Genomics 67, 54–68 (2000)doi:10.1006/geno.2000.6235, available online at http://www.idealibrary.com on
MEPE, a New Gene Expressed in Bone Marrow andTumors Causing Osteomalacia
Peter S. N. Rowe,* ,1 Priyal A. de Zoysa,* Rong Dong,* Huei Rong Wang,*Kenneth E. White,† Michael J. Econs,† and Claudine L. Oudet‡
*Centre for Molecular Osteo-Renal Research, Department of Biochemistry and Molecular Biology, Royal Free and University CollegeMedical School, Rowland Hill Street, Hampstead, London NW3 2PF, United Kingdom; ‡Institut de Genetique et de Biologic
Moleculaire et Cellulare, CNRS/INSERM/ULP, Illkirch, France; and †Department of Medicine,Indiana University School of Medicine, Indianapolis, Indiana 46202
Received January 18, 2000; accepted April 20, 2000
sCo
ehib1wclmha
Oncogenic hypophosphatemic osteomalacia (OHO)is characterized by a renal phosphate leak, hy-pophosphatemia, low-serum calcitriol (1,25-vitamin-D3), and abnormalities in skeletal mineralization.Resection of OHO tumors results in remission of thesymptoms, and there is evidence that a circulatingphosphaturic factor plays a role in the bone disease.This paper describes the characterization and clon-ing of a gene that is a candidate for the tumor-se-creted phosphaturic factor. This new gene has beennamed MEPE (matrix extracellular phosphoglyco-protein) and has major similarities to a group ofbone–tooth mineral matrix phospho-glycoproteins(osteopontin (OPN; HGMW-approved symbol SPP1),dentin sialo phosphoprotein (DSPP), dentin matrixprotein 1 (DMP1), bone sialoprotein II (IBSP), andbone morphogenetic proteins (BMP). All the pro-teins including MEPE contain RGD sequence motifsthat are proposed to be essential for integrin–recep-tor interactions. Of further interest is the findingthat MEPE, OPN, DSPP, DMP1, IBSP, and BMP3 allmap to a defined region in chromosome 4q. Refinedmapping localizes MEPE to 4q21.1 between ESTsD4S2785 (WI-6336) and D4S2844 (WI-3770). MEPE is525 residues in length with a short N-terminal signalpeptide. High-level expression of MEPE mRNA oc-curred in all four OHO tumors screened. Three of 11non-OHO tumors screened contained trace levels ofMEPE expression (detected only after RT-PCR andSouthern 32P analysis). Normal tissue expression
as found in bone marrow and brain with very-low-evel expression found in lung, kidney, and humanlacenta. Evidence is also presented for the tumor
Sequence data from this article have been deposited with theEMBL/GenBank Data Libraries under Accession Nos. AJ276396 andHSA276396.
1 To whom correspondence should be addressed. Telephone: 144(0)171-830-2938 or 144 (0)171-794-0500, Ext. 4938. Fax: 144 (0)171-794-9645. E-mail: [email protected]. Internet: http://www.ucl.
A number of familial diseases that result in disor-ders of phosphate uptake, vitamin D metabolism, andbone mineralization have been characterized. Recentlya gene (PHEX) that is defective in patients with X-linked hypophosphatemic rickets (HYP) was clonedand characterized (HYP Consortium et al., 1995; Rowet al., 1996a, 1997; Francis et al., 1997). Oncogenicypophosphatemic osteomalacia (OHO) has many sim-
larities to HYP with an overlapping pathophysiology,ut different primary defects (Drezner, 1990; Rowe,997, 1998a). Both disorders result in renal phosphateasting, impaired vitamin D metabolism, rickets (in
hildren), and osteomalacia. Tumor-acquired osteoma-acia is rare, and the tumors are mainly of mesenchy-
al origin, although a number of different tumor typesave also been reported (Francis and Selby, 1997; Io-kimidis et al., 1994; Lyles et al., 1980; Rowe, 1994,
1997; Shane et al., 1997; Weidner and Cruz, 1987).Surgical removal of the tumor(s) results in the disap-pearance of disease symptoms, and extensive researchhas confirmed that an unknown circulating phospha-turic factor(s) plays a role in the pathogenesis of thedisease (Miyauchi et al., 1988; Aschinberg et al., 1977;Popovtzer, 1981; Rowe et al., 1996b; Cai et al., 1994).
