ORIGINAL ARTICLE Bone Morphogenetic Protein 4 Gene Therapy in Mice Inhibits Myeloma Tumor Growth, But Has a Negative Impact on Bone Marita Westhrin,1,2† Toril Holien,1,3† Muhammad Zahoor,2 Siv Helen Moen,1,2 Glenn Buene,1,2 Berit Størdal,1 Hanne Hella,1 Huipin Yuan,4 Joost D de Bruijn,4,5 Anton Martens,6 Richard WJ Groen,6 Fatima Bosch,7,8 Ulf Smith,9 Anne-Marit Sponaas,1 Anders Sundan,1,2 and Therese Standal1,2,3 1 Department of Clinical and Molecular Medicine, Faculty of Medicine, Norwegian University of Science and Technology (NTNU), Trondheim, Norway 2 Centre of Molecular Inflammation Research (CEMIR), Norwegian University of Science and Technology, Trondheim, Norway 3 Department of Hematology, St. Olavs Hospital, Trondheim, Norway 4 Kuros Biosciences BV, Bilthoven, The Netherlands 5 The School of Engineering and Materials Science, Queen Mary University of London, London, UK 6 Department of Hematology, Cancer Center Amsterdam, VU University Medical Center, Amsterdam, The Netherlands 7 Center of Animal Biotechnology and Gene Therapy and Department of Biochemistry and Molecular Biology, School of Veterinary Medicine, Universitat Autònoma de Barcelona, Barcelona, Spain 8 Centro de Investigación Biomédica en Red de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM), Madrid, Spain 9 Department of Molecular and Clinical Medicine, Sahlgrenska University Hospital, Gothenburg, Sweden ABSTRACT Multiple myeloma is characterized by accumulation of malignant plasma cells in the bone marrow. Most patients suffer from an osteolytic bone disease, caused by increased bone degradation and reduced bone formation. Bone morphogenetic protein 4 (BMP4) is important for both pre- and postnatal bone formation and induces growth arrest and apoptosis of myeloma cells. BMP4-treatment of myeloma patients could have the potential to reduce tumor growth and restore bone formation. We therefore explored BMP4 gene therapy in a human-mouse model of multiple myeloma where humanized bone scaffolds were implanted sub- cutaneously in RAG2−/− γC−/−mice. Mice were treated with adeno-associated virus serotype 8 BMP4 vectors (AAV8-BMP4) to express BMP4 in the liver. When mature BMP4 was detectable in the circulation, myeloma cells were injected into the scaffolds and tumor growth was examined by weekly imaging. Strikingly, the tumor burden was reduced in AAV8-BMP4 mice compared with the AAV8-CTRL mice, suggesting that increased circulating BMP4 reduced tumor growth. BMP4-treatment also prevented bone loss in the scaffolds, most likely due to reduced tumor load. To delineate the effects of BMP4 overexpression on bone per se, without direct influence from cancer cells, we examined the unaffected, non-myeloma femurs by μCT. Surprisingly, the AAV8-BMP4 mice had sig- nificantly reduced trabecular bone volume, trabecular numbers, as well as significantly increased trabecular separation compared with the AAV8-CTRL mice. There was no difference in cortical bone parameters between the two groups. Taken together, BMP4 gene therapy inhibited myeloma tumor growth, but also reduced the amount of trabecular bone in mice. Our data suggest that care should be taken when considering using BMP4 as a therapeutic agent. © 2019 The Authors. JBMR Plus published by Wiley Periodicals, Inc. on behalf of American Society for Bone and Mineral Research. KEY WORDS: BMPs/TGF-βs; Cancer; Osteoblast; Osteoclasts; Tumor-induced bone disease This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Received in original form July 17, 2019; revised form October 11, 2019; accepted October 17, 2019. Accepted manuscript online October 23, 2019. Address correspondence to: Toril Holien, Department of Clinical and Molecular Medicine, Faculty of Medicine, Norwegian University of Science and Technology (NTNU), Trondheim, Norway. E-mail: toril.holien@ntnu.no; Therese Standal, Department of Clinical and Molecular Medicine, Faculty of Medicine, Norwegian University of Science and Technology (NTNU), Trondheim, Norway. E-mail: therese.standal@ntnu.no † MW and TH contributed equally to this work. JBMR® Plus (WOA), Vol. 4, No. 1, January 2020, e10247. DOI: 10.1002/jbm4.10247 © 2019 The Authors. JBMR Plus published by Wiley Periodicals, Inc. on behalf of American Society for Bone and Mineral Research. 1 of 9 n 1. Introduction and ligated with the iRFP PCR product, using T4 DNA ligase (all Fer- mentas, Thermo Fisher Scientific). The resulting plasmid, pLVX- EF1α-iRFP-IRES-ZsGreen1, was used together with TransLenti Viral M ultiple myeloma is a hematological cancer caused by accu- mulation of malignant plasma cells in the bone marrow.(1) Nearly all patients suffer from a severe osteolytic bone disease, Packaging Mix (Open Biosystems) and Genejuice (Novagen, Merck Life Science AS, Oslo, Norway) to transfect 293 T packaging cells causing pain and fractures.(2) The bone disease is caused by (Open Biosystems, Thermo Fisher Scientific). Supernatants con- increased osteoclast activity and a lack of bone repair due to taining lentivirus were used to transduce the KJON myeloma cell too few and dysfunctional osteoblasts.(2) Currently, bisphospho- line. Positively transduced cells expressed both iRFP and ZsGreen1 nates are the most common drugs used to treat the bone dis- fluorescent proteins and were sorted on a FACSAria™ Fusion flow ease, and there is a lack of treatment options that can promote cytometer (BD Biosciences, San Jose, CA, USA) to obtain a pure bone formation. New, efficient drugs to treat myeloma have population of iRFP-positive cells for in vivo studies. been developed the last decades, including immunomodulatory agents, proteasome inhibitors, histone deacetylase inhibitors, 2.3 Cell viability in vitro and monoclonal antibodies.