Significant differences in biological parameters between prodrugs cleavable by carboxypeptidase G2 that generate 3,5-difluoro-phenol and -aniline nitrogen mustards in gene-directed enzyme prodrug therapy systems.

Nine new nitrogen mustard compounds derived from 2,6-difluoro-4-hydroxy- (3a-e) and 2,6-difluoro-4-amino- (4a-d) aniline were synthesized as potential prodrugs. They were designed to be activated to their corresponding 3,5-difluorophenol and -aniline (4)-nitrogen mustards by the enzyme carboxypeptidase G2 (CPG2) in gene-directed enzyme prodrug therapy (GDEPT) models. The compounds were tested for cytotoxicity in the MDA MB-361 breast adenocarcinoma. The cell line was engineered to express stably either CPG2 tethered to the cell surface stCPG2-(Q)3 or beta-galactosidase (beta-Gal) as control. The cytotoxicity differentials were calculated between CPG 2-expressing and -nonexpressing cells and yielded different results for the two series of prodrugs despite their structural similarities. While the phenol compounds are ineffective as prodrugs, their aniline counterparts exhibit outstanding activity in the tumor cell lines expressing CPG2. [3,5-Difluoro-4-[bis(2-chloroethyl)amino]phenyl]carbamoyl-l-glutamic acid gave a differential of >227 in MDA MB361 cells as compared with 19 exhibited by 4-[(2-chloroethyl)(2-mesyloxyethyl)amino]benzoyl-l-glutamic acid, 1a, which has been in clinical trials.

A novel vascular endothelial growth factor-directed therapy that selectively activates cytotoxic prodrugs.

We have generated fusion proteins between vascular endothelial growth factor (VEGF) and the bacterial enzyme carboxypeptidase G2 (CPG2) that can activate the prodrug 4-[(2-chloroethyl)(2-mesyloxyethyl)amino]benzoyl-L-glutamic acid (CMDA). Three asparagine residues of CPG2 were mutated to glutamine (CPG2(Q)3) to prevent glycosylation during secretion, and truncations of VEGF(165) were fused to either the C- or N-terminal of CPG2. The K(m) of the fusion proteins (37.5 microM) was similar to that of secreted CPG2(Q)3 (29.5 microM) but greater than that of wild-type CPG2 (8 microM). The affinity of the fusion proteins for VEGF receptor-2 (VEGFR2) (K(d)=0.5-1.1 nM) was similar to that of [(125)I]VEGF (K(d)=0.5 nM) (ELISA) or slightly higher (K(d)=1.3-9.6 nM) (competitive RIA). One protein, VEGF(115)-CPG2(Q)3-H(6), possessed 140% of the enzymic activity of secreted CPG2(Q)3, and had a faster half-maximal binding time for VEGFR2 (77 s), than the other candidates (330 s). In vitro, VEGF(115)-CPG2(Q)3-H(6) targeted CMDA cytotoxicity only towards VEGFR-expressing cells. The plasma half-life of VEGF(115)-CPG2(Q)3-H(6) in vivo was 3 h, comparable to equivalent values observed in ADEPT. We conclude that enzyme prodrug therapy using VEGF as a targeting moiety represents a promising novel antitumour therapy, with VEGF(115)-CPG2(Q)3-H(6) being a lead candidate.

Appropriate subcellular localisation of prodrug-activating enzymes has important consequences for suicide gene therapy.

Escherichia coli B nitroreductase (NR) has been expressed stably in MDA-MB-361 human breast adenocarcinoma cells either as the wild-type protein (wtNR), which is distributed evenly between the cytoplasmic and nuclear compartments, or targeted to the mitochondrion (mtNR). Whereas bacterial NR is active as a dimer, a proportion of wtNR is monomeric. In contrast, mtNR is mostly dimeric, suggesting that it adopts a more stable, native conformation. Despite this, when tested in gene-directed enzyme prodrug therapy cell cytotoxicity studies, cells expressing wtNR or mtNR had similar sensitivity to the prodrug CB1954 and mounted similar bystander killing effects. Furthermore, when short prodrug exposures were given, wtNR was more efficient at killing cells than mtNR. These data demonstrate that the site of enzyme expression and prodrug activation is an important variable that requires consideration in suicide gene therapy approaches.

