The Mechanism of Action of Anti-tumour Platinum Compounds
The Mechanism of Action of Anti-tumour Platinum Compounds
A wide range of studies carried out by cell biologists and biochemists lead to the conclusion that platinum anti-tumour compounds exert their potency by stopping the replication of DNA, thereby preventing cell division. This is most likely achieved by the formation of a complex between the bases of the DNA and the platinum compounds. The exact nature of this binding is not yet understood.
As a result of observations made by Rosenberg in the early 1960s, a range of platinum compounds have been discovered which have significant anti-tumour properties (1, 2). The two prototype compounds originally investigated are the cis isomers of diamminoplatinum(II) dichloride and diammino-platinum(IV) tetrachloride. The stercochemistries are shown in Figure 1. The trans isomers of these complexes are ineffective as anti-tumour agents. The cis -diammino-platinum(II) dichloride has now had an extensive clinical trial as an anti-tumour drug (3). The cis isomers are highly effective cell poisons which inhibit the replication of rapidly dividing cells. There is no evidence that the compounds so far examined are selective for tumour cells; indeed, many classes of rapidly growing cells are damaged by these agents. The effectiveness of killing cells depends upon the dose of drug received but not in a linear way. It has been demonstrated that at low doses some types of cell have a mechanism for recovery from the damage inflicted by the drug. In these respects the platinum compounds are not dissimilar from other anti-tumour agents such as alkylating agents or X-irradiation.
In addition to their anti-tumour properties the cis platinum compounds display a variety of other biological effects which are significant to our understanding of their mechanism of action. For example, they induce bacterial cells of certain strains of E. Coli to form long filaments (2, 4). This is apparently an inhibition of the process of cross-wall formation at the division stage of the growth cycle. Bacteriophage virus particles that attack bacterial cells are inhibited (5) and, possibly most significant of all, the cis platinum isomers are highly effective at inducing the production of phage particles from lysogenic strains of E. Coli bacteria (6).
It is always important to establish the mechanism of action of a drug in order to provide sound guidelines for the design and synthesis of more effective analogues, and also to throw light upon the toxic properties and unwanted chemical side-reactions with the expectation of thereby discovering measures to ameliorate the effects. In this article we discuss the evidence for the mechanism of action of platinum anti-tumour compounds, beginning with a summary of the types of analogue compounds now known to possess anti-tumour activity. The rational synthesis and testing of a number of analogues is essential to studies of mechanism of action. The structural requirements of a molecule for effective anti-tumour action are thereby defined and this is often the first step in understanding how a chemical acts; this can never provide a complete answer, however, for drug action can be divided into three stages. First, there must be transport of the chemical to a site of action. During this stage the chemical structure may be modified by metabolism or by change of its chemical environment. The drug may become permanently bound, rendered inoperative or excreted by the system. The second stage is attack at the site of action leading usually to the inhibition of some vital biochemical pathway. The third stage which may take place is the recovery of the organism due to its ability either to eliminate or repair damage or to develop a mechanism for by-passing the blocked pathway. After discussion of the chemistry of analogue compounds we assemble the evidence which indicates the site of attack of the platinum compounds. Finally, the nature of the binding to the target site is considered.
Analogues and their Chemistry
Much of the work in this area has concentrated upon analogues of platinum(II) compounds (7). Although some platinum(IV) complexes display activity they have been as yet relatively little studied; it has been suggested that platinum(IV) compounds may be reduced in vivo to the divalent oxidation state, but this has not been verified experimentally. Activity has also been sought among complexes of other metal ions such as palladium-(II), rhodium(III) and iridium(III) but only marginal activity has been discovered so far.
A large number of analogues of the platinum(II) compounds have now been screened and the following empirical rules for activity have emerged (7). First, the complex must be neutral. Possibly this requirement enables the drug to cross membranes by ensuring it is lipid soluble. Secondly, the molecule must contain a pair of cis leaving groups open to substitution by incoming groups. In cis -[Pt(NH3)2Cl2] the chloride ions are replaced relatively rapidly, in a low chloride medium, by nucleophiles such as H2O, ≥N, -NH2, or -S−, all of which are present in a biological milieu. However, in the presence of high concentration of chloride ion, as found in blood plasma, the replacement of chloride ion will be very slow except by the strongest nucleophiles. Thus it may be that the chloride ions remain bound to platinum(II), preventing it from binding to biological molecules during transport. Replacement will then occur only at the target where the local chloride ion concentration drops to a low value. However, in spite of this some analogues have been discovered, for example [Pt(NH3)2(methylmalonate)], which have reasonable activity but do not possess good leaving groups. Methylmalonate, being a chelated ligand, will be thermodynamically rather stable (7). In this case it could be that the organism is able to metabolise off the methylmalonate ligand.
