J. Cell Set. 13, 429-439 (i973) 429 Printed in Great Britain

LETHALITY OF FOR CULTURED MAMMALIAN CELLS BY INTERFERENCE WITH

K. ISHII* AND H. GREEN Department of , Massachusetts Institute of Technology, Cambridge, Massachusetts, 02139, U.S.A.

SUMMARY Adenosine at low concentration is toxic to mammalian cells in culture. This may escape notice because some sera (such as calf or human) commonly used in culture media, contain . In the absence of serum deaminase, adenosine produced inhibition of growth of a number of established cell lines at concentrations as low as 5 x io~* M, and killed at 2 x io~5 M. This effect required the presence of cellular , since a mutant line deficient in this was 70-fold less sensitive to adenosine. The toxic substance is therefore derived from adenosine by phosphorylation, and is probably one of the adenosine . The toxic effect of adenosine in concentrations up to 2 x io~* M was completely prevented by the addition of or of potentially convertible to uridine, suggesting that the adenosine was interfering with endogenous synthesis of uridylate. In the presence of adenosine, the conversion of labelled aspartate to uridine nucleotides was reduced by 80-85 %> and labelled orotate accumulated in both the cells and in the culture medium. The lethality of adenosine results from inhibition by one of its products of the synthesis of uridylate at the stage of phosphoribosylation of orotate.

INTRODUCTION Though adenosine is not an intermediate on the endogenous pathway of biosynthesis, it can be efficiently utilized through the purine salvage pathways as the sole purine source in cultured mammalian cells whose endogenous purine synthesis is blocked by (Green & Ishii, 1972). The route of its utilization under these conditions is predominantly through deamination to and successive conversion to and IMP. We report here that at least part of the reason for this is that calf and other mammalian sera contain sufficiently active adenosine deaminase to deaminate, under cell culture conditions, most added adenosine within some hours. If this is avoided by the use of serum lacking the deaminase, a part of the added adenosine is utilized through phosphorylation and even at quite low concen- tration has marked inhibitory effects on the cells, attributable to interference with pyrimidine synthesis. • Present address: Institute for Virus Research, Kyoto University, Sakyo-Ku, Kyoto 606, Japan. 43 o K. Ishii and H. Green

MATERIALS AND METHODS Cell culture Cells were cultivated as monolayers in the Dulbecco-Vogt medium (which contains no or pyrimidines) supplemented with 10% serum. The lines used were 3T3 and 3T6 (Todaro & Green, 1963); 3T6-TG8, lacking hypoxanthine phosphoribosyl transferase (Long et al. 1973); 3T6-TM, lacking adenosine kinase (Chan, Ishii, Long & Green, 1973); 3T6-DF8, lacking phosphoribosyl transferase (Kusano, Long & Green, 1971); and HeLa. Effects of adenosine on cell growth were tested on exponentially growing cultures at a cell below 5 x 103 per ml of medium. The cultures were observed over a period of 1 week, and growth was compared to that of the control in absence of adenosine according to an arbitrary scale ( + , + +, and + + +). Killing of the cells was indicated by cell detachment from the monolayers.

Adenosine deaminase activity of sera u [ C]adenosine (0-2-2-5 /*Ci) was diluted with unlabelled adenosine to IO~*-IO~3 M in 0-3 ml of a solution containing serum-free medium, 50 mM phosphate buffer, pH7-i, and 10% 6 serum. The Km of calf serum adenosine deaminase is 3-3 x io" M (Cory, Weinbaum & Suhadolnik, 1967). During incubation at 37 °C, 50-/1I samples were taken at intervals and added to 12 ml of cold o-oi NHCI; I ml of the solution was immediately applied to a 15 x o-6 cm column containing 0-4 ml of Dowex-soW-X8, 200-400 mesh, H+ form, previously washed with distilled . The column was eluted first with 14-4 ml of 0-5 M LiCl solution in 001 N HC1, which removed inosine and hypoxanthine; adenosine was then eluted with 48 ml of 0-2 M LiOH solution. The eluates were collected in fractions and 1 ml of each counted by liquid scintillation.

