Download - Matrix glycoprotein SC1/ECM2 augments B lymphopoiesis

1997 90: 3404-3413

Nancy A. Jenkins, Yoshiaki Tomiyama, Yuji Matsuzawa and Paul W. KincadeKenji Oritani, Yuzuru Kanakura, Keisuke Aoyama, Takafumi Yokota, Neal G. Copeland, Debra J. Gilbert, Matrix Glycoprotein SC1/ECM2 Augments B Lymphopoiesis

http://bloodjournal.hematologylibrary.org/content/90/9/3404.full.htmlUpdated information and services can be found at:

(3057 articles)Hematopoiesis and Stem Cells Articles on similar topics can be found in the following Blood collections

http://bloodjournal.hematologylibrary.org/site/misc/rights.xhtml#repub_requestsInformation about reproducing this article in parts or in its entirety may be found online at:

http://bloodjournal.hematologylibrary.org/site/misc/rights.xhtml#reprintsInformation about ordering reprints may be found online at:

http://bloodjournal.hematologylibrary.org/site/subscriptions/index.xhtmlInformation about subscriptions and ASH membership may be found online at:

Copyright 2011 by The American Society of Hematology; all rights reserved.20036.the American Society of Hematology, 2021 L St, NW, Suite 900, Washington DC Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by

only.For personal use at OSAKA U SEIMEIKAGAKU on September 24, 2012. bloodjournal.hematologylibrary.orgFrom

Matrix Glycoprotein SC1/ECM2 Augments B LymphopoiesisBy Kenji Oritani, Yuzuru Kanakura, Keisuke Aoyama, Takafumi Yokota, Neal G. Copeland, Debra J. Gilbert,

Nancy A. Jenkins, Yoshiaki Tomiyama, Yuji Matsuzawa, and Paul W. Kincade

The extracellular matrix produced by stromal cells plays a the full length of SC1/ECM2 and Ig was found to recognizecritical role in lympho-hematopoiesis. It was recently discov- pre-B cells in a divalent cation-dependent manner, and toered that matrix glycoprotein SC1/ECM2 is a component of augment mitogen-dependent proliferation of mature B cells,that matrix and preliminary evidence suggested that it could as well as the cloning of pre-B cells, but to have no influencecontribute to the nurturing environment for B-lymphocyte on myeloid progenitor cells. Although SC1/ECM2 is normallyprecursors. A fusion protein prepared from the amino termi- a secreted protein, we show that it is also capable of aug-nal portion of SC1/ECM2 and the constant region of human menting lymphopoiesis when expressed as a transmem-Ig preferentially bound to pre-B cells. Furthermore, the clon- brane protein on fibroblasts. Although the C-terminal por-ing efficiency of interleukin-7–dependent B-cell precursors tion of SC1/ECM2 has sequence homology to osteonectin/was increased in a dose-dependent manner by addition of SPARC, the unique N-terminal one fifth of the protein wasthis fusion protein. We now report the complete cDNA se- sufficient to augment lymphocyte growth.quence for murine SC1/ECM2 and its localization to the cen- q 1997 by The American Society of Hematology.tral region of chromosome 5. A fusion protein prepared from

E to establish bioassay systems specific to each molecule. Werecently developed a new cloning strategy that is capable

XTRACELLULAR matrix molecules within bone mar-row play a fundamental role in lympho-hematopoie-

sis.1-6 They include various collagens, proteoglycans, and of identifying cell-surface or soluble factors that recognizehematopoietic cells.12 With this approach we identified sev-glycoproteins. This matrix functions to retain hematopoietic

cells within bone marrow (BM), to localize growth factors eral stromal cell-derived matrix components whose N-termi-nal portions could bind to pre-B cells. One of them wasaround cells, and to give signals for proliferation or differen-

tiation to hematopoietic precursors. For example, fibronectin, matrix glycoprotein SC1/ECM2. SC1/ECM2 cDNA wasoriginally isolated from a rat brain expression library usingcollagens, and laminin are ligands of integrins that are signal

inducers as well as adhesion molecules.3,5 Hyaluronan is a polyclonal antibodies directed against synaptic junction gly-coproteins.13 The deduced amino acid sequence suggestedligand of CD44, and in several systems facilitates cell-cell

adhesion and cell migration.4 Such molecules may be im- that SC1/ECM2 is a secreted and calcium-binding protein,and it is considered to be a matrix component in brain. Aportant in lympho-hematopoiesis, because antibodies to

VLA-4 or CD44 completely blocked the production of human homolog of SC1/ECM2 was isolated by means of amonoclonal antibody (MoAb) directed to high endotheliallymphoid and myeloid cells in long-term BM cultures.7,8

Proteoglycans also influence the differentiation of hemato- venule cells.14 It is associated with basal, lateral, and apicalsurfaces of endothelial venule cells and was shown to bepoietic cells9 and this may be attributed in part to their ability

to immobilize growth factors such as granulocyte-macro- anti-adhesive for endothelial cells on fibronectin-coated sur-faces.15 Our studies showed that SC1/ECM2 is also ex-phage colony-stimulating factor (GM-CSF), interleukin-3

(IL-3), and IL-7.10,11 pressed by stromal cells and that a fusion protein containingthe amino terminal portion of SC1/ECM2 is recognized byHowever, the complexity of the matrix provides a major

obstacle for detailed functional studies because it is difficult pre-B cells. Furthermore, the preliminary analysis suggestedthat SC1/ECM2 might represent a matrix protein in BM thatis important for lymphopoiesis. This possibility has now

From the Oklahoma Medical Research Foundation, Oklahoma been explored with a more detailed investigation of SC1/City; the Departments of Internal Medicine II and Hematology/ ECM2 structure and function.Oncology, Osaka University Medical School, Osaka, Japan; andMammalian Genetics Laboratory, ABL-Basic Research Program,

