Persistence of DNA adducts, hypermutation and acquisition of cellular resistance to alkylating agents in glioblastoma

ABSTRACT

Glioblastoma is a lethal form of brain tumour usually treated by surgical resection followed by radiotherapy and an alkylating chemotherapeutic agent. Key to the success of this multimodal approach is maintaining apoptotic sensitivity of tumour cells to the alkylating agent. This initial treatment likely establishes conditions contributing to development of drug resistance as alkylating agents form the O⁶-methylguanine adduct. This activates the mismatch repair (MMR) process inducing apoptosis and mutagenesis. This review describes key juxtaposed drivers in the balance between alkylation induced mutagenesis and apoptosis. Mutations in MMR genes are the probable drivers for alkylation based drug resistance. Critical to this interaction are the dose-response and temporal interactions between adduct formation and MMR mutations. The precision in dose interval, dose-responses and temporal relationships dictate a role for alkylating agents in either promoting experimental tumour formation or inducing tumour cell death with chemotherapy. Importantly, this resultant loss of chemotherapeutic selective pressure provides opportunity to explore novel therapeutics and appropriate combinations to minimise alkylation based drug resistance and tumour relapse.

KEYWORDS:

• glioblastoma
• O⁶-methylguanine
• cellular resistance
• mismatch repair
• MGMT
• DNA adducts
• DNA repair
• alkylating agents

Introduction: A pharmacological paradox

Alkylating agents have been used for decades by the biomedical community to induce tumours in experimental animals and to treat patients with tumours. Insights into the factors that determine this biological outcome are relevant for cancer prevention and cancer therapy.¹ What is becoming apparent is the molecular underpinnings of this dual and contrasting influence are remarkably similar for tumour induction and tumour ablation. Understanding this commonality may provide key insights to aetiology of alkylation resistance and also enhance our knowledge of tumour recurrence post radiotherapy. Specifically, there are 4 relevant major observations:

For the purpose of tumour induction in animals the objective is to facilitate the methylation of the guanine bases in DNA in the 6 position (to form O⁶-MeG) inducing mispairing with thymine, generating subsequent mutations. At a suitable dose for example in animals, this will lead to cancers of the colon several months after the administration of the alkylating agents (azoxymethane or dimethylhydrazine). Tumours can be induced in a dose-dependent manner in a variety of other animal tissues.² In those tissues sensitive to the alkylating agents, a correlation between the persistence and formation of O⁶-MeG adducts and the risk of subsequent appearance of tumours has been observed.^3,4 Here the tumour yield was shown to be associated with the cumulative amount of O⁶-MeG present in the DNA and the accumulative increase in cell proliferation.³

In contra-distinction to animal tumour inducing studies, the treatment of human tumours, for example high and low grade gliomas, also has as a core objective, the facilitation of methylation of guanine bases within DNA with a view to induce cellular apoptosis as a consequence of mispairing of O⁶-MeG with thymine. A number of alkylating agents including temozolomide (TMZ), dacarbazine and bifunctional alkylating agents have been used clinically. Adduct formation can occur for example with TMZ, a key part of the standard Stupp treatment regimen.⁵ One of the key clinical challenges faced in this approach has been the gradual development of tumour resistance with subsequent tumour regrowth with a median 2.5 month survival benefit and 2 year survival rate of 26.5% .⁶ In this context; the contemporary research focus has been upon the role and characteristics of the DNA repair protein, O⁶-methylguanine-DNA methyltransferase (MGMT) potentially as a site for drug resistance as well as deficiencies in the mismatch repair (MMR) pathway.

An additional anomaly relates to MGMT that removes the methyl group from O⁶-MeG and restores the integrity of guanine bases in DNA. Thus it can be predicted that a decreased influence of MGMT would enhance the effectiveness of the alkylating agent. However it could also be argued that the greater the effectiveness of this enzyme in curtailing the persistent presence of O⁶-MeG, the lower the probability of mispairing and associated generation of unwanted mutations. Treatment of MGMT inactivated glioblastomas with an alkylating agent frequently induces a hypermutator phenotype.⁷ Tumours that followed an evolutionary path to high grade glioma at recurrence had a signature consistent with alkylating agent induced mutagenesis.⁸ If these mutations are a consequence of persistence of O⁶-MeG, it could be argued that MGMT mediated removal of adducts will lower the probability of thymine mispaired mediated mutations. Then it could be concluded, that the greater the expression of MGMT the fewer the unwanted mutations. Indeed in the rat colon, inhibition of MGMT increases the number of tumours after administration of an alkylating agent.⁹ Furthermore, elimination of MGMT expression in knockout mice showed a higher frequency of colon cancers in response to an alkylating agent (AOM), than did the wild type.¹⁰ In contrast in the human, the reverse appears to be the case and it would appear that decreased tissue expression of MGMT is protective and this lowered expression has been a target for reducing drug and or cellular resistance.¹¹ Regardless it does draw focus to the relationship between the effectiveness of cellular DNA adduct management and MGMT expression.

