An Lysophosphatidic Acid Receptors 1 and 3 Axis Governs Cellular Senescence of Mesenchymal Stromal Cells and Promotes Growth and Vascularization of Multiple Myeloma

Masahiko Kanehira 1, Tohru Fujiwara 1 2, Shinji Nakajima 3, Yoko Okitsu 1, Yasushi Onishi 1, Noriko Fukuhara 1, Ryo Ichinohasama 4, Yoshinori Okada 5, Hideo Harigae 1 2

ABSTRACT

Mesenchymal stromal cells (MSCs) are multipotent progenitor cells and there is much interest in how MSCs contribute to the regulation of the tumor microenvironment. Whether MSCs exert a supportive or suppressive effect on tumor progression is still controversial, but is likely dependent on a variety of factors that are tumor‐type dependent. Multiple myeloma (MM) is characterized by growth of malignant plasma cells in the bone marrow. It has been shown that the progression of MM is governed by MSCs, which act as a stroma of the myeloma cells. Although stroma is created via mutual communication between myeloma cells and MSCs, the mechanism is poorly understood. Here we explored the role of lysophosphatidic acid (LPA) signaling in cellular events where MSCs were converted into either MM‐supportive or MM‐suppressive stroma. We found that myeloma cells stimulate MSCs to produce autotaxin (ATX), an indispensable enzyme for the biosynthesis of LPA, and LPA receptor 1 (LPA1) and 3 (LPA3) transduce opposite signals to MSCs to determine the fate of MSCs. LPA3silenced MSCs (siLPA3‐MSCs) exhibited cellular senescence‐related phenotypes in vitro, and significantly promoted progression of MM and tumorrelated angiogenesis in vivo. In contrast, LPA1‐silenced MSCs (siLPA1MSCs) showed resistance to cellular senescence in vitro, and efficiently delayed progression of MM and tumor‐related angiogenesis in vivo. Consistently, anti‐MM effects obtained by LPA1‐silencing in MSCs were completely reproduced by systemic administration of Ki6425, an LPA1 antagonist. Collectively, our results indicate that LPA signaling determines the fate of MSCs and has potential as a therapeutic target in MM.

Key words. Hematologic malignancies • Marrow stromal stem cells • Mesenchymal stem cells • Marrow stromal cells • Tissue‐specific stem cells • Adult stem cells • Bone marrow stromal cells

SIGNIFICANCE STATEMENT

In tumor biology, it has remained unsolved whether tumor‐surrounding tissue is tumor‐supportive or tumor‐suppressive. This study clearly showed how tumor‐surrounding tissues will be converted into tumor‐supportive or tumor‐suppressive. Two cognate receptors, LPA1 and LPA3, transduce opposite signals to tumor‐surrounding cells and determine the fate of the cells toward pro‐senescent and anti‐senescent, respectively. The pro‐senescent STEM CELLS 2016;00:00‐00 www.StemCells.com cells readily converted into tumor‐ cells and delayed tumor growth. Furthermore, chemical agent which inhibits supportive cells and promote tu‐ LPA1 signal efficiently delayed tumor growth in mice. In summary, this study mor growth. Conversely, anti‐ not only unveiled the mechanism by which tumor‐surrounding tissue is essenescent cells were resistant to tablished but also provides a concept of new antitumor drugs.

INTRODUCTION

Mesenchymal stromal cells (MSCs) are multipotent adult stem cells capable of differentiating into several cells in mesoderm lineage such as osteoblasts, adipocytes and chondrocytes, and have been used in over 1,500 clinical trials [1]. In the bone marrow, it is thought that MSCs support hematopoiesis by contributing the hematopoietic niche [2]. Similarly, MSCs are thought to home to and contribute to the tumor microenvironment [3], where they can exert supportive or suppressive effects on tumor progression [4‐6]. This discrepancy would come from the secretory profile of MSCs which depends on heterogeneous stimuli from tumor cells and surrounding tissue, comprises MSCs as well as endothelial cells, macrophages, osteoblasts, osteoclasts, and other types of cells at variable frequencies [7, 8]. Unveiling the mechanisms underlying conversion of MSCs into either tumor‐supportive or tumorsuppressive will help understanding the nature of tumor stoma.
Multiple myeloma (MM) is a monoclonal tumor of plasma cells caused by the clonal expansion of a terminally differentiated B cells within the bone marrow. Despite recent progress in new therapeutic options, such as high‐dose chemotherapy followed by autologous stem cell transplantation, and development of drugs, the disease remains incurable [9‐12]. During MM progression, the bone marrow microenvironment plays crucial roles as a stroma via the promotion of growth, survival, and drug resistance of myeloma cells. Accumulating evidence indicates that MSCs in the bone marrow can act as stroma for myeloma cells and facilitate MM progression. MSCs are known to secrete several growth factors, cytokines, chemokines, exosomes, and matrixdegrading enzymes, which promote tumorigenesis and metastasis of myeloma cells [12‐15]. It has been proposed that MM‐supportive stroma is established via the mutual crosstalk between MSCs and myeloma cells in the bone marrow, but the details are still unknown [16, 17].
Lysophosphatidic acid (LPA) is a bioactive small glycerophospholipid derived from plasma membrane phospholipid which stimulates cell proliferation, migration, and survival. Extracellular LPA stimulates signaling via at least 6 cognate G protein‐coupled receptors (LPA1–6) [18, 19]. Although the etiological significance of the difference remains unknown, MM patients have higher LPA levels in their plasma than do healthy subjects [20]. Here we postulated that cellular events mediated via LPA signaling may contribute to the progression of MM.
We demonstrated that MSCs in which LPA3 signaling was impaired efficiently acquired cellular senescencerelated phenotypes and transdifferentiated into tumorassociated fibroblasts (TAFs) in response to stimulation by myeloma cells. The LPA3‐silenced MSCs significantly promoted MM progression and tumor‐related angiogenesis in an FGF2‐dependent manner. In contrast, MSCs in which LPA1 signaling was impaired were refractory to cellular senescence and transdifferentiation into TAFs and had less MM‐promoting abilities. Our results indicate that LPA‐induced cellular events in MSCs are critical steps for MM tumorigenesis through the establishment of tumor stroma.

