DNA Synthesis as a Therapeutic Target: The First 65 Years

This article was inspired by my reading of The Emperor of All Maladies: A Biography of Cancer, by Siddhartha Mukherjee, M.D. (1) and in particular, by reading two reviews of the book by scientists whom I admire (2, 3). These reviewers were transfixed, as was I, by accounts in the early chapters of Sidney Farber and his use of folic acid analogs, beginning in 1947, in the experimental treatment of acute lymphocytic leukemia in desperately ill children. However, both reviewers and I were struck by how little information was presented about mechanisms of action of the early antimetabolites, including the folate analogs used by Farber, and at how many of those early antimetabolites were mentioned only briefly, if at all. Gertrude Elion and George Hitchings at Burroughs-Wellcome rate a page for their synthesis of 6-mercaptopurine (6-MP) and other purine analogs, but Charles Heidelberger, who synthesized 5-fluorouracil (FU) and 5-fluorodeoxyuridine (FUdR), receives no mention at all.

To be sure, the genetics of cancer is probably easier to write about for the educated lay public than the enzymes and metabolic pathways, which must be understood if one is to grasp the actions of aminopterin or FU, as well as mechanisms of toxicity and resistance. Mukherjee writes lucidly about carcinogens, oncogenes, tumor suppressors, and the chain of mutations that lead from a normal cell to a cancer cell. However, given the attention paid to Mukherjee's book, which I consider excellent scientific writing, despite these omissions, and given the number of important compounds that resulted from early development of antimetabolites in cancer treatment, some brief comments about the early nucleotide antagonists—the “pioneer” antineoplastic drugs (Fig. 1)—and their actions seem in order. This topic is of special interest to me, as both my Ph.D. adviser and my postdoctoral mentor made seminal contributions to understanding biochemical actions of the early antimetabolites. At the University of Washington in the late 1950s, Frank Huennekens and his group were characterizing enzymes that involve folate coenzymes. Of particular note, Osborn, Freeman, and Huennekens (4) reported that aminopterin and methotrexate are potent inhibitors of dihydrofolate reductase (DHFR), with this almost certainly being the basis for their action against leukemias and other neoplasias. The finding that enzymes could act as specific targets for the action of drugs was a revelation to me and helped lead me to Frank's laboratory for my Ph.D. thesis.

Figure 1.

Structures of the pioneer antineoplastic drugs.

During that period, I became aware of Seymour Cohen's work at the University of Pennsylvania. In 1954, Cohen and his coworkers had discovered thymineless death, a process in which bacteria starved of thymine or thymidine undergo rapid loss of viability, a phenomenon quite different from the consequences of deprivation of an essential amino acid, for example, where growth would cease, but cells would not die. Then, in 1958, Cohen's laboratory reported the basis for the biochemical actions of FU and FUdR (5). Metabolic conversion of either the base or the nucleoside to FUdR 5′-monophosphate (FdUMP) generates an analog of deoxyuridine monophosphate, the substrate for thymidylate synthase (TS). The analog is a potent inhibitor of the enzyme, and it causes in mammalian cells a phenomenon akin to thymineless death; growing cells are starved for an essential DNA component, and this causes them to die. At about this time, Cohen was also discovering that virus infection stimulates the establishment of new metabolic pathways in infected cells, and this is what led me to his laboratory for postdoctoral work.

My aim in writing this article is to introduce one-half dozen nucleic acid antimetabolites, developed in the 1940s and 1950s as potential anticancer agents, and to briefly describe their molecular and metabolic actions. Hundreds or thousands of papers have been published about each compound, so this overview will be cursory. However, this small family of pioneer drugs (Table 1), which include the folate antagonists and the fluorinated pyrimidines, has several common features. First, they are all synthetic analogs, synthesized in the 1940s and 1950s (with the exception of HU, which was synthesized in 1859 and is not an analog). Second, despite problems with toxicity and drug resistance, the analogs to be discussed are still in clinical use, despite the dozens of newer drugs that have been developed, some taking their lead from these pioneer drugs. Third, most of these compounds have found uses, often unexpected, in treating diseases other than cancer. Fourth, these compounds have found use as research reagents, and in particular, their use in the clinic and in the laboratory has led to discovery of biological processes of fundamental importance. Finally, advances in understanding structures and mechanisms of target enzymes and metabolism of these pioneer drugs have opened the door to more effective, more specific, and less toxic ways of using these drugs and their derivatives. Important work, designed to improve the effectiveness of these drugs, is under way, 65 years after Farber's seminal discoveries.

