cDNA Cloning and Molecular Characterization of MSE55, a Novel ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry andMolecular Biology, Inc.

Vol. 267, No. 20, Issue of July 15, pp. 13986-13992, 1992 Printed in U.S.A.

cDNA Cloning and Molecular Characterization MSE55, of a Novel Human Serum ConstituentProtein That Displays Bone Marrow Stromal/Endothelial Cell-specific Expression* (Received for publication, October 9,1991)

Wadie F. BahouS, AlanD. Campbell#,and Max S. WichaQ From the $Division of Hematology, State University of New York, Stony Brook, New York 11794-8151ana‘ the §Division of Hematology/Oncology, University of Michigan, Ann Arbor, Michigan 11794-8151

Hemonectin is a lineage-specific cytoadhesive protein that may be involved in the developmentally regulated adhesion of granulocytic cells to bone marrow stroma. Immunoblot analysis using an anti-hemonectin antibody recognizes two distinct immunoreactive species in endothelial cell lysates (-Mr 65,000) andhuman serum (-Mr 55,000). Initial characterization of the 55kDa protein has now been completed by isolating the cDNA from a human endothelial cell expression library. Sequence analysis of overlapping clones identifies acomposite sequence spanning 2030 nucleotides with an open reading frame of 1173 base pairs. No significant sequence similarity wasobserved on analysis of current GenBank databases. The open reading frame wasexpressed as a recombinant protein inEscherichia coli and used as an immunogen for theproduction of a specific polyclonal antibody. Immunoblotting with this antibody identifies a single immunoreactive species of apparent M, 55,000 in HUVEC lysates and human serum, confirming that a secreted form normally circulates as a serum constituent protein. This antibody fails to recognize purified hemonectin, suggesting that the M, 55,000 protein is not hemonectin. Cross-species Southern blot analysis reveals persistent hybridizing fragmentsin all species tested, suggestive of a developmentally conserved function. Northern blot analysis demonstratesexpression limited to endothelial andbone marrow stromalcells, but not poly(A) RNA from monkey liver, spleen, brain, lung, and kidney. On this basis, we have designated this novel protein MSES5, for marrow gtromal/gndothelial cell protein with molecular a mass of 55,000 daltons. Its tissuespecific expression may suggest a functional role in hematopoiesis.

has been convincingly demonstrated (2). In long-term bone marrow culture, the stromal cell layer is composed of distinct cellular types which collectively support hematopoiesis. The majority of these stromal cells express type IV collagen, laminin, andvon Willebrand factor, suggestive of an endothelial cell lineage of origin (3). The role of the ECM in promoting the growth and differentiation of hematopoietic cells in vitro has also been demonstrated (4). We have previously described a novel ECM cytoadhesive protein that promotes the lineage-specificattachment of granulocytic cells to bone marrow stroma (5). Hemonectin mediates the adhesion of HL-60 cells and developing bone marrow granulocytic cells but not circulating mature granulocytes, suggesting that developmentally regulated adhesion to this protein may be involved in the mechanism of granulocyte release into the peripheral circulation (6). In this paper, we show that an antibody to hemonectin recognizes two immunologically distinct proteins: a 55-kDa protein found in human serum and a 65-kDa species found in lysates isolated from human endothelial cells. cDNAcloning and initial characterization of the 55,000-dalton protein have now been completed, confirming its unique identity and tissue-specific expression restricted to human endothelial and bone marrow stromal cells. Peptide sequence analysis of three tryptic fragments obtained from a purified hemonectin fraction are not contained within the primary translation product, and immunoblotting with an antibody specifically directed against this recombinant protein fails to recognize purified hemonectin, suggesting that this protein is not hemonectin. Pending further characterization, we term this proteinMSE55, for human bone =arrow $romal/gndothelial cell protein of apparent M, 55,000. MATERIALSANDMETHODS

The hematopoietic microenvironment is an elaborate meshwork of soluble growth factors, extracellular matrix (ECM),’ and stromal cells that maintains the dynamic repopulation of circulating blood elements (1).The importance of this adherent stromal cell layer in maintaining effective hematopoiesis * Supported in part by Grants HL02431 (to W. F. B.) and HL35255 (to M. S. W.) from the National Institutes of Health and a Merit Review (to A. D. C.) from the Veteran’s Administration Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequencefs)reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank withaccessionnumberfs) M88338. The abbreviation used is: ECM, extracellular matrix.

