Targeting O ⁶-methylguanine-DNA methyltransferase with specific inhibitors as a strategy in cancer therapy

Abstract

O⁶-methylguanine-DNA methyltransferase (MGMT) repairs the cancer chemotherapy-relevant DNA adducts, O⁶-methylguanine and O⁶-chloroethylguanine, induced by methylating and chloroethylating anticancer drugs, respectively. These adducts are cytotoxic, and given the overwhelming evidence that MGMT is a key factor in resistance, strategies for inactivating MGMT have been pursued. A number of drugs have been shown to inactivate MGMT in cells, human tumour models and cancer patients, and O⁶-benzylguanine and O⁶-[4-bromothenyl]guanine have been used in clinical trials. While these agents show no side effects per se, they also inactivate MGMT in normal tissues and hence exacerbate the toxic side effects of the alkylating drugs, requiring dose reduction. This might explain why, in any of the reported trials, the outcome has not been improved by their inclusion. It is, however, anticipated that, with the availability of tumour targeting strategies and hematopoetic stem cell protection, MGMT inactivators hold promise for enhancing the effectiveness of alkylating agent chemotherapy.

Introduction

DNA alkylating agents have been used in cancer therapy for almost 30 years. Table 1 lists the two categories of these agents, which as a group are frequently referred to as “O ⁶-alkylating agents”. The methylating nitrosamide, N-methyl-N-nitrosourea (MNU), which is a highly neurotropic carcinogen [1], was paradoxically one of the first anticancer drugs used in brain tumour therapy [2]. Later, streptozotocin, which is a glucose derivative of MNU, was introduced in the clinic and is still in use for the therapy of islet-cell carcinomas [2]. MNU is quite unstable and, at neutral and alkaline pH, decomposes spontaneously reacting immediately with cellular nucleophils. The agent was replaced by procarbazine and dacarbazine (DTIC), which are much more stable needing metabolic activation by cytochrome P450 enzymes to generate alkylating species. The newest generation drug is temozolomide (Temodal, Temodar^®). This is a triazene derivative that does not need metabolic activation, decomposing spontaneously into the active form, methyltriazenoimidazole carboxamide (MITC), which releases carbonium ions that alkylate DNA.

Table 1.

Alkylating anticancer drugs

	Clinical application
Methylating anticancer drugs
Streptozotocine (zanosar)	Metastatic cancer of the pancreatic islet cells
Procarbazine (natulan)	Hodgkin’s Lymphoma (MOPP regimen) malignant gliomas (PCV regimen)
Dacarbacine (DTIC)	Metastatic melanoma (ABVD regimen) Hodgkin’s lymphoma, sarcoma (MAID regimen)
Temozolomide (temodal)	Malignant gliomas, melanoma
Chloroethylating anticancer drugs
ACNU (nimustine)	Malignant gliomas (PCV regimen)
BCNU (carmustine)	Malignant gliomas (PCV regimen), medulloblastoma, astrocytoma, melanoma, multiple myeloma, malignant lymphoma (Hodgkin’s and non-Hodgkin)
CCNU (lomustine)	Malignant gliomas (PCV regimen), Hodgkin’s lymphoma, melanoma, bronchial carcinoma
MeCCNU (semustine)	Malignant gliomas (PCV regimen), lymphomas, colorectal cancer, stomach cancer.
HeCNU (elmustine)	Malignant gliomas (PCV regimen)
Estramustine	Prostate cancer
Fotemustine	Melanoma

The chloroethylating agents encompass carmustine (BCNU), nimustine (ACNU), semustine (methyl-CCNU), lomustine (CCNU) and the second generation drug fotemustine. These monofunctional nitrosourea derivatives generate, among other lesions, O ⁶-chloroethylguanine in DNA. Within several hours after formation, this unstable adduct undergoes intramolecular rearrangement, forming the N1-O ⁶-ethenoguanine adduct and subsequently a N1-guanine-N3-cytosine interstrand DNA crosslink [3]. These crosslinks are highly toxic (see Fig. 1), activating the apoptotic pathway, as do crosslinks induced by bifunctional drugs such as cyclophosphamide [4].

Fig. 1.

O⁶-MeG and O⁶-chloroethylguanine driven cell death pathways, and protection by MGMT. During DNA replication, O⁶-MeG mispairs with thymine forming O⁶-MeG-thymine [142]. Mismatch repair removes thymine from O⁶-MeG-T mispairs. Due to the mispairing properties of O⁶-MeG, thymine is again inserted, which results in a futile repair cycle. This may result in single-strand DNA repair patches that block replication. In a subsequent round of replication this eventually results in DNA double-strand breaks [143] that are potent activators of the apoptotic pathway [144]. O⁶-chloroethylguanine in DNA is an unstable adduct undergoing a slow intramolecular rearrangement, forming the cyclic etheno adduct and subsequently a N1-guanine-N3-cytosine interstrand crosslink. If not repaired by the crosslink repair system, which involves the p53 regulated proteins DDB2 and XPC [145], these crosslinks are highly toxic, activating the apoptotic pathway [4]. MGMT repairs the initially formed O⁶-MeG as well as O⁶-MeG mispaired with thymine [20]. Therefore, its resynthesis exerts protection even some time after the primary lesion O⁶-MeG was induced. MGMT also repairs the O⁶-chloroethylguanine adduct by transferring the chloroethyl group to its own cysteine

Both methylating and chloroethylating agents damage cellular macromolecules via a unimolecular nucleophilic substitution reaction (S_N1 reaction), and they thus have a strong electrophilic affinity towards oxygen atoms in DNA. Among these, the O ⁶ position of guanine is biologically very likely the most important. O ⁶-alkylguanine is repaired by the suicide enzyme O ⁶-methylguanine-DNA methyltransferase (MGMT), which protects against a substantial portion of the toxic and mutagenic effects of methylating and chloroethylating agents (Fig. 1). Although there is some controversy about whether or not MGMT is also able to protect against cyclophosphamide toxicity [5–7], MGMT inactivating agents are effective only with the O ⁶-alkylating agents. Therefore, this review refers solely to this group of anticancer drugs and the preclinical development and clinical application of MGMT-inactivating agents.

Cytotoxicity mechanisms of the O⁶-alkylating agents

Although methylating agents generate 13 adducts in DNA [8], it has been shown that, under most circumstances, the minor product O ⁶-MeG, amounting to less than 8% of total alkylations, is the major toxic lesion. For the chloroethylating agents, O ⁶-chloroethylguanine is also a minor lesion to which most of the toxicity is attributed. The most compelling evidence supporting the mechanisms of the toxic effects of O ⁶-alkylating agents is that, in the vast majority of situations, repair by MGMT (see below) almost completely abolishes cell killing, particularly in the lower dose range of the agents. In vitro experiments indicated that, at high dose levels other repair pathways, e.g. base excision repair removing other lesions, e.g. N-alkyl purines, may become saturated and hence these lesions may contribute to cytotoxicity. In this case, protection by MGMT is less important for determining overall toxicity. For methylating and chloroethylating agents, the pathways leading to cell death, which is executed mainly by apoptosis in both melanoma [9] and glioblastoma cells [10], are outlined in Fig. 1.