An aligned counter intuitive feature relates to the induction of apoptosis that occurs in animal experiments. In the clinical setting this is a relatively straight-forward goal and apoptosis is a consequence of the administration of an alkylating agent. However we have shown that in the rat colon the expectation is reversed. After alkylation with doses of the agent similar to those employed for inducing tumours, the genomic pathways associated with cell cycle and DNA damage and repair processes are disrupted in favour of apoptosis.¹² collective processes that either inhibit tumour formation or help facilitate tumour removal.

Finally, the pharmacological regimen of administration (dose, frequency, and route) of alkylating agents to either induce tumours in animals or as chemotherapeutic agents in humans also deserves comment. In experimental animal models, to ensure a reasonable tumour incidence, a common approach to inducing colon cancers involves injections (intraperitoneal or subcutaneously) of the alkylating agent at significantly spaced intervals of for example 2, 4 or 8 weeks.¹³ or once weekly injections for either 1, 5, 10 or 20 weeks.³ In contrast it has been demonstrated in TMZ resistant GBM cancer stem cell lines that a 7 d on 7 d off schedule showed the highest toxicity compared to other schedules examined.¹⁴ In humans, multiple dosing, at defined intervals, with schedules very different than those seen in animal experimentation, is a characteristic of the chemotherapeutic approach to treating tumours. The standard of care has been extensively documented.¹⁵ and has involved daily concurrent administration of TMZ and radiotherapy followed by cycles of maintenance adjuvant chemotherapy daily for 5 d every 28 d. It can be assumed then that the effectiveness of alkylating agents to either induce or ablate tumours shares a commonality in as much as there is a requirement for repeated presentation of the agent. However the key distinction is markedly different administration schedules. This conclusion also raises the possibility of time based tissue responses and interactions with repeated administration of alkylating agents being pivotal in either tumour induction or tumour ablation. It could be argued that the timing and dose of the alkylating agents, and its relationship to radiation dose, volume and frequency is also very important. A more complete understanding of that relationship is likely to improve progression free survival.

Lastly it seems possible that the tumour promoting characteristics employed by biomedical researchers are not in some way underpinning the resistance to treatment seen in the clinical setting. At its core an anomaly rests with the adduct-induced mispairing of O⁶-MeG with thymine to drive tumour formation in one setting and to destroy tumours in another. As highlighted by Ghosal and Chen.¹⁶ chromosomal aberrations and mutations can lead to uncontrolled cellular proliferation and development of cancer on the one hand and cellular senescence and cell death which prevents cancer development on the other.

The argument is developed below that it is the alkylating dose response relationships and the temporal based intervals between alkylation and DNA repair, which include dose, timing of repeat treatments, and interactions with radiotherapy dose that are critical. The unwanted adduct persistence over time provides coherence to these apparent anomalies and potential avenues for increasing pharmacological effectiveness. Inherent in this argument is the probable cellular linkage between drug resistance and tumour recurrence.

Current limitations in alkylation based therapy in human glioblastoma

Malignant gliomas have very poor prognoses with 5-year survival rates less than 5%.¹⁷ It has been known for some time that alkylating agents may produce second tumours in patients treated for primary cancer.¹⁸ and the recurrence of tumours and development of new tumours is a major contributor to cancer mortality. In glioblastoma there is acquired resistance to chemotherapeutic agents, summarized by Kohsaka and Tanaka.¹¹ Alkylating agents, including TMZ, are used in the treatment of malignant glioblastoma and can act by generating a methyldiazonium ion, which in turn methylates the 6 position of the guanine base in DNA. While several sites may be methylated on DNA bases, the adduct formation at the 6 position is the most critical leading to the formation of O⁶-MeG, a major mutagenic and carcinogenic adduct. The newly formed O⁶-MeG can mismatch with thymine and as a consequence activate the mismatch repair (MMR) enzyme pathway. Importantly, with O⁶-MeG formation this process enters a futile cycle leading to double strand DNA breaks and ultimately to the activation of caspase-3 and induction of apoptosis.^19,20

The MMR pathway is one of the major cellular processes associated with DNA repair.²¹ Defects in the MMR occurs in approximately 10–15% of all sporadic colorectal cancers, and a deficient MMR pathway results in inherited predisposition to colon cancer, also known as Lynch syndrome.²² MMR is a highly conserved cellular process involved in the detection and repair of mismatched base pairing, double strand breaks and induction of apoptosis and consists of 4 genes: MLH1, MSH2, MSH6, and PMS2. Modifications to the chromatin structure allow MMR proteins to gain access to DNA to enable detection of mismatched bases and subsequent repair. Mismatched bases are initially detected by the dimeric MSH2 and MSH6 complex, which then recruits MLH1 and PMS2 in an ATP-dependent manner.