MATERIALS AND METHODSNCI-Frederick Cancer Research and Development Center, Freder-ick, MD. Cells. 293T, a human renal cell carcinoma cell line, was kindly

provided by Dr T. Hirano (Osaka University, Osaka, Japan) andSubmitted January 14, 1997; accepted June 17, 1997.Supported in part by the National Cancer Institute, Department maintained in Dulbecco’s modified Eagle’s medium (DMEM) sup-

plemented with 10% fetal calf serum (FCS; GIBCO, Grand Island,of Health and Human Services, under contract with Advanced Bio-Science Laboratories, Inc. Supported in part by grants from the NY). DW34, a mouse stromal cell–dependent pre-B cell line, was

kindly provided by Dr S. Nishikawa (Kyoto University, Kyoto, Ja-Ministry of Education, Science and Culture, the Inamori Foundation,Senri Life Science Foundation, and Mochida Memorial Foundation pan) and maintained on a mouse stromal cell line, P5, in RPMI1640

medium supplemented with 5% FCS and 5 1 1005 mol/L 2-mercap-as well as Grant No. AI-33085 from the National Institutes of Health.Address reprint requests to Kenji Oritani, MD, The Second De- toethanol.16 Mouse stromal cell clones (OP42, BMS2, and NIH3T3),

a mouse fibroblast cell line (L), mouse pre-B cell clones (1A9,partment of Internal Medicine, Osaka University Medical School,2-2 Yamadaoka, Suita, Osaka 565, Japan. BCB8, BCB9, and BCB10), mouse lymphoma cell lines (7OZ/3,

WEHI231, WEHI279, AKR1, and EL4), a mouse erythroid cell lineThe publication costs of this article were defrayed in part by pagecharge payment. This article must therefore be hereby marked (GM86), and a mouse myelomonocytic leukemia cell line (WEHI3)

were maintained as previously described.17,18‘‘advertisement’’ in accordance with 18 U.S.C. section 1734 solely toindicate this fact. Cloning of full-length SC1/ECM2 cDNA. The isolation and se-

quencing of a 5*-portion cDNA of murine SC1/ECM2 was describedq 1997 by The American Society of Hematology.0006-4971/97/9009-0029$3.00/0 in our previous report.12 Sense and antisense primers, 5*-ATCCAG-

3404 Blood, Vol 90, No 9 (November 1), 1997: pp 3404-3413

AID Blood 0034 / 5h40$$$661 10-03-97 11:47:36 bldas WBS: Blood


SC1/ECM2 AUGMENTS B LYMPHOPOIESIS 3405

Fig 1. Amino acid sequenceof murine homolog of SC1/ECM2. The box marks a potentialsignal peptide. Potential N-gly-cosylation sites are indicatedwith circles. The osteonectin-like sequence is underlined withdashes, and the EF-hand Ca-binding site is underlined with abold line.

CCACCTCTCCGCAGATCT-3* and 5*-CTTCAGGTCCACCTC- kit (Boehringer Mannheim); washing was done to a final stringencyof 1.01 SSCP (phosphate-buffered sodium chloride sodium citrate),GAAGCTGTA-3*, were designed from this sequence to perform

polymerase chain reactions (PCR). Because BMS2 stromal cells 0.1% sodium dodecyl sulfate (SDS), 657C. A major fragment of 7.4kb was detected in Sca I–digested C57BL/6J DNA and a majorexpressed this gene, a cDNA library was made from mRNA of

BMS2 cells using oligo dT and a TimeSaver cDNA synthesis kit fragment of 9.0 kb was detected in Sca I–digested M spretus DNA.The presence or absence of the 9.0-kb Sca I M spretus–specific(Pharmacia, Uppsala, Sweden) and ligated into pEFBOS vector. Sus-

pensions of pools containing several thousand colonies were sub- fragment was followed in backcross mice. A description of theprobes and RFLPs for the loci linked to SCI/ECM2 including Fgf5,jected to PCR reactions with the above two primers. After 35 cycles

of PCR (947C for 1 minute, 557C for 2 minutes, 727C or 3 minutes), Gfil, and Adrbk2 has been reported previously.21,22 Recombinationdistances were calculated as described23 using the computer programPCR samples were separated in 1.5% agarose gels and screened with

a specifically amplified band (307 bp). A positive pool was divided SPRETUS MADNESS. Gene order was determined by minimizingthe number of recombination events required to explain the alleleinto progressively smaller pools and rescreened until a single clone

was isolated. The nucleotide sequence of the insert was then deter- distribution patterns.Fusion proteins. An Ig/pEFBOS vector was used to produce Igmined using an automated DNA sequencer (Aplied Biosystems, Fos-

ter City, CA). fusion proteins.12 This Ig cassette was mutated to diminish Fc-medi-ated bindings.24 An HT/pEFBOS vector was used to produce fusionNorthern blot analysis. Total RNAs from various cell lines were

isolated using TRIzol Reagent (GIBCO), electrophoresed through a proteins composed of an HPC4 epitope tag and a transmembranedomain of tissue factor.12 The full-length cDNA of SC1/ECM2 wasformaldehyde agarose gel, and transferred onto a nylon membrane

(Amersham, Arlington Heights, IL). A blot containing poly(A)/ amplified by PCR with 5*-CCCGCGGCCGCCACCTCTCCGCAG-ATCTAGCCAGC-3* and 5*-CCCCTCGAGAAGAGGTTTTCA-RNAs from adult mouse tissues was acquired from Clontech (Palo

Alto, CA). The SC1/ECM2– or b-actin–containing cDNA frag- TCTATATCCTC-3*. PCR samples were digested with Not I andXho I, and ligated into the Not I and Xho I sites of Ig/pEFBOS andments were labeled with [32P]dCTP using a random primed DNA

labeling kit (Boehringer Mannheim, Indianapolis, IN) and hybridized HT/pEFBOS vectors (F-SC1-Ig/pEFBOS and F-SC1-HT/pEFBOS).In some experiments, an N-SC1-HT/pEFBOS plasmid was used toto the membrane overnight. Blots were then washed and autoradio-

graphed. produce fusion proteins containing N-terminal portion of SC1/ECM2(93 amino acids). All constructs were confirmed by sequencing.Interspecific backcross mapping. Interspecific backcross prog-

eny were generated by matching (C57BL/6J 1 Mus spretus) F1 Immunoprecipitation and immunoblotting. Supernatants of cul-tures of 293T cells transfected with CD44- or F-SC1-Ig/pEFBOSfemales and C57BL/6J males as described.19 A total of 205 N2 mice

were used to map the SC1/ECM2 locus (see text for details). DNA plasmids were immunoprecipitated with proteinG-sepharose (Phar-macia). Western blot was performed according to our previouslyisolation, restriction enzyme digestion, agarose gel electrophoresis,