It is important to note that chloroethylating agents induce apoptosis in the post-treatment cell cycle following the passage of cells through S-phase. This process is mismatch repair (MMR)-independent, whereas methylating agents require two cell cycles, with MMR acting after the first, in order to generate DNA double-strand breaks (DSBs) that trigger apoptosis [11].

The DNA repair protein MGMT

The DNA repair protein MGMT protects cells against the toxic effects of methylating and chloroethylating agents. It belongs to the group of enzymes that repair DNA by damage reversal. Human cells contain just one MGMT species, and the MGMT gene consists of one non-coding and four coding exons, is about 145 kb long and is located at chromosomal position 10q26. The gene encodes an mRNA of 866 nucleotides that codes for a protein containing 207 amino acids with a molecular weight of 24 kDa. MGMT is a relatively stable protein, having a half-life of >24 h [12]. There are reports that phosphorylation of MGMT affects its activity [13, 14].

Repair occurs in a one-step reaction that does not involve excision of the alkylated base from DNA, instead the methyl or chloroethyl group at the O ⁶ position of guanine is transfered to the cysteine residue (Cys145) in the active centre of the MGMT molecule (Fig. 2a) [15, 16]. This results in restoration of guanine in the DNA and irreversible inactivation of MGMT. Therefore, MGMT is often referred to as a “suicide enzyme”.

Fig. 2.

Dealkylation of O ⁶-methylguanosine and the pseudosubstrate O ⁶-benzylguanine by MGMT. a Following DNA methylation, the methyl group is transferred from O ⁶-methylguanosine to the active centre of MGMT. The same happens if the free base O ⁶-MeG is used as a pseudo-substrate. b In case of O ⁶-BG, the benzyl group is transferred to Cys 145 of MGMT. Alkyl group transfer to MGMT inactivates the protein and subjects it to degradation via ubiquitination

Because of the stoichiometry of the repair reaction, repair capacity is determined initially by the number of active MGMT molecules in a cell and, if damage levels exceed the levels of pre-existing protein molecules, the rate of de novo synthesis following their inactivation. Protection against cell killing is a linear function of MGMT activity up to a level of MGMT expression of about 200,000 molecules per cell [17], above which the toxicity of other lesions probably becomes dominant. There is some evidence that, after alkyl group transfer, MGMT undergoes ubiquitination and proteasome-mediated degradation [18]. De novo synthesis rates were first studied in rat liver [19]. In glioma cells in vitro, the MGMT activity is recovered within 1–2 cell cycle following a single temozolomide treatment (Kaina, unpublished data), but this effect will likely be dose- and cell line-dependent.

The important consideration is that, because of the stoichiometry and autoinactivation mechanism, cells can be depleted of MGMT and consequently can be more susceptible to killing by O ⁶-alkylating agents. Since O ⁶-MeG can be repaired if either cytosine or thymine is the opposite base [20] and cell killing by methylating agents requires two rounds of DNA synthesis, it would be predicted that sustained inactivation of MGMT is essential for enhancing toxicity. Recently, it has been shown that apoptosis occurs not only in the 2nd, but also the 3rd and 4th cell cycle following treatment with temozolomide [11]. It remains to be established if MGMT is able to attenuate the killing effect if it becomes replenished several cell cycles after treatment. For the chloroethylating agents, MGMT ablation needs only to extend until the formation of the interstrand crosslink, which cannot be repaired by MGMT, but may be processed by other pathways (see Fig. 1). There is one report that MGMT can become covalently bound to the intermediate 1-O ⁶-ethenoguanine in DNA [21], but the physiological significance of this, if any, has yet to be established.

The action of MGMT is not restricted to O ⁶-alkylguanine lesions in DNA; free bases can also be substrates and can therefore also inactivate the protein—a characteristic that has allowed the development of MGMT-inactivating drugs (Fig. 2). Indeed, as is discussed below, an enormous range of O ⁶-alkylguanine bases and adducts in synthetic oligonucleotides are known to completely or partially inactivate (or inhibit) MGMT.

No other functions of MGMT have been described. Thus, in adherent cells, no effect on cell growth is seen whether they express low or very high MGMT levels, indicating that MGMT itself is not linked to regulation of proliferation. In addition, although MGMT is frequently upregulated in cells in more aggressively growing tumours [22], there is no evidence that MGMT has a direct stimulatory or inhibitory effect on tumour growth, although there is one report that the methylated (inactivated) MGMT can bind to the oestrogen receptor and affects growth rate [23]. It has also been shown that, in CD34⁺ stem cells, which have low MGMT activity [24], very high level expression of a mutant form of MGMT that is inactivator-resistant inhibits stem cell proliferation, but the basis of this has not been established [25].

Variable expression of MGMT in tumours and normal tissue

The activity of MGMT has been determined in various human normal and tumor tissues including brain, colon, ovary, testis and breast [26, 27]. Expression was highly variable, particularly in the tumour tissue [28]. There were high levels of MGMT activity in colon cancer, pancreatic carcinoma and lung cancer, while brain tumours and malignant melanoma generally express low levels. MGMT activity in tumour was often higher than that in surrounding normal tissue [27]. For ovarian cancer, MGMT expression correlated with grading and staging [22] and, in glioblastomas, MGMT activity was shown to increase in recurrences [29].

In normal tissues, the expression of MGMT is tissue- and cell-type-regulated. Thus, in rat and human, liver expresses the highest MGMT level, followed by lung and kidney. The MGMT activity also differs between individuals and in the same individual as a function of time. Thus, a long-term study of peripheral blood mononuclear cells (PBMC) from healthy individuals revealed high inter-individual (7.6-fold) and intra-individual (1.4- to 3.5-fold over a 42-day period) variation of MGMT expression [30]. A comparison of the inter-individual MGMT activity in normal lung and colon samples also revealed variation [28].

MGMT and resistance to chemotherapy

The mechanism of cell killing by O ⁶-alkylguanine suggests that, while high levels of MGMT activity will result in resistance to O ⁶-alkylating drugs, low levels do not necessarily cause sensitivity. This reflects the fact that several downstream events are involved in converting O ⁶-alkylguanine into a killing event (see Fig. 1), all of which might be modulated to attenuate cell kill signalling and execution. Nevertheless, there is overwhelming evidence that MGMT is clearly a very important marker of O ⁶-alkylating drug resistance in cultured rodent and human cells, in animal models, including transgenic and knockout mice, and in a wide range of human tumour xenografts.

There is also evidence for the protective effect of MGMT in patients treated with O ⁶-alkylating drugs. Thus, low MGMT expression levels, assessed by immunostaining, correlated with DTIC response in malignant melanoma [31], but this was not the case with temozolomide in another study [32]. Silencing of MGMT by promoter hypermethylation of CpG islands can occur early in human tumorigenesis, and this has frequently been observed in several tumour types including colon, brain, lung, head and neck cancer, and lymphomas [33]. In human glioblastoma, MGMT promoter methylation is related to a better therapeutic response of the patients [34–36]. The same is true for MGMT activity with high MGMT-expressing gliomas (>30 fmol/mg protein) responding poorly to O ⁶-alkylating drug-based therapy [29]. Since low MGMT activity in gliomas (asctocytomas WHO III and glioblastomas) is considered to correlate with promoter methylation [37], and several trials showed MGMT promoter methylation correlates with better outcome of therapy (for review, see [38]), the methylation status of the MGMT promoter is currently being used to predict those patients who are likely to have successful temozolomide or combined temozolomide/CCNU/ACNU chemotherapy. It should be noted that, in current glioma therapy, temozolomide is administered concomitantly with radiotherapy (RT), the outcome of which also correlates with MGMT promoter methylation [35]. It should be further noted that in some studies on glioma [39] and melanoma [40] MGMT expression did not correlate with MGMT promoter methylation status.

