Background
Cancer cells harbour genetic mutations and exhibit phenotypic traits that are not found in normal cells. These cancer-specific features may provide avenues for the development of targeted therapies that selectively kill cancer cells [
1]. The development of such strategies is a major focus of current cancer research. One such approach is synthetic lethality, which is when an acquired defect in a particular factor or pathway renders cancer cells sensitive to inhibition of another second specific factor. Still, this approach is often limited to a certain context or cancer type, for example, inhibiting poly (ADP-ribose) polymerase (PARP) in BRCA1- and BRCA2-deficient breast and ovarian cancers [
2,
3]. An alternative though related strategy that may be more effective in treating a wider range of cancers is cancer phenotypic lethality, which is to target factors and cell processes that are non-essential in normal cells but that become essential for cell growth following the acquisition of hallmark cancer traits [
1,
4‐
6]. However, until such pathways are identified and validated for therapeutic potential, radiotherapy- and chemotherapy-based treatments that are often associated with side-effects and resistance will remain the mainstay of treatments.
Oxidative stress, which arises when there is an imbalance between the production of reactive oxygen species (ROS) and the ability of a cell to counteract their levels or effects, is a hallmark cancer trait that can drive both carcinogenesis and continuing tumour evolution [
4,
7]. Paradoxically, ROS are the basis of much of the cytotoxicity of radiotherapy and several chemotherapy treatments. Hence, there may be several normally non-essential oxidative stress response factors and pathways that become ‘conditionally essential’ in cancer cells and/or significantly affect therapy responses.
ROS can react with all components of DNA to cause numerous types of lesions [
8]. However, the free deoxyribonucleoside triphosphate (dNTP) pool is reportedly 190–13,000 times more susceptible to modification than DNA [
9]. This suggests that a significant proportion of oxidative-stress-induced DNA damage arises via misincorporation of oxidised dNTPs during DNA replication rather than direct DNA modification. Oxidised DNA bases do not majorly disrupt DNA structure; however, they can subsequently lead to secondary types of DNA damage such as DNA mispairs [
10], DNA single-strand breaks (SSBs) that arise when DNA glycosylases remove damaged bases during base excision repair (BER), and DNA double-strand breaks (DSBs) [
11] that arise through poorly defined mechanisms possibly due to DNA replication stress. DSBs in particular are highly genotoxic and cytotoxic if not repaired correctly. This leads to the prediction that the pathways involved in preventing oxidised DNA base misincorporation could be critical in either promoting or suppressing cancer development and evolution depending on context [
12].
Mut T Homologue 1 (MTH1) is a Nudix hydrolase family enzyme member that hydrolyses selected oxidised dNTP and NTP substrates to the corresponding mono-phosphate products and inorganic pyrophosphate to prevent their misincorporation into DNA and RNA respectively [
13‐
15]. Primary substrates of MTH1 are dNTPs containing 8-oxo-7,8-dihydroguanine (8-oxoGua), one of the most common types of ROS-induced lesions [
16], and 2-hydroxy-adenine. Supporting the idea of a role in cancer cell maintenance, MTH1 levels are elevated in various cancers [
17‐
21], while lower MTH1 levels in U2OS osteosarcoma cells and non-small cell lung cancer (NSCLC) patient samples correlates with increased levels of DNA oxidation [
22,
23]. MTH1 overexpression in oncogene-expressing human cells promoted transformation [
11,
24,
25], while knockdown leads to DNA-replication-associated DNA damage response (DDR) activation and senescence [
11,
26]. Accordingly, the first developed MTH1 inhibitors appear to selectively inhibit cancer cell growth [
22,
27]. Collectively, these findings suggested that as cells undergo malignant transformation and acquire the trait of oxidative stress, MTH1 becomes essential for maintaining genome integrity and cell viability. This implies that targeting MTH1 activity could form the basis of a new targeted therapy strategy [
1,
28]. However, other recent data challenged these observations and conclusions. In these studies, MTH1 deficiency did not hinder the growth of HeLa, SW480 or U2OS cells, and highly specific MTH1 inhibitors displayed only weak cancer cell cytotoxicity [
29‐
31]. The title of the recent review, “MTH1 as a chemotherapeutic target: the elephant in the room” [
32], highlights the fact that the differing opinions and conclusions remain unresolved. Hence, it has become critically important to undertake further work to shed light on these contradictory findings and better understand the relevance of MTH1 in cancer and therapy.
