Eur. J. Biochem. 169,299-305 (1987) @] FEBS 1987

Hydroxyurea : effects on deoxyribonucleotide pool sizes correlated with effects on DNA repair in mammalian cells Andrew COLLINS and David J. OATES Cancer Research Campaign Mammalian Cell DNA Repair Group, Department of Zoology, University of Cambridge

(Received April 3/July 15, 1987) - EJB 87 0395

We have measured deoxyribonucleotide pool sizes in different cell types: normal human, transformed human (HeLa), and the permanent hamster line CHO-K1. The range of sizes of the four DNA precursor pools in CHO cells is far greater than in human cells. It is a general rule that hydroxyurea causes rapid depletion of pools (except for dTTP) until the pool present in smallest amount is exhausted; this suggests a tight coupling of the pools to DNA replication (the presumed main cause of the depletion). The effect of hydroxyurea on DNA repair after ultraviolet irradiation (namely, a relatively small accumulation of incomplete repair sites blocked at the resynthesis stage) is probably accounted for by the reduced availability of DNA precursors. However, depletion of the dCTP pool is not an adequate explanation for the observed enhancement by hydroxyurea of the inhibitory effect of cytosine arabinoside; we suggest other possible modes of action. Ultraviolet irradiation has only small effects on the levels of deoxyribonucleotides.

The regulation of deoxyribonucleotide metabolism has ramifications far beyond the simple provision of precursors for DNA synthesis. Various inherited disorders, including the Lesch-Nyhan syndrome, gout, and certain immunodeficiency syndromes, have their basis in a genetic defect of purine or pyrimidine metabolism [l , 21. Many anticancer, antiviral, anti- psoriatic and immunosuppressive drugs act via nucleotide metabolism [3 , 41. One of these drugs, hydroxyurea, is widely used experimentally as an inhibitor of DNA synthesis; it is a potent inhibitor of ribonucleotide reductase, the enzyme responsible for de novo production of deoxyribonucleotides (dNTPs) by reduction of ribonucleotides at the level of diphosphates [ 51.

Hydroxyurea is also employed in investigations of DNA repair, not without controversy. By suppressing replicative DNA synthesis, hydroxyurea makes it possible to detect the very small amounts of DNA synthesis associated with the cellular repair of DNA damage. It has been widely believed that hydroxyurea has no effect on DNA repair, but it is clear that hydroxyurea in fact delays the resynthesis step (presum- ably by limiting substrate availability) and holds open the gaps introduced into DNA at damage sites by a repair endonuclease [6]. Hydroxyurea can be used in a different way in repair studies: to increase the effectiveness of another DNA synthesis inhibitor, cytosine arabinoside. It is generally assumed [6,7] that hydroxyurea acts by decreasing the concen- tration of dCTP, since aCTP (phosphorylated on entry to the cell) inhibits DNA polymerase a competitively with dCTP.

There are yet other possible influences of hydroxyurea on DNA repair. The accumulation of incomplete repair sites in

Correspondence to A. Collins, Department of Biochemistry, Uni- versity of Aberdeen, Marischal College, Aberdeen, Scotland AB9 1 AS

Abbreviations. aCTP, arabinocytosine 5’-triphosphate; aCyd, cytosine arabinoside; HEL, human embryonic lung; CHO, Chinese hamster ovary.

Enzymes. Ribonucleotide reductase (EC 1.17.4.1); DNA poly- merase a and /3 (EC 2.7.7.7).

the presence of hydroxyurea might be due to an inhibition of ligation rather than resynthesis. The resistance of resynthesis, as measured by [3H]dThd incorporation, to inhibition by hydroxyurea might be partly accounted for by an elevated specific activity of cellular [3H]dTTP if the endogenous dTTP pool is diminished by hydroxyurea [8].

It should be possible to substantiate or eliminate the various hypothesized influences of hydroxyurea on DNA re- pair by a comparative study, in different cell types, of dNTP pool sizes, effects of hydroxyurea on these pools, and effects of hydroxyurea on DNA repair after ultraviolet irradiation; the present study was undertaken with this aim. In addition, we have examined the effect of ultraviolet irradiation itself on dNTP pools.

