Background
Multiple myeloma (MM) arises from the clonal growth of malignant plasma cells in the bone marrow (BM) [
1]. Treatment options for MM are continuously improving, leading to significantly increased response rates as well as prolonged survival [
1,
2]. Despite this progress, myeloma remains a difficult-to-treat disease with the vast majority of patients eventually relapsing. Therefore, the identification of novel drug targets and introduction of additional therapeutic agents are urgently needed to improve the efficacy of existing therapies, to overcome drug resistance and to unravel additional drugable key players in the pathophysiology of MM.
The polycomb complex protein BMI-1 (BMI-1) constitutes a pleiotropic factor with implications in the regulation of the cell cycle, DNA damage response, apoptosis, senescence as well as stem cell self-renewal and differentiation [
3]. BMI-1 was originally discovered as a cooperation factor for v-myc avian myelocytomatosis viral oncogene homolog (MYC) in lymphomagenesis and constitutes a central component of the polycomb repressive complex 1 (PRC1), an epigenetic repressor complex which acts through histone H2A mono-ubiquitination at lysine 119 [
4‐
8]. Overexpression of BMI-1 was frequently observed in diverse human malignancies and associated with tumour initiation and propagation, disease progression and poor prognosis [
9‐
13]. Moreover, BMI-1 was shown to mediate the growth and survival of cancer stem cells in several solid and haematological malignancies [
14‐
17].
BMI-1 represents an attractive drug target in myeloma as well. Upregulation of BMI-1 has been reported previously in MM, and silencing of BMI-1 by small hairpin (sh) RNA significantly impaired the proliferation and colony formation of myeloma cells [
18,
19]. Furthermore, silencing of BMI-1 induced apoptosis in vitro and in vivo through upregulation of BCL2-like 11 (
Bim) expression in MM cells [
19]. More recent results demonstrated that shRNA-mediated silencing of BMI-1 also sensitizes myeloma cells to bortezomib, which was attributed to increased expression of p21 and BCL2-associated X protein (
Bax) [
20]. However, despite the identification of BMI-1 as an attractive drug target in myeloma and various other malignancies, inhibitors specifically targeting BMI-1 have not been available so far.
Kreso et al. recently reported targeting of colorectal carcinoma with PTC-209, a novel small molecule inhibitor of BMI-1 [
21]. Treatment of colorectal cancer cells reduced BMI-1 protein levels and significantly impaired tumour growth in vitro and in vivo. Importantly, PTC-209 also targeted cancer-initiating cells [
21]. Treatment of chronic and acute myeloid leukemia cells with PTC-209 was likewise shown to impair tumour growth and survival [
22,
23]. We therefore aimed to investigate the pre-clinical activity of PTC-209 in myeloma and to explore the impact of BMI-1 inhibition on the tumour microenvironment.
Discussion
In spite of the recent advances in the treatment of MM, the recurrence of myeloma after response to existing therapies is a major drawback on the way to cure. The identification of novel therapy targets and subsequent implementation of new anti-myeloma therapeutics is therefore urgently needed. Based on previous reports, inhibition of the polycomb complex protein BMI-1 might represent an attractive treatment approach for myeloma [
19,
20], but therapeutic agents targeting BMI-1 are not available for clinical use so far. In the current study, we investigated the anti-MM activity of PTC-209, a novel small molecule inhibitor of BMI-1.
Our initial analysis of publically available GEP datasets confirmed the overexpression of
BMI-1 in MM. Overexpression of
BMI-1 has been reported in various malignancies, including MM [
18], and is typically associated with poor survival [
9‐
13]. We likewise observed a significant elevated expression of
BMI-1 in MM as well as in MGUS and SMM patients. Of note,
BMI-1 expression was further elevated in relapsed TT3, but not TT2 patients. This suggests that the use of distinct treatment strategies such as the addition of bortezomib in TT3 specifically impacts BMI-1 levels. According to this assumption, shRNA-mediated silencing of BMI-1 was shown to sensitize MM cells to bortezomib [
20]. Our observation of increased
BMI-1 expression in relapsed TT3 patients suggests that further BMI-1 upregulation might confer a more aggressive phenotype during the progression of MM as it was shown in the progression of several other tumour entities [
9,
12,
24‐
28]. This is also evidenced by an association of high
BMI-1 expression with worse overall survival in relapsed and/or refractory patients treated with bortezomib or dexamethasone (Fig.
