Serglycin Constitutively Secreted by Myeloma Plasma Cells Is a Potent Inhibitor of Bone Mineralization in Vitro*

Although the biological significance of proteoglycans (PGs) has previously been highlighted in multiple myeloma (MM), little is known about serglycin, which is a hematopoietic cell granule PG. In this study, we describe the expression and highly constitutive secretion of serglycin in several MM cell lines. Serglycin messenger RNA was detected in six MM cell lines. PGs were purified from conditioned medium of four MM cell lines, and serglycin substituted with 4-sulfated chondroitin sulfate was identified as the predominant PG. Flow cytometry and confocal microscopy showed that serglycin was also present intracellularly and on the cell surface, and attachment to the cell surface was at least in part dependent on intact glycosaminoglycan side chains. Immunohistochemical staining of bone marrow biopsies showed the presence of serglycin both in benign and malignant plasma cells. Immunoblotting in bone marrow aspirates from a limited number of patients with newly diagnosed MM revealed highly increased levels of serglycin in 30% of the cases. Serglycin isolated from myeloma plasma cells was found to influence the bone mineralization process through inhibition of the crystal growth rate of hydroxyapatite. This rate reduction was attributed to adsorption and further blocking of the active growth sites on the crystal surface. The apparent order of the crystallization reaction was found to be n = 2, suggesting a surface diffusion-controlled spiral growth mechanism. Our findings suggest that serglycin release is a constitutive process, which may be of fundamental biological importance in the study of MM.

Serglycin is a unique proteoglycan (PG), 4 containing a small peptide core of 17.6 kDa, rich in serine/glycine repeats. These repeats are found in the central part of the core protein and carry eight glycosaminoglycan (GAG) side chains. Although serglycin does not contain a transmembrane domain, this PG was initially discovered at the cell membrane of rat L2 yolk sac tumor cells (1) and was the first PG gene to be cloned (2). Serglycin is synthesized mainly by cells of hematopoietic origin (3), although recent evidence demonstrates expression by other cell types such as endothelial (4) and pancreatic acinar cells (5) as well as in embryonic stem cells (6). It is packaged into secretory granules or vesicles in platelets, mast cells, neutrophils and other leukocytes and is secreted upon activation (3). In other cell types such as lymphocytes and hematopoietic tumor cells (7)(8)(9)(10)(11), endothelial cells (4), and uterine decidua (12) it is constitutively secreted along with granule proteins. Serglycin may contain chondroitin sulfate (CS) or heparin chains. In most blood cells serglycin bears CS chains, with the exception of connective tissue mast cells, where heparin chains are attached to the core protein (3,4,7,8,13,14).
The function of serglycin has not been well studied, but it is likely involved in the packaging of proteins into secretory granules and/or directing the secretion of these molecules (15). Recent studies have demonstrated that serglycin plays a key role in the storage of mast cell proteases and granzyme B and the maturation of secretory granules of mast cells and cytotoxic T-lymphocytes (16,17). Serglycin binds a wide variety of important secretory proteins (such as intragranular platelet factor 4, macrophage inhibitory protein-1␣ (MIP-1␣) (14), mast cell chymases (15), extracellular matrix components (such as fibronectin and collagen) (7,18), and cell surface molecules (such as CD44) (19,20). It is proposed that serglycin, through its GAG chains, serves as a carrier of other molecules during secretion, directing their binding to other specific ligands with high affinity. A novel mechanism for the role of serglycin in the delivery of granzyme B has been proposed (21). Granzyme B complexed with serglycin is liberated by serglycin at the cell surface of target cells by binding to higher affinity sites on cell surface through electrostatic transfer.
Multiple myeloma (MM) is a B-cell malignancy characterized by slow proliferation of malignant plasma cells within the bone marrow. The survival of MM cells is dependent on the growth conditions provided by the bone marrow microenvironment. PGs such as syndecan-1, present on myeloma cell surface but also shed in extracellular matrix (22), are important for myeloma cell biology and bone homeostasis (23)(24)(25)(26)(27)(28).
Bone destruction is a common complication of MM and is associated with severe morbidity. Several osteoclast-activating factors are implicated in myeloma bone disease; however, the disruption of the balance between the receptor activator of NF-B ligand (RANKL) and osteoprotegerin is of major importance (29). The interaction between the receptor activator of NF-B (RANK) expressed on the surface of cells of osteoclastic lineage and RANKL expressed on stromal cells plays a key role in the development and activation of osteoclasts, whereas osteoprotegerin, a decoy receptor for RANKL secreted from stromal cells, inhibits RANKL-RANK signaling. MM cells stimulate osteoclastogenesis by triggering a coordinated increase in RANKL and a decrease in osteoprotegerin in the bone marrow (29). Radiographic examinations show radiolucent lesions without calcification in many patients, suggesting that along with enhanced bone resorption, mineralization is impaired in MM bone lesions. However, the mechanisms of impaired bone formation in bone lesions of patients with MM are poorly understood (30).
Bone mineralization is the ion deposition in the tissue, which in vertebrates appears as hydroxyapatite (HAP) formation. HAP (Ca 5 (PO4) 3 OH), is the most stable calcium phosphate salt under physiological conditions. It has been suggested as the model compound for the study of hard tissue calcification such as bone. The role of various biological molecules as modulators of biomineralization has also been appreciated. PGs are involved in the calcification process acting either as promoters or inhibitors (31)(32)(33). PGs accumulated in the bone marrow in patients with MM may directly influence bone mineralization process.