METHODS

Cell culture

The human myeloma cell lines OPM‐2, IM‐9, and RPMI8226 were obtained from American Type Culture Collection (Manassas, VA). All myeloma cell lines were routinely cultured in RPMI‐1640 (Sigma‐Aldrich, St. Louis, MO) supplemented with 10% (v/v) FBS (Biological Industries, Kibbutz Beit‐Haemek, Israel) and 1% penicillin/streptomycin (Sigma‐Aldrich). Primary human bone marrow‐derived MSCs at passage 1 were kindly provided by Texas A&M Health Science Center (Temple, TX). Human MSCs were obtained from three donors, 21year‐old female donor 1, 22‐year‐old male donor 2, and 24‐year‐old male donor 3; MSCs from donor 1 were used at passage 3 in this study unless otherwise noted. MSCs were cultured in MEM‐alpha (Thermo Fisher Scientific, Waltham, MA) supplemented with 17% (v/v) FBS and 2 mM L‐glutamine (Thermo Fisher Scientific). To expand MSCs, a frozen vial (passage 2, 106 cells) was thawed and seeded in a 150‐mm dish (Corning Inc., Corning, NY). After 24 h, adherent (i.e., viable) cells were recovered with trypsin/EDTA, re‐seeded at a density of 60 cells/cm2 and cultured by changing media every 3 days. After 9 days, MSCs were collected and used for further experiments. Conditioned medium (CM) of three myeloma cell lines was obtained as follows: OPM‐2, IM‐9, or RPMI‐8226 cells (5 × 105) were cultured in 75‐cm2 flasks until they grew to confluence, and the media were centrifuged, passed through a 0.45m filter, and stored at 4°C until use. To obtain the CM of MSCs, MSCs (1 × 105) were cultured in 6‐well plates for 24 h, and the cells were aspirated, washed twice with PBS, and stimulated individually with CM from the three myeloma cell lines. After 48 h, the media were centrifuged, passed through a 0.45‐m filter, and stored at 4°C until use. MSCs were also stimulated with CM from IM‐9 myeloma cells in the presence of the TLR4 inhibitor peptide VIPER (Bio‐Techne, Minneapolis, MN) or a selective NF‐B inhibitor JSH‐23 (Sigma‐Aldrich) for 48 h in the same manner as described above.

Quantification of the concentration and enzymatic activity of ATX in media

ATX concentrations in the CM from the three myeloma cell lines and the CM from MSCs were measured using a Human ENPP‐2/Autotaxin Quantikine ELISA Kit (BioTechne) according to the manufacturer’s instructions. The enzymatic activity of ATX in all CM samples was also quantified using FS‐3 (Echelon Biosciences, Salt Lake City, UT). In brief, FS‐3, a fluorescent substrate of ATX, was mixed with the CM and incubated for 24 h at 37°C, and changes in fluorescent intensity were measured using a FlexStation 3 Multi‐Mode Microplate Reader (Molecular Devices, Sunnyvale, CA).

RNAi knockdown

Transfection with siRNAs was performed to achieve in vitro gene‐silencing. At 24 h before transfection, 4 × 105 MSCs were plated onto 100‐mm dishes. MSCs were transfected with 75 pmol of siRNAs using Lipofectamine RNAi MAX (Thermo Fisher Scientific) according to the manufacturer’s instructions. All siRNAs used in this study were siGENOME SMARTpool siRNAs purchased from Dharmacom (GE Healthcare Dharmacon, Lafayette, CO). The SMARTpool siRNAs used in this study were as follows: Non‐Targeting siRNA Pool #1, Human LPAR1 (1902), Human LPAR2 (9170), Human LPAR3 (23566), Human LPAR4 (2846), Human LPAR5 (57121), Human LPAR6 (10161), and Human FGF2 (2247).

Flow cytometry

The percentage of MSCs at the G1 phase of cell cycle was analyzed using a FACSAria II cell sorter (BD Biosciences, San Jose, CA) by staining DNA and RNA with 7AAD (Bio‐Techne) and Pyronin Y (Sigma‐Aldrich), respectively. The percentage of MSCs at the S phase of cell cycle was analyzed using the APC‐BrdU Flow Kit (BD Biosciences) and a FACSAria II cell sorter (BD Biosciences) according to the manufacturer’s instruction. The detailed procedure is described in Supplemental methods.

Quantitative RT‐PCR

Total RNA was extracted from MSCs using RNeasy Plus Mini Kit (Qiagen, Germany) and converted into cDNA using the SuperScript III First‐Strand Synthesis System for RT‐PCR (Bio‐Techne) according to the manufacturer’s instruction. Quantitative RT‐PCR was performed using SYBR Green ER qPCR SuperMix (Thermo Fisher Scientific) according to the manufacturer’s instruction.The sequences of primers and detailed procedure aredescribed in Supplemental methods.

In vitro cell proliferation assay

Cell proliferation assays were performed using the CellTiter 96 Aqueous Non‐Radioactive Cell Proliferation Assay (Promega, Madison, WI) according to the manufacturer’s instructions or by directly counting the number of cells. In brief, MSCs from 3 donors at passage 3 were silenced with siRNAs that targeted the 6 LPA receptors (LPA1–6) and seeded onto 96‐well plates or 6well plate at a density of 2.5 × 104 cells/200 l in triplicate or 4.0 × 105 cells/3 ml, respectively, before stimulating with IM‐9 CM. After 48 h, 20 l of the reagent was added to each well of 96‐well plate, before incubating at 37°C for 30 min and measuring the absorbance at 490 nm using an iMark Microplate Reader with Microplate Manager 6 software (Bio‐Rad Laboratories, Hercules, CA). At the same time, cells were recovered with trypsin/EDTA and counted manually under a microscope using a hemocytometer.