MATERIALS AND METHODS

Cell culture

Normal fibroblasts from human embryonic lung (HEL, from Gibco) were grown in Eagle’s minimal essential medium buffered with bicarbonate, with non-essential amino acids and 10% foetal calf serum (Gibco). HeLd and CHO cells were grown in the same medium but with 3% foetal calf and 3% newborn calf serum (Gibco).

Measurement of deoxyribonucleotide pools

100-mm dishes (Nunc) were inoculated with equal numbers of cells and, as they approached confluence, the medium was changed to ensure a fully proliferative state when the experiment was performed one day later. One dish was sacrificed to count the number of cells. In some experiments inhibitors were added from 10 x stock solutions in phosphate- buffered saline (0.2 g KCI, 0.2 g KH2P04, 8 g NaCl, 1.15 g Na2HP04 per litre); in others, the medium was removed for ultraviolet irradiation with a germicidal lamp emitting

300

Incubation t h e (hours)

Fig. 1. Deoxyribonucleoside triphosphate pool sizes in log-phase normal human fibroblasts ( H E L cells) ; effect of addition of 10 mM hydroxyurea to culture medium. Bars represent the standard error of the mean, or the range of duplicate determinations

0 0.5 1.0 0 0.5 r0 0 0.5 1.0 0 0.5 1.0 Incubation time (hours)

Fig. 2. Deoxyribonucleoside triphosphate pool sizes in log-phase HeLa cells; effect of addition of 10 m M hydroxyurea to culture medium. Bars represent the standard error of the mean, or the range of duplicate determinations

at 254 nm and medium was then replaced. After the desired incubation period, medium was removed as completely and quickly as possible by aspiration, and the cells were frozen by placing the dish on an aluminium plate chilled by solid COz. 100 pl of ice-cold 12% HC104 was dispersed over the cell layer, which, while still frozen, was scraped from the dish and transferred to a 1.5-ml centrifuge tube. After 30 min on ice, tubes were centrifuged. The supernatant was neutralised with 4 M KOH/0.4 M KH2P04, left on ice for 10 min, centrifuged again, and the supernatant treated with NaI04 and methylamine to digest ribonucleotides [9]. The remaining deoxyribonucleotides were analysed (either at once or after storage at - 70°C) by HPLC on a Radial-PAK SAX cartridge (Waters) using a Waters Z-module with an LDC HPLC system. Elution was by a linear gradient over 35 min of 0.22 - 0.44 M ammonium phosphate, pH 5.5, at a flow rate of 2 ml/ min. The order of elution (detected by ultraviolet absorption) was: dTTP, dCTP, dATP, dGTP. Sometimes the dTTP peak

was superimposed on a trailing edge of unidentified ultraviolet-absorbing material which was always found to elute in the first few minutes of the run; the determination of dTTP was, as a result, less precise than that of the other nucleotides. Calibration was done against standard de- oxyribonucleotide solutions (Sigma).

D N A break accumulation

Cells in 35-mm dishes were prelabelled with [3H]dThd (0.1 ICi/ml for 2 days) and the medium was changed on the day of the experiment. Cells were preincubated for 30min with inhibitors (added from 10 x stock solutions), irradiated with ultraviolet light (or, for controls, mock-irradiated) and further incubated with inhibitors for 30 min. DNA breaks accumulating during this incubation were measured by the alkaline unwinding assay in conjunction with hydroxyapatite chromatography [lo]. In this procedure, the background level

301

Incubation time (hours)

Fig. 3. Deoxyribonucleoside triphosphate pool sizes in log-phase CHO cells; effect of addition of 10 mM hydroxyurea to culture medium. Bars represent the standard error of the mean, or the range of duplicate determinations

of breaks in unirradiated cells is subtracted from the breaks seen in irradiated samples. Background break frequencies were <0.2,0 and Q0.5 per 1 GDa for HEL, HeLa and CHO cells in these experiments.

RESULTS

Sizes of deoxyribonucleotide pools in cells incubated with hydroxyurea

Proliferating cultures of normal human embryonic lung fibroblasts (HEL), transformed human cells (HeLa), and Chinese hamster ovary cells (CHO-Kl), were incubated with 10 mM hydroxyurea. At intervals up to 1 h, sample cultures were quickly frozen and concentrations of the four DNA precursors were estimated by HPLC.