Table 1.

Therapeutic uses of “pioneer” anticancer drugs^*

Agent	Target enzyme	Uses in cancer treatment	Other clinical use
Folate analogs (aminopterin, methotrexate)	DHFR	Acute lymphocytic leukemia; choriocarcinoma; breast, head, neck, and lung cancers; osteosarcoma; bladder cancer	Psoriasis; rheumatoid arthritis; organ transplant immunosuppression
Fluorinated pyrimidines (FU; FUdR)	TS	Breast, colon, esophageal, stomach, pancreas, head and neck; premalignant skin lesion (topical)
Arabinosylcytosine (araC)	DNA polymerase	Acute myelogenous and acute lymphocytic leukemias; non-Hodgkins lymphoma
Purine analogs (6-MP; 6-thioguanine)	(DNA synthesis)	Acute lymphocytic and acute myelogenous leukemias; small-cell non-Hodgkins lymphoma	Crohn's disease
Hydroxyurea (HU)	Ribonucleotide reductase (RNR)	Chronic myelogenous leukemia; polycythemia vera; essential thrombocytosis	Sickle-cell disease

^*Clinical applications listed from ref. (6).

My own contributions, early in my career, to understanding the biochemical actions of these agents—folate antagonists and fluorinated pyrimidines—were modest but perhaps, bear brief retelling to reveal my motivation in writing this article. Both methotrexate and FdUMP were shown to behave kinetically, as irreversible inhibitors of their respective target enzymes. In 1961, Werkheiser (7) showed that one could essentially titrate DHFR with aminopterin, one of the folate analogs used by Farber. As shown in Fig. 2, a plot of reaction rate versus enzyme concentration in the presence of a fixed amount of aminopterin showed a break point, presumably representing equivalence between aminopterin and its binding sites on DHFR and suggesting that inhibition was irreversible. This was useful information in the era before highly purified enzymes and antibodies to the same were readily available, as the break point resembled a titration end point, allowing estimation of the amount of the target enzyme in tissues, such as the white blood cells of leukemia patients being treated with folate analogs.

Figure 2.

Titration of DHFR by aminopterin. Enzyme activity was measured as a function of enzyme concentration in the presence and absence of aminopterin as indicated (7).

As a graduate student, I tried to duplicate Werkheiser's results, working with partially purified DHFR from chicken liver. However, my data showed a pattern, suggesting reversible inhibition. Moreover, the enzyme-substrate complex could be readily dissociated by passage through a hydroxylapatite column (8). In graduate school courses, I had been taught that competitive inhibitors are substrate analogs that bind reversibly at the catalytic site. So, it was a relief to learn that aminopterin, which behaved kinetically as a noncompetitive inhibitor, at least under some circumstances, was actually binding reversibly, albeit tightly enough under some circumstances to appear kinetically irreversible.

Several years later, my graduate student, John Erickson, reinvestigated the DHFR-folate antagonist interaction, this time using the enzyme encoded by bacteriophage T4, which he had obtained as the earliest form of DHFR to be purified to homogeneity. Erickson observed that the UV difference spectrum of enzyme-bound versus free aminopterin was remarkably similar to the difference spectrum of free aminopterin at pH 2.8 versus pH 7 (Fig. 3). This suggested that binding of folate analogs to DHFR involved protonation of the 4-amino group of the analog (which replaces a hydroxyl in folate or dihydrofoate), thereby allowing formation of an additional hydrogen bond to stabilize the enzyme-inhibitor complex (9). Later, crystallographic analyses of DHFR supported this prediction and in fact, showed that the 4-amino group forms two hydrogen bonds with residues on the protein (10).