Lung cancer is the leading cause of cancer death worldwide [
33]. Despite improvements in survival rates for many other cancer types in recent years, NSCLC therapy responses and patient outcomes have not significantly improved [
33,
34]. In our study, we addressed two main objectives to enable the assessment of the potential of MTH1 inhibition as a NSCLC targeted therapy strategy. First, we assessed if MTH1 deficiency alone is genotoxic or cytotoxic to several lung cell lines, and whether these effects were highly selective to NSCLC cells relative to normal cells. Second, we evaluated potential new combination therapy strategies by testing if targeting MTH1 enhanced the effects of current therapeutic agents. Thus, we tackle the currently opposing and contentious opinions on the significance of MTH1 in cancer biology and therapy [
1,
35].
Methods
Cell lines and chemicals
A549 (CCL-185; wildtype p53), H522 (CRL-5810; p53 mutation c.572delC, p.P191fs*56), H23 (CRL-5800; p53 mutation c.738G > C, p.M246I) and MRC-5 (CCL-171; wildtype p53) cell lines were purchased fully authenticated from ATCC (p53 mutation information provided at
https://www.atcc.org/Documents/Learning_Center/~/media/5F7B1CCACF724E3398BE56BFBEE3EFE4.ashx). MRC-5 and A549 cell lines were cultured in EMEM (ATCC) and DMEM-high glucose (ThermoFisher Scientific) media respectively, and H522 and H23 cells in RPMI 1640 medium ATCC modification. All media were supplemented with 10% (
v/v) FBS (ThermoFisher Scientific). Cells were cultured at 37 °C in a humidified atmosphere (95% air / 5% CO
2) and maintained at a low passage by not passaging beyond 6 months’ post resuscitation. Cisplatin, gemcitabine, VP-16 (etoposide), phleomycin, hydroxyurea and MTH1 small molecule inhibitors (TH287, TH588) were purchased from Sigma-Aldrich.
siRNA transfections
These were performed using DharmaFECT 1 reagent (GE Healthcare) according to manufacturer’s instructions. Briefly, a transfection complex was prepared by incubating together for 20 min at room temperature, 7.5 μl DharmaFECT 1 reagent, 125 μl Opti-MEM medium (ThermoFisher Scientific) and siRNA diluted in 125 μl Opti-MEM medium (final siRNA concentrations were 20 and 15 nM for H522 and remaining cell lines, respectively). 3 X 105 cells were plated with the transfection complex and incubated for 24 h in Opti-MEM medium, after which the transfection media was replaced with standard medium. MTH1-siRNA oligonucleotide 5′- > 3′ sequences (ThermoFisher Scientific) were sense CAUCUGGAAUUAACUGGAUtt and antisense AUCCAGUUAAUUCCAGAUGaa. Silencer Select Negative Control 1 siRNA (ThermoFisher Scientific) was used as scramble siRNA control.
Modified alkaline comet assay
DNA damage was assessed using Formamidopyrimidine-DNA glycosylase (Fpg)-modified comet assay that is a slight modification of the original method [
36]. Briefly, slides containing cells embedded in 0.6% low melting point agarose were incubated overnight at 4 °C in lysis buffer (2.5 M NaCl, 100 mM Na
2EDTA, 10 mM Tris-base, 1% Triton X-100, pH 10) (chemicals purchased from Sigma-Aldrich). Lysed cells were treated with Fpg (final concentration 0.8 U/gel) for 30 min at 37 °C, and subjected to alkaline electrophoresis in buffer containing 300 mM NaOH, 1 mM Na
2EDTA, pH 13. Following neutralization with 0.4 M Tris-base, pH 7.5, slides were stained with 2.5 μg/ml propidium iodide (PI) and dried at 37 °C. Comets were visualised at Χ200 magnification using an Olympus BH-2-RFL-T2 fluorescent microscope fitted with an excitation filter of 515–535 nm and a 590 nm barrier filter, and images were captured via a high performance CCDC camera (COHU MOD 4912–5000/0000). % tail DNA was calculated using Komet software (Andor Technology). For radiation treatments, the Xstrahl RS320 X-Ray Irradiator system was used to expose agarose-embedded cells on ice (assessments of immediate DNA damage) or cells in suspension that were then cultured for 24 h (recovery samples). In the MTH1 inhibitor experiments, cells growing in complete medium were treated with 10 μM TH287 or TH588 for 24 h prior to collection.