There is a growing view that the repair enzyme MGMT, responsible for the removal of the alkylating group from the O⁶ position of the guanine base, is a key player when resistance to the alkylating drug occurs. It can be argued that by removing the methyl group the original integrity of the guanine base and its pairing is restored; thus escaping the strand breakage apoptosis cascade that effectively renders the alkylating agent less effective. It follows that removal of the alkylating agent generated adducts by MGMT acts to limit the effectiveness of the chemotherapeutic agent. Not surprisingly, and as highlighted by Johannessen et al.²³ increasing levels of MGMT expression correlate well with in vitro and in vivo glioma resistance to alkylating agents. Additionally, patients lacking MGMT expression (epigenetic silencing) benefit from alkylation treatment compared to those with MGMT proficient tumours.²³ Not surprisingly the properties of MGMT have come into focus in assessing resistance to therapeutic alkylating agents.

Role of MGMT in cellular resistance to alkylation therapy in glioblastoma

The MGMT protein is a principle defence mechanism against the cytotoxic, mutagenic and carcinogenic effects of exogenously derived DNA alkylating agents.²⁴ MGMT amount and activity is normally distributed in the population however it is differentially expressed in different patient's brain tumours. MGMT is expressed in many human tissues.²⁵ and its expression is believed to be over expressed in common human tumours.²⁶ Its expression is induced upon exposure to genotoxic stress.²⁷ Tumour cells that express MGMT display a resistance to alkylating agents.²⁸ and as summarised by Sarkaria et al.²⁰ in high grade glioma a deficiency in MGMT can increase the sensitivity to alkylating agents. It has been demonstrated using stem-like glioblastoma cells that TMZ induced cell death exhibited an inverse correlation with MGMT expression levels.²⁹ Moreover TMZ induced cell death correlated with the efficiency of knockdown of MGMT.²⁹ Consistent with the results of in vitro studies, epigenetic mediated decreases in MGMT expression by way of promoter region methylation is associated with greater survival in brain cancer.^30–32 and an independent indicator of better prognosis.³³ Interestingly MGMT promoter methylation status was more reliable a predictor of prognosis than MGMT protein or gene expression levels.³⁴

By way of contrast a variety of lines of evidence suggest a role of MGMT or its overexpression is not an exclusive characteristic of all cellular resistance to alkylation. For example when two malignant cell lines were made resistant with TMZ exposure, MGMT was not overexpressed.³⁵ Subsequently Perazzoli et al.³⁶ concluded that MGMT does not play a key role in TMZ resistance in either A172 or LN229 glioblastoma cell lines both of which have low MGMT expression. Furthermore, Perazzoli et al.³⁶ concluded that it does not play an essential part in SF268 or SK-N-SH cell lines both of which have high levels of MGMT expression. In animal experiments tumour yield after treatment with an alkylating agent was not related to cumulative changes in MGMT activity.³ Furthermore, elimination of MGMT expression in knockout mice was associated with lower, not higher, frequencies of alkylation induced aberrant crypt foci (ACF).³⁷ In addition as noted by Kohsaka and Tanaka.¹¹ the survival time of patients who have methylated promoters of MGMT is still short. Moreover TMZ sensitivity was not restored in TMZ resistant GBM patients with competitive substrate inhibition of MGMT.³⁸ Additionally the findings of Rivera et al.³⁹ suggest that MGMT methylation appears to be a predictive biomarker for radiation response. This raises the possibility that the genomic methylation protective influence is not selective just for the alkylating agents.

In summary, there is accumulating evidence for MGMT promoter methylation status correlating with responses of patients to TMZ. It is a strong prognostic marker in paediatric and adult patients with glioblastoma treated with TMZ.⁷ Furthermore, it can be concluded, that while there is considerable support for a role of MGMT in cellular resistance to alkylation therapy, that role may not be totally exclusive or act in tandem in its contribution to the overall cellular resistance. As indicated by Kohsaka and Tanaka.¹¹ there is a likelihood of other mechanisms also being involved in TMZ resistance. In this context four observations are noteworthy:

•	In animal experiments while tumour yield is not associated with changes in cumulative MGMT activity, it is associated with the cumulative amount of O6-MeG in DNA.3
•	A correlation between MGMT expression and the expression of the MMR complex has been demonstrated.36
•	In alkylation induced ACF in the colon, genetic ablation of both MGMT and MMR, independently suppressed ACF and in combination had a multiplicative effect.40
•	In the absence of MGMT expression, loss of MMR (MSH2) can promote TMZ resistance.41

Collectively these findings highlight the need to better understand the role of MGMT expression, capacity and function relative to the MMR apoptotic process for O⁶-MeG removal in the cellular resistance to alkylation.

