Background
Lung cancer is the leading cause of cancer-related mortality, resulting in over 1 million deaths each year worldwide. This is mainly due to the difficulty of early detection and a lack of successful treatment methods, thus more effective therapy options are needed. Non-small cell lung cancer (NSCLC) accounts for approximately 85 % of all lung cancers, and standard chemotherapy for NSCLC includes pemetrexed (multitargeted antifolate, MTA; commercial name ‘Alimta’), as a single agent or in combination therapy (reviewed in [
1]). The combination of MTA with cisplatin was recently recommended as the gold standard therapy for adenocarcinoma lung cancer patients with good performance status [
2]. In addition, MTA-cisplatin combination therapy is also the recommended treatment regimen for malignant pleural mesothelioma (MPM) (reviewed in [
3].
MTA is a folic acid antagonist inhibiting the synthesis of precursor purine and pyrimidine nucleotides required for DNA and RNA synthesis. MTA thereby interferes with the proliferation and survival of replicating cancer cells. Prolonged treatment with MTA induces replicative stress in the form of single stranded DNA, which, if not repaired, can lead to the formation of double stranded DNA breaks [
4]. Cisplatin mainly induces intrastrand crosslinks and to a lesser extent also interstrand crosslinks. Repair of intrastrand crosslinks is mainly dependent on functional nucleotide excision repair, which is impaired upon nucleotide depletion [
5].
Induction of DNA lesions or DNA replication stress leads to the activation of the DNA damage response (DDR) (reviewed in [
6]). During DDR, phosphorylation of histone variant H2AX (γH2AX) serves as a key mediator for the assembly of DNA repair proteins at the sites of DNA damage as well as for the activation of checkpoint proteins. Consequently, analysis of γH2AX is frequently used as a surrogate marker for DDR activation. Prolonged cell cycle arrest after DNA damage induction results at the molecular level in DNA double strand break formation [
7] and at the cellular level in a terminal proliferation halt, i.e. senescence [
8]. By permanently arresting proliferation of damaged cells, senescence serves as a barrier to cancer development.
An increasing body of evidence suggests that epithelial-to-mesenchymal transition (EMT) plays a crucial role in the tumorigenesis, drug resistance, relapse and metastasis of various cancers (reviewed in [
9]). Diverse stimuli can lead to the activation of the EMT signaling cascade thereby inducing the expression of mesenchymal markers, e.g. vimentin. Furthermore, the expression of N-Cadherin is associated with acquisition of EMT phenotype whereas E-cadherin is associated with an epithelial phenotype (reviewed in [
9]). Cells can advance only partially through the EMT program, resulting in a mixture of epithelial and mesenchymal traits (reviewed in [
9]). For example in breast cancer, a hybrid epithelial/mesenchymal (hybrid E/M) state was identified at the single cell level, which was associated with increased mammosphere formation and poor predicted outcomes in all breast cancer subtypes [
10]. In the context of lung cancer, it was shown that IL-6-induced EMT increases adenocarcinoma tumor growth and metastasis
in vivo [
11] and we have recently shown that blocking EMT abrogates resistance to MTA in NSCLC [
12]. Mesenchymal cells are characterized by a loss of cell-to-cell contact and a spindle-shaped morphology (reviewed in [
13]).
Expression of NANOG, Sox2, CD44 is associated with stemness in various tissues and has allowed the identification of normal stem cells and subsequently also of cancer stem cells (CSCs; reviewed in [
9]. For lung cancer, CSCs were identified by means of numerous markers, e.g. drug-resistant side-population, CD133+, ALDH
high and EpCAM+ cells (for references, see [
14]). However, similar to the latest discoveries concerning the EMT status, more recent findings indicate that increased plasticity might also be present within cancer populations, enabling bidirectional interconvertibility between CSCs and non-CSCs (reviewed in [
15]).
In this study, we aimed to optimize the MTA-cisplatin anticancer modality and subsequently performed an in-depth molecular and cellular analysis to elucidate the molecular mechanisms underlying the observed benefit of sequential combination therapy. We demonstrated that prolonged MTA pretreatment improved the combination therapy’s efficiency. This effect correlated with the induction of persistent DNA damage, increased apoptosis and senescence initiation. The occurrence of resistant clones was thereby diminished, however those that did remain featured an epithelial-to-mesenchymal phenotype and were enriched for stem cell traits.