ROS level measurements
30,000 cells per well were seeded in triplicate in black 96 well plates (Porvair) and cultured for 24 h. Cells were washed with 200 μl PBS prior to the addition of 1 μl H2DCF-DA (ThermoFisher Scientific) and incubated in the dark for 30 min in a humidified atmosphere at 37 °C. 200 μl PBS was then added to all the wells, and the relative ROS-induced fluorescence intensities were measured immediately on a FLUOstar OPTIMA Microplate Reader (BMG Labtech; 485 nm excitation and 520 nm emission wavelengths). 30-min pre-treatment with 9.8 mM hydrogen peroxide was used for positive controls (relatively high dose used to overcome the scavenging of extracellular hydrogen peroxide by sodium pyruvate in the media [
37]). Samples without seeded cells used as blanks.
WST-1 cell proliferation assay
WST-1 is a water-soluble tetrazolium salt that is cleaved to a formazan dye in a mechanism mainly dependent on NAD(P)H production by metabolically active cells. 2 days after transfection, 1 X 104 cells (2 X 104 for H522) were seeded in triplicate for each sample in clear flat bottom 96 well plates, and left for 3 days before performing the assay according to manufacturer’s instructions (Sigma-Aldrich). Briefly, 10 μl Cell Proliferation Reagent WST-1 was added to each well containing 100 μl media and incubated for 30 min to 4 h. Absorbance values (that ranged between 0.5–2) were determined on an ELx808 microplate reader (BIO-TEK Instruments) at 450 nm against a blank control background. Cell proliferation (%) was determined by calculating (mean absorbance of sample / mean absorbance of control) X 100. 2-day VP-16 (Sigma-Aldrich) treatments were used as positive controls.
Annexin V/PI apoptosis assay
Double staining with annexin V-FITC (apoptosis marker)/PI combined with flow cytometry was applied as described in manufacturer’s instructions (Affymetrix). VP-16 was used as a positive control, while 4 untreated negative control cells were included for instrumental compensation and gating: no stain, PI only, annexin V-FITC only, and both PI and annexin V-FITC. Samples were analysed on a BD FACSCanto™ II flow cytometer (BD Biosciences) using BD FACSDiVa™ v6.1.3 software. At least 10,000 events were acquired per sample.
Western blot analysis
Standard techniques were used. Briefly, protein samples were prepared using Laemmli buffer lysis and sonication (15 s at 14 μm using Soniprep 150). MTH1, MTH2 and α-tubulin antibodies were purchased from Abcam, and CHK1 (2G1D5), phospho-CHK1 (Ser345), phospho-CHK2 (Thr68) from Cell Signaling Technology. Polyclonal secondary antibodies were horseradish-peroxidase-conjugated, and detection was performed using ECL substrate (Pierce) and X-ray film (CL-XPosure). Band intensities were quantified using densitometry GeneSnap or Image J 1.49 version software.
Data analysis and statistical tests
GraphPad Prism software (version 7 and 6.05) was used to calculate mean ± standard deviation (S.D) or standard error mean (SEM). Unless otherwise stated in the figure legend, data was evaluated by one-way ANOVA followed by post-hoc Tukey’s multiple comparison test to compare values between two or more groups. P-value < 0.05 was considered as statistically significant. The number of independent experimental repeats are indicated in each figure or figure legend.