The increasing awareness of the potential biological effects of PGs in MM prompted us to study secreted PGs in this disease. We postulated that, apart from syndecan-1, MM cells secrete additional PGs into the extracellular milieu. Therefore, we isolated and quantified PGs in conditioned medium from MM cells. Interestingly, we found a highly constitutive secretion of serglycin in all MM cells lines studied. We characterized the serglycin secreted from MM cells and found that it is the major PG secreted by these cells. Serglycin was associated to the cell membrane in a GAG-dependent manner but also intracellularly to cytoplasmic vesicles or granules. The expression of serglycin by MM cells was also confirmed in patient material by immunohistochemistry and immunoblotting. Serglycin isolated from MM cell lines was a potent inhibitor of bone mineralization process having an inhibitory effect on HAP crystal growth in vitro.

EXPERIMENTAL PROCEDURES
Cell Lines and Cultures-The human myeloma cell line JJN3 was a gift from Jennifer Ball (Department of Immunology, University of Birmingham, UK). The U266 cell line was purchased from American Type Culture Collection (ATCC, Manassas, VA). The OH-2 and IH-1 cell lines were established by us (34,35). As kind gifts we received CAG (J. Epstein, Little Rock, AK) and INA-6 cell lines (M. Gramatzki, University of Erlangen-Nuremberg, Erlangen, Germany). All of these cell lines were cultured in RPMI 1640 supplemented with 10% fetal calf serum. IH-1 and OH-2 were supplemented with 10% human serum and 2 ng/ml IL-6. The INA-6 cell line was also supplemented with 2 ng/ml IL-6. All culture media contained 2 mmol/liter glutamine and 40 g/ml gentamycin. The cell lines were cultured in a humidified atmosphere containing 5% CO 2 at 37°C.
Preparation of Conditioned Medium-Conditioned medium was prepared from four MM cell lines. The JJN3, U266, CAG, and INA-6 cell lines grown to high density were centrifuged, and the supernatant was collected, passed through a sterile filter, and stored frozen until further analysis (Ϫ20°C).
Isolation of PGs by DEAE-Sephacel-One liter of cell culture medium prepared as described above was brought to a final concentration of 10 M formamide, 50 mM sodium acetate buffer, pH 6.3, and protease inhibitors (10 mM EDTANa 2 , 10 mM benzamidine hydrochloride, 1 mM phenylmethylsulfonyl fluoride) were added (36). Medium was passed through a DEAE-Sephacel column (40-ml bed volume) equilibrated with the above formamide-sodium acetate buffer. The column was washed with 3 bed volumes of the above buffer containing 0.2 M NaCl to elute the nonproteoglycan components, and PGs were fractionated by gradient elution ranging from 0.2 M to 0.9 M NaCl. Fractions were collected, and aliquots were precipitated in 80% ethanol, 1.3% potassium acetate. Precipitates were dissolved in distilled water and were monitored and quantified by terms of GAG content using the DMB method (37) and a commercial syndecan-1 enzyme-linked immunosorbent assay (Diaclone, Besancon, France). Aliquots of the fractions were also precipitated and subjected to SDS-PAGE analysis using a 4 -16% gradient resolving gel (Bio-Rad). Gels were stained with 0.1% toluidine blue in 0.1 M acetic acid followed by silver staining (Pierce).
The major PG population eluting with 0.55-0.7 M NaCl was pooled and subjected to gel permeation chromatography on Sephacryl S-400 (GE Healthcare) columns before and after treatment with trypsin. Trypsin digestion (1 mg/ml trypsin, Invitrogen) was performed overnight at 37°C in phosphatebuffered saline (PBS) containing 0.02% (w/v) EDTANa 2 . The columns were eluted with 4 M guanidine hydrochloride, 50 mM sodium acetate, pH 5.8. Fractions were collected and aliquots precipitated as above, and PGs were monitored by the DMB method and by SDS-PAGE analysis.
Characterization of Serglycin by N-terminal Sequencing-Samples of the major PG population of the four MM cell lines were treated with chondroitinase ABC in 50 mM Tris-HCl, pH 7.5, containing the protease inhibitors (10 mM EDTANa 2 , 1 mM phenylmethylsulfonyl fluoride, 10 mM benzamidine hydrochloride). Samples were electrophoresed on 4 -16% gradient resolv-ing gel and transferred to Immobilon P membranes at a constant current of 80 mA at 4°C for 12 h in 50 mM Tris-HCl, pH 8.3. The membranes were stained with Coomassie R-250, and the first nine amino acids from the N terminus of the protein bands observed were sequenced on an automated sequencer (Protein Analysis Center, Karolinska Institutet).
Expression of Serglycin in Six Myeloma Cell Lines by Reverse Transcriptase Polymerase Chain Reaction-Total RNA was isolated with the High Pure total RNA isolation kit (Roche Applied Science). Total RNA was first reverse-transcribed into cDNAs using the first-strand cDNA synthesis kit (GE Healthcare). The expression of serglycin in MM cell lines was examined by using specific primers (5Ј-AATGCAGTCGGCTTGT-CCTG and 3Ј-GCCTGATCCAGAGTAGTCCT) to generate a 272-base pair fragment that bridged the first intron and included about half of the serine-glycine repeat region (4). Three hundred nanograms of the reverse transcriptase reaction products was then subjected to PCR amplification in a total volume of 50 l (10 mM Tris-HCl, 50 mM KCl, pH 8.3, containing dNTP mix (each at 0.16 mM), both downstream and upstream primers (each at 1 M), 1.5 mM MgCl 2 , and 2.5 units of Tag polymerase. The PCR amplification was carried out for 35 cycles: 94°C for 1 min, annealing temperature 51°C for 1 min, 72°C for 2 min (Peltier Thermal Cycler 200, MJ Research, SDS Promega). The PCR products were electrophoresed in 2% agarose gel containing 1 g/ml ethidium bromide. Bands visible in UV light were cut out and extracted using the Qiax gel extraction kit (Qiagen, Hilden, Germany). The identity of the PCR products was confirmed by sequence analysis. Sequencing was carried out by the dideoxy chain termination method (Cybergene, Stockholm, Sweden).
