytes, unirradiated, or UV-irradiated 4 Jmy2 . and incubated for 1 h at 378C with dThd, dCyd, dAdo and dGuo, 10y4 M each; or with aphidicolin 15 mM.; or with no addition. From w5x with permission. Fig. 5. HeLa cells irradiated with UV 1 Jmy2 . and incubated at 378C for 30 min a., for 5 min with 15 mM aphidicolin b. or for 30 min with 15 mM aphidicolin c. before preparation for SCGE.

190

A.R. Collins et al.r Mutation Research 375 (1997) 183193

with reaction buffer made up without ATP and MgCl 2 . Lymphocytes and HeLa cells, irradiated with different UV doses, were embedded, lysed and incubated with uvrABC Fig. 8.. Occurrence of breaks is

Fig. 7. Oxidative damage to HeLa cell DNA; H 2 O 2 dose response. Cells were treated with H 2 O 2 at various concentrations, and assayed for DNA damage either immediately a. or after a 2-h incubation b.. DNA strand breaks were detected by the standard comet assay v .. To reveal oxidised purines, DNA was digested in the gel. with FPG for 15 min `., 30 min I. or 45 min ^.. From w20x with permission.

clearly UV-dependent, but as UV dose increases, the index of damage does not rise beyond a certain level of breaks, suggesting that another factor may be limiting the activity of the enzyme under these conditions. 3.4. Monitoring the remoal of DNA damage An alternative approach to studying cellular DNA repair is to measure the rate and extent of removal of damage. The availability of lesion-specific endonucleases means that it is possible to follow, individually, the removal of different kinds of damage. In the case of oxidative damage, SSBs and oxidised bases are repaired with different kinetics. Fig. 9 illustrates the removal of SSBs and 8-OH-gua in HeLa cells and lymphocytes on incubation after H 2 O 2 treat-

Fig. 6. Strand breaks and endonuclease III-sensitive sites in H 2 O 2 -treated HeLa cells. A: cells treated with 100 mM H 2 O 2 with or without incubation for repair. B: cells incubated for 1 h after treatment with doses of H 2 O 2 as shown and incubated in gel. with buffer or endonuclease III. From w4x with permission.

A.R. Collins et al.r Mutation Research 375 (1997) 183193

191

of a cell extract, or a purified protein, for reconstituting repair from its components, and for characterising repair-defective mutant cell lines. The principle of the assay is that cells containing appropriate DNA damage, lysed after embedding in agarose, provide the substrate for repair enzymes or cell extract added onto the gel. We prepared a DNA-free cell extract from a repair-competent lymphoblastoid cell line and incubated it with UV-irradiated HeLa nucleoids w17x. In the presence of ATP, incision occurred, resulting in UV dose-related DNA breaks Fig. 10.. If, after allowing accumulation of such breaks i.e., incomplete repair sites., more extract was added and incubated together with ATP and deoxyribonucleoside triphosphates, the comets showed a marked reduction in breaks, indicating that repair synthesis and ligation had occurred in vitro. Thus, the two stages of repair degradation and resynthesis can be examined separately or in sequence. Because of its speed and simplicity, this test is ideal for testing repair activi-

Fig. 8. UV dose responses for DNA damage in a. HeLa cells and b. lymphocytes. Cells were irradiated with various doses of UV and then assayed for DNA damage by the comet assay, including incubation with uvrA, uvrB and uvrC combined. for 45 min I., 60 min ^., or with no incubation v .. From w20x with permission.

ment. At time zero, SSBs predominate, saturating the assay, so that extra FPG-sensitive sites are not detectable. After 1 h, when most SSBs have been repaired, FPG sites are measurable. Evidently, FPGsensitive lesions are more persistent than SSBs, and there is no sign of their removal over 4 h of incubation in either HeLa cells or lymphocytes. We found a similar pattern of repair in cells treated with endonuclease III to detect oxidised pyrimidines w5x. 3.5. In itro DNA repair While several proteins that take part in mammalian nucleotide excision repair have now been characterised, undoubtedly others remain to be identified. The enzymes and other factors involved in base excision repair of alkylation and oxidation damage are less well understood. The comet assay provides a useful system for assessing the repair activity

Fig. 9. Time course of repair of oxidative damage; a. HeLa cells, b. lymphocytes. Cells were treated with 100 mM H 2 O 2 and incubated at 378C. At intervals, samples were assayed for DNA damage; SSBs `., and SSBs plus FPG-sensitive sites v .. Also shown are untreated controls, without ,. or with %. FPG digestion.