The relative roles of DNA repair mechanisms, O⁶-MeG adducts and apoptosis

A great deal of our understanding of the protective roles of MGMT and apoptosis has come from studies in which the influence of exogenous alkylating agents on tumour formation has been explored. In pioneering studies on alkylating agents, Dedrick and Morrison.¹⁸ proposed that the intrinsic carcinogenic potency was related to total lifetime exposure to the active species. As highlighted earlier, the oncogenic impact of adduct formation can be reduced with DNA repair by MGMT. An additional and important decrease in the oncogenic impact of O⁶-MeG can be achieved by apoptosis. What is emerging is a view that MGMT mediated repair of DNA and cellular apoptosis acting in concert in a highly prescribed fashion.

In a previous study.⁴² it was demonstrated in the rat that three key hallmarks of the role played by MGMT and cellular apoptosis following the administration of an alkylating agent:

It can be concluded that the DNA repair mechanisms that are evoked after the administration of an alkylating agent are highly ordered and sequenced. Moreover the capacity to remove DNA based O⁶-MeG adducts rests primarily with apoptosis, at concentrations commonly employed to induce tumours in animals. It should be emphasized that under these conditions programmed cell death proceeds only after the MGMT repair process is saturated. Consistent with this view is the observation that administration of an alkylating agent to mice, deficient in MGMT leads to a 20 fold higher level of apoptosis in the colon, than was seen in the wild-type mice.⁴⁰

 These conclusions are consistent with the results of earlier in vitro studies that support a role of O⁶-MeG in triggering apoptosis by post replication MMR mediated DNA double strand breaks.^45–47 However, the failure of apoptosis to completely remove all of the O⁶-MeG DNA base adducts offers the potential for their cellular persistence during new rounds of DNA replication and cell division providing the setting for base transition mutations and enhanced mutation load Fig. 2.

Figure 1. A predicted influence of low and high doses of alkylating agent on adduct removal and apoptosis. With low dose of the alkylating agent, MGMT can facilitate DNA repair by effective removal of the DNA adducts within tumour cells. Under these conditions the enzymatic removal of adducts will not trigger the MMR process nor its attending apoptotic response in tumour cells (left hand panel). With higher doses of the alkylating agent the MGMT capacity is depleted, the MMR process activated and tumour cells containing DNA adducts removed by apoptosis. The potential for persistence of adducts with the higher dose of the alkylating agent is highlighted (right hand panel).

Figure 2. A predicted influence of high (left hand panel) and low MGMT activity or expression on apoptosis. High activity of MGMT will effect DNA repair by enzymatic removal of adducts with only a minor contribution of the MMR mediated apoptotic process and minimal removal of tumour cells. With a smaller influence of MGMT, the greater the capacity for activation of the MMR apoptotic process and the greater capacity for removal of tumour cells containing DNA adducts.

•	Firstly, MGMT effectively prevents the accumulation of O6-MeG adducts at low doses of alkylating agent without the necessity for the recruitment of apoptosis Fig. 1
•	Secondly, this MGMT mediated repair is rapidly saturated (within a few hours) as the doses of alkylating agent are increased to levels that are commonly employed to induce tumours. The paradoxical nature of this suicidal enzyme with unsustainable induction has been highlighted earlier.43 Importantly considerable time elapses after a single dose of the alkylating agent before enzyme activity is restored.42 In the mouse colon the MGMT recovery to pre-treatment activity is approximately 152 hours.3 At the higher doses of alkylating agent, the tissue response is predominantly one of apoptosis. Thus in the intact animal MGMT is fast responding but of low capacity, a feature entirely consistent with its suicidal enzymatic characteristics Fig. 1.
•	Thirdly at higher single doses of alkylating agent, despite the activation of the apoptotic processes, adducts persist for up to 36 to 48 hours after administration of the alkylating agent throughout a period when cell proliferation is active Fig. 2. The latter observation is consistent with the earlier studies of Hong et al.44 where they noted that in the rat colon after the administration of the alkylating agent AOM, apoptosis declines well before the peak DNA adduct level was reached. Moreover Jackson et al.3 demonstrated that the cumulative total O6-MeG levels and cell proliferation indices were associated with tumour induction in the distal colon.

Mutations in DNA repair pathways; a role of cellular persistence of O⁶MeG?