Data for HEL cells are shown in Fig. 1. At the start of the incubation, the largest pool is that of dTTP; dCTP and dATP are each about half the size of dTTP, and dGTP is present at about one quarter the level of dTTP. Within 10 min of adding hydroxyurea dGTP becomes undetectable. Levels of dCTP and dATP fall too, but measurable amounts remain even after 1 h. The dTTP pool seems to behave erratically (though this may be an artefact, see Materials and Methods) but in comparison with the other nucleotides, the dTTP pool is left relatively intact by hydroxyurea treatment.

This is true of the dTTP pool in HeLa cells, too (Fig. 2). In this case, the starting levels of dCTP, dATP and dGTP are roughly the same (half that of dTTP), and all three fall to low levels, around lo%, within 0.5 h. (The apparent rise in dATP and dGTP at 1 h is of dubious significance, representing a single sample).

Fig. 3 depicts CHO cells, similarly treated. There is a very marked difference in the starting levels of the pools in these cells compared with HEL or HeLa; dGTP is present at only one quarter of the level of dATP or dTTP, and dCTP is more than 10 times as abundant as dATP or dTTP. In the presence of hydroxyurea, dGTP falls to an undetectable level, dATP is partially depleted, dCTP shows little change and dTTP, if anything, increases.

Effects of hydroxyurea and cytosine arabinoside on D N A repair after ultraviolet irradiation

The same cell types (HEL, HeLa and CHO-K1) were examined for their response to ultraviolet irradiation in the

1 3 10 Hydroxywea concentration (mM)

01 I I 1

10-6 10-5 10-4 Cytosine arabinoside conc. (MI

Fig. 4. DNA breaks accumulated by normal .fibroblasts ( H E L cells) incubated with inhibitors after ultraviolet irradiation. Inhibitor: (A) hydroxyurea; (B) cytosine arabinoside alone (-) and plus 10 mM hydroxyurea (----). Ultraviolet dose: (0,O) 3 J m - or (A, A) 10 J rn-’

presence of hydroxyurea and/or cytosine arabinoside (aCyd). These DNA synthesis inhibitors tend to prevent the comple- tion of repair at sites of DNA damage, so that the breaks introduced by the repair process, normally transient, accumulate to readily measurable numbers.

Figs 4-6 show the accumulation of DNA breaks after ultraviolet irradiation (3 or 10 J m-2). In all cases (and in

302

I

1 3 10 Hydroxyurea concentration (mM)

I

Fig. 5. D N A breaks accumulated by HeLa cells incubated with inhibi- tors after ultraviolet irradiation. Details as for Fig. 4

similar experiments not depicted here), more breaks are detected after incubation with both inhibitors than with either alone. The low levels of DNA breaks seen on incubation after ultraviolet irradiation in the absence of any inhibitor have been documented elsewhere [ l l , 121. They are comparable with the frequencies seen in the present experiments at the lowest concentrations of either hydroxyurea or aCyd (< 1 break/l GDa at 10 J m-2). We have previously shown [lo] that 0.1 mM aCyd with 10 mM hydroxyurea gives maximal inhibition of excision repair in various human cell lines. Given equal ultraviolet doses, HeLa and HEL cells accumulate more DNA breaks under these conditions than do CHO cells (Figs 4 - 6).

Eflecls of ultraviolet irradiation on deoxyribonucleotide pools

Table 1 gives the concentrations of DNA precursor pools in unirradiated HEL, HeLa and CHO-K1 cells and in parallel cultures irradiated with ultraviolet light (10 J m-2) and in- cubated for 2 h. (In preliminary experiments, in which cells were irradiated with 2 J m-2 or 10 J rnp2 and incubated for 30 min or 2 h, it was found that significant changes occurred only at 2 h after the higher ultraviolet dose, so later experi- ments concentrated on these conditions). Levels of dTTP, dCTP and dATP tend to increase after irradiation, and dGTP tends to fall, but the changes are not substantial.