Conversion of 14C-labelled aspartate to uridine nucleotides 3T6 cells were inoculated into 100-mm Petri dishes and grown in medium containing 10% horse serum. On the following day, when the cells were in exponential growth, fresh medium was supplied with or without adenosine at io~* M. After 3 h incubation, the medium was renewed with addition of adenosine as before. Uniformly labelled [14C]aspartate (167 mCi/mM) was added to a concentration of 5 /tCi/ml of medium. After 2 h incubation, the medium was collected, and cold perchloric acid (PCA) was added to 5 %. The cell layers were washed with serum-free medium, and 4 ml of cold 5% PCA were added. After 10 min, the cell layers were detached with a rubber policeman. Each extract was centrifuged at 3000 rev/min and the pellet was washed with cold PCA. The washing and first supernatant were combined and neutralized with potassium carbonate. After standing overnight at 4 °C the precipitate was centrifuged. The supernatant was acidified by the addition of to 2 x io~3 N (pH 3-5), and washed charcoal (2% w/v) added (Smith & Khorana, 1963). The charcoal was washed with io~3 M formic acid and the adsorbed nucleotides eluted twice with 50% containing 0-5 N hydroxide. The eluates were combined, evaporated to dryness, disolved in water and applied to thin-layer chromatographic plates. Chromatography in the first dimension using isopropanol:HC1:water (70:15:15) (Wyatt, 1955) separated the nucleo- sides and nucleotides of , and orotate, together with orotate itself, from the and nucleotides of the other bases. Chromatography in a second dimension, using isopropanol:water: (85:15:1-3) (Wyatt, 1955) moved the nucleosides and bases away from the nucleotides. For measurement of radioactivity in the pyrimidine fractions, the plastic-backed cellulose layers were cut into 8-mm slices along the second dimension and counted in dioxane-based scintillator solution. The identity of the labelled products was established by comparison with the mobility of unlabelled standards located under u.v. illumination. In some cases radioautographs were prepared from chromatograms in which the radioactive pyrimidines and known unlabelled markers had migrated in the same track. Adenosine lethality and pyrimidine biosynthesis 431

100Q

Time, h Fig. 1. Deamination of adenosine by calf serum. [14C]adenosine was added to produce concentrations of ICC'-IO"6 M to a mixture of serum-free medium, phosphate buffer (pH 71) and 10 % calf or horse serum. Incubation was carried out at 37 °C. Ordinate shows the amount of labelled adenosine remaining with increasing incubation time. O—O—O, horse serum with io"3 M adenoaine, A; #—#—0, calf serum with io~3, io~4 and io~6 M adenosine, B, C, D respectively.

For further confirmation of the identity of labelled orotate, the labelled spot obtained in the second dimension chromatography (Fig. 2, p. 435) was eluted, evaporated and rechromato- graphed, either in a third solvent system containing n-butanohmethanol: ammonia: water (60:20:1:20) (Randerath & Randerath, 1967), or in a fourth solvent consisting of ethanol and 1 M ammonium , 1:1 (P. Cashian, personal communication). The latter gave ex- cellent resolution of orotate from and of orotate from dihydroorotate.

RESULTS The adenositie deaminase activity of mammalian serum The presence of an adenosine deaminase of low specific activity has been demon- strated in calf serum, and the enzyme has been purified (Cory et al. 1967). We have confirmed that relative to cell extracts the deaminase activity of calf serum is very low; for example, an extract of 3T6 cells contains per unit of protein about 300 times more deaminase activity than that of unfractionated calf serum. Yet considering the relative amounts of cells and medium employed in cell cultures and the time scale involved, the serum activity may be very appreciable. Fig. 1 shows the results of an experiment in which labelled adenosine was added to culture medium containing 10% calf serum or 10% horse serum, and the amount of adenosine remaining in the medium was followed with time (no cells were present). In the presence of calf serum, adenosine at the highest initial concentration (io~3 M) was half destroyed in about 8 h. At io"4 M, the half- was about 90 min, and at icr5 M, about 35 min. Most of the radioactivity lost from adenosine was recovered as hypoxanthine, indi- cating that calf serum also contains an enzyme capable of deribosylating inosine, 432 K. Ishii and H. Green

Table i. Effect of adenosine on 3T6

Medium supplemented with

mmol/1. Calf serum Horse serum o-o + + + + + + O-OO2 + + + + + + 0-005 + + + + + o-oi + + + ± 0-02 + + + Killed O-2O + + + Killed i-o ± Killed 2-O Killed Killed Growth assessed on an arbitrary scale, +, + +, + + +, compared with control without added adenosine.