Southern blot transfer, and hybridization were performed essentially reported methods using horseradish peroxidase–conjugated goat an-tihuman Ig (Southern Biotechnology Associates, Birmingham, AL)as described.20 All blots were prepared with Hybond-N/ nylon mem-

brane (Amersham). The probe, anÇ2.7-kb Xba I fragment of mouse and the ECL chemiluminescence detection system (Amersham).7Flow cytometry analysis. Antibody incubation and washing stepscDNA, was labeled with [32P] dCTP using a nick-translation labeling



ORITANI ET AL3406

Fig 2. Amino acid sequence homology among SC1/ECM2, osteonectin/SPARC, and QR1. Periods indicate gaps introduced to align se-quences. Boxes with dashed lines represent amino acid identities. Boxes with solid lines represent positions of cysteine residues.

were performed at room temperature in Hanks’ balanced salt solutions ated in semisolid agar containing 25 mg lipopolysaccharide (LPS).The granulocyte-macrophage progenitor assay (CFU-c) was performed(GIBCO) containing 1% bovine serum albumin (Sigma, St Louis, MO)

and 0.02% sodium azide. Cells were stained with supernatants of cul- with 25 mL of 10-fold concentrated WEHI3-conditioned medium.CD44- and F-SC1-Ig fusion proteins were purified with protein Atures of 293T cells transfected with the F-SC1-Ig/pEFBOS plasmid.

Fluorescein isothionate (FITC)-conjugated goat antihuman IgG (South- columns (Pierce, Rockford, IL) and added to cultures at the indicatedconcentrations. All colony assays were performed in 35-mm dishesern Biotechnology Associates) was used as a second antibody. Superna-

tants containing soluble CD44-Ig were used for negative controls. In and incubated at 377C, 5% CO2. Colonies were scored on day 6.Cell adhesion assay. NIH3T3 cells that overexpressed N-termi-some experiments, fresh or cultured BM cells were stained with FITC-

conjugated antimouse Mac1 (Boehringer Mannheim) and phycoerythrin nal or full length of SC1/ECM2 were obtained by stable transfectionwith N-SC1-HT/pEFBOS or F-SC1-HT/pEFBOS plasmids, respec-(PE)-conjugated anti-mouse CD45RA (Pharmingen, San Diego, CA).

Surface expression of fusion proteins composed of SC1/ECM2, an tively. Each NIH3T3 clone was plated at 105 per well in 24-wellplates and allowed to grow for 2 days before adhesion assays. BCB10HPC4 epitope tag, and a transmembrane domain of tissue factor was

evaluated by the stainings with an HPC4 antibody.25 The surface immu- cells were suspended in Hanks’ balanced salt solution supplementedwith 3% FCS and adjusted to a concentration of 106/mL beforenofluorescence was evaluated by a FACScan flow cytometer (Becton

Dickinson, Mountain View, CA). being applied to NIH3T3 monolayers. After a 30-minute coculture,unbounded cells were removed by three time washes. Before eachColony-forming cell assays. Mouse BM cells were prepared and

suspended in 1 mL of assay medium as previously described.26 The aspiration, plates were agitated on a vortex mixture. Bound cellswere then recovered and counted. The percentage of bound cellssemisolid agar colony-forming unit assay for B-lymphocyte precursors

(CFU-IL7) was performed with 1 ng recombinant murine IL-7 (R&D was determined as follows: Percent Bound Å 100 1 No. of BoundCells/No. of Applied to Each Well.Systems, Minneapolis, MN). Clonable B cells (CFU-B) were enumer-




(Fig 1, Genbank/EMBL accession number AB000127). Theinsert cDNA sequence was 2,730 bp and contained an openreading frame that encoded 650 amino acids. The proteinpossesses a highly hydrophobic stretch of 17 amino acids at

Fig 3. Expression of SC1/ECM2 in various cells and tissues. (Aand C) Total RNA (15 mg/lane) and (B) Poly(A)" RNA (2 mg/lane)were isolated from the indicated cell lines and tissues, and subjectedto Northern blot analysis. The lower panel is a control for equalloading where the same blot was probed with b-actin.

Coculture of BM cells and fibroblasts. A BM cell cocultureassay was performed as previously described with minor modifica-tions.27 Each stably transfected NIH3T3 clone was plated on a 24-well plate at subconfluent density. BM cells (2 1 104/well) werethen added to cultures in the presence of 1 ng/mL IL-7. NIH3T3cells as well as nonadherent populations were obtained from each Fig 4. SC1/ECM2 maps in the central region of mouse chromo-well on day 5. Viable BM-derived cells were easily distinguished some 5. SC1/ECM2 was placed on mouse chromosome 5 by interspe-from NIH3T3 cells by size and were counted by trypan blue dye cific backcross analysis. The segregation patterns of SC1/ECM2 andexclusion. flanking genes in 94 backcross animals that were typed for all loci

are shown at the top of the figure. For individual pairs of loci, morethan 94 animals were typed (see text). Each column represents theRESULTSchromosome identified in the backcross progeny that was inherited

Isolation of murine homolog SC1/ECM2 cDNA and pri- from the (C57BL/6J Ì M spretus) F1 parent. The shaded boxes repre-sent the presence of a C57BL/6J allele and white boxes representmary structure of the predicted protein. Although our ini-the presence of a M spretus allele. The number of offspring inheritingtial studies suggested that SC1/ECM2 might be importanteach type of chromosome is listed at the bottom of each column. Afor support of B lymphopoiesis, functional experiments werepartial chromosome 5 linkage map showing the location of SC1/only conducted with the amino terminal portion of the mole- ECM2 in relation to linked genes in shown at the bottom of the figure.