:p O 1 3 10

._ Hydroxyurea concentration (mM)

I

I I

10-6 10-5 10-4 Cytosine arabinoside conc. (M)

Fig. 6. D N A breaks accumulatedby CHO cells incubated with inhibitors after ultraviolet irradiation. Details as for Fig. 4

DISCUSSION

Cell-type differences in D N A precursor pools

The literature dealing with DNA precursor pool measure- ments is notable for the wide variations in estimates for the same cell type obtained in different laboratories. Table 2 summarises data for CHO cells. The discrepancies are not explained by differences in methods (for example, Leeds et al. [13] used the same enzymic technique as Skoog et al. [14]). The relative amounts of the different dNTPs are consistent, however, with roughly equal pools of dTTP and dATP, a very large pool of dCTP and a very small pool of dGTP.

In the case of HeLa cells, the data of Bray and Brent [15] for cells in early G1 (dTTP, 17 pmol; dCTP, 9 pmol; dATP, 5 pmol; dGTP, 6 pmol/106 cells) are similar to ours (see Fig. 2); Leeds et al. [13] again report higher levels. Pools of the four DNA precursors are more uniform in HeLa cells than they are in CHO cells, with dTTP present in rather larger quantity than the others.

Snyder [16] reported DNA precursor concentrations in proliferating human fibroblasts of 47, 12, 24, and 5 pmol/106 cells for dTTP, dCTP, dATP and dGTP, quite close to our figures. Normal (untransformed) human cells thus resemble HeLa cells in the proportions of the four dNTPs, except that dGTP is least abundant.

Eflects of hydroxyurea on D N A precursor pools

The different distributions of DNA precursor pools in these three cell types provide a useful framework for in-

303

Table 1. dNTP pool size in different cell types; effects of ultraviolet irradiation All cultures were in a proliferating state. Estimates were made by HPLC on samples 2 h after irradiation/mock irradiation at 10 J m-'. Mean values are shown with standard errors where data from three or more experiments are available. In square brackets are the ratios of mean dNTP concentrations with/without irradiation

Cell type Irradiation dTTP dCTP dATP dGTP

HEL - 31 f 3.4 30 f 7.8 15 f 0.9 10 k 2 . 2 (normal human) + 33 3 5 f 9 21 f 3.9 7.3 2.4

11 f 0.8 12 k 2 . 1 (transformed + 24f 1.2 17 f 4.2 18 f 4.4 7.3 f 1.4 human) [1.67] [1.55] [ 1.641 [0.58]

CHO - 20 f 6 271 f 35 32 f 10 4.2 f 1.4 (hamster) + 20 271 33 3.8

[1.06] [3.17] [1.40] [0.73] HeLa - 16 f 1.1 11 f 0 . 9

[1.00] [1.02] [1.03] [0.90]

Table 2. Sizes of dNTP pools in CHO cells; comparisons of different determinations

dTTP dCTP dATP dGTP Reference Cell state

58 223 39 12 [321 log phase 15" 180" 12" 1.2" ~141 peaks during cell cycle 4.2" 32" 4.2" 0.6" 1331 log phase

130 1100 95 15 ~ 3 1 log phase (?) 11 158 12 2.3 this paper log phase

a Data in original source were in terms of pmol/pg DNA, converted approximately to pmol/106 cells for this table using the factor 3 pg/ lo6 cells.

vestigating the effects of hydroxyurea. A detailed time course of changes in dNTP pool sizes in mouse embryo cells in- cubated with hydroxyurea was plotted by Skoog and Nordenskjold [17]. For CHO cells, however, there are data only for a 0.5-h incubation with hydroxyurea [18]. Snyder [16] documented the changes in dNTPs at 3 h after adding hydroxyurea to human fibroblasts. Here we follow changes in dNTP concentrations at frequent intervals over the first hour of incubation with hydroxyurea, examining three dif- ferent cell types.

In no case does the pool of dTTP show a convincing fall in the presence of hydroxyurea, and it may even rise. Similar behaviour was seen in mouse embryo cells [17], CHO cells [18], normal human cells [16] and mouse 3T6 cells [19]. Nicander and Reichard [I91 studied the turnover of the dTTP pool under inhibitory conditions: whereas the blocking of replication at the level of polymerisation (using aphidicolin) revealed a continuing degradation of dTTP, hydroxyurea pre- vented this turnover so that the dTTP pool was stabilised.