Characterization of Glycosaminoglycan Moiety of Serglycin by SDS-PAGE and Capillary Electrophoresis-The isolated populations were analyzed by SDS-PAGE before and after treatment with chondroitinase ABC and/or chondroitinase AC II or a mixture of heparitinases. Digestion with chondroitinase ABC was performed in 50 mM Tris-HCl, pH 7.5, at 37°C for 6 h using 0.01 unit/10 g of uronic acid. Chondroitinase ACII digestion was performed in the same buffer at pH 6.0 under the same conditions. Treatment with heparin lyases I, II, and III was performed in 20 mM sodium acetate, pH 7.0, containing 1 mol of calcium acetate at 37°C overnight. The gels (4 -16%) were stained with toluidine blue followed by silver staining. Aliquots of the digests were also analyzed by capillary electrophoresis using an HP 3D CE (Agilent Technologies) (38). Separation and analysis were carried out on an uncoated fused silica capillary tube (75 m inner diameter, 55 cm total length, 50 cm effective length to the detector) at 25°C. The capillary tube was washed for 1 min with 0.1 M NaOH before each run and then with the operating buffer (15 mM orthophosphate, pH 3.0) for 4 min. Samples were introduced by the pressurized mode at the cathode (reverse polarity) and separated at 20 kV. Detection was performed at 232 nm using a diode array detector. Quantification was performed using the HP ChemStation software provided by the manufacturer with the instrument.
Polyclonal Antibody Production and Immunoblotting-Polyclonal antibody production was performed by Invitrogen. Briefly, a keyhole limpet hemocyanin-conjugated 25-amino acid peptide of serglycin (RLRTDLFPKTRIQDLNRIFPLSEDY), starting immediately before the serine-glycine repeat region, was synthesized. Two rabbits were immunized according to standard protocol. Bleeds from week 8 and 10 were pooled, and affinity purification against the peptide was performed.
The antibody was tested on serglycin isolated from MM cell lines and was used for serglycin detection in patients bone marrow biopsies. Serglycin preparations isolated from all four MM cell lines and nonfractionated condition medium as well as bone marrow samples (1 l of bone marrow plasma) from patients with MM and healthy volunteers were diluted in Laemmli sample buffer (Bio-Rad), electrophoresed on a 4% stacking-10% resolving gel, and transferred to Immobilon P membranes at constant current 80 mA at 4°C for 12 h in 50 mM Tris-HCl, pH 8.3. The membranes were washed with 0.15 M NaCl in 0.01 M phosphate buffer, pH 7.2, containing 0.1% Tween 20 (PBS-T) and blocked with 5% dry defatted milk in PBS-T. They were immersed in polyclonal antibody produced against serglycin peptide diluted 1:10000 in PBS-T and incubated for 60 min at room temperature. The membranes were washed and incubated with peroxidase-conjugated secondary antibody (goat anti-rabbit) diluted 1:2000 in PBS-T for 60 min at room temperature. The membranes were washed, and bands were visualized by the enhanced chemiluminescence technique (ECL, Amersham Biosciences) according to the instructions of the manufacturer. For semiquantitative analysis, the density of immunoreactive bands was estimated following image analysis of immunoblots with the UNIDocMv program (version 99.03) for Windows (UVI Tech, Cambridge, UK).
Flow Cytometry-All six MM cell lines were subjected to flow cytometry to detect the presence of serglycin on the cell surface. Approximately 5 ϫ 10 5 cells were washed in PBS and incubated with polyclonal rabbit anti-serglycin antibody (10 g/ml) or nonimmune rabbit IgG (10 g/ml) in 0.1% bovine serum albumin in PBS. After washing, the cells were incubated with fluorescein isothiocyanate-conjugated donkey anti-rabbit antibody (DAKO Cytomation, Copenhagen). In a FACSscan flow cytometer (BD Biosciences), 10000 cells from each sample were analyzed on a single-cell basis. As a specificity control, anti-serglycin antibody (10 g/ml) was incubated with the synthetic serglycin peptide (50 g/ml) or purified serglycin (50 g/ml) for 10 min prior to labeling of the cells. The cells were also treated with 0.1 unit of chondroitinase ABC for 30 min at 37°C prior to labeling with anti-serglycin or control antibody. Results were depicted as frequency distribution histograms.
Confocal Microscopy-MM cell lines were subjected to confocal microscopy analysis to detect the presence of serglycin. Briefly, for extracellular staining, 10 5 cells were washed in PBS. Cells were fixed in 2% formaldehyde in PBS supplemented with 5% human serum for 15 min at room temperature. Nonspecific staining was blocked with 5% human serum in PBS for 15 min. After washing, cells were incubated with polyclonal rabbit antiserglycin antibody (10 g/ml) or normal rabbit IgG (10 g/ml) in PBS/serum for 30 min, where after nuclei were counterstained with 1 mg/liter bisbenzidine (Fluka, Steinheim, Germany). Cells were washed and incubated with Alexa-647-conjugated donkey anti-rabbit antibody (5 g/ml) for 30 min. Cells were washed, resuspended in 100 l of PBS, and examined in a Petri dish. For intracellular staining, the above protocol was applied, but all incubations and washes were performed in 5% human serum supplemented with 0.3% saponin. Examination of immunocytochemical staining was performed with a Leica TCS NT confocal laser-scanning microscope equipped with an ArKr laser, permitting the detection of signals from the fluorochromes, with emission wavelengths at 488 nm and 568 nm. For each experiment the excitation and sensitivity of the detector were adjusted so that the corresponding negative controls with normal IgG gave barely detectable signals. Subsequently all settings were kept equal for the analysis of anti-serglycin-labeled cells. As a specificity control, anti-serglycin antibody (10 g/ml) was incubated with the synthetic serglycin peptide (50 g/ml) or purified serglycin (50 g/ml) for 10 min prior to labeling the cells.