(3) Nevertheless, multiple myeloma Cell viability was measured using the Cell Titer-Glo assay remains an incurable disease.(4) (Promega, Madison, WI, USA), that measures the ATP content in Bone morphogenetic proteins (BMPs) is a large subgroup of wells. Cells were seeded in 96 well optical plates (104 cells/well) ligands in the transforming growth factor (TGF)-β family.(5) In and treated as indicated in the figure legends. Cell Titer-Glo vitro, several BMPs induce growth arrest and apoptosis in multi- reagent was added following the manufacturer’s instructions ple myeloma cell lines as well as in primary myeloma cells from and luminescence was measured using Victor 1420 multilabel patients.(6–11) BMP-signaling is also important for both pre- and counter (PerkinElmer Inc., Waltham, MA, USA). To distinguish postnatal bone formation.(12) For example, combined deletion between effects on cell division and cell death, we measured of BMP2 and BMP4 in mesenchymal stem cells (MSC) leads to apoptosis by annexin V labelling. In brief, naïve, KJON cells were severely impaired osteogenesis in mice(13) and inhibiting BMP- seeded in 96 well plates (5 × 104 cells/well) and treated as indi- signaling reduces osteoblast differentiation in mouse and cated in the figure legends. The cells were stained using Apotest human cells.(14–16) On the other hand, two separate studies FITC kit (Nexins Research, Kattendijke, The Netherlands). Then, found increased bone mass and bone strength in mice treated cells were incubated with annexin V FITC (0.2 μg/mL) on ice for with soluble BMPR1A-Fc fusion protein.(17,18) BMPR1A-Fc has 1 hour. Propidium iodide (PI) (1.4 μg/mL) was added 5 minutes high affinity to BMP2 and BMP4 and acts as a decoy receptor that before cells were analyzed using an LSRII flow cytometer inhibits their binding to receptors on the surface of cells. Thus, (BD Biosciences, San Jose, CA, USA). Cells negative for both the effects of a given BMP in the context of multiple myeloma annexin-V and PI staining were considered viable. is not entirely clear. In this study we wanted to clarify if BMP4 could have therapeutic potential in multiple myeloma patients, 2.4 RAG2−/− γC−/− by preventing tumor growth and restoring bone homeostasis. We therefore evaluated the effects of AAV-based BMP4 gene We used RAG2−/− γC−/− BALB/c female mice as described previ- therapy in a human-mouse scaffold model of multiple myeloma. ously.(21) The mice lack B, T and NK cell immunity and were kept in specific pathogen free (SPF) unit. Here the mice were housed 2. Materials and Methods in IVC-cages, with free access to bedding material, nesting material and enrichment objects. Mice were given sterile food (RM1 2.1 Cell culture and reagents #801002, Special Diets Services, Essex, UK,) and water ad libitum, and were caged in groups of 3–5 mice. Mice were maintained at The human myeloma cell line KJON(19) was cultured in RPMI with a room temperature of 21–22 C and 55% humidity with a 12 hour 5% heat inactivated human serum (HS) and 2 ng/mL interleukin light/dark cycles including 1 hour dusk/dawn. All mice were of (IL)-6 (Gibco, Thermo Fisher Scientific, Waltham, MA, USA). approximately the same age at the beginning of experiments. in vitro experiments with KJON cells were performed with 2% HS in RPMI and IL-6 (1 ng/mL). The mouse myeloma cell line 2.5 Human-mouse scaffold myeloma model and NS0 was generously provided by Dr Z. Eshhar (Weizmann Insti- imaging tute of Science, Israel), and the cells were grown in 10% FCS in RPMI. All cells were cultured at 37  C in a humidified atmosphere We used a modified version of a previously described xenograft containing 5% CO2 and were tested for mycoplasma every mouse model with a humanized bone environment.(21) Briefly, 3 months. Recombinant murine (rm) BMP4 (Cat# 5020-BP), human bone marrow-derived mesenchymal stromal cells (hMSC) recombinant human (rh) BMP4 (Cat# 314-BP), rhM-CSF (Cat# from healthy donors were seeded on biphasic calcium phosphate 216-MC), rhRANKL (Cat# 390-TN), neutralizing BMP4 antibody (BCP) scaffolds and differentiated toward osteoblasts for 1 week (Cat# MAB50201), and rat isotype control (MAB006) were from in vitro. Then, four cell-containing scaffolds were inserted subcuta- R&D Systems (Bio-Techne, Abingdon, UK). neously on the back of 20 14-week old RAG2−/− γC−/− female mice and left for further differentiation of the cells. After 8 weeks, mice were treated with recombinant adeno-associated virus (AAV) of 2.2 Generation of iRFP-labeled KJON myeloma cells serotype 8, AAV8-CTRL (n = 10) or AAV8-BMP4 (n = 10) (1012 viral The iRFP sequence was amplified from piRFP (gift from Vladislav Ver- particles in 100 μL), by tail vein injections. The BMP4 is a murine khusha, Addgene plasmid# 31857; http://n2t.net/addgene:31857; codon-optimized sequence. These viruses contain the human RRID:Addgene_31857).(20) with PCR primers with overhangs contain- α1-antitrypsin (hAAT1) promoter to ensure transgene expression ing restriction sites for SpeI and NotI (Sigma). pLVX-EF1α-IRES- in the liver. The AAVs were produced and purified as described ZsGreen1 (Cat# 631982, Clontech, Takara Bio USA, CA, USA) was previously.