Nicander and Reichard [19, 201 report a similar effect of inhibitors on dCTP levels; with hydroxyurea, the dCTP concentration in 3T6 cells remains high, whereas rapid turn- over continues in cells incubated with aphidicolin. There is, they suggest, a regulatory mechanism that maintains the pyrimidine deoxyribonucleotide pools at a fairly constant level when de novo synthesis is not occurring.

The relative stability of the dCTP pool in the presence of hydroxyurea may be a feature of rodent cells, but it is not the case for human cells in which, as we have shown, dCTP is rapidly depleted. Our results suggest that, apart from dTTP which seems to be regulated in a unique way, the behaviour

of the dNTP pools on addition of hydroxyurea follows a general rule : namely, that the nucleotide present in smallest amount disappears first. In the case of normal human and CHO cells, this is dGTP; in HeLa, dCTP, dATP and dGTP all fall to a low level within 0.5 h. Once synthesis of dNTPs is blocked with hydroxyurea, DNA replication presumably continues until the limiting nucleotide(s) is depleted. The other pools then more slowly undergo further depletion (seen most clearly in Fig. l), implying that DNA replication is the main drain on them, but that there is also some secondary metabolic process leading to dNTP turnover. Thus, even in non-S-phase cells, hydroxyurea would be expected to deplete DNA pre- cursor pools, rather slowly, which explains the otherwise puzzling ability of hydroxyurea to inhibit DNA repair in such cells [8].

There is an attractive but unproven hypothesis [21] that DNA replication is accomplished by an enzyme complex that includes ribonucleotide reductase as well as DNA polymerase, etc, and that the deoxyribonucleotides are synthesised at the site of polymerisation. Thus dNTPs may exist at a high con- centration in a very small compartment of the cell [22]. There may also be a pool of dNTPs that have leaked out from the site of replication, or have been synthesised via salvage pathways; but the rapid depletion of one or more dNTPs, reported here, suggests the existence, at least for those dNTPs, of a single pool tightly coupled with DNA replication.

Hydroxyurea and the repair of ultraviolet-damaged D N A

How do these findings concerning deoxyribonucleotide pool sizes help with interpreting the effects of inhibitors of

304

DNA repair? There are fragmentary data in the literature on the ability of various inhibitors to cause incomplete repair sites to accumulate. Figs 4-6 present, for the first time, results of experiments done in parallel with different cell types and with different concentrations of hydroxyurea and cytosine arabinoside alone and together. Though protocols described elsewhere differ widely (in ultraviolet dose, inhibitor concen- trations, time of incubation, etc.), the present results are entirely consistent with those of Snyder et al. [23], Collins et al. [24] and Squires et al. [lo] using normal human cells; and with comparative data for different cell types incubated with a low concentration of cytosine arabinoside [25]. It is clear, for instance, that hydroxyurea alone has little effect in normal human cells (its effect on HeLa and CHO cells is more signifi- cant; see also [26]).

The potentiation by hydroxyurea of the cytosine- arabinoside-induced accumulation of DNA breaks after ultraviolet is well known [7, 11, 23, 271. It is generally considered that hydroxyurea acts by depleting the pool of dCTP, with which aCTP competes in inhibiting DNA polymerase CI, but until now data to substantiate this idea have been lacking; indeed it has been known for many years that the dCTP pool does not fall on addition of hydroxyurea to mouse embryo cells [5]. We have now demonstrated that hydroxyurea causes a severe and rapid fall in dCTP in HeLa cells and a much slower decline in HEL cells. HEL and HeLa cells show similar enhancement by hydroxyurea of the number of breaks accumulated with cytosine arabinoside present after irradiation (Figs 4, 5).