Immunohistochemistry-Reactivity to the rabbit anti-serglycin antibody and CD138 (syndecan-1) was also studied using parallel sections of routinely fixed (4% buffered formalin) and paraffin-embedded bone marrow. For this purpose five tissues with MM and five tissues with benign proliferation of plasma cells (including one case with monoclonal gammopathy of unknown significance (MGUS)) were retrieved from the archives at Karolinska University Hospital, Huddinge. Sections (3 m) were deparaffinized, rehydrated, decalcified in formic acid, and subjected to epitope retrieval in a microwave oven using ChemMate target retrieval solution (Dako, Copenhagen, Denmark) diluted 1:10 according to manufacturer instructions. Endogenous peroxidase quenching was performed with the ChemMate peroxidase-blocking solution (Dako). Blocking was performed with protein block solution (Dako), and sections were incubated with a 28 g/ml dilution of the primary antibody against serglycin or a 1:200 dilution of CD138/syndecan-1 mouse monoclonal antibody in Tris-buffered saline. The staining to demonstrate serglycin and syndecan-1 was performed using the ChemMate detection kit peroxidase/DAB (Dako) in a DakoCytomation TechMate Autostainer. The sections were counterstained with hematoxylin.
Bone Marrow Plasma Samples from Patients and Controls-Sixteen patients with newly diagnosed multiple myeloma were included for the analysis of serglycin in the bone marrow plasma. Before treatment, after informed consent, bone marrow plasma was aspirated from the crista iliaca or sternum. Aspirates were centrifuged and plasma was frozen at Ϫ70°C until analysis. Fourteen healthy volunteers were included as controls. For all controls, bone marrow plasma was collected and stored in an identical manner as described for patients with MM.
In Vitro Effect of Serglycin in Hydroxyapatite Crystal Growth-Analytical reagent grade chemicals and calibrated A-grade volumetric glassware were used. Stock solutions of calcium dichloride, potassium dihydrogen phosphate, and sodium chloride were prepared from the respective crystalline solids (Merck) using triply distilled CO 2 -free water. The pH was measured using a combined glass/Ag/AgCl electrode (Metrohm, 6.0202.100), standardized before and after each experiment with National Bureau of Standards (NBS) standard buffer solutions. The HAP seed crystals were prepared from solutions of Ca(NO 3 ) 2 , H 3 PO 4 , and NH 3 , aged in the mother liquor for 8 months, filtered, and dried as described previously (39). Their specific surface area was found to be 34.6 m 2 /g as determined by a multiple point Brunauer, Emmett, and Teller (BET) method (PerkinElmer Life Sciences, Sorptometer 212D). The synthetic HAP crystals were characterized by scanning electron microscopy (Jeol GSM 5200, Leo Supra 35VP), x-ray powder diffraction (Philips PW 1830/1840, CuK ␣ radiation, nickel filter), infrared spectroscopy (KBr pellet method, FT-IR PerkinElmer Life Sciences 16-PC), and chemical analysis. The crystals showed the characteristic powder x-ray diffraction pattern and the infrared spectrum of stoichiometric HAP, and from the chemical analysis the molar ratio of calcium to phosphates was found to be 1.67 Ϯ 0.02.
Crystal growth experiments were performed at constant composition conditions (i.e. the number of moles of all the reactants was maintained constant during the reaction) (39). Experimental conditions were chosen to physicochemically resemble the biological ones (i.e. the ionic strength was 0.15 M in NaCl, and the pH was equal to 7.40, both similar to that of human blood; the temperature was adjusted to 37 Ϯ 0.1°C). The working solutions were prepared in a covered thermostated doublewalled glass reactor with volume totaling 15 ml. Pure gas nitrogen (99.99%; presaturated with water at 37°C prior to introduction) was bubbled through the working solution before and during the course of the experiment to exclude atmospheric carbon dioxide. HAP formation is followed by proton release, and therefore we had a sensitive means of monitoring the crystallization reaction. A pH drop as much as 0.005 pH units as a result of HAP precipitation triggered the addition of reactants from two mechanically coupled burettes of an appropriately modified pH-stat system (Metrohm 614 Impulsomat). The composition of the titrant solutions in the two burettes was CaCl 2 and NaCl in the first and KH 2 PO 4 and KOH in the second. The molar concentration ratio of the titrants corresponds to the stoichiometry of the precipitating HAP, Ca:P:OH ϭ 5:3:1. As a result, the HAP crystallization took place under conditions of constancy on the supersaturation and on the concentration of the reactants. This was verified by chemical analysis of aliquots withdrawn randomly during the crystal growth process. In all cases, the concentration remained constant throughout the course of the experiment to within Ϯ3%. The samples taken were filtered through membrane filters (Millipore 0.22 m), and the solids removed by filtration were examined by scanning electron microscopy, x-ray powder diffraction, infrared spectroscopy, specific surface area, and chemical analysis. The rates of crystallization in the presence (R i ) and absence (R 0 ) of serglycin were calculated from the rates of titrant addition and corrected for surface area changes. The reproducibility of the measured crystal growth rates of HAP formation was better than 2%. The HAP crystal growth experiments in the presence of serglycin isolated from the CAG cell line were done in a way so that the crystal surface could have sufficient time to interact with the foreign substance prior to onset of the crystallization process. Thus, the synthetic HAP crystals (3 mg) were in the reactor in a solution of phosphates and serglycin at the desired pH and ionic strength 2 h before the experiment was started by adding the appropriate quantity of CaCl 2 solution. The procedure described here is advantageous NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 compared with the classical constant composition method, because adsorption phenomena do not interfere with kinetic measurements (40).

