Heparanase influences expression and shedding of syndecan-1, and its expression by the bone marrow environment is a bad prognostic factor in multiple myeloma

Multiple myeloma (MM) is a B-cell neoplasia characterized by the growth of a clone of malignant plasma cells (PCs) in the bone marrow. Syndecan-1, a transmembrane heparan sulfate proteoglycan (HSPG), is expressed by both normal and malignant PCs and is now widely used to identify and purify these cells.^1,2  We have previously shown that among the 10 known HSPGs, syndecan-1 is the main one present on the surface of PCs.³  Syndecan-1 binds to extracellular matrix proteins and to growth factors that have a HS-binding domain, thus promoting their activity.⁴  In particular, it has been shown that syndecan-1 mediates hepatocyte growth factor (HGF) binding and promotes c-met signaling in myeloma cells (MMCs).⁵  We and others have shown that syndecan-1 is necessary for the MMC growth activity of HGF and epidermal growth factor (EGF) family ligands that have a HS-binding domain.^3,5–7  Syndecan-1 concentrates high amounts of growth factors at the cell surface and then likely facilitates the activation of their receptors. The ectodomain of syndecan-1 can be cleaved from the MMC surface into a soluble form by metalloproteinases that remain to be identified in MMCs.^8,9  Soluble syndecan-1 is present at high levels in the serum of patients with MM and is an indicator of poor prognosis.^8,10,11  It also accumulates in the bone marrow microenvironment where it might sequester heparin-binding growth factors in the proximity of the tumor.¹²  Accordingly, it has been shown that soluble syndecan-1, which retains biologically active HS chains, can promote tumor growth in a myeloma severe combined immunodeficiency–hu mouse model.¹³ 

The function of HSPGs is regulated by extracellular enzymes that modulate the structure of HS chains. One such enzyme is heparanase (HPSE), encoded by only one gene in mammalian cells.^14,15  HPSE is an endoglucuronidase that cleaves HS chains at only a few sites, resulting in HS fragments of 10 to 20 sugar units long.^14,16  These fragments are long enough to bind growth factors and can be more biologically active than the native HS chains.¹⁷ HPSE is overexpressed in numerous tumors where it increases the angiogenic and metastatic potential of tumor cells.¹⁸  In patients with MM, HPSE activity could be found in bone marrow aspirates.¹⁹  HPSE increases microvessel density¹⁹  and promotes growth and metastasis of MMCs to bone in vivo.²⁰ 

Here, we show for the first time that HPSE is involved in the regulation of syndecan-1 gene expression and controls the production of soluble syndecan-1 in myeloma cell lines. In patients with newly diagnosed MM, HPSE is expressed mainly by cell fractions of the bone marrow microenvironment, especially monocytes and osteoclasts, but weakly by MMCs. High HPSE expression in bone marrow of patients with MM correlates with a shorter event-free survival (EFS) and overall survival (OAS).

Whole bone marrow (WBM) samples and purified MMCs were obtained from 39 patients with MM at diagnosis (presenting at the University Hospitals of Heidelberg, Germany, and Montpellier, France) after written informed consent was given in accordance with the Declaration of Helsinki. These patients were treated with high-dose chemotherapy (HDC) and autologous stem cell transplantation (ASCT). The study was approved by the ethics boards of both universities. According to the Durie-Salmon classification, 8 patients were in stage IA, 6 in stage IIA, 22 in stage IIIA, and 3 in stage IIIB. Four patients had IgAκ MM, 3 IgAλ MM, 19 IgGκ MM, 7 IgGλ MM, 4 Bence-Jones κ MM, 1 Bence-Jones λ MM, and 1 nonsecreting MM. MMCs were purified with anti-CD138 magnetic cell sorting microbeads (Miltenyi-Biotec, Paris, France). Normal WBM samples were obtained from healthy donors after informed consent was given. For 7 additional patients with newly diagnosed MM, BM environment (ENV) cells were obtained by removing myeloma cells with CD138 Miltenyi microbeads (< 1% plasma cells). For 5 other patients with newly diagnosed MM, BM T cells, monocytes, and polymorphonuclear cells were purified as described previously.²¹  Osteoclasts and BM stromal cells (BMSCs) were generated in vitro as described.²¹  Immature dendritic cells (DCs) were generated from leukapheresis products of patients with MM, as previously described.²²  Briefly, 8 × 10⁶ G-CSF–mobilized leukapheresis cells were plated in 2 mL X-VIVO15 medium (BioWittaker, Walkersville, MD) per well in 6-well flat-bottomed plates (Nunc, Roskilde, Denmark). Nonadherent cells were discarded by gentle rinsing after a 2-hour incubation at 37°C in 5% CO₂. Adherent cells were cultured in X-VIVO15 medium with 2% human albumin, 100 ng/mL GM-CSF (LEUKINE; Berlex, Montville, NJ) and 25 ng/mL IL-4 (Cellgenix, Freiburg, Germany) for 5 days. Mature DCs were obtained by a further 24-hour maturation with GM-CSF, IL-4, and TNFα. Human IL-6–dependant XG myeloma cell lines (HMCLs) were obtained in our laboratory.^23,24  They were routinely maintained in RPMI1640, 10% fetal calf serum, and 2 ng/mL IL-6 (Abcys, Paris, France). U266, SKMM, OPM2, LP1, and RPMI8226 HMCLs were purchased from ATTC (Rockville, MD).

