Background
Multiple myeloma (MM) is characterized by the accumulation of malignant plasma cells (PCs) in the bone marrow (BM) [
1]. MM cells reside in the BM microenvironment similar to their normal counterparts where they receive essential survival and growth signals from the surrounding stromal compartments [
2]. Direct interactions between myeloma cells and bone marrow stromal cells (BMSCs) or osteoclasts [
3] as well as the crosstalk through various soluble factors play a major role in the pathophysiology of MM leading to clinical features such as myeloma bone disease, anaemia and immunosuppression [
1].
MM cells are closely connected with BMSCs, modify their function and induce the production of tumour promoting factors [
4]. Within the growth signalling network of the myeloma microenvironment, IL-6, IGF-1 and insulin play a prominent role [
5-
7]. An imbalance of the RANKL to OPG ratio promotes osteoclast activation [
8] while the secretion of osteoblast inhibitory molecules such as DKK-1 and activin A impairs osteogenesis, leading to myeloma induced bone disease [
9,
10].
Bone morphogenetic proteins (BMPs) in turn stimulate bone formation [
11] and are at the same time potent inhibitors of MM cell survival [
12,
13]. Higher BMP expression in malignant plasma cells is associated with superior overall survival [
13]. Less is known, however, about the role of BMP antagonists in MM. In solid malignancies, upregulation of BMP antagonists such as gremlin was shown to promote the growth and survival of tumour cells [
14]. We therefore speculated that deregulation of BMP antagonistic proteins in the myeloma microenvironment might be involved in the pathophysiology of MM. To address this question, we analysed the BMP antagonist expression profile in myeloma, identifying insulin like growth factor binding protein 7 (IGFBP7) as a potential gene of interest. IGFBP7 is a secreted protein with IGF-1, insulin [
15], VEGFA [
16] and activin A [
17] binding properties and has been described as BMP antagonist [
18]. Methylation dependent silencing of IGFBP7 was reported as associated with unfavourable outcome in lung-, breast-, and pancreatic cancer, as well as hepatocellular carcinoma [
19-
23]. On the contrary, high
IGFBP7 expression was linked to poor prognosis in oesophageal adenocarcinoma as well as head and neck squamous cell carcinomas [
24,
25]. Recent studies also suggested a role for IGFBP7 in haematological malignancies. In acute lymphoblastic leukemia (ALL)
IGFBP7 expression was associated with adverse outcome and shown to interfere with leukemia cell proliferation [
26,
27]. IGFBP7 was also reported to be involved in the crosstalk between BMSCs and ALL cells, mediating asparaginase-resistance in B-lineage ALL cells [
27]. Based on these results we were interested in the prognostic and pathophysiologic role of IGFBP7 in MM.
Discussion
Various soluble factors have been identified in the signaling crosstalk between myeloma and stromal cells including IGF-1, IL6, DKK-1, RANKL and activin A [
1,
2,
5-
10]. In this study we identified IGFBP7, a secreted factor with BMP antagonistic activity [
18], and binding sites for IGF-1, insulin [
15], VEGFA [
16] and activin A [
17] as a potential novel player in the pathogenesis of MM, including myeloma bone disease. BMPs are known to induce apoptosis in MM cells [
12,
13], so upregulation of BMP antagonists might be envisaged to be used as a protective mechanism by myeloma cells. Median
IGFBP7 expression levels, although
IGFBP7 displayed a heterogenous expression profile, were found to be downregulated in malignant plasma cells compared to their normal counterparts, suggesting that myeloma cells do not rely on the BMP antagonistic activity of IGFBP7.
We show that
IGFBP7 downregulation in MM cells can be explained by promoter methylation. Such a downregulation of
IGFBP7 by promoter methylation was previously observed in solid tumours including melanoma, prostate, lung and breast cancer as well as T-ALL, suggesting a common mechanism in cancer [
19,
21,
30-
32]. Plasma cells from MGUS patients already display reduced
IGFBP7 expression, indicating that deregulation of IGFBP7 might occur early in the course of the disease.
IGFBP7 belongs to the family of insulin like growth factor binding proteins (IGFBPs) consisting of six high- and several low-affinity members [
33]. IGFBPs were described to act via IGF dependent and independent mechanisms [
33]. Contrary to other family members, in particular IGFBP3, IGFBP7 possesses low binding affinity for IGFs [
15,
34]. Accordingly, unlike IGFBP3 [
6], IGFBP7 was not able to antagonize the growth stimulatory activity of IGF-1 and insulin on MM cell lines.