In a pioneering study Goth and Rajewsky described the persistence of O⁶-ethylguanine in rat brain DNA and its consequences in carcinogenesis with the agent ethyl-nitroso-urea.⁴⁸ In mouse models there is a positive correlation with cumulative O⁶-MeG amounts and tumour load in the colon following treatment with an alkylating agent.³ We have suggested previously that apoptotic deletions of cells with O⁶-MeG adduct enhances a tissues capacity to handle oncogenic lesions.⁴² What becomes critical is the kinetics of the retention of unrepaired O⁶-MeG adducts in relation to cell replication. We have shown by histochemical and mass spectrometric analysis that adducts are detectable in the cancer prone colon from 36 to 48 hours after single administration of the alkylating agent, a period when MGMT is ineffective and importantly cell proliferation active.^42,49 The significance of the latter findings relates to mutagenesis which can be a consequence of unrepaired O⁶-MeG being present at the time of DNA replication. It is the pairing with thymine in the next round of replication that establishes the GC>AT transition mutations. Mutations of this type are found in genes involved in human colon cancer tissues including the oncogene K-RAS⁵⁰ and the tumour suppressor gene.^51,52 It can be assumed that the greater the number of adducts entering this cycle the greater the possibility of a mutation occurring in cancer promoting genes. However in this context the significance of mutations in the MMR pathway genes are becoming evident.

By way of summary, the enzyme MGMT directly repairs DNA by facilitating the removal of the methyl moiety from O⁶-MeG and while this enzyme is effective at low doses of the alkylating agent, it is the apoptotic process that has the dominant role at higher exposures to these agents.⁴² The apoptotic response is dependent on the MMR process, which is triggered by O⁶-MeG perturbations during the first and second cell cycle after treatment.⁵³ This dependency is well illustrated in a variety of settings including human embryonic kidney cells that are deficient in a functional MMR. It was shown that expression of one of the key MMR genes, hMLH1, in those cells corrected the MMR defect, and resulted in these cells being about 100 fold more sensitive to killing by an alkylating agent than cells not corrected.⁵⁴ Consistent with this finding is the observation that the loss of the MMR process and its attending inability to signal apoptosis sensitizes animals to adduct induced colon cancer precursor lesions.⁴⁰

Importantly, inactivation of MMR confers tolerance to methylating agents.⁵⁵ As indicated by Noonan et al.⁵³ cells deficient in the MMR pathway are resistant to killing by the alkylating agents, become methylation tolerant and show an enhanced response to the mutagenic and carcinogenic effects of alkylating agents. Magnetic resonance imaging (for tumour growth measurement) and immunochemistry showed that glioblastomas negative for MSH6 had an increased growth rate with alkylation treatment.⁵⁶ Moreover it has been suggested that the tolerance to mispairing of O⁶-MeG with thymine is the presumed basis for resistance to TMZ.²⁰ Two additional considerations highlighted by Johannessen and Bjerkvig.⁴¹ deserve comment. Firstly MMR deficiency promotes TMZ resistance in the absence of MGMT expression.⁴¹ Secondly, the presence of a link between MGMT promoter methylation and a hypermutator phenotype consequent to MMR mutations occurs predominately in MGMT promoter methylated tumours, suggesting escape from the MGMT mediated alkylating therapy sensitivity is by virtue of selection for MMR deficiency.⁵⁷

As highlighted by Johannessen and Bjerkvig.⁴¹ alkylating agent induced mutations will leave O⁶-MeG:T mispairs undetected and increase the resistance to TMZ in GBMs. The possibility arises that in those tissues where the alkylated adducts persist with time, the chances of transition mutations and mutational load are enhanced. If those mutations impair the efficacy of the MMR pathway, then the combined responses of methylation tolerance, hypermutation and enhanced mutagenic sensitivity would be predicted Fig. 3.

Figure 3. A speculated series of predicted events in the alkylation mediated formation of drug resistance and hypermutation. With an initial dose of alkylating agent, sufficient to inactivate MGMT and stimulate MMR, apoptosis is activated removing DNA adducts in tumour cells. The potential for persistence of adducts that may induce transition mutations within the tumour cells is highlighted (left hand panel). As the dose schedules continue, so does the probability of persistent adducts promoting mutations within MMR gene. When this occurs, the tumour cells are potentially primed for clonal initiation (middle panel). Once primed with an ineffective MMR, subsequent presentation of the alkylating agent will no longer have the capacity to induce apoptosis and as such the tumour cells will display drug resistance. In addition the DNA adducts that otherwise would have been removed by apoptosis, now become available for transition mutations leading to hypermutation (right hand panel).

Repair based mutations mediated by the unwanted persistence of O⁶-MeG; the underpinning contributor to alkylation based drug resistance and hypermutation?