1b) [
29]. These results confirmed
BMI-1 overexpression in all stages of MM, from the onset of the disease until progression in response to therapy, underlining its central role as an attractive drug target in MM.
PTC-209 demonstrated significant anti-myeloma activity in all HMCLs analysed. In line with the effect of shRNA-mediated silencing of BMI-1 [
19], we observed a significant impact on the colony formation of myeloma cells, suggesting that targeting BMI-1 also affects the viability of tumour-propagating cells. Recent reports indicated that PTC-209 targets cancer-initiating cells in colorectal and biliary tract cancer. In particular, PTC-209 impaired sphere formation in both entities as well as growth of aldehyde dehydrogenase-positive (ALDH
+) cells in certain biliary tract cancer cell lines [
21,
30]. Future studies therefore have to clarify whether BMI-1 inhibition specifically targets tumour-propagating cells in MM as well.
Similar to shRNA-mediated BMI-1 inhibition in breast and lung adenocarcinoma cells [
31,
32], the growth-inhibiting effect of PTC-209 was associated with deregulation of
CCND1,
MYC,
CDKN1A and
CDKN1B. These genes are known to be implicated in the proliferation of MM cells and their deregulation therefore likely explains the accumulation of cells in the G1 phase and the impaired entry into the S and G2M phase of the cell cycle. Induction of apoptosis was accompanied by a rapid increase of
NOXA expression and subsequent reduction of MCL-1 protein levels. Prior studies reported that silencing of BMI-1 in MM cells was linked to increased expression of either Bim or Bax. However, in these studies, upregulation of Bim and Bax reached significance 48 h post BMI-1 silencing [
19,
20]. In the current study, upregulation of
NOXA (but not Bim or Bax) was already observed 5 h post treatment with PTC-209, suggesting that NOXA might be upstream of Bim and Bax in the initiation of apoptosis after impairing BMI-1. According to this assumption, upregulation of NOXA leads to increased binding of NOXA to MCL-1, thereby releasing Bim from MCL-1 which subsequently mediates Bax (and Bak)-dependent induction of apoptosis [
33,
34]. Similar to our results, a time-dependent increase of NOXA prior to Bim protein levels was observed in chronic lymphatic leukemia cells in response to histone deacetylase inhibitors (HDACi). HDACi were shown to induce a rapid increase of
NOXA mRNA levels, which subsequently triggers MCL-1 binding and induces apoptosis [
35]. Moreover, BMI-1 was shown to mediate the survival of memory CD4 T cells as well as mantle cell lymphoma cells via direct binding to the
NOXA gene locus and repression of NOXA mRNA expression through histone modifications [
14,
36]. These findings suggest that early upregulation of
NOXA might release and activate Bim and Bax to exert their apoptotic effects upon BMI-1 inhibition.
Importantly, the anti-MM activity of PTC-209 was upheld in the presence of major myeloma growth factors (IGF-1 and IL-6) as well as in co-culture with BMSCs. Moreover, we observed synergistic activity of PTC-209 with pomalidomide, carfilzomib and dexamethasone, suggesting that inhibition of BMI-1 might improve current treatment strategies. A similar observation was made when BMI-1-silenced myeloma cells were treated with bortezomib. Concurrent treatment enhanced the anti-proliferative and apoptotic activity of bortezomib via pronounced induction of p21 and Bax [
20].
In addition to its direct anti-myeloma effect, we demonstrated that PTC-209 showed a significant impact on stromal compartments as well. Little is known about the role of BMI-1 in the fate of BM environmental cells. Low expression levels of BMI-1 were associated with senescence in endothelial cells of the human cornea [
37]. Moreover, BMI-1 was shown to promote the angiogenic activity of glioma and hepatocellular carcinoma cells [
38‐
40]. In the current study, PTC-209 significantly inhibited osteoclast and tube formation in vitro. Increased osteoclast activity and formation as well as induction of angiogenesis are prominent features of the myeloma microenvironment. The interaction of myeloma cells and these compartments is implicated in tumour growth, progression and drug resistance [
41]. Interfering with these manifestations therefore impairs MM cell growth and survival. Treatment with PTC-209 might thus not only target MM via direct effects on tumour cells, but also by impairing the crosstalk between tumour and stromal cells.