In CHO cells the pool of dCTP is especially large. We might therefore expect cytosine arabinoside to be less effective at inhibiting DNA repair in these cells compared with human cells. The lack of any sign of breaks reaching a plateau with increasing cytosine arabinoside concentration (plus hydroxy- urea) in Fig. 6, suggests that is so. The use of this method to quantify total repair sites in CHO cells may therefore result in underestimation. The failure of hydroxyurea to cause sig- nificant reduction in the dCTP pool in CHO cells (Fig. 3) is at odds with its ability to potentiate the cytosine-arabinoside- induced break accumulation after irradiation (Fig. 6). It is possible that the dCTP pool available to repair enzymes is only a small part of the total dCTP pool, so that its depletion by hydroxyurea could not be detected. Nicander and Reichard [28] have suggested that the dCTP pool (in mouse 3T6 fibroblasts) is in two functionally distinct parts; and much of the dCTP may be located in the cytoplasm [13].

There are other ways in which hydroxyurea and cytosine arabinoside may act synergistically or additively on DNA repair, which may apply also to HeLa and HEL cells. It is possible that direct inhibition of DNA polymerase CI by aCTP, together with indirect inhibition by hydroxyurea acting through substrate provision, retard repair synthesis in a pro- portion of repair sites that would otherwise progress to the point at which ligation is possible. Alternatively, hydroxyurea may be acting on certain sites that are insensitive to inhibition by cytosine arabinoside, perhaps a subset of sites at which repair synthesis is accomplished by DNA polymerase p (which obviously requires dNTPs but is resistant to aCTP).

Hunting and Dresler [29] recently investigated the role of hydroxyurea in inhibiting repair (in the absence of cytosine arabinoside), using permeabilised human fibroblasts. Hydroxyurea had no direct effect on repair DNA synthesis or on the ligation of repair patches after ultraviolet irradiation. Mimicking the effect of hydroxyurea on ribonucleotide re- ductase by reducing the concentrations of dNTPs in their

permeable cell assay did produce a decrease in repair synthesis and slower ligation. They conclude that hydroxyurea may act simply by reducing dNTP levels; but that a damage-induced reduction in dNTPs may contribute to its effectiveness. Newman and Miller [30] reported that the pool of dCTP in CHO cells falls by about 90% 1 h after ultraviolet irradiation. However, we (and R. D. Snyder, personal communication) have found no such marked decrease in any of the four dNTP pools in a wide range of cell types; Table 1 shows that in most cases the pools increase after irradiation, but the changes (in either direction) are barely significant.

It is notoriously difficult to show a reduction in repair DNA synthesis in the presence of hydroxyurea. The explana- tion in terms of an elevated specific activity of [3H]dThd leading to an overestimate of DNA synthesis from 3H incor- poration data is not supported by the relative constancy of the dTTP pool in all three cell types examined. However, there is evidence [8, 161 that hydroxyurea increases the uptake of [3H]dThd into the cellular pool of dTTP. In addition, resyn- thesis at a reduced rate but for an increased time (at repair sites held open by hydroxyurea) will accomplish substantial incorporation [6] (see also [31] for the analogous argument for aphidicolin inhibition).

Conclusion

By documenting in detail the changes in cellular dNTP pools with time after adding hydroxyurea we have tried to clarify the involvement of this simple drug in the highly complex metabolic system centred around DNA replication and repair. By studying several cell types, we have shown the imprudence of certain generalisations that have crept into the body of knowledge. One of these, the idea that hydroxyurea potentiates the inhibitory effect of cytosine arabinoside on DNA repair by decreasing the dCTP pool, is shown to be credible only in certain cases. Another generalisation some- times made (most recently in [19]) is that hydroxyurea causes a specific loss of purine dNTPs. This may be a reasonable statement for mouse cells, but a more general conclusion is that, while dTTP is not decreased, the sizes of the other dNTP pools fall in concert, until replication comes to a halt when the smallest pool (not necessarily a purine) is depleted.

The action of hydroxyurea on DNA repair is not wholly resolved; but its main features are seen to be generally consis- tent with what is now known about the effects of hydroxyurea on dNTP metabolism.

We are grateful to Dr R. T. Johnson and Prof. H. M. Keir for advice and support. The research was funded by the Cancer Research Campaign; AC holds the Campaign’s Michael Sobell Fellowship.