Identification of Serglycin Secreted by MM Cell
Lines-The secretion of serglycin by MM cell lines was examined following isolation of PGs by combined anion-exchange and gel perme-ation chromatographies and subsequent characterization of fractionated PGs. PGs were isolated from conditioned medium of four MM cell lines (three IL-6-independent cell lines, U266, JJN3, and CAG, and IL-6-dependent INA-6). One liter of culture medium was chromatographed on a DEAE-Sephacel column, and PGs were fractionated by linear gradient elution ranging from 0.2 to 0.9 M NaCl. As shown in Fig. 1, the major PG population was eluted with 0.55-0.7 M NaCl in all cases and was pooled as shown by the bars (on each panel). The major PG population was almost completely separated from shed syndecan-1, which was eluted with a lower NaCl concentration and represents a minor population. In a control experiment, including normal culture medium with fetal calf serum, only trace amounts of GAGs (Ͻ15 g/liter) were detected. The pooled PG populations from DEAE-Sephacel were then analyzed by gel permeation chromatography on a Sephacryl S-400 column. In all cases the isolated PGs were eluted close to the void volume of the column as homogeneous populations with K av ranging from 0.13 to 0.18 (Fig. 2). Rechromatography on the same column after treatment with trypsin showed elution profiles identical to those before treatment, revealing that these PGs were trypsin-resistant. This is consistent with the behavior of serglycin, which retains the GAG attachment region intact after trypsin digestion, resulting in only a small decrease in overall size due to some proteolysis of the core protein, which could not detected following such a chromatographic procedure (Fig. 2).
The nature of isolated PGs was examined by SDS-PAGE before and after treatment with various GAGdegrading enzymes. PGs were resistant to the action of a mixture of heparin lyases (figure not shown), but their GAG chains were completely removed by treatment with chondroitinase ABC (Fig. 3A), suggesting that these PGs contain chondroitin sulfate and/or dermatan sulfate chains. Samples treated with chondroitinase ABC were also transferred to Immobilon P membrane, and the membrane was stained with Coomassie R-250. A major band at 28 kDa was apparent in all preparations   (Fig. 3B). The band at 60 kDa and other minor protein bands were derived from the enzyme preparation; the 60-kDa band corresponds to bovine serum albumin included as stabilizer in the enzyme preparation. Although the cDNA for serglycin predicts a molecular size of 14.4 kDa for the mature protein, it should be noted that the core protein of serglycin isolated from various sources exhibits a molecular size of ϳ28 kDa on SDS-PAGE even in the presence of ␤-mercaptoethanol. The increased size of the core protein is most probably attributable to the residual stubs of GAGs remaining after enzymic depolymerization. The 28-kDa bands were excised, and their N-terminal amino acid sequence was analyzed. In all preparations the obtained sequence was identical with that of human mature serglycin starting immediately after the signal peptide (Fig. 3B). In some blots the bands of ϳ28 kDa appeared as doublets. Both bands were sequenced, and their N-terminal sequences were identical with that of serglycin. It had previously been suggested (4) that these doublets are characteristic of serglycin core proteins substituted with a different number of GAG chains.
Detection of Serglycin mRNA in MM Cell Lines-The expression of serglycin by MM cell lines was also confirmed by reverse transcription PCR. In addition to the four MM cell lines used for the isolation of serglycin, two more IL-6 dependent cell lines, OH-2 and IH-1, were included, and total RNA was amplified using specific primers for serglycin. The reverse transcription PCR products obtained (Fig. 4) were subjected to sequence analysis, which confirmed that all six MM cell lines synthesize mRNA for serglycin.
Quantification of Serglycin-We went on to quantify the serglycin secreted by four MM cell lines and to compare its levels with that of other PGs. Serglycin preparations were quantified by terms of GAG content using the DMB method and whale chondroitin 4-sulfate as GAG standard. Other PGs fractionated on a DEAE-Sephacel column were collected, and their GAG content was also measured. In this pool of all other secreted PGs the amount of shed syndecan-1 was estimated by enzymelinked immunosorbent assay. We found that serglycin represented the vast majority of PGs secreted by these cells; its concentration ranged from 0.75 mg GAG/liter to 1.86 mg GAG/ liter of culture medium (Fig. 5). Other PGs amounted to less than 30% of total secreted PGs. We noted that in contrast to the constant secretion of serglycin, the concentration of shed syndecan-1 was highly variable between cell lines, and in some of them syndecan-1 was hardly detected (syndecan-1 in U266, 190 g/liter; in CAG, 30 g/liter; in JJN3, 16 g/liter; in INA-6, 12 g/liter).
Identity of GAG Chains and Disaccharide Composition-To characterize GAGs on serglycin, preparations were digested by chondroitinase ABC and AC II in combination or with chondroitinase AC II alone, and the digestion mixtures were sub-   NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 jected to capillary electrophoresis. CS is the only GAG attached to myeloma-derived serglycin in our cell lines. This was demonstrated by the complete liberation of disaccharides by chondroitinase AC II. Combined digestion with chondroitinase ABC and AC II did not increase the absolute amount of disaccharide obtained following degradation. Furthermore, products corresponding to heparin, heparan sulfate, or hyaluronan were not observed following treatment with the appropriate enzymes and capillary electrophoresis (data not shown). CS in serglycin preparations was almost completely constituted by 4-sulfated disaccharides (percentage ranging from 87 to 93%), whereas only minute amounts of 6-sulfated (2-10%) and nonsulfated units (3-5%) were detected (Fig. 6). Surprisingly, in contrast to other serglycin preparations described in the literature, our serglycin contained only trace amounts of disulfated units (mainly (4,6)-disulfated).