Myelosuppression is the dose-limiting toxicity of temozolomide and indeed the other O ⁶-alkylating drugs. Thus, it was shown that 7% of patients had grade 3 or 4 haematological toxic effects following concomitant radiotherapy and temozolomide, and 14% following adjuvant temozolomide [41]. One study showed that there were lower levels of MGMT activity in the PBMC of patients who suffered the highest degree of bone marrow toxicity [42] suggesting that this parameter may be used to identify patients best likely to tolerate dose intensification.

MGMT inhibitors

The finding that MGMT activity has a major impact on the sensitivity of tumour cells to O ⁶-alkylating agents stimulated the search for strategies to inhibit MGMT activity in the tumour tissue in order to increase the response to these agents. The concept of MGMT inhibition was established more than 25 years ago and was first achieved by pretreatment with O ⁶-alkylating agents themselves. Thus, pretreatment of colon carcinoma cells or normal human fibroblasts with the model methylating agent N-methyl-N′-nitro-N-nitrosoguanidine (MNNG) increased carmustine-induced toxicity. This was explained by the ability of MNNG to generate O ⁶-MeG in DNA and this resulted in MGMT depletion so that more DNA crosslinks were generated when the carmustine “challenge” dose was applied [43–45]. In further studies, it was shown that O ⁶-MeG added as a free base to the cell culture medium was capable of inhibiting MGMT in HeLa cells [46], in human fibroblasts and tumour cells [47], and in nude mice HT29 xenografts following intraperitoneal administration. [48]. This was consistent with the fact that O ⁶-MeG acts as a weak substrate for MGMT [46].

In the following years, a huge number of O ⁶-guanine derivatives and related compounds have been described and used for inactivating MGMT in a variety of experimental settings. The most potent of these agents, as demonstrated in numerous in vitro and in vivo studies, was O ⁶-benzylguanine (O ⁶-BG, Figs. 2b, 3), which is more than 2,000-times more effective than O ⁶-MeG. In vitro, complete inactivation of MGMT activity was observed as soon as 15 min after addition of O ⁶-BG to the culture medium, and this resulted in a substantial increase in the cytotoxicity of CCNU [7, 49]. Also, in rodent liver and kidney, O ⁶-BG was shown to almost completely (>95%) inactivate MGMT. In further experiments, a series of O ⁶-substituted purine and S6-substitute thiopurine derivatives were synthesised, along with many other compounds that were considered to be suitable candidates for effective MGMT inactivation. The order of potency of some of the most effective “pseudosubstrate” agents for MGMT inactivation were O ⁶-(p-Y-benzyl)-guanine (where Y is H, F, Cl, or CH₃) > O ⁶-benzyl-2′-deoxyguanosine > O ⁶-(p-Y-benzyl)guanosine (where Y is H, Cl, or CH₃) ≥ several 9-substituted O ⁶-benzylguanine derivatives ≥ O ⁶-allylguanine > O ⁶-benzylhypoxanthine > O ⁶-methylguanine [50]. The 7-substituted benzylguanine derivatives (2-amino-6-(p-Y-benzylthio)purine (where Y is H or CH₃), 2-amino-6-[(p-nitrobenzyl)thio]-9-beta-d-ribofuranosylpurine, and 7-benzylguanine were inactive.

Fig. 3.

Chemical structure of O ⁶-BG and O ⁶-BTG and the corresponding folate and glucose derivatives

Where this has been assessed, O ⁶-MeG analogues inactivate MGMT by alkyl group transfer to Cys 145 in the active centre of the protein (Fig. 2b) [16, 51], and it is presumed that all such agents act in this way. It should be noted that it has been shown for very few compounds that the O ⁶-alkyl group, or its equivalent, is actually transferred to the MGMT active site cysteine residue, so it is formally possible that a portion of the agents are strong competitive inhibitors of MGMT rather than inactivators. MGMT is a DNA binding protein that, on recognizing substrates in alkylated DNA, an arginine “finger” flips the base out of the minor groove and into the active site pocket, so it is intriguing to consider precisely how this mechanism acts on free bases.

While considerable effort has been invested in developing and assessing novel MGMT inactivating agents, O ⁶-BG has remained the most extensively used agent for experimental purposes and in pre-clinical and clinical trials. The only other inactivator that has entered clinical trials is O ⁶-(4-bromothenyl) guanine (O ⁶-BTG, Lomeguatrib, previously called PaTrin-2; see Fig. 3). This is about 10 times more potent than O ⁶-BG with an IC₅₀ of 3.4 nM compared to 180 nM for O ⁶-BG [52]; in vivo, an IC₅₀ for O ⁶-BTG of 4 nM and for O ⁶-BG of 35 nM was reported for HeLa S3 cells [53]. These agents are discussed extensively below.

Pharmacokinetic parameters of O⁶-BG and O⁶-BTG

The metabolism, distribution between plasma and cerebrospinal fluid and elimination of O ⁶-BG has been analysed in a number of studies. O ⁶-BG is metabolised in the liver to O ⁶-benzyl-8-oxoguanine (8-oxo-BG), which is mediated by CYP1A2 and, to a 200-fold lesser extent, CYP3A4 [54]. 8-oxo-BG, which also inactivates MGMT, is further metabolised by de-benzylation into 8-oxoguanine, which is also accomplished by CYP1A2 [55]. The half-life, clearance and the area under curve (AUC) of O ⁶-BG and 8-oxo-BG have been analysed in rats [56], non-human primates [57, 58], adult humans [59] and children [60]. The data are compiled in Table 2. In addition, the penetration O ⁶-BG into the cerebrospinal fluid was analysed. In non-human primates, O ⁶-benzyl-8-oxoguanine (8-oxo-BG) showed a very high CSF:plasma ratio (36%) compared to O ⁶-BG (4.3%) [57]. This more extensive penetration of 8-oxo-BG into the cerebrospinal fluid was verified in another study [61]. 8-oxo-BG and O ⁶-BG showed comparable MGMT inactivation in kidney and brain of mice, while O ⁶-BG was more effective than 8-oxo-BG in MGMT inactivation in D456 human brain tumor xenografts [62]. For O ⁶-BTG, only one study evaluating the pharmacokinetic parameters in patients with advanced solid tumors is available [63].

Table 2.