RNA was extracted with the RNeasy Kit (Qiagen, Hilden, Germany), the SV-total RNA extraction kit (Promega, Mannheim, Germany), and Trizol (Invitrogen, Karlsruhe, Germany) in accordance with the manufacturer's instructions. Biotinylated complementary RNA (cRNA) was amplified with double in vitro transcription, according to the Affymetrix small sample labeling protocol (Affymetrix, Santa Clara, CA). The biotinylated cRNA was fragmented and hybridized to the human genome U133 set (for HMCLs) or U133 Plus 2.0 (for patient samples) microarrays according to the manufacturer's instructions (Affymetrix). Fluorescence intensities were quantified and analyzed using the GCOS software (Affymetrix).

RNA was extracted using the RNeasy Kit (Qiagen). We generated cDNA from 100 ng total RNA using Superscript II reverse transcriptase (Invitrogen, Cergy-Pontoise, France). For real-time reverse transcriptase–polymerase chain reaction (RT-PCR), we used Assay-on-Demand primers and probes and the TaqMan Universal Master Mix from Applied Biosystems (Courtaboeuf, France) according to the manufacturer's instructions. Gene expression was measured using the ABI Prism 7000 Sequence Detection System. For comparative analysis of gene expression, data were obtained by using the ΔΔCt method derived from a mathematical approach previously described. For each sample, the C_T value for the gene of interest was determined and normalized to its respective C_T value for β2-microglobulin (B2M) (ΔC_T = C_T − C_TB2M) or GAPDH (ΔC_T = C_T − C_TGAPDH) and compared with a cell type used as a positive control. The formula used was 1/2^{ΔCT sample − ΔCT control cell line}. Ct values were collected during the log phase of the cycle. The results were expressed as the relative mRNA levels to control cell sample mRNA.

Cells were lysed in 10 mM Tris-HCl (pH 7.05), 50 mM NaCl, 50 mM NaF, 30 mM NaPPi, 1% Triton X-100, 5 μM ZnCl₂, 100 μM Na₃VO4, 1 mM DTT, 20 mM β-glycerophosphate, 20 mM PNPP, 20 μg/mL aprotinin, 2.5 μg/mL leupeptin, 0.5 mM PMSF, 0.5 mM benzamidine, 5 μg/mL pepstatin, and 50 nM okadaic acid. Lysates were resolved on 12% sodium dodecyl sulfate–polyacrylamide by gel electrophoresis (SDS-PAGE) and transferred to a nitrocellulose membrane (Schleicher and Schuell, Kassel, Germany). Membranes were blocked for 2 hours at room temperature in 140 mM NaCl, 3 mM KCl, 25 mM Tris-HCl (pH 7.4), 0.1% TBS-T, 5% nonfat milk, and then immunoblotted with a rabbit polyclonal antiheparanase antibody (InSight Biopharmaceuticals, Rehovot, Israel) at a 1:1000 dilution. As a control for protein loading, we used a mouse monoclonal anti–β-actin antibody (Sigma, St Louis, MO). The primary antibodies were visualized with goat anti–rabbit (Sigma) or goat anti–mouse (Bio-Rad, Hercules, CA) peroxidase-conjugated antibodies by an enhanced chemiluminescence detection system.