cule expressed as an Ig fusion protein. Furthermore, a com- Recombination distances between loci in centiMorgans are shownto the left of the chromosome, and the position of loci in humanplete sequence of the murine homologue of SC1/ECM2 hadchromosomes, where known, are shown to the right. References fornot been previously described. Therefore, we determined thethe human map positions of loci cited in this study can be obtainedfull-length cDNA sequence by screening a BMS2 library by from GDB (Genome Data Base), a computerized database of human

PCR with sense and antisense primers designed from the linkage information maintained by The William H. Welch MedicalLibrary of The Johns Hopkins University (Baltimore, MD).original 5*-portion cDNA sequence of murine SC1/ECM212



ORITANI ET AL3408

Fig 5. A full length of F-SC1-Ig fusion protein recognizes the BCB10 pre-B cell line. Either the F-SC1- or CD44-Ig/pEFBOS plasmid wastransfected into 293T cells. (A) Supernatants from each transfected sample were immunoprecipitated with proteinG-Sepharose. Western blotwas perfomed by using horseradish peroxidase–conjugated goat antihuman Ig and the ECL chemiluminescence detection system. Sampleswere evaluated under reducing conditions. (B) BCB10 cells were stained with supernatants from each transfected sample, followed by FITC-goat antihuman IgG (shaded histograms). Hanks’ solution with (right) or without (left) 5 mmol/L EDTA was used as washing buffer. Negativecontrol stainings obtained with CD44-Ig are also shown (open histograms).

the extreme amino-terminal end that is appropriate for a ble exception and it contained SC1/ECM2 transcripts onlyin the mouse.signal peptide,28 but lacks an internal hydrophobic membrane

spanning domain (Fig 1). There are four potential N-glyco- Chromosomal mapping of the murine SC1/ECM2 gene.The mouse chromosomal location of SC1/ECM2 was deter-sylation sites at positions 29, 148, 168, and 462. In addition,

an EF hand Ca-binding domain is present near the carboxy mined by interspecific backcross analysis using progeny de-rived from matings of [(C57BL/6J 1 M spretus)F1 1terminal end (DPNKDKHITLKEW), suggesting that SC1/

ECM2 may bind calcium.29 The deduced amino acid se- C57BL/6J] mice. This interspecific backcross mapping panelhas been typed for over 2,300 loci that are well distributedquence of murine SC1/ECM2 showed high homology with

rat SC1/ECM2 (86.2% identity), and 232 carboxy-terminal among all the autosomes as well as the X chromosome.19C57BL/6J and M spretus DNAs were digested with severalamino acids (amino acid 419-650) showed especially high

homology with osteonectin (62.2%) and QR1 (72.5%). All enzymes and analyzed by Southern blot hybridization forinformative restriction fragment length polymorphisms14 of the cysteines in this area were completely conserved

among these three molecules30,31 (Fig 2). (RFLPs) using a mouse cDNA SC1/ECM2 probe. The 9.0-kb Sca I M spretus RFLP (see Materials and Methods) wasCell and tissue distribution of SC1/ECM2. Because of

very limited information available concerning SC1/ECM2 used to follow the segregation of the SC1/ECM2 locus inbackcross mice. The mapping results indicated that SC1/expression in rodents,13 we conducted Northern blot analyses

of mRNA obtained from panels of cell lines and tissues (Fig ECM2 is located in the central region of mouse chromosome5 linked to Fgf5, Gfi1, and Adrbk2. Although 94 mice were3). This showed that SC1/ECM2 was expressed by two BM-

derived stromal cell lines (BMS2 and P5), as well as a splenic analyzed for every marker and are shown in the segregationanalysis (Fig 4), up to 156 mice were typed for some pairsstromal cell clone (OP42) and a fibroblast cell line

(NIH3T3). In contrast, SC1/ECM2 transcripts were not de- of markers. Each locus was analyzed in pairwise combina-tions for recombination frequencies using the additional data.tectable in any lympho-hematopoietic cell lines (1A9,7OZ/

3, WEHI231, EL4, and WEHI3). SC1/ECM2 was widely The ratios of the total number of mice exhibiting recombi-nant chromosomes to the total number of mice analyzed forexpressed in various tissues including heart, brain, spleen,

lung, muscle, kidney, and testis. Although SC1/ECM2 tran- each pair of loci and the most likely gene order are: centro-mere- Fgf5 - 3/156 - SC1/ECM2 - 0/135 - Gfi1 - 2/121 -scripts were particularly abundant in heart, brain, and lung,

expression was not detected in liver. This pattern of expres- Adrbk2. The recombination frequencies (expressed as ge-netic distances in centiMorgans [cM]) { the standard error)sion is similar in most respects to that described for a human

SC1/ECM2 homolog.14 The kidney represents the most nota- are - Fgf5 - 1.9 //0 1.1 - [SC1/ECM2,Gfi1] - 1.7 //0 1.2 -




Adrbk2. No recombinants were detected between SC1/ECM2and Gfi1 in 135 animals typed in common, suggesting thatthe two loci are within 2.2 cM of each other (upper 95%confidence limit). Therefore, SC1/ECM2 maps to the centralregion of mouse chromosome 5. As summarized in Fig 4,this shares homology with human chromosomes 4q, 1p, and22q. In particular, Gfi1 has been placed on human 1p22. Thetight linkage between SC1/ECM2 andGfi1 in mouse suggeststhat SC1/ECM2 will reside on 1p as well.Recognition of BCB10 cells by SC1/ECM2. We cloned