Pharmacokinetic parameters for O ⁶-BG, O ⁶-BTG and 8-oxo-BG

Organism	Substance	T1/2 (h)	AUC (µM h)	C L	C max	V D	Reference
Rat	O6-BG i.v.	1.6		160 ml/h/kg		405 ml/kg	[56]
	8-oxo-BG	1.2		312 ml/h/kg		507 ml/kg	[56]
Nonhuman primates	O6-BG 200 mg/m2 i.v.	1.6	212	68 ml/min/m2			[57]
	8-oxo-BG	14	2420	6.4 ml/min/m2			[57]
Nonhuman primates (cerebrospinal fluid)	O6-BG (1 mg intraventricular)	0.52	319	0.22 ml/min	412 µM		[58]
	8-oxo-BG	0.76	5		1.9 µM		[58]
Adult humans	O6-BG 10–80 mg/m2 i.v.	0.078–0.1	0.28–2.3	35.9–37.7 l/h/m2	0.22–1.82 µg/ml		[59]
	8-oxo-BG	2.8–9.2	3.45–65.65		0.42–4.33  µg/ml		[59]
	O6-BTG (10-40 mg/m2 i.v.)	Initial 0.43 terminal 2.97	259.9–499.2	85.2–131.8 l/min (C L/F)			[63]
	O6-BTG (10 mg/m2 oral)		52.7–70.6
Children	O6-BG 120 mg/m2 i.v.	1.42	13.25	760 ml/min/m2	11 µM		[60]
	8-oxo-BG	6	378.3		35 µM		[60]

T_1/2 Half-life, AUC area under the curve, C_L clearance, C_max peak concentration in plama or cerebrospinal fluid, V_D volume of distribution, F bioavailability

Preclinical studies: human tumour xenograft models

O⁶-benzylguanine

The effectiveness of O ⁶-BG in sensitising cancer cells to the therapeutic effects of O ⁶-alkylating agents has been clearly demonstrated in a large number of human tumour xenograft models (Table 3). Initial experiments showed that pretreatment with O ⁶-BG increased carmustine sensitivity in athymic mice bearing either human medulloblastoma (D341 Med) or human glioblastoma (D-456 MG) xenografts, leading to regression of 18/20 xenografts [64]. Pretreatment with O ⁶-BG (35 mg/kg) completely depleted MGMT activity in mice bearing human melanoma (A375P) xenografts, and the combination of temozolomide (40 mg/kg) given together with O ⁶-BG on 5 consecutive days produced a significant tumour growth delay in comparison to temozolomide alone. This, however, was not the case upon combination of a single O ⁶-BG pretreatment and a single dose of 200 mg/kg temozolomide [65]. In MGMT-proficient human gastric adenocarcinoma xenografts, O ⁶-BG completely suppressed MGMT activity for up to 12 h and O ⁶-BG given at a single dose of 90 mg/kg i.p. 2 h before carmustine (25 mg/kg) produced a significant growth delay [66]. Similar results were obtained using human glioma SF767 xenografts to test different schedules of O⁶-BG administration. Treatment with O ⁶-BG (80 mg/kg) completely ablated xenograft MGMT activity for 12 h, which recovered to 40% of control activity after 24 h. Combined treatment with O ⁶-BG and 15 mg/kg carmustine resulted in significant inhibition of tumour growth compared to carmustine alone [67]. An even stronger sensitising effect was observed after combining the O⁶-BG bolus treatment (80 mg/kg) with additional low-dose O⁶-BG pre- and post-treatment (8 mg/kg for 24 h), which led to ablation of MGMT activity for 24 h [68].

Table 3.

Preclinical studies with O ⁶-BG

Tumor xenograft model	Anticancer drug	Readout	Reference
Medulloblastoma (D341 Med)	Carmustine	Tumour regression	[64]
Glioblastoma multiforme (D-456 MG)
Melanoma (A375P)	TMZ	Tumour growth delay	[65]
Gastric adenocarcinoma (BGC-823)	Carmustine	Tumour growth delay	[66]
Glioma (SF767)	Carmustine	Tumour growth delay	[67]
Glioma (SF767)	Carmustine	Tumour growth delay	[68]
Medulloblastoma (D341MED) glioma (D-245 MG)	Carmustine	Tumour growth delay	[69]
Pancreatic tumours (MIA PaCa-2, CFPAC-1, PANC-1, CAPAN-2 and BxPC-3)	TMZ or carmustine	Tumour growth delay	[70]
Malignant glioma (D-456 MG)	Cyclophosphamide plus carmustine TMZ plus irinotecan	Tumour growth delay	[71]
Metastatic neuroblastoma	TMZ plus irinotecan	Prolonged survival	[72]
Malignant glioma (F98)	Carmustin	Prolonged survival	[73]
Medulloblastoma (Daoy)	Carmustine	Tumour growth delay	[74]
Malignant glioma (U87MG)	Carmustine	Tumour growth delay	[75]
CNS tumor (D-54M/D-245 MG)	Carmustine	Tumour growth delay	[76]
Intracranial malignant glioma (D-456 MG)	Carmustine i.a versus i.p	Prolonged survival	[77]
Melanoma	TMZ	Tumour growth delay	[78]
Medulloblastoma (Daoy)	Carmustine	Tumour growth delay	[79]

In MGMT-proficient medulloblastoma xenografts, O ⁶-BG produced tumour growth delays following carmustine treatment [69]. The same was found in MGMT-proficient pancreatic xenografts using carmustine or temozolomide as the therapeutic agent [70]. Addition of O ⁶-BG also enhanced the response of xenografts to combined chemotherapy. Thus, in MGMT-proficient malignant glioma xenografts (D-456 MG), O ⁶-BG produced growth delay upon combined treatment with either cyclophosphamide plus carmustine or temozolomide plus irinotecan [71]. Also, in the case of metastatic neuroblastoma xenografts, the survival of mice after combined temozolomide/irinotecan therapy was enhanced by O ⁶-BG from 10 to 56% [72].

O⁶-BG also enhanced the response of xenografts in the rat brain. Thus, in MGMT-proficient malignant glioma xenografts (F98), O⁶-BG (50 mg/kg) prolonged the median survival (to 34 days) compared to O⁶-BG alone (22 days), carmustine alone (25 days) or the non-treated control group (23.5 days) [73], demonstrating that O⁶-BG can effectively abrogate carmustine resistance. In addition to O⁶-BG, its derivative O⁶-benzyl-2′-deoxyguanosine was analysed in a MGMT-positive medulloblastoma xenograft (Daoy), and this agent showed a greater enhancing effect compared to O⁶-BG. Thus, O⁶-benzyl-2′-deoxyguanosine (134 mg/m²) given i.p. 1 h before BCNU (25 mg/m²) produced increased growth delay and survival compared to an equimolar dose of O⁶-BG (90 mg/m²) combined with BCNU or treatment with BCNU alone [74].

Interestingly, O ⁶-BG was also found to sensitise xenografts expressing very low MGMT activity. Thus, 1-h pretreatment of athymic mice with O ⁶-BG (40 mg/kg) reduced MGMT activity in subcutaneous human glioma (U87MG) xenografts from 4.3 to 0.9 fmol/mg protein, and produced tumour growth delays of 23.3 days upon temozolomide (35 mg/kg) and 11.8 days upon carmustine (10 mg/kg) treatment [75]. Similarly, O ⁶-BG produced substantial tumour growth delays in MGMT-deficient glioma xenografts (D-54 MG and D-245 MG) upon carmustine treatment [76]. The data suggest that even very low levels of MGMT are already adequate for protecting cells against killing by O ⁶-alkylating drugs. If so, complete and protracted inactivation of MGMT prior to and following alkylating agent administration are likely to be necessary in order to achieve maximal cytotoxicity. However, even in these circumstances, alternative resistance mechanisms may override the lack of MGMT activity in a chronic administration setting.