Table 2. Inhibition by adenosine of growth of different cell lines in medium free from adenosine deaminase

Lowest inhibitory adenosine concentration Cell line mol/1. (xio«) 3T6 5 3T3 2 HeLa 35 3T6-TG8(HPT-) 20 3T6-DF8(APT-) S 3T6-TM(AK-) 350 probably inosine phosphorylase. Inosine and hypoxanthine together accounted for all the radioactivity lost from adenosine. In contrast to calf serum, horse serum was found to be completely free from deaminase. Incubation of io~3 M adenosine with medium containing 10% horse serum led to no detectable loss of adenosine (Fig. 1) or appearance of labelled inosine. Other sera tested and found to possess deaminase activity were foetal calf serum, human serum and y-globulin-free calf serum. The activity of the calf serum enzyme was unaffected by heating at 60 °C for 30 min.

The effect of adenosine on cultured cells in the absence of serum deaminase Several established lines were found to grow as well in medium supplemented with 10% horse serum as with 10% calf serum. However, when adenosine was added to the cell cultures in medium containing horse serum, it was found to be quite toxic at very low concentration. Table 1 shows a comparison of the effects of adenosine on 3T6 in medium containing the 2 kinds of serum. In 10% calf serum there was no effect on cell growth at low or moderate adenosine concentrations, while in medium containing 10 % horse serum there was definite inhibition of growth Adenosine lethality and pyrimidine biosynthesis 433 at 0-005 mil and killing of the cells at 0-02 mM. To obtain any effect on cells in medium containing calf serum, adenosine had to be added at 1 HIM or higher. A comparison of the relative sensitivity of a number of lines to adenosine in medium containing horse serum is shown in Table 2. 3T3 and HeLa cells were slightly more and slightly less sensitive, respectively, than wild type 3T6. Sublines of 3T6 lacking the enzyme hypoxanthine phosphoribosyl transferase (3T6-TG8) or lacking adenine phosphoribosyl transferase (3T6-DF8) were approximately as sensi- tive to adenosine as the wild type. However, a 3T6 subline deficient in adenosine kinase (3T6-TM) was much more resistant to adenosine, as an approximately 70- fold higher concentration was required to produce inhibition of growth. This suggests strongly that the toxic effect of low adenosine concentrations requires direct con- version of the adenosine to AMP, a reaction which cannot be carried out to any degree by the T36-TM (AK~) line (Chan et al. 1973). Of course, adenosine can be converted to AMP indirectly even in this line through the pathway adenosine -> ino- sine -y hypoxanthine -> IMP ->• adenylosuccinate -> AMP. However, as inosine and hypoxanthine are not toxic to 3T6 or 3T6-TM in concentrations up to 2 mM, the inhibitory effects of adenosine in low concentrations appear to depend on the direct conversion of adenosine to AMP by adenosine kinase. Other bases and nucleosides tested at 1 mM and found to have no inhibitory effect on the growth of 3T6 included uridine, , , and ; and at 0-4 mM. Adenine was toxic to 3T6 cells at 1 mM, but not at lower concentrations. If the enzyme adenine phosphoribosyl transferase, which converts adenine to AMP is less active than adenosine kinase, which converts adenosine to AMP, the AMP levels in the cell would not be driven up as readily by adenine as by adenosine. All tests of the effect of purines and pyrimidines on cell lines were carried out at cell concentrations below 5 x io3 per ml of medium. At higher cell concentration, the toxic effects of adenosine could be observed on the day following its addition, but the cells often recovered and grew progressively, though they were delayed compared with controls. Their recovery was probably due to destruction of the adenosine by the cellular adenosine deaminase, for in separate experiments it was found that in culture medium supplemented with 10% horse serum, and containing 4X io6 3T6 cells/ml, adenosine added to o-i mM was half destroyed in 2-5 h.