SA‐‐gal assay

MSCs silenced with siRNAs against LPA1 and LPA3 were stained with SA‐‐gal solution [1 mg/ml X‐gal, 4 mM potassium ferricyanide, 4 mM potassium ferrocyanide, 2mM MgCl2 in PBS (pH6.0)] for 24 h at 37°C and then observed using light microscopy. SA‐‐gal activity was determined on the basis of absorbance at an OD of 650 nm using a FlexStation 3 Multi‐Mode Microplate Reader (Molecular Devices).

Animal studies

Procedures were performed using 4‐week‐old female athymic BALB/c‐nu/nu mice. All mice were purchased from CLEA Japan, Inc. (Tokyo, Japan) and maintained under sterile conditions. Studies involving animals, including housing and care, method of euthanasia, and experimental protocols, were performed according to protocols approved by Tohoku University’s Institutional Committee for the Use and Care of Laboratory Animals. For the tumor xenograft experiments, 1 × 106 IM‐9 cells were co‐transplanted with 4 × 105 MSCs silenced with siRNAs against LPA1, LPA3, FGF2 or both LPA3 and FGF2 in 100 l PBS into the left flank of nude mice under ketamine/xylazine anesthesia. Mice injected only with either IM‐9 cells or MSCs were used as controls. Tumor growth was measured every 2 days using calipers, and tumor volume was calculated according to the following formula: tumor volume (mm3) = 0.52 × [length (mm)] × [width (mm)]2. All mice were euthanized after 20 days and the tumor masses were collected and weighed. See also Supplemental methods.

Bioluminescent imaging

To generate myeloma cell line stably expressing luciferase protein, we first constructed an expression vector containing a luciferase gene. The firefly luciferase gene was amplified by PCR using plasmid pGL4.10 [luc2] (Promega) as a template and using the primer pair which creates NotI and EcoRI restriction sites: forward 5’‐AAGCGGCCGCCATGGAAGARGCCAAAAACAT‐3’, reverse, 5’‐TTGAATTCTTACACGGCGATCTTGCCGC‐3’. The obtained fragment was subcloned into the NotI/EcoRI site of pQCXIN vector (Clontech Laboratories, Mountain View, CA). IM‐9 myeloma cells were transfected with a luciferase‐expressing plasmid (pQCXIN‐luc) with an Amaxa Cell Line Nucleofector Kit C using Nucleofector 2b (Lonza, Switzerland), according to the manufacturer’s instructions. After 48 h, transfected cells were selected by adding 1000 g/ml G‐418 (Roche Diagnostics, Indianapolis, IN) to the medium. At 25 days after transfection, IM‐9 cells that stably transduced the luciferase gene (IM‐9‐luc+ cells) were used for in vivo experiments. To monitor tumor growth noninvasively and quantitatively, 1 × 106 IM‐9‐luc+ cells were co‐transplanted with 4 × 105 MSCs silenced with siRNAs against LPA1, LPA3, FGF2 or both LPA3 and FGF2 in 100 l PBS into the left flank of nude mice under ketamine/xylazine anesthesia. At 10, 15, and 20 days after IM‐9‐luc+ injection, the mice were intraperitoneally administered with 150 mg/kg sterile D‐Luciferin (PerkinElmer, Waltham, MA) dissolved in PBS. At 10 min after substrate injection, the mice were anesthetized using 2% isoflurane and imaged for 1 s at the maximal light collection rate and the highest resolution using the IVIS Spectrum CT Preclinical In Vivo Imaging System (PerkinElmer).

ELISA

FGF2 concentrations in culture media were determined using ELISA kit (Bio‐Techne) according to the manufacturer’s instructions. LPA levels in sera were detemined using human lysophosphatidic acid (LPA) ELISA kit (Cusabio Biotech, College Park, MD) according to the manufacturer’s instructions. The detailed procedure is described in Supplemental methods.

Histopathological analysis

At 20 days after co‐transplantation of IM‐9 with MSCs into mice, the mice were euthanized and the tumor masses were collected. Before pathological examinations, the collected tumor masses were fixed in 15% formalin neutral buffer, embedded in paraffin blocks, and sectioned using a microtome to prepare tissue slides. Tissue slides were stained with hematoxylin and eosin (H&E) and immunostained with anti‐human CD138 (Agilent Technologies, Santa Clara, CA), antihuman kappa light chain (Agilent Technologies), antihuman lambda chain (Agilent Technologies), or anti‐Ki67 antibodies (Agilent Technologies). To analyze tumorrelated angiogenesis, the collected tumor masses were fixed in 4% paraformaldehyde, embedded in OCT compound (Sakura Finetech, Tokyo, Japan), and frozen at −80°C. Cryosections were prepared from the frozen blocks using a cryostat and stained with anti‐mouse CD31 antibody (BioLegend, San Diego, CA). To evaluate tumor‐related angiogenesis, the CD31‐positive area was quantified in 9 randomly selected fields from 3 mice ineach group of mice using Image J software.

Immunocytochemistry

MSCs silenced with siRNAs against LPA1 or LPA3 were grown in 6‐well plates to 50%–60% confluency. After 24 h, the MSCs were stimulated with IM‐9 CM for 72 h. The MSCs were then fixed in 4% paraformaldehyde and stained with anti‐human ACTA2 (‐SMA) antibody (LifeSpan BioSciences, Seattle, WA).

Generation of retroviral vector for LPA1 expression

The entire open reading frame (ORF) of human LPA1 was subcloned into pQCXIN retroviral vector (Clontech Laboratories). The pQCXIN vector expressing LPA1 was transfected into HEK293gp cells (Riken Bioresource Center, Japan). After 48 hours, culture supernatants were collected and used for infection of MSCs. The sequences of primers and detailed procedure are described in Supplemental methods.

MSCs isolation from bone marrow aspirates

Mononuclear cells in total bone marrow aspirates were labelled with MSC phenotyping kit (Miltenyi Biotec GmbH, Germany) according to the manufacturer’s instruction and Lin‐/CD90+/CD105+ cells were sorted using a FACSAria II cell sorter with FACS Diva software (BD Biosciences). The detailed procedure is described in Supplemental methods.