BMSCs of BMI-1
−/− mice were shown to undergo a shift from osteogenesis to adipogenesis. Moreover, BMI-1
−/− mice displayed an osteopenic phenotype characterized by skeletal growth retardation, decreased osteoblast numbers and activity [
42,
43]. We also observed reduced osteoblast activity and formation in the presence of PTC-209. Considering the negative regulation of osteoblast development by myeloma cells, blockade of BMI-1 could aggravate these effects probably leading to skeletal-related side effects. We therefore aimed to identify the underlying mechanism for the decreased osteoblast formation. As BMI-1 is known for its close interaction with the Wnt signalling pathway [
31], a major signalling pathway in osteogenesis [
44], we speculated that the osteoblast inhibition observed in our study might be related to this connection. We indeed revealed a significant induction of
DKK1 expression in developing osteoblasts during PTC-209 treatment and that blockade of DKK1 with a specific antibody, at least in part, reversed the suppressive effect of PTC-209 on osteoblast activity. This suggests that combination therapy with anti-DKK1 antibodies might overcome the osteoblast suppressive effects of BMI-1 inhibition. In line with our results, silencing of BMI-1 in breast cancer cells was shown to impair Wnt signalling via downregulation of Wnt ligands (e.g. Wnt3a) and upregulation of Wnt inhibitors including DKK1. BMI-1 knockout was shown to upregulate DKK1 and to target cancer cells via subsequent downregulation of
MYC and
CCND1 [
31]. Interestingly, short-term treatment (5 h) with PTC-209 was found to induce
DKK1 expression in myeloma cells as well (up to 5.0 ± 1.9-fold increase,
P < 0.05) (data not shown). This assumes that targeting Wnt signalling via BMI-1 blockade might also target myeloma cells. In line with this, inhibition of the Wnt signalling pathway was recently shown to affect the survival of mantle cell lymphoma-initiating cells [
45]. Considering the proposed roles of Wnt signalling in disease progression and therapy resistance [
46‐
48], BMI-1 inhibition could significantly improveme existing therapies by overcoming drug resistance.
Numerous small molecule inhibitors are currently in clinical development to improve the treatment opportunities for MM patients. These include Bruton’s tyrosine kinase [
49], mitogen-activated protein kinase (MAPK) signalling cascade [
50], phosphoinositol-3 kinase/AKT [
51] and Bcl-2 family inhibitors [
52] among others. Future studies have to clarify which of these agents provide the most potent anti-tumour activity, define predictive markers for an individualized treatment approach and examine the additive value in combination with standard regimens. Targeting of BMI-1 represents a promising novel therapeutic strategy among these evolving arsenals of specific inhibitors due to its universal expression pattern in MM and its impact on the myeloma microenvironment. Further studies evaluating the role of BMI-1 inhibition in myeloma and the applicability of more selective inhibitors (e.g. PTC596) in vitro and in vivo are therefore warranted.
Methods
Reagents
PTC-209, pomalidomide and carfilzomib were obtained from SelleckChem, dissolved in DMSO and stored at −80 °C. Dexamethasone was obtained from Sigma-Aldrich, dissolved in PBS and stored at −20 °C. Recombinant human IGF-1, IL-6, receptor activator of nuclear factor-kappa B ligand (RANKL) and macrophage colony-stimulating factor (M-CSF) were obtained from Peprotech, dissolved in PBS/BSA 0.1 % and stored at −20 °C. Goat anti-human DKK1 neutralizing antibody and normal goat IgG were purchased from R&D Systems, dissolved in PBS and stored at −20 °C.
Cell lines and culture conditions
Human multiple myeloma cell lines (HMCLs) U266, KMS-12-BM, OPM-2, NCI-H929, SK-MM-1 and RPMI8226 were obtained from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). MM.1S and MM.1R cells were kindly provided by Dr. Steven Rosen (Northwestern University, Chicago, IL). A human bone marrow mesenchymal stromal cell line immortalized by enforced expression of telomerase (BMSC TERT+) was kindly provided by Dr. Dario Campana (St. Jude Children’s Research Hospital, Memphis, TN). All cell lines were cultivated in RPMI-1640 medium supplemented with 10 % heat-inactivated fetal bovine serum, 2 mM L-glutamine and 100 U/ml penicillin/streptomycin (Gibco). BMSC TERT+ cells were supplemented with 1 μM hydrocortisone (Sigma-Aldrich). For co-culture experiments, 1 × 105 BMSC TERT+ cells were seeded in 24-well plates and cultured overnight before 2 × 105 MM cells were added per well for 72 h.