Protection from adenosine inhibition and lethality by pyrimidines Because of the possibility that the toxic effects of adenosine might result from some type of imbalance in the supply of purines or pyrimidines, other bases and nucleosides were examined for their ability to counteract the effect of adenosine. Table 3 shows that the toxic effect of adenosine in medium containing horse serum was completely abolished at all concentrations up to 0-2 mM by the addition of 1 mM uridine. The same was true for all other lines tested (3T6-DF8, HeLa). At concentrations of 1 mM, cytidine, and were as effective as uridine (Table 4). The pyrimidine bases of these nucleosides can very probably be converted into uridine through the action of deaminases and phosphorylases known to be present in mammalian cells (Creasy, 1963; Pontis, Degerstedt & Reichard, 28 CE L 13 434 K- Ishii and H. Green

Table 3. Protection of T,T6 from adenosine toxicity by uridine

Growth of 3T6 Adenosine, mmol/1. — Ur +Ur(imM) o-o + + + + + + o-ooi + + + + + + O-OO2 + + + + + + 0-005 + + + + + o-oi ± + + + 0-02 Killed + + + 0-05 Killed + + + 01 Killed + + + 0-2 Killed + + + 0-5 Killed + i-o Killed + 2-0 Killed ± Growth assessed on an arbitrary scale ( + , ++, + + +) against control culture without adenosine.

Table 4. Ability of other pyrimidines to protect 3 T6 cells against the lethal effects of adenosine (o-i mM)

I II Protection No protection Uridine Cytidine Orotidine Deoxyuridine Orotate Deoxycytidine Uracil Hypoxanthine Inosine Guanine Guanosine Deoxy guanosine All compounds were tested at 1 mM, except guanine which was at 0-4 mM, and deoxyguano- sine, at o-2 mM. Those in group I were equally effective at 1 mM, but at lower concentrations uridine was more effective than the others. Compounds in group II failed to prevent killing of the entire cell population.

1961). Uracil was also effective but cytosine was not. Orotate, orotidine and all of the purines tested were ineffective (Table 4). The toxicity of adenosine concentrations of 0-5 mM or higher could not be prevented in 3T6-TM by any of the compounds, including uridine.

Effect of adenosine on biosynthesis of uridine nucleotides In view of evidence suggesting that purines and pyrimidines may utilize the same permeation system for entry into the cell (Hakala & Kenny, 1972), the possibility was considered that uridine might protect against the toxic effects of adenosine by com- Adenosine lethality and pyrimidine biosynthesis 435

05

Origin Slice no. Fig. 2. Chromatography of pyrimidine nucleotides and nucleosides labelled with u [ C]aspartate (second dimension). A, cell layer; B, culture medium. # %t cells labelled with aspartate during the period 2-5 h after addition of 01 mM adenosine; O—O—O, cells labelled with aspartate without added adenosine. petitively inhibiting its entrance into the cell. An experiment was therefore performed to measure incorporation of [3H]adenosine present in the medium at a concentration of o-i mM, in the presence and absence of 1 mM uridine. The incorporation of labelled adenosine into perchloric acid (PCA)-insoluble form was higher in the presence of 1 mM uridine than in its absence at all times over the 24-h period examined. This finding gave no support to the idea that uridine prevented the uptake of adenosine, and it seemed more likely that the toxicity of adenosine over the range reversible by uridine (see Table 3) was due to interference with pyrimidine synthesis. As a test of this possibility, we determined the effect of added adenosine on the conversion of [14C]aspartate to acid-soluble uridine nucleotides. Adenosine (o-i mM) was added to a culture of growing 3T6 cells; 3 h later, fresh medium containing the same adenosine concentration and 5 /iCi/ml of [14C]aspartate was added. After further incubation, the medium was removed, the cells were extracted with 5 % PCA and the uridine nucleotides separated by 2-dimensional thin layer chromato- graphy on cellulose. It was found that the radioactivity in the uridine nucleotides was lowered to approximately 15 % of control levels (Fig. 2 and Table 5). Chromato- graphy also disclosed a labelled spot corresponding to orotate or orotidine in the