A problem encountered in the systemic application of O ⁶-BG is that MGMT depletion in normal tissue cells also sensitises them to killing by the cytotoxins. To counter this exacerbated dose-limiting toxicity, the dose of the O ⁶-alkylating agent needs to be reduced even though this may impact on the therapeutic effectiveness of the agent. Routes of administration that limit the exposure of normal tissues to the MGMT inactivator and/or the cytotoxin might circumvent this problem. Indeed, in athymic rats carrying intracranial human malignant glioma (D-456 MG) xenografts, intraarterial (i.a.) administration of O ⁶-BG was superior to intraperitoneal (i.p.) application. O ⁶-BG depleted MGMT activity in the xenografts to a similar level with a dose of 2.5 mg/kg (i.a.) or 10 mg/kg (i.p.). In addition, i.a. application allowed the use of higher carmustine doses (25 mg/kg vs 6.25 mg/kg), resulting in longer median survival (59–61 days) compared with i.p. application (37 days) [77]. In human melanoma xenografts in nude rats, it was shown that 3.5 mg/kg O ⁶-BG depleted the tumor MGMT by 93.5%, and when combined with regional administered temozolomide, a significant reduction in tumour growth was achieved [78]. In addition to O ⁶-BG, 9-substituted derivatives of BG (O ⁶-benzyl-2′-deoxyguanosine and O ⁶-benzylguanosine) showed additional tumour growth delay in human medulloblastoma (Daoy) tumour xenografts upon carmustine treatment, but this was not the case with O ⁶-benzyl-9-cyanomethylguanine. The effectiveness of these guanosine derivatives results from their efficient cellular uptake and catabolism into O ⁶-BG [79].

O⁶-bromothenylguanine (lomeguatrib, PaTrin-2)

The effect of O ⁶-BTG on MGMT activity and growth inhibition following O ⁶-alkylating drug administration has been analysed in several human tumour xenografts (Table 4). O ⁶-BTG (20 mg/kg) depleted MGMT in human melanoma (A375M) xenografts 2 h after i.p administration [80]. Combining O ⁶-BTG (20 mg/kg i.p. daily) and temozolomide increased the tumour quintupling time by 8.7 days. Interestingly, the toxicity of this combination was less than that of combined temozolomide and O ⁶-BG [80].

Table 4.

Preclinical studies with O ⁶-BTG

Tumour xenograft model	Anticancer drug	Readout	Reference
Human melanoma (A375M)	TMZ (1×)	Depletion of MGMT tumour quintupling time	[80]
Human melanoma (A375M)	TMZ (5×, spaced 4 h)	Tumour quintupling time higher after 5× then 1×	[81]
Human melanoma (A375M)	TMZ (5×, spaced 0, 4, 12, 24 h)	Tumour quintupling time independent of spacing	[82]
Breast carcinosarcoma (MCF-7)	TMZ	Depletion of MGMT tumour quintupling time	[83]
Primary leukaemia blasts	TMZ	Increased sensitivity	[84]

Shorter periods of temozolomide treatment (every 4 h) increased tumour growth delay to 33.6 days versus 23.2 days using the same total dose as a single administration, but with increased systemic toxicity [81]. In a third study, nude mice bearing human melanoma (A375M) xenografts were treated a total of 5 times with temozolomide (100 mg/kg i.p.) at intervals of 4, 12 or 24 h, and this resulted in tumour quintupling times of 16.8, 5.9 and 6.2 days, respectively. Combining these schedules with O ⁶-BTG (20 mg/kg i.p.) given 1 h prior to temozolmide resulted in quintupling times of 22.1, 21.3 and 22.3 days, respectively. It was concluded that MGMT inactivation by O ⁶-BTG is more promising in enhancing the activity of temozolomide than compressed temozolomide scheduling [82]. A further study showed that human breast carcinoma (MCF-7) xenografts expressed very high levels of MGMT and were completely resistant to temozolomide given daily at 200 mg/kg for 5 days. O ⁶-BTG (20 mg/kg i.p.) completely inactivated tumour MGMT and in combination with this schedule temozolomide substantially increased tumour quintupling time by 22 days without significant increase in toxicity [83]. O ⁶-BTG was also shown to enhance the toxicity of temozolomide in long-term cultured cells obtained from patients with acute lymphoblastic leukaemia and acute myeloblastic leukaemia. This suggests that O ⁶-BTG may benefit leukemia patients treated with temozolomide or dacarbazine [84].

Clinical studies

O⁶-benzylguanine

Phase I trials

A large number of phase I and II clinical trials involving O ⁶-BG are listed (http://www.clinicaltrials.gov). Thus, from 31 trials, 16 are already completed, 2 are recruiting, 9 are active but not recruiting, and 4 have been suspended or terminated (as of April 2010).

Initial phase I trials determined the dose of O ⁶-BG needed for complete depletion of tumour MGMT activity and defined the maximum tolerated dose of the anticancer drug when combined with O ⁶-BG (Table 5). In adult patients, O ⁶-BG doses of 100 mg/m² [85, 86] to 120 mg/m² [87, 88] are necessary for complete inactivation of MGMT. Interestingly, depletion of MGMT activity occurred in PBMCs at lower doses and did not correlate with tumour MGMT [87]. PBMC should thus not be considered a surrogate for events in tumour tissue, except perhaps to confirm that the patient did receive the inactivator. Another study showed that a bolus of 120 mg/m² O ⁶-BG followed 1 h later by a continuous infusion of 30 mg/m²/day for 48 h completely depleted MGMT activity in progressive malignant glioma [89]. Using an O ⁶-BG dose of 100 mg/m² [86] or 120 mg/m² [90], the maximal tolerated dose of carmustine was 40 mg/m², which is considerably lower than the dose of 120 mg/m² without the MGMT inhibitor. Using bolus infusion of O ⁶-BG (120 mg/m²,1 h) on days 1, 3, and 5, combined with a continuous infusion of O ⁶-BG at 30 mg/m²/day, the maximal tolerated dose of temozolomide was 200 mg/m² on day 1 and 50 mg/m²/day on days 2–5 [91]. The total dose (400 mg/m²) was only slightly less than the maximum tolerated dose of a single temozolomide application without MGMT inhibitor, which was 472 mg/m² [89]. O ⁶-BG was also administered (continuous infusion of 30 mg/m²/day) together with carmustine wafers (Gliadel), which allow topical application of BCNU [92]. The results of more extensive clinical trials using these strategies are awaited with interest.

Table 5.