HPSE activity in HMCLs was measured with a commercial assay based on the measurement of HPSE-induced degradation of biotinylated-HS (b-HS) fragments (Takara Bio, Kyoto, Japan). Briefly, 10 × 10⁶ cells were collected and suspended in 1 mL extraction buffer. After cell debris removal by centrifugation, the supernatant was incubated with biotinylated-HS (b-HS) for 45 minutes at 37°C. The mixture was then incubated on a FGF-coated plate and undegraded b-HS was detected by HRP-streptavidin. This b-HS fragment is specifically designed so that it cannot bind to FGF when being degraded by HPSE, thus resulting in a lower HRP signal. Correspondence between absorbance at 450 nm and HPSE activity was determined by comparison to a standard curve established using unlabeled heparan sulfate as a standard substitute.

MMCs were transfected with 1 nM SMARTpool HPSE siRNA or 1 nM nontargeting siRNA (siGLO) (Dharmacon, Lafayette, CO). For transfections, Amaxa nucleofection technology was used (Amaxa, Cologne, Germany). U266 myeloma cells were resuspended in the nucleofector T solution, available as part of the Amaxa cell optimization kit, and were electroporated using the T-001 protocol. Briefly, 4 × 10⁶ cells were divided into aliquots of 100 μL with 15 μg siRNA and were transferred to a cuvette and nucleofected with the Amaxa Nucleofector device. Cells were immediately transferred into wells containing 37°C prewarmed culture medium in 24-well plates. All experiments were performed in duplicate. At days 1 to 2 after electroporation, (1) the gene expression level of HPSE, syndecan-1, B2M, and GAPDH were analyzed by real-time PCR; (2) membrane syndecan-1 expression levels at the cell surface were analyzed with a FACS-Calibur fluorescence-activated cell sorter (Becton Dickinson, Mountain View, CA), using PE-conjugated anti-CD138 (Beckman Coulter, Marseilles, France); (3) cells were counted and viability was assessed using annexin V–FITC (Boehringer, Mannheim, Germany); (4) soluble syndecan-1 was measured by enzyme-linked immunoabsorbent assay (ELISA) from cell line culture supernatants.

HMCLs in the exponential growth phase were cultured for 3 days. The cell count, the density of membrane syndecan-1, and the concentration of soluble syndecan-1 were measured every day for 3 days. Soluble syndecan-1 was measured using an ELISA kit with a detection level of 8 ng/mL (Diaclone, Besançon, France).

HPSE expression was assessed by immunohistochemistry on bone marrow biopsies from 20 patients with newly diagnosed myeloma, after written informed consent was given. Biopsies were fixed in Bouin fixative, decalcified, and routinely processed for paraffin embedding. Sections of Bouin-fixed, paraffin-embedded HMCLs (XG-2 and XG-7) were used as positive and negative controls, respectively. Tissue sections were deparaffined in xylene and rehydrated through graded concentrations of ethanol and distilled water. Epitope retrieval was performed by incubating the slides for 40 minutes in 10 mM EDTA buffer solution (Euromedex, Souffelweyersheim, France), pH 7.2, at 99.8°C, in a double boiler. The slides were then incubated for 30 minutes in a blocking solution containing 10% goat serum (Dako Cytomation S.A.S., Trappes, France). Slides were washed, and endogenous peroxidase activity was quenched with 3% hydrogen peroxide in methanol followed by 15-minute incubation in CAS Block (Zymed Laboratories, San Francisco, CA). The slides were stained with a rabbit polyclonal anti-HPSE antibody (InSight Biopharmaceutical) at a concentration of 4.5 μg/mL. All stainings were performed with the NEXES Ventana Medical Systems automaton using the iVIEW DAB Detection Kit (Ventana Medical Systems, Illkirch, France) with goat antirabbit secondary Abs and diaminobenzidine as chromogen, according to the manufacturer's instructions. The immunohistochemical reaction was counterstained with hematoxylin. Heparanase staining was determined by 2 pathologists (P.R. and Thérèse Rousset, CHU, Montpellier, France). Each section stained with anti-HPSE was compared with an adjacent section stained with irrelevant rabbit serum as a negative control.

Images were visualized using an Olympus BX-51 microscope (Olympus, Tokyo, Japan) equipped with one of the following: a UplanApo 40×/0.85 NA air objective (Figure 6, except panel Bvi) or a UplanApo 10×/0.40 NA air objective (Figure 6Bvi). All images were taken at room temperature using an Olympus CMA-D3 camera (Tokyo, Japan) and were processed using Adobe Photoshop (Adobe Systems, San Jose, CA).

