Incorporation of nucleoside analogs into nuclear or mitochondrial DNA is determined by the intracellular phosphorylation site

Nucleoside analogs used in cancer chemotherapy and in treatment of virus infections are phosphorylated in cells by nucleoside and nucleotide kinases to their pharmacologically active form. The phosphorylated nucleoside analogs are incorporated into DNA and cause cell death or inhibit viral replication. Cellular DNA is replicated both in the nucleus and in the mitochondria, and nucleoside analogs may interfere with DNA replication in both these subcellular locations. In the present study we created a cell model system where nucleoside analogs were phosphorylated, and thereby pharmacologically activated, in either the nucleus, cytosol, or mitochondria of cancer cells. The system was based on the reconstitution of deoxycytidine kinase (dCK)-deficient Chinese hamster ovary cells with genetically engineered dCK targeted to the different subcellular compartments. The nucleoside analogs phosphorylated by dCK in the mitochondria were predominantly incorporated into mitochondrial DNA, whereas the nucleoside analogs phosphorylated in the nucleus or cytosol were incorporated into nuclear DNA. We further show that the nucleoside analogs phosphorylated in the mitochondria induced cell death by an apoptotic program. These data showed that the subcellular site of nucleoside analog phosphorylation is an important determinant for incorporation of nucleoside analogs into nuclear or mitochondrial DNA.     

Bystander effects of nucleoside analogs phosphorylated in the cytosol or mitochondria

The efficiency of nucleoside kinase suicide gene therapy for cancer is highly dependent on "bystander" cell killing, i.e., the transfer of cytotoxic phosphorylated nucleoside analogs to cells adjacent to those expressing the suicide enzyme. We have recently studied the possible use of mitochondrial nucleoside kinases as suicide genes. In the present study, we investigated if nucleoside analogs phosphorylated in the mitochondrial matrix cause bystander killing. We used deoxycytidine kinase-deficient Chinese hamster ovary cells reconstituted with deoxycytidine kinase targeted to either the cytosol or mitochondria matrix and determined the bystander cell killing when these cells were incubated with the nucleoside analogs 1-beta-D-arabinofuranosylcytosine and 2',2'-difluorodeoxycytidine. A bystander effect occurred when nucleoside analogs were phosphorylated in the cytosol, but not when these compounds were phosphorylated in the mitochondria. These findings suggest that nucleoside kinases targeted to the mitochondrial matrix have limited use in suicide gene therapy when efficient bystander cell killing is required.     

Nucleoside analog cytotoxicity and bystander cell killing of cancer cells expressing Drosophila melanogaster deoxyribonucleoside kinase in the nucleus or cytosol.

We have recently shown that the overexpression of Drosophila melanogaster multisubstrate deoxyribonucleoside kinase (Dm-dNK) in cancer cell lines increases the cells' sensitivity to several cytotoxic nucleoside analogs and the enzyme may accordingly be used as a suicide gene in combined gene/chemotherapy treatment of cancer. To further characterize the enzyme for possible use as a suicide gene, we constructed a replication-deficient retroviral vector that expressed either the wild-type enzyme that localizes to the cell nucleus or a mutant (arg247ser) that localizes to the cytosol. A thymidine kinase-deficient osteosarcoma cell line was transduced with the recombinant virus and we compared the sensitivity and bystander cell killing when the cell lines were incubated with the pyrimidine nucleoside analogs (E)-5-(2-bromovinyl)-2'-deoxyuridine and 1-beta-D-arabinofuranosylthymine. In summary, we showed that the cells' sensitivity and the efficiency of bystander cell killing were not dependent on whether Dm-dNK was located in the nucleus or cytosol.     

Optimization and evaluation of electroporation delivery of siRNA in the human leukemic CEM cell line

In order to study nucleoside analog activation in the CEM cell line, a transfection protocol had to be optimized in order to silence an enzyme involved in nucleoside analog activation. Hematopoetic cell lines can be difficult to transfect with traditional lipid-based transfection, so the electroporation technique was used. Field strength, pulse length, temperature, electroporation media, siRNA concentration, among other conditions were tested in order to obtain approximately 70–80% mRNA and enzyme activity downregulation of the cytosolic enzyme deoxycytidine kinase (dCK), necessary for nucleoside analog activation. Downregulation was assessed at mRNA and enzyme activity levels. After optimizing the protocol, a microarray analysis was performed in order to investigate whether the downregulation was specific. Additionally two genes were differentially expressed besides the downregulation of dCK. These were however of unknown function. The leakage of intracellular nucleotides was also addressed in the electroporated cells since it can affect the DNA repair mechansism and the efficiency of nucleoside analogs. Three of these pools were increased compared to untreated, unelectroporated cells. The siRNA transfected cells with reduced dCK expression and activity showed reduced sensitivity to several nucleoside analogs as expected. The multidrug resistance to other drugs, as seen in nucleoside analog-induced resistant cells, was not seen with this model.     