192

A.R. Collins et al.r Mutation Research 375 (1997) 183193

ties in fractions of cell extract during purification. For instance, fractions can be tested for their ability to complement an extract deficient in a particular step in repair. 3.6. Molecular epidemiology The comet assay has been used in various pilot studies to test its potential as an indicator of oxidative stress in human subjects. For example, the effect of strenuous exercise in inducing DNA breaks in lymphocytes has been investigated w18,19x. We recently undertook an exercise in molecular epidemiology to test the antioxidant hypothesis, i.e., the idea that dietary antioxidants can protect against cancer by limiting free radical attack on DNA. One hundred men, aged 5059 years, half smokers and half nonsmokers, received a daily supplement of vitamin C, vitamin E and b-carotene, or a placebo, for 20 weeks. Blood samples were taken at intervals, and lymphocytes were isolated and run in the comet assay with and without an endonuclease III digestion. This enzyme clearly revealed extra sites in both smokers and non-smokers. After 20 weeks, there was a striking decrease in oxidised pyrimidines in those who had received supplements; smokers and nonsmokers did not differ significantly in the amount of damage Fig. 11..

Fig. 11. Effect of dietary antioxidant supplementation on levels of oxidised pyrimidines in lymphocytes of smokers and non-smokers. Endogenous damage was measured after 20 weeks of supplementation with 100 mg vitamin C, 280 mg vitamin E and 25 mg b-carotene per day or placebo..

3.7. Conclusions For measuring DNA strand breaks, the comet assay is already recognised as being among the most sensitive methods available; it has further advantages of speed, simplicity and the fact that observations are made at the level of single cells. The extensions of the method described here making use of repair enzymes, both endogenous and exogenous add greatly to the assays potential applications. It is important to remember, however, that the nature of comets, and the physicochemical events underlying their formation, are not fully understood or agreed upon. This is an unsatisfactory situation, and it is to be hoped that, while continuing to use the assay in routine genotoxicity testing, its devotees will find time to look more closely at fundamental aspects of the comet assay.

Acknowledgements
Fig. 10. Repair of UV-damaged DNA by human cell extract. HeLa cells were irradiated with different doses of UV before embedding in agarose and lysing. The DNA was then incubated with buffer ^., or with buffer plus cell extract derived from GM1899A lymphoblastoid cells `. for 45 min, or with buffer plus cell extract for 45 min and then with buffer, extract and deoxyribonucleoside triphosphates 5 mM each. for a further 45 min I.. ATP was present in all cases. From w17x with permission.

We thank Dr. Serge Boiteux and Dr. Geoff Margison for gifts of FPG, and Dr. Richard Cunningham for the endonuclease III over-producing cell strain. The UvrABC proteins were obtained from Dr. P. van de Putte; purification of these proteins was done with the support of the EC Concerted Action on DNA Repair and Cancer. We are grateful to Dr. Susan Duthie, Catherine Gedik, Iona Fleming and Ma Ai-

A.R. Collins et al.r Mutation Research 375 (1997) 183193

193

guo for their substantial contributions to the work described here. M.D. was in receipt of an ICRETT Fellowship from the UICC. Work done by R.S. was funded by a Royal Society Collaborative Grant. We acknowledge the support of the Scottish Office Agriculture and Fisheries Department and EC contract CIPA-CT94-0129.