(22) After another 2 weeks, the mice were whole-body cut with SpeI and NotI restriction enzymes, treated with FastAP, irradiated using a dose of 2 Gy photons on the day before n 2 of 9 WESTHRIN ET AL. JBMR Plus (WOA) injection of 106 KJON cells into three of the scaffolds. The fourth further processing. The femurs were then homogenized using scaffold was used as a non-tumor cell control. An overview of Metal Bead Lysing Matrix (MP Biomedicals, LLC, OH, USA) and Tri- the experimental set-up is shown in Fig. 1. Treatment with AAV zol (Thermo Fisher Scientific). Samples of mRNA were reversely is regarded relatively safe,(23) and in line with this, AAV8 treatment transcribed and RT-PCR analysis was performed using TaqMan did not cause any adverse events in mice. However, one mouse in Gene Expression Arrays (Applied Biosystems, Thermo Fisher the AAV8-BMP4 group passed away for unknown reason. Tumor Scienctific) as described previously.(24) The primers are listed in load (iRFP intensity) was measured weekly using Pearl Imager Supplementary Table S1. Genes with a Ct value ≥36 were consid- and analyzed with the accompanying Image Studio software (LI- ered as not detected. StepOne Software v2.1 (Applied Biosys- COR Biosciences, Lincoln, NE, USA). At week 6 post myeloma cell tems) was used to analyze the samples and the comparative Ct injection, the mice were sacrificed, and serum, organs and bones method was used to estimate relative changes in gene expres- were harvested. All mice were euthanized before tumors reached sion using Gapdh as housekeeping gene. 1 cm3 (− scaffold) or when the tumor affected animal well-being. Animal handling and procedures were approved by the Norwe- 2.8 Osteoclast differentiation gian food safety authority (FOTS7692). The experiment was not blinded. Peripheral blood mononuclear cells (PBMCs) were isolated from healthy donors using Lymphoprep (Axis-Shield, Oslo, Norway). CD14+ cells were further purified using magnetic beads 2.6 Western blotting (Miltenyi). Cells were seeded out in 96 well plates and cultured Mouse liver cells were prepared by cutting the liver in smaller pieces in aMEM with 10% heat inactivated HS and M-CSF (30 ng/mL) using a scalpel, before they were dissociated using gentleMACS™ for 2 days. At this point rhBMP4 (20–200 ng/mL) or rhRANKL Dissociator (Miltenyi Biotech, Bergisch Gladbach, Germany). Pelleted (100 ng/mL) was added as indicated in the figure legend. When cells were lysed in lysis buffer (50 mM Tris–HCl (pH 7.5), 1% IGEPAL multinuclear cells were visible with light microscopy, cells were CA-630 (Sigma-Aldrich), 150 mM NaCl, 10% glycerol, 1 mM Na3VO4, fixed and stained for TRAP using Acid Phosphatase, Leukocyte 50 mM NaF and protease inhibitor cocktail (Roche, Basel, Switzer- (TRAP) kit, (Merck KGaA, Darmstadt, Germany). TRAP positive land)). For serum samples, 5 uL of serum was used per well. The cells with 3 or more nuclei were counted. samples were denatured in NuPage LDS sample buffer (Invitrogen, Thermo Fisher Scientific) supplemented with 25 mM 2.9 Scaffold bone analysis dithiothreitol (DTT) for 10 minute at 70 C before they were sepa- The scaffolds were harvested at end point and decalcified using rated on 4–12% Bis-Tris polyacrylamide gels with MES buffer Osteosoft (Merck). After approximately 4 weeks, when scaffolds (Invitrogen), and transferred to a nitrocellulose membrane using were soft to the touch, they were embedded in paraffin, sec- the iBlot Dry Blotting System (Invitrogen). The membrane was tioned (3.5 μm) and stained with hematoxylin and eosin. Images blocked using nonfat dry milk (5%) diluted in Tris-buffered saline were acquired using Nikon Microscope ECLIPSE Ci-S. The amount with 0.01% Tween 20 (TBS-T). The primary antibodies were: Mouse of bone and total scaffold perimeter were quantified using NIS anti-BMP4 (Cat# ab93939, RRID:AB_10562295) and mouse anti- Elements (BR 4.00.00, Nikon). GAPDH (Cat# ab8245, RRID:AB_2107448) (Abcam, Cambridge, UK). Blots were incubated with horseradish peroxidase (HRP) conjugated 2.10 μCt analysis secondary antibodies (DAKO Cytomation, Glostrup, Denmark) and developed with SuperSignal West Femto Maximum Sensitivity Sub- Femurs were harvested and examined by ex vivo μCT using a μCt strate (Thermo Fisher Scientific). Images were obtained with Odys- scanner (Skyscan 1176, Bruker, Kontich, Belgium) as sey FC and analyzed using Image Studio Software (LI-COR). described.(25) Images were acquired using the following settings: 18-μm voxel resolution, 0.5-mm aluminum filter, 50-kV voltage, and 500-μA current, 252 ms exposure time, rotation 0.5 degrees, 2.7 Real-time RT-PCR frame averaging = 4. Images were reconstructed and analyzed Liver cells were dissociated as described in the western blotting using SkyScan software programs NRecon (version 1.6.9.4), Data- section and mRNA was isolated from the cells. Femurs from all Viewer (version 1.4.4), and CT Analyzer (version 1.12.10.0). Femo- mice were harvested, flushed and kept in liquid nitrogen until ral trabecular analysis region of interest (ROI) was determined by Fig. 1. Overview of experimental set up. In brief, 14 weeks old RAG2−/− γC−/− mice were implanted with calcium phosphate scaffolds containing human MSCs. AAV8-BMP4 or AAV8-CTRL were administered 8 weeks post scaffold implantation by tail-vein injection (1012 viral particles/100 μL saline/mouse). 