Background
Lung cancer (LC) is the tumor type with the highest number of cancer-associated deaths worldwide [
1]. LC is histologically categorized into non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC) of which NSCLC constitutes about 85 % of all cases and is further divided into adeno-, squamous cell- and large cell carcinoma [
1]. Surgery, if possible, is the treatment of choice for stage I, II and IIIa NSCLC with chemotherapy primarily being used as adjuvant or neoadjuvant treatment [
2]. For non-resectable or advanced NSCLC, which constitutes the majority of cases, multimodal chemotherapy alone or in combination with radiotherapy is the main treatment option [
2]. The chemotherapy regimen usually consists of a cisplatin or a carboplatin doublet combined with gemcitabine, vinorelbine, paclitaxel, pemetrexed or docetaxel [
2]. The primary mechanism of cisplatin action at clinically relevant doses is to induce DNA damage. This is achieved through covalent crosslinking of platinum to the cellular DNA, leading to the formation of crosslinks in the same DNA strand (intra-strand crosslink) or between the two different strands, so called inter-strand crosslinks, ICLs [
3]. Subsequently, the ICLs physically impede the progress of the replication fork and transcriptional machinery causing replication stress and blocked transcription process, leading to activation of the intra-S checkpoint, and if the lesions are too extensive, induction of cell death [
3].
Cisplatin resistance is still a major obstacle for the clinical management of NSCLC. At the molecular level, a cisplatin-refractory phenotype can be a result of: (I) failure to reach the DNA (pre-target resistance), (II) impeded induction of DNA lesions (on-target resistance), (III) malfunctioning of cell death pathways (post-target resistance), and (IV) activation of pro-survival signaling pathways that are not directly influenced by cisplatin, but abolish its death-inducing capacity (off-target resistance), reviewed in [
4].
Although the molecular mechanisms underlying cisplatin refractoriness have been investigated for over a decade, only two biomarkers that can predict cisplatin sensitivity and distinguish responders from non-responders have reached the clinic, excision repair cross-complementing rodent repair deficiency, complementation group 1 (
ERCC1) and ribonucleotide reductase M1 (
RRM1), respectively. NSCLC cases whose specimen lacked ERCC1 expression had a more prominent response to adjuvant cisplatin treatment and hence ERCC1 expression holds promise as a predictive biomarker. [
5]. Low RRM1 mRNA expression was linked to a better response to a cisplatin/gemcitabine regimen [
6]. However, neither
ERCC1 nor
RRM1 were correlated to cisplatin sensitivity when basal mRNA expression was analyzed in 12 NSCLC cell lines [
7] reflecting the complexity in finding biomarkers which can predict cisplatin responsiveness.
Other studies have aimed to characterize signaling cascades which could drive cisplatin-survival and hence constitute putative resistance-driving networks in lung cancer by focusing on short term effects of continuous cisplatin treatment i.e. from hours up to a few days, or by creating resistant sub-lines after repeated cisplatin pressure which also could generate new driving mutations [
4,
8]. In this study, we explored the intrinsic properties of the cisplatin-surviving sub-population of NSCLC cells 9 days after a single one hour-treatment. This treatment regimen was chosen to reflect the short pulse of drug used clinically, where administration time is typically 30 minutes to two hours (
http://www.cisplatin.org/treat.htm).
Using this approach, we found a heterogeneous gene expression pattern when analyzing three biological replicates of cisplatin-surviving NSCLC clones. Among the different biological replicates we identified genes in diverse cellular pathways in these cisplatin-survivors e.g. dickkopf-1 (DKK1), X-ray repair cross-complementing protein 2 (XRCC2), formin 1 (FMN1) and lectin, galactoside-binding, soluble 9 (LGALS9). Through bioinformatics analysis, we identified TCF4, EZH2, DNAJB6 and HDAC2 as co-regulated, upstream regulators of DKK1, which may form a signaling circuit that enhances the effect of DKK1 in enabling survival after cisplatin treatment. By siRNA-mediated knockdown of DKK1 in NSCLC and ovarian cancer cells, the colony forming capacity and/or cell survival upon cisplatin treatment was reduced significantly. In contrast, plasmid-based overexpression of FMN1 did not clearly increase cisplatin sensitivity of NSCLC cells. Thus our data suggest that DKK1 should be further explored as a potential biomarker of cisplatin refractoriness and/or as a target for cisplatin-sensitizing strategies in NSCLC and other tumor types.