the N-terminus of SC1/ECM2 on the basis of its ability tobind to the BCB10 pre-B cell line.12 To evaluate whetherthe native form of SC1/ECM2 also recognizes pre-B cells,we made a fusion protein composed of the full length ofSC1/ECM2 and a human IgG1 cassette (F-SC1-Ig). The pro-tein was confirmed by Western blot, and the molecularweight of this fusion protein was approximately 150 kD (Fig5A). As shown in Fig 5B, F-SC1-Ig could bind to BCB10cells, when the cells were stained in Hanks’ solutions thatcontained Ca2/ and Mg2/. Because this staining was com-pletely blocked by addition of EDTA, divalent cations facili-tated recognition of BCB10 cells by F-SC1-Ig. Because inte-grins require divalent cations for binding to their ligands,some integrins could be the receptors for SC1/ECM2 on pre-B cells. However, monoclonal antibodies to a4 (PS/2), a5(MFR5), aL (FD441.8), b1 (9EG7), or b2 (M18/2) did notblock the bindings of F-SC1-Ig to BCB10 cells (data notshown). Therefore, the receptor for SC1/ECM2 was not at-tributed to any of these integrins.Influence of SC1/ECM2 on production of B-lineage lym-

phocytes. Semisolid agar cloning assays were used to eval-uate the direct influence of SC1/ECM2 on lympho-hemato-poiesis. As shown in Fig 6, the cloning efficiency of IL-7responding pre-B cells (CFU–IL-7) was increased in a dose-dependent manner by addition of full length of F-SC1-Ig.Mature mitogen responsive B cells (CFU-B) were even moredramatically influenced and clonal proliferation was in-creased approximately 100%. In contrast, F-SC1-Ig had noinfluence on the responsiveness of myeloid progenitor cells(CFU-c) to colony-stimulating factors and CD44-Ig fusionprotein had no significant effect on any of these cultures Fig 6. The effects of F-SC1-Ig on colony assays. A F-SC1-Ig/

pEFBOS plasmid or a CD44-Ig/pEFBOS plasmid was transfected intoand served as a negative control. F-SC1-Ig did not elicit293T cells. The F-SC1-Ig and the CD44-Ig were purified with a proteinlymphocyte colony formation when cultures were estab-A column and added to (A) CFU-IL7, (B) CFU-B, and (C) CFU-c colonylished in the absence of IL-7 or mitogen (data not shown). assays of mouse BM cells at the indicated concentrations. Results

Therefore, SC1/ECM2 augments survival and/or prolifera- represent mean numbers of colonies per 105 BM cells Ô SD (n ! 3).Statistically significant differences from control (CD44-Ig) values aretion of B-lineage lymphocytes and their precursors, but hasindicated by one (P Ú .05) or two (P Ú .01) asterisks. Similar resultsno obvious influence on myeloid lineage cells.were observed in three independent experiments.A membrane-anchored form of SC1/ECM2 augments lym-

phopoiesis. SC1/ECM2 is normally a secreted protein butmight be immobilized in tissues through interaction withtransmembrane or extracellular matrix molecules. Therefore, or morphology of the fibroblasts. In addition, overexpressionwe investigated whether an immobilized form of the mole- of membrane-anchored N-terminal portion or full length ofcule would have functional activity. NIH3T3 fibroblasts that SC1/ECM2 on NIH3T3 cells did not change adhesion tonormally express the soluble form were transfected with N- BCB10 cells (Table 1). However, the transfected fibroblastsSC1- or F-SC1-HT/pEFBOS plasmids. This resulted in large had increased ability to support growth of pre-B cells. Prolif-quantities of the protein localized to the cell surface, where erating foci appeared within 5 days when murine BM wasit could be detected by means of the HPC4 MoAb (Fig 7). cultured on control fibroblasts in the presence of IL-7. The

recovered cells were determined to be B-lymphocyte lineageThere was no obvious influence of this treatment on growth



ORITANI ET AL3410

Fig 7. Expession of membrane-anchored SC1/ECM2 on stably transfected NIH3T3 cells. The N- or F-SC1-HT/pEFBOS plasmid or HT/pEFBOSvector was stably transfected into NIH3T3 cells. Each clone was stained with an HPC4 antibody. Shaded histograms depict stainings obtainedwith the HPC4 antibody, and open histograms represent results obtained with the second antibody alone.

by flow cytometry (presence of CD45RA and absence of cells and their progeny ex vivo and substantial progress hasbeen made in identifying transmembrane, secreted, and extra-Mac1, Fig 8). Significantly more lymphocytes were recov-

ered from cocultures initiated with NIH3T3 cells that ex- cellular matrix molecules that could be important. Of particu-lar interest are those which promote survival of blood cellpressed immobilized SC1/ECM2 (Fig 9A). This response

was similar when the full length or just the amino terminal precursors, and a preliminary study suggested that SC1/ECM2might represent such a protein.12 We now describe the com-portions of SC1/ECM2 were present. It would have been

possible in this model for immobilized SC1/ECM2 to influ- plete coding sequence of the murine SC1/ECM2 gene andlocalize it to chromosome 5. The amino terminal portion ofence lymphocyte precursors indirectly via some other cell

type present in the initial marrow inoculum. Therefore, co- the molecule was sufficient for recognition of lymphocytescultures were also performed with the stromal cell dependent and it promoted clonal proliferation of mature, as well asDW34 pre-B cell line.16 Survival and expansion of DW34 immature cells of the B lineage. Furthermore, SC1/ECM2cells was again much more efficient on transfected fibro- augmented B lymphopoiesis when present in either solubleblasts (Fig 9B). Therefore, an immobilized form of this ma- and immobilized forms. Detailed information is not availabletrix glycoprotein can deliver signals in addition to those about mechanisms through which SC1/ECM2 has this influ-provided by the constitutively produced soluble SC1/ECM2 ence, but structural similarity to the well-studied osteonectin/made by this fibroblast cell line. SPARC protein provides some basis for speculation.