Clinical studies with O ⁶-BG (phase I)

O6-BG dose	Anticancer drug	Readout	Reference
100 mg/m2	–	Complete depletion of MGMT for 18 h	[85]
100 mg/m2	Carmustine (40 mg/m2)	Maximal tolerated dose	[86]
120 mg/m2	–	Complete depletion of MGMT	[87]
120 mg/m2	–	Complete depletion of MGMT after 6 but not 18 h	[88]
Bolus of 120 mg/m2 + continuous infusion of 30 mg/m2/day	TMZ (472 mg/m2)	Complete depletion of MGMT maximal tolerated dose	[89]
120 mg/m2	Carmustine (40 mg/m2)	Maximal tolerated dose	[90]
Bolus of 120 mg/m2 + continuous infusion of 30 mg/m2/day	TMZ (200 mg/m2 on day 1 + 50 mg/m2/day on days 2–5)	Maximal tolerated dose	[91]
Continuous infusion of 30 mg/m2/day	Carmustine (8 wafers)	Maximal tolerated dose	[92]
120 mg/m2/day	TMZ (>55 mg/m2/day)	Antitumour activity	[93]
	100 mg/m2/day	Maximal tolerated dose (in children)
Bolus of 120 mg/m2 + continuous infusion of 30 mg/m2/day	TMZ (200 mg/m2), irinotecan (80 mg/m2)	Maximal tolerated dose	[94]
Bolus of 120 mg/m2 + continuous infusion of 30 mg/m2/day	TMZ (407–562 mg/m2)	Maximal tolerated dose (in children)	[95]
O6-BG (120 mg/m2)	Carmustine (58 mg/m2)	Maximal tolerated dose (in children)	[96]

In all cases, dose limitation was myelosuppression manifested as neutropenia, leukopenia, and thrombocytopenia [86, 89, 91, 93, 94]. That O ⁶-BG exacerbated this toxicity provides strong circumstantial evidence for MGMT being a major resistance mechanism, at least in the corresponding normal human bone marrow stem cells. Strategies to overcome this problem have been proposed and some have been used in clinical trials (see below).

Phase I trials with O ⁶-BG were also performed in children suffering from CNS tumours. The maximal tolerated dose of temozolomide given 30 min after infusion of 120 mg/m²/day O⁶-BG for 5 consecutive days was 100 mg/m²/day [93]. When O ⁶-BG was given i.v. at a dose of 120 mg/m² followed by 48 h continuous infusion at 30 mg/m²/day, the maximum tolerated total dose of temozolomide (given 6 h after O ⁶-BG bolus) was 407–562 mg/m² [95]. Antitumour activity was observed at 120 mg/m²/day O ⁶-BG combined with temozolomide doses of >55 mg/m²/day. In another trial, a combination of O ⁶-BG (120 mg/m²) and carmustine showed a maximal tolerated dose of 58 mg/m² [96].

In addition to alkylating drugs, O ⁶-BG has been examined for its effect on irinotecan-induced toxicity, showing a maximal tolerated dose of irinotecan of 80 mg/m² when combined with 200 mg/m² TMZ [94].

Phase II trials

Several phase II trials have evaluated the safety and toxicity of O ⁶-BG in combination with O ⁶-alkylating agents and these are outlined in Table 6. One study in patients with recurrent glioblastoma multiforme showed that O ⁶-BG can be coadministered safely with carmustine wafers: the overall survival was 82% after 6 months, 47% after 1 year and 10% after 2 years [97]. A similar study evaluated a combination of temozolomide and O ⁶-BG in patients with glioblastoma multiforme and anaplastic glioma: only 3% of the former and 16% of the latter responded to therapy, and in 48% of the patients, grade 4 hematologic events were observed [98]. Combined treatment with O ⁶-BG and carmustine every 6 weeks in 18 patients with CNS tumours also failed to show any impact on clinical outcome [99]. In another study, in 17 patients with multiple myeloma, 1 complete response and 3 partial responses were observed [100]; in 18 patients with chemo-naive advanced melanoma, 1 complete response, 4 stable disease and 13 progressive disease were observed, and in 18 prior-chemotherapy patients, no responses, 3 stable disease and 15 progressive disease were observed [101]. In 12 patients with advanced soft tissue sarcoma, there were also no responders [102]. It should be noted that in all these trials the response of patients receiving O ⁶-BG with temozolomide or carmustine was not compared with the alkylating drug only group, which makes assessment of the data difficult.

Table 6.

Clinical studies with O ⁶-BG (phase II)

tumour	Anticancer drug	Readout	Reference
Glioblastoma multiforme	O6-BG (Bolus of 120 mg/m2 + continuous infusion of 30 mg/m2/day) + carmustine wafers	Safety of application no impact on clinical outcome	[97]
Glioblastoma multiforme anaplastic glioma	O6-BG (Bolus of 120 mg/m2 + continuous infusion of 30 mg/m2/day) + TMZ (472 mg/m2)	Safety of application no impact on clinical outcome	[98]
CNS tumours	O6-BG (120 mg/m2) + 40 mg/m2 carmustine	No impact on clinical outcome	[99]
Multiple myeloma	O6-BG (120 mg/m2) + 40 mg/m2 carmustine	No impact on clinical outcome	[100]
Advanced melanoma	O6-BG (120 mg/m2) + 40 mg/m2 carmustine	No impact on clinical outcome	[101]
Advanced soft tissue sarcoma	O6-BG (120 mg/m2) + 40 mg/m2 carmustine	No impact on clinical outcome	[102]

O⁶-bromothenylguanine

Phase I and II trials

In two phase I studies, the combination of O ⁶-BTG and temozolomide or irinotecan was analysed in various solid tumours to define the biologically effective (i.e. MGMT-inactivating) dose, the maximum tolerated dose and dose-limiting toxicity of the combination (Table 7). In the first study, O ⁶-BTG was administered to patients with advanced solid tumours [63]. Within 4 h, depletion of MGMT activity was observed in PBMCs (≥95%) and tumour biopsies (≥92%) at doses of ≥10 mg/m²/day i.v. or ≥20 mg/m²/day orally. The maximal tolerated dose of temozolomide in combination with O ⁶-BTG was 150 mg/m² (vs 200 mg/m² without O ⁶-BTG) and dose-limiting toxicity was myelosuppression [63]. In another study, O ⁶-BTG was administered to 24 patients with metastatic colorectal cancer [103], and the maximum tolerated dose was 80 mg/day when combined with 300 mg/m² irinotecan; dose-limiting toxicity was neutropaenia and diarrhoea. At this dose level, O ⁶-BTG administration resulted in complete MGMT inactivation in PBMCs [103]. Besides use in therapy of solid tumours, pilot studies were performed to analyse O ⁶-BTG in refractory acute leukaemia. The data indicate that also in this cancer, O ⁶-BTG (40 mg/m²/day orally, day 0–10) can suppress MGMT activity and that after combined treatment with temozolomide (150 mg/m²/day orally, day 1–7) six out of eight patients showed partial or complete disappearance of blast cells in peripheral blood and in bone marrow [104].

Table 7.