Zmmunoblotting and Cell Line Preparation-Permanently cloned bone marrow stromal cell lines from murine long-term bone marrow culture were kindly provided by Dr. P. Anklesaria (Worcester, MA) and propagated as previously described (7). Human umbilical vein endothelial cells (HUVEC) were propagated formpooled primary cultures of human umbilical veins (8).Endothelial cells were grown on gelatin-coated plates, and themedia were supplemented with 10% fetal calf serum, 100 pg/ml endothelial cell growth factor (Collaborative Research, Inc., Bedford, MA), 100pg/ml porcine intestinal heparin, penicillin (100 units/ml),and streptomycin (100 pg/ml). After rinsing twice with Dulbecco’s phosphate-buffered saline, confluent cells in third or fourth passage were directly harvested with a rubber policeman. Lysates from HUVECs, the Philadelphia-chromosome positive (Ph+) erythroleukemic cell line K562 (9), and the T cell acute lymphoblastic leukemic cell line MOLT-4 (10) were prepared by solubilization in 0.1% sodium dodecyl sulfate (SDS), 50 mM Tris, pH 7.5, and 10 mM phenylmethylsulfonyl fluoride. Samples were size-fractionated on an 8%SDS-polyacrylamide gel, transferred to nitrocellulose filters, and blocked overnight at 4 “C in BLOTTO

13986

Characterization Molecular (BLOTTO is 5% Carnation non-fat powdered milk, 10 mM Tris, pH 7.5, 140 mM NaCl). Immunoblotting and development of filters was performed as previously described ( l l ) , using the specific antibody diluted in BLOTTO, and a 1500 dilution of the appropriate peroxidase-conjugated second antibody (Jackson Immunoresearch, West Grove, PA). Immunoscreening and cDNA Cloning-Construction of both HUVEC cDNAlibraries cloned into theexpression vector X g t l l has been previously described (12). Immunoscreening of recombinant phage was performed by the method of Young and Davis (13), except that preincubation and antibody treatment of filters was performed in BLOTTO. To remove any background anti-Escherichia coli antibody, the anti-hemonectin antibody (5) diluted in BLOTTO was adsorbed against nitrocellulose filters presoaked with lysates prepared from wild-type Xgtll-infected E. coli. Immunoreactive phage clones were plaque-purified, andthe DNA was purified from minilysates by standard methods (14). The initial antibody-positive cDNA insert was directly amplified from pure phage lysates by the polymerase chain reaction (X), using synthetic oligonucleotides 69 bp upstream from the Xgtll EcoRI cloning site (forward primer 5'-TGGCTGAATATCGACGGTTTCCAT) and 79 bp downstream from the Eco RI site (reverse primer 5'-CAGACATGGCCTGCCCGGTTATTA) (11). After EcoRI digestion, the cDNA insert was directly labeled in 1% low melt agarose by random hexamer priming (16) and used for further screening by standard methods (14). Full length cDNA inserts were isolated, digested with EcoRI, directly subcloned into M13mp18, and bidirectionally sequenced using the dideoxy chain termination method (17), and T7 DNA polymerase (Sequenase, U. S. Biochemical). Sequence analysis was performed with the IBI-Pustell software package (International Biotechnologies, Inc., New Haven, CT) (18) and theWisconsin Genetics Computer Group package (19). mRNA Analysis-Total cellular RNA from appropriate cell lines and tissues was isolated by immediate solubilization in guanidine hydrochloride, and the RNA was serially ethanol-precipitated followed by phenol extraction and further ethanol precipitation (20). The RNA was quantified by optical density at 260 nm and by direct visual estimation of ribosomal RNA by ethidium bromide staining in a denaturing formaldehyde gel. Rhesus monkey poly(A) RNA was purchased from Clontech Laboratories (Palo Alto, CA). Analysis of transcripts was performed by Northern transfer tonylon membranes (Schleicher and Schuell) and subsequent hybridization with the radiolabeled insert at 68 "C overnight using a solution containing 6 X SSC without formamide and 50 pg/ml sonicated salmon sperm DNA (20). Blots were washed to high (0.1 X SSC, 0.1% SDS, 1mM EDTA, pH 8.0, and 10 mM sodium phosphate a t 68 "C for 1 h) or low (0.1 X SSC, 0.5%SDS, 1 mM EDTA, pH 8.0, and 10 mM sodium phosphate at 55 "C for 30 min) stringency according to the manufacturer's recommendations and analyzed by autoradiography with an intensifying screen at -70 "C for 3-10 days. DNA Extraction and Southern Blotting-DNA from normal human volunteers or various animal species was extracted from peripheral blood leukocytes as previously described (21) and quantitated by absorption spectrophotometry at 260 nm. Approximately 5-10 pg of DNA was digested with the appropriate restriction enzyme (Stratagene, La Jolla, CA), and samples were size-fractionated by electrophoresis in a 0.8% agarose gel and transferred to nylon membranes by Southern transfer (22). Hybridization, wash conditions, and autoradiography were completed as described above. For some experiments, filters were stripped according to the manufacturer's recommendations, and adequacy was confirmed by overnight exposure to Kodak XAR-5 film. Fusion Protein Expression and Antibody Preparation-A portion of the cDNA encoding the open reading frame (ORF) was amplified by polymerase chain reaction using synthetic oligonucleotide primers with added EcoRI restriction sites (underlined). The forward primer corresponded to nucleotide positions 348-371 ( 5 ' - C A C K TCGATGCCCGGCCCCCAGGGGGGCAG)and thereverse primer to positions 1932-1903 (5'-CACGAATTCTGGGATGAGCTGGGG GCAGGCGGTCCCTTT). The polymerase chain reaction product was gel-purified, digested with EcoRI, and ligated in-frame into the Staphylococcalfusion vector pRIT2T (Pharmacia LKBBiotechnology Inc.) (23), which had been specifically engineered with a collagenspecific recognition sequence between carrier protein and the EcoRI cloning site.2 Chimeric fusion proteins consisting of Staphylococcal protein A and the primary translation product were made as previously described (11).The insoluble fusion protein was solubilized in