Statistical analyses

All data are presented as mean and SD of the indicated number of samples and replicates. Significance testing was performed with the Student’s two‐tailed t test. Pvalues of < 0.05 were considered statistically significant.

RESULTS

MSCs upregulate and secrete autotaxin (ATX) in response to myeloma cells via a Toll‐like receptor 4 (TLR4)‐NF‐B‐dependent pathway Since patients with MM show higher levels of plasma LPA, we measured the ATX levels in conditioned medium (CM) from three myeloma cell lines (IM‐9, OPM‐2, and RPMI‐8226) and MSCs. ATX was detected in the conditioned media from MSCs but not all CM from the three myeloma cell lines (Figure 1A). Notably, ATX production from MSCs was significantly promoted by stimulation with the CM from the three myeloma cell lines (Figure 1A). The enzymatic activity of the ATX was measured by FS‐3 assay. Conditioned medium from myeloma cell‐stimulated MSCs increased ATX activity almost 2‐fold (Figure 1B). We then explored the mechanism by which myeloma cells stimulated MSCs to produce ATX. Blockage of NF‐B and TLR4 activation using antagonists markedly reduced ATX production from MSCs stimulated with the CM from IM‐9 (IM‐9 CM) in a dose‐dependent manner (Figure 1C and 1D). In summary, we conclude that ATX production from MSCs is stimulated by myeloma cells via a TLR4‐NF‐Bdependent pathway (Figure 1E).

LPA1 and LPA3 signaling regulate cellular senescence in MSCs positively and negatively, respectively

Previously, we reported that LPA signaling contributes to the progression of cellular senescence in MSCs [21]. To identify the LPA receptor(s) responsible for cellular senescence in MSCs, we silenced the LPA1–6 genes in MSCs using siRNAs and monitored their proliferation. First, we confirmed the accuracy and specificity of siRNAs for LPA1–6 in MSCs. As shown in Supporting information Figure S1A, siRNAs used in this study efficiently and specifically silenced their target genes. LPA6 mRNA was absent in MSCs. Well‐known characteristics of cellular senescence are upregulation of senescenceassociated ‐galactosidase (SA‐‐gal) activity and marked changes in morphology [22]. As shown in Figure 2A, SA‐‐gal activity was decreased from 100% (control) to 57% in LPA1‐siRNA‐transfected MSCs (siLPA1‐MSCs). Conversely, SA‐‐gal activity was increased from 100% (control) to 229% in LPA3‐siRNA‐transfected MSCs (siLPA3‐MSCs). In addition, siLPA1‐MSCs retained a small and more spindle‐shaped morphology compared with control‐MSCs, although siLPA3‐MSCs had a large and flattened morphology compared with control‐MSCs (Figure 2B). The cell size was quantified by measuring the forward scatter (FSC) intensity. Similar to the morphological observations, siLPA1‐MSCs and siLPA3‐MSCs were smaller and bigger in size than control‐MSCs, respectively (Figure 2B). Another characteristic of cellular senescence is cell cycle arrest. As shown in Figure 2C, proliferation of MSCs from 3 different donors was enhanced from 100% in control‐MSCs to 137% in siLPA1MSCs and decreased from 100% in control‐MSCs to 75% in siLPA3‐MSCs. Consistently, number of MSCs from 3 different donors was enhanced from 100% in controlMSCs to 117% in siLPA1‐MSCs and decreased from 100% in control‐MSCs to 81% in siLPA3‐MSCs (Supporting information Fig. S2). To give evidence to our results, the percentage of MSCs arrested their cell cycle in G1 phase and in proliferation (S phase) was determined. The percentage of cells in the G1 phase (DNA content2n/Pyronin Yhi) was significantly decreased by 69.6% in siLPA1‐MSCs compared with control‐MSCs (80.1%, P < 0.001; Supporting information Fig. S3A). In a further analysis, the percentage of cells in the S phase was significantly increased by 9.2% in siLPA1‐MSCs compared with control‐MSCs (7.8%, P = 0.003; Supporting information Fig. S3B). Quantitative RT‐PCR (qRT‐PCR) analysis demonstrated that the mRNA levels of CDK2 and Cyclin E1 (CCNE1) were upregulated by approximately 1.5–2‐fold compared with control‐MSCs (Supporting information Fig. S3C). On the other hand, the percentage of cells in the G1 phase was significantly increased by 77.8% in siLPA3‐MSCs compared with control‐MSCs (65.5%, P = 0.0039; Supporting information Fig. S4A). In addition, the percentage of MSCs in the S phase was significantly inhibited by 0.3% in siLPA3‐MSCs compared with control‐MSCs (4.3%, P < 0.001; Supporting information Fig. S4B). qRT‐PCR analysis demonstrated that the mRNA levels of CDK4 and Cyclin D1 (CCND1) were downregulated by approximately 50% in siLPA3‐MSCs compared with control‐MSCs (Supporting information Fig. S4C). These results strongly suggest that LPA1 and LPA3 signaling have opposing effect in cellular senescence in MSCs by acting positive and negative regulator, respectively.