Discussion
In this study we tested the potential of a new targeted therapy strategy for NSCLC, whilst simultaneously analysing opposing opinions within the field regarding the conditionally essential requirements for MTH1 in cancer cells and whether the current pursuit of MTH1 inhibitor development is likely to yield effective therapeutic agents [
1,
32,
35]. We show that MTH1 does indeed have a NSCLC-specific role for maintaining genome stability. The basis of this cancer-specificity remains unclear, as DNA oxidation levels in MTH1-deficient lung cells do not correlate with background ROS levels. This goes against the current model [
32], and suggests that perhaps the NSCLC-specific effect could be due to downstream defects in removing the oxidatively damaged DNA induced in the cancer cell lines [
56,
57]. Despite the functional role for MTH1 in NSCLC cells, we show that MTH1 deficiency ultimately does not cause NSCLC death, either alone or when combined with other therapeutic agents. One possibility for this could have been that the cell culture media used contained sodium pyruvate, a ROS scavenger. However, we do not believe this to be the case as sodium pyruvate scavenges extracellular ROS rather than intracellular endogenous ROS [
37,
58], and from what we can tell, other MTH1 studies that reported cytotoxic effects associated with MTH1 also used media containing sodium pyruvate [
11,
22,
29]. Ultimately, our work argues that MTH1 inhibitors will likely not be effective therapeutic agents. Instead, given that we show that MTH1 deficiency in NSCLC cells induces non-cytotoxic DNA oxidation and DDR alterations, we propose that treating NSCLC patients with MTH1 inhibitors could actually provide an environment for further mutation accumulation to drive cancer heterogeneity and evolution. In accordance with this proposition, MTH1-knockdown in human B lymphoblastoid cells induces a higher mutation rate but not cell death after UVA-induced oxidative stress [
59], while MTH1 overexpression repressed the DNA-replication-dependent mutator phenotype in mismatch-repair-defective colorectal cancer cells [
60].
The increases in oxidatively damaged DNA in NSCLC cell lines following MTH1 knockdown was relatively small (Fig.
2). However, the alterations in DDR signaling indicate that this was enough to disrupt DNA replication and lead to secondary types of DNA damage such as DSBs (Fig.
3). One proposed model for how this occur is that oxidised DNA bases induce DNA replication stress, which is defined as defective DNA replication fork progression [
61,
62], and that this somehow subsequently leads to DSBs [
22,
27]. It is possible that DSBs can arise from replication fork run-off at BER-induced SSBs, which would lead to the generation of one-ended broken DNA replication forks in a mechanism analogous to DSBs arising from Top1-DNA adducts [
63]. Alternatively, DNA replication forks may stall at sites of oxidatively damaged DNA and be cleaved by endonucleases such as Mus81 to also generate one-ended broken replication forks [
64,
65]. No matter how they arise, these one-ended DNA DSBs on replicating chromosomes would lead to Chk2 activation [
65], and are potentially highly genotoxic or cytotoxic as they may be very difficult to resolve. Concordantly, one-ended DNA DSBs are linked to various types of mutations [
66‐
69].
It is unclear why the DDR signaling alterations varied between the MTH1-defective NSCLC lines (Fig.
3), but given that different cancers already harbour many other mutations and potentially DDR defects, the signaling variances may simply reflect the differing abilities and deficiencies in DDR functions in different cancers. Furthermore, ultimately the ATM/CHK2 and ATR/CHK1 pathways are interlinked, as ATM-activating DSBs can subsequently lead to ATR activation if they are resected to generate ssDNA overhangs [
70], and processing of ATR-activating stalled forks can generate DSBs [
64]. The induction of phosphorylated CHK2 in H522 cells suggests that DNA DSBs arise. Why this should only occur or be detectable in H522 cells remains unclear. The reasons for decreased total CHK1 levels in A549 and H23 cell cultures were surprising. Although DDR activation following MTH1 knockdown was previously observed, including phosphorylation of H2AX, 53BP1, ATM and DNA-PKcs [
11,
22,
24,
27], to our knowledge this is the first time that a MTH1-knockdown-associated ‘switching off’ of the DDR has been observed. Induced deficiency of a DDR factor may indicate that MTH1 knockdown in A549 and H23 cells initially induces DNA replication fork stalling and ATR/CHK1 activation, but that the bulk of the cells efficiently turned off this cell cycle checkpoint signaling by lowering CHK1 levels to continue proliferating. Given that CHK1 levels were decreased within 4 days of MTH1 knockdown, this would not be enough for a mutation and clonal expansion to occur within the population, suggesting CHK1 suppression occurs through another mechanism that may involve changes in gene expression (epigenetic), RNA processing, post-translational modifications and/or proteosomal degradation. Accordingly, various stresses have previously been linked to CHK1 degradation [
71,
72].