Previous studies indicated a role for IGFBP7 as tumour suppressor in several solid malignancies including melanoma, hepatocellular carcinoma, thyroid, breast, bladder and colon cancer by either direct induction of apoptosis and/or inhibition of tumour growth by deregulation of p16, p21, p53 and ERK signaling [
19,
26,
35-
38]. However, results regarding the role of IGFBP7 in solid tumours are inconsistent and recent studies provide a more complex picture of IGFBP7 in different entities. In glioblastoma, IGFBP7 was shown to stimulate tumour cell proliferation and angiogenesis of brain endothelial cells [
39,
40]. Similarly, elevated
IGFBP7 expression was detected in specific stages of colorectal cancer and silencing of IGFBP7 reduced proliferation as well as colony formation in colorectal cancer cell lines [
41]. Expression of
IGFBP7 was also detected in cancer-associated fibroblasts, endothelial cells, mesenchymal tumours and malignant epithelial cells with a mesenchymal phenotype. In the latter, loss of IGFBP7 significantly impaired the anchorage independent growth of tumour cells. Furthermore,
IGFBP7 expression in cancer stromal cells supported the growth of colon cancer cells [
41]. These findings establish a complex variety of functions for IGFBP7 depending on the malignancy investigated.
In haematological malignancies, IGFBP7 was found to be associated with BAALC expression in T-ALL and to correlate with poor survival [
26]. Similarly, in B-cell ALL
IGFBP7 expression in BMSCs was found to be associated with asparaginase resistance and decreased leukemia-free survival [
27]. In our study, high expression levels of
IGFBP7 predicted poor outcome in two large and independent cohorts of MM patients. Whereas the mean values and standard deviations (5.564 ± 3.213 vs. 3.155 ± 1.966) as well as the percentage of patients expressing
IGFBP7 as defined by the PANP algorithm (marginal and present, 44.9 vs. 12.9%) significantly differed between the HM- and LR-cohort, the prognostic impact of
IGFBP7 expression was comparable between both cohorts. Observed differences in frequency and height of expression are most likely due to different amplification protocols used (double vs. single amplification) [
42-
44].
IGFBP7 expression in MM cells was linked to prognostically adverse chromosomal aberrations such as translocation t(4;14) and amplification 1q21. The multiple myeloma SET domain (MMSET) protein, specifically overexpressed in t(4;14) myeloma, is a histone methyl transferase shown to modulate DNA methylation, thereby inducing the activation of specific target genes [
45]. Interestingly,
IGFBP7 expression was found to be regulated by MMSET in myeloma cells [
45] which is in line with our observation of higher IGFBP7 transcript levels in t(4;14) cases. The association between MMSET expression and expression of
IGFBP7 was confirmed by
in-silico analysis of two open-source GEP datasets. This suggests that IGFBP7 represents a novel marker of high-risk myeloma, defined by an epigenetic signature and regulated by MMSET.
The growth attenuating effects of IGFBP7 we observed
in vitro were rather small and evident only at high IGFBP7 concentrations [
46]. Moreover, we show that, in clinical myeloma samples,
IGFBP7 expression is associated with higher myeloma cell proliferation, in turn associated with adverse prognosis. The phenotype of MMSET expressing myeloma seems to overrule the weak inhibitory activity of IGFBP7 on myeloma cell proliferation. Thus, rather than being mechanistically involved in the poor outcome, IGFBP7 might rather be seen as a marker associated with a high-risk disease phenotype.
We show upregulation of IGFBP7 in BMSCs while undergoing osteogenic differentiation and treatment with exogenous rhIGFBP7 further stimulating osteoblast activity. IGFBP7 treatment led to a significant downregulation of DKK1, a major osteoblast inhibitory molecule in MM [
9]. Furthermore, we were able to show that IGFBP7 can neutralize the inhibitory effects of activin A on osteoblast development. Activin A mediated inhibition of osteogenesis leads to uncoupling of the delicate balance between bone formation and degradation thus contributing to myeloma related bone destruction [
10,
47]. Our results suggest that the downregulation of IGFBP7 in the myeloma microenvironment enables activin A to release its full destructive potential.
Possibly, the lack of IGFBP7 in the microenvironment is aggravated by low or absent intrinsic expression of
IGFBP7 in myeloma cells. In contrast, in myeloma cases with high
IGFBP7 expression the downregulation of this molecule in the microenvironment might be counterbalanced by production in myeloma cells, possibly explaining the correlation of higher intrinsic
IGFBP7 expression and absence of advanced bone disease. In line with this, prior studies reported osteolytic lesions to occur less frequently in myeloma cases harboring a translocation t(4;14) [
48].