Cytotoxic activity of 2',2'-difluorodeoxycytidine, 5-aza-2'-deoxycytidine and cytosine arabinoside in cells transduced with deoxycytidine kinase gene

Deoxycytidine nucleoside analogs must be first phosphorylated to become active anticancer drugs. The rate-limiting enzyme in this pathway is deoxycytidine kinase (dCK). Cells deficient in this enzyme are resistant to these analogs. To evaluate the potential of dCK to be used as suicide gene for deoxycytidine nucleoside analogs, we transduced both human A-549 lung carcinoma and murine NIH3T3 fibroblast cell lines with this gene. The dCK-transduced cells showed an increase in cytotoxicity to the analogs, cytosine arabinoside (ARA-C), and 5-aza-2'-deoxycytidine (5-AZA-CdR). Unexpectedly, the related analog, 2',2'-difluorodeoxycytidine (dFdC), was less cytotoxic to the dCK-transduced cells than the wild-type cells. For the A-549-dCK cells, the phosphorylation of dFdC by dCK was much greater than control cells. In accord with the elevated enzyme activity, we observed a 6-fold increased dFdC incorporation into DNA and a more pronounced inhibition of DNA synthesis in the A-549-dCK cells. In an attempt to clarify the mechanism of dFdC, we investigated its action on A549 and 3T3 cells transduced with both cytidine deaminase (CD) and dCK. We reported previously that overexpression of CD confers drug resistance to deoxycytidine analogs. In this study, when the CD-transduced cells were also transduced with dCK they became relatively more sensitive to dFdC. In addition, we observed that dFdU, the deaminated form of dFdC, was cytotoxic to the A-549-dCK cells, but not the wild-type cells. Our working hypothesis to explain these results is that the mitochondrial thymidine kinase (TK2), an enzyme reported to phosphorylate dFdC, acts as an important modulator of dFdC-induced cell toxicity. These findings may further clarify the action of dFdC and the mechanism by which it induces cell death.     

Selective assays for thymidine kinase 1 and 2 and deoxycytidine kinase and their activities in extracts from human cells and tissues.

Human cells salvage pyrimidine deoxyribonucleosides via 5'-phosphorylation which is also the route of activation of many chemotherapeutically used nucleoside analogs. Key enzymes in this metabolism are the cytosolic thymidine kinase (TK1), the mitochondrial thymidine kinase (TK2) and the cytosolic deoxycytidine kinase (dCK). These enzymes are expressed differently in different tissues and cell cycle phases, and they display overlapping substrate specificities. Thymidine is phosphorylated by both thymidine kinases, and deoxycytidine is phosphorylated by both dCK and TK2. The enzymes also phosphorylate nucleoside analogs with very different efficiencies. Here we present specific radiochemical assays for the three kinase activities utilizing analogs as substrates that are by more than 90 percent phosphorylated solely by one of the kinases; i.e. 3'-azido-2',3'-dideoxythymidine (AZT) as substrate for TK1, 1-beta-D-arabinofuranosylthymidine (AraT) for TK2 and 2-chlorodeoxyadenosine (CdA) for dCK. We determined the fraction of the total deoxycytidine and thymidine phosphorylating activity that was provided by each of the three enzymes in different human cells and tissues, such as resting and proliferating lymphocytes, lymphocytic cells of leukemia patients (chronic lymphocytic, chronic myeloic and hairy cell leukemia), muscle, brain and gastrointestinal tissue. The detailed knowledge of the pyrimidine deoxyribonucleoside kinase activities and substrate specificities are of importance for studies on chemotherapeutically active nucleoside analogs, and the assays and data presented here should be valuable tools in that research.     