This paper presents the cloning and characterizationof an OHO tumor-derived factor that is expressed inbone marrow and brain. We have named this newprotein MEPE (matrix extracellular phosphoglycopro-tein). The high levels of expression found in OHO tu-mors indicate that it may play a role in the pathophys-
Clinical Profile of Patients BD, ND, EM, DS, LS, andLA with OHO
Patients BD (hemangiopericytoma) and ND (intracranial mixedconnective tissue variant of phosphaturic mesenchymal tumor) havebeen described in earlier publications (Rowe et al., 1996b; David etal., 1996). Both patients exhibited classical tumor osteomalacia andpresented with low serum phosphate, radiological osteomalacia, andlow serum 1,25-vitamin D3 (calcitriol). Patient EM was an elderly
ale with a large hemangiopericytoma on his left upper arm andith classic OHO. The tumor was removed at the Royal Nationalrthopaedic Hospital (Stanmore, UK; Mr. Briggs, surgeon, and Dr.. Stamp, consultant) and was processed as described for BD andD. Patient DS was a young female (22 years; also with classic OHO)
rom the United States, and her forearm tumor (hemangiopericy-oma) was resected at the University of Pittsburgh Medical CenterProfessor Frederick Gilkey, consultant) and sent to the Unitedingdom on dry ice.Two skin polyp tumors originated from patients (LS and LA) with
inear sebaceous nevus syndrome (LSNS) and associated rickets.oth patients LS (Great Ormond Street Hospital for Sick Children,K; Mr. Mayo and Dr. Scher) and LA (University of Paris Hospitalrofessor Michelle Garabedian and Dr. Luce Condamine) were cov-red in skin polyps with differing morphologies and presumed dif-ering phosphaturic activities.
Tumor Conditioned Medium (TCM)
Tumor samples from BD, ND, EM, LA, and LS were collectedimmediately after resection. Samples were then cut into ;1-mm3
pieces, and some were frozen in liquid nitrogen. The remainingpieces of tumor tissue were processed for tissue culture as describedpreviously (Rowe et al., 1996b). Control skin biopsies were removedand treated in the same way. For patient ND, intracranial tumormaterial, adjacent subdura, and dura (furthest from tumor) wereprocessed (Rowe et al., 1996b).
OHO Tumor cDNA Expression Library and CloneIsolation
Poly(A)1 RNA was extracted from patient BD using streptavidin-Magnesphere paramagnetic particle technology (PolyATract System,Promega). The purified mRNA was then used to generate a cDNAtemplate using the cDNA synthesis kit from Stratagene. Linkerprimers were added to the cDNA to generate a 59 EcoRI compatiblecDNA end and an XhoI-compatible 39 cDNA end for cloning intoLambda ZAP II uni bacteriophage vector. Recombinant bacterio-phages were plated out and amplified on Escherichia coli XL1-Bluemrf9. Total primary clones numbered 800,000 with 6% wildtyperepresentation.
Rabbit antisera raised against patient BD preoperation serum(Rowe et al., 1996b) were extensively preabsorbed with normal hu-man serum and E. coli lysate to remove E. coli antibodies andbackground human serum-derived antibodies. Briefly, 80-mm-diam-eter nitrocellulose filters were added to whole E. coli lysate (Strat-agene), and a second set of filters were soaked with normal humanserum (10 ml). The impregnated filters were incubated for 10 min atroom temperature in sequence with 250 ml of 1:1000 diluted anti-rabbit preoperation antiserum in 1% BSA; 20 mM Tris–HCl (pH 7.5),150 mM NaCl (TBS); 0.02% NaN3. The preabsorbed preoperationantiserum (pre-abs-anti-op) was then used to screen the cDNA li-brary and all Western blots described.