Figure 3.

Analysis of aminopterin binding to T4 phage DHFR by UV difference spectrophotometry. Solid line, Spectrum of 3.6 μM aminopterin in the presence of 2.8 μM DHFR versus 3.6 μM free aminopterin. Dashed line, Spectrum of 2.9 μM aminopterin at pH 2.8 versus 2.9 μM aminopterin at pH 7.0 (9).

The inhibition of TS by FdUMP presented similarities but led to a different conclusion. Cohen et al. (5) had reported irreversible inhibition by FdUMP in enzyme preparations from T-even phage-infected Escherichia coli. However, Hartmann and Heidelberger (11) reported the inhibition as reversible and competitive in preparations from Ehrlich ascites tumor cells. Seymour asked me to investigate the apparent discrepancy. As it turned out, there was no discrepancy (12). When T2 or T6 phage-induced TS was incubated with FdUMP for 5 min, inhibition was irreversible, and titration via Werkheiser plots could easily be seen. However, when reactions were initiated by adding enzyme to a mixture of substrate (dUMP) and inhibitor (FdUMP) in the presence of the cofactor, 5,10-methylenetetrahydrofolate (mTHF), the data showed classical, competitive inhibition, as determined in part by nice Lineweaver-Burk plots. Evidently, at the substrate/inhibitor ratios used, 1000 or more, FdUMP could bind readily and irreversibly at the active site but only after competing with dUMP for access to that site. Later work, in the laboratories of Daniel Santi and Peter Danenberg (13), confirmed that FdUMP is not only a tightly binding substrate analog but is also a covalently bound inhibitor, one of the earliest, well-understood cases of mechanism-based inhibition. Analysis of the covalently bound FdUMP-TS complex opened the door to mechanistic understanding of the TS reaction and to development of other antagonists.

Now, I would like to briefly discuss the biochemistry of each class of these pioneer drugs and to describe how each compound has found use in the research laboratory, as well as in the clinic. These much-abbreviated discussions may be viewed as filling gaps that reviewers noted in discussing Dr. Mukherjee's otherwise excellent book.

Folate antagonists

As described in Mukherjee's book, a group at Lederle Laboratories synthesized several folate analogs in the 1940s and provided them to Sidney Farber. The most effective antiproliferative agents were aminopterin and methotrexate. The aminopterin molecule is identical to that of folic acid, except for substitution of an amino group for a hydroxyl on the pteridine ring. Methotrexate has one additional modification—a methyl group on the nitrogen of the p-aminobenzoic acid moiety. Although aminopterin is more closely related to folate than is methotrexate, and both are comparable in potency as DHFR inhibitors, it is methotrexate that is used in the clinic. This may relate to relative tendencies of each molecule to form polyglutamates after uptake into cells, a modification that helps prevent efflux of the molecule from the cell.

As tetrahydrofolate (THF) coenzymes are involved in the biosynthesis of purine nucleotides and of methionine, as well as formation of the methyl group of thymine, inhibiting THF synthesis is expected to interfere with synthesis of DNA, RNA, and protein. How, then, does one explain the preferential toxicity of methotrexate toward proliferating cells? An answer came when Wahba and Friedkin (14) discovered that TS, the enzyme that converts deoxyuridine monophosphate to thymidine monophosphate, uses mTHF as both a one-carbon donor and an electron donor, resulting in transfer to the pyrimidine ring of a methylene group and its reduction to the methyl level. Hence, the flux rate through the TS reaction should be a determinant of methotrexate toxicity; the higher the rate of DNA replication, the higher the rate at which mTHF is converted to DHF, which cannot be recycled to THF and then to mTHF and must accumulate if the DHFR reaction is blocked.