10 weeks post implantation 106 fluorescently labeled KJON myeloma cells were injected into 3 out of 4 scaffolds. Empty scaffold refers to scaffold with MSCs, but without tumor cells. JBMR® Plus BMP4 AND MYELOMA 3 of 9 n identifying the distal end of the femur and calculating 10% of the Continuous drug exposure may generate acquired resistance total femur length toward the femora mid-shaft, where we in multiple myeloma.(26) To examine if the tumor cells became then analyzed an ROI of 15% of the total femur length. Analy- resistant to BMP4 during the treatment, we isolated live cells sis of bone structure was completed using adaptive thresh- from tumors from both AAV8-BMP4 and AAV8-CTRL mice. Cells olding (mean of minimum and maximum values) in CT were treated with BMP4 in vitro and cell viability was measured Analyzer. Thresholds for analysis were determined manually by quantifying ATP levels using CellTiter Glo. Cells from both based on grayscale values (0–255, where 0 = black and groups were equally sensitive to the BMP4 treatment, indicating 255 = white) and were set as 36 to 255. Cortical analyses were that they did not acquire resistance to BMP4 during the experi- performed 35% above the distal end of the femur toward the ment (Supplementary Fig. S2). femora mid-shaft, also with a 15% ROI with the threshold values set as 80 to 255. 3.2 The presence of myeloma cells reduces the amount Images were generated using CtVox (version 3.3) of bone in scaffolds (Skyscan 1.1.6.0). Human bone is generated on the scaffolds by osteoblasts that differentiate from human MSCs seeded on the scaffolds before 2.11 Statistical analysis implantation.(21) In this model, similar to what happens in mul- Statistical analysis was performed using GraphPad Prism 7.04. To tiple myeloma, bone formation in the scaffolds is impaired by compare two groups we used the unpaired, two-tailed t-test. To the presence of myeloma cells.(21) This was also the case here, compare more than 2 groups we performed 1- or 2-way ANOVA as we found less bone in scaffolds with myeloma cells com- with Dunn’s/Dunnett’s test or Bonferroni Post hoc test, respec- pared with the scaffolds without myeloma cells (Fig. 3A, left, tively. Results were considered significant when p < .05. p < .05). Interestingly, this difference was lost in the AAV8-BMP4-treated mice (Fig. 3A, right panel). This could potentially be explained by an increased availability of BMP4 3. Results in the AAV8-BMP4 treated mice, promoting osteoblast differen- tiation and bone formation in the humanized bone scaffolds. 3.1 BMP4 gene therapy reduces tumor growth in vivo However, when we quantified amount of bone in scaffolds BMP4 induces apoptosis in approximately half of primary mye- without tumors, we found no significant difference between loma cell samples tested, as well as in a few myeloma cell lines.(7,9) AAV8-BMP4 treated mice compared with AAV8-CTRL mice Here, we wanted to use a cell line which is both sensitive to BMP4 (Fig. 3A). Thus, BMP4 did not have any beneficial effect on bone and relies on a supportive tumor microenvironment to grow formation in the scaffolds. Representative images of scaffolds in vivo. KJON is a slowly proliferating, IL-6 dependent cell line that are shown (Fig. 3B, Supplementary Fig. S3). is sensitive to BMP9.(10,19) BMP4 reduces viability in this cell line in vitro, as shown by a reduction in annexin V/PI negative cells 3.3 High circulating levels of BMP4 reduces mouse (Fig. 2A). Representative plots showing gating for annexin V/PI trabecular bone volume staining are shown in Fig. 2B. To investigate if BMP4 would also reduce in vivo tumor growth, we utilized a human-mouse scaffold The myeloma cells were confined to the scaffolds, as we could model of multiple myeloma(21,22) where we injected AAV8-BMP4 not detect plasma cells in the spleen or murine bone marrow or AAV8-CTRL viral particles intravenously. When recombinant when examined by imaging or flow cytometry (data not BMP4 was detectable in blood plasma (data not shown), fluores- shown). cently labeled KJON myeloma cells were injected directly into We next aimed to investigate the effects of the elevated levels the scaffolds. of circulating BMP4 on bone alone, without influence from the Tumor growth was examined by weekly imaging for 6 weeks cancer cells. First, we established that the tumors did not affect and we found a significant reduction in tumor load in the mouse bones by some soluble factor secreted from the AAV8-BMP4 mice compared with the AAV8-CTRL mice (p < .01, tumor cells. This was done by comparing cortical and trabecular Fig. 2C,D). At this time point, serum levels of BMP4 were high bone parameters in femurs obtained from un-implanted control (50–200 ng/mL) in the AAV8-BMP4 mice (Fig. 2E). The BMP4 was mice (untreated RAG2−/− γC−/−) with the AAV8-CTRL mice. As biologically active, since addition of sera obtained from the shown in Table 1 there was no difference in trabecular or cortical AAV8-BMP4-treated mice to the murine myeloma cell line NS0, bone parameters between the groups of mice when examined led to reduction in cell viability, e.g. ATP levels, and adding a by ex vivo μCT. Thus, the presence of tumor/scaffold had no BMP4-neutralizing antibody restored viability (Fig. 2F). In contrast, apparent effect on mouse bone and the femurs in the NS0 viability was not affected by sera obtained from AAV8-CTRL AAV8-CTRL mice can be considered “naïve”. mice. Further supporting successful transduction, we could detect We next examined the effect of BMP4 on mouse bone by com- the pro-form of BMP4 in liver lysates from AAV8-BMP4 mice, but paring properties of femurs