Methods
Patients
Patients presenting with previously untreated MM (n = 332) or monoclonal gammopathy of unknown significance (MGUS; n = 22) at the University Hospitals of Heidelberg and Montpellier (HM group) as well as 10 healthy normal donors have been included in the study approved by the Heidelberg ethics committee (#229/2003 and S-152/2010) after written informed consent was obtained in accordance with the Declaration of Helsinki. Patients were diagnosed, staged and response to treatment assessed according to standard criteria. Normal bone marrow plasma cells and myeloma cells were purified as previously published [
13,
44]. Memory B cells (MBCs; n = 11) and polyclonal plasmablasts (PPCs; n = 10) were obtained as reported [
49]. Aliquots of unpurified whole bone marrow from myeloma patients (n = 154) and healthy donors were obtained after NH
4 lysis.
Bone disease was assessed by conventional X-ray and whole body CT scan and graded as 0 (normal bone structure), 1 (osteopenia/osteoporosis), 2 (1–3 osteolyses) or 3 (major structural damage, >3 osteolyses) as described previously [
44].
The prognostic impact of
IGFBP7 expression data was validated on an independent cohort of 701 newly diagnosed, therapy-requiring patients treated within the total therapy 2 or 3 protocol, respectively (LR group), for whom gene expression data are publicly available (see below) [
50,
51]. All patients in both cohorts, i.e. the HM- and the LR-group, received high-dose chemotherapy followed by autologous stem cell transplantation.
Cells 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 cells were kindly provided by Dr Steven Rosen (Northwestern University, Chicago, IL). All HMCLs 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).
A human bone marrow mesenchymal stromal cell line immortalized by enforced expression of telomerase (hBMSC TERT+) was kindly provided by Dr Dario Campana (St. Jude Children’s Research Hospital, Memphis, TN) and cultured in RPMI-1640 medium supplemented with 10% FBS, 2 mM L-glutamine, 100 U/ml penicillin/streptomycin and 1 μM hydrocortisone (Sigma-Aldrich). Primary human BMSCs (Lonza) were cultured in α-MEM (Gibco) supplemented with 10% FBS, 2 mM L-Glutamine, 100 U/ml penicillin/streptomycin and 1 ng/ml FGF-2 (Peprotech).
For co-culture experiments, 0.25 × 106 TERT+ hBMSCs were seeded in 6-well plates and cultured over night before 0.5 × 106 MM cells were added per well for 72 hours in the presence or absence of transwell inserts (0.4 μm pore size; BD). Direct contact cultures were MACS sorted by CD138 selection previous to RNA isolation of the CD138-negative population. Purity of the negative sort was ≥ 90% by cytological assessment.
Cytotoxicity assay
Cell Counting Kit-8 (Sigma) was used to analyze the impact of recombinant human IGFBP7 (rhIGFBP7; Peprotech) on myeloma cell growth following the manufacturer’s directions. In brief, 2 × 10
4 cells per well were seeded in flat-bottomed 96-well plates (Iwaki) in the presence of rhIGFBP7 (0–10 μg/ml). Neutralization experiments with recombinant human IGF-1 (10 or 100 ng/ml) (R&D Systems), insulin (10 or 100 ng/ml) (Roche Diagnostics) and IGFBP7 (1 or 10 μg/ml) were performed in serum free Syn-H medium (ABCell-Bio) according to Sprynski
et al. [
6]
BrdU assay
Proliferation of myeloma cells was assessed by BrdU assay (Calbiochem) following the manual. BrdU label was added for the last 19 hours of the culture period. Proliferation was determined by absorbance measurement at 450 nm using a HTS 7000 Bio Assay Reader (Perkin Elmer).
Flow cytometry
Intracellular staining of IGFBP7 was performed using the BD Cytofix/Cytoperm™ Plus Kit (BD Biosciences) according to the manual. After fixation and permeabilization, cells were incubated with the primary goat-anti-human IGFBP7 antibody (sc-6064; Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) or the corresponding isotype control (sc-3887; Santa Cruz) for 30 min at 4°C. Thereafter cells were washed and incubated for 30 min at 4°C with a secondary PE-conjugated donkey-anti-goat antibody (sc-3857; Santa Cruz). Analysis was performed on a FACScan (BD Biosciences).
Induction of apoptosis was determined by FACS analysis of Annexin V/7-AAD stainings (BD Biosciences). HMCLs were seeded at a concentration of 2.5 ×105/ml and treated with either PBS/BSA 0.1% (control) or rhIGFBP7 (10–20 μg/ml) for 72 hours. Cells were incubated for 15 min with Annexin V and 7-AAD in the dark before performing analysis.