In suicide gene therapy, the site of subcellular localization of the activating enzyme is more important than the rate at which it activates prodrug.

The bacterial enzyme carboxypeptidase G2 (CPG2) can be expressed both intracellularly (CPG2*) or tethered to the outer surface (stCPG2(Q)3) of mammalian cells, where it is able to activate mustard prodrugs for use in suicide gene therapy protocols. Here we compare the properties of CPG2 expressed in these two locations. CPG2 is active as a dimer, and one of the mutations required to block glycosylation of stCPG2(Q)3 destabilizes the dimers. Some of the mutations to this site partially correct the dimerization defect and recover a proportion of the activity. Surface tethering also recovers some enzyme activity, but through an unknown mechanism. The efficacy of CPG2 in these two locations is compared with the tumor cell lines A2780, SK-OV-3, and WiDr, which are sensitized to the prodrug 4-([2-chloroethyl][2-mesyloxyethyl]amino)benzoyl-L-glutamic acid (CMDA) by both CPG2* and stCPG2(Q)3 expression in suicide gene therapy protocols in vitro. We find that stCPG2(Q)3 is a more efficient mediator of CMDA-dependent cell killing than CPG2*. Lower levels of stCPG2(Q)3 activity are required to give cell killing that can only be achieved by higher levels of CPG2*. In bystander effect assays, low levels of stCPG2(Q)3 are required for efficient killing, whereas relatively high levels of CPG2* activity are required. Also, shorter exposures to prodrug are required for cell killing when stCPG2(Q)3 is expressed compared with when CPG2* is expressed. These data demonstrate that the location of the enzyme in the cell is more important than the enzyme activity as the determinant in mediating cytotoxicity.

Regressions of established breast carcinoma xenografts by carboxypeptidase G2 suicide gene therapy and the prodrug CMDA are due to a bystander effect.

The role of the bystander effect in the treatment of a human breast carcinoma xenograft was studied by suicide gene therapy with carboxypeptidase G2 (CPG2) and CMDA. Cells expressing enzymatically active surface-tethered bacterial CPG2 [stCPG2(Q)3] were mixed with control beta-galactosidase (beta-Gal)-expressing cells to give stCPG2(Q)3:beta-Gal ratios of, respectively: group 1, 0:100; group 2, 10:90; group 3, 50:50; and group 4, 100:0. Four days after injection of the cells into nude mice, the prodrug 4-[(2-chloroethyl)(2-mesyloxyethyl)amino]benzoyl-L-glutamic acid (CMDA) was administered. Tumor growth delay correlated well with the levels of stCPG2(Q)3 expression: group 1, 0 day delay; group 2, 10 days; group 3, 16 days; and group 4, 90 days. Similarly, the number of cures was strongly correlated to the levels of stCPG2(Q)3 activity: group 1, zero of six cured; group 2, one of six cured; group 3, three of six cured and group 4, four of six cured. There was a good correlation between CPG2 enzyme activity in the tumors and the number of cures. The majority of cells from groups 2 and 3 were apoptotic whereas those from group 1 were not, indicating a substantial bystander effect in the tumors. These results suggest that a bystander effect plays a major role in suicide gene therapy regimens with stCPG2(Q)3 and CMDA.

Self-immolative anthracycline prodrugs for suicide gene therapy.

Four novel potential prodrugs derived from daunorubicin (8, 10) and doxorubicin (12, 14) were designed and synthesized. They are self-immolative prodrugs for suicide gene therapy activation by the enzyme carboxypeptidase G2 (CPG2) subsequently releasing the corresponding anthracyclines, by a 1,6-elimination mechanism. A mammary carcinoma cell line (MDA MB 361) was engineered to express CPG2 intracellularly (CPG2) or extracellularly, tethered to the outer cell membrane (stCPG2(Q)3). The prodrugs derived from doxorubicin showed prodrug/drug cytotoxicity differentials of 21-fold (compound 12) and 23-fold (compound 14). Prodrug 12 underwent an 11-fold activation when assayed in the cell line expressing externally surface-tethered CPG2.

Prodrugs for antibody- and gene-directed enzyme prodrug therapies (ADEPT and GDEPT).