obtained from AAV8-BMP4 treated not from AAV8-CTRL mice (Supplementary Fig. S1A,B), and we also mice with the AAV8-CTRL treated mice. Surprisingly, the observed that AAV8-BMP4 treatment increased downstream BMP AAV8-BMP4 mice had significantly reduced trabecular bone vol- targets such as Smad7 and Id1 in the liver (Supplementary Fig. S1C, ume (bone volume per tissue volume; % BV/TV, p = .017), trabec- D). To examine if the BMP4 transgene would alter the expression ular numbers (Tb.N/mm, p = .016) as well as significantly of endogenous BMP4 or other BMPs, we analyzed the mRNA increased trabecular separation (Tb.Sp (mm)) compared with expression of mouse Bmp4, Bmp2, Bmp6, and Gdf2 (BMP9) in the the AAV8-CTRL mice (p = .0005, Fig. 3C-F). Thus, high levels of cir- liver. The expression of these did not change (Supplementary culating BMP4 were detrimental for trabecular bone. In contrast, Fig. S1E-H). Taken together, these data demonstrated that AAV8- there were no differences in cortical bone parameters between BMP4 gene therapy lead to the production of high amounts of cir- the two groups (Supplementary Fig. S4). Moreover, since culating, biologically active BMP4 that reduced tumor growth. RAG2−/− γC−/− mice are immunocompromised, we performed n 4 of 9 WESTHRIN ET AL. JBMR Plus (WOA) Fig. 2. BMP4 inhibited myeloma cell growth in vivo. (A) the myeloma cell line KJON was treated with different doses of rmBMP-4. After 48 h, the numbers of viable cells were determined by labeling with annexin V-FITC and propidium iodide (PI). The amount of annexin V/PI negative cells was plotted, n = 3 independent experiments. ****; p < .0001, 1-way ANOVA, Dunnett’s multiple comparisons test. (B) the figure shows representative dot plots of the results presented in a. cells in the lower left quadrant, which were negative for both annexin V and PI, were considered viable. (C) To estimate tumor burden, the amount of near-infrared fluorescent protein (iRFP) in each scaffold was measured weekly using the pearl imager system in AAV8-CTRL mice (n = 30) and AAV8-BMP4 mice (n = 27), p < .01, 2-way ANOVA, Bonferroni post test. (D) Representative images of tumor burden in AAV8-CTRL (top) and AAV8-BMP4 (bottom) treated mice are shown. (E) Amount of BMP4 in the serum was estimated at end point by semi-quantitative western blotting. (F) Serum from AAV8-CTRL or AAV8-BMP4 mice, 4% final serum concentration, was added to cultures of NS0 murine myeloma cells and incubated for 48 h. the graph shows relative ATP-levels compared to medium control (not shown) as a measure of reduced cell viability. The reduction in cell viability was counteracted by a BMP4 neutralizing antibody, p < .0005, Bonferroni post test. all error bars represent SEM. a small study using the same viral vectors in female immuno- 3.4 Effects of BMP4 gene therapy on osteoblasts and competent C57BL6/N mice (n = 4/5 group). Although the reduc- osteoclasts tion in bone volume (% BV/TV) in AAV8-BMP4 compared to AAV8-CTRL mice was not significant, there was a significant In an attempt to determine whether the BMP4-induced effects on increase in trabecular separation (Tb.Sp) and a reduction in tra- bone were caused by increased bone resorption or decreased becular numbers (Tb.N/mm) in AAV8-BMP4 treated mice com- bone formation in the RAG2−/− γC−/− mice we measured the bone pared to AAV8-CTRL mice, supporting that high levels of BMP4 degradation marker C-terminal telopeptide of Type I collagen may have a negative impact on trabecular bone also in immuno- (CTX-1) and bone formation marker type I pro-collagen N-terminal competent mice (Supplementary Fig. S5). pro-peptide (PINP) in mouse sera (Supplementary Fig. S6). JBMR® Plus BMP4 AND MYELOMA 5 of 9 n Fig. 3. Legend on next page. n 6 of 9 WESTHRIN ET AL. JBMR Plus (WOA) Table 1. Trabecular Bone Parameters in the Femurs of Untreated myeloma patients.(27) This suggests that BMPs can have anti- RAG2−/− γC−/− vs AAV8-CTRL Mice tumor effects in patients. We show here that BMP4 treatment p-value inhibited tumor growth in vivo, in line with previous studies exam- Mean  SEM (unpaired t-test) ining effects of BMP4 on tumor cell survival in vitro. Importantly, more than half of the patient-derived primary cells we tested AAV8-CTRL in vitro were sensitive to BMP4, which implies that BMP4 could AAV8-CTRL RAG2−/− γC−/− vs RAG2−/− γC−/− have beneficial effects in a large group of patients.(7,9) Multiple BV/TV (%) 2.425  0.55 3.291  0.94 .412 myeloma is a very heterogeneous disease, and the malignant cells Tb.Th (mm) 0.05684  0.00 0.05575  0.00 .5448 harbor different genetic aberrations.(27) About 50% of all myeloma Tb.Sp (mm) 0.4873  0.03 0.4668  0.04 .6882 patients have cancer cells that are hyperdiploid.(27) Despite this Tb.N/mm 0.4166  0.09 0.5843  0.16 .3302 high number of hyperdiploid cells, very few cell lines have been established with this genotype. Here, we used the hyperdiploid myeloma cell line, KJON, which has a relatively slow growth rate and relies on addition of recombinant IL-6 in the absence of a sup- However, there were no significant differences between the porting microenvironment, thus resembling what takes place in groups. We also analyzed the murine femurs to examine if high patients.