Gene Expression Profiles were analyzed with our bioinformatics platform (RAGE, remote analysis of microarray gene expression; http://rage.montp.inserm.fr) designed by T. Reme (INSERM U475, Montpellier, France) or SAM (Significance Analysis of Microarrays) software. SAM analysis was applied to 22 530 probesets with at least 3 presences in the 19 HMCLs samples and a variation coefficient of 20 or greater. We performed a 2-class comparison between HPSE^pos and HPSE^neg HMCLs, with 1000 permutations and a false discovery rate of 7.7%. Statistical comparisons were made with the nonparametric Mann-Whitney test or the Student t test for pairs. The statistical significance of differences in event-free survival between groups of patients was estimated by the log-rank test. An event was defined as relapse or death (for EFS) or as death (for OAS). The survival curves were plotted using the Kaplan-Meier method.

Expression of HPSE was evaluated with Affymetrix U133 set microarrays on 19 HMCLs. Eleven of 19 HMCLs displayed a “present” call for the HPSE probeset (HPSE^pos) with a median expression level of 90 (range, 48-448; Figure 1A) The other 8 cell lines displayed an “absent” call (HPSE^neg) with a low expression level (median, 6; range, 2-28) (Figure 1A). The call (present or absent) is determined by Affymetrix GCOS software and indicates whether a gene is reliably expressed.²⁵  The data were validated by real-time RT-PCR (r = 0.87, P < .01; Figure 1B). HPSE expression at the protein level was evaluated by Western blot analysis. HPSE was detected in 3 of 3 HMCLs with a present Affymetrix detection call (U266, XG-2, and XG-20) but not in those with an absent Affymetrix call (XG-1 and XG-7; Figure 1C). The 50-kDa MW active form of the enzyme was predominant, but the 65-kDa MW proenzyme could also be detected (Figure 1C). Furthermore, HMCLs were tested for HPSE activity using an assay based on the measurement of HPSE-induced degradation of biotinylated-HS fragments (b-HS). In this assay, undegraded b-HS is bound to an FGF-coated plate, and bound b-HS is detected by HRP-streptavidin. The b-HS fragment is specifically designed so that it does not bind to FGF when degraded by HPSE, thus resulting in a lower HRP signal. HPSE activity was high in the HPSE^pos XG-2 and XG-20 HMCLs (0.6 U/mL and 0.82 U/mL, respectively). It was intermediate in U266 (0.14 U/mL) and below the detection limit (0.05 U/mL) in the HPSE^neg XG-1 and XG-7 HMCLs (Figure 1D). Thus, HPSE activity correlated with HPSE expression in HMCLs.

Because the HPSE signal expression in HMCLs is either present or absent, these HMCLs are of interest to study the biologic function of HPSE in MM. We compared the gene expression profile of the 11 HPSE^pos HMCLs with those 8 HPSE^neg HMCLs in a supervised analysis performed with the SAM software using 1000 permutations. Only probesets that were present in more than 3 samples and had at least a 2-fold change in expression were retained in the analysis. This resulted in a list of 41 probesets that were all overexpressed in HPSE^pos HMCLs, with a false discovery rate (FDR) of 7.7%. These 41 probesets interrogated 33 genes and 3 expressed sequence tags (ESTs), which are listed in Table 1Interestingly, syndecan-1 was among those 41 genes.

As shown in Figure 2A syndecan-1 gene was significantly overexpressed 4.3-fold in HPSE^pos HMCLs (median = 1320) compared with HPSE^neg HMCLs (median = 304). Because syndecan-1 is expressed as a membrane protein that is cleaved by metalloproteinases, we quantified membrane expression and soluble syndecan-1 production in HMCLs (Figure 2B-2C). Using FACS analysis, no significant difference in membrane syndecan-1 expression was found between HPSE^pos HMCLs (median MFI = 656) and HPSE^neg HMCLs (median MFI = 799) (Figure 2B). The rate of soluble syndecan-1 production in culture supernatant of HMCLs was measured during the exponential growth phase, between day 2 and day 3 of culture. Indeed, soluble syndecan-1 accumulated in culture medium in parallel with cell count increase, as shown for XG-2 and U266 in Figure 2D. Soluble syndecan-1 production was below the detection level (< 8 ng/mL) in 6 of 8 HPSE^neg HMCLs, whereas it was detected in 10 of 11 HPSE^pos HMCLs with a median production of 35 ng/10⁶cells/24 hours (P = .02; Figure 2C). The rate of soluble syndecan-1 production did not correlate with the density of membrane syndecan-1 in the 19 HMCLs (data not shown). This data suggested that HPSE may influence syndecan-1 gene expression and/or the shedding of soluble syndecan-1.