Antibody- and gene-directed enzyme prodrug therapy are two-step targeting strategies designed to improve the selectivity of antitumour agents. The approaches are based on the activation of specially designed prodrugs by antibody-enzyme conjugates targeted to tumour-associated antigens (ADEPT) or by enzymes expressed by exogenous genes in tumour cells (GDEPT). Herein the design, synthesis, physico-chemical and biological properties, kinetics and clinical trials of the prodrugs and the enzymes carboxypeptidase G2 and nitroreductase are reviewed for ADEPT and GDEPT.

Self-immolative nitrogen mustard prodrugs for suicide gene therapy.

Four new potential self-immolative prodrugs derived from phenol and aniline nitrogen mustards, four model compounds derived from their corresponding fluoroethyl analogues and two new self-immolative linkers were designed and synthesized for use in the suicide gene therapy termed GDEPT (gene-directed enzyme prodrug therapy). The self-immolative prodrugs were designed to be activated by the enzyme carboxypeptidase G2 (CPG2) releasing an active drug by a 1, 6-elimination mechanism via an unstable intermediate. Thus, N-[(4-¿[4-(bis¿2-chloroethyl¿amino)phenoxycarbonyloxy]methyl¿pheny l)c arbamoyl]-L-glutamic acid (23), N-[(4-¿[4-(bis¿2-chloroethyl¿amino)phenoxycarbonyloxy]methyl¿pheno xy) carbonyl]-L-glutamic acid (30), N-[(4-¿[N-(4-¿bis[2-chloroethyl]amino¿phenyl)carbamoyloxy]methyl¿+ ++phen oxy)carbonyl]-L-glutamic acid (37), and N-[(4-¿[N-(4-¿bis[2-chloroethyl]amino¿phenyl)carbamoyloxy]methyl¿+ ++phen yl)carbamoyl]-L-glutamic acid (40) were synthesized. They are bifunctional alkylating agents in which the activating effects of the phenolic hydroxyl or amino functions are masked through an oxycarbonyl or a carbamoyl bond to a benzylic spacer which is itself linked to a glutamic acid by an oxycarbonyl or a carbamoyl bond. The corresponding fluoroethyl compounds 25, 32, 42, and 44 were also synthesized. The rationale was to obtain model compounds with greatly reduced alkylating abilities that would be much less reactive with nucleophiles compared to the corresponding chloroethyl derivatives. This enabled studies of these model compounds as substrates for CPG2, without incurring the rapid and complicated decomposition pathways of the chloroethyl derivatives. The prodrugs were designed to be activated to their corresponding phenol and aniline nitrogen mustard drugs by CPG2 for use in GDEPT. The synthesis of the analogous novel parent drugs (21b, 51) is also described. A colorectal cell line was engineered to express CPG2 tethered to the outer cell surface. The phenylenediamine compounds were found to behave as prodrugs, yielding IC50 prodrug/IC50 drug ratios between 20- and 33-fold (for 37 and 40) and differentials of 12-14-fold between CPG2-expressing and control LacZ-expressing clones. The drugs released are up to 70-fold more potent than 4-[(2-chloroethyl)(2-mesyloxyethyl)amino]benzoic acid that results from the prodrug 4-[(2-chloroethyl)(2-mesyloxyethyl)amino]benzoyl-L-glutamic acid (CMDA) which has been used previously for GDEPT. These data demonstrate the viability of this strategy and indicate that self-immolative prodrugs can be synthesized to release potent mustard drugs selectively by cells expressing CPG2 tethered to the cell surface in GDEPT.

Gene-directed enzyme prodrug therapy: quantitative bystander cytotoxicity and DNA damage induced by CB1954 in cells expressing bacterial nitroreductase.

Clones of human colon carcinoma (WiDr), ovarian carcinoma (SK-OV-3), and Chinese hamster V79 cells expressing the nitroreductase enzyme (NR) from E. coli B were 52-, 225- and 177-fold respectively more sensitive to a 24-h incubation with the prodrug 5-(aziridin-1-yl)-2,4-dinitrobenzamide (CB1954) than the parent lines. The IC50s of non-NR-expressing bystander cells were measured in the presence of differing proportions of NR-expressing cells. The shift in IC50 was used to calculate a value for the bystander effect, termed the transmission efficiency (TE), which is the decrease in IC50 due to bystander effect as a percentage of the maximum decrease possible. The percentage of NR-expressing cells for which the TE was 50%, (the TE50) is a single datum of bystander efficacy. WiDr and V79 cell lines, had a similar TE50 of approximately 2%. SK-OV-3 gave a lower value of 0.3%. These TE50 correlate with concentrations of cytosolic NR activity, which is distinguished from endogenous DT diaphorase activity by kinetic differences. A novel method is described which enables both DNA crosslinks and drug-induced single-strand breaks to be simultaneously quantified in a sedimentation assay. Using this technique, bystander DNA damage was demonstrated in V79 cells, of approximately 50% of that in activator cells.