The loss of MMR was proposed by Fink et al.⁵⁸ to result in drug resistance by directly impairing apoptosis activation and indirectly by increasing the mutation rate through the genome. Subsequently Hunter et al.¹⁷ demonstrated somatic mutations in the mismatch repair gene, MSH6, in two gliomas exhibiting large numbers of somatic mutations recurrent after alkylating treatment. In a more recent study.⁸ on mutational analysis on recurrent glioma, Johnson et al used genome sequence analysis to understand to what extent mutations in initial tumours differ from subsequent recurrent tumours and how alkylating chemotherapy altered the mutational profile of recurrent tumours. They found evidence for somatic mutations in the MMR genes in hypermutated tumours that were not present in the original tumours. As summarised by Johnson et al.⁸ resistance to alkylating agents, in this case TMZ, occurs in part through acquired mutations that inactivate the MMR pathway and continued exposure can result in hypermutation. Hypermutation is also associated with glioblastoma arising from germline biallelic mismatch repair deficiency.⁵⁹ and as highlighted by Shlien et al.⁶⁰ high grade tumours of this type have massive numbers of substitution mutations, greater than all childhood and most cancers. Lynch syndrome (hereditary nonpolyposis colorectal cancer (HNPCC), is also the result of germline mutations in at least one of the genes involved in MMR, most frequently mutations in MLH1, resulting in microsatellite instability.^22,61 and as indicated by Kerr and Midgley.⁶² cells with defective MMR tend to accumulate DNA mutations at a higher rate. It has been shown by Ding et al.⁶³ that relapse in acute myeloid leukaemia is also associated with the addition of new mutations. As highlighted by Ding et al.⁶³ the clonal evolution is shaped in part by the chemotherapy to establish and maintain remissions and “eradication of the founding clone and all of its subclones will be required to achieve cures”.

There are two important considerations that can be drawn from studies demonstrating hypermutation. Firstly, the observations correlate with the lack of benefit of repeat alkylator seen in patients with glioblastoma who relapse and possibly the lack of benefit to date in the use of ‘targeted chemotherapy’, much of which is based on a gene profile on the original tumour. Secondly, mutations that inactivate the MMR pathway and the attending hypermutated state can be viewed as a predicted consequence of persistence of DNA adducts Fig. 3. As proposed by Hunter et al.¹⁷ mutational inactivation of MSH6 produces both resistance to the alkylating agents in gliomas and accelerated mutagenesis in resistant clones. The latter, a consequence of exposure to alkylating agents in the face of defective MMR.¹⁷ Collectively there emerges a possible unifying model and molecular basis for the development of cellular resistance to alkylating agents in glioblastoma, summarised schematically in Fig. 3.

Toward a unifying mechanism?

In describing a model for the induction of tumours by chemical carcinogens, Jackson et al.³ suggested the involvement of four factors. In particular, the formation of DNA adducts, the lack of repair of the DNA adducts, a decrease in cell loss due to apoptosis and a subsequent round of DNA synthesis with fixation of mutation at sites where the DNA adducts have persisted. It is the dose response combined with temporal considerations of these interwoven processes of alkylation, DNA repair and accelerated mutagenesis that provides a coherent explanation for resistance to alkylation and the attending hypermutation:

•	An argument can be developed to suggest that at lower concentrations of the alkylating agent MGMT is effective in repairing DNA adducts and under these conditions apoptosis plays a smaller role in DNA repair Fig. 1 The lower concentration of the alkylating could also include that interval during which time the drug levels are rising to a higher steady state concentration. By nature of the limited catalytic capacity of MGMT, as the concentration of alkylating agent increases, MMR mediated apoptosis, becomes dominant and as such facilitates the death of tumour cells Fig. 2 Not unexpectedly, under conditions of minimal silenced MGMT activity, the apoptotic response would be predicted to be greater than that seen than when MGMT has a higher activity. Fig. 2 In this setting MGMT methylation would be associated with longer mean survival and as such a prognostic marker. Of importance however and discussed earlier, was the observation that MMR mutations occur predominately in MGMT promoter methylated tumours. As MGMT inactivated glioblastomas treated with an alkylating agent may induce a hypermutator phenotype.7 under conditions of reduced MGMT expression, it is possible that escape from the MGMT mediated alkylating sensitivity is now by virtue of the selection for MMR deficiency.57
•	At higher concentrations of the alkylating agent, MGMT plays only a minor role in DNA repair and apoptosis dominates. Fig. 1 Under the conditions of higher drug concentrations, differences in MGMT activities will be of a lesser consequence and the effective removal of DNA adduct will be dependent to a greater extent upon the efficacy of apoptosis. It is predictable that the observed impact of MGMT activity (including its methylation status) to the alkylating agent will be strongly influenced the prevailing cellular concentration achieved for the drug.
•	We and others have established that in those tissues prone to tumour formation after exposure to an alkylating agent, the O6-MeG adducts formed persist with time. This persistence sets the stage for the mispairing of the O6-MeG adduct with thymine and transition mutations.Fig. 3 With repeated exposures to the alkylating agent, the potential exists for one of the presentations of the alkylating agent to embed mutations that lead to inactivation of the MMR processes. If this occurs the cells may be primed for a potential environment of high adduct load with minimal DNA repair capacity. The potential consequences of this setting will be concomitant drug tolerance, hypermutation. Fig. 3 as well as accelerated mutagenesis in resistant clones as a consequence of continued exposure to alkylating agents as described by Hunter et al.17