The other cis ligands, the amines, are kinetically inert to substitution. This has been shown in an experiment using a doubly labelled radioactive analogue, namely [14C-ethylenediamine195PtCl2] (8). The intact drug could be recovered from the urine of rats and the tissue distribution of both the metal ion and the ligand was virtually identical. Thus it appears that the action of the pair of cis amine groups is to block substitution of one side of the platinum molecule. Nevertheless, the amine ligands are important in controlling activity (9). Table I lists a series of analogues in which the amine groups have been varied in a systematic way. Also shown are the results of tests with an experimental animal tumour system, the PC6 plasma cell tumour, at the Chester Beatty Research Institute. The therapeutic index of the best analogue is over thirty times greater than that of the cis -[Pt(NH3)2Cl2] towards this tumour system. But of particular interest is the fact that the anti-tumour potency remains at about 2.5 mg/kg until the cyclobutyl compound, and thereafter decreases. There is still an increase in the therapeutic index because the toxicity measured by the LD50, decreases to a greater extent. Thus there is an optimum size of the alicyclic substituent on the amine. It is possible that the effect of the amine group in these compounds is upon the transport properties of the drug thereby affecting its toxic side-reactions rather than its therapeutic efficacy.
|LD50 (mg/kg)||ID90 (mg/kg)||TI|
The ID90 is the minimum dose that causes tumour regression
The LD50 is the dose which kills 50 per cent of the animals
The ratio of LD50 and ID90 is the therapeutic index (TI)
(Data from Reference 9)
The Target Molecule
Evidence for the nature of the true target molecule is often, of necessity, rather indirect. Merely to measure the distribution of, say, a radioactively labelled platinum drug among different classes of biological molecules does not reveal which of the different sites occupied is the one leading to inhibition of cell division. This is clearly brought out by Figure 2. The extent of binding of cis -and trans -[Pt(NH3)2Cl2] to different cellular macromolecules of experimental tumour HeLa cells has been measured, using atomic absorption spectroscopy, as a function of the concentration of platinum (10). It can be seen that all cell components are bound to a degree, and, even more remarkable, that approximately twice as much of the trans isomer is bound as of the cis isomer although cell kill is only found with the cis isomer. Thus it is usually necessary to correlate the extent of observed binding with the inhibition of some biochemical pathway. It is now clear from a number of pieces of evidence that DNA replication is the first process inhibited by cis -[Pt(NH3)2Cl2] and that the drug brings this about by binding directly to DNA itself. It was possible to show that, in mammalian cells growing in culture, DNA replication was inhibited at therapeutic doses of cis -[Pt(NH3)2Cl2] before RNA synthesis or protein synthesis (11). DNA synthesis can be monitored by measuring the uptake and incorporation of one of its component molecules which has been radioactively labelled, namely tritiated thymidine, [3H]TdR. Similarly RNA and protein synthesis can be followed by measuring the uptake of radioactive components unique to them, for example, tritiated uridine [3H]UdR, and tritiated lysine. Figure 3 shows the results obtained for HeLa cells by Pascoe and Roberts. The concentration of cis -[Pt(NH3)2Cl2] is held to a level which ensures cell survival and does not result in other potentially lethal biochemical consequences. Thus inhibition of DNA synthesis is marked while RNA and protein synthesis continue unchecked. This type of study has also been carried out by Harder and Rosenberg (12) using a different species of mammalian cells in culture, and very similar results were obtained with identical conclusions being reached. A significant additional point was brought out, however, in this work. The inhibition of DNA synthesis which occurred at very low levels of cis -[Pt(NH3)2Cl2] persisted even after removal of the agent from the culture medium. The inhibition was found to be irreversible.