To address this question, we used siRNA technology. Electroporation of the U266 HMCL with a HPSE-specific siRNA resulted in a 87% inhibition of HPSE expression as determined by real-time RT-PCR in 5 independent experiments (P < .01), whereas electroporation with a nontargeting control siRNA (siRNA-co) did not affect the mRNA level of HPSE (Figure 3A, first panel). Of major interest, HPSE down-regulation induced a 38% reduction of syndecan-1 gene mRNA expression (P < .01), without affecting GAPDH mRNA expression (Figure 3A, second and third panels). The HPSE nontargeting siRNA-co did not affect syndecan-1 or GAPDH gene expression. We then investigated whether targeting HPSE mRNA may influence membrane syndecan-1 expression or soluble syndecan-1 production. Electroporation of HPSE-specific siRNA had no effect on the expression of membrane syndecan-1 (Figure 3B). In contrast, HPSE down-regulation resulted in nearly 80% inhibition (P < .01) of soluble syndecan-1 production (Figure 3C). No inhibition was found with the control siRNA (Figure 3C). Of note, the siRNA-HPSE did not significantly affect the survival and proliferation of U266 cells (data not shown). Those data indicate that targeting HPSE mRNA results in a decrease of syndecan-1 gene expression and soluble syndecan-1 production without effecting the expression of membrane syndecan-1.

To determine whether HPSE is expressed in the bone marrow of patients with MM, we evaluated the gene expression level of HPSE in the whole bone marrow (WBM) (including MMCs and cells of the microenvironment) of 39 patients with newly diagnosed MM as well as in the MMCs purified from the BM of the same 39 patients, using U133 Plus 2.0 Affymetrix microarrays. A high HPSE expression was detected in the WBM samples (Figure 4A). The HPSE expression in the WBM samples (median, 76; range, 25-221) was 7.6-fold higher (P < .001) than that found in the corresponding MMCs (median, 10; range, 8-26). Furthermore, HPSE had an Affymetrix present call in 92% (36 of 39) of the WBM samples, whereas it had a present call in only 10% (4 of 39) of the purified MMC samples (Figure 4A). HPSE was also expressed in the WBM of 7 healthy donors (median, 84; range, 70-110; data not shown). To confirm those data, HPSE expression was analyzed by real-time RT-PCR in the WBM of 7 additional patients with newly diagnosed MM, as well as in the corresponding purified MMCs and the corresponding microenvironment cells depleted from MMCs (< % MMCs, noted ENV; Figure 4B). For each patient, the WBM sample was used as a reference and was assigned the arbitrary value of 100. For 4 of 7 patients (P1-P4), HPSE was almost exclusively expressed by cells from the environment (ENV). For the other 3 patients (P5-P7), HPSE was expressed by both MMCs and cells from the environment (Figure 4B). Because HPSE was much more frequently expressed in HMCLs derived from patients with extramedullary myeloma (58% of presence; see Figure 1A) compared with BM primary MMCs, we looked for HPSE expression in purified MMCs from 5 patients with plasma cell leukemia (PCL1-5). HPSE expression was in the same range than that found in purified MMCs from patients with intramedullary myeloma (Figure 4B). Environment cells depleted from MMCs (ENV) and purified MMCs were also analyzed by Western blot. HPSE protein was detected in 5 of 5 environment cell samples. It was barely detectable in 3 of 3 purified MMC samples, but its expression was much lower than that found in the environment cells (Figure 4C), thus confirming the Affymetrix and real-time RT-PCR data.

To identify the cell populations expressing HPSE in the myeloma microenvironment, bone marrow T lymphocytes (CD3⁺), monocytes (CD14⁺), polymorphonuclear cells (CD15⁺), and MMCs (CD138⁺) were purified from the BM of 5 patients with newly diagnosed myeloma. In addition, BM stromal cells (BMSCs, n = 5) and osteoclasts (n = 7) were generated from cells of patients with MM in vitro. For each population, the gene expression level of HPSE was assayed using Affymetrix U133 plus 2.0 microarrays. As shown in Figure 5A, HPSE was mainly expressed by CD14⁺ monocytes and osteoclasts. HPSE expression in monocytes (median, 819) and osteoclasts (median, 355) was, respectively, 200-fold and 88-fold higher than that in MMCs (median, 4). HPSE was not expressed by BMSCs. All data were validated by real-time RT-PCR (Figure 5B). Of note, no HPSE expression could be detected in monocyte-derived immature or mature dendritic cells using real-time RT-PCR (data not shown). Thus, HPSE is highly expressed by the BM environment, mainly by monocytes and osteoclasts, and occasionally by primary MMCs.