Methods
Cell culture and reagents
The NSCLC cell lines A549 (CCL-185) and H460 (HTB-177) were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured in Dulbecco’s’ modified Eagle’s medium nutrient mixture F-12 Ham (Cat. #D6421, Sigma-Aldrich, St. Louis, MO, USA), supplemented with 10 % fetal bovine serum (Cat. #10270-106; Life Technologies, Grand Island, NY, USA), 1 % Penicillin/Streptomycin solution (Cat. #P0781, Sigma-Aldrich) and 1 % L-Glutamine (Cat. #25030-024, Sigma-Aldrich) at 37 °C in a humidified 5 % CO2 incubator. Cell lines were previously DNA fingerprinted (Microsynth, Bern, Switzerland). Medium was changed every 3 days.
Pemetrexed/MTA (commercial name ‘ALIMTA’; Cat #VL7640) was purchased from Eli Lilly (Suisse) S.A. (Vernier/Geneva, Switzerland). Cisplatin (commercial name ´Cisplatin Ebewe´) was purchased from Sandoz Pharmaceuticals AG (Steinhausen/Cham, Switzerland).
Drug response and senescence associated β-galactosidase assay
To determine cell growth during the treatment and the initial recovery phase, 1×106 cells were seeded into 150 mm × 20 mm tissue culture treated plates (Cat. #20151, SPL Life Sciences Co., Ltd, Korea). Parallel experiments were performed in triplicate and samples were subsequently processed for flow cytometry as described below. Starting at the day after seeding, i.e. day 0, cells from one plate per treatment were harvested using TrypLE (Cat. #12604021, Invitrogen, Grand Island, NY, USA). Cell titers were determined using a hemocytometer and trypan blue (Sigma-Aldrich) (final concentration 0.1 %) for dead cell exclusion. The cells were washed in phosphate-buffered saline and processed for analysis by flow cytometry as described below. Resistant clones on recovery day 10 were counted on a centered surface of 25 cm2, using a 5 mm × 5 mm grid for orientation. To determine cell growth during the extended recovery period, cells were harvested at day 10 of the recovery period, reseeded at a density of 10’000 cells per 150 mm × 20 mm plate and cell titers were subsequently determined as described above. Experiments were repeated independently three times.
Senescent cells were visualized by using the senescence associated β-galactosidase assay (Cat. #20151, Cell Signaling Technology, MA, USA). In detail, 2000 cells were recovered at day 10 after each treatment and were seeded in tissue culture treated 6-well plates, fixed at day 17 and stained overnight according to the manufactures protocol. An inverted light microscope (Eclipse TS100, Nikon Instruments Inc., Melville NA, USA) equipped with a 10x objective was used for visual quantification of senescent cells. Experiments were repeated independently three times.
Gene expression analysis
Total RNA was isolated and purified with RNeasy Mini Kit (Cat. #74106, Qiagen, Hilden, Germany) and subsequent reverse transcription performed using High Capacity cDNA Reverse Transciption Kit (Cat. #4368814, Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s instructions. Conditions for reverse transcription: 10 min at 25 °C, 120 min at 37 °C, 5 s at 85 °C, hold at 4 °C. Quantitative real-time PCR (qRT-PCR) analyses were performed on a 7500 Fast Real-Time PCR System (Applied Biosystems) with commercially available TaqMan “Assay on Demand” primer/probes (Applied Biosystems) and TaqMan FastUniversal PCR Master Mix (Life Technologies). TaqMan primer list: E-Cadherin (Hs01023894_m1), N-Cadherin (Hs00983056_m1), Vimentin (Hs00185584_m1), CD44 (Hs 01075861_m1 CD44), Oct4B (Hs 00742896_s1 POU5F1), NANOG (Hs04260366_g1), CD133 (Hs 01009254_m1 PROM1), H3F3A (Hs02598544_g1), PP1A (Hs04194521_s1). qRT-PCR cycle conditions were as follows: 50 °C for 2 min, 95 °C for 10 min, 40 cycles of 95 °C for 15 s, 60 °C for 1 min. Gene expression was assessed by ΔΔCt values of the target genes versus control genes H3F3A and PP1A. Experiments were performed as three technical replicates and repeated independently three times.