Methods
Cell lines and culture conditions
In the present study human NSCLC cell lines U-1810 and U-1752 (gifts from Uppsala University, Sweden [
9]), A549, H23, H125, H157, H661 and H1299 (ATCC, Manassas, VA, USA) were used. Cells were cultured at 37 °C and 5 % CO
2 in RPMI-1640 medium containing 2 mM L-glutamine, supplemented with 10 % heat-inactivated fetal bovine serum (both from Invitrogen, Stockholm, Sweden). In addition, the human ovarian cancer cell lines A2780 and its cisplatin-resistant subline A2780 cis (Sigma-Aldrich, Stockholm, Sweden) were used and cultured as above. To maintain the cisplatin resistance of A2780 cis cells, 1 μM cisplatin was added to the culture medium every 3
rd-4
th passage. All cell lines used in the study were established and already published on (see above). No ethical permits were therefore required for their use in the current study.
NSCLC cells were seeded in duplicate in Cell + culture dishes (Sarstedt, Landskrona, Sweden) at a density of 500 cells/100 mm dish and were after 24 h treated with cisplatin (2.5-20 μM, Hospira Nordic AB, Stockholm, Sweden) for one hour. Cells were rinsed in PBS after treatment and allowed to form colonies over a 9-days period. The resulting colonies were visualized by staining with crystal violet (0.5 % crystal violet in 25 % methanol) or collected for RNA extraction (see below). For clonogenic survival analyses, colonies consisting of at least 50 cells were counted under a light microscope using duplicate plates from three independent experiments. For retreatment experiments, cell colonies were instead trypsinized and pooled, counted and seeded in 96-well plates for MTT or in new Cell + plates for treatment the next day using the same setup as in the first treatment.
RNA extraction and gene expression analysis
In order to have enough RNA for the gene expression analysis all the surviving clones from each biological replicate were pooled and subjected to total RNA extraction using Trizol (Invitrogen) as described [
10]. Cleanup was performed using the RNeasy Mini kit (Qiagen, Sollentuna, Sweden) and RNA quality was analyzed using an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA). Analysis of gene expression was performed using Affymetrix® whole transcript GeneChip® Human Gene 1.0 ST arrays (Affymetrix, Santa Clara, CA, USA), which contains probes for 28 869 genes. cDNA was prepared from 500 ng total RNA, labeled and hybridized to arrays using standard protocols (
http://www.affymetrix.com/support/technical/product_updates/wt_1_1_assay.affx). Primary array processing was performed using the Affymetrix GeneChip® Command Console® Software (AGCC, version 1.1) and subsequent analysis was conducted using the Affymetrix Expression Console (EC, version 1.1).
Post-acquisition data processing was carried out using previously described methods (
http://www.affymetrix.com/estore/browse/level-1-instruments-software-landingpage.jsp?expand=true&parent=35854&category=35919). Briefly, probe logarithmic intensity error estimation (PLIER) was used to enhance probe signals by summarization; perfect match GC composition-based background correction (PM GCBG) was applied for background correction and global median to normalize the signals. For further analysis, genes with signal intensity below 10 after background correction were excluded to avoid taking genes whose alterations are not easily distinguished from noise into subsequent analyses. In addition, genes corresponding to uncharacterized proteins, hypothetical proteins prefixed with the letters LOC, and small nucleolar RNAs (
SNORD) were also excluded from the analysis since in this study we aimed to focus on well annotated, protein-coding mRNAs. The raw data presented and used in this article is deposited in NCBI's Gene Expression Omnibus (GEO) [
11] as described in the Availability of supporting data section. Hierarchical clustering analysis was performed using Partek Genomics Suite version 6.6 (Partek Inc., St. Louis, MO, USA) in which clustering was based on rows and columns using Euclidean distance for row/column dissimilarity and average linkage as row/column method.
Quantitative real-time PCR (q-RT-PCR)
For the q-RT-PCR validation of gene expression data, 500 ng of the same RNA batch was used as template for cDNA synthesis using Reverse Transcription Reagents with random hexamer primers (Applied Biosystems, Stockholm, Sweden) as previously described [
12]. To quantify mRNA expression levels, cDNA, Fast SYBR®Green Master Mix (Applied Biosystems) and the following primers (
DKK1, forward: CGG GAA TTA CTG CAA AAA TGG AAT ATG TG, reverse: AAG CTT TCA GTG ATG GTT TCC TCA ATT;
XRCC2, forward: GGC GAT GTG TAG TGC CTT CCA TA, reverse: TTT CTT TCA AGG AAC TTC TAC CTT CAA GTC;
LGALS9, forward: AGC TCC AGT GGA ACC AGG TTT G, reverse: TCA TTT CCA CTG AAG CCA GTC TGA A;
ERCC1, forward: CTG CTT GTC CAG GTG GAT GTG AAA, reverse: GAT ACA CAT CTT AGC CAG CTC CTT GAG.