Sensitization of colorectal and pancreatic cancer cell lines to the prodrug 5-(aziridin-1-yl)-2,4-dinitrobenzamide (CB1954) by retroviral transduction and expression of the E. coli nitroreductase gene.

Expression of genes encoding prodrug-activating enzymes can increase the susceptibility of tumor cells to prodrugs, and may ultimately achieve a better therapeutic index than conventional chemotherapy. CB1954 is a weak, monofunctional alkylating agent which can be activated by Escherichia coli nitroreductase to a potent dysfunctional alkylating agent which crosslinks DNA. We have inserted the nitroreductase gene into an LNCX-based retroviral vector, to allow efficient gene transfer and expression in colorectal (LS174T) and pancreatic (SUIT2, BxPC3, and AsPC1) cancer cell lines. A clone of LS174T cells expressing nitroreductase showed > 50-fold increased sensitivity to CB1954, and nitroreductase-expressing clones of pancreatic tumor lines were up to approximately 500-fold (SUIT2) more sensitive than parental cells. Concentrations of CB1954 minimally toxic to nontransduced cells achieved 100% cell death in a 50:50 mix of parental cells with SUIT2 cells expressing nitroreductase; and marked "bystander" cell killing was seen with just 10% of cells expressing nitroreductase. Significant bystander cell killing was dependent on a high cell density. In conjunction with regional delivery of vectors and tumor selectivity of cell entry and/or gene expression, nitroreductase and CB1954 may be an attractive combination for prodrug-activating enzyme gene therapy of colorectal and pancreatic cancer.

Mustard prodrugs for activation by Escherichia coli nitroreductase in gene-directed enzyme prodrug therapy.

Twenty nitrogen mustard analogues derived from 5-(aziridin-1-yl)-2,4-dinitrobenzamide (CB 1954, 1) were evaluated as candidate prodrugs for gene-directed enzyme prodrug therapy (GDEPT) in Chinese hamster V79 cell lines engineered to express Escherichia coli nitroreductase (NR). Structural variations within the series included the use of N-dihydroxypropyl and (N-dimethylamino)ethyl carboxamide side chains, the use of chloro, bromo, mesyl, and iodo leaving groups on the mustards, and regioisomeric changes. The compounds were assayed for cytotoxicity (IC50) with the NR-expressing and controls of non-NR-expressing cell lines. The proportion of NR-expressing cells required in a mixture for nonexpressing cells to experience 50% of their cytotoxicity (termed the TE50) was used to assess the compounds' ability to induce a bystander effect. This study suggests that 5-[N,N-bis(2-bromoethyl)amino]-2,4-dinitrobenzamide (8), 5-[N,N-bis(2-iodoethyl)amino]-2,4-dinitrobenzamide (9), 2-[N,N-bis(2-bromoethyl)-amino]-3,5-dinitrobenzamide (13), and 2-[N,N-bis(2-iodoethyl)amino]-3,5-dinitrobenzamide (14) showed considerable improvements over 1, exhibiting greater potency, higher IC50 ratios, and lower TE50s, and are thus superior prodrugs to 1 for GDEPT.

Involvement of DT-diaphorase (EC 1.6.99.2) in the DNA cross-linking and sequence selectivity of the bioreductive anti-tumour agent EO9.