Flow cytometry
For analysis by flow cytometry, cells were harvested as described above. Subsequently, cells were washed with phosphate-buffered saline, pH 7.4, fixed and permeabilized with Cytofix/Cytoperm solution (Cat. #554714, BD Biosciences (San Jose, CA, USA)). Staining with mouse Alexa Fluor 488 anti-γH2AX (Ser139) (Cat. #613406, BioLegend, San Diego, CA, USA) antibody was performed in phosphate-buffered saline (Pharmacy, University Hospital Bern, BE, Switzerland) supplemented with 0.5 % saponin (Sigma-Aldrich) and 1 % bovine serum albumin (Sigma-Aldrich) on a rotating wheel (3 rpm) overnight at 4 °C. Subsequently, cells were treated with 100 μg/ml RNase A (Sigma-Aldrich) and DNA was stained simultaneously with 0.5 μg/ml 4’,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich). Cell fluorescence was measured on a LSR2 upgraded flow cytometer (BD Biosciences) and analyzed using FlowJo V10 (Tree Star, Inc. (Ashland, OR, USA)). Flow cytometric analysis of the different time points was performed in batches, e.g. cells of the treatment period were analyzed first, followed by the early recovery phase (rec d1-3), recovery day 6 and 10 and at last cells of recovery day 14. Each analysis was accompanied by an untreated control. Approximately 5 % of the cells of the untreated controls were placed in the F/S-high compartment and used as normalization standard to compensate for variations in photomultiplier settings. Buffer treated controls were used to set the gating threshold for γH2AX to 5 % as describe before [
16]. When indicated, cells were labeled with 10 mM 5-ethynyl-20-deoxyuridine for 60 min and stained (Alexa Fluor 647) according to the manufacturer’s instructions (C35002; Life Technologies).
To compensate for slight shifts in linear DAPI fluorescence intensity due to treatment-induced changes in FSC/SSC signal intensity, gates for cell cycle distribution analysis were adjusted according to the peak of the G1 subpopulation.
Statistical analysis
Data are presented as the mean ± standard deviation of at least three independent experiments if not stated differently. Statistical differences were assessed using unpaired t test with Welch’s and P values < 0.05 were considered significant.
Discussion
A recent retrospective analysis of three phase III randomized controlled trials indicated that concomitant MTA-cisplatin combination therapy should be considered as the gold standard treatment for advanced adenocarcinoma (reviewed in [
1]). The present
in vitro study provides evidence that suggests that pretreatment with MTA prior to cisplatin is significantly more efficient than concomitant treatment with both drugs. These results are in agreement with the study of Kano et al., which revealed that exposure of A549 cells to MTA for 24 h and subsequent cisplatin treatment for 24 h followed by a 72 h recovery phase synergistically reduced cell survival whereas the inversed treatment regimen or the simultaneous application of MTA and cisplatin resulted in antagonistic effects [
26]. Similarly, 24 h pretreatment with MTA sensitized A549 cells to a subsequent treatment with a histone deacetylase inhibitor and increased the survival-benefit of the combination treatment in a patient-derived lung cancer mouse xenograft model [
27]. In addition, MTA pretreatment followed by treatment with a protein kinase Cβ inhibitor was shown to synergistically reduce cell growth of four NSCLC cell lines whereas the inverse treatment schedule had an antagonistic effect [
28]. Thus, the present results are in agreement with the general consensus in the literature indicating that MTA pretreatment sensitizes NSCLC cells to a variety of cytotoxic drugs. The present study provides for the first time an in-depth analysis of the effects on NSCLC cells of the combined treatment of MTA and cisplatin over an extended recovery period.
24 h after the end of the treatment phase, cell numbers were ~5-fold lower after treatment #1 compared to treatment #3. Although a short-term recovery analysis would therefore suggest treatment #1 to be the most efficient anticancer treatment, treatment #3 was more effective over a more extended recovery period. Tumor growth in animal models is usually monitored over periods of several weeks. However, further studies will be needed to demonstrate the superior anticancer efficiency of treatment #3 in vivo.