However, many tissues undergo cell proliferation as part of their normal function—hair follicles, intestinal epithelium, cells of the immune system—and these cells are also sensitive to DHFR inhibition. This is the principal basis for the toxicity of methotrexate. Various means have been developed to deal with this lack of selectivity, notably providing leucovorin (5-formyl-THF) as a means of repleting THF, even when DHFR is inhibited. Although leucovorin rescue is widely used in methotrexate chemotherapy, the basis for its selective “rescue” of nontumor cells is not entirely clear (15).

Virtually all anticancer drugs suffer from the problem of resistance. Among the several processes leading to methotrexate resistance, such as mutation of the target enzyme or loss of transport into cells, one of the most vexing is a dramatic increase in DHFR levels in tumor cells. The target enzyme accumulates to an extent that its activity can be blocked only by administering lethal quantities of the drug. Robert Schimke (16) made the important discovery that this results from selective gene amplification. After prolonged exposure to the drug, the gene encoding DHFR replicates independently of whole genome replication until hundreds of gene copies can be present. More than one mechanism may be involved, as the amplified genes can be clustered together on a single giant chromosome or else, distributed in hundreds of minichromosomes. Whatever the mechanism, or mechanisms, this discovery rapidly led to findings that gene amplification occurs commonly as a response to cytotoxic agents—as with metallothionein gene amplification in response to heavy metal toxicity, for example. Also, selective gene amplification has been discovered as a mechanism in normal metazoan development, where for example, ribosomal RNA genes become amplified during oogenesis in amphibians, and also in carcinogenesis, where oncogene amplification has been observed in several instances.

I mentioned earlier that most of these pioneer drugs have found use as research reagents. Among folate analogs, aminopterin is a component of hypoxanthine-aminopterin-thymidine (HAT medium), widely used in generating monoclonal antibodies. HAT selects for cells that have active salvage pathways for nucleotide biosynthesis. Aminopterin blocks de novo synthetic pathways to purine nucleotides and deoxythymidine triphosphate (dTTP), and cells survive only if they have enzymes permitting the use of hypoxanthine for purine synthesis and of thymidine for dTTP synthesis.

Finally, I also mentioned earlier that the pioneer drugs have served as templates for development of second- and third-stage drugs. Among the many DHFR inhibitors that have found use in the clinic, perhaps the most widely used is trimethoprim, also developed by the Burroughs-Wellcome group. Trimethoprim is more distantly related to DHF than is methotrexate in the structural sense, but it is highly specific for the DHFRs of prokaryotes and lower eukaryotes and is in wide use as a broad-spectrum antibiotic.

Fluorinated pyrimidines

As suggested earlier, the folate antagonists came into existence during the 1940s, as the result of untargeted investigations, when little was known about the biochemical functions of folic acid. By contrast, FU and FUdR were synthesized rationally, as a direct result of the finding that tumor cells take up and metabolize uracil more rapidly than nontumor cells. Charles Heidelberger (17) reasoned that as the fluorine atom has a van der Waals radius close to that of hydrogen, fluorinated derivatives of uracil might also be preferentially taken up and metabolized by tumor cells to lethal products. As indicated earlier in this article, both analogs exert their principal cytotoxic effects through conversion to the corresponding deoxyribonucleoside monophosphate, FdUMP, a potent inhibitor of TS.

Examination of the TS-FdUMP interaction revealed that inhibition was irreversible but only in the presence of the folate cofactor, mTHF. This suggested that FdUMP was acting as a true mechanism-based inhibitor, with the structure of the inhibited enzyme revealing the TS reaction mechanism. Indeed, crystallization of the ternary FdUMP-mTHF-TS complex revealed a structure close to that of the putative transition state for the reaction, supporting a mechanism that begins with nucleophilic attack by a cysteine thiol on carbon-6 of the pyrimidine substrate (18).

These developments made TS an attractive target for rational drug design. Modeling the catalytic site and designing molecules to fit that site could, in principle, lead to drugs with greater target specificity than the existing fluoropyrimidines. The several compounds that emerged turned out to be analogs of folate coenzymes, not the nucleotide substrate, dUMP. One such analog is pemetrexed, which is used in treating mesothelioma and lung cancer. Although reasonably effective, pemetrexed has three target enzymes, including DHFR and TS, so it does not have the specificity envisioned at the beginning of the search.