(19) Such cells are usually not able to grow in a mouse BMP4 levels had altered osteoblast differentiation. Again, mRNA bone marrow microenvironment,(28) but in this model the human expression of osteocyte- and osteoblast-specific markers mesenchymal cells provide the support needed. The model also (Sclerostin (Sost), Dickkopf-related protein 1 (Dkk1), Runt-related recapitulates the tumor-induced bone loss, which is a characteris- transcription factor 2 (Runx2) and Osterix (Sp7)) did not differ tic feature of myeloma. Thus, we here show that BMP4 gene ther- between the groups (Supplementary Fig. S7). Although this may apy inhibited myeloma growth in a relevant in vivo model. suggest that osteoblast differentiation was not significantly Importantly, although we found a reduction of tumor size by affected, the variation in gene expression within groups were BMP4 gene therapy, it did not promote bone formation. To further high, which makes it hard to conclude on this matter. investigate if BMP4 can have a beneficial effect on tumor-bearing For the osteoclast-specific markers, Cathepsin K (Ctsk) and bone it would be interesting to explore the effect of BMP4 gene Nuclear factor of activated T-cells, cytoplasmic 1 (Nfatc1), we therapy in a syngeneic myeloma mouse model, perhaps also in found a significant increase in Ctsk in the AAV8-BMP4 mice combination with anti-resorptive treatment such as Denosumab (p < .05, Fig. 3G,H). We therefore investigated if BMP4 had an or bisphosphonates. osteoclast-promoting effect in vitro. Indeed, addition of recom- The bone-inducing effect of BMPs has been known since the binant human (rh) BMP4 to CD14+ osteoclast-precursors 1960s, and BMP2, BMP4-7 and BMP9 have all been appreciated increased osteoclast differentiation (Ctrl vs rhBMP4 (200 ng/ for their osteogenic potential.(12,29) We were therefore surprised mL), p < .05, Fig. 3I). Taken together our results suggest that to find that trabecular bone was significantly reduced in BMP4 has a negative impact on bone, at least in part, by increas- AAV8-BMP4 treated mice compared with AAV8-CTRL mice, and ing osteoclast numbers. that high levels of circulating BMP4 failed to promote the amount of bone in the scaffolds. To examine if this effect was 4. Discussion somehow specific for RAG2−/− γC−/− BALB/c mice, we performed a small study in immunocompetent C57BL6/N mice using the In this study we wanted to explore BMP4 gene therapy as a same AAV8-BMP4 vector. Also in these mice, overexpression of potential treatment for multiple myeloma in a human-mouse BMP4 decreased the amount of trabecular bone, while cortical model. We found that BMP4 gene therapy inhibited myeloma bone was unaffected. Our results are thus in contrast to previous tumor growth, but surprisingly reduced trabecular bone in mice. studies demonstrating that BMPs, including BMP4, promote BMPs, like TGF-β, usually act as tumor suppressors.(5) In multiple osteoblast differentiation and bone formation.(13–15,30) Further, myeloma, several different BMPs inhibit growth in vitro.(6–10) The in another study researchers found that injection of AAV2- abundance of different BMPs in the bone marrow is not known. BMP4 into the skeletal muscle of immunocompetent rats It was shown that BMP6 mRNA is expressed by both normal and resulted in new bone induced by endochondral ossification malignant plasma cells, and that high levels of BMP6 in myeloma already at week 3.(31) In contrast to our study, where we used cells was associated with a favorable prognosis in multiple AAV8-BMP4 to increase circulating levels of BMP4, it is likely that FIGURE 3 BMP4 effect on human and murine bone. (A) Amount of bone/scaffold perimeter is presented for both empty and tumor scaffold for AAV8-CTRL treated mice (n = 10) and AAV8-BMP4 treated mice (n = 9). The left femur from each mouse, AAV8-CTRL (n = 10) and AAV8-BMP4 (n = 9), was harvested and examined by ex vivo μCT. *; p < .05, 1-way ANOVA, Dunn’s multiple comparisons test. (B) Representative images of scaffolds stained with H&E. tumor cells (t, blue), bone (b, pink) and scaffolds (s, black) are presented (larger sections of the scaffolds can be seen in Supplementary Fig. S3). (C) Trabecular volume as a proportion of tissue volume (BV/TV, %), (D) trabecular number (Tb. N, mm−1) and (E) trabecular separation (Tb.Sp, mm) was assessed. error bars represent SEM (A-E). *; p < .05, ***; p < .005, two-tailed unpaired t-test. (F) Representative images for an AAV8-CTRL mouse and an AAV8-BMP4 mouse are shown. Femur cDNA from AAV8-CTRL mice (n = 10) and AAV8-BMP4 mice (n = 8) was used for comparative RT-PCR using TaqMan assays for the osteoclast specific markers Ctsk (F) and Nfatc1 (G-H). The relative gene expression was analyzed using the ΔΔCt method with Gapdh as housekeeping gene. Error bars represent min to max (G-H). *; p < .05, two-tailed unpaired t-test. (I) Cells were differentiated with M-CSF (30 ng/mL) in the presence or absence of rhBMP4 or RANKL (100 ng/mL) as indicated. TRAP positive cells with more than 2 nuclei were counted as osteoclasts and num- ber of osteoclasts are presented as relative to M-CSF-treated cells. Presented is the mean of six independent experiments (RANKL, n = 3) and error bars represent SEM. *; p < .05, 1-way ANOVA, Dunnett’s multiple comparisons test. representative images, 20X and 40X magnification, are shown to the right. JBMR® Plus BMP4 AND MYELOMA 7 of 9 n the AAV2-BMP4 mainly increased BMP4 locally, thus explaining Acknowledgments the different outcomes. On the other hand, and in line with our data, BMP4 overexpression in bone caused severe osteopenia This work was supported by funds from the Norwegian Cancer and increased osteoclast number in mice.