LPA3 reduction in MSCs promotes MM progression in a murine xenograft model

To explore the relationship between MSC cellular senescence and MM progression, we established a xenograft model of human MM by co‐transplanting IM‐9 myeloma cells with MSCs in athymic nude mice (Figure 3A, left). Considering that IM‐9 myeloma cells rapidly and reproducibly formed palpable tumor masses in mice compared with other MM cell lines (OPM‐2 and RPMI‐8226), we used IM‐9 myeloma cells for our in vivo experiments. For pathological diagnosis, the tumor specimens were immunostained with anti‐ light chain, anti‐ light chain, and anti‐CD138, specific markers for MM [23], and anti‐Ki‐67, a proliferation marker. The tumor specimens were positive for light chain, CD138, and Ki‐67, and pathologically diagnosed as MM (Figure 3A, right). IM‐9 only produces ‐light chain and therefore tumor specimens were negative for ‐light chain. Before starting the xenograft experiments, we confirmed that the gene‐silencing effects of both siLPA1 and siLPA3 lasted for at least 20 days in MSCs (Supporting information Fig. S5). In this xenograft model, the tumor volumes were significantly higher in the group co‐transplanted with siLPA3‐MSCs than in the group cotransplanted with control‐MSCs (Figure 3B, left). Transplantation with only IM‐9 myeloma cells or MSCs did not lead to the formation of any palpable masses in mice. Next, the tumor masses were collected at 20 days after transplantation and weighed. In agreement with the tumor volume measurements, the weight of the tumor masses in the group co‐transplanted with siLPA3MSCs was approximately 2‐fold higher than that in the group co‐transplanted with control‐MSCs (Figure 3B, middle and right). Next, we monitored tumor growth at different time points by co‐transplanting luciferaseexpressing IM‐9 myeloma cells (IM‐9‐luc+) with either siLPA3‐MSCs or control‐MSCs. In agreement with the results in Figure 3B, siLPA3‐MSCs significantly promoted MM progression compared with control‐MSCs (Supporting information Fig. S6). Next, we monitored tumorrelated angiogenesis in the xenograft model. Macroscopically, tumors of mice co‐transplanted with siLPA3MSCs exhibited abundant vascularization (Figure 3C, top). Similar to the macroscopic appearance, the tumor tissues established by co‐transplantation with siLPA3MSCs exhibited profound neovascularization compared with those established by co‐transplantation with control‐MSCs (Figure 3C, bottom and right). These results clearly demonstrate that siLPA3‐MSCs promoted MM progression and tumor‐related angiogenesis compared with control‐MSCs. Thus, LPA3 signaling in MSCs can act as a negative regulator of MM progression.

LPA1 reduction in MSCs delays MM progression

We next evaluated whether siLPA1‐MSCs were MMsuppressive as opposed to siLPA3‐MSCs. As shown in Figure 4A, tumor growth was significantly delayed in the group co‐transplanted with siLPA1‐MSCs compared with the group co‐transplanted with control‐MSCs, in both volume and weight. In agreement with the results, siLPA1‐MSCs significantly delayed MM cell growth compared with control‐MSCs (Supporting information Fig. S7). As last, we monitored tumor‐related angiogenesis. As shown in Figure 4B, tumor tissues established by cotransplantation with siLPA1‐MSCs exhibited significantly poorer angiogenesis than that established by cotransplantation with control‐MSCs. In order to analyze the importance of LPA1 signaling in MSCs for MM progression, we prepared LPA1‐overexpressing MSCs and evaluated their abilities to promote MM progression. As shown in Supporting information Figure S8A and B, LPA1‐overexpressing MSCs (MSCs‐LPA1+) readily acquired senescence‐related phenotypes, such as upregulation of SA‐‐gal activities and morphological changes. And as expected, MSCs‐LPA1+ significantly promoted MM progression compared to control MSCs (MSCs‐Null) at the same extent as siLPA3‐MSCs (Supporting information Fig. S8C). According to these results, we conclude that LPA1 signaling in MSCs can act as a positive regulator of MM progression.

siLPA3‐MSCs readily transdifferentiate into tumor‐associated fibroblasts (TAFs) in response to stimulation from myeloma cells

We hypothesized that the MM‐promoting effect elicited by siLPA3‐MSCs might be attributable to transdifferentiation of siLPA3‐MSCs into tumor‐associated fibroblasts (TAFs). As shown in Figure 5A, compared with controlMSCs, siLPA3‐MSCs strongly expressed ‐SMA, a differentiation marker for TAFs, after stimulation with IM‐9 CM, although siLPA1‐MSCs expressed less ‐SMA than control‐MSCs. The mRNA levels for LPA1–5 did not change in MSCs after stimulation with CM from three myeloma cell lines or LPA (Supporting information Fig. S1B). Next, we attempted to determine the factor released from TAFs that is responsible for MM progression. TAFs are known to secrete several factors that can facilitate MM progression, such as IGF‐1, CXCL12 (SDF1), VEGF, and FGF2 [24‐26]. Compared with controlMSCs, siLPA3‐MSCs did not exhibit any significant upregulation of the mRNAs for IGF‐1 and CXCL12 despite stimulation with IM‐9 CM. VEGF mRNA was upregulated in both control‐MSCs and siLPA3‐MSCs by stimulation with IM‐9 CM. Notably, siLPA3‐MSCs stimulated with IM‐9 CM exhibited approximately 2‐fold higher mRNA expression levels for FGF2 than unstimulated siLPA3MSCs and control‐MSC (Figure 5B). In agreement with the qRT‐PCR results, siLPA3‐MSCs significantly upregulated FGF2 protein levels after stimulation with IM‐9 CM (Supporting information Fig. S9). Next, we tested whether FGF2 could be a factor responsible for siLPA3MSC‐induced MM progression. As shown in Figure 5C, siLPA3‐MSCs significantly promoted tumor growth compared with control‐MSCs in the xenograft model, but the MM‐promoting effect observed in siLPA3‐MSCs was completely eliminated by silencing FGF2 in siLPA3MSCs, in both volume and weight. In agreement with the results, siLPA3‐MSCs promoted MM cell growth compared with control‐MSCs, but the effect was completely eliminated by silencing FGF2 in siLPA3‐MSCs (Supporting information Fig. S10). Finally, we monitored tumor‐related angiogenesis in the mice. As shown in Figure 5D, siLPA3‐MSCs promoted the formation of neovasculature in the tumor tissue, but the effect was completely eliminated by silencing FGF2 in siLPA3MSCs. These results suggest that the signaling through LPA1 and LPA3 determines the senescent state of MSCs. Furthermore, pro‐senescent MSCs promote MM progression and tumor‐related angiogenesis via transdifferentiation into TAFs and FGF2 production and, conversely, anti‐senescent MSCs delay MM progression (Figure 5E).