The cDNA sequence of SC1/ECM2 (Fig 2) is well con-DISCUSSION served between mice and rats (86.2% identity at the amino

acid level), and especially in the C-terminal portion (98.3%A thorough understanding of the marrow microenviron-identity in a stretch of 232 amino acids). This C-terminalment should eventually make it possible to propagate stemportion of SC1/ECM2 also has high homology with osteo-nectin, QR1, testican and transforming growth factor-b–in-Table 1. Overexpression of SC1/ECM2 Does Not Influenceduced protein TSC36/glioma-secreted follistatin-related pro-Adhesion Between Stromal and Pre-B Cellstein, and all of 14 cysteine residues were completelyTransfectant Cell Adhesion (%)conserved among these molecules.30-34 This part of osteonec-

NIH3T3 HT/pEFBOS clone 1 13.8 { 0.50 tin consists of three domains, a follistatin-like domain, aNIH3T3 HT/pEFBOS clone 2 12.9 { 0.61

high a-helix content, and an EF-domain.35 The EF region isNIH3T3 N-SC1-HT/pEFBOS clone 1 13.6 { 0.68an autonomously folding and crystalliable domain that bindsNIH3T3 N-SC1-HT/pEFBOS clone 2 13.3 { 0.60calcium and collagen IV.36NIH3T3 F-SC1-HT/pEFBOS clone 1 14.1 { 0.61The high degree of sequence homology in the C-terminalNIH3T3 F-SC1-HT/pEFBOS clone 2 12.5 { 0.80

portion of SC1/ECM2 suggests that the protein may haveOne representative experiment is shown where BCB10 cells, a pre-Bmany functions in common with osteonectin/SPARC. Forcell line, was added to wells containing NIH3T3 cells stably transfectedexample, it might be expected to be anti-adhesive in somewith HT/pEFBOS, N-SC1-HT/pEFBOS, or F-SC1-HT/pEFBOS plasmid.

Similar results were obtained in two independent experiments. circumstances, to regulate cell proliferation, and to bind cy-




Fig 8. Surface phenotypes offreshly isolated or cultured BMcells. Freshly isolated mouse BMcells or BM cells cocultured onNIH3T3 cells were subjected totwo-color flow cytometry analy-sis. Myeloid cells were simulta-neously detected by anti-Mac1antibody, and B-lineage cellswere stained for a CD45RAmarker. Quadrants are indicatedto show levels of backgroundstainings observed with appro-priate irrelevant control antibod-ies.

tokines.37 However, the amino terminal portion of SC1/ is that SC1/ECM2 binds and presents this essential chemo-kine to developing and activated B lymphocytes.ECM2 had similarity to neither osteonectin nor QR1 and

might be expected to mediate unique functions. Indeed, ap- The DW34 lymphocyte clone was also used as a biologicalindicator to identify the stromal cell-surface ectoenzymeproximately one fifth of the molecule (93 amino acids at N-

terminus) was sufficient for recognition by lymphoid cells known as BST1.44 Although the function of BST1 has notbeen fully characterized, BM stromal cell lines derived fromand augmentation of their expansion. Interactions between

SC1/ECM2 and lymphocytes are divalent cation depen- patients with multiple myeloma and rheumatoid arthritis hadgreatly elevated expression of BST1. It will be interestingdent,12 but the calcium-binding EF domain is localized near

the C-terminus and need not to be involved. Divalent cations to investigate SC1/ECM2 production in such conditions be-cause abnormalities of BM stromal cells have been reportedmay therefore be important for function of the counter-recep-

tor for SC1/ECM2 on lymphocytes. in some diseases with monoclonal or polyclonal B-cell pro-liferation, and the BM microenvironment plays importantOsteonectin/SPARC is known to bind cytokines such as

platelet-derived growth factor (PDGF).38 Therefore, it will roles in the pathogenesis of those diseases.45Although we originally isolated SC1/ECM2 on the basisbe important to learn if this is true for SC1/ECM2 and which

domains are required. It is interesting that growth of the of its recognition of pre-B cells, this matrix molecule mayalso influence activities of mature B lymphocytes. NorthernDW34 pre-B cell clone was augmented by both full-length

and truncated forms of SC1/ECM2. This clone has pre- blot analysis of spleen did not show high SC1/ECM2 expres-sion, but the signal could have been substantially dilutedviously been shown to be responsive to a chemokine known

as PBSF/ SDF-1, which is the natural ligand of fusin.39-42 by lymphocyte-derived mRNA, that does not contain SC1/ECM2 transcripts. A monoclonal antibody to murine SC1/Lymphocyte development is disrupted in mice that have been

targeted for the PBSF gene.43 One possibility to be explored ECM2 will be useful in identifying positive cells in periph-

Fig 9. Augmentation of pre-B cell growth by SC1/ECM2 overexpressed on the surface of stromal cells.An N-SC1-HPC4-TF/pEFBOS plasmid or an F-SC1-HPC4-TF/pEFBOS plasmid was stably transfectedinto NIH3T3 cells. (A) BM cells (2 Ì 104) or (B) a pre-B cell line, DW34 (1 Ì 103), were cocultured in thepresence of 1 ng/mL IL-7. Each dot represents meansof recovered nonadherent cells from cocultures witheach transfected clone. Statistically significant differ-ences from control (cocultures with NIH3T3 trans-fected with HPC4-TF/pEFBOS plasmid) values are in-dicated by one (P Ú .05) or two (P Ú .01) asterisks.Similar results were observed in four independentexperiments.



ORITANI ET AL3412

2. Greenberger JS: The hematopoietic microenvironment. Criteral lymphoid tissues inasmuch as the human homolog isRev Oncol Hematol 11:65, 1991relatively restricted to endothelial cells within the tonsil.143. Long MW: Blood cell cytoadhesion molecules. Exp HematolFunctions have yet to be demonstrated for SC1/ECM2 in

20:288, 1992that tissue and possible roles should be investigated in heart,4. Kincade PW: Cell adhesion mechanisms utilized for lympho-brain, and lung, because these tissues have particularly high poiesis, in Shimizu Y (ed): Lymphocyte Adhesion Molecules. Aus-levels of expression. tin, TX, Landes RG, 1993, p 249

Collagens, laminin, fibronectin, hyaluronan, proteogly- 5. Verfaillie C, Hurley R, Bhatia R, McCarthy JB: Role of bonecans, and a variety of glycoproteins are present within the marrow matrix in normal and abnormal hematopoiesis. Crit Revextracellular matrix of BM and there is reason to believe Oncol Hematol 16:201, 1994