How is DNA synthesis inhibited so effectively by cis -[Pt(NH3)2Cl2]? Several answers are possible. The molecule may bind and inhibit an enzyme vital to the synthesis. Alternatively, the drug may bind DNA itself and interfere with synthesis. The first possibility, inhibition of an enzyme, is generally thought to be an unlikely one. The drug would have to inhibit all the enzyme present, and sufficient drug must be administered to inhibit any new enzyme manufactured by the cell. The persistent inhibition at low concentration demonstrated by Harder and Rosenberg argues against this mechanism. Indeed, these workers looked for but were unable to detect any inhibition of one of the enzymes involved in DNA synthesis, namely a DNA polymerase.
This leaves the second possibility, that cis -[Pt(NH3)2Cl2] irreversibly binds DNA itself in such a way as to inhibit its replication. Certainly there is now plenty of direct evidence for the binding of the drug directly to DNA. But in addition there is a mass of indirect evidence of the type referred to earlier which supports the idea that the attack of cis -[Pt(NH3)2Cl2] on DNA leads to significant biochemical consequences.
The early observation of Rosenberg and his co-workers (2, 4) that cis -[Pt(NH3)2Cl2] induces filamentous growth in cells of the bacterium E. Coli is probably an indication of the ability of an agent to react with DNA leading to inhibition of DNA synthesis with no accompanying stopping of RNA or protein synthesis. Filamentous growth occurs in response to a variety of agents, for example, U.V. and X-irradiation or alkylating agents, all of which damage DNA. Such agents have been called “radiomimetic” and often, as a result of this attack on the DNA of a cell, they evoke a range of common biological and biochemical effects (13).
Further important evidence of direct attack on DNA is provided by the results of Reslová (6) who investigated the ability of platinum compounds to induce the growth of phage from lysogenic strains of E. Coli bacteria. Phage are bacterial viruses, Figure 4, which cannot reproduce themselves alone but need to take over the replication apparatus of a bacterium. By injecting into it a piece of DNA, or RNA, they instruct the bacterial cell to replicate phage particles rather than daughter bacterial cells. Figure 5 shows the life cycle of a lysogenic bacterial virus (14). After it has injected its chromosomal material into the host bacterial cell the phage may multiply immediately, eventually causing the rupture (lysis) of the cell and the release of many new phage particles. This is the lytic cycle. However, sometimes the genetic material of the phage becomes part of the circular strand of DNA which is the host cell’s genetic material. In this case the phage becomes dormant. Its DNA is duplicated along with the bacterial DNA when the bacterium divides. The bacterium is then said to be lysogenic. The release of the phage DNA to direct synthesis of new phage particulars is normally a very rare event, say 1 in 10,000 divisions of the lysogenic bacterium. However, radiomimetic agents can cause the phage DNA to be released, and phage particles to be synthesised, with consequent cell lysis, a readily observed process. Reslová (6) was able to show that there exists an excellent correlation between the anti-tumour activity of platinum compounds and their ability to induce lysogenic E. Coli to enter the lytic cycle. This is striking evidence of a correlation between the anti-tumour activity of platinum compounds in higher cells and their ability to bind DNA and induce phage from simple bacterial cells. The correlation is better than that between the ability of platinum compounds to induce filamentous growth in bacteria and their anti-tumour activity.
Thus the weight of evidence now points to the direct binding of cis -[Pt(NH3)2Cl2] to DNA and the subsequent inhibition of DNA replication as the important primary biochemical event leading to anti-tumour activity.
Mode of Binding to the Target Molecule
We have seen that although both cis - and trans -[Pt(NH3)2Cl2] bind to DNA, only the cis isomer inhibits DNA replication and leads to radiomimetic effects. Thus the precise mode of binding is of paramount importance to elicit anti-tumour activity.
The chemistry of the two isomers gives a clue to the possible nature of the binding to DNA that is required. Both isomers could undertake mono-functional binding by loss of one chloride ion and attachment to a single incoming group. However, it is difficult to see how this would lead to differences between the biochemical effects they each cause. On the other hand, both isomers are capable of bi-functional binding by loss of two chloride ligands each and their requirements for the stereochemical disposition of the incoming groups are very different. The cis isomer requires two groups about 3.0 Å apart on the same side of the molecule whereas the trans isomers can only bind two groups about 5 Å apart approaching from opposite sides of the molecule. Therefore there must be a high degree of selectivity for binding sites on DNA by these two molecules.