Quantitative RT-PCR
Total RNA was isolated using RNeasy kit (Qiagen) and cDNA synthesis was performed with M-MuLV reverse transcriptase (New England Biolabs). IGFBP7, p16, p21, p27, Runx2, Dlx-5, Col1A and DKK-1 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.
Methylation
For IGFBP7 induction experiments HMCLs were seeded at 2 ×105/ml and cultivated in the presence of 0.1% DMSO (control) or 500 nM 5-aza-2′-deoxcytidine (aza; Sigma) for 96 hours. Trichostatin A (TSA; Sigma) was added at a concentration of 50 ng/ml for the last 24 hours of the culture period.
Analysis of the IGFBP7 promoter methylation status was performed by varionostic GmbH (Ulm, Germany) using pyrosequencing technology.
Osteogenic differentiation
For osteoblast development, BMSCs were seeded at a density of 12500–25000 per cm2 and grown to 70-80% confluence. Osteoblast differentiation was initiated by changing the medium to α-MEM supplemented with 15% FBS, 2 mM L-Glutamine, 100 U/ml penicillin/streptomycin, 10 nM dexamethasone (Sigma), 50 μg/ml ascorbic acid (Sigma) and 10 mM β-glycerophosphate (Sigma). Osteogenic medium was changed every 3–4 days. Recombinant human IGBFP7 (1 or 10 μg/ml) and/or activin A (50 or 100 ng/ml) (Peprotech) were added with every medium change. Cells treated with PBS/BSA 0.1% served as control.
Alkaline phosphatase activity assay
Alkaline phosphatase activity was determined at day 14 of differentiation using the AttoPhos® AP Fluorescent Substrate System (Promega) following the manual. Fluorescence was measured at 595 nm with excitation at 430 nm. A standard curve was obtained by diluting the supplied Calibrator solution (500 ng BBT/ml) in AttoPhos buffer.
Gene expression profiling
Gene expression profiling was performed as previously published [
44] using U133 2.0 plus arrays according to the manufacturer’s instructions (Affymetrix, Santa Clara, CA, USA). Expression data are deposited in Array Express under the accession numbers E-MTAB-317 as well as E-TABM-1138 and Gene Expression Omnibus GSE24080 (the latter two for the LR group). When different probe sets were available for the same gene, we chose the probe set yielding the maximal variance and the highest signal.
Interphase fluorescence in situ hybridization
Analyses were performed on CD138-purified plasma cells as described [
52] using probes for the detection of numerical aberration of the chromosomal regions 1q21, 5p15/5q35, 9q34, 13q14.3, 15q22, 17p13, as well as the IgH-translocations t(11;14)(q13;q32), and t(4;14)(p16.3;q32) (Kreatech, Amsterdam, The Netherlands and MetaSystems, Altlussheim, Germany).
Statistical analysis
Correlations (Spearman and Pearson) and survival analysis (Cox regression analysis) were performed using SPSS statistics 17.0. 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.
Gene expression analyses were performed on GC-RMA preprocessed data sets. To assess presence (“expressed”) or absence (“not expressed”) of gene expression, the “Presence-Absence calls with Negative Probesets” (PANP) algorithm was used [
53]. For the analysis of the prognostic value of IGFBP7, patients with high and low
IGFBP7 expression in the HM cohort were delineated using maximally selected rank statistics as implemented in the maxstat R package (
http://cran.r-project.org/web/packages/maxstat/index.html). The survival function was assessed by means of the Kaplan-Meier method (rms package in R). The log rank test was used to test for statistically significant survival curve differences. An effect was considered as statistically significant if the
P-value of its corresponding statistical test was not higher than 5%. All computations were performed using R 3.1.0 (
http://www.r-project.org/) and Bioconductor 2.14.
For validation of the impact of IGFBP7 expression on survival, cel-files from 701 patients treated within the total therapy 2 or 3 protocol (LR group), were preprocessed as described above (GC-RMA), and the same analysis strategy followed to obtain an optimal cut-off (maxstat), i.e. the optimal cut-off was derived as the mean of the cut-offs derived for EFS and OS. This separate analysis yielded two different cut-off values for the HM- and LR-cohort. As the generation of two different cut-offs for the HM and LR-group, i.e. delineating different fractions of patients between in the different cohorts, might give an over-optimistic validation, we likewise applied the fraction of patients defined as “high IGFBP7 expressing” from the HM-cohort (46/247, 18.6%) to the LR-cohort (131/701, 18.6%).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AB, DH, NZ, HL and BK designed the research. AB performed the experiments. AS, SL, DH and BK performed gene expression profiling and analysed the data. AB, MS, DHe and DH performed the statistical analysis. AB, DH and NZ wrote the paper. All authors reviewed and approved the final version of the manuscript.