The chemistry of the mitomycin C-related drug indoloquinone EO9 would suggest that its mechanism of action is likely to involve DNA damage after reductive activation. The ability of this agent to induce DNA damage in intact cells has been examined using alkaline filter elution. After treatment with pharmacologically relevant concentrations of EO9, both DNA strand breaks and interstrand cross-links were detected in rat Walker tumour cells and human HT29 colon carcinoma cells. These cell lines express relatively high levels of DT-diaphorase (NAD(P)H: quinone acceptor oxidoreductase), which is believed to be involved in EO9 activation. The extent of DNA damage was increased by approximately 30-fold under hypoxia in BE colon carcinoma cells that express non-functional DT-diaphorase, but this dramatic hypoxia enhancement was not seen in HT-29 cells. These data are consistent with cytotoxicity studies that indicate that DT-diaphorase appears to be important in EO9 activation under aerobic conditions, but other enzymes may be more relevant under hypoxia. The involvement of DT-diaphorase in DNA damage induction was further investigated using cell-free assays. DNA cross-links were detectable in plasmid DNA co-incubated with EO9, cofactor and DT-diaphorase but not in the absence of this enzyme. In contrast, using a Taq polymerase stop assay, monofunctional alkylation was detected in plasmid DNA without metabolic activation, although the sequence selectivity was altered after reduction catalysed by DT-diaphorase.

Bioactivation of dinitrobenzamide mustards by an E. coli B nitroreductase.

A nitroreductase isolated and purified from Escherichia coli B has been demonstrated to have potential applications in ADEPT (antibody-directed enzyme prodrug therapy) by its ability in vitro to reduce dinitrobenzamides (e.g. 5-aziridinyl 2,4-dinitrobenzamide, CB 1954 and its bischloroethylamino analogue, SN 23862) to form cytotoxic derivatives. In contrast to CB 1954, in which either nitro group is reducible to the corresponding hydroxylamine, SN 23862 is reduced by the nitroreductase to form only the 2-hydroxylamine. This hydroxylamine can react with S-acetylthiocholine to form a species capable of producing interstrand crosslinks in naked DNA. In terms of ADEPT, SN 23862 has a potential advantage over CB 1954 in that it is not reduced by mammalian DT diaphorases. Therefore, a series of compounds related to SN 23862 has been synthesized, and evaluated as potential prodrugs both by determination of kinetic parameters and by ratio of IC50 against UV4 cells when incubated in the presence of prodrug, with and without the E. coli enzyme and cofactor (NADH). Results from the two studies were generally in good agreement in that compounds showing no increase in cytotoxicity in presence of enzyme and cofactor were not substrates for the enzyme. None of the analogues were activated by DT diaphorase isolated from Walker 256 carcinoma cells. For those compounds which were substrates for the E. coli nitroreductase, there was a positive correlation between kcat and IC50 ratio. Two compounds showed advantageous properties: SN 25261 (with a dihydroxypropylcarboxamide ring substituent) which has a more than 10-fold greater aqueous solubility than SN 23862 whilst retaining similar kinetic characteristics and cytotoxic potency; and SN 25084, where a change in the position of the carboxamide group relative to the mustard resulted in an increased cytotoxicity ratio and kcat compared with SN 23862 (IC50 ratios 214 and 135; kcat values of 75 and 26.4 sec-1, respectively). An analogue (SN 25507) incorporating both these structural changes had an enhanced kcat of 576 sec-1. This study elucidates some of the structural requirements of the enzyme and aids identification of further directions in the search for suitable prodrugs for an ADEPT nitroreductase system.

Virtual cofactors for an Escherichia coli nitroreductase enzyme: relevance to reductively activated prodrugs in antibody directed enzyme prodrug therapy (ADEPT).

A nitroreductase enzyme has been isolated from Escherichia coli that has the unusual property of being equally capable of using either NADH or NADPH as a cofactor for the reduction of its substrates which include menadione as well as 5-(aziridin-1-yl)-2,4-dinitrobenzamide (CB 1954). This property is shared with the mammalian enzyme, DT diaphorase. The nitroreductase can, like DT diaphorase, also use simple reduced pyridinium compounds as virtual cofactors. The intact NAD(P)H molecule is not required and the simplest quaternary (and therefore reducible) derivative of nicotinamide, 1-methylnicotinamide (reduced), is as effective as NAD(P)H in its ability to act as an electron donor for the nitroreductase. The structure-activity relationship is not identical to that of DT diaphorase and nicotinic acid riboside (reduced) is selective, being active only for the nitroreductase. Irrespective of the virtual cofactor used, the nitroreductase formed the same reduction products of CB 1954 (the 2- and 4-hydroxylamino derivatives in equal proportions). Nicotinic acid riboside (reduced), unlike NADH, was stable to metabolism by serum enzymes and had a plasma half-life of seven minutes in the mouse after an i.v. bolus administration. NADH had an unmeasurably short half-life. Nicotinic acid riboside (reduced) could also be produced in vivo by administration of nicotinic acid 5'-O-benzoyl riboside (reduced). These results demonstrate that the requirement for a cofactor need not be a limitation in the use of reductive enzymes in antibody directed enzyme prodrug therapy (ADEPT). It is proposed that the E. coli nitroreductase would be a suitable enzyme for ADEPT in combination with CB 1954 and a synthetic, enzyme-selective, virtual cofactor such as nicotinic acid riboside (reduced).