w10x

w11x

References
w1x Cook, P.R., I.A. Brazell and E. Jost 1976. Characterization of nuclear structures containing superhelical DNA, J. Cell Sci., 22, 303324. w2x Singh, N.P., M.T. McCoy, R.R. Tice and E.L. Schneider 1988. A simple technique for quantitation of low levels of DNA damage in individual cells, Exp. Cell Res., 175, 184 191. w3x Gedik, C.M., S.W.B. Ewen and A.R. Collins 1992. Singlecell gel electrophoresis applied to the analysis of UV-C damage and its repair in human cells, Int. J. Radiat. Biol., 62, 313320. w4x Collins, A.R., S.J. Duthie and V.L. Dobson 1993. Direct enzymic detection of endogenous oxidative base damage in human lymphocyte DNA, Carcinogenesis, 14, 17331735. w5x Collins, A.R., A.-G. Ma and S.J. Duthie 1995. The kinetics of repair of oxidative DNA damage strand breaks and oxidised pyrimidines. in human cells, Mutation Res. DNA Repair., 336, 6977. w6x Olive, P.L., J.P. Banath and R.E. Durand 1990. Heterogene ity in radiation-induced DNA damage and repair in tumor and normal cells measured using the comet assay, Radiat. Res., 122, 8694. w7x Collins, A.R., M. Dusinska, C.M. Gedik and R. Stetina 1996. Oxidative damage to DNA: do we have a reliable biomarker? Env. Health Persp., 104 Suppl. 3., 465469. w8x Ostling, O. and K.J. Johanson 1984. Microelectrophoretic study of radiation-induced DNA damages in individual mammalian cells, Biochem. Biophys. Res. Commun., 123, 291 298. w9x Green, M.H.L., A.P.W. Waugh, J.E. Lowe, S.A. Harcourt, J. Cole and C.F. Arlett 1994. Effect of deoxyribonucleosides

w12x

w13x

w14x w15x

w16x w17x

w18x

w19x

w20x

on the hypersensitivity of human peripheral blood lymphocytes to UV-B and UV-C irradiation, Mutation Res., 315, 2532. Yew, F.F.-H. and R.T. Johnson 1979. Ultraviolet-induced DNA excision repair in human B and T lymphocytes. II. Effect of inhibitors and DNA precursors, Biochim. Biophys. Acta, 562, 240251. Munch-Petersen, B., G. Tyrsted and B. Dupont 1973. The X deoxyribonucleoside 5 -triphosphate dATP and dTTP. pool in phytohemagglutinin-stimulated and non-stimulated human lymphocytes, Exp. Cell Res., 79, 249256. Collins, A.R.S., M.J. Ord and R.T. Johnson 1981. Correlations of DNA damage and repair with nuclear and chromosomal damage in HeLa cells caused by methylnitrosamides, Cancer Res., 41, 51765187. Doetsch, P.W., W.D. Henner, R.P. Cunningham, J.H. Toney and D.E. Helland 1987. A highly conserved endonuclease activity present in Escherichia coli, bovine, and human cells recognizes oxidative DNA damage at sites of pyrimidines, Mol. Cell. Biol., 7, 2632. Ames, B.N. 1989. Endogenous oxidative DNA damage, aging and cancer, Free Rad. Res. Commun., 7, 121128. Boiteux, S., E. Gajewski, J. Laval and M. Dizdaroglu 1992. Substrate specificity of the Escherichia coli Fpg protein formamidopyrimidine-DNA glycosylase.: excision of purine lesions in DNA produced by ionizing radiation or photosensitization, Biochemistry, 31, 106110. Cerutti, P.A. 1974. Effects of ionizing radiation on mammalian cells, Naturwissenschaften, 61, 5159. Collins, A.R., I.M. Fleming and C.M. Gedik 1994. In vitro repair of oxidative and ultraviolet-induced DNA damage in supercoiled nucleoid DNA by human cell extract, Biochim. Biophys. Acta, 1219, 724727. Tice, R.R., P.W. Andrews and N.P. Singh 1990. The single cell gel assay: a sensitive technique for evaluating intercellular differences in DNA damage and repair, Basic Life Sci., 53, 291301. Hartmann, A., U. Plappert, K. Raddatz, M. Grunert-Fuchs and G. Speit 1994. Does physical activity induce DNA damage?, Mutagenesis, 9, 269272. Dusinska, M. and A. Collins 1996. Detection of oxidised purines and UV-induced photoproducts in DNA of single cells, by inclusion of lesion-specific enzymes in the comet assay, Alt. Lab. Animals, 24, 405411.