Serglycin in Multiple Myeloma
Specificity of Polyclonal Antibody-To examine the expression of serglycin in clinical samples, we prepared a polyclonal antibody against a stretch of the peptide representing part of the human serglycin peptide core (Fig. 7A). The specificity of the polyclonal antibody was verified by Western blotting. Isolated serglycin from all four MM cell lines digested with chondroitinase ABC (Fig. 7B, left panel) and nondigested culture medium from the CAG cell line (Fig. 7B, right panel) were transferred to Immobilon P membrane, and immunolabeling was performed. The antibody recognized only the 28-kDa protein band in the chondroitinase ABC-digested samples. Interestingly, the nondigested sample was labeled on the top of the separating gel, where intact serglycin remained, indicating that the antibody is able to recognize intact serglycin. When the blot was incubated with anti-serglycin antibody, which had been preincubated with the peptide used for immunization in rabbits, no signal for serglycin core protein was obtained (data not shown).
Serglycin Is Present on the Myeloma Cell Surface-We also employed this antibody to analyze the presence of serglycin on the MM cell surface by means of flow cytometry. The signal was specific, since preincubation of the antibody with synthetic serglycin peptide or serglycin isolated from MM cell lines completely inhibited cell surface signal (figure not shown). A control antibody was also included for all experiments. Although the signal intensity for serglycin varied somewhat between cell lines, five cell lines in our panel expressed serglycin on the cell surface. No surface-bound serglycin was found on U266 cells (Fig. 8A). Pretreatment of cells with chondroitinase ABC partly reduced the level of serglycin on the cell surface (Fig. 9), suggesting that the binding of serglycin on the cell surface was at least partly dependent on the GAG moiety of serglycin.
Confocal Microscopy Analysis of Serglycin-Confocal microscopy analysis of nonpermeabilized cells, employing polyclonal antibody against serglycin or control antibody, indicated the presence of this PG on the entire MM cell surface (shown in Fig.  8B). In keeping with the findings in flow cytometry, the signal from the cell lines was somewhat varying; CAG, JJN3, and INA-6 cells stained brightly in a halo-like pattern. OH-2 and IH-1 stained weakly in a spotted pattern on the cell surface. As in the flow cytometry findings, no surface serglycin could be detected in U266 cells. CAG and U266 cells were permeabilized to demonstrate possible intracellular staining for serglycin. CAG cells stained for serglycin in an area near the nucleus, whereas in U266 cells, serglycin was widely present in the entire cytoplasm (Fig. 10).
Distribution of Serglycin in Bone Marrow-When applied to bone marrow tissues from patients with MM, reactivity with the anti-serglycin antibody appeared both in the cell cytoplasm and extracellularly (Fig. 11A). Different cell lines contained this PG, with the strongest reactivity found in megakaryocytes and platelets. Moderate to strong reactivity was seen in myeloid cells and in plasma cells, whereas lymphocytic cells appeared to be unreactive. The serglycin reactivity seemed specific, because no binding could be demonstrated after preincubation of the antibody with excess amounts of the epitope (figure not shown) or incubation with nonimmune control rabbit IgG (Fig. 11B). Plasma cell reactivity for serglycin was seen both in MM infiltrates (Fig. 11A) and in benign plasma cells (Fig. 11D), the latter cells could be identified by typical cartwheel chromatin texture and CD138 membrane reactivity (Fig. 11C). A limited number of bone marrow plasma samples from patients with MM (16 cases), as well as 14 control samples from healthy volunteers,  were analyzed by immunoblotting (Fig. 12A). In all cases the reactivity of antibody against serglycin was observed on the top of the gels, showing a polydisperse high molecular size population with the typical appearance of PGs. The density of immunoreactive bands was estimated, and the quantitative data are presented in Fig. 12B. Bone marrow samples from healthy individuals appeared to contain similar amounts of serglycin, in contrast to samples from MM patients, which exhibited diversity in serglycin amounts. Some MM samples contained very low amounts of serglycin (as shown in Fig. 12A, lanes 5 and 6), but almost 30% (5 of 16 cases) of the examined bone marrow biopsies from MM patients contained markedly increased amounts of serglycin in the extracellular milieu (Fig. 12, A, lane  7, and B, encircled cases).
Serglycin Is a Potent Inhibitor of Hydroxyapatite Crystal Growth-When serglycin isolated from the CAG cell line was present in the supersaturated solutions, the rate of HAP crystal growth decreased. At higher concentrations of serglycin in the supersaturated solution a more drastic decrease of the crystallization rate of HAP is observed (Fig. 13). The kinetic and thermodynamic data and the experimental conditions employed in the experiments are illustrated in Table 1. The driving force for crystallization may be expressed in terms of the change in Gibbs free energy of transfer from unstable supersaturated to saturated solutions (e.g. thermodynamic equilibrium) shown in Equation 1, where IP is the ionic product of the precipitating salt, k s o ϭ 2.35 ϫ 10 Ϫ59 its thermodynamic solubility product at 37°C, is the number of ions in the formula unit of the precipitating calcium phosphate phase (e.g. 9 for HAP), R g is the gas constant, T is the absolute temperature, and ⍀ is the supersaturation ratio.
The ionic product, IP, was calculated from the concentrations of all ionic species in the working solution during crystallization. For the computation of the solution speciation, dissociation constants for phosphoric acid and the equilibrium constants for ion pair formation of calcium, potassium, and sodium, which are present in the solution with the phosphate anions, were involved. The procedure involved successive approximations of the ionic strength (40) using expressions for mass balance, electroneutrality, and ion association equilibrium constants; for this reason computer programmed software was developed.