From the covalent linkage of drugs to novel inhibitors of ribonucleotide reductase: Synthesis and biological evaluation of valproic esters of 3′-C-methyladenosine

We synthesized a series of serum-stable covalently linked drugs derived from 3′-C-methyladenosine (3′-Me-Ado) and valproic acid (VPA), which are ribonucleotide reductase (RR) and histone deacetylase (HDAC) inhibitors, respectively. While the combination of free VPA and 3′-Me-Ado resulted in a clear synergistic apoptotic effect, the conjugates had lost their HDAC inhibitory effect as well as the corresponding apoptotic activity. Two of the analogs, 2′,5′-bis-O-valproyl-3′-C-methyladenosine (A160) and 5′-O-valproyl-3′-C-methyladenosine (A167), showed promising cytotoxic activities against human hematological and solid cancer cell lines. A167 was less potent than A160 but had interesting features as an RR inhibitor. It inhibited RR activity by competing with ATP as an allosteric effector and concomitantly reduced the intracellular deoxyribonucleoside triphosphate (dNTP) pools. A167 represents a novel lead compound, which in contrast to previously used RR nucleoside analogs does not require intracellular kinases for its activity and therefore holds promise against drug resistant tumors with downregulated nucleoside kinases.     

Potential of ribozymes against deoxycytidine kinase to confer drug resistance to cytosine nucleoside analogs

Hematopoietic toxicity is the dose-limiting side effect produced in cancer chemotherapy with deoxycytidine nucleoside analogs. Deletion of the deoxycytidine kinase (dCK), results in a drug resistance phenotype to these analogs. An interesting gene therapy strategy to confer drug resistance to cytosine nucleoside analogs would be to specifically inactivate the dCK in normal hematopoietic stem cell. In this study, we designed hammerhead ribozymes that can specifically cut and downregulate the murine dCK mRNA. Three different ribozymes were identified and shown to cleave in vitro the dCK RNA. After introduction of ribozyme cDNA into murine L1210 leukemic cells by retroviral transfer, two of the ribozymes showed some capacity in reducing dCK activity. However, analysis of transduced L1210 clones showed that the significant reduction in the dCK mRNA was not sufficient to confer drug resistance to cytosine arabinoside. Nevertheless, these results provide a new avenue of modulating the dCK enzyme activity and with improved modifications may have the potential for use in gene therapy to confer drug resistance to deoxycytidine analogs.     

Regulation of ribonucleotide reductase activity in intact mammalian cells

An intact cell assay system based upon Tween-80 permeabilization was used to investigate the regulation of ribonucleotide reductase activity in Chinese hamster ovary cells. Models used to explain the regulation of the enzyme have been based upon work carried out with cell-free extracts, although there is concern that the properties of such a complex enzyme would be modified by extraction procedures. We have used the intact cell assay system to evaluate, within whole cells, the current model of ribonucleotide reductase regulation. While some of the results agree with the proposals of the model, others do not. Most significantly, it was found that ribonucleotide reductase within the intact cell could simultaneously bind the nucleoside triphosphate activators for both CDP and ADP reductions. According to the model based upon studies with cell-free preparations, the binding of one of these nucleotides should exclude the binding of others. Also, studies on intracellular enzyme activity in the presence of combinations of nucleotide effectors indicate that GTP and perhaps dCTP should be included in a model for ribonucleotide reductase regulation. For example, GTP has the unique ability to modify through activation both ADP and CDP reductions, and synergistic effects were obtained for the reduction of CDP by various combinations of ATP and dCTP. In general, studies with intact cells suggest that the in vivo regulation of ribonucleotide reductase is more complex than predicted from enzyme work with cell-free preparations. A possible mechanism for the in vivo regulation of ribonucleotide reductase, which combines observations of enzyme activity in intact cells and recent reports of independent substrate-binding subunits in mammalian cells is discussed.     

Nucleoside 5'-triphosphate analogs as positive and negative effectors of mammalian ribonucleotide reductase

Ribonucleotide reductase activity is strongly regulated by nucleoside 5'-triphosphates acting as positive and negative effectors. With the use of dGTP analogs, araGTP and dITP, it was found that the structural requirements of dGTP to serve as a positive effector of ADP reductase were not the same as the requirements for dGTP to serve as a negative effector of CDP and ADP reductase activities. The dTTP analogs methylenedTTP and dideoxyTTP also gave different responses in terms of activating GDP reductase activity and inhibiting CDP and ADP reductase activities. Etheno-ATP and etheno-dATP were inactive as positive and negative effectors, respectively, of CDP reductase activity. DideoxyATP was less active than dATP as a negative effector. Formycin ATP was a very poor substitute for ATP as a positive effector of CDP reductase. These studies indicate that the effector sites are very specific in terms of binding nucleoside triphosphates as positive or negative modulators of ribonucleotide reductase activity.     