6. Klein G: The extracellular matrix of the hematopoietic micro-that this is much more than an inert framework.3,5,6,46 Inenvironment. Experimenta 51:914, 1995addition to the ability to immobilize growth factors, the ma-7. Miyake K, Weissman IL, Greenberger JS, Kincade PW: Evi-trix may deliver signals for survival and expansion, as ap-

dence for a role of the integrin VLA-4 in lympho-hemopoiesis. Jpears to be the case with SC1/ECM2. The possibility willExp Med 173:599, 1991need to be investigated whether SC1/ECM2 delivers signals8. Miyake K, Medina K, Hayashi S, Ono S, Hamaoka T, Kincaderelated to differentiation and if it influences lymphocyte mi- PW: Monoclonal antibodies to Pgp-1/CD44 block lympho-hemopoi-gration. Changes in transfected fibroblast morphology were esis in long term bone marrow cultures. J Exp Med 171:477, 1990

not obvious. Moreover, the amounts of F-SC1-Ig fusion pro- 9. Luikart SD, Kenny PM, Oegema TR: Human bone marrowteins that bind to B-lineage lymphocytes are not substantial, heparan sulfate induces leukemia cell differentiation. Connect Tissueand overexpression of SC1/ECM2 on a fibroblast did not Res 31:99, 1995

10. Gordon MY, Riley SM, Watt SM, Greaves MF: Compartmen-change adhesion to a pre-B cell line (Fig 5 and Table 1).talization of a haematopoietic growth factor (GM-CSF) by glycos-Therefore, SC1/ECM2 may not function as an importantaminoglycans in the bone marrow microenvironment. Natureadhesion molecule and it will be interesting to learn if it can326:403, 1987be antiadhesive with respect to lymphocytes, as it has been11. Roberts R, Gallagher J, Spooncer E, Allen TD, Bloomfieldshown to be for endothelial cells.15 Our findings showed

F, Dexter TM: Heparan sulphate bound growth factors: A mecha-that, in contrast to B-lineage precursors, growth of myeloid nism for stromal cell mediated haemopoiesis. Nature 332:376, 1988progenitors from BM was not influenced by SC1/ECM2. 12. Oritani K, Kincade PW: Identification of stromal cell productsThis does not exclude possible roles in myeloid cell matura- which interact with pre-B cells. J Cell Biol 134:771, 1996tion, and it is still possible that SC1/ECM2 might interact 13. Johnston IG, Paladino T, Gurd JW, Brown IR: Molecularwith megakaryocytic and erythroid progenitors, as well as cloning of sc1/ecm2: A putative brain extracellular matrix glycopro-

tein showing partial similarity to osteonectin/BM40/SPARC. Neuronstem cells in that organ.2:165, 1990We compared our interspecific map of chromosome 5 with14. Girard JP, Springer TA: Cloning from purified high endothe-a composite mouse linkage map that reports the map location

lial venule cells of hevin, a close relative of the antiadhesive extracel-of many uncloned mouse mutations (provided by Mouselular matrix protein SPARC. Immunity 2:113, 1995Genome Database, a computerized database maintained at15. Girard JP, Springer TA: Modulation of endothelial cell adhe-the Jackson Laboratory, Bar Harbor, ME). SC1/ECM2 sion by hevin, an acidic protein associated with high endothelialmapped in a region of the composite map that lacks mouse venules. J Biol Chem 278:4511, 1996

mutations with a phenotype that might be expected for an 16. Nishikawa S-I, Ogawa M, Nishikawa S, Kunisada T, Kodamaalteration in this locus (data not shown). Mapping of human H: B lymphopoiesis on stromal cell clone: Stromal cell clones actingSC1/ECM2 has not been reported, but tight linkage to the on different stages of B cell differentiation. Eur J Immunol 18:1767,

1988murine Gfi1 gene suggests that it might be located on 1p.17. Miyake K, Medina K, Ishihara K, Kimoto M, Auerbach R,An examination of human mutations in this location (OMIM)

Kincade PW: A VCAM-like molecule on murine bone marrow stro-did not reveal an obvious candidate for a condition that canmal cells mediates binding of lymphocyte precursors in culture. Jbe attributed to SC1/ECM2-related defects.Cell Biol 114:557, 1991NIH3T3 cells produce SC1/ECM2 transcripts and presum-18. Oritani K, Wu X, Medina K, Hudson J, Miyake K, Gimbleably secrete the mature protein. Nonetheless, transfection of JM, Burstein SA, Kincade PW: Antibody ligation of CD9 modifiesthese fibroblasts with membrane-bound forms of SC1/ECM2 production of myeloid cells in long-term cultures. Blood 87:2252,

markedly increased their ability to support lymphocyte 1996expansion. As additional information accrues about the mo- 19. Copeland NG, Jenkins NA: Development and applications oflecular requirements for lympho-hematopoiesis, it should be- a molecular genetic linkage map of the mouse genome. Trends Genet

7:113, 1991come possible to engineer ever more efficient supporting20. Jenkins NA, Copeland NG, Taylor BA, Lee BK: Organiza-microenvironments.

tion, distribution, and stability of endogenous esotropic murine leu-ACKNOWLEDGMENT kemia virus DNA sequences in chromosomes of Mus musculus. J

Virol 43:26, 1982We thank Mary Barnstead for excellent technical assistance.21. Benovic JL, Onorato JJ, Arriza JL, Stone WC, Lohse M,

REFERENCES Jenkins NA, Gilbert DJ, Copeland NG, Caron MG, Lefkowitz RJ:Cloning, expression, and chromosomal localization of b-adrenergic1. Campbell A, Wicha MS, Long M: Extracellular matrix pro-

motes the growth and differentiation on murine hematopoietic cells receptor kinase 2. J Biol Chem 266:14939, 199122. Bell DW, Taguchi T, Jenkins NA, Gilbert DJ, Copeland NG,in vitro. J Clin Invest 75:2085, 1985