Clinical studies with O ⁶-BTG

tumour model	Anticancer drug	Readout	Reference
Phase I
Advanced solid tumours	O6-BTG (≥10 mg/m2/day i.v. or ≥20 mg/m2/day)	Depletion of MGMT in PBMCs and tumour biopsies, maximal tolerated dose	[63]
	TMZ (150 mg/m2)
Metastatic colorectal cancer	O6-BTG (80 mg/day) Irinotecan (300 mg/m2)	Depletion of MGMT in PBMCs and maximal tolerated dose	[103]
Refractory acute leukaemia	O6-BTG (40 mg/m2/day, day 0–10), temozolomide (150 mg/m2/day, day 1–7)	Depletion of MGMT, response rate	[104]
Metastatic melanoma	O6-BTG (40 mg) for 10 or 14 day plus TMZ (75–100 mg/m2) day 1–5	Toxicity, adverse effects, response rate	[107]
Colorectal cancer	O6-BTG (120 mg)	Complete MGMT inactivation in the tumour after 12 h	[109]
Prostate cancer	O6-BTG (120 mg)
CNS cancer	O6-BTG (160 mg)
Phase II
Metastatic melanoma	O6-BTG (40-80 mg) 5×	Response rate, median time to disease progression	[105]
	TMZ (125 or 200 mg/m2) 5×
Metastatic colorectal carcinoma	O6-BTG (40 mg) 5×	Median time to disease progression, pharmacokinetics of TMZ and O 6-BTG depletion of MGMT activity in PBMCs	[106]
	TMZ (50–200 mg/m2) 5×
Metastatic melanoma	O6-BTG (40–80 mg) 5×	Pharmacodynamic analysis for [105] and [107], depletion and recovery of MGMT activity in PBMCs and tumour biopsies	[108]
	TMZ (125 or 200 mg/m2) 5×

In one phase II trial, over 100 patients with metastatic melanoma were treated with temozolomide alone or a combination of O ⁶-BTG and temozolomide on days 1–5 every 28 days for up to 6 cycles [105]. Combination with O ⁶-BTG did not significantly influence the overall response rate (13.5 vs 17.3%) or the median time to disease progression (65.5 vs 68 days). In another phase II study in 19 patients with stage IV metastatic colorectal carcinoma, O ⁶-BTG and temozolomide orally for 5 consecutive days resulted in the same outcome as the temozolomide alone group. In both groups, the median time to progression was 50 days and the commonest adverse effects were gastrointestinal and haematologic toxicity [106].

It was suggested that the inability of O ⁶-BTG to enhance clinical response to temozolomide might be a consequence of scheduling, since tumour biopsies showed recovery of MGMT activity within 24 h [105]. This would have been missed in the previous phase I studies, because MGMT activity was determined in the tumour at early time points after treatment. Based on this, higher daily doses of O ⁶-BTG and an extended dosing period beyond that of temozolomide was assessed in additional phase I studies. Thirty-two patients with metastatic melanoma were treated orally with O ⁶-BTG (40 mg) for 10 or 14 days and temozolomide (75–100 mg/m²) on days 1–5. Due to haematologic toxicity, the optimal extended O ⁶-BTG dosing schedule was O ⁶-BTG for 10 days combined with temozolomide (75 mg/m²). However, this extended O ⁶-BTG dosing schedule also showed no advantage compared to temozolomide alone in the treatment of melanoma [107]. This study showed furthermore that, while MGMT activity was completely inactivated in PBMC and tumours biopsied on the last day of treatment with O ⁶-BTG, it thereafter quickly recovered in tumours, indicating that even more protracted dosing with O ⁶-BTG would be needed for extensive ablation of MGMT activity [108].

In order to establish if different doses of O ⁶-BTG would be needed to deplete MGMT activity in prostate, colorectal or brain tumours [109], a total of 32 patients were given a single dose of O ⁶-BTG orally approximately 12 h before resection of their primary tumour. Complete inactivation of MGMT in prostate and colorectal cancers required a dose of 120 mg, and in CNS tumours, a dose of 160 mg O ⁶-BTG [109]. This indicates that the doses used in the phase II studies (40–80 mg) may have been lower than optimal.

An additional factor that thwarts attempts to improve the therapeutic index of O ⁶-alkylating agents in melanoma is very likely the inherent drug resistance of melanomas. Melanoma cells undergo apoptosis in response to methylating and chloroethylating agents, and the upstream pathways have been elucidated [9]. However, cell death execution is inefficient due to silencing of downstream pro-apoptotic pathways (Roos and Kaina, unpublished data). Therefore, MGMT inhibition together with strategies aimed at reactivating apoptotic signaling should be considered in future trials in order to enhance the therapeutic response of DTIC, temozolomide and other O ⁶-alkylating agents for melanoma therapy.

MGMT inhibitor targeting

As has been described above, potent MGMT inactivating agents have been developed that are without any toxicity per se in animals, inhibit the growth of human tumour xenografts in nude mice when combined with O ⁶-alkylating drugs and, in patients, inactivate MGMT in tumour, PBMC and other tissues and have no side effects per se. However, in clinical trials, they have not yet resulted in any improvement in the therapeutic efficacy (in terms of overall survival) of methylating or chloroethylating anticancer drugs in glioma and melanoma therapy (see Tables 1, 2, 3 and 4). A likely major reason for this is that dose reduction of the alkylating drug is necessary for patients to tolerate the increased systemic side effects and that this dose reduction effectively decreases tumour cell kill. It would therefore be highly desirable to develop strategies for specifically targeting the MGMT-inactivating agent selectively to the tumour. A simple, but technically sophisticated, approach is to administer the inhibitor locally. This has been done in an individual trial with a patient suffering from glioblastoma multiforme. After dissection of a recurrence that expressed MGMT (175 fmol/mg protein; for comparison, the average MGMT expression level of pretreatment gliomas is 37 fmol/mg protein [29]), an Ommaya reservoir was implanted into the tumour cavity and used to administer O ⁶-BG directly into the brain prior to systemic temozolomide. No systemic or neuronal toxicity was observed due to intracranial O ⁶-BG administration, which indicates that this approach is feasible and tolerated by the patient [110]. Currently, treatment of a larger group of patients is awaiting approval for assessing the feasibility, costs and benefit of such treatment.

In addition to topical delivery of O ⁶-BG two chemical modification strategies have been assessed for targeting MGMT inactivators to tumours. These are (1) conjugation of O ⁶-BG to folate and (2) conjugation of O ⁶-BG and O ⁶-BTG to β-d-glucose. The first approach is based on the observation that tumour cells often exhibit high levels of folate receptors. The O ⁶-BG-folate conjugates were shown to be effective MGMT inactivators and to predominantly kill cells expressing high folate receptor levels in comparison to low level expressing cells [111]. In addition to O ⁴-benzylfolic acid (Fig. 3), folate ester derivatives of O ⁶-benzyl-2′-deoxyguanosine and O ⁶-[4-(hydroxymethyl)benzyl]guanine were synthesised. The former is a potent MGMT inactivator that effectively sensitised human tumour cells to BCNU [112]. Despite these encouraging findings, no results of in vivo studies with xenografts have been published to date.