Crystal growth rates were found to be proportional to the   relative solution supersaturation, , with respect to HAP, defined as and R ϭ ks n where R is the crystal growth rate, k the rate constant, s a function of the number of the active growth sites on the crystal surface, and n the apparent order of the crystallization reaction. Logarithmic plots according to Equation 3 yielded straight lines in the presence and absence of serglycin (Fig. 14A). From the slope of the linear plots, a value of 1.54 Ϯ 0.16 was obtained for HAP crystallization in the absence of any additive, whereas in the presence of serglycin the corresponding value is 1.50 Ϯ 0.39.
These obtained values of the slopes correspond to an apparent second order (n ϭ 2) crystallization reaction, which suggests a surface diffusion-controlled mechanism (41). In the presence of serglycin, no changes were observed in the HAP overgrown morphology, as verified from scanning electron microscopy studies (Fig. 14, B and C). Furthermore, no traces of other calcium phosphate solid phases were detected from the spectroscopic infrared and crystallographic x-ray patterns of the precipitated material. It is therefore noted that serglycin affected only the crystallization kinetics of HAP, causing no changes in the mechanism of crystal growth or the morphology of the crystals formed nor favoring the formation of another calcium phosphate salt. As seen from Table 1, when serglycin was present in the supersaturated solution, the rate of HAP crystal growth decreased significantly. The presence of a foreign compound in the supersaturated solution in which a crystal growth process is taking place very often results in the interaction of the solute species with the surface of the precipitating solid. When the solute species have functional groups such as carboxyl and/or amino groups, they may adsorb reversibly on the crystal surface, which contains centers of positively and negatively charged ions, and therefore the solute species may be accommodated on the surface. It has been suggested that the adsorption of inhibitor molecules at active growth sites on the crystal surfaces accounts for the reduction of the crystal growth rate. Thus, the growth centers are blocked, and the adsorbed mole- FIGURE 13. Effect of serglycin on HAP crystal growth rate. HAP crystal growth was studied in the presence of various concentrations of serglycin; the growth rate, R i , was measured, and the ratio, R i /R 0 (where R 0 is the rate of HAP crystal growth in blank samples), was plotted against the respective concentrations.   NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46

Serglycin in Multiple Myeloma
cules prevent the crystal lattice species from incorporation in the crystal.
Assuming that the basic principles of the Langmuir adsorption model are valid (i.e. the adsorption energy is the same for all adsorption sites, absence of interaction between the adsorbed molecules and monolayer coverage of the surface), then at equilibrium the rates of adsorption and desorption of the solute on the surface are equal, k ads ͑1 Ϫ ͒c eq ϭ k des (Eq. 4) where k ads and k des are the specific rate constants for adsorption and desorption, respectively, is the fraction of the crystal surface active growth sites occupied by the adsorbed molecules, and c eq is the equilibrium solution concentration of the additive. Thus, the interaction and the kinetic results may be interpreted by the Langmuir formalism. Therefore the growth rates depend on the surface coverage, , where R i and R 0 are the crystal growth rates in the presence and the absence of the additive. A combination of Equations 4 and 5 gives the following.
In Equation 6, k aff is the affinity constant (equal to k ads /k des ); it is a measure of the affinity of the adsorbate for the adsorbent. As may be seen in Fig. 15, a straight line was obtained for serglycin, suggesting the validity of the assumed model. The affinity constant, as determined from the slope of the linear plot of R 0 /(R 0 Ϫ R i ) against 1/c eq according to Equation 6, is 0.031 liter/mg.

DISCUSSION
We have shown for the first time that serglycin is the major proteoglycan synthesized and secreted by MM cell lines. To our knowledge, the presence of serglycin in MM has previously been demonstrated only through a Western blot of one mouse MM cell line (X63) (19), but they did not provide detailed evidence for the protein synthesis and its subcellular distribution in various MM cell lines. Here we show by several techniques, including isolation and detailed biochemical characterization, that serglycin is the major PG secreted by myeloma cells. Serglycin isolated from myeloma cell lines was larger than 250 kDa and carried exclusively CS side chains, as shown previously in other cell lines of the hematopoietic lineage. The CS chains consisted predominantly of 4-sulfated disaccharides (up to 93%) and, only to a limited extent, 6-sulfated and nonsulfated units. MM cell serglycin contained only trace amounts of disulfated disaccharides, in contrast to serglycin secreted by hematopoietic cells of mouse origin, which contains significant amounts of oversulfated disaccharides (19).
The combined FACS and confocal microscopy studies in nonpermeabilized MM cells demonstrated that serglycin binds to the cell surface in variable amounts. Although immunostaining has suggested that serglycin is present on the rat L2 yolk sac tumor cell membrane as well as within the cytoplasm (1), it seems unlikely that serglycin is a transmembrane PG, because there are no hydrophobic sequences in the mature protein. In our study, chondroitinase treatment greatly reduced the level of cell surface serglycin, indicating that intact GAG chains mediate serglycin-cell surface interaction. However, chondroitinase treatment did not completely remove serglycin from the cell surface. A possible explanation is that CS side chains, in interaction with other protein, are partially protected from enzymatic degradation. Thus, this finding could provide indirect evidence for a serglycin-interacting partner molecule on the cell surface. In U266 cells, serglycin was not present on the  cell surface but was abundant in the cytoplasm, as described below. Even after incubation of cells with purified serglycin, no serglycin was detected on the U266 surface (data not shown). This could suggest that U266 cells lack a necessary partner molecule for serglycin on the cell membrane.