28-3 43° K. Ishii and H. Green

Table 5. Effect of adenosine on pyrimidine synthesis from [14C]aspartate*

Control, + Adenosine, cpm x io~* cpm x io"1 A. Undine Uridine nucleotides Orotate nucleotides Orotate

Cell layer 21-7 0-4 63 5'4 Medium 7'5 S'2 57 292 Expt. 1 Sum 292 56 12-0 34'6 . Total 348 466 Cell layer 132 0 22 2'5 Medium 8-o 44 96 309 Expt. 2 Sum 212 44 n-8 33'4 Total 256 452 Cell layer IS-4 O-2 26 22 Medium 196 IO5 167 35-8 Expt. 3 • Sum 3S'O 10-7 193 38-0 , Total 457 573 • Summary of 3 experiments of the type described in Fig. 2. extracts of cells grown in the presence of adenosine, while none was present in cells grown in medium to which no adenosine was added (Fig. 2 A). NO significant label was found in positions corresponding to uridine or thymidine in either extract, consistent with a relatively small pool of nucleosides compared with nucleotides. Examination of the culture medium at the time the cells were harvested showed small amounts of label migrating in the region occupied by the uridine nucleotides, and the amounts were similar whether adenosine had been present or not; but in the medium containing adenosine there was a large labelled spot corresponding to orotate or orotidine. The radioactivity in this spot exceeded that of the entire uridine nucleotide spot (Fig. 2B and Table 5). No discrete labelled spot with this mobility was obtained from the medium of cultures grown without added adenosine. The identity of the compound was established by elution from the chromatogram, addition of unlabelled orotate or orotidine and chromatography in the third or fourth solvent systems. The labelled compound whether isolated from the culture medium or the cell layer had a mobility which corresponded perfectly with the u.v. spot for orotate; no radioactivity could be found with the mobility of orotidine or dihydroorotate.