There are three different chemical groupings in the structure of DNA, namely the backbone phosphates, the sugar residues, and the organic bases, purines and pyrimidines. The ultra-violet spectrum of DNA after reaction with either the cis or trans isomer of [Pt(NH3)2Cl2] shows marked changes, shown in Figure 6 (15). This is conclusive evidence that the platinum compounds bind to the organic bases since the ultra-violet spectrum is due entirely to electronic excitation within these groups. However, it does not eliminate the possibility of there also being binding to the sugar and phosphate moiety. Sugar platinum complexes are not known and this mode of binding seems unlikely. Phosphate complexes of platinum have been described recently but they are not of high stability and are rather labile (16). Thus it is unlikely that binding to phosphate groups will produce the irreversible and persistent inhibition observed.
A number of studies have been directed towards understanding the ways in which the cis and trans isomers bind to the purine and pyrimidine bases. A spectrophotometric study indicated that the three bases adenine (A), guanine (G), and cytidine (C) will bind both isomers, whereas the fourth base present in DNA, thymine (T), will only do so very slowly at room temperature, generating a blue complex that is readily visible if formed in DNA (17, 18). Kinetic studies reveal that the rate of reaction with guanosine has a half-life of 6.7 h for cis -[Pt(NH3)2Cl2], whereas reaction rates with other bases are at least twice as slow. Adenosine-5-mono-phosphate is, however, an interesting exception reacting with a half-life of 2 hours although adenosine itself is slow to react (19). This pointed to guanosine as the base in DNA most likely to bind platinum preferentially. This has been confirmed by measurements of the extent of binding of cis -[Pt(NH3)2Cl2] to DNA with varying GC/AT ratios. Stone, Kilman and Sinex monitored the extent of binding by measuring the buoyant density of DNA (20); the density increases on binding of platinum. As the GC content of the DNA species increased, so the extent of binding increased. This has been confirmed by Munchausen and Rahn (21), who monitored the degree of binding using 195Pt, a radioisotope. It was also observed that complexes between cis -[Pt(NH3)2Cl2] and adenosine, or guanosine are stable to mild acid hydrolysis, conditions which completely depurinate DNA. They were able to extract platinum-purine complexes from DNA in this way and to analyse them for the amount of guanosine and adenosine. Apparently the cis isomer binds preferentially to guanosine, although there is a second slower reaction with adenosine. This is a remarkable parallel with the alkylating agents, which produce a preponderance of alkylations at guanosine in the 7 positions (13). Figure 7 shows the three bases that are known to bind cis - and trans -[Pt(NH3)2Cl2].
However, this body of evidence still does not reveal a clear-cut difference between the mode of binding of the cis and trans isomer nor does it reveal the reason why binding to guanosine should be so lethal in the case of the cis isomer. There are several modes of binding of a bi-functional reagent to DNA which might occur. Binding of two bases on opposite strands of DNA by one molecule of platinum would constitute an inter-strand cross-link. Presumably this would prevent separation of the two strands and inhibit DNA replication. Alternatively the linkage could occur between two adjacent bases on a single strand, forming an intra-strand crosslink. Finally, bidentate chelation to one base, for example, between the oxygen atom at position 6 and the N-7 position of guanosine, might take place.
Pascoe and Roberts (11) have measured the degree of inter-strand cross-linking of HeLa cell DNA in vivo and in vitro by both cis - and trans -[Pt(NH3)2Cl2]. DNA was grown in such a way that one strand of the double helix incorporated a heavy radioactive label, namely 5-bromo-2′-deoxyuridine (BUdR). Centrifugation of the DNA on an alkaline caesium chloride gradient led to the separation of a light strand and a heavy radioactive strand. If the strands have been cross-linked so that they can no longer be separated then a radioactive hybrid of intermediate density is obtained (Figure 8). In this way the degree of inter-strand cross-linking can be monitored.