The other LPA receptors besides LPA1 and LPA3 in MSCs are negligible in MM tumorigenesis

To eliminate possibilities that LPA receptors other than LPA1 and LPA3 are responsible for MM progression, MSCs were silenced with siRNA for LPA2, 4, 5 or 6 and monitored their senescent state and MM‐promoting ability. As shown in Supporting information Figure S11A and B, MSCs silenced with LPA2, 4, 5 or 6 did not show characteristics of senescent cells, such as upregulation of SA‐‐gal activities and morphological changes. Consistent with these observations, MSCs silenced with LPA2, 4, 5 or 6 did not enhance their MM‐promoting abilities (Supporting information Fig. S11C).

MSCs derived from MM patients highly express LPA1

In order to provide clinical and biological evidences to our findings, we compared the expression levels of LPA1 and LPA3 in MSCs derived from MM patients and healthy subjects. MSCs were collected by sorting Lin/CD90+/CD105+ population from mononuclear cells in bone marrow (Supporting information Fig. S12A). Interestingly, MSCs from MM patients expressed higher level of mRNA for LPA1 (150%, P = 0.03) compared to MSCs from healthy subjects (100%; Supporting information Fig. S12B). In addition, MM patients showed higher serum LPA levels than healthy subjects consistent with a previous report (P = 0.005; Supporting information Fig. S13) [20]. These findings strongly suggest that higher expression of LPA1 in MSCs determine tumorigenesis of MM not only in experimental but also in clinical setting.

Systemic administration of LPA1 antagonists exerts same anti‐MM effects as observed in siLPA1‐MSCs

In addition, we confirmed whether systemic administration of LPA1‐specific antagonist could result in the same effect as co‐transplantation with siLPA1‐MSCs using murine xenograft model. Ki16425 is known to preferentially antagonize LPA1 because the Ki values for LPA1 and LPA3 are 0.34 M and 0.93 M, respectively [27]. Tumor growth in the group administered Ki16425 was significantly delayed compared with that in the group treated with the vehicle, in both volume and weight (Figure 6A). In agreement with the results, Ki16425 administration significantly delayed myeloma cell growth compared with vehicle treatment (Supporting information Fig. S14). As expected, Ki16425 administration significantly inhibited angiogenesis in tumor tissues than vehicle treatment (Figure 6B). To generalize the effect of Ki16425 in MM treatment, we also evaluated the anti‐MM effect of Ki16425 in another murine xenograft model using OPM‐2. As shown in Supporting information Fig. S15, systemic administration of Ki16425 was also delayed growth of OPM‐2. Similarly, anti‐MM effect observed by administration of Ki16425 was also obtained by systemic administration of AM095, another LPA1 antagonist (Supporting information Fig. S16). All mice systemically administered Ki16425 and AM095 did not show any obvious abnormalities, such as weight loss, anorexia or diarrhea (data now shown). These results provide the possibility that retarding the LPA1 signaling by systemic administration of Ki16425 would be applicable for MM therapy without any detrimental effects.