Gilks CB, Zweidler-McKay P, Grimes HL, Tsichlis PN, Testa JR: 35. Engel J, Taylor W, Paulsson M, Sage H, Hogan B: CalciumChromosomal localization of a gene, GFI1, encoding a novel zinc binding domains and calcium induced conformational transition offinger protein reveals a new syntonic region between man and ro- SPARC/BM-40/Osteonectin, an extracellular glycoprotein expresseddents. Cytogenet Cell Genet 70:263, 1995 in mineralized and nonmineralized tissues. Biochemistry 26:6958,23. Green EL: Linkage, recombination, and mapping, in: Genetics 1987

and Probability in Animal Breeding Experiments. New York, NY, 36. Maurer P, Hohenadl C, Hohenester E, Gohring W, Timpl R,Oxford University Press, 1981, p 77 Engel J: The C-terminal portion of BM-40 (SPARC/Osteonectin) is24. Lund J, Winter G, Jones PT, Pound JD, Tanaka T, Walker an autonomously folding and crystalliable domain that binds calcium

MR, Artymiuk PJ, Arata Y, Gurton DR, Jefferis R, Wood JM: and collagen IV. J Mol Biol 253:347, 1995Human Fc gamma RI and Fc gamma RII interact with distinct but 37. Lane TF, Sage EH: The biology of SPARC, a protein thatoverlapping sites on human IgG1. J Immunol 147:2657, 1991 modulates cell-matrix interactions. FASEB J 8:163, 199425. Rezaie AR, Fiore MM, Neuenschwander PF, Esmon CT, 38. Raines EW, Lane TF, Iluera-Arispe ML, Ross R, Sage EH:

Morrissey JH: Expression and purification of a soluble tissue factor The extracellular glycoprotein SPARC interacts platelet-derivedfusion protein with an epitope for an unusual calcium-dependent growth factor (PDGF)-AB and -BB and inhibits binding of PDGFantibody. Protein Expr Purif 3:453, 1992 to its receptors. Proc Natl Acad Sci USA 89:1281, 199226. Medina KL, Smithson G, Kincade PW: Suppression of B 39. Tashiro K, Tada H, Heilker R, Shirozu M, Nakano T, Honjo

lymphopoiesis during normal pregnancy. J Exp Med 178:1507, 1993 T: Signal sequence trap: A cloning strategy for secreted proteins27. Kaisho T, Oritani K, Ishihara K, Tanabe M, Muraoka O, Ochi and type I membrane proteins. Science 261:600, 1993

T, Hirano T: Human bone marrow stromal cell lines from myeloma 40. Nagasawa T, Kikutani H, Kishimoto T: Molecular cloningand rheumatoid arthritis that can support murine pre-B cell growth. and structure of a pre-B-cell growth-stimulating factor. Proc NatlJ Immunol 149:4088, 1992 Acad Sci USA 91:2305, 199428. Heijne G: A new method for predicting signal sequence cleav- 41. Bleul CC, Farzan M, Choe H, Parolin C, Clark-Lewis I, So-

age sites. Nucleic Acids Res 14:4683, 1986 droski J, Splinger TA: The lymphocyte chemoattractant SDF-1 is a29. Kretsinger RH: The informational role of calcium in the cyto- ligand for LSESTR/fusin and blocks HIV-1 entry. Nature 382:829,sol. Adv Cycl Nucl Res 11:1, 1979 199630. Mason IJ, Taylor A, Williams JG, Sage H, Hogan BLM: 42. Oberlin E, Amara A, Bacherlerie F, Bassia C, Virelizier J-L,Evidence from molecular cloning that SPARC, a major product of Arenzana-Seisdedos F, Schwartz O, Heard J-M, Clark-Lewis I,mouse embryo parental endoderm, is related to an endothelial cell Legler DF, Loetcher M, Baggiolini M, Moser B: The CXC chemo-culture shock glycoprotein of Mr 43,000. EMBO J 5:1465, 1986kine SDF-1 is the ligand for LESTR/fusin and prevents infection by31. Guermah M, Crisanti P, Laugier D, Dezelee P, Bidou L,T-cell-line-adapted HIV-1. Nature 382:833, 1996Pessac B, Calothy G: Transcription of a quail gene expressed in43. Nagasawa T, Hirota S, Tachibana K, Takakura N, Nishikawaembryonic retinal cells is shut off sharply at hatching. Proc Natl

S, Kitamura Y, Yoshida N, Kikutani H, Kishimoto T: Defects of B-Acad Sci USA 88:4503,1991cell lymphopoiesis and bone-marrow myelopoiesis in mice lacking32. Alliel PM, Perin JP, Jolles P, Bonnet FJ: Testican, a multido-the CXC chemokine PBSF/CDF-1. Nature 382:635, 1996main testicular proteoglycan resembling modulators of cell social44. Kaisho T, Ishikawa J, Oritani K, Inazawa J, Tomizawa H,behaviour. Eur J Biochem 214:347, 1993

Muraoka O, Ochi T, Hirano T: BST-1, a surface molecule of bone33. Shibanuma M, Mashimo J, Mita A, Kuroki T, Nose K: Clon-marrow stromal cell lines that facilitates pre-B-cell growth. Procing from a mouse osteoblastic cell line of a set of transforming-Natl Acad Sci USA 91:5325, 1994growth-factor-b1-related genes, one of which seems to encode a45. Cohen PL, Eisenberg RA: Lpr and gld: Single gene modelsfollistatin-related polypeptide. Eur J Biochem 217:13, 1993

of systemic autoimmunity and lymphoproliferative disease. Annu34. Zwijsen A, Blockx H, Arnhem WV, Willems J, Fransen L,Rev Immunol 9:243, 1991Devos K, Raymackers J, Voorde A, Slegers H: Characterization46. Flaumenhaft R, Rifkin B: Extracellular matrix regulation ofof a rat C6 glioma-secreted follistatin-related protein (FRP). Eur J

Biochem 225:937, 1994 growth factor and protease activity. Curr Opin Cell Biol 3:817, 1991