RRM1, forward: CCT ATG AGG GCT CTC CAG TTA GCA A, reverse: CCA GTC CCA TAG GTC TGT AGG AGT AAC;
18S, forward: GCT TAA TTT GAC TCA ACA CGG GA, reverse: AGC TAT CAA TCT GTC AAT CCT GTC C) (from DNA technology, Risskov, Denmark) or
FMN1 (cat.# QT01330315, Qiagen) were mixed in a final volume of 10 μl. The Fast PCR program was used on the ABI Prism 7900HT Sequence detection system (Applied Biosystems), which is initiated at 95 °C for 20 s, followed by 45 amplification cycles (95 °C, 1 s; 60 °C, 20 s). For each biological sample two technical replicates were used and the relative RNA expression obtained by applying the 2
−ΔΔCt method [
13] in which 18S rRNA was used as an internal control.
Immunoblotting
Proteins were extracted using RIPA buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.1 % Na-deoxycholate and 1 % NP-40. Thirty microgram of total protein was loaded onto ready-to-use 4-12 % Bis-Tris gels (NuPAGE, Invitrogen), separated by electrophoresis and thereafter blotted onto nitrocellulose membrane (Trans-Blot, Bio-Rad, Hercules, CA, USA). After blocking in Odyssey blocking buffer, diluted 1:1 with TBST (LI-COR Biosciences, Lincoln, NE, USA), primary antibodies recognizing phosphoserine 9 GSK3B, phosphoserine 473 AKT, total AKT and PI3-kinase (5558, 9271, 4685 and 4257, respectively, Cell Signaling Technology, Danvers, MA, USA), p21WAF1/Cip1 or Bcl-2 (sc-756 and sc-509, Santa Cruz Biotechnology, Dallas, TX, USA) was added. To control for loading differences, GAPDH (ab9484, Abcam, Cambridge, UK) or β-tubulin (Sigma-Aldrich) was used. To visualize primary antibody binding on the membranes, secondary goat-anti-mouse or goat-anti-rabbit antibodies directly conjugated to infrared dyes, IRDye (LI-COR Biosciences) were applied and resulting protein expression levels analyzed by the Odyssey®Sa Infrared Imaging System (LI-COR Biosciences).
Ingenuity Pathway Analysis
Ingenuity Pathway Analysis tool (IPA; Ingenuity Systems, Redwood city, CA) was used to create in silico interaction networks of DKK1 based on published, publically available data, showing direct upstream transcription regulators of DKK1 as well as proteins downstream of DKK1.
MTT cell viability assay
To assess cytotoxic response of cisplatin, MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium) cell viability assay was used in a 96-well format as previously described [
14]. Three technical replicates were made for each biological sample and assayed after a continuous exposure to cisplatin for 72 h. For NSCLC cells, 5 000 cells/well were used and in A2780 and A2780 cis experiments, 15 000 cells were seeded per well. Cell viability was assessed by adding the MTT reagent as indicated [
14] and is given as % of untreated cells whose viability was set to 100 %. For the NSCLC cells, cisplatin sensitivity was calculated using the area under curve (AUC) from the survival curve.
DKK1 siRNA transfection
To inhibit DKK1 expression in U-1810, A549 and A2780/A2780 cis cells, 50 nM siRNA against DKK1 (si1 = s22721: Sense: GCU UCA CAC UUG UCA GAG Att, Antisense: UCU CUG ACA AGU GUG AAG Cct; si2 = s22722: Sense: GGC UCU CAU GGA CUA GAA Att, Antisense: UUU CUA GUC CAU GAG AGC Ctt, Invitrogen) or non-targeting siRNA (NT, 4390843, Invitrogen) was added to the cells during 72 h (U-1810, A549) or 96 h (A2780, A2780 cis) using Dharmafect 1 (0.1 %) from Dharmacon (Thermo Scientific, Lafayette, CO, USA). Cells were subsequently detached and frozen for RNA extraction or were re-plated for cell death and signaling profiling analysis (collected 24-72 h after cisplatin exposure), for MTT or for colony formation capacity after cisplatin treatment.