Bacteriophage Lambda ZAP II uni OHO cDNA clones were platedout on E. coli XL1-Blue mrf9 and incubated for 3 h at 37°C. Hy-bond-N1 filters preincubated with 10 mM isopropyl b-D-Thiogalac-opyranoside (IPTG) were then placed on top of the developing
laques and incubated for a further 3 h at 42°C. Filters were then
emoved and washed with TBS supplemented with Tween 20 andhen blocked with 1% BSA in TBS with 0.02% NaN3 overnight at°C. Pre-abs-anti-op was then added to the blocked filters and left forh at room temperature. Subsequent washes of the filters and
ncubation with goat anti-rabbit alkaline phosphatase conjugate,ollowed by visualization using 5-bromo-4-chloro-3-indolyl phos-hate/nitroblue tetrazolium, were as described in Stratagene’s Pi-oblue Immunoscreening kit.
Phagemid Rescue, Construction, and Purification ofFusion Protein
Positive bacteriophage plaques were removed from agarose platesafter tertiary screening. The agarose plugs containing the lyticplaques were then added to 0.5 ml of SM buffer supplemented with0.02% chloroform and left at 4°C overnight. Rescue and transforma-tion of bacteriophage clones to pBluescript II SK(2) phagemids werecarried out using ExAssist phage as described by Stratagene. Thehost cells for the purified phagemid were E. coli SOLR cells. PlasmidDNA was then prepared using standard techniques (Rowe et al.,1994a) and sequenced using ABI fluorescence automated sequencingand standard vector-specific primers.
The cDNA coding sequence (deduced from the largest cDNA clone,pOHO11.1) of MEPE was subcloned into the prokaryote expressionvector plasmid pCAL-n-EK (Stratagene vector), and the constructwas transformed into E. coli BL21 (DE3) and E. coli XL1-Blue mrf9(strains obtained from Stratagene). The region selected for cloninglacked the first (N-terminal) 95 amino acid residues of MEPE andended with the natural stop codon of MEPE (a total of 430 residues;Figs. 1 and 2). A valine residue linked the N-terminal truncatedrec-MEPE to a vector-derived calmodulin-binding protein tag (CBP-tag). The method of ligation-independent cloning (LIC) was used asdescribed in the Stratagene Affinity cloning and protein purificationkit (Catalog Nos. 214405 and 214407). Two primers were designedfrom the MEPE sequence 59 and 39 ends, respectively, with addi-tional overhang linker sequence as follows (boldface sequence repre-sents linker):
Forward59 GAC GAC GAC AAG. GTG AAT AAA GAA TAT AGT ATC AGT AALinker MEPE specific
Reverse59 GGA ACA AGA CCC GT. CTA GTC ACC ATC GCT CTC ACT 39Linker MEPE specific
The amplified region is indicated in Fig. 1. For LIC, the amplifiedproduct was then treated with Pfu polymerase and dATP as de-scribed by Stratagene and then directly annealed to linearizedpCAL-n-EK plasmid vector with complementary linker overhangs.The construct was then transformed into competent E. coli XL1-bluemrf9 cells and competent E. coli BL21 (DE3). Clones were thenselected on ampicillin plates, and plasmids were prepared and se-quenced. Fusion protein encoded in pCAL-n-EK:MEPE was inducedin strain E. coli BL21 (DE3) using IPTG using Stratagene’s pCAL-n-EK induction protocol. Recombinant MEPE was purified usingcalmodulin affinity chromatography resin as described by the Strat-agene affinity purification protocol.