Special function of deoxycytidine kinase (dCK) in the activation of chemotherapeutic nucleoside analogs and in the inhibition of cell proliferation

Deoxycytidine kinase (dCK) plays a central role in the deoxynucleoside salvage processes, phosphorylating dC, dA, and dG to their monophosphates. In mammalian cells, the major source of dTTP comes also from dC via dCMP deaminase. Moreover, based on its broad substrate specificity, this enzyme is responsible for the activation of several nucleoside analogues of therapeutical importance, influencing the sensitivity of malignant tissues towards chemotherapy. The expression of dCK is highest in different lymphoid cells/tissues, in embryonic cells and in most malignant cells (2, 7, 13-15, 18). The activity of dCK is not cell cycle-regulated. In contrast to this, dCK activity was found to be elevated several fold upon short-term treatments of normal human lymphocytes with therapeutic nucleoside analogs, and other genotoxic agents as well as by DNA damaging agents including the DNA polymerase inhibitor aphidicolin, the topoisomerase II inhibitor etoposide and gamma-irradiation, which might be a potentially important phenomenon with respect to the clinical practice, too. These findings indicated that the main trigger of activation could be the damaged DNA itself, and the biological relevance might be to supply the dNTPs for the enhanced DNA repair. Activation of dCK was paralleled by elevated levels of intracellular dATP, raising the possibility that dCK activation is linked to the induction of apoptosis. With regard to the mechanism of enzyme activation, no changes were found in the protein and mRNA levels of dCK upon stimulation, while the activation process was calcium dependent and comprised a protein phosphorylation step. A positive correlation was found between the enzymatic activity and the native immunoreactivity of dCK, strongly arguing that dCK undergoes a conformational change during activation, which results in the formation of a catalytically more active steric structure (8-11, 22, 26, 32-34, 35, 36).     

Protein phosphatase 2A regulates deoxycytidine kinase activity via Ser-74 dephosphorylation

Deoxycytidine kinase (dCK) is a critical enzyme for activation of anticancer nucleoside analogs. Its activity is controlled via Ser-74 phosphorylation. Here, we investigated which Ser/Thr phosphatase dephosphorylates Ser-74. In cells, the PP1/PP2A inhibitor okadaic acid increased both dCK activity and Ser-74 phosphorylation at concentrations reported to specifically target PP2A. In line with this, purified PP2A, but not PP1, dephosphorylated recombinant pSer-74-dCK. In cell lysates, the Ser-74-dCK phosphatase activity was found to be latent, Mn²⁺-activated, responsive to PP2A inhibitors, and diminished after PP2A-immunodepletion. Use of siRNAs allowed concluding definitively that PP2A constitutively dephosphorylates dCK in cells and negatively regulates its activity.     

A simple method to purify ribonucleotide reductase

Assays of ribonucleotide reductase in extracts of Detroit 98 (human) cells were found to be complicated by the rapid depletion of the substrate (CDP) by nucleoside diphosphate kinase. Assays of either 100,000g supernatants or ammonium sulfate-fractionated extracts resulted in the conversion of >90% of the substrate to CTP within 2 min. It was therefore desirable to separate nucleoside diphosphate kinase from ribonucleotide reductase. Chromatography of the fractionated extract on an ATP-agarose column resulted in the delivery of nondissociated ribonucleotide reductase in the void volume and the retention of >99.9% of the nucleoside diphosphate kinase. The kinase could be eluted by 2 mm ATP. The ribonucleotide reductase was recovered from this commercially available gel with an apparent yield of >200%. It could be accurately assayed with only minimal extraneous depletion of substrate. Furthermore, it was stable to storage at ?80°C. Tris-HCl was found to inhibit the enzyme. When HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid)-Na buffer was used in place of Tris-HCl, the rate of CDP reduction was increased by 2.5-fold. Since the above procedure selectively removes nucleoside diphosphate kinase from crude preparations of ribonucleotide reductase, it should have general applicability for purifying ribonucleotide reductase from other sources.     

Studies on the mechanism of adenosylcobalamin-dependent ribonucleotide reduction by the use of analogs of the coenzyme.