Overexpression of FMN1 and assessment of cisplatin sensitivity
FMN1 was overexpressed in U-1810 cells by transfecting cells with the FMN1 open reading frame cDNA integrated in the pCMV6-AC-GFP plasmid (OriGene, Rockville, Maryland, USA), using Lipofectamine LTX reagent (Invitrogen, Germany). Briefly, U1810 cells were seeded in 6-well plates and transfected with 2 μg of pCMV6-AC-GFP FMN1 plasmid for 24 h. As a control, cells only treated with Lipofectamine were used. The next day, media was removed, and normal growth media (RPMI-1640) was added to each well for another 24 h. Western blot analysis was used to confirm the overexpression of FMN1 at the point of cisplatin treatment using a FMN1 antibody (Abcam, Cambridge, UK). To assess the effect on proliferation and cisplatin sensitivity, cells were seeded in 96-well plates (8000 cells/well), and the next day treated with indicated concentrations of cisplatin for 72 h. The cytotoxicity of cisplatin was determined with (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium) (MTT) assay as described above. Survival of cells is given by comparing the absorbance in treated cells relative to the absorbance in cells only treated with Lipofectamine. Three separate transfections were performed with triplicate technical repeats in the MTT. Data presented is the mean ± SEM.
Statistical analysis
Data given is the mean ± S.D. from three separate experiments, unless otherwise indicated. A two-tailed unpaired Student’s t-test was used. P<0.05 was considered for statistical significance.
Discussion
Platinum-based compounds e.g. cisplatin and carboplatin constitute the standard chemotherapy regimen for NSCLC. Unfortunately a large proportion of the cases display intrinsic resistance to these platinum drugs and for yet another fraction, a platinum-refractory phenotype typically develops during the treatment course [
40]. In this study, we aimed to identify molecular determinants which drives a cisplatin-refractory phenotype and hence could be used either as biomarkers of response or as sensitizing targets for cisplatin in NSCLC. Our approach of studying gene expression alterations in cisplatin-surviving NSCLC clones is different from previous reports using either very high, non-clinically achievable cisplatin doses in short term treatment schedules [
41] or tumor cell models of acquired resistance [
42]. The latter is mostly reported to result in resistance mechanisms involving up-regulation of membrane-associated drug efflux pumps such as ATP-binding cassette proteins and copper-extruding P-type ATPases [
4,
8].
In our three biological replicates only
FMN1 showed altered expression in all three replicates, illustrating that NSCLC cells surviving cisplatin pulse treatment have heterogeneous clonogenic survival capacity and gene expression patterns. A possible reason may be that few prominent long term effects are seen on the RNA level 9 days after cisplatin treatment. However, one interpretation of this outcome is that cisplatin treatment can result in the expansion of different resistant clones in different experiments. This clonal evolution hypothesis has been demonstrated after epidermal growth factor receptor (
EGFR)-ablative therapy, where a very low number of Kirsten rat sarcoma viral oncogene homolog (
KRAS)-mutated colorectal cancer cells emerged to become the dominant clone among the surviving cells [
43,
44]. Hence, we speculate that even small initial variations in cisplatin responsiveness can induce certain clones to become dominant. Optimally, if not limited by the minimal amounts needed for the analysis method, it would be interesting to analyze multiples of clones separately to explore their differences and heterogeneity further.
We and others have shown that a chemotherapy-refractory cancer stem cell phenotype is evident in certain NSCLC cell lines [
14,
45]. However, this sphere-forming capacity after enrichment in stem cell media was not found in the NSCLC cell line used in this study U-1810, suggesting that they might not contain an appreciable proportion of stem-like cells and that the heterogenic response of chemotherapy in this particular cell line likely is governed by other signaling cascades. We observed the same cisplatin response in clonogenic and MTT assays upon retreatment (Additional file
4), therefore we could verify that using our single-treatment setup, we were most likely only studying the primary effects in the surviving clones that were selected due to intrinsic refractoriness.
The one gene that was down-regulated in all three biological replicates was formin 1 (
FMN1), a protein which enhances formation of cell-cell adhesion [
16]. As cisplatin disrupts cell-cell adhesion before it induces apoptosis [
46], one may speculate that the fraction of cells with low
FMN1 expression may be less responsive to the adhesion-disruptive effects of cisplatin, and consequently survive. However, by overexpressing
FMN1 we were not able to sensitize NSCLC cells to cisplatin indicating that either FMN1 is not directly involved in regulating cisplatin sensitivity or it acts in concert with other signaling aberrations to confer survival advantage if down-regulated, which not is recapitulated when forced overexpression is used.