(32) The same study also Society (#5793765 and #4500930), the Norwegian Research found that overexpression of noggin, a BMP antagonist, had Council (#223255 and #193072), the Liaison Committee for edu- opposite effects. BMP2 and BMP4 have high affinity for BMPR1A, cation, research and innovation in Central Norway (90061001) and treating mice with a decoy receptor for these BMPs (i.e. a and Ministerio de Economía y Competitividad (MINECO) BMPR1A Fc-fusion protein) also led to increased numbers of (SAF2014-54866R), Spain. The authors would like to thank Kris- osteoblasts and reduced numbers of osteoclasts, resulting in tine Misund for cloning the pLVX-EF1α-iRFP-IRES-ZsGreen1 plas- higher bone mass in these mice.(17,18) Supporting these data, mid and Tonje Nedal for technical assistance. Parts of the in vivo conditional deletion of Bmpr1a in osteoblast-lineage cells has work was provided by the Comparative Medicine Core Facility been shown to increase bone mass in mice.(33,34) In conclusion, (CoMed), Norwegian University of Science and Technology our results that BMP4 has a negative effect on trabecular bone (NTNU). CoMed is funded by the Faculty of Medicine at NTNU in mice is supported by previous reports. and Central Norway Regional Health Authority. A weakness of our study is the lack of bone histomorpho- metric data. We can therefore only speculate on how BMP4 acted at the cellular level to reduce trabecular bone. Expression of Authorship Contribution mRNA in femurs flushed of bone marrow showed significantly higher expression of Ctsk in the AAV8-BMP4 treated mice, which MW and TH designed the study, performed experiments, ana- may indicate that there were more osteoclasts in these bones. lyzed data and wrote the paper; SHM, MZ and AMS performed There was, however, a great variation in gene expression within mouse experiments; GB assisted with mouse experiments and groups and no differences in serum CTX-1 levels, so we could not analyzed data; BS and HH performed experiments; YH/JDdB pro- make a conclusion on this matter. While BMPs in bone have vided scaffolds; AM/RWJG provided mice and designed the mainly been studied in relation to effects on the osteoblast line- study; FB provided virus and designed the study; US/AS age, a few studies have earlier shown that BMP2, BMP4, BMP5, designed the study; TS designed the study, analyzed data and and BMP6 promote osteoclast differentiation in vitro and/or wrote the paper. All authors revised the manuscript before in vivo, supporting our findings.(32,35,36) An indirect effect of submission. BMP2 and BMP4 on osteoclasts was shown in a study where Bmpr1a was conditionally deleted in osteocytes leading References to decreased RANKL expression in osteocytes and reduced osteoclast differentiation.(33) The importance of osteoblastic 1. Bianchi G, Munshi NC. Pathogenesis beyond the cancer clone(s) in BMP-signaling for RANKL expression was also shown in other multiple myeloma. Blood. 2015;125:3049–58. studies.(17,34) Together these studies and our data presented 2. Yaccoby S. Advances in the understanding of myeloma bone disease herein support that BMP-signaling impacts bone remodeling and tumour growth. Br J Haematol. 2010;149:311–21. by influencing both osteoblast and osteoclast differentiation. 3. Kumar SK, Rajkumar SV. The multiple myelomas - current concepts in cytogenetic classification and therapy. Nat Rev Clin Oncol. 2018;15: AAV mediated gene delivery methods are considered safe 409–21. and have high gene delivery efficacy and are therefore promis- 4. Ravi P, Kumar SK, Cerhan JR, et al. Defining cure in multiple myeloma: ing tools for gene therapy.(37) In adult mice, the same a comparative study of outcomes of young individuals with myeloma AAV8-BMP4 vector as used here was shown to increase insulin and curable hematologic malignancies. Blood Cancer J. 2018;8:26. sensitivity and protected mice on a high-fat diet from obe- 5. Davis H, Raja E, Miyazono K, Tsubakihara Y, Moustakas A. Mechanisms sity.(22) In line with this study,(22) we found that while control of action of bone morphogenetic proteins in cancer. Cytokine mice had a weight gain of about 5% in the 8 weeks from viral Growth Factor Rev. 2016;27:81–92. injections until culling, the weight of BMP4 treated mice 6. Kawamura C, Kizaki M, Yamato K, et al. Bone morphogenetic protein- remained unchanged during the course of the study. However, 2 induces apoptosis in human myeloma cells with modulation of it is unlikely that such a small difference in body weight will STAT3. Blood. 2000;96:2005–11. lead to a dramatic effect on trabecular bone. Moreover, in the 7. Hjertner O, Hjorth-Hansen H, Borset M, Seidel C, Waage A, Sundan A. Bone morphogenetic protein-4 inhibits proliferation and induces C57BL6/N immunocompetent mice there was no difference in apoptosis of multiple myeloma cells. Blood. 2001;97:516–22. body weight between the groups, and we could still observe 8. Ro TB, Holt RU, Brenne AT, et al. Bone morphogenetic protein-5, −6 a reduction in trabecular bone. and −7 inhibit growth and induce apoptosis in human myeloma Taken together, BMP4 gene therapy inhibited myeloma tumor cells. Oncogene. 2004;23:3024–32. growth, but also reduced the amount of trabecular bone in mice. 9. Holien T, Vatsveen TK, Hella H, et al. Bone morphogenetic proteins It is still unclear how BMP4 affects tumor-bearing bone and the induce apoptosis in multiple myeloma cells by Smad-dependent possibility of combining BMP4 therapy with anti-resorptive treat- repression of MYC. Leukemia. 