The effectiveness of FU and FUdR as antineoplastic agents stimulated a search for other useful fluorinated pyrimidine analogs. The most effective, also synthesized by Heidelberger (17), is 5,5,5-trifluorothymidine (F3TdR). Again, because of the atomic size similarity of fluorine to hydrogen, the CF₃ group is comparable with a methyl group. F3TdR has found most of its use as an antiviral agent. It is an excellent substrate for herpes virus-coded thymidine kinase, with its lethal effects evidently occurring as a result of incorporation of the analog into viral DNA.

As mentioned earlier, the principal therapeutic target for FU and FUdR is TS. Unlike the folate analogs, FU and FUdR must undergo metabolic activation to yield the active inhibitor, FdUMP. Despite the fact that FUdR is activated in just one step, namely, phosphorylation by thymidine kinase, whereas FU has a complex metabolism, FU is in much wider use in the clinic. For both analogs, however, there are metabolic complexities that limit effectiveness and raise the possibility of additional modes of action.

Both analogs can be incorporated into DNA, as the result of phosphorylation of FdUMP to the di- and then the triphosphate. In addition, inhibition of TS causes dUMP to accumulate, and this leads to accumulation of dUTP and its incorporation into DNA. It has long been thought that this is the basis for toxicity of fluorinated pyrimidines; dUMP accumulation in DNA would stimulate the DNA-uracil base excision repair system. If repair is initiated on both DNA strands at nearby sites, the result can be dsDNA breakage, normally a lethal event. However, the fluorinated derivative of dUTP also accumulates in TS-inhibited cells, and recent evidence (19) indicates that toxicity results more directly from FdUMP incorporation into DNA than from excessive activity of the DNA-uracil repair system. There is evidence also for dUTP incorporation into DNA being a major component of the lethal action of folate antagonists as well.

Of concern, with respect to specificity, is incorporation into RNA. This is more of a concern for FU than for FUdR, as the metabolism of FU generates fluorinated ribonucleotides, as well as deoxyribonucleotides. In messenger RNA, FU can be misread as cytosine (20), leading to mRNA coding errors in translation, and this can occur at comparable rates in tumor and nontumor tissue. There is also evidence that fluoro-UMP residues in RNA interfere with normal pre-mRNA splicing patterns and tRNA base modifications.

A final source of complexity, not directly related to fluorinated pyrimidines but possibly relevant, involves regulatory properties of the TS molecule itself. It has long been known that TS regulates its own synthesis through interaction with a specific site on its own mRNA. More recently, TS has been found to regulate expression of the tumor suppressor p53 and the oncogene c-myc (21), acting similarly as a translational repressor.

araC

In the early 1950s, arabinosyl nucleosides were isolated as natural products (22). Originally called spongouridine and spongothymidine, these were found in quantity in a marine sponge, Cryptotethya crypta, and were promptly shown to be 1-β-d-arabinosyl derivatives of uracil and thymine, respectively. Although both compounds displayed little biological activity in early experiments, their presence in natural environments spurred interest in synthesizing other arabinosyl nucleosides. Two such compounds, araC and arabinosyladenine (araA), were synthesized in the late 1950s and found to have potent activities. As seen in Table 1, araC is active against several leukemias and non-Hodgkins lymphoma. araA is used in treating herpes virus infections.