To further clarify the contribution of MMCs in the HPSE expression, bone marrow biopsies from 20 patients with MM were stained using the polyclonal anti-HPSE antibody. The anti-HPSE antibody efficiently labeled the XG-2 HMCL which has a present Affymetrix detection call (Figure 6Ai-ii), unlike the XG-7 HMCL which has an absent call (Figure 6Aiii). As expected from the Affymetrix data, HPSE was expressed by polymorphonuclear cells and myeloid cells within the BM of 20 of 20 patients (as shown for one patient in Figure 6Biv-v). In addition, a strong staining of megakaryocytes was found (Figure 6Bvi). Regarding MMCs, HPSE expression was heterogeneous among patients. Data are summarized in Table 2. HPSE expression lacked in the MMCs of 3 of 20 patients (as illustrated for one patient in Figure 6Bvii). In the majority of the patients, a strong heterogeneity was found with few positively stained MMCs close to negative tumor cells (Figure 6Bviii-ix; Table 2). This heterogeneous HPSE expression among MMCs is in agreement with previously published data.¹⁹  HPSE expression level in the positively stained MMCs was low. In 2 of 20 samples, we found a higher percentage of more intensely labeled MMCs (Figure 6Bx; Table 2). Of note, a strong nuclear staining appeared in the MMCs from those 2 patients.

Taken together, those data indicate that HPSE is mainly expressed by cells of the BM environment of patients with MM.

We investigated the connection of clinical parameters with HPSE gene expression for 30 of the 39 patients with newly diagnosed MM who were treated with HDC and ASCT. Patients were classified into 2 groups: 15 HPSE^low patients with the lowest HPSE expression in the WBM (median expression, 60) and 15 HPSE^high patients with the highest HPSE expression (median expression, 125) (Figure 7A, left). The HPSE^high group had an increased percentage of patients with elevated C-reactive protein and β2-microglobulin (P ≤ .05; Table 3). Other clinical parameters (lactate dehydrogenase, serum calcium, albumin, hemoglobin, age, presence of bone lesions, and cytogenetic abnormalities) were not significantly different. According to the International Staging System (ISS), the HPSE^high group included a significantly higher frequency of patients with stage III MM and a lower frequency of patients with stage I MM. The stages according to the Durie-Salmon classification were not significantly different between the 2 groups (Table 3). HPSE^high patients had a shorter EFS (50% of survival, 6 months; P = .017) than HPSE^low patients (50% of survival not reached) (Figure 7A). The OAS was also significantly shorter in HPSE^high patients (P = .023). In a Cox proportional-hazard model monitoring for the highest or lowest HPSE expression in the WBM as well as ISS stage, both parameters were significantly predictive for EFS (P = .029 and P = .002, respectively). If HPSE expression was tested together with ISS staging, ISS staging remained an independent prognostic factor (P = .02) unlike HPSE expression. Classifying these 30 patients into 2 groups according to HPSE expression in MMCs did not result in any significant differences in the EFS or OAS between the 2 groups (Figure 7B). This is in agreement with the low HPSE expression in MMCs compared with that found in the microenvironment. Those data indicate that HPSE expression in the WBM, but not in the MMCs, is linked to the prognosis of patients with myeloma.

Given the major role of the syndecan-1 HSPG in MM, mainly as a coreceptor for numerous MMC growth factors, we investigated the expression and possible function of HPSE in myeloma. We found that HPSE is expressed in 11 of 19 HMCLs and influences syndecan-1gene mRNA expression: (1) when comparing the gene expression profile of the 11 HPSE^pos HMCLs to that of the 8 HPSE^neg HMCLs, syndecan-1 was one of the genes overexpressed with the highest fold change in HPSE^pos HMCLs; (2) the silencing of HPSE expression with siRNAs induced a down-regulation of syndecan-1 gene mRNA. We used the siRNA method because neutralizing antibodies to HPSE are not presently available commercially. Interestingly, although HPSE expression does not interfere with membrane syndecan-1 expression, it is strongly correlated with the rate of soluble syndecan-1 production by HMCLs. Seven of 8 HPSE^neg HMCLs failed to produce detectable soluble syndecan-1, whereas 10 of 11 of HPSE^pos HMCLs produced large amounts. In addition, targeting HPSE mRNA with siRNA induced a strong inhibition of soluble syndecan-1 production. One possible mechanism to explain the effect of HPSE on the shedding of syndecan-1 could be that degradation of HS chains by the HPSE enzyme would render membrane syndecan-1 more easily accessible to metalloproteinases. This hypothesis is supported by the observation that treatment of MM cells with the bacterial enzyme heparitinase, which completely degrades HS chains, also induced an increase in soluble syndecan-1 production by HMCLs (data not shown). Of note, membrane syndecan-1 expression in the 19 HMCLs did not correlate with the rate of soluble syndecan-1 production, and the HPSE inhibition by siRNA did not affect membrane syndecan-1 expression. One explanation is that the down-regulation of HPSE yields a decreased mRNA expression and concomitantly reduces shedding of syndecan-1 protein, which taken together does not influence the net amount of syndecan-1 on the surface of HMCLs.