2012;26:1073–80. ment such as denosumab or bisphosphonates should be 10. Olsen OE, Wader KF, Misund K, et al. Bone morphogenetic protein-9 explored in future studies. Whether other BMPs that are also suppresses growth of myeloma cells by signaling through ALK2 but is inhibited by endoglin. Blood Cancer J. 2014;4:e196. potent inhibitors of multiple myeloma cell survival and prolifera- tion will have similar impact on bone remains to be investigated. 11. Olsen OE, Sankar M, Elsaadi S, et al. BMPR2 inhibits activin and BMP signaling via wild-type ALK2. J Cell Sci. 2018;131. 12. Wu M, Chen G, Li Y-P. TGF-β and BMP signaling in osteoblast, skeletal development, and bone formation, homeostasis and disease. Bone Res. 2016;4:16009. Disclosures 13. Bandyopadhyay A, Tsuji K, Cox K, Harfe BD, Rosen V, Tabin CJ. Genetic analysis of the roles of BMP2, BMP4, and BMP7 in limb patterning and There are no conflicts of interest to declare. skeletogenesis. PLoS Genet. 2006;2:e216. n 8 of 9 WESTHRIN ET AL. JBMR Plus (WOA) 14. Abe E, Yamamoto M, Taguchi Y, et al. Essential requirement of BMPs- 26. Abdi J, Chen G, Chang H. Drug resistance in multiple myeloma: latest 2/4 for both osteoblast and osteoclast formation in murine bone findings and new concepts on molecular mechanisms. Oncotarget. marrow cultures from adult mice: antagonism by noggin. J Bone 2013;4:2186–207. Miner Res. 2000;15:663–73. 27. Chng WJ, Glebov O, Bergsagel PL, Kuehl WM. Genetic events in the 15. Mishina Y, Starbuck MW, Gentile MA, et al. Bone morphogenetic pro- pathogenesis of multiple myeloma. Best Pract Res Clin Haematol. tein type IA receptor signaling regulates postnatal osteoblast func- 2007;20:571–96. tion and bone remodeling. J Biol Chem. 2004;279:27560–6. 28. Burger R, Guenther A, Bakker F, et al. Gp130 and ras mediated signal- 16. Standal T, Abildgaard N, Fagerli U-M, et al. HGF inhibits BMP-induced ing in human plasma cell line INA-6: a cytokine-regulated tumor osteoblastogenesis: possible implications for the bone disease of model for plasmacytoma. Hematol J. 2001;2:42–53. multiple myeloma. Blood. 2007;109:3024–30. 29. Urist MR. Bone: formation by autoinduction. Science (New York, N.Y.). 17. Baud’huin M, Solban N, Cornwall-Brady M, et al. A soluble bone mor- 1965;150:893–9. phogenetic protein type IA receptor increases bone mass and bone strength. Proc Natl Acad Sci U S A. 2012;109:12207–12. 30. Khan MP, Khan K, Yadav PS, et al. BMP signaling is required for adult skeletal homeostasis and mediates bone anabolic action of parathy- 18. Ko FC, Van Vliet M, Ellman R, et al. Treatment with a soluble bone roid hormone. Bone. 2016;92:132–44. morphogenetic protein type 1A receptor (BMPR1A) fusion protein increases bone mass and bone formation in mice subjected to Hin- 31. Luk KD, Chen Y, Cheung KM, Kung HF, Lu WW, Leong JC. Adeno- dlimb unloading. JBMR Plus. 2017;1:66–72. associated virus-mediated bone morphogenetic protein-4 gene therapy for in vivo bone formation. Biochem Biophys Res Commun. 19. Våtsveen TK, Børset M, Dikic A, et al. VOLIN and KJON—Two novel 2003;308:636–45. hyperdiploid myeloma cell lines. Genes Chromosomes Cancer. 2016;55:890–901. 32. Okamoto M, Murai J, Yoshikawa H, Tsumaki N. Bone morphogenetic proteins in bone stimulate osteoclasts and osteoblasts during bone 20. Filonov GS, Piatkevich KD, Ting LM, Zhang J, Kim K, Verkhusha VV. development. J Bone Miner Res. 2006;21:1022–33. Bright and stable near-infrared fluorescent protein for in vivo imag- ing. Nat Biotechnol. 2011;29:757–61. 33. Kamiya N, Shuxian L, Yamaguchi R, et al. Targeted disruption of 21. Groen RWJ, Noort WA, Raymakers RA, et al. Reconstructing the BMP signaling through type IA receptor (BMPR1A) in osteocyte human hematopoietic niche in immunodeficient mice: opportunities suppresses SOST and RANKL, leading to dramatic increase in bone for studying primary multiple myeloma. Blood. 2012;120:e9. mass, bone mineral density and mechanical strength. Bone. 2016; 91:53–63. 22. Hoffmann JM, Grünberg JR, Church C, et al. BMP4 gene therapy in mature mice reduces BAT activation but protects from obesity by 34. Kamiya N, Ye L, Kobayashi T, et al. Disruption of BMP signaling in browning subcutaneous adipose tissue. Cell Rep. 2017;20:1038–49. osteoblasts through type IA receptor (BMPRIA) increases bone mass. J Bone Miner Res. 2008;23:2007–17. 23. Phillips JE, Gersbach CA, Garcia AJ. Virus-based gene therapy strate- gies for bone regeneration. Biomaterials. 2007;28:211–29. 35. Kaneko H, Arakawa T, Mano H, et al. Direct stimulation of osteoclastic bone resorption by bone morphogenetic protein (BMP)-2 and 24. Westhrin M, Moen SH, Holien T, et al. Growth differentiation factor expression of BMP receptors in mature osteoclasts. Bone. 2000;27: 15 (GDF15) promotes osteoclast differentiation and inhibits osteo- 479–86. blast differentiation and high serum GDF15 levels are associated with multiple myeloma bone disease. Haematologica. 2015;100:e511–4. 36. Wutzl A, Brozek W, Lernbass I, et al. Bone morphogenetic proteins 25. Johnson RW, Brennan HJ, Vrahnas C, et al. The primary function of 5 and 6 stimulate osteoclast generation. J Biomed Mater Res. 2006; gp130 signaling in osteoblasts is to maintain bone formation and 77:75–83. strength, rather than promote osteoclast formation. J Bone Miner 37. Kotterman MA, Schaffer DV. Engineering adeno-associated viruses Res. 2014;29:1492–505. for clinical gene therapy. Nat Rev Genet. 2014;15:445–51. JBMR® Plus BMP4 AND MYELOMA 9 of 9 n