Arabinose is epimeric to ribose at carbon 2. This alters the conformation of the sugar in nucleotides synthesized from either araC or araA. In fact, for both nucleosides, the active metabolite is the respective triphosphate (araCTP or araATP), formed by three sequential phosphorylations of the nucleoside. The target is DNA synthesis. Originally, it was thought that araCMP, once incorporated into DNA, acted as a chain terminator, because of misalignment of the 3′-hydroxyl group for accepting the next nucleotide. However, araCMP has been found in DNA internucleotide linkage after araC treatment of cells (23), so clearly replicative DNA polymerases must extend DNA chains from araC termini. Gel-based DNA polymerase assays using defined DNA templates and primers establish that replication is inhibited at the araCMP insertion step and at extension from an araCMP terminus (23, 24). In addition, assays in which an araCMP residue is placed in the template strand show inhibition of polymerases in bypassing this block when that strand is replicated. Hence, araC interferes with DNA synthesis by a complex mechanism involving (at least) analog incorporation, replicative chain extension, and bypass of araC in template DNA. Experiments with T7 phage RNA poymerase indicate that araCMP residues in DNA can interfere with transcriptional initiation but only when inserted near the transcriptional start point (25). Most evidence points to DNA replication as the site of action for araC.

Clinical resistance to araC results primarily from enzymatic deamination of intermediates en route to araCTP. araC is a good substrate for deoxycytidine deaminase, whereas araCMP is susceptible to deamination by dCMP deaminase. This kind of resistance is often prevented by coadministration of a deaminase inhibitor, such as tetrahydrouridine, an inhibitor of deoxycytidine deaminase.

Thiolated purines

These compounds, notably 6-MP and 6-thioguanine, were synthesized in the 1940s and 1950s by the Burroughs-Wellcome group, led by George Hitchings and Gertrude Elion. This effort, which generated analogs that are widely used as immunosuppressants and in treating leukemias, viral diseases, microbial infections, and gout, rates a brief mention in Mukherjee's book. Given the spectacular success of this effort, resulting in a share of the 1988 Nobel Prize to Elion and Hitchings, it is instructive to read Hitchings' comments, in a 1968 award address, about the opportunity that they saw (26): “To put the program into perspective, one might have a brief look at the state of the art as it was 25 yr ago. Chemotherapy was in limbo. A few empiricists were dutifully poisoning infected mice with whatever came to hand, but the main stream of scientific thought held the field as beneath notice as an area devoid of intellectual challenge.” Later, in the same address, he identified nucleic acid synthesis as a “quiet backwash where a small group might work relatively undisturbed by the pressures of intense competition.”

Among the several purine analogs synthesized by the Elion-Hitchings group, the most successful anticancer agent is 6-MP, although 6-thioguanine has uses in the clinic as well. 6-MP was approved for drug use by the U.S. Food and Drug Administration in 1953, barely 2 years after its synthesis and characterization in bacterial systems had been reported.

The basis for the selective antiproliferative effect of these purine analogs is still rather mysterious. Both compounds are excellent substrates for hypoxanthine guanine phosphoribosyltransferase (HGPRT) but less good substrates for the kinases that would phosphorylate the inosinic acid and guanylic acid analogs, resulting from HGPRT action. Hence, in the case of 6-MP, the thio analog of inosine monophosphate accumulates, and this acts as an allosteric inhibitor of purine nucleotide biosynthesis at several points—principally, the first committed step in the de novo pathway, namely, the conversion of 5-phosphoribosyl-1-pyrophosphate to phosphoribosylamine. Hence, an effect of 6-MP is to inhibit DNA and RNA synthesis, through limitation of its purine nucleotide precursors (27). However, this effect would seem to interfere with all cells, whether or not proliferating. 6-MP has been reported to be incorporated into DNA as the thiol analog of dGMP (28), which then leads to DNA strand breakage through action of the mismatch repair system. There is also a report that the 6-thio analog of dGTP interferes with telomerase activity in vitro (29). It seems likely that the selectivity of 6-MP is related to one or more of its interactions with DNA, through thiolated analogs of dGTP, but the specific steps have not yet been identified.

6-Thiopurines and in particular, 6-thioguanine, have found extensive use in eukaryotic cell genetics. HGPRT is encoded by a gene on the X chromosome, meaning that male cells have but one copy of the gene. Hence, cells can become resistant to 6-thioguanine as the result of a single mutation. This makes 6-thioguanine resistance an excellent selectable marker for determination of mutation frequencies and rates in mammalian cells.