-

A. Yang and D. Ginsburg, personal communication.

of MSE55

13987

8 M guanidine hydrochloride and affinity-purified by incubation of crude bacterial lysates (diluted 1:40 in Tris saline-Tween-20 50 mM Tris, pH 7.5, 150 mM NaCl, and 0.05%Tween-20) with 1ml of I&coated Sepharose beads (Pharmacia), with vigorous stirring at 4 "C overnight. Beads were sedimented and extensively washed with Tris saline-Tween-20, and adsorbed protein samples were size-fractionated by electrophoresis in reducing buffer on a Coomassie-stained SDS-polyacrylamide gel (24). For the production of a polyclonal antibody, approximately 100 pg of the chimeric recombinant protein was excised as a discrete band from the gel, suspended in complete Freund's adjuvant, and used as immunogen in a rabbit, with two subsequent boosts using incomplete Freund's adjuvant. RESULTS

Endothelial Cell Immunoreactivity-Human umbilical vein endothelial cells were isolated and propagated as described above. Lysates from K562 cells, MOLT-4 cells, and HWECs were solubilized, size-fractionated in an 8% SDS-polyacrylamide gel, and transferred to nitrocellulose filters. An immunoblot performed with the guinea piganti-rabbit hemonectin antibody (5) reveals distinct immunoreactivity with lysates from K562 cells (9), HUVEC, and control rabbit hemonectin but not with lysates from MOLT-4 cells (10) (Fig. L4). Additionally, an immunoreactive protein of different size is noted in human serum. As judged by molecular weightstandards, the identical immunoreactivespecies inboth endothelial and K562 cells has a molecular massof approximately 65,000 daltons, compared with the 55,000-dalton protein identified in human serum. cDNA Cloning and SequenceAnalysis-Since a distinct immunoreactive species was identified in endothelial cells, a HUVEC cDNAlibrary constructed in the bacterial expression vector Xgtll was screened with the polyclonal antibody. ApA