DISCUSSION Purine nucleotides have been shown to affect the synthesis of pyrimidines in bacteria (Anderson & Meister, 1966) but their effects are thought to be exerted at an early stage of the pyrimidine biosynthetic pathway. In cultured fibroblasts, allo- purinol, an analogue of hypoxanthine, interfered with the late stages of the biosyn- thetic pathway (Kelley, Beardmore, Fox & Meade, 1971) and caused of orotidine and orotate in man (Fox, Royse-Smith & O'Sullivan, 1970; Kelley & Adenosine lethality and pyrimidine biosynthesis 437 Beardmore, 1970). Even at 1 mM, did not have any toxic effect on the growth of cultured cells. Allopurinol and were found to be inhibitors of orotidylic decarboxylase in cell extracts (Kelley & Beardmore, 1970). Krooth (1964) has shown an inhibitory action of adenosine on the growth of mutant human diploid fibroblasts obtained from persons with . Owing to the very reduced activity of the last 2 in the pathway of pyrimidine synthesis (OMP pyrophosphorylase and OMP decarboxylase) the cells suffered from, partial deficiency of uridine, and grew somewhat more rapidly in the presence of added uridine. The growth of these strains was inhibited by adenosine, and the inhibition was relieved by uridine. Measurement of OMP decarboxylase activity in cells grown in the presence of adenosine showed slight reduction of the enzyme activity in wild type cells and inhibition to about one third of the activity in mutant cells. Since the addition of adenosine nucleotides did not inhibit the activity of the enzymes in extracts, the effect of adenosine on the cells was considered to be on the enzyme synthesis. Adenosine did not kill the mutant fibroblasts but only inhibited their growth reversibly. Wild type human diploid fibroblasts were completely un- affected (Krooth, 1964), but in view of our results this may only have been because adenosine deaminase was present in the serum supplement. In experiments on wild type human diploid fibroblasts (strain SB), we found that after preliminary adaptation for satisfactory growth in medium supplemented with horse serum, growth of the cells was strongly inhibited at adenosine concentrations of 0-02 mM and higher, and there was considerable cell killing, though the cultures were not totally destroyed. Our experiments on the effect of adenosine on pyrimidine synthesis in 3T6 indi- cates failure of reaction catalysed by OMP pyrophosphorylase. The effect of adenosine could hardly be due solely to reduced synthesis of the enzyme, as was thought to be the case for OMP decarboxylase in the experiments of Krooth, since the effect on nucleotide synthesis and the orotate excretion took place within 2-5 h after addition of adenosine. This was the earliest time tested since it is known that the cellular ATP pool does not equilibrate with externally added adenosine for at least 1*5 h (Emerson, 1971). Our results are more consistent with an effect of an adenosine nucleotide on the activity of OMP pyrophosphorylase or possibly on the availability of phosphoribosyl pyrophosphate. In the presence of a lethal concentration of adeno- sine, most of the free pyrimidines were in the form of orotate liberated into the medium (Table 5). Taking this orotate into consideration, it should be noticed that there was no overall reduction in pyrimidine synthesis in the presence of adenosine; on the contrary, there was some increase, implying that the scarcity of uridine nucleotides activated the reactions leading to orotate. These results suggest that orotic aciduria in man or animals might follow from a affecting the level of adenosine nucleotides. The mammalian organism seems to have gone to considerable length to protect cells from adenosine. The intestinal mucosa is differentiated for the production of a deaminase (Imondi, Lipkin & Balis, 1970) which presumably acts on absorbed adenosine. The serum deaminase would destroy any adenosine produced by nucleo- 438 K. hhii and H. Green tidase in cells and liberated into the blood. Fibroblast adenosine deaminase is extremely active, but in the absence of the serum enzyme, it is unable to protect small numbers of cells from the toxic effects of adenosine. It may be of interest that a fibroblast line (NBL-6, American Type Culture Collection) derived from the horse (the only species of the 3 examined whose serum did not contain adenosine deaminase) was also sensitive to adenosine killing in culture. We have not identified the nucleotide which interferes with the phosphoribosylation of orotate. The concentrations of AMP, ADP, ATP, cyclic AMP and other adenosine nucleotides may be elevated in cells exposed to adenosine. Cyclic AMP and its derivatives are known to suppress cell growth (Sheppard, 1971; Johnson, Friedman & Pastan, 1971) but are believed to do this at concentrations lower than those which produce toxic effects. In the absence of the serum deaminase, adenosine inhibits cell growth at concen- trations comparable to those at which some analogues of adenosine are toxic. The action of those analogues which can be phosphorylated (Schnebli, Hill & Bennett, 1967; Acs & Reich, 1967) may in part be due to effects similar to those we have described here for adenosine itself. These results also bear on the role of adenosine kinase. It seems clear that the function of this enzyme is not to act on exogenous adenosine but on adenosine generated within the cell through the action of . The kinase therefore has a regulating function which is to prevent excessive deamination of adenosine, a process which would lead to purine loss from the cells (Green & Ishii, 1972; Chan et al. 1973). On the other hand, when cells are exposed to exogenous adenosine, the kinase can phosphorylate an excessive amount, the eventual nucleotide product being raised to lethal concentration.

This investigation was aided by grants from the National Cancer Institute.