HU

The last among our pioneer drugs, born in the 1950s and still used in the clinic, is HU, which is distinctive among our agents in not being a metabolite analog. However, it shares with the others DNA synthesis as its metabolic target. As noted earlier, HU was synthesized in 1869, but nearly a century passed before its pharmacological properties were recognized. A major use was in treating chronic myelogenous leukemia, for which it has now been largely supplanted by imatinib (Gleevec). Furthermore, this is another anticancer drug that has found unexpected uses—in this case, treatment of sickle cell disease. Its mechanism in this latter use is not well understood, but one of its effects is to stimulate synthesis of fetal hemoglobin.

As an anticancer agent, it is well known that the target for HU is RNR and that HU acts by scavenging a catalytically essential free radical (30). RNR catalyzes reduction of the four canonical ribonucleoside diphosphates—ADP, CDP, GDP, and UDP—to their respective deoxyribonucleoside diphosphates. This is the first reaction committed to DNA synthesis. Because of the need to coordinate supplies of their four dNTPs with their rates of uptake in DNA synthesis and because of the need to coordinate DNA replication with the cell cycle, RNR is heavily regulated, by both allosteric and genetic processes.

RNR was found in the 1960s to be inhibited by HU, but its target on the enzyme was not identified until the early 1980s. Electron paramagnetic resonance (EPR) spectroscopy showed that the active RNR molecule contains a free radical and that HU destroys the radical (31). The enzyme protein is an α₂β₂ tetramer. A large subunit (α₂) contains the catalytic site, and the small subunit (β₂) contains a catalytically essential free radical. A combination of EPR spectroscopy and site-directed mutagenesis identified the radical as a tyrosine radical cation (32), formed in conjunction with a binuclear iron center. A chain of electron carriers, mostly amino acid residues, carries an electron in a cysteine thiol group in the catalytic center, some 30 Å away, in the large subunit, from the tyrosyl radical, reducing it transiently to a tyrosine residue. The active site thiyl group, which has been created in the active site, then initiates the reduction of the ribonucleotide substrate (33). Treatment of the enzyme molecule with HU causes a similar one-electron reduction of the tyrosyl radical, but under these circumstances, the cysteine in the catalytic center is not activated. The identity of RNR as the target for HU is confirmed by the fact that virtually all cell lines that become HU-resistant in culture overproduce the small (radical-containing) subunit of RNR (34).

A large number of compounds, which like HU, contain a carbamate functional group, have been reported to be superior to HU as RNR inhibitors (35). However, as far as I know, none has supplanted HU in the clinic—or in the laboratory, where HU is widely used as a DNA synthesis inhibitor. Its advantage over other dNTP synthesis inhibitors is that it lowers all four dNTP pools, more or less in concert, whereas FUdR and other compounds lead to dNTP pool imbalances, which can have secondary effects. There are a few DNA polymerase inhibitors, such as aphidicolin, in fairly common laboratory use as replication inhibitors, but as they inhibit the different replicative polymerases differentially, their effects may be harder to evaluate.

Concluding remarks

As I look back over this essay, I realize that it barely begins to describe the first 65 years since the introduction of the pioneer anticancer drugs. Instead, the focus has been on the first 25 or 30 years, when the drugs were discovered, biochemical targets were identified, therapeutic efficacy was assessed along with toxicity, and pathways for metabolic activation and development of drug resistance began to be identified. The more recent history, which is foreclosed by space considerations, deals with interactions of drugs, enzymes, and metabolic pathways with newly discovered signaling systems; processes that actually kill targeted cells; development of more effective drug combinations based on increasingly sophisticated understanding of metabolism and drug effects; and structural analysis of target enzymes for discovery of new drugs. It is unlikely, were George Hitchings available to assess the current scene, that he would still consider nucleic acid synthesis a quiet backwash where a small group might work relatively undisturbed or that he would consider cancer chemotherapy beneath notice as an area devoid of intellectual challenge.