Catalytic properties of NAD(P)H:quinone acceptor oxidoreductase: study involving mouse, rat, human, and mouse-rat chimeric enzymes.

NAD(P):quinone acceptor oxidoreductase (quinone reductase) (DT-diaphorase, EC 1.6.99.2) is involved in the process of reductive activation of cytotoxic antitumor quinones and nitrobenzenes. In this study, we initially examined the relative abilities of mouse, rat, and human quinone reductases to reduce two prodrugs, CB 1954 [5-(aziridin-1-yl)-2,4-dinitrobenzamide] and EO9 [5-(1-aziridinyl)-3-(hydroxymethyl)-2-(3-hydroxy-1-propenyl)-1- methyl-1H-indole-4,7-dione]. By using Escherichia coli-expressed quinone reductases and evaluating them under identical conditions, we confirmed previous finding showing that the human enzyme is not as effective as the rat enzyme in reducing CB 1954 and EO9, although the two enzymes have similar NAD(P)H-menadione reductase activities. Interestingly, although the amino acid sequence of mouse quinone reductase is more homologous to that of the rat enzyme, we found that the mouse enzyme behaves similarly to the human enzyme in its ability to reduce these compounds and to generate drug-induced DNA damage. To determine the region of quinone reductase that is responsible for the catalytic differences, two mouse-rat chimeric enzymes were generated. MR-P, a chimeric enzyme that has mouse amino-terminal and rat carboxy-terminal segments of quinone reductase, was shown to have catalytic properties resembling those of rat quinone reductase, and RM-P, a chimeric enzyme that has rat amino-terminal and mouse carboxyl-terminal segments of quinone reductase, was shown to have catalytic properties resembling those of mouse quinone reductase. In addition, MR-P and RM-P were found to be inhibited by flavones with Ki values similar to those for rat and mouse quinone reductases, respectively. Based on these results, we propose that the carboxyl-terminal portion of the enzyme plays an important role in the reduction of cytotoxic drugs and the binding of flavones.

Self-immolative prodrugs: candidates for antibody-directed enzyme prodrug therapy in conjunction with a nitroreductase enzyme.

The synthesis and properties of some prodrug candidates for antibody-directed enzyme prodrug therapy (ADEPT) are described. These compounds have been designed to generate the corresponding active drug upon interaction with a bacterial nitroreductase that can be conjugated to antibodies that recognize tumor-selective antigens. The active drugs included in the study are actinomycin D, mitomycin C, doxorubicin, 4-[bis(2-chloroethyl)amino]aniline and 4-[bis(2-chloroethyl)amino]phenol. The prodrugs were all 4-nitrobenzyloxycarbonyl derivatives of these drugs, which upon enzymatic reduction, generated the drug through self-immolation of the 4-(hydroxyamino)benzyloxycarbonyl group. In the case of actinomycin D, the ratio of the dose required between drug and prodrug to give the same cytotoxicity was greater than 100. The prodrug was also much less toxic (20-100x) than actinomycin D to mice in vivo. Therefore this self-immolative prodrug has a potential application in the treatment of cancer using an ADEPT-type approach.

Identification, synthesis and properties of 5-(aziridin-1-yl)-2-nitro-4-nitrosobenzamide, a novel DNA crosslinking agent derived from CB1954.