A series of 17 analogs of 5'-deoxy-5'-adenosylcobalamin(adenosylcobalamin) have been synthesized with modifications in the base or ribose moiety of the nucleoside ligand. These analogs have been examined for their effects on reactions catalyzed by the ribonucleotide reductase of Lactobacillus leichmannii. All the analogs are inhibitors of ATP reduction in the presence of adenosylcobalamin as coenzyme, and hence all are bound to the catalytic site. Only the 3-beta-D-ribofuranosyladenine analog (isoadenosylcobalamin) showed substantial activity as a coenzyme in ATP reduction, giving a rate of 59% of that obtained with the adenosylcobalamin. Lesser rates of reduction were obtained with nebularyl-, 2'-deoxyadenosyl-, tubercidyl-, isopropylideneadenosyl-, L-adenosyl-, and ara-adenosylcobalamin, coenzyme activity decreasing in that order. Other analogs showed no significant coenzyme activity. The rate of hydrogen exchange into water from the 5'-methylene group of the nucleoside ligand appeared to parallel the coenzyme activity in those analogs examined, but only the four cobalamins with highest coenzyme activity (adenosyl, isoadenosyl, nebularyl, 2'-deoxyadenosyl) gave detectable amounts of "active coenzyme B12," THe rapidly formed paramagnetic intermediate of ribonucleotide reduction. The enzyme system produced the slowly formed paramagnetic species characterized by a doublet EPR spectrum only with adenosyl- and isoadenosylcobalamin. By contrast the enzymic degradation of analogs to cob(II)alamin and 5'-deoxynucleoside occurred not only with those analogs active as coenzymes and in the exchange reaction but also with a number of coenzymically inactive analogs, and the rate of degradation was unrelated to the rate of ribonucleotide reduction for those analogs with coenzyme activity.     

Mitochondrial purine and pyrimidine metabolism and beyond

ABSTRACT

Carefully balanced deoxynucleoside triphosphate (dNTP) pools are essential for both nuclear and mitochondrial genome replication and repair. Two synthetic pathways operate in cells to produce dNTPs, e.g., the de novo and the salvage pathways. The key regulatory enzymes for de novo synthesis are ribonucleotide reductase (RNR) and thymidylate synthase (TS), and this process is considered to be cytosolic. The salvage pathway operates both in the cytosol (TK1 and dCK) and the mitochondria (TK2 and dGK). Mitochondrial dNTP pools are separated from the cytosolic ones owing to the double membrane structure of the mitochondria, and are formed by the salvage enzymes TK2 and dGK together with NMPKs and NDPK in postmitotic tissues, while in proliferating cells the mitochondrial dNTPs are mainly imported from the cytosol produced by the cytosolic pathways. Imbalanced mitochondrial dNTP pools lead to mtDNA depletion and/or deletions resulting in serious mitochondrial diseases. The mtDNA depletion syndrome is caused by deficiencies not only in enzymes in dNTP synthesis (TK2, dGK, p53R2, and TP) and mtDNA replication (mtDNA polymerase and twinkle helicase), but also in enzymes in other metabolic pathways such as SUCLA2 and SUCLG1, ABAT and MPV17. Basic questions are why defects in these enzymes affect dNTP synthesis and how important is mitochondrial nucleotide synthesis in the whole cell/organism perspective? This review will focus on recent studies on purine and pyrimidine metabolism, which have revealed several important links that connect mitochondrial nucleotide metabolism with amino acids, glucose, and fatty acid metabolism.     

Regulation of ribonucleotide reductase activity in mammalian cells

Mammalian ribonucleotide reductase catalyzes the rate-limiting for the de novo synthesis 2'-deoxyribonucleoside 5'-triphosphates. There is some suggestion that this step may also be the rate-limiting step of DNA synthesis. It is apparent that the level of the enzyme, ribonucleotide reductase, varies through the cell cycle and is highest in those tissues with the greatest proliferation rate. This increase in activity is associated with increased protein synthesis. The purified enzyme has been shown to be subject to strict allosteric regulation by the various nucleoside triphosphates and it has been proposed that allosteric regulation plays an important role in the level of ribonucleotide reductase activity which is expressed. All experimental data relating to this point, however, do not support the role of deoxyribonucleoside triphosphates as a major factor in determining cellular reductase activity during normal cell division. Several naturally occurring factors have been isolated from cells which lower ribonucleotide reductase activity in vitro. These factors have been found in tissues of low growth fraction and appear to be absent or low in tissues or high growth fraction such as tumor, regenerating liver and embryonic tissues. The expression of intracellular ribonucleotide reductase activity is therefore controlled at various levels and by various factors and the prevailing mode of regulation may vary throughout the cell cycle transverse and also in the various types of cells.     