DISCUSSION

In this study, we identified a novel mechanism for LPA1/3 signaling with MSCs in establishment of tumor stroma for MM. Recent researches have gradually demonstrated that tumor progression is promoted by bidirectional signaling between tumor cells and MSCs, which is mediated by several adhesion molecules, cytokines and exosomes [7, 8]. However, the mechanisms that allow tumor cells to convert MSCs into tumor stroma are poorly understood. In this study, we identified ATX as a MSC‐derived autocrine factor and LPA1/3 as major LPA receptors which transduce opposite signals in MSCs. Moreover, our report demonstrates that LPA1/3 activation and senescent state in MSCs are important for determining the fate of MSCs toward tumorsupportive or tumor‐suppressive stroma.
First, we showed that MSCs produce ATX in response to myeloma cells via a TLR4‐NF‐B‐dependent pathway. Although the factor produced by myeloma cells remains unidentified, it could be an endogenous ligand for TLR4. MSCs are known to express a number of TLRs and ligation of TLRs such as TLR2, TLR3, and TLR4 leads to tumor growth, angiogenesis, and tumor invasion through production of inflammatory mediators in tumor microenvironment [28]. Previous studies have shown that TLR ligands that link inflammatory events and tumor development are released from tumor cells, including high mobility group box 1 (HMGB1) [29]. Further studies are awaited to determine the factor which is secreted from myeloma cells and stimulates TLR4 in MSCs. It has also been reported that inhibition of NF‐B signaling ameliorates MM in a murine model [30]. In addition, some drugs that inhibit NF‐B signaling, such as dexamethasone and bortezomib, are widely used to treat MM in routine clinical practice [31, 32]. Thus, the effects obtained by these drugs may be attributable to the inhibition of ATX production from MSCs, as well as to anti‐inflammation or inhibition of proteasomes.
LPA signaling provokes MSCs to induce marked changes in their phenotypes. In this study, we observed that silencing of LPA3 in MSCs was sufficient to halt their cell cycle and to induce cellular senescence. In addition, siLPA3‐MSCs acquired properties related to TAFs in response to the stimulation from myeloma cells, thereby promoting MM progression. Considering the mechanism that allows senescent MSCs to transdifferentiate into TAFs, MSCs possess significant plasticity in certain conditions [33, 34]. Previously, it was reported that ‐SMA+ TAFs originate from bone marrow MSCs and exhibit cellular senescence‐related phenotypes after being recruited by inflammatory conditions in gastric cancer [35]. Recently, it was reported that normal quiescent fibroblasts were induced to transdifferentiate into TAFs by the transcription factor TWIST1 in response to IL‐6, an inflammatory cytokine [36]. According to these report, the most plausible explanation is that senescent MSCs are likely to be destined to transdifferentiate into TAFs by taking advantage of stimulation from myeloma cells as the driving force. The precise mechanism that renders senescent MSCs more susceptible to transdifferentiate in TAFs should be further clarified. Intriguingly, it has been reported that MSCs from MM patients exhibit early senescent profiles compared with MSCs from healthy subjects [37]. Our results have clearly proven the relationship between cellular senescence in MSCs and MM progression. Cellular senescence has long been recognized as a phenotype that occurs in vitro, but it cannot be directly linked with the aging that occurs in vivo. However, much evidence has been obtained to correlate cellular senescence in vitro with aging in vivo [22, 38]. It is widely considered that the incidence rates of MM increase proportionally in elderly persons, thus, many investigations have addressed the relationship between MM progression and aging. Our results may give a new paradigm to understand the relationship between MM progression and aging.
During MM progression, several studies have demonstrated that angiogenic factors, such as HGF, FGF2, and VEGF, influence the tumorigenic phenotype. Moreover, the plasma concentrations of these angiogenic factors are significantly elevated in MM patients compared with healthy subjects and strongly correlate with the severity of MM [39‐41]. In the present study, we showed that siLPA3‐MSCs stimulated with myeloma cells increased FGF2 expression, which was not observed in control‐MSCs or untreated siLPA3‐MSCs. Importantly, FGF2 silencing completely inhibited MM progression and tumor‐related angiogenesis when siLPA3MSCs were co‐transplanted in vivo. FGF2 is widely known as a strong angiogenic factor and it also acts as a mitogen in many types of tumor cells by activating the ERK‐MAPK pathway [42]. Therefore, FGF2 seems to be a major factor produced by TAFs that is responsible for the propagation of myeloma cells and for establishing the microenvironment which surrounds myeloma cells in vivo. Furthermore, interrupting the FGF2 signaling pathway could be a promising therapeutic opportunity in MM.
Conversely, our study showed that LPA1 silencing in MSCs promoted cell cycle progression and inhibited cellular senescence and transdifferentiation into TAFs. These effects were in contrast to those observed during LPA3 silencing in MSCs. To the best of our knowledge, this is the first study to demonstrate that LPA1 and LPA3 transduce the opposite signals to MSCs and strongly influence their phenotypic destinations. This study provides valuable insights that help us understand the function and impact of MSCs during MM tumorigenesis in vivo. Previously, we showed that the addition of Ki16425, a chemical agent that preferentially antagonizes LPA1 rather than LPA3, could delay cellular senescence in MSCs in vitro, which is consistent with our finding that LPA1 silencing inhibited the progression of cellular senescence in MSCs in vitro [21]. In spite of the close affinities for LPA1 and LPA3, Ki16425 exhibited almost same effects as observed in siLPA1‐MSCs. This discrepancy might be attributed to different patterns of G protein coupling between LPA1 and LPA3. It was reported that LPA1 associates to G12/13, Gq or Gi, and that LPA3 associates to Gq or Gi [19]. This difference might give diversity to signaling pathway or turnover of these receptors, such as internalization, trafficking, degradation, etc. Further studies are needed, however, to explore this hypothesis. In this study, MSCs derived from MM patients expressed higher levels of LPA1 than healthy subjects, although LPA3 expression was comparable in both groups. Taking the result into account, our data using murine xenograft models suggest that the expression balance between LPA1 and LPA3 in MSCs determines the progression of MM. Although it is still obscure whether LPA1 upregulation in MSCs and serum LPA level correlate with progression of MM, it would be meaningful to investigate the relationship among LPA1 expression level in MSCs, serum LPA level and stages of MM progression.
Furthermore, in the present study, we showed that systemic administration of Ki16425 significantly delayed the progression of MM and tumor‐related angiogenesis in vivo. In a previous study, Ki16425 significantly inhibited liver and lung metastases in a mouse model of breast cancer [43]. In addition, Ki16425 has been reported to be a promising agent for inhibiting bone metastasis by tumor cells, and it is currently being evaluated in preclinical studies. When combined with these results, our findings suggest the intriguing possibility that use of an antagonist antibody of LPA1 or pharmacological modulation of LPA1/3 signaling could yield effective therapies to combat MM. Further studies of LPA signaling in MSCs will provide new insights into the control of tumor‐stroma interactions in MM and other types of tumors.

CONCLUSIONS

In conclusion, our observations demonstrate that LPA1 and LPA3 transduce opposite signals in MSCs to determine the fate of MSCs into MM‐supportive and MMsuppressive stroma, respectively. LPA1/3 signals are sufficient for pro‐senescent and anti‐senescent state in MSCs and senescent MSCs can readily transdifferentiate into TAFs in response to myeloma cells and promote MM progression and angiogenesis. Conversely, antisenescent MSCs are relatively resistant to transdifferentiate into TAFs and delay MM progression and angiogenesis. Consistent with these results, systemic administration of LPA1 antagonist exerts almost same antiMM and anti‐angiogenic effects as observed in antisenescent MSCs. To date, the role of MSCs in tumor development, either supportive or suppressive, remains controversial. This study has deciphered the mechanisms by which MSCs are converted into tumorsupportive or tumor‐suppressive stroma. The universality of our concept that modulation of LPA1/3 axis is effective for regression of other types of tumors awaits further studies using various tumor cells and experimental conditions.