HPSE expression has been documented in several solid tumors. In particular, by real-time RT-PCR, increased levels of HPSE mRNA have been found in human breast, colon, lung, prostatic, and ovarian tumors, compared with corresponding normal tissues.¹⁶  To demonstrate the biologic significance of HPSE in multiple myeloma, we analyzed HPSE expression in the bone marrow of patients with MM. Here, we show that HPSE is expressed in the BM of patients with MM, both at the mRNA and protein levels, and that cells of the environment are the main source of HPSE. Among the different cell components of the BM, the highest HPSE expression was found in monocytes. We detected no HPSE expression in monocyte-derived immature or mature dendritic cells that were generated in vitro. Our findings are in agreement with recently published data showing that HPSE protein is present both in monocytes and dendritic cells unlike HPSE mRNA, which is exclusively expressed in monocytes. This indicates that HPSE gene expression and synthesis mainly occur in monocytes or in the very early steps of differentiation toward immature dendritic cells.²⁶ HPSE was also expressed in T cells and polymorphonuclear cells at a lower level. Because MMCs survive and proliferate in a tumor niche involving BMSCs and osteoclasts which cannot be harvested in vivo, we produced them in vitro and found that osteoclasts highly expressed HPSE unlike BMSCs. It is likely that the MMCs themselves can also be a source of HPSE, at least in some patients. However, overall, HPSE expression in MMCs was much lower than that found in the environment cells. First, the HPSE Affymetrix signal was 7.6-fold lower in MMCs compared with WBM samples, and it was reliably detectable in tumor cells of only 10% of the patients (4 of 39). Second, HPSE protein expression measured by Western blot was barely detectable in MMCs compared with environment cells. Third, using immunohistochemistry, we found that HPSE was expressed by the environment cells of 20 of 20 patients, whereas most MMCs were negative or expressed low levels of HPSE. In the majority of the patients, few positively stained MMCs were surrounded by HPSE^neg MMCs. This heterogeneity, which is in agreement with the previous report by Kelly et al,¹⁹  might explain why we found a low HPSE expression in the majority of the purified MMC samples compared with the BM environment samples. Note that in 2 of 20 patients (patients 19 and 20) we found groups of intensely labeled myeloma cells. This is in agreement with the Affymetrix data showing that 10% of the patients expressed reliable amounts of HPSE mRNA in the MMCs. Contrary to primary MMCs, HPSE expression was detectable in 58% (11 of 19) of the HMCLs. We already pointed out that HMCLs may express some genes (HB-EGF, IL-6, and APRIL) that are originally expressed by the BM environment but not by primary MMCs themselves.^6,27  Of note, HPSE expression in MMCs from patients with plasma cell leukemia was in the same range than that found in MMCs from patients with intramedullary myeloma.

A major finding of this study is that HPSE expression in the WBM of patients with MM is an indicator of poor prognosis. Thus, HPSE mRNA expression in the WBM holds prognostic value even though its median expression in patients with MM did not differ from that of healthy donors. In agreement with this observation, it was previously reported that healthy persons and MM patients have similar amounts of IGF-1 in the circulation, whereas MM patients with low serum IGF-1 levels have a favorable prognosis.²⁸  Despite the low number of patients whose HPSE expression and clinical data were available and the short follow-up (median, 2 years), the difference in EFS and OAS between HPSE^high and HPSE^low patients was highly significant with HPSE expression predicting a less favorable disease outcome. HPSE^high patients had higher CRP and β2-microglobulin levels than did HPSE^low patients. HPSE expression was also linked to ISS staging, the HPSE^low group having a higher percentage of stage 1 and a lower percentage of stage 3 patients than the HPSE^high group. The prognostic value of HPSE expression in the WBM was not independent of ISS staging. Of note, the percentage of MMCs in the WBM in the HPSE^low group (median 4.5; range, 0.03-45) was not statistically significantly different from that in the HPSE^high group (median, 2.9; range, 0.5-12). Classifying patients in 2 groups according to HPSE expression in purified MMCs did not yield any difference in EFS or OAS, neither in this set of 30 patients (Figure 6B) nor in a larger set of MMCs from 104 patients with newly diagnosed MM (data not shown). This again emphasizes that HPSE is mainly expressed by the BM environment and that this expression is biologically relevant. This is the first study showing that a gene measured in the bone marrow microenvironment has prognostic value in MM.