5-(Aziridin-1-yl)-4-hydroxylamino-2-nitrobenzamide, the active form of 5-(aziridin-1-yl)-2,4-dinitrobenzamide (CB1954), can react spontaneously with oxygen, and in aqueous solution yields 5-(aziridin-1-yl)-2-nitro-4-nitrosobenzamide and hydrogen peroxide. Mild biological reducing agents such as NAD(P)H, reduced thiols and ascorbic acid rapidly re-reduced the nitroso compound to the hydroxylamine. Both compounds were equally efficient at inducing cytotoxicity and DNA interstrand crosslinking in cells when exposed in phosphate-buffered saline (PBS). Neither agent was capable of inducing cross-links in isolated DNA. When acetyl coenzyme A was included in the incubation, crosslink formation was seen with the hydroxylamine, but not with the nitroso compound. Thus, the nitroso compound is acting as a prodrug for the hydroxylamine, and needs to be reduced to this compound to exert its cytotoxic effects. In vivo anti-tumour tests showed that neither compound was effective in its own right. This may be due to the rapid reduction of the nitroso to the hydroxylamine, and the reaction of the hydroxylamine with serum proteins. The chemical synthesis of the 5-(aziridin-1-yl)-2-nitro-4-nitrosobenzamide, and an improved synthesis of 5-(aziridin-1-yl)-4-hydroxylamino-2-nitrobenzamide is described. These results emphasize the potential efficacy of the in situ activation of prodrugs such as CB1954 either by endogenous enzymes such as DT diaphorase, or by antibody directed enzyme prodrug therapy (ADEPT).

The bioactivation of CB 1954 and its use as a prodrug in antibody-directed enzyme prodrug therapy (ADEPT).

Walker cells in vivo or in vitro are exceptionally sensitive to the monofunctional alkylating agent CB 1954 (5-(aziridin-1-yl)-2,4-dinitrobenzamide). The basis of the sensitivity is that CB 1954 forms DNA interstrand crosslinks in Walker cells but not in insensitive cells. Crosslink formation is due to the aerobic reduction of CB 1954 to form 5-(aziridin-1-yl)-4-hydroxylamino-2-nitrobenzamide by the enzyme DT diaphorase. The 4-hydroxylamine can not crosslink DNA directly but requires further activation by a non-enzymatic reaction with a thioester (such as acetyl coenzyme A). As predicted from their measured DT diaphorase activities, a number of rat hepatoma and hepatocyte cell lines are also sensitive to CB 1954. However, no CB 1954-sensitive tumours or cell lines of human origin have been found. This is because the rate of reduction of CB 1954 by the human form of DT diaphorase is much lower than that of the Walker enzyme (ratio of kcat = 6.4). To overcome this intrinsic resistance of human cells towards CB 1954 a number of strategies have been developed. First, analogues have been developed that are more rapidly reduced by the human form of CB 1954. Second, the cytotoxicity of CB 1954 can be potentiated by reduced pyridinium compounds. Third, a CB 1954 activating enzyme can be targeted to human tumours by conjugating it to an antibody (ADEPT). A nitroreductase enzyme has been isolated from E. coli that can bioactivate CB 1954 much more rapidly than Walker DT diaphorase and is very suitable for ADEPT. Thus CB 1954 may have a role in the therapy of human tumours.

The bioactivation of 5-(aziridin-1-yl)-2,4-dinitrobenzamide (CB1954)--I. Purification and properties of a nitroreductase enzyme from Escherichia coli--a potential enzyme for antibody-directed enzyme prodrug therapy (ADEPT).

A nitroreductase enzyme has been isolated from Escherichia coli B. This enzyme is an FMN-containing flavoprotein with a molecular mass of 24 kDa and requires either NADH or NADPH as a cofactor. Partial protein sequence analysis showed extensive homology with the "classical nitroreductase" of Salmonella typhimurium and a nitroreductase induced in Enterobacter cloacae. In common with the Salmonella enzyme, the E. coli B enzyme is capable of reducing nitrofurazone. The E. coli nitroreductase is also capable of reducing the anti-tumour agent CB1954 [5-(aziridin-1-yl)-2,4-dinitrobenzamide], a property shared with the mammalian enzyme DT diaphorase [NAD(P)H dehydrogenase (quinone)] as isolated from Walker cells. The reduction of CB1954 by the E. coli enzyme results in the generation of cytotoxic species. Both enzymes also share the properties of being able to reduce quinones and are both inhibited by dicoumarol. The nitroreductase is a more active enzyme against CB1954 (kcat = 360 min-1) than Walker DT diaphorase (kcat = 4 min-1) and also has a lower Km for NADH (6 vs 75 microM).