There have been several contradictory and opposing reports on the cytotoxicity of MTH1 deficiency using various siRNA and shRNA sequences, cell lines and inhibitors [
11,
22,
24,
26,
27,
29‐
31,
38], which were recently summarized and compared in a review article [
32]. A critical finding of our work is, that despite MTH1 deficiency causing genomic instability in NSCLC cells and decreased H23 cell proliferation, there was a lack of cytotoxicity associated with MTH1 knockdown in all NSCLC cell lines (Fig.
5). A simple explanation for this finding is that the levels of MTH1 knockdown in our experiments were not sufficient enough to induce MTH1 deficiency. Or, as already discussed, other factors may be able to sufficiently compensate for MTH1 [
43]. However, we do not believe either of these possibilities is the basis of the disparities, as the 1.5- to 2-fold increase in oxidatively damaged DNA damage (Fig.
2) is comparable to that in other studies that did detect loss of cancer cell viability [
11,
22,
26,
38]. Also, we performed our experiments 4 days after transfection, which was before MTH1-proficient cells could take over culture (as confirmed by Western blot, Fig.
1). Hence, we suggest that the increased levels of genomic instability in MTH1-deficient NSCLC cells is not sufficiently high enough to induce cell death, rather it could promote further mutations and heterogeneity. Overall, our data therefore indicates that MTH1 inhibition will likely not be a successful therapeutic strategy for many NSCLC patients even when used in combination treatments. However, it remains possible that the effects of MTH1 deficiency vary considerably depending on circumstances. For example, MTH1 inhibition may be more effective on cancer cells that exhibit very high oxidative stress or particular possess mutations, and combining MTH1 inhibition with other specific agents or inhibitors (for example, Chk2 inhibitors) may prove to be selectively toxic.
The MTH1 small molecule inhibitors, TH287 or TH588, were proposed to be effective for cancer cell killing due to MTH1 inhibition [
22]. In our studies, TH287 and TH588 did induce apoptosis in 2 out of the 3 NSCLC cell lines tested, but this did not entirely correlate with increases in oxidatively damaged DNA levels (Fig.
6). This suggests that the effects on cell viability may have been distinct from MTH1 inhibition. Accordingly, the cytotoxicity of TH588 to melanoma cells was recently suggested to correlate to endogenous ROS levels but be independent of MTH1, as TH588 treatment induced melanoma cell death but MTH1 knockdown did not, and TH588-induced apoptosis is not rescued by overexpressing MTH1 or introducing the bacterial homolog of MTH1 that is not inhibited by TH588 [
73]. It was also recently proposed that TH287 or TH588 at the dose we used exert much of their cytotoxic effects through tubulin polymerisation defects [
30], though this conclusion was subsequently challenged as TH588 does not induce cellular changes commonly associated with Paclitaxel-induced tubulin defects [
38]. Repeating the TH287 and TH588 treatments at a lower dose of 3 μM, which does not induce tubulin polymerisation defects [
30], and over a longer time period may more specifically assess the consequences of TH287/TH588-induced MTH1 inhibition on NSCLC cells. Nonetheless, other highly specific MTH1 inhibitors were found to not be cytotoxic to cancer cells [
29,
31]. This not only supports the conclusion that MTH1 is not essential for NSCLC cell viability, but also strengthens the argument that MTH1 inhibitors may not make effective therapeutic agents.