MMR mutations, hypermutations and the pharmacological paradox

Earlier in this review we highlighted the apparent paradox, namely that alkylating agents have been used to induce tumours in experimental animals as well as to destroy tumours in humans. Furthermore, both applications have at their core the significance of O⁶-MeG mispairing with thymine. We also highlighted the differences in pharmacological presentation of the alkylating agents to produce tumours in animal models on the one hand and to aid in their removal in humans on the other. It should be noted that in both settings the dosing schedules were arrived at empirically. It could be predicted in the animal studies, based on the discussion above, that single doses on the alkylating agent well-spaced (weeks) will enable an unimpeded persistence of adducts with time. This setting provides an optimal conditions for adduct induced transition mutations and the likelihood of hypermutation and tumour formation. In contrast in humans, regular (daily) presentation of the alkylating agent as a chemotherapeutic will maximise continuously the apoptotic response which will act to decrease the presence of adduct persistence and their availability for transition mutations.

In regard to the paradox, it seems reasonable to conclude that the ultimate expression and utility of the O⁶-MeG thymine mispairing in either tumour promotion, or tumour killing, is determined in part by the adopted intervals in the dosing schedule.

A role for combination therapies

There is a need to improve survival in glioblastoma patients. Despite the optimization in dosage regimens, resistance to the alkylating agent will eventually occur and as such provides a pharmacological challenge to prevent clonal evolution. Not surprisingly, in recent years many avenues involving combination therapies have been explored, with a common view to enhance the effectiveness of alkylating agents including TMZ. These approaches have been summarized extensively in recent reviews.^11,64,65 Moreover as highlighted recently.⁶⁶ a number of drugs (not related to the alkylating agents) in routine use may also have potential antineoplastic or radiosensitising properties.

Therapeutic targeting of MGMT

Based in part on the observation that tumour cells that express more MGMT are more resistant to TMZ treatment, therapeutic targeting of MGMT expression or enzymatic activity to improve the efficacy of the alkylating agent, or overcoming resistance of the agent has been studied.^7,11,64,65 The number of potential combination therapies being explored in either preclinical or clinical trials is quite extensive.

Is the persistence of DNA adducts a combination therapeutic target?

Based upon the discussion so far one avenue of pursuit will be to minimize further the probability of adduct induced mutations that impair the MMR pathway. In doing so, to preserve the apoptotic properties of the alkylating agents, and at the same time to overcome drug resistance and formation of the founding clones.

As discussed a potential contributor to drug resistance and hypermutation with an alkylation based chemotherapeutic approach are mutations in the MMR genes, mutations that are a consequence of cellular persistence the O⁶-MeG adduct. In theory, combination therapies have the potential to offset this mutational escape.

We have shown previously that with higher concentrations of an alkylating agent the first line barrier of defence, MGMT is completely depleted within a very short time leaving a dependence on apoptosis as the principle avenue for removal of cells with an adduct load by the MMR pathway.⁴² Despite the induction of apoptosis in animals, adducts may still be present in tissues up to one or two d after the administration of the alkylating agent and most importantly at a time when cell proliferation is occurring.^42,49 Of importance Jackson et al.³ have, based on animal experiments suggested tumour yield is associated with the cumulative amount of O⁶-MeG present in DNA over the treatment period. It follows that one approach is to reduce the persistence of adducts after alkylation and/or to inhibit DNA repair through mechanisms other than the MMR process. Intriguingly, similar challenges and considerations have come from studies on the aetiology of colon cancer, where the focus has been upon apoptotic based modulation of O⁶-MeG levels in the colon that have arisen from dietary insults.

The most commonly inherited colorectal cancer is that of Lynch syndrome (HNPCC) a consequence of germline mutations in genes associated with MMR with an attending microsatellite instability.⁶⁷ Support for the association between methylation tolerance and defective MMR in colorectal cancer cell lines has been provided.⁶⁸ The colon cancer cell line HCT116 exhibits microsatellite instability and displays a mutation in the hMLH1 gene.⁶⁹ and importantly is resistant to TMZ.⁶⁹ However these cell lines, despite resistance to the alkylating agents and gene mutations associated with MMR, are capable of undergoing programmed cell death. In particular HCT116 cells display an apoptotic response to the histone deacetylases (HDAC) inhibitor butyrate and its structural analogues.⁷⁰ One key aspect of studies on colon cancer has been focus upon apoptosis mediated by the short chain fatty acid, butyrate, with a view to decrease the persistent mutagenic influence of DNA alkylation.