B

-88-

-45-

-29AMulEwwEcRl ANTIBODY

umysE55

FIG. 1. Immunoblots using the ( A ) anti-hemonectin or ( B ) anti-MSE55 antibodies. A, total cellular lysates (-5 mg/ml) were prepared from human umbilical vein endothelial cells (HUVEC) ( l a n e 2), K562 cells (lane 3), and MOLT-4 cells ( l a n e 4 ) . Human serum (diluted 1:25) ( l a n e I), purified rabbit hemonectin (1pg, lane 6 ) , or individual cell lysates (50pl) were size-fractionated on an 8% SDSpolyacrylamide gel, transferred to nitrocellulose filters, and blotted with the guinea pig anti-hemonectin antibody (5)at a 1:lOOO dilution. Immunoreactive species of different size were seen in human serum (M.55,000)when compared to lysates from HUVEC or K562 cells (Mr-65,000).The anti-hemonectin antibody fails to recognize recombinant MSE55. B, the ORF of MSE55 (see Fig. 2 and text) was ligated in-frame into theexpression plasmidpBluescript (Stratagene), and recombinant fusion proteins made as previously described. Total lysates (-320 mg/ml, 20 plllane) prepared from bacteria transformed with the wild-type plasmid ( l a n e 5)or the plasmid construct encoding the recombinant MSE55 fusion protein (lane 6 ) , purified rabbit hemonectin ( l a n e 4 ) , total cellular lysates (50 pl) from HUVEC ( l a n e 2), K562 cells ( l a n e 3), or human serum (diluted 1:lOO) were sizefractionated on an 8%SDS-PAGE gel, transferred to nitrocellulose, and immunoblotted with the specific rabbit anti-MSE55 antibody at a 1:2000 dilution. Western blot with a specific anti-MSE55 antibody recognizes one distinct band of M,55,000 in human serum and lysates from K562 and HUVECs. No immunoreactivity is seen with rabbit hemonectin or bacterial lysates transformed with wild-type plasmid. Molecular weight standards are indicated.

13988

Molecular Characterization of MSE55

proximately 1.5 million clones from a random hexamerprimed library were initially screened with the subsequent identification of phage clone H87. The radiolabeled cDNA insert of H87 was used as probe to rescreen an oligo(dT)primed library. Positive phage were identified at a frequency of approximately 1:5000, and 10 such clones ranging in size from 1.2 to 2.2 kb were chosen for further analysis. Fig. 2A displays the restriction map of seven representative phage cDNA inserts. As judged by Northern blot analysis (see below), three inserts(H115, H136, and H142) were predicted to be near full length clones, were subcloned into M13mp18, and sequenced by dideoxy chain termination (17). Two of the clones (H115 and H136) were sequenced on both strands, and a third (H142) was sequenced on the sense strand only. All three sequences were identical except for two silent nucleotide changes, 996 bp (H136, CTG, Leu; H115 and H142, CTC, Leu) and 1428 bp (H136, GCG, Ala; H115 and H142, GCT, Ala). Neither of these mismatches results in an alternative translation product. These single base pair differences could represent either cloning artifacts (e.g. reverse transcriptase errors) or DNA sequence polymorphisms. The first 192 base pairs of H115 represent aunique sequence which subsequently converges with H136 at nucleotide 69 (numbered with respect to H136). Two other clones €3144 and H142 overlap with H136 at nucleotides 1 and 3, respectively, confirming the authenticity of this 5’ region (Fig. 2B). That the diverging sequence of H115 may represent an unprocessed intron is suggested by the presence of a typical splice acceptor sequence at the converging junction (sequence not shown) (25). This was confirmed by Northern blotting utilizing this 5’ segment as probe, which revealed no message in human endothelial cells (not shown, see “Tissue and Cell Expression” below). Two upstream methionines (positions 186 and 318) are present with a purine (G) at position -3 that fit the consensus for translation initiation (26) but are followed by in-frame stop codons. Within this context, reinitiation at downstream start sites is not problematic (26). Thethird methionine represents the start of the long ORF. With regard to its long 5”untranslated region (>300 bp) and the multiplicity of upstream ATGs, this cellular gene displays two featuresdescribed in a large number of proto-oncogenes described to date (26). Protein Analysis-The composite sequence of overlapping clones consists of 2030 nucleotides with a long open reading frame of 1173 base pairs, encoding a 391-amino acid protein of predicted molecular mass of 42,000 daltons. A homology search through the GenEmbl and GenBank Databases (Version 7.0) using Tfasta and Wordsearch algorithms (19) confirmed the unique identity of the sequence and revealed no significant homology to previously described proteins. Of note, the predicted protein sequence did not contain three recently sequenced peptides from a purified rabbit hemonectin preparati~nA . ~ Kyte and Doolittle hydropathy profile (27) revealed no evidence for a long hydrophobic region, suggesting that it is unlikely to be an integral membrane protein. The lack of a signal peptide has been previously described for other serum proteins, including plasminogen activator inhibitor 2 (28) and ovalbumin. In the latterexample, an internal noncleaved hydrophobic sequence has been identified, confirmed only after extensivemutagenesis studies (29). Whether a similar mechanism is operational in MSE55 is presently unknown. A series of proline/alanine repeats with the consensus PAANPPA are found inthe internalportion of the protein (Fig. 3, A and C ) . This eight-domain repeat is striking in that no

:’A. Campbell and M. Wicha, unpublished data.

A. FA

I

5’

HI49

I

I

P

A S P

AS

P

i

3