Analysis of each individual experiment revealed
DKK1,
XRCC2 and
LGALS9 as top scored differentially up-regulated genes in cisplatin-surviving clones from replicates 1 and 3, respectively. It is well documented that cisplatin treatment activates multiple DNA damage signaling cascades, and here we found an increased expression of
XRCC2, a member of the homologous recombination repair pathway, in cisplatin-refractory residual NSCLC clones. This up-regulation might be due to inherent properties of the cells, or alternatively, a selective pressure on the surviving clones to up-regulate proteins involved in DNA repair to withstand the damage. In line with our data, mouse embryonic fibroblasts deficient in
XRCC2 are reported to be hypersensitive to cisplatin treatment [
47], further pointing towards a connection between high
XRCC2 expression levels and cisplatin resistance. Albeit
LGALS9 has not yet been implicated in NSCLC or in a chemotherapy-refractory phenotype of other tumor cells, various galectins such as galectin-1 and -3 were reported to have a role in driving a chemotherapy-refractory phenotype [
48,
49].
Importantly, we demonstrate that
DKK1 has a role in the intrinsic platinum responsiveness of NSCLC, as siRNA-mediated ablation of
DKK1 sensitized NSCLC cells to cisplatin and reduced their clonogenic survival potential.
DKK1 is a secreted protein with dual anti- and pro-survival functions in different tumor types. For instance,
DKK1 may act as a tumor suppressor through inhibition of Wnt/β-catenin signaling and is reported to activate apoptosis in multiple tumor types e.g breast cancer, renal cell carcinoma, melanoma and choriocarcinoma [
50‐
53]. In head and neck cancer cells, decreased
DKK1 expression was associated with acquired cisplatin resistance [
42], whereas overexpression of
DKK1 in a glioma cell line sensitized these cells to DNA damaging agents including cisplatin [
54]. Some of these data are opposed to our study in which DKK1 was upregulated in cisplatin-surviving NSCLC clones and its knockdown conferred cisplatin sensitivity. These differences could possibly be attributed to tumor type specific divergences in signaling cascades, or in mechanisms of acquired cisplatin resistance. Our results of cisplatin sensitization from NSCLC were however validated also in ovarian cancer cells which were sensitized to cisplatin upon siRNA knockdown of
DKK1. Yet we could not sensitize the acquired cisplatin-resistant subclone A2780 cis at the level of knockdown achieved in our experiments. Our interpretation is that DKK1 regulates intrinsic cisplatin resistance, still it may not be the main driver of acquired cisplatin resistance.
Multiple studies have demonstrated an oncogenic role of
DKK1 in diverse tumor types such as multiple myeloma, hepatoblastoma, Wilm’s tumor and hormone-resistant breast cancer [
55‐
57]. Moreover, high serum level of DKK1 has been detected in patients with NSCLC and esophageal carcinoma where it was associated with tumor progression and poor outcome of these malignancies suggesting that
DKK1 in these tumor malignancies may have an oncogenic role [
18,
20,
58]. Using the cBioPortal for Cancer Genomics (cbioportal.org) [
59,
60] which integrates data from several databases including The Cancer Genome Atlas, we found that
DKK1 was altered at the level of either mRNA upregulation, mutation, homozygous deletion or amplification in a total of 6-9 % of lung adeno- or squamous cell carcinoma patients [
61,
62]. In the adenocarcinoma population, the mentioned alterations in DKK1 were also linked to a significantly reduced overall survival [
62], further supporting the importance of DKK1 in NSCLC. Yet it remains to be demonstrated if DKK1 regulate intrinsic cisplatin sensitivity
in vivo. Such studies could be performed using NSCLC patient-derived xenografts in mice. To demonstrate that DKK1 is a predictor of cisplatin refractoriness
in vivo in NSCLC patients is more challenging as it would require a biopsy of primary tumor and metastasis prior and post cisplatin treatment which is not a standard routine in our clinic at present. Hence a controlled clinical trial would be required in order to adequately address this issue.
Through bioinformatics analysis of
DKK1, we identified a number of putative transcription regulators of this gene. Specifically, ectopic expression of the Wnt signaling components TCF4 as well as active β-catenin induce transcription of the
DKK1 gene, and the
DKK1 promoter contains several
TCF4 response elements, which fits well with our data of co-regulated
TCF4 and
DKK1 [
63].