Cloning of Additional N-Terminal Region of MEPEby PCR
The Lambda ZAP II uni OHO cDNA library described above andused for cloning MEPE was also used to extend and clone additional59 end-terminal MEPE sequence. A total of 5 ml or 100,000 plaque-forming units of library was used as a substrate for PCR with MEPEgene-specific primer “GSP2” (59-T CTT CCC CCA GGA GTT TAATC-39) and vector-specific “T3-pol” primer (59-GGC CGC AAT TAA
CCC TCA CTA AAG G-39). The amplified PCR product was then
56 ROWE ET AL.
FIG. 1. Complete cDNA sequence of MEPE. Sequences used for PCR primer design are boxed, and the total number of amino acidresidues is 525 (1989 bp of cDNA sequence). The prokaryotic expression vector pCal-n-EK contained all in-frame residues from pOHO11.1(truncated rec-MEPE) from residue L96 to the MEPE stop codon (TAG), at 1625–1627 bp (see Materials and Methods). The singlepolyadenylation sequence AA(T/U)AAA is underlined in boldface. The region of shared localized homology with DMP1, DSPP, and OPN isunderlined in wavy line format (MEPE ASARM motif C-terminus), RGD residues are enclosed in an ellipsoid, a glycosaminoglycanattachment site is boxed (SGDG), a tyrosine kinase site is double-underlined, the signal peptide is defined by an arrowed line at the
N-terminus, and N-glycosylation motifs are enclosed in an arrow-boxed cartouche.
FIG. 1—Continued
eudltHcm
lcwp
rda(GwT
trtuslbnpnmowS
nM
58 ROWE ET AL.
separated using NuSieve agarose electrophoresis, visualized usingUV, excised, purified, and TA cloned into InVitrogen plasmid vectorpCR2.1 using conventional techniques (Rowe et al., 1994a). Theligated molecule was then transformed into E. coli XL1-blue compe-tent cells using Stratagene’s protocol and then sequenced using ABIautomated sequencing.
SDS–PAGE and Western Blotting of TCMand rec-MEPESamples were denatured at 70°C for 10 min in NuPAGE LDS
sample buffer, and proteins were resolved by 4–12% gradient SDS–PAGE with a Novex NuPAGE system, on bis-acrylamide gels (pH6.4) using Mops–SDS running buffer. After electrophoresis, proteinswere stained by incubating the gels in 7.5% acetic acid supplementedwith Sypro Orange. Visualization of proteins was achieved with UVillumination using a Bio-Rad FluorImager gel imaging system.
For Western blotting, acrylamide gels were equilibrated in trans-fer buffer at room temperature for 1 h, and the proteins were elec-troblotted onto PVDF membranes (Amersham). Membranes werethen incubated with pre-anti-op, post-anti-op, or calmodulin conju-gated to alkaline phosphatase. Secondary antibody (goat anti-rabbit-IgG conjugated to horseradish peroxidase (HRP)) was then added topre- and post-anti-op antisera-screened blots, and immunopositivebands were visualized by enhanced chemiluminescence (ECL1; Am-rsham) and a Bio-Rad FluorImaging-Multianaylst system. Calmod-lin affinity binding was visualized using the colorometric systemiscussed earlier for clone detection (Stratagene). Biotinylated mo-ecular weight markers (Amersham) were used as internal controlso assess transfer and molecular weight. Streptavidin conjugated toRP was added to the secondary antibody (goat anti-rabbit IgG
onjugated to HRP) to facilitate visualization of the biotinylatedarkers via chemiluminescence.
Glycoprotein StainingProteins were separated by SDS–PAGE and blotted onto PVDF
membranes as described above. Specific glycoprotein detection wascarried out using an Immuno-Blot kit for glycoprotein detection(Bio-Rad), and Amersham biotinylated markers were added as in-ternal controls. Subsequent blocking and detection were carried outas described earlier using the Enhanced Chemiluminescence kit(Amersham) and streptavidin horseradish peroxidase. Primary an-tibody and secondary goat anti-rabbit HRP were not used.