Cytotoxicity assay
Viability was determined by using Cell Counting Kit 8 (Sigma-Aldrich) following the manufacturer’s directions. In brief, HMCLs (2 × 104), BMSC TERT+ cells (1 × 104) and PBMCs (2.5 × 105) were incubated in flat-bottomed 96-well plates (ThermoFisher Scientific) in the presence of PTC-209 (0.01–10 μM) alone, or in combination with either pomalidomide (1–5 μM) or carfilzomib (1–5 nM). Viability assessment in the presence of recombinant human IGF-1 (10 ng/ml) and IL-6 (10 ng/ml) was performed in serum-free Syn-H medium (ABCell-Bio). After 96 h, cells were incubated with WST-8, and absorbance was measured at 450 nm using a HTS 7000 Bio Assay Reader (Perkin Elmer).
HMCLs (2 × 103) either treated or untreated with PTC-209 at 1 μM were plated in duplicates in 1.1 ml methylcellulose-based medium (MethoCult Classic, StemCell Technologies) per 6-well and incubated for 14 days (37 °C, 5 % CO2). At the end of the incubation period, the number of colonies consisting of >40 cells was scored using an inverted microscope with ×4, ×10 and ×20 planar objectives.
Flow cytometry
Induction of apoptosis was determined by Annexin V/7-AAD staining (BD Biosciences). HMCLs were seeded in the presence or absence of BMSC TERT+ cells and treated with 0.1 % DMSO (control), PTC-209 (1 μM), pomalidomide (1 μM) and/or carfilzomib (5 nM) for 72 h. Cells were incubated with Annexin V and 7-AAD for 15 min in the dark before performing analysis.
Cell cycle analysis was performed after treatment of HMCLs with PTC-209 (1 μM) for 24 h using the FxCycle™ PI/RNase Staining solution (ThermoFisher Scientific) following the manufacturer’s instructions.
Intracellular staining of BMI-1 and MCL-1 was performed using the BD Transcription Factor Buffer Set (BD Biosciences) according to the manual. After fixation and permeabilization, cells were incubated with a mouse anti-human MCL-1 antibody (ab197529, Abcam), mouse anti-human BMI-1 antibody (562528, BD Biosciences) or the corresponding isotype controls for 40 min at 4 °C. Thereafter, cells were washed and analysed. All analyses were performed on a FACScan (BD Biosciences).
Quantitative RT-PCR
Total RNA was isolated using RNeasy kit (Qiagen), and cDNA synthesis was performed with M-MuLV reverse transcriptase (New England Biolabs). CDKN1A, CDKN1B, MYC, CCND1, NOXA, TRAP, cathepsin K and DKK1 expression levels were analysed by quantitative PCR (qPCR) using TaqMan Universal PCR Master Mix and pre-designed TaqMan gene expression assays (Applied Biosystems). RPLPO served as endogenous control. Reactions were carried out in 25 μl volumes and run on the ABI Prism 7300 platform (Applied Biosystems). All samples were run at least in duplicates.
PARP ELISA
Levels of cleaved PARP were analysed by using a commercially available ELISA kit (Invitrogen) following the manual. In brief, cell lysates were incubated with cleaved PARP detection antibody for 3 h at room temperature on an orbital shaker. Subsequently, wells were washed and incubated with anti-rabbit IgG HRP for 30 min at room temperature. After an additional wash step, 100 μl stabilized chromogen was added per well, the plate was incubated for 30 min at room temperature in the dark and finally mixed with 100 μl stop solution per well. Absorbance was measured at 450 nm, and levels of cleaved PARP were determined in relation to a standard curve.
Osteogenic differentiation
BMSCs were seeded at a density of 25,000 per square centimetre and grown to 70–80 % confluence. Osteoblast differentiation was initiated by changing the medium to alpha-MEM supplemented with 15 % FBS, 2 mM L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 10 nM dexamethasone, 50 μg/ml ascorbic acid and 5 mM β-glycerophosphate (Sigma-Aldrich). Osteogenic medium was changed every 3–4 days. PTC-209, normal goat IgG (1 μg/ml) and/or anti-DKK1 neutralizing antibody (1 μg/ml) were added with every medium change. Cells treated with 0.1 % DMSO served as control. Osteoblast formation was assessed by alkaline phosphatase activity assay and alizarin red S staining as described previously [
53,
54].