Our immunostaining study of permeabilized cells revealed that serglycin is present intracellularly in MM cell lines. Although our aim has not been to characterize the intracellular serglycin distribution in detail, it was interesting to note that CAG cells demonstrated increased perinuclear staining, whereas in U266 cells serglycin was abundant in the entire cytoplasm. This could open the way for further studies of the exact distribution of serglycin in MM cells.
Serglycin is secreted constitutively by various cell types including three hematopoietic tumor cell lines, which were found to secrete 55-70% of their serglycin amounts under control conditions and to totally release serglycin when stimulated by phorbol myristate acetate (7,8). Serglycin is suggested to be an important molecule for the packaging of biologically active molecules such as proteases, cytokines, and growth factors. The interaction of serglycin with these molecules may be crucial for their stability and bioavailability after secretion. Serglycin may also act as stabilizer of ligand-receptor binding through its interaction with other cell surface molecules such as CD44, forming a more stable supramolecular complex.
Binding of serglycin on the myeloma cell surface as observed in our study may also suggest that serglycin is involved in myeloma cell adhesion on extracellular matrix containing fibronectin and collagen. Such adhesion of myeloma cells in the bone marrow is critical for tumor spread and survival. Serglycin is known to bind extracellular matrix proteins such as collagen type I and fibronectin, and these interactions may be responsible for the localization and immobilization of growth factors and cytokines in the extracellular milieu. Several studies have already indicated the importance of adequate concentrations of diffusible cytokines and growth factors by immobilized PGs in the regulation of cell proliferation and adhesion (42-44). Schick et al. (6) similarly propose that serglycin may interact with various growth factors and cytokines during embryogenesis and could be involved in the transmission of cytokine and growth factors signals from the inner cell mass to the trophectoderm or vice versa. Serglycin may also play important role as an intermediate molecule in homophilic or heterophilic cellcell aggregation in MM, because it is known to bind CD44 on the cell surface in a manner dependent both on GAG chains and core proteins promoting cell-cell aggregation (as described by Toyama-Sorimachi et al. (19)).
Winberg et al. (45) demonstrated that serglycin can be covalently linked to MMP-9 through one or more disulfide bridges in THP-1 cells. This association is suggested to be important for the transport, targeting, and regulation of the enzyme. In several cancer cell lines, MMP-9 has been found bound to the plasma membrane (46 -49). This interaction of MMP-9 with plasma membrane is poorly understood, although it has been shown that MMP-9 binds to the CD44 receptor on the cell surface (46,49). However, these studies did not show how MMP-9 links to the receptor. Abecassis et al. (50) demonstrated that RhoA induces MMP-9 expression in human mi-crovascular endothelial cells (HMEC-1). Confocal analyses revealed clustering of MMP-9 in advancing lamellipodia at the forefront of endothelial cells, where this proteinase colocalized with RhoA and CD44. It is interesting to note that endothelial cells synthesize and secrete as major CS-bearing PG serglycin (4). Therefore, the binding of MMP-9 to CD44 may be mediated through serglycin, which covalently links to MMP-9 and interact through its GAG chains with plasma-anchored CD44.
Our finding that all cell lines expressed such high levels of serglycin suggests that secretion of serglycin may be a common finding in MM. Immunohistochemical examinations confirmed that myeloma cells synthesize and secrete serglycin to the extracellular milieu. Serglycin is also found in benign plasma cells; however, they secrete only low amounts because of the relatively low number of such cells. We also performed a rough semiquantitative screening for the accumulation of serglycin in the bone marrow aspirates from a limited number of patients with newly diagnosed MM. Serglycin was found to be markedly elevated in 30% of patients as compared with control healthy volunteers. The observed variability of serglycin levels in bone marrow aspirates might be a consequence of varying amounts of infiltrating malignant plasma cells and tumor burden. These promising results tempted us to develop a more convenient and accurate methodology for the estimation of serglycin in bone marrow plasma or in blood serum, with a goal of examining the possible correlations of serglycin accumulation with disease progression and prognosis (in progress in our laboratory).
Recent evidence has demonstrated that MM cells affect bone homeostasis through suppression of new bone formation (29). Oshima et al. (30) demonstrated that myeloma plasma cells secrete soluble factors that inhibit osteoblast differentiation and mineralized nodule formation. A canonical Wingless-type (Wnt) signaling pathway soluble inhibitor secreted Frizzledrelated protein 2 (sFRP-2), which is constitutively synthesized and secreted by most MM cells and suppresses bone morphogenetic protein-2 (BMP-2)-induced osteoblast differentiation and mineralized nodule formation. It is well known that PGs are also involved in bone mineralization, acting either as promoters or inhibitors of HAP crystal growth. Our in vitro experiments demonstrate that serglycin is a potent inhibitor of HAP crystal growth and may be a key molecule in myeloma bone disease. Serglycin reduces the crystal growth rate of HAP due to its adsorption and further blocking of the active growth sites on the crystal surface. The crystallization kinetics were interpreted in terms of the Langmuir adsorption model. The apparent order of the crystallization reaction was found to be n ϭ 2, suggesting a surface diffusion controlled spiral growth mechanism. The kinetic affinity constant, which in this case may be regarded as an "inhibition" constant, obtained for serglycin (0.031 liter/mg) is comparable with that estimated previously for hyaluronan (0.078 liter/mg), a potent inhibitor of HAP crystal growth (34). This constant is much higher than that of the A1D1 (0.00074 liter/mg) and A1 (0.00329 liter/mg) ultracentrifugation fractions derived from guanidine extracts of chick sternal cartilage, which are much less effective in HAP crystal growth inhibition (33). Our results suggest that serglycin, constitutively secreted by myeloma plasma cells and accumulated NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 in the bone marrow, inhibits bone mineralization in sites of bone remodeling, further contributing to the imbalance of the bone resorption-bone formation process that occurs in patients with MM.