DNAJB6 is known to activate
DKK1 expression and also had an increased expression in our data demonstrating a regulation which fits earlier reported alterations [
64]. In contrast,
EZH2 and
HDAC2 which cause repression of
DKK1 according to literature [
65‐
67], also showed increased expression in our data. However, at least the
HDAC2 effects were reported to be p53-dependent [
67] and might therefore not apply in this cell system since U-1810 cells lack p53 expression due to a truncating mutation at p53 codon 172 [
33]. Nevertheless, additional validation experiments using siRNA/overexpression of these proteins are needed to confirm a role for these transcriptional regulators in the observed increased
DKK1 expression in the cisplatin-refractory NSCLC clones.
IPA analysis identified p21
WAF1/Cip1 to be a putative downstream effector protein of
DKK1, and p21
WAF1/Cip1 is reported to negatively regulate the cell cycle, i.e. to have a tumor suppressor role [
68]. In rat mesenchymal stem cells, addition of recombinant DKK1 protein decreased p21
WAF1/Cip1 mRNA levels as well as the β-gal staining, both indicators of senescence [
23]. This is in line with our data where DKK1 knockdown increased p21
WAF1/Cip1. Another IPA-retrieved report show however that transgenic mice with ectopic expression of DKK1 in intestinal crypts has an up-regulated p21
WAF1/Cip1, possibly as a consequence of repression of c-myc expression [
22]. Overexpression or silencing of
p21
WAF1/Cip1
induced or reduced, respectively, the cytotoxicity of cisplatin in NSCLC A549 cells, signifying its importance in cisplatin response in NSCLC [
69]. After cisplatin treatment, an increased expression of p21
WAF1/Cip1 is commonly seen in p53 wild type cell lines [
70], like we see in A549 cells (Additional file
3C). Although p21
WAF1/Cip1 was decreased in the p53-mutant U-1810 cells after cisplatin, the level was higher after si
DKK1 combined with cisplatin. Data from A549 cells support this elevated p21
WAF1/Cip1 level in si
DKK1-ablated, cisplatin-treated samples, despite their differential response to cisplatin. Therefore we speculate that p21
WAF1/Cip1 could contribute to the reduced growth after
DKK1 knockdown and cisplatin by induction of G1 arrest and senescence.
GSK3B is a negative regulator of Wnt signaling pathway and inhibition of GSK3B activity, i.e. increased p-Ser9, has previously been shown to confer resistance to cisplatin in lung and ovarian cancer cells [
29,
71,
72]. The mRNA expression was co-regulated with DKK1 in the cisplatin-refractory cells but we did not see any change in the phospho-GSK3B at the time point studied after si
DKK1. Still, the previously reported
DKK1-regulation (Fig.
4b) of both
GSK3B (up) and
DLG4 (down) was confirmed in replicate 1.
No major changes were seen when we analyzed the PI3K/AKT proteins which are known to be involved in cisplatin-refractoriness [
35]. We did however see an almost 2-fold down-regulation in expression of the anti-apoptotic protein Bcl-2 in both DKK1 siRNA and DKK-1 ablated and cisplatin-treated samples in both U-1810 and A549 cells (Fig.
5e and Additional file
3C). A reduced Bcl-2 allows for activation of pro-apoptotic BAK/BAX, which is required for proper cisplatin response [
37], i.e. increased cisplatin-induced apoptosis. This could serve as a mechanism for the sensitization since elevated levels of Bcl-2 and other proteins within the same family e.g. BCL-XL and MCL1 correlate with cisplatin resistance as well as tumor recurrence in NSCLC and other cancers [
73‐
76]. Small molecule inhibitors for Bcl-2-like proteins are also tested in clinical trials together with cisplatin [
77]. Yet the importance of this down-regulation and the role of DKK1 in regulating cisplatin-induced apoptotic signaling would require further studies.
Apart from DKK1's role as a Wnt-signaling antagonist, DKK1 overexpression correlates to an accumulation of β-catenin in the cytoplasm or nucleus in clinical samples from hepatocellular carcinoma [
78]. We analyzed the total level of β-catenin protein (data not shown) in the samples from Fig.
5e but did however not detect any differences at this time point.
Authors' contributions
RL, LL, KV, PH, and DZ designed the study. HS, LL, DZ and BM performed the experiments. HS and LL summarized the data and drafted the manuscript. PH, LL, MN and KV designed the revision experiments and MN and LL conducted these experiments with PH/KV helping out in their evaluation. All authors gave input to the manuscript and contributed to the discussion/conclusions of the study.