Intra-strand cross-links will not be detected in this way. The results obtained are shown in Table II. There is a striking difference between the extent of cross-linking for equivalent doses of the cis- and trans -[Pt(II)(NH3)2Cl2] and also between DNA when treated outside the cell, in vitro, and when treated within the cell, in vivo. Combining these data with measurements of the total amount of platinum bound per molecule of DNA it can be shown that only 1 in 400 DNA reactions of cis -[Pt(NH3)2Cl2] result in inter-strand cross-links, with a parallel value of 1 in 400 reactions for the trans isomer. This contrasts with 1 in 8 reactions of sulphur mustard with DNA in vivo which generate inter-strand cross-links. This casts doubt upon the idea that inter-strand crosslinks are the major lesions leading to cell death. However, the number of inter-strand cross-links which must be put into HeLa cell DNA in order to kill a given fraction of the cell population is of the same order for both isomers, being about 350 for the cis and 200 for the trans isomer. Therefore, even if inter-strand cross-linking is not the most common lesion in DNA it appears to parallel the major lesion responsible for cell death. Presumably one of the other modes of bi-functional attack is the significant antitumour lesion.
|Compound||Dose required to produce 10 per cent DNA inter-strand cross-linking (μM)|
|in vitro||in vivo|
It is, even so, of some interest to consider the likely nature of these cross-links. Since the leaving groups of the cis isomer are only about 3 Å apart the platinum ion must bind two groups on opposite strands of DNA with approximately this separation. However, studies of the binding of cis -[Pt(NH3)2Cl2] to guanosine, have shown that the N-7 position of guanosine is a favoured site. The N-3 position of cytosine is the preferred one and on adenosine N-3, and N-7 can be occupied. There is also indirect evidence for the involvement of the 6-amino group in binding platinum ion (17). No involvement of the 6-oxygen group in guanosine has been detected. The only way in which inter-strand cross-linking could occur utilising any of these binding sites and without major disruption of the helical structure of DNA would be for a cis platinum complex to bridge two 6-amino groups of adenosine groups on opposite chains. Figure 9 is a photograph of a DNA model showing that the 6-amino groups are vertically above one another and separated by 3.5 Å. Evidence has been obtained for the ability of cis -[Pt(NH3)2Cl2] to link two adenosine groups together in this way in a simple dinucleotide, the dimer, adenosine-sugar-phosphate-sugar-adenosine (22).
Model of DNA showing the relative disposition of the 6-amino groups of two adenine bases on opposite strands of DNA
Photograph by courtesy of the Chester Beatty Research Institute
A recent X-ray structure (23) of a complex formed between cis -[Pt(NH3)2I2] and inosine-5′-phosphate, an analogue of guanosine lacking the 2-amino groups, confirms the early spectrophotometric work showing binding at the N-7 position (Figure 10). Such a binding cannot lead to an inter-strand crosslink without the unwinding of a short section of the DNA helix because the N-7 groups of guanosine residues on opposite strands are over 8 Å apart with bulky groups between (Figure 11). This might better be regarded as a model of a possible intra-strand link. No wholly satisfactory methods of estimating the number of intra-strand links or of bi-functional binding to a single base have been devised. Recently, Roos (24) has studied the competitive inhibition of the intercalation of DNA by the dye, 9-amino-acridine, with cis -[Pt(NH3)2Cl2]. Acridine dyes are large flat aromatic molecules which bind to DNA by sliding between adjacent Watson-Crick base pairs. Roos postulates that, if platinum makes a link between two bases vertically above one another, intercalation of the dye will be prevented. He is able to show an inhibition of the binding of acridine dye after attack of cis -[(14C-ethylenediamine)PtCl2] on DNA. This must be some form of cross-linking between bases from separate rungs of DNA. Since the work of Pascoe and Roberts (11) shows that inter-strand links have such a low frequency, Roos’ results point to the presence of intra-strand links. Although he has not examined the effect of a trans isomer on the intercalation of a dye, it is difficult to envisage a bridge of the same type being possible.
The third mode of bi-functional attack, binding to a single purine molecule between, say, 6-0 and N-7 of guanosine, has not been identified in reactions between the cis isomer and purines. Conceivably it could occur in guanosine if the proton were simultaneously displaced from the N-1 position. But as this proton is involved in the H-bonding of a base-pair in DNA itself, it could be difficult to remove. Thus this mode of binding remains to be detected.