Induction of a new ribonucleotide reductase after infection of mouse L cells with pseudorabies virus.

The mammalian ribonucleotide reductase consists of two nonidentical subunits, protein M1 and M2. M1 binds nucleoside triphosphate allosteric effectors, whereas M2 contains a tyrosine free radical essential for activity. The activity of ribonucleotide reductase increased 10-fold in extracts of mouse L cells 6 h after infection with pseudorabies virus. The new activity was not influenced by antibodies against subunit M1 of calf thymus ribonucleotide reductase, whereas the reductase activity in uninfected cells was completely neutralized. Furthermore, packed infected cells (but not mock-infected cells) showed an electron paramagnetic resonance spectrum of the tyrosine free radical of subunit M2 of the cellular ribonucleotide reductase. These data given conclusive evidence that on infection, herpesvirus induces a new or modified ribonucleotide reductase. The virus-induced enzyme showed the same sensitivity to inhibition by hydroxyurea as the cellular reductase. The allosteric regulation of the virus enzyme was completely different from the regulation of the cellular reductase. Thus, CDP reduction catalyzed by the virus enzyme showed no requirement for ATP as a positive effector, and no feedback inhibition was observed by dTTP or dATP. The virus reductase did not even bind to a dATP-Sepharose column which bound the cellular enzyme with high affinity.     

Structural basis for activation of the therapeutic L-nucleoside analogs 3TC and troxacitabine by human deoxycytidine kinase

L-nucleoside analogs represent an important class of small molecules for treating both viral infections and cancers. These pro-drugs achieve pharmacological activity only after enzyme-catalyzed conversion to their tri-phosphorylated forms. Herein, we report the crystal structures of human deoxycytidine kinase (dCK) in complex with the L-nucleosides (-)-beta-2',3'-dideoxy-3'-thiacytidine (3TC)--an approved anti-human immunodeficiency virus (HIV) agent--and troxacitabine (TRO)--an experimental anti-neoplastic agent. The first step in activating these agents is catalyzed by dCK. Our studies reveal how dCK, which normally catalyzes phosphorylation of the natural D-nucleosides, can efficiently phosphorylate substrates with non-physiologic chirality. The capability of dCK to phosphorylate both D- and L-nucleosides and nucleoside analogs derives from structural properties of both the enzyme and the substrates themselves. First, the nucleoside-binding site tolerates substrates with different chiral configurations by maintaining virtually all of the protein-ligand interactions responsible for productive substrate positioning. Second, the pseudo-symmetry of nucleosides and nucleoside analogs in combination with their conformational flexibility allows the L- and D-enantiomeric forms to adopt similar shapes when bound to the enzyme. This is the first analysis of the structural basis for activation of L-nucleoside analogs, providing further impetus for discovery and clinical development of new agents in this molecular class.     

Reversion of hydroxyurea resistance, decline in ribonucleotide reductase activity, and loss of M2 gene amplification

A key rate-limiting reaction in the synthesis of DNA is catalyzed by ribonucleotide reductase, the enzyme which reduces ribonucleotides to provide the deoxyribonucleotide precursors of DNA. The antitumor agent, hydroxyurea, is a specific inhibitor of this enzyme and has been used in the selection of drug resistant mammalian cell lines altered in ribonucleotide reductase activity. An unstable hydroxyurea resistant population of mammalian cells with elevated ribonucleotide reductase activity has been used to isolate three stable subclones with varying sensitivities to hydroxyurea cytotoxicity and levels of ribonucleotide reductase activities. These subclones have been analyzed at the molecular level with cDNA probes encoding the two nonidentical subunits of ribonucleotide reductase (M1 and M2). Although no significant differences in M1 mRNA levels or gene copy numbers were detected between the three cell lines, a strong correlation between cellular resistance, enzyme activity, M2 mRNA and M2 gene copies was observed. This is the first demonstration that reversion of hydroxyurea resistance is directly linked to a decrease in M2 mRNA levels and M2 gene copy number, and strongly supports the concept that M2 gene amplification is an important mechanism for achieving resistance to this antitumor agent through elevations in ribonucleotide reductase.