Short chain fatty acids

The short chain fatty acid, butyrate induces apoptosis (through HDAC inhibition) and inhibits proliferation in colon cancer cell lines.^70,71 The pathways for activation of the HDAC response are well described.⁷² Butyrate is a colonic fermentation product of resistant starch (RS) and RS has been shown to protect against AOM-induced carcinogenesis in rodent models of colorectal cancer, summarized by Fung et al.⁷³ In particular it is believed that butyrate plays a major role in reducing tumour formation in the AOM treated rat.⁷⁴ Of particular interest is the observation that delivery of butyrate to the large bowel by means of high amylose starch enhanced apoptosis and reduced levels of O⁶-MeG after a single dose of the alkylating agent.⁷⁵

Of relevance in the context of glioma, is the antiepileptic drug valproic acid (VPA). It is a small branched fatty acid, a structural analogue of butyrate and like butyrate, displays HDAC inhibition and inhibits tumour growth and metastasis in rodents.⁷⁶ In particular in common with butyrate, VPA inhibits class I and class II HDACs.⁷⁷ A summary of some of the antineoplastic effects of VPA have been described recently.⁶⁶ The question arises; can VPA mediated apoptosis, stimulated independently of the MMR process and thereby retard cellular resistance to TMZ in glioma?

Not surprisingly there is considerable interest in this question and many preclinical and clinical studies evaluating VPA are in progress. Particular focus has been experimentally in the context of the potential enhancement of radiation therapy.⁷⁸ as well as a therapeutic addition to standard radiation therapy with TMZ.⁷⁹ A comprehensive review of the mechanisms of action of VPA has been described by Berendensen et al.⁸⁰ and includes the observation that VPA inhibits a subset of HDAC and cellular kinases. Moreover VPA inhibits DNA repair and as such potentiates chemotherapy or radiation therapy.⁸⁰ It has been observed that VPA is a neuroprotective agent in a variety of settings. Experimentally VPA acts as a radiosensitizer in brain cancer cells and it protects hippocampal neurons from radiation induced damage in cell culture and animal models.⁷⁸ Finally Nakada et al.⁶⁴ have highlighted other potential features of VPA. These include greater accessibility of anticancer drugs to DNA, p53 independent synergistic induction of apoptosis with TMZ, induction of autophagy independent of caspase and enhanced bioavailability of TMZ by reducing the clearance of the methylating metabolite.

The combination of VPA with chemotherapy and radiotherapy in glioblastoma would appear to provide a rational option for investigation and to assess efficacy in well-designed clinical trials.^80,81 Retrospective studies of GBM in children and adults showed a significant survival benefit associated with VPA.^15,82,83 In contrast in a prospective clinical trial in newly diagnosed glioblastoma, progression free survival and overall survival in patients taking VPA were not different from those without the drug.⁸⁴ However this meta-analysis omitted compliance, dose and timing of VPA therapy, with the result that the effect of VPA remains unknown.⁸⁵ At this stage the utility of combined VPA treatment with alkylation based chemotherapy and radiotherapy will await the outcomes of a randomised clinical trial currently underway (trial number: ACTRN12616001350415).

Our accrued knowledge and understanding of alkylation based cellular resistance should permit the targeting of agents for use as combination therapies. The objective being to preserve the core beneficial effects of the alkylating agents, to reduce the acquired cellular resistance and to enhance their effectiveness against aggressive tumours.

Summary

The acquisition of cellular resistance and the formation of a hypermutated state in recurrent glioblastoma provide a significant pharmacological challenge. At its core it is an appreciation of the interwoven characteristics of dose interval and temporal responses that occur with DNA alkylation. This includes the interplay between DNA repair, persistence in adduct formation, apoptosis, hypermutation and malignancy. By contrasting the key paradoxical characteristics of tumour induction in experimental animals, with the known characteristics of drug resistance and hypermutation in humans, two fundamental drivers become apparent. The first relates to the cellular significance of persistence of O⁶-MeG after alkylation of DNA and in its unrepaired prolonged state its ability to mediate transition mutations. The second relates to the pivotal role of MMR in drug resistance and hypermutation. It is the interaction between these two processes that is fundamental to the acquisition of cellular resistance and hypermutation.

Impairment of MMR function, through mutations mediated by persistence of the alkylation adducts should be viewed as a candidate responsible for the attending loss of chemotherapeutic selective pressure. The latter provides the stimulus to explore combination therapies, particularly those that target impairment of MMR function, to enhance the effectiveness of alkylating agents and reduce resistance for drugs used in treatment in glioblastoma.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgement

The authors would like to thank Professor Stuart Pitson and Saiful Islam for their valuable input during the final stages of the manuscript.