Northern and Southern AnalysesFor Northern analysis, two blots from Clontech (MTN and MTN
III) containing poly(A)1 RNAs from heart, brain, placenta, lung,iver, skeletal muscle, kidney, pancreas, stomach, thyroid, spinalord, lymph node, trachea, adrenal gland, and bone marrow tissuesere screened with MEPE cDNA amplified with specific internalrimers Pho433-111F (59-GGT TAT ACA GAT CTT CAA GAG AGA
G-39) and PHO877-111R (59-GTT GGT ACT TTC AGC TGC ATCACT-39) (Fig. 1). Purified and amplified MEPE cDNA was thenradiolabeled using [a-32P]dCTP (3000 Ci/mmol) in conjunction withthe MegaPrime labeling kit from Amersham. Hybridization (60°C),prehybridization (60°C), and washing of blots were carried out usingpublished methods (Rowe et al., 1996a). Southern analysis was car-ied out using genomic digests of DNA extracted from blood asescribed previously (Rowe et al., 1994b). Control GA3PDH (glycer-ldehyde 3-phosphate dehydrogenase) primers were G3PDH199F59-TTCCATGGCACCGTCAAGGCTGA-39) and G3PDH859R (59-AGACCACCTGGTGCTCAGTGTA-39). Control transferrin primersere TRANSF561F (59-CACGTAAACCTCTTGAGAAAGCA-39) andRANSF1503R (59-GGTCTCTGCTGATACACACTCT-39).
Primer ExtensionPrimer extension of MEPE was performed using the Primer Ex-
tension System (Promega, Southampton, UK). Primer labeling, an-
nealing, and extension steps were performed according to the man-ufacturer’s protocol (Promega Technical Bulletin 113). The MEPEcomplementary primer (PEP-1) used had the sequence 59-CACA-CAGCTTTGCTTAGTTTTCTC-39 and was 59-end-labeled using[g-32P]ATP (Amersham, Little Chalfont, UK). A 1-ml aliquot of a 1 in10 dilution of the labeled PEP-1 was annealed to 125 ng of BDpoly(A)1 RNA or to a water blank (the 2RNA control) in an 11-mlvolume at 63°C. Following reverse transcriptase (RT) primer exten-sion, the reactions were stopped by adding 23 TBE/urea loadingbuffer. fX174 HinfI molecular weight standards were labeled ac-cording to the manufacturer’s protocol and diluted 203, and either0.25- or 1-ml aliquots equivalents were loaded after dilution in load-ing buffer. Aliquots of 20 ml were denatured at 90°C for 10 min priorto loading onto a 1.0-mm, 10-well, 6% polyacrylamide TBE/urea geland run on an XCell II gel system (Novex) for 50 min at a constant180 V. Following electrophoresis, the gel was fixed for 30 min in 10%glacial acetic acid, 10% methanol and dried for autoradiography at270°C.
Reverse Transcription-PCR
MEPE internal primers, as highlighted in Fig. 1 (Pho433-111F andPHO877-111R), were used to amplify poly(A)1 RNA using reverseranscription-PCR and the Perkin–Elmer–Roche RNA PCR kit. Di-ect PCR amplification of three human tissue cDNA panels (Clon-ech) that contained the following human tissue cDNAs was alsondertaken—Panel I (K1420-1): heart, brain, placenta, lung, liver,keletal muscle, kidney, and pancreas; Panel IV (K1426-1): spleen,ymph node, thymus, tonsil, bone marrow, fetal liver, and peripherallood leukocyte; and human tumor Panel (K1422-1): breast carci-oma, lung carcinoma I, colon adenocarcinoma I, lung carcinoma II,rostatic adenocarcinoma, colon adenocarcinoma II, ovarian carci-oma, and pancreatic carcinoma. The cDNAs were normalized by theanufacturers against a selection of housekeeping genes and by
urselves against GA3PDH and transferrin. All amplified productsere resolved using agarose gel electrophoresis, and verified byouthern blotting and sequencing.