Controversial results concerning the effect of cell density, e.g. gap junction intercellular communication, on survival of cancer cells after treatment with DNA damaging agents including cisplatin are reported in the literature [
29‐
31]. At the start of the 24 h cisplatin treatment, cell density was 2.7-fold higher after extensive pemetrexed pretreatment (treatment #3, day 2) compared to treatment #1 (day 0) (Fig.
1b). However, cell density was still >4 times below full confluence (Fig.
1b, untreated control, day 3), which is associated in NSCLC cell lines with DNA damage resistance [
29]. Hence, an alternative experimental setting will be necessary to elucidate if survival after pretreatment with MTA prior to cisplatin challenge is significantly altered by variations in cell culture density.
While increased senescence was observed after treatment #1, as demonstrated by increased β-galactosidase activity, cell size and granularity, these effects were much more pronounced after treatment #3. Consistent with these findings of senescence induction, it has been shown that MTA-treated malignant mesothelioma cells also undergo accelerated senescence [
32]. Interestingly, the same study revealed that conditioned media from senescent MPM cells triggered the emergence of EMT-like, clonogenic and chemoresistant cell subpopulations. Senescent cancer cells can be cleared by the immune system, however the role of senescence in cancer progression is still debated (reviewed in [
17]). Senescence serves as a physiological barrier against tumor initiation and progression. On the other hand, senescent cancer cells might be able to overcome their dormant state representing a dangerous potential for tumor relapse. Thus, it will be crucial to utilize an immune competent animal model to determine the schedule-dependent anticancer efficiency of the combination therapy.
We observed that the three tested treatment regimens predominantly induced senescence rather than apoptosis. In contrast, Yang et al. [
33] showed that 60 % of A549 cells became apoptotic after treatment with 1 μM MTA for 72 h. Of note is that, the present study estimated the fraction of apoptotic cells based on the sub-G1 DNA content determined by flow cytometry whereas Yang et al. performed their analysis by comet assay and quantification of caspase activity.
It is well-established that the cell cycle status plays a critical role in the efficiency of combination chemotherapy. Unperturbed cells are maximally sensitive to cisplatin treatment during late G1/early S-phase and least sensitive during peak DNA synthesis (reviewed in [
34]). We showed that treatment with 1 μM MTA alone for 24 h resulted in the accumulation of cells in S-phase, which is in agreement with a previous study [
33]. At the start of the 24-h cisplatin treatment, the fraction of F/S-low cells in S-phase was highest in treatment #3 (day2) compared to treatment #2 (day1) and treatment #1 (untreated). The effect was even more pronounced in F/S-high cells. In agreement with our observation, Yang et al. also reported a significant S-phase arrest of A549 cells after treatment with 1 μM MTA for 48 h [
35]. Thus, we observed increased cisplatin sensitivity upon MTA-induced S-phase arrest, suggesting that it is not the S-phase status
per se but is more likely to be the MTA-induced perturbation of DNA synthesis, which sensitizes cells to subsequent cisplatin treatment. Indeed, it has been reported that cells blocked at the G1/S-boundary remained sensitive to cisplatin after release of the block [
36].
Comprehensive analysis revealed that 10 days after treatment #1, the cell cycle distribution of the F/S-low subpopulation was similar to the untreated control whereas cells were mainly arrested in S/G2/M-phase after treatment #3. Notably, at day 14 of the recovery phase, all F/S-low subpopulations displayed normal cell cycle distribution after the three treatment regimens although this fraction was significantly lower after treatment #3. In summary, during the extended recovery period, a subpopulation of cells in all three treatment groups overcomes cell cycle arrest and successfully completes mitosis, as also indicated by the presence of proliferating clones at later recovery time points. Thus, this small fraction of cells is resistant to even the most efficient treatment regimen tested in this study, even though it is smaller in number than for other treatments.