REFERENCES 1. Giblett, E. R., Ammann, A. J., Ward, D. W., Sandman, R. &

Diamond, L. K. (1915) Lancer i, 1010-1013. 2. Stanbury, J. B., Wyngaarden, J. B., Fredrickson, D. S., Goldstein,

J. L. & Brown, M. S. (1983) The metabolic basis of’ inherited disease, 5th edn, pp. 1043 - 1226, McGraw-Hill, New York.

3. Foster, R. W. & COX, B. (1980) Basic pharmacology, passim, Butterworths. London and Boston.

4. Nguyen, B. T., EISayed, Y . M. & Sadee, W. (1984) Cancer Res. 44.2212 - 2211.

5. Reichard, P. (1972) Adv. Enz. Regul. 10, 3- 16. 6. Collins, A. R. S. & Johnson, R. T. (1984) Adv. Radial. Biol. I / ,

71 - 129.

305

7. Dunn, W. C. & Regan, J. D. (1979) Mol. Pharmacol. 15, 367-

8. Collins, A. R. S., Schor, S. L. & Johnson, R. T. (1977) Mutat.

9. Garrett, C. & Santi, D. V. (1979) Anal. Biochem. 99, 268-273. 10. Squires, S., Johnson, R. T. & Collins, A. R. S. (1982) Mutat. Res.

11. Collins, A. R. S., Downes, C. S. &Johnson, R. T. (1980) J . Cell. Physiol. 103, 179- 191.

12. Johnson, R. T., Collins, A. R. S., Squires, S., Mullinger, A. M., Elliott, G. C., Downes, C. S. & Rasko, I. (1987) J . Cell Sci.

13. Leeds, J. M., Slabaugh, M. B. & Mathews, C. K . (1985) Mol.

14. Skoog, K . L., Nordenskjold, B. A. & Bjursell, K. G. (1973) Eur.

15. Bray, G. & Brent, T. P. (1972) Biochim. Biophys. Acta269, 184-

16. Snyder, R. D. (1984) Mutat. Res. 131, 163-172. 17. Skoog, L. & Nordenskjold, B. (1971) Eur. J . Biochem. 19, 81 -

18. Skoog, L. & Bjursell, G. (1974) J . Biol. Chem. 249, 6434-6438. 19. Nicander, B. & Reichard, P. (1985) J. Biol. Chem. 260, 5376-

374.

Res. 42, 41 3 -432.

95, 389 - 404.

SUPPI. 6 , 263-288.

Cell. Bid. 5 , 3443 - 3450.

J . Biochem. 33,428 -432.

191.

89.

5381.

20. Nicander, B. & Reichard, P. (1985) J . Biol. Chem. 260, 9216-

21. Reddy, G. P. V. & Pardee, A. B. (1980) Proc. Natl Acad. Sci. USA

22. Wickremasinghe, R. G. & Hoffbrand, A. V. (1983) FEBS Lett.

23. Snyder, R. D., van Houten, B. & Regan, J. D. (1984) Chem. Biol.

24. Collins, A. R. S., Squires, S. & Johnson, R. T. (1982) Nucleic

25. Collins, A,, Jones, C. & Waldren, C. (1982) J . Cell Sci. 56,423 -

26. Collins, A. R. S. (1977) Biochim. Biophys. Acta 478,461 -473. 27. Cleaver, J. E. (1981) J . Cell. Physiol. 108, 163-173. 28. Nicander, B. & Reichard, P. (1983) Proc. Natl Acad. Sci. USA

29. Hunting, D. J. & Dresler, S. L. (1985) Carcinogenesis 6, 1525-

30. Newman, C. N. & Miller, J. H. (1983) Biochem. Biophys. Res.

31. Collins, A. (1983) Biochim. Biophys. Acta 741, 341 -347. 32. Hunting, D. & Henderson, J. F. (1981) Can. J . Biochem. 59,723 -

33. Meuth, M. (1981) Somat. Cell Genet. 7, 89-102.

9222.

77, 3312-3316.

159,175-179.

Interactions 50, 1 - 14.

Acids Res. 10, 1203-1213.

440.

80, 1347-1351.

1528.

Comrnun. 116, 1064-1069.

727.