REFERENCES

Gunn, W.G., et al., A crosstalk between myeloma cells and marrow stromal cells stimulates production of DKK1 and interleukin‐6: a
Uccelli, A., L. Moretta, and V. Pistoia, potential role in the development of lytic bone Mesenchymal stem cells in health and disdisease and tumor progression in multiple ease. Nat Rev Immunol, 2008. 8(9): p. 726‐36. myeloma. Stem Cells, 2006. 24(4): p. 986‐91.
Fajardo‐Orduna, G.R., H. Mayani, and J.J. Nakajima, S., et al., Induction of thymic Montesinos, Hematopoietic Support Capacity stromal lymphopoietin in mesenchymal stem of Mesenchymal Stem Cells: Biology and Clinicells by interaction with myeloma cells. Leuk cal Potential. Arch Med Res, 2015. Lymphoma, 2014. 55(11): p. 2605‐13.
Yang, X., et al., One cell, multiple roles: Aikawa, S., et al., Lysophosphatidic acid contribution of mesenchymal stem cells to as a lipid mediator with multiple biological tumor development in tumor microenvironactions. J Biochem, 2015. 157(2): p. 81‐9. ment. Cell Biosci, 2013. 3(1): p. 5.
Yung, Y.C., N.C. Stoddard, and J. Chun, Karnoub, A.E., et al., Mesenchymal stem LPA receptor signaling: pharmacology, physicells within tumour stroma promote breast ology, and pathophysiology. J Lipid Res, 2014. 55(7): p. 1192‐1214.
Sasagawa, T., et al., Abnormal serum Rhee, K.J., J.I. Lee, and Y.W. Eom, Mesenlysophospholipids in multiple myeloma pachymal Stem Cell‐Mediated Effects of Tumor tients. Lipids, 1999. 34(1): p. 17‐21. Kanehira, M., et al., Targeting lysophos‐16(12): p. 30015‐33.
Klopp, A.H., et al., Concise review: Dissectassociated senescence of human marrow ing a discrepancy in the literature: do mesenstromal cells. PLoS One, 2012. 7(2): p. chymal stem cells support or suppress tumor e32185. growth? Stem Cells, 2011. 29(1): p. 11‐9. 22 Ohtani, N. and E. Hara, Roles and mech‐
Shain, K.H., W.S. Dalton, and J. Tao, The anisms of cellular senescence in regulation of tumor microenvironment shapes hallmarks of tissue homeostasis. Cancer Sci, 2013. 104(5): mature B‐cell malignancies. Oncogene, 2015. p. 525‐30. 34(36): p. 4673‐82.
Santra, M., J.D. Shaughnessy, Jr., and 8 Attar‐Schneider, O., et al., Multiple mye‐W.T. Bellamy, Expression of multiple myeloma loma and bone marrow mesenchymal stem associated markers in bone marrow spicules cells' crosstalk: Effect on translation initiation. using a novel immunohistochemical tech‐Mol Carcinog, 2015. nique. Biotech Histochem, 2011. 86(2): p. 9 Dimopoulos, M.A., et al., Current treat‐119‐23. ment landscape for relapsed and/or refracto‐
Vanderkerken, K., et al., Insulin‐like ry multiple myeloma. Nat Rev Clin Oncol, growth factor‐1 acts as a chemoattractant 2015. 12(1): p. 42‐54. factor for 5T2 multiple myeloma cells. Blood,
Ayed, A.O., L.J. Chang, and J.S. Moreb, 1999. 93(1): p. 235‐41. Immunotherapy for multiple myeloma: Cur‐Ria, R., et al., Bone marrow angiogenesis rent status and future directions. Crit Rev and progression in multiple myeloma. Am J Oncol Hematol, 2015. Blood Res, 2011. 1(1): p. 76‐89. Rollig, C., S. Knop, and M. Bornhauser,
Azab, A.K., et al., CXCR4 inhibitor Multiple myeloma. Lancet, 2015. 385(9983): AMD3100 disrupts the interaction of multiple p. 2197‐208. myeloma cells with the bone marrow micro‐Kuehl, W.M. and P.L. Bergsagel, Molecuenvironment and enhances their sensitivity to lar pathogenesis of multiple myeloma and its therapy. Blood, 2009. 113(18): p. 4341‐51. premalignant precursor. J Clin Invest, 2012.
Ohta, H., et al., Ki16425, a subtype‐122(10): p. 3456‐63. selective antagonist for EDG‐family lysophosReagan, M.R. and I.M. Ghobrial, Multiple phatidic acid receptors. Mol Pharmacol, 2003. myeloma mesenchymal stem cells: characteri‐64(4): p. 994‐1005. zation, origin, and tumor‐promoting effects.
Sun, Z., S. Wang, and R.C. Zhao, The Clin Cancer Res, 2012. 18(2): p. 342‐9. roles of mesenchymal stem cells in tumor
Roccaro, A.M., et al., BM mesenchymal inflammatory microenvironment. J Hematol stromal cell‐derived exosomes facilitate mulOncol, 2014. 7: p. 14. tiple myeloma progression. J Clin Invest, 2013.
Apetoh, L., et al., The interaction be‐123(4): p. 1542‐55. tween HMGB1 and TLR4 dictates the outcome
Wang, J., et al., Bone marrow stromal of anticancer chemotherapy and radiothera‐cell‐derived exosomes as communicators in py. Immunol Rev, 2007. 220: p. 47‐59. drug resistance in multiple myeloma cells. Blood, 2014. 124(4): p. 555‐66.
Tatetsu, H., et al., Dehydroxymethylepoxyquinomicin, a novel nuclear factorkappaB inhibitor, induces apoptosis in multiple myeloma cells in an IkappaBalphaindependent manner. Mol Cancer Ther, 2005. 4(7): p. 1114‐20.
Mitsiades, N., et al., The proteasome inhibitor PS‐341 potentiates sensitivity of multiple myeloma cells to conventional chemotherapeutic agents: therapeutic applications. Blood, 2003. 101(6): p. 2377‐80.
Salem, K., et al., Combination chemotherapy increases ONO-7300243 cytotoxicity of multiple myeloma cells by modification of nuclear factor (NF)‐kappaB activity. Exp Hematol, 2013. 41(2): p. 209‐18.
Xouri, G. and S. Christian, Origin and function of tumor stroma fibroblasts. Semin Cell Dev Biol, 2010. 21(1): p. 40‐6.
Baulida, J. and A. Garcia de Herreros, Snail1‐driven plasticity of epithelial and mesenchymal cells sustains cancer malignancy. Biochim Biophys Acta, 2015. 1856(1): p. 55‐61.
Quante, M., et al., Bone marrow‐derived myofibroblasts contribute to the mesenchymal stem cell niche and promote tumor growth. Cancer Cell, 2011. 19(2): p. 257‐72.
Lee, K.W., et al., Twist1 is a key regulator of cancer‐associated fibroblasts. Cancer Res, 2015. 75(1): p. 73‐85.
Andre, T., et al., Evidences of early senescence in multiple myeloma bone marrow mesenchymal stromal cells. PLoS One, 2013. 8(3): p. e59756.
Ohtani, N., D.J. Mann, and E. Hara, Cellular senescence: its role in tumor suppression and aging. Cancer Sci, 2009. 100(5): p. 792‐7.