The second approach exploits the finding that a common characteristic of tumour cells is increased glucose consumption [113], which is related to elevated glucose uptake via the up-regulation of glucose transporters [114]. On this basis, MGMT inactivators have been conjugated to d-glucose. The in vitro testing of such conjugates showed that a short spacer of 1–4 carbon atoms between d-glucose and the N9 of O ⁶-BG completely abolished or strongly attenuated MGMT inactivation. One possible explanation is that d-glucose in the immediate vicinity of the O ⁶-BG moiety prevents access of the free base to the active site in MGMT. Extending the linker to more than six carbon atoms restored most of the activity [115, 116]. A linker of eight carbons appeared to be optimal, since it retained activity and, serendipitously, significantly increased water solubility. It was also shown that for MGMT inactivation, linking the N9 of O ⁶-BG via a C8 spacer to β-d-glucose was superior to α-d-glucose. The free base inhibitors (see Fig. 3) inactivated MGMT in cell extracts and in living cells in the sequence: O ⁶-BTG > O ⁶-I-BG (O ⁶-iodobenzylguanine) > O ⁶-BG > O ⁶-iodothenylguanine, and the same order of potency was seen with the corresponding glucose conjugates. O ⁶-(2-fluropyridimylmethl)guanine-C8-β-d-glucose was significantly less effective than the other inactivators [53]. The conjugates were not cytotoxic per se in cell culture, penetrated quickly into living cells and depleted MGMT within ~45 min (unpublished data). When given 1 h prior to and after treatment with the alkylating agent, O ⁶-BTG-C8-β-d-glucose was similar to O ⁶-BTG in its ability to sensitise MGMT-expressing CHO and HeLa cells to fotemustine and temozolomide toxicity in colony formation experiments [53].

These in vitro studies demonstrate that the glucose conjugates are able to enter cells and inactivate MGMT resulting in substantial potentiation of the killing effect of O ⁶-alkylating agents. Although it is likely that glucose transporters (sodium–glucose-linked transporters (SGLTs) or glucose transporters (GLUTs)) are involved in the uptake, the transport mechanism still needs to be elucidated. Another issue is the possible expression of these transporters in proliferating haematopoetic progenitor cells and other normal proliferating cells in the body, which could result in additional side effects when using these targeted inhibitors. Our preliminary data show that cells from different tumour types display substantial differences in uptake efficiency of O ⁶-BG-C8-β-d-glucose. This suggests that the transporter/s involved are expressed to different extents in cancers, and hence that it will be essential to identify the transporter/s involved in order to screen individual tumour biopsies for expression.

Stem cell protection

The myelosuppressive effects of the O ⁶-alkylating agents are likely due to the low levels of MGMT expressed in haematopoetic stem cells or a proliferating precursor pool in the bone marrow. This is supported by the consistent demonstrations that myelosuppression is more extensive following administration of MGMT-inactivating agents. To attenuate the severity of myelosuppression, gene transfer and high level expression of MGMT in haematopoetic stem cells, in the context of autologous bone marrow transplantation, is a feasible proposition. Such a strategy could additionally involve tumour sensitisation using MGMT-inactivating agents but only if an inactivating agent-resistant MGMT was employed. It is this latter approach that has received most attention in many in vitro and preclinical studies [117]. An additional benefit of this approach would likely be the protection of bone marrow stem cells/hematopoetic precursor cells against the mutagenic and carcinogenic effects of O ⁶-alkylating agents such as temozolomide, which was suggested by in vivo experiments performed in mice [118] and by case reports of long-time surviving patients [119–121].

Initial studies involved the transfer of mgmt gene into CD34⁺ stem cells harvested from peripheral blood and this provided haematoprotection against the toxicity of O ⁶-alkylating agents [122–126]. The simultaneous myeloprotection–tumour sensitisation strategy was investigated using the E. coli alkyltransferase, Ada, which is not inhibited by O ⁶-BG [127] and versions of the human MGMT cDNA encoding G156A or P140K, which are also resistant to inactivation by O ⁶-BG [128–130] and, in the case of P140K, O ⁶-BTG [131]. Resistance to O ⁶-BG inactivation clearly varies between the wt and both mutant MGMT proteins. Thus, in MGMT-transduced haematopoetic K562 cells, the wt MGMT was already inhibited by 0.1 μM O ⁶-BG, the G156A mutant had an IC₅₀ of 15 μM and the P140K mutants remained active up to 1 mM O ⁶-BG [132].

A number of preclinical studies have investigated various strategies to achieve expression of the mutant MGMT in haematopoetic cells and to assess its impact on the combined administration of MGMT inactivator and O ⁶-alkylating agents. Retroviral transduction of bone marrow cells with MGMT(G156A) and transplantation into mice protects these animals against the myelotoxic effects of O ⁶-BG-BCNU treatment [133]. It also increased tumour kill in human xenograft models because it allowed administration of increased doses of the cytostatic drug [134]. The same was also observed using transduction of MGMT(P140K) in combination with O ⁶-BG and BCNU [135, 136], temozolomide [137] or ACNU [138] as well as combined O ⁶-BTG and temozolomide treatment [131].

One problem with MGMT gene transfer is the low cellular transduction rate of the viruses used. Thus, alternative viral systems like foamy viral [139] or lentiviral vectors [140] carrying MGMT(P140K) have been investigated. In addition to the transduction of mutated MGMT, transduction of MGMT (P140K) together with multi-drug resistance 1 gene (mdr1) have been investigated [141], and significantly higher survival of transduced HL60 cells following treatment with O ⁶-BG/temozolomide and paclitaxel has been reported. These preclinical studies have been translated into the clinic. Of the three phase I clinical trials of MGMT-based myeloprotective gene therapy described (http://www.clinicaltrials.gov), one has already been suspended, one is ongoing and the third was completed this year with the results yet to be published. The outcome of these studies is awaited with great interest.

Conclusions

The ability of MGMT-inactivating agents to sensitise mammalian cells and human tumour xenografts to the toxic or therapeutic effect of O ⁶-alkylating drugs has been clearly demonstrated in virtually all experimental systems in which the strategy has been assessed. However, despite all the indications that MGMT is a major resistance factor in human tumours, and also in bone marrow, there is at present no convincing evidence that this strategy (i.e. systemic MGMT depletion) has been able to improve the outcome in cancer patients treated with these agents.

Ongoing studies intended to limit exposure of normal tissues to MGMT inactivators and or O ⁶-alkylating agents involve alternative delivery routes, topical applications and tumour-targeted inactivators. These hold considerable promise for the success of drugs and drug combinations the efficacy of which is attenuated by MGMT. The alternative strategy of protecting the bone marrow against the collateral toxicity of O ⁶-alkylating agents, which itself is increased by MGMT inactivators, currently involves the delivery and expression of an inactivator-resistant MGMT variant in bone marrow stem cells. MGMT gene transfer is just beginning to enter the clinic, and several problematic aspects have still to be resolved. These include the regulation of MGMT expression levels by an appropriate vector system, the growth inhibitory side effects of mutant forms of MGMT and the perceived hazard of viral genomic integration. On the other hand, this approach has additional potential benefits: it may allow dose escalation even if MGMT inactivators are not used, and it may attenuate the long-term side effects of O ⁶-alkylating agents in the bone marrow, which include myeloproliferative disorders and leukemias.

Our increasing understanding of the mechanisms of cell killing by O ⁶-alkylating agents has indicated the possibility that gliomas and malignant melanomas may be inherently resistant to a wide range of these and other anticancer drugs. Therefore, a concerted strategy might be required in which inactivation of MGMT is combined both with inhibition of other repair pathways involved in protection against temozolomide- and chloroethylation-induced toxicity and with abrogation of downstream drug resistance factors, e.g. by reactivating and enhancing apoptotic pathways.