There is now a considerable body of evidence suggesting that the anti-tumour property of cis -[Pt(NH3)2Cl2] arises from its ability to bind to DNA in a way which inhibits DNA synthesis. The predominant mode of binding which brings this about is probably a bi-functional reaction of the platinum complex, after loss of two chloride ions, at the N-7 positions on two adjacent guanine bases. Inspection of models of DNA requires this to be a link between two bases on one strand. However, there is no indication why this mode of binding should be so effective at blocking DNA replication. It is presumed that the trans isomer is unable to form this type of intra-strand link.
We have not, in this article, touched upon the problem of whether this type of platinum link in DNA can be removed by a cellular repair process. Some cells are known to be able to remove damage inflicted by other radiomimetic agents. This is an area of high interest currently and may well be important for understanding further aspects of drug action. The fact that cells can acquire resistance to drugs is an important limitation on their clinical usefulness. It is now necessary to ask whether cells develop this resistance by use of a repair mechanism.
B. Rosenberg, L. Van Camp, J. E. Trosko and V. H. Mansour, Nature, 1969, 222, (5191), 385
B. Rosenberg and L. Van Camp and T. Krigas, Nature, 1965, 205, (4972), 698
For a full account of the clinical work see “Recent Results in Cancer Research”, Vol. 48, ed. T. A. Connors and J. J. Roberts, Springer-Verlag, Berlin, 1974
B. Rosenberg and L. Camp. Van, E. B. Grimley and A. J. Thomson, J. Biol. Chem., 1967, 242, (6), 1347
K. V. Shooter, R. Howse and R. K. Merrifield and A. B. Robins, Chem.-Biol. Interact., 1972, 5, (5), 289
S. Reslov á, Chem.-Biol. Interact., 1971 – 2, 4, (1), 66
For a recent review see M. J. Cleare and J. D. Hoeschele, Platinum Metals Rev., 1973, 17, (1), 2
D. M. Taylor and J. D. Jones and A. B. Robins, Biochem. Pharmacol., 1973, 22, (7), 833
T. A. Connors, M. Jones, W. C. J. Ross, P. D. Braddock and A. R. Khokhar and M. L. Tobe, Chem.-Biol. Interact., 1972, 5, (6), 415
J. J. Roberts and J. M. Pascoe, Nature, 1972, 235, (5336), 282
J. M. Pascoe and J. J. Roberts, Biochem. Pharmacol, 1974, 23, (9), 1345 and 1359
H. C. Harder and B. Rosenberg, Int. J. Cancer, 1970, 6, (2), 207
See A. Loveless, “Genetic and Allied Effects of Alkylating Agents”, Butterworths, London, 1966
See J. D. Watson, “Molecular Biology of the Gene”, W. A. Benjamin, Inc., New York, 1970
P. Horáček and J. Drobník, Biochim. Biophys. Acta, 1971, 254, (2), 341; S. Mansy, Ph.D. Thesis, Michigan State University, 1972
J. A. Stanko, reported in Ref. 3
S. Mansy and B. Rosenberg and A. J. Thomson, J. Am. Chem. Soc., 1973, 95, (5), 1633
I. A. G. Roos and A. J. Thomson, reported in Ref. 3
A. B. Robins, Chem.-Biol. Interact., 1973, 6, (1), 35, and 1973, 7, (1), 11
P. J. Stone and A. D. Kelman and F. M. Sinex, Nature, 25 Oct., 1974, 251, 736
L. Munchausen and R. A. Rahn, submitted to Biochim. Biophys. Acta
I. A. G. Roos and A. J. Thomson and S. Mansy, J. Am. Chem. Soc., 1974, 96, (20), 6484
D. M. L. Goodgame, I. Jeeves and F. L. Phillips and A. C. Skapski, Biochim. Biophys. Acta, 1975, 378, (1), 153
I. A. G. Roos . To be published in proceedings of conference “Metals in Medicine”, Sydney, August, 1975
The author has benefited over a number of years from discussions with Dr J. J. Roberts. Rustenburg Platinum Mines Limited have provided generous financial support for the author’s own work. Acknowledgement is also made of the skill of Mr Moreman of the Chester Beatty Research Institute in taking the colour photographs of the DNA model.