Human MEPE Chromosomal Localization
PCR primer design. PCR primers specific for human MEPE wereMEPE1 forward, 59-GAAAGGCTCCTGGGGTAGAC-39, and MEPE1reverse, 59-CCAGGATTCTTGGCTCACTC-39. The human-specificMEPE1 primer pair amplified a single 249-bp product from humangenomic DNA, but not from murine or Chinese hamster ovary cellDNA. Genomic PCR products were sequenced to verify their ampli-fication from the human MEPE gene.
PCR amplification of human MEPE. PCRs with somatic cell orradiation hybrid DNAs were performed in 25-ml volumes with 25–50
g of DNA and 100 ng of each MEPE primer in the presence of 2 mMgCl2 using the GeneAmp Kit (Perkin–Elmer). The PCR products
were resolved by electrophoresis on a 1.5% agarose gel (FMC Bio-products, Rockland, ME) and visualized by ethidium bromide stain-ing. All mapping experiments were repeated on separate days toverify the results.
Analysis of somatic cell hybrid mapping. MEPE was localized toa single chromosome using the NIGMS Human/Rodent Somatic CellHybrid Mapping Panel 2, Version 3 (Drwinga et al., 1993) (CoriellCell Repositories, Camden, NJ) by PCR with the protocols describedabove.
Radiation hybrid panel. The GeneBridge 4 radiation hybrid map-ping panel (Gyapay et al., 1996) (Research Genetics, Inc., Huntsville,AL) was used for fine localization of MEPE. Individual hybrids werescored by the presence or absence of PCR products after gel electro-phoresis. Chromosomal assignment was performed through theWhitehead Institute/MIT Center for Genome Research server (http://
Screening of TCM proteins from patient ND, osteo-sarcoma cell lines HTB96 (U-2 OS), and SaOS2 withpre-abs-anti-op revealed two distinct immunopositivebands at ;54–57 and ;200 kDa (Fig. 2). Patient ND’sumor sample and adjacent subdura tissue gave muchtronger 54- to 57-kDa signals than dura brain sampleonditioned medium (tissue furthest from tumor), ando staining for the 200-kDa band was found in the duraonditioned medium (Fig. 2A). SaOS2 conditioned me-ium had a reduced signal for the 200-kDa band rela-ive to ND TCM and HTB96 (Fig. 2A). Skin conditionededium (patients ND and BD), TCM derived fromonphosphaturic tumor material (patients LA and LS),nd medium controls were all negative when Westernlots were screened with both pre-abs-anti-op and post-bs-anti-op.Recombinant truncated MEPE (rec-MEPE) stained
ositively with pre-abs-anti-op, and this could be com-eted out with added rec-MEPE. A positive band of4–57 kDa was obtained with Sypro Orange protein-tained and pre-abs-anti-op-screened rec-MEPE (seeigs. 2A and 2B). This was the same size as the 55- to7-kDa band (pre-abs-anti-op Western screened) foundith patient ND tumor-conditioned medium and osteo-
arcoma cell lines HTB96 and SaOS2. RecombinantEPE contains an additional 4.5-kDa CBP-tag at the-terminus that decreases mobility and results in anpparent increase in molecular weight on SDS–PAGEels. The size of tumor-derived nascent protein andec-MEPE is likely to vary due to posttranslationalodification (proteolytic processing, phosphorylation,
nd glycosylation) of the tumor-derived form.TCM Western blots from OHO tumor patients BD
nd EM screened with pre-abs-anti-op gave results
FIG. 2. Photographs showing Western blot screened with preabel (4–12% gradient SDS–PAGE) (B) Lanes in A and B contain samubdura brain matter adjacent to tumor from OHO patient ND; 3, dur, media control, DM; 5, purified recombinant MEPE; 6, nonphosphaine conditioned medium HTB96; 8, osteosarcoma cell line conditi
arkers with sizes indicated in kDa.