It has been previously shown that aneuploidy is associated with poor prognosis in NSCLC [
37,
38]. Interestingly, 14 days after treatment #1, a significant fraction of (p53-proficient) A549 cells acquired abnormal chromosome content, i.e. became aneuploid, which was less pronounced after treatment #3. We previously observed that induction of persistent DNA damage in p53-deficient HCT116 cells due to REV3 inhibition results in prolonged mitosis and subsequent return to interphase without cell division resulting in aneuploidy [
23], which was not observed in p53-proficient HCT166 cells nor A549 cells, independently of p53 status. Thus, results from the present study suggest that concomitant MTA-cisplatin treatment might have an adverse effect on the protective function of p53 upon the induction of aneuploidy. Treatment #3 not only significantly reduced the number of surviving cancer cells but also decreased the fraction of aneuploid cells in the remaining, therapy-resistant population.
Treatment of A549 cells with 1 μM MTA for 24/48 h led to a time dependent increase in H2AX phosphorylation, which is in agreement with the literature [
33,
35]. Simultaneous MTA-cisplatin treatment considerably increased H2AX phosphorylation. Interestingly, after MTA pretreatment for 24 h (treatment #2, day1), subsequent cisplatin-MTA treatment for 24 h (treatment #2, day2) did not further increase H2AX phosphorylation levels compared to MTA treatment alone (treatment #3, day2). Furthermore, after MTA pretreatment for 48 h (treatment #3, day2), subsequent cisplatin-MTA treatment for 24 h (treatment #3, day3) resulted in decreased H2AX phosphorylation levels in F/S-low and also F/S-high cells independently of the cell cycle status. In this context, it has been shown that nucleotide depletion leads to sequestration of replication protein A (RPA), thereby restricting its availability for nucleotide excision repair [
5], the primary DNA repair system processing cisplatin adducts. Thus, we hypothesize that MTA-induced nucleotide depletion induces the sequestration of RPA to single stranded DNA formed at stalled replication forks. Insufficient RPA levels limit nucleotide excision repair of cisplatin adducts and subsequent H2AX phosphorylation [
21]. Additional cisplatin treatment may overload molecular pathways required for the maintenance of the high γH2AX levels induced by MTA pretreatment, but further experiments will be required to elucidate the exact molecular mechanisms involved.
An initial increase in γH2AX phosphorylation was observed during the recovery phase after treatment #3. In this context, it has been shown that reduced nucleotide levels in combination with cisplatin treatment lead to nucleotide excision repair-mediated single-strand gaps, which are likely to be converted into double-strand breaks in the subsequent S-phase [
39]. Similarly, treatment with 5-Fluorouracil leads to the incorporation of 5-Fluorouracil and uracil during S-phase generating DNA repair-dependent, persistent DNA strand breaks during the successive G2/M-G1-phase, thereby interfering with the replication machinery in the subsequent S-phase [
40]. Thus, the increased levels of H2AX phosphorylation at the extended recovery time points after treatment #3 might be due to the persistence of complex DNA damage or repair intermediates. In this context, it has been shown before that the persistence of H2AX phosphorylation 24 h after cisplatin treatment correlated with the loss of clonogenic potential [
41]. In summary, the analysis of H2AX phosphorylation levels have provided a first insight into the molecular mechanisms underlying the increased efficiency of the combination therapy after prolonged MTA pretreatment. Further studies will be necessary to elucidate the exact nature of the resulting DNA damage after prolonged MTA pretreatment. It has been reported that cell lines contain subpopulations, which can be identified based on various features. For example, from parental A549 cells, a subpopulation with increased sphere formation capacity has been identified based on co-expression of EpCAM/CD166/CD44 [
42]. The current experiments revealed the existence of three distinct subpopulations in parental A549 cells by virtue of their capacity to give rise to colonies with divergent morphologies, and we focused on resistant cells surviving the combination treatment. Although the absolute numbers of all the clone types decreased after combination therapy, type 3 clones characterized by an enlarged cell size and mesenchymal morphology were relatively enriched, most prominently by treatment #3. Formation of type 1&2 clones was almost completely abolished by treatment #3. Further analysis of the subpopulation surviving treatment #3 revealed not only the expression of EMT markers but also that of stemness genes was increased indicating that the resistant clones have characteristic features associated with hybrid-E/M state. In agreement with these findings, it has been shown that cisplatin-resistant NSCLC sublines displayed a putative stemness signature with increased expression of CD133/CD44 compared to the corresponding parental cells [