REFERENCES Acs, G. & REICH, E. (1967). Tubercidin and related pyrrolopyrimidine antibiotics. In Anti- biotics (ed. D. Gottlieb & P. D. Shaw), pp. 494-498. New York: Springer-Verlag. ANDERSON, P. M. & MEISTER, A. (1966). Control of Escherichia coli carbamyl phosphate syn- thetase by purine and pyrimidine nucleotides. , N.Y. 5, 3164-3169. CHAN, T-S., ISHII, K., LONG, C. & GREEN, H. (1973). Purine excretion by mammalian cells deficient in adenosine kinase.^. cell. Physiol. 81, 315-321. CORY, J. C, WEINBAUM, G. & SUHADOLNIK, R. J. (1967). Multiple forms of calf serum adeno- sine deaminase. Archs Biochem. Biophys. 118, 428—433. CREASY, W. (1963). Studies on the of 5-iodo-2'-deoxycytidine in vitro. Purification of deaminase from mouse kidney. J. biol. Chem. 238, 1772-1776. EMERSON, C. P. (1971). Regulation of the synthesis and stability of ribosomal RNA during contact inhibition of growth. Nature, New Biol. 232, 101-106. Fox, R. M., ROYSE-SMITH, D. & O'SULLIVAN, W. J. (1970). Orotidinuria induced by allo- purinol. Science, N.Y. 168, 861-862. GREEN, H. & ISHII, K. (1972). On the existence of a guanine nucleotide trap, the role of adeno- sine kinase and a possible cause of excessive purine production in mammalian cells. J. Cell Sci. n, 173-177- HAKALA, M. T. & KENNY, L. N. (1972). Common mechanism for the passage of purine and pyrimidine nucleosides through the plasma membrane of sarcoma 180 cells. Fedn Proc. Fedn Am. Socs exp. Biol. 31, 457. Adenosine lethality and pyrimidine biosynthesis 439 IMONDI, A. R., LIPKIN, M. & BALIS, M. E. (1970). Enzyme and template stability as regulatory mechanisms in differentiating intestinal epithelial cells. J. biol. Chem. 245, 2194-2198. JOHNSON, G. S., FRIEDMAN, R. M. & PASTAN, I. (1971). Restoration of several morphological characteristics of normal fibroblasts in sarcoma cells treated with adenosine-3': 5'-cyclic monophosphate and its derivatives. Proc. natn. Acad. Sci. U.S.A. 68, 425-429. KELLEY, W. N. & BEARDMORE, T. D. (1970). Allopurinol: Alteration in pyrimidine metabolism in man. Science, N.Y. 169, 388-390. KELLEY, W. N., BEARDMORE, T. D., FOX, I. H. & MEADE, J. C. (1971). Effect of allopurinol and on pyrimidine synthesis in cultured human fibroblasts. Biochem. Pharmac. 20, 1471-1478. KROOTH, R. S. (1964). Properties of diploid cell strains developed from patients with an in- herited abnormality of uridine biosynthesis. Cold Spring Harb. Symp. quant. Biol. 29, 189-212. KUSANO, T., LONG, C. & GREEN, H. (1971). A new reduced human-mouse somatic cell hybrid containing the human gene for adenine phosphoribosyl transferase. Proc. natn. Acad. Sci. U.S.A. 68, 82-86. LONG, C, CHAN, T.-S., LEVYTSKA, V., KUSANO, T. & GREEN, H. (1973). Absence of demon- strable linkage of human genes for enzymes of the purine and pyrimidine salvage pathways in human-mouse somatic cell hybrids. Biochem. Genet. 9, 283-297. PONTIS, H., DEGERSTEDT, G. & REICHARD, P. (1961). Uridine and deoxyuridine phosphorylases from Ehrlich ascites tumour. Biochim. biophys. Acta. 51, 138-147. RANDERATH, K. & RANDERATH, E. (1967). Thin layer separation methods for derivatives. In Methods in Enzymology, vol. \t (ed. L. Grossman & K. Moldave), pp. 323- 347. New York and London: Academic Press. SCHNEBLI, H. P., HILL, D. L. & BENNETT, L. L. (1967). Purification and properties of adenosine kinase from human tumor cells of type H.Ep. No. 2. J. biol. Chem. 242, 1997-2004. SHEPPARD, J. R. (1971). Restoration of contact-inhibited growth to transformed cells by di butyryl adenosine 3': s'-cyclicmonophosphate. Proc. natn. Acad. Sci. U.S.A. 68, 1316-1320. SMITH, M. & KHORANA, H. G. (1963). Preparation of nucleotides and derivatives. In Methods in Enzymology, vol. 6 (ed. S. P. Colowick & N. O. Kaplan), pp. 645-669. New York and London: Academic Press. TODARO, G. J. & GREEN, H. (1963). Quantitative studies of the growth of mouse embryo cells in culture and their development into established lines. J. Cell Biol. 17, 299-313. WYATT, G. R. (1955). Separation of nucleic acid components by chromatography on filter paper. In The Nucleic Acids, vol. 1 (ed. E. Chargaff & J. N. Davidson), pp. 243-265. New York and London: Academic Press. {Received 2 January 1973)