The fact that HPSE controls syndecan-1 expression and soluble syndecan-1 production may explain at least in part the poor prognosis of HSPE^high patients, as increased levels of soluble syndecan-1 have been described as a bad prognostic factor in patients with MM.^10,11  Soluble syndecan-1 can promote tumor growth in a murine model,¹³  presumably because it can sequester heparin-binding growth factors that support MMC growth as well as angiogenic factors such as VEGF,^5,29,30  thus providing MMCs with an increased amount of growth factors. Consistently, Kelly et al¹⁹  have shown HPSE to increase microvessel density, which in turn correlates with poor prognosis in MM.^31,32  In an independent series of 20 patients with newly diagnosed MM, we did not find any significant correlation between HPSE expression in the WBM and soluble syndecan-1 concentration in the plasma (data not shown), likely because soluble syndecan-1 can be retained by the BM matrix and hence only partially circulates.³³  The impact of HPSE expression on survival may further be linked to the modification of syndecan-1 HS chains by the endoglycosidase enzymatic activity of HPSE. The ability of HPSE to degrade HS chains into small fragments is critical to promote tumor growth in epithelial tumors,^16,34  and HPSE enzymatic activity is present in the BM aspirates of patients with myeloma.¹⁹  HPSE may promote tumor growth by releasing heparin-binding MMC growth factors or angiogenic factors that are trapped within the environment through the HS chains of syndecan-1. Accordingly, it has been shown that HPSE can release HS-bound growth factors such as FGF2 from the extracellular matrix.³⁵  Alternatively, HPSE may improve the HS-mediated effect of heparin-binding growth factors. Indeed, small fragments resulting from the HPSE cleavage were reported to be more biologically active than the native HS chains from which they are derived.¹⁷  Finally, besides its endoglycosidase activity, it is now becoming clear that HPSE might have other nonenzymatic functions which may also have an impact on tumor growth. It was not within the scope of our study to elucidate the mechanism involved in the regulation of syndecan-1 gene expression and shedding by HPSE, but one can speculate that it might involve such nonenzymatic functions. These nonenzymatic functions include stimulation of endothelial cell invasion via AKT pathway activation,³⁶  enhanced cell adhesion of glioma,³⁷  lymphoma,³⁸  and T cells mediated by β1-integrin,³⁹  and induction of VEGF expression.⁴⁰  These mechanisms are poorly understood but might involve a yet undefined receptor that could lead to activation of transduction pathways and mediate diverse biologic effects.

Altogether, these data provide evidence that HPSE can modulate syndecan-1 activity through the cleavage of its HS chains but also through the regulation of syndecan-1 mRNA expression and shedding. HPSE, mainly produced by the environment, is an indicator of poor prognosis for patients with MM. Therefore, HPSE inhibitors such as PI-88, which is currently in clinical trials,⁴¹  will be of interest for the treatment of patients with MM.

This work was supported by grants from the Ligue Nationale Contre Le Cancer (équipe labellisée), Paris, France. It is part of a national program called Carte d'Identité des Tumeurs (CIT) funded by the Ligue Nationale Contre le Cancer (http://www.ligue-cancer.net).

Contribution: K.M. designed the research, performed the experiments; and wrote the paper; D.H., M.H., E.J., M.B., T.M., J.-F.R., and H.G. collected bone marrow samples and clinical data and participated in writing the paper; P.R., V.P., and M.L. contributed in performing the experiments; T.R., M.J., and J.D.V. participated in analyzing the data and in writing the paper; and B.K. is the senior investigator, who designed the research and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Bernard Klein, INSERM U475, 99, rue Puech Villa, 34197 Montpellier Cedex 5, France; e-mail: klein@montp.inserm.fr.