43]. Interestingly, expression of stem cell markers NANOG, Oct4 and Sox2 were significantly upregulated in those cells as were the EMT markers c-Met and Beta-catenin, which is in agreement with a hybrid E/M state. Further, a hybrid-E/M state was also identified in tumors of different origin, e.g. breast cancer [
10] and clear cell renal cell carcinoma [
44]. Since the hybrid-E/M state was associated with poor survival in breast cancer [
10], it is tempting to speculate that the hybrid-E/M state might allow tumor cells to more readily differentiate either to a more mesenchymal phenotype (associated with invasiveness, metastasis and chemoresistance) or an epithelial phenotype (associated with metastatic colonization/proliferation) [
45] depending on the demands of the tumor microenvironment [
44]. A recent publication revealed the existence of a hybrid-E/M state in various lung adenocarcinoma cell lines [
46]. Lung adenocarcinoma cell lines containing a significant fraction of hybrid-E/M cells were highly invasive despite gene and protein expression of epithelial markers. It was proposed that the hybrid-E/M state is controlled by a regulatory network consisting of mir-200/Zeb and Snail/miR34 [
47]. Thus, our study adds to the growing body of literature identifying the hybrid-E/M state and its regulatory switch as a potential therapeutic target for cancer therapy (reviewed in [
48]). From a pragmatic point of view, we established in this study a protocol to efficiently select a subpopulation consisting of highly resistant lung cancer cells characterized by a hybrid-E/M state, which, at least in breast cancer, is associated with poor survival. Additional experiments will be necessary to elucidate whether the increase in cells featuring a hybrid-E/M state is due to a treatment-induced selection process or if the treatment regimen induces gene expression changes in clone type 3 cells, e.g. increased expression of stem cell genes.
Our study was limited to the analysis of the cell line A549 containing an activating mutation of the KRAS oncogene. In lung adenocarcinoma, oncogenic KRAS mutations are highly prevalent (~25 %) but therapy choices are very limited (reviewed in [
49]) indicating that our findings might be of relevance to improve treatment of a significant fraction of lung adenocarcinoma patients. However, further analysis of cell lines and primary cultures containing alternative mutational signatures will be necessary to evaluate if these findings are also of relevance for different lung cancer subsets.
Currently, MTA is administered as a daily 10-min infusion, which results in a relative rapid clearance from the body. However, MTA is rapidly converted intracellularly to its active form by polyglutamation. Thus, active MTA polyglutamate derivatives are sustained in tumor cells long after the MTA blood levels have declined, explaining the clinical efficiency of the initial phase I trials with single-agent MTA therapy, which was administered on day 1 every 21 days for up to 6 cycles (reviewed in [
50]). The present study reveals that prolonged MTA pretreatment increases the efficiency of the combination therapy
in vitro. Hence, it is tempting to speculate that a delayed cisplatin administration might also increase the efficiency of the MTA-cisplatin combination therapy in the clinical setting.
Conclusions
The present study has revealed that the efficiency of the MTA-cisplatin combination therapy can be augmented in vitro by optimizing the treatment schedule. The increased efficiency after prolonged MTA pretreatment is attributed to the induction of persistent DNA damage, which in turn results in increased apoptosis and senescence initiation thereby ultimately diminishing the occurrence of resistant clones. Therefore, our study warrants further experiments to elucidate whether an optimization of the standard therapy schedule might also potentiate the current combination treatment regimen in vivo. In addition, the remaining, therapy resistant cells not only exhibited a mesenchymal phenotype but also featured an increased expression of stem cell markers, both distinct hallmarks of a hybrid-E/M state. Thus, our research is the starting point for future in-depth analysis of this therapy-resistant subpopulation in lung cancer.
Authors’ contributions
CCT, RP, RAS and TMM substantially contributed to the conception and design of this study. CCT, LF, PD and TMM carried out the molecular biology studies and performed the data acquisition. CCT and TMM participated in the interpretation of the data and drafted the manuscript. RP and PD participated substantially in the interpretation of the data and critically revised the manuscript for important intellectual content. RAS was involved in drafting the manuscript, revising it critically for intellectual content and has given final approval of the version for publication. All authors read and approved the final manuscript.