Osteoclast differentiation
Human PBMCs were obtained from voluntary healthy donors and cultured (2.5 × 106/ml) overnight in 24-well plates to remove non-adherent cells. Subsequently, osteoclast differentiation was initiated by changing the medium to alpha-MEM supplemented with 10 % FBS, 2 mM L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 ng/ml RANKL and 25 ng/ml M-CSF in the presence or absence of PTC-209. Medium was changed every 3–4 days. At day 14, cultures were fixed and stained using the Acid Phosphatase Leukocyte Kit (Sigma-Aldrich) according to the manufacturer’s instructions. Cells were evaluated using an inverted microscope and multinucleated TRAP-positive cells (>3 nuclei) were scored as mature osteoclasts.
Tube formation was assessed following the manual of the Angiogenesis Starter Kit (Gibco). Briefly, human umbilical vein endothelial cells (HUVECs) were seeded in LVES-supplemented medium 200 at a density of 2.5 × 10
3 cells per square centimetre. For in vitro tube formation, 80 % confluent cultures were harvested, and HUVECs (50,000 per 24-well plate) were plated on Geltrex® Matrix pre-coated wells in the presence or absence of PTC-209. Tube networks were documented 19 h post-seeding by using an inverted microscope. Analysis was performed using the Angiogenesis Analyzer for Image J (
http://image.bio.methods.free.fr/ImageJ/?Angiogenesis-Analyzer-for-ImageJ.html).
Gene expression analysis
For the illustration of
BMI-1 expression in CD138
+ purified bone marrow samples as well as for the assessment of overall survival in different MM patients, following datasets from the Gene Expression Omnibus (GEO) were selected: GSE2658 (consisting of 559 samples from newly diagnosed MM patients treated within the TT2 and TT3 protocol), GSE5900 (22 healthy controls, 44 MGUS patients and 12 SMM patients), GSE6477 (15 healthy controls, 22 MGUS patients, 24 SMM patients, 73 newly diagnosed and 28 relapsed MM patients), GSE31161 (346 TT2 baseline and 127 TT2 relapse samples as well as 433 TT3 baseline and 29 TT3 relapse samples) and GSE9782 (264 relapsed and/or refractory MM patients treated with bortezomib or dexamethasone). The GSE9782 dataset was established by Mulligan et al. (2007) and contains data from 156 patients treated within the APEX phase 3 trial, 57 patients treated within a companion study to receive bortezomib after progressive disease with dexamethasone, 44 patients treated within the SUMMIT phase 2 trial and 7 patients treated within the CREST phase 2 trial [
29].
Expression and clinical data were downloaded into R using the Bioconductor GEOquery package. For GSE6477, GSE5900 and GSE31161, raw CEL files were downloaded from GEO and analyses were performed on gcrma-normalized samples in R. In case of GSE2658, corresponding raw CEL files were downloaded from GEO study GSE24080 in order to perform a gcrma-normalization on the raw data using the ‘affy’ package from Bioconductor (since raw CEL files are not provided in the GSE2658 dataset). For GSE9782, mas5 expression sets were retrieved for analysis using the GEOquery package, since raw CEL files are not provided for this study.
Statistical analysis
Survival analysis was performed in R using the ‘survival’ package. In order to split the patients into two groups with different survival probabilities exhibiting higher or lower BMI-1 expression, the maximally selected rank statistics, implemented in the maxstat R package, was applied to the BMI-1 expression data. The statistical significance of differences in overall survival between the two groups was calculated by the log-rank test, and survival curves were plotted using the Kaplan-Meier method. For the analysis of in vitro experiments, two-tailed unpaired t test was performed using Prism 5 (GraphPad Software Inc., La Jolla, CA, USA). P values <0.05 were considered to be statistically significant. Graphs represent the mean ± standard deviation of at least three independent experiments performed in triplicates. Synergistic, additive or antagonistic drug activities of combination treatments were determined by using CompuSyn software.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AB and HL designed the research. AB performed the experiments and analysed the data. KS and WS performed the analysis of the GEP datasets. AB, NZ and HL wrote the paper. All authors reviewed and approved the final version of the manuscript.