dentical to those for patients ND and BD, but with
dditional faint positive bands at lower molecularass (48–52 kDa), as well as the 55- to 57-kDa band
omigrating with rec-MEPE (data not shown). In allamples, the major Sypro Orange-stained protein bandt 66 kDa was negative when screened with pre-abs-nti-op. Glycoprotein screening of duplicate proteinlots (see Materials and Methods) gave the same re-ults as screening with preoperation antisera, and both4- to 57- and 200-kDa bands stained positive, confirm-ng that these proteins are glycosylated.
In conclusion preabsorbed preoperation antiserumpecifically detects proteins derived from OHO-TCM.he major proteins detected fall into three distinctolecular size ranges: 48–52 kDa, 54–57 kDa (major
ands), and 200 kDa. All OHO-TCM samples wereositive for the 54- to 57-kDa protein (EM, ND, andD), and all proteins detected by preabsorbed preop-ration antisera stained positive when screened forlycoprotein status. Non-OHO tumor control tissuesnd media were negative when screened with preab-orbed preoperation antisera.
solation of cDNA Clones from Tumor cDNA Library
Expression screening of a Lambda ZAP II cDNAibrary (constructed from patient ND poly(A)1 RNA),
using pre-abs-anti-op, resulted in the isolation of 9positive clones. A total of 500,000 clones were screened,indicating an approximate representation of 0.054%.The 9 clones were sequenced, and 3 clones were com-pletely homologous to human clusterin (HGMW-ap-proved symbol CLU). The remaining 6 clones (MEPE)contained identical sequence of varying lengths(pOHO11.1, 430 amino acid residues (1655 bp);pOHO6.2, 297 amino acid residues; pOHO6.1, 291 res-idues; pOHO4.2, 231 residues; pOHO8.1, 227 residues;pOHO11.2, 227 residues). Additional 59 MEPE se-
ed preoperation antisera (A) and equivalent Sypro Orange staineds from 1, tumor cells from intracranial tumor (OHO patient ND); 2,rain matter—furthest material from brain tumor (OHO patient ND);ic tumor conditioned medium from patient LA; 7, osteosarcoma celld medium SaOS2 cells. Lane M contains protein molecular mass
sorbplea btur
one
quence was obtained by using gene-specific primer
arsM
pHprNd
60 ROWE ET AL.
GSP2 and vector-specific primer T3-pol to PCR-amplifyOHO library cDNA (see Materials and Methods). Thecomplete MEPE cDNA sequence is 1989 bp as con-firmed by primer extension, Northern blotting, andsequencing. All 9 clones (MEPE and clusterin) werein-frame with the microbial b-galactosidase promoter,thus producing identical protein sequence of varyinglengths for clusterin and MEPE. No other clones wereisolated using preabsorbed preoperation antisera ex-pression screening.
Computer Prediction and Structure Analysisof MEPE
Complete MEPE cDNA encodes 525 amino acid res-idues and 1989 bp of nucleic acid sequence. The largestcDNA clone isolated for MEPE contained the entire 39end of the gene with a poly(A)1 tail and a single poly-denylation sequence (AA(T/U)AAA) (Fig. 1). The openeading frame overlapped and extended the othermaller MEPE cDNA clones isolated, with a predicted
FIG. 3. GCG peptide structure (Rice, 1995; and see also http://wwrediction for MEPE. The primary amino acid backbone is shownydrophilicity/hydrophobicity regions are represented as ellipsoidsentagon. The N-glycosylation sites are represented as ellipsoids oegions on the primary backbone. The signal peptide is indicated-terminus. Predicted PHEX, NEP, and PC cleavage sites are shoelineated at the C-terminus.
R of 47.3. Additional 59 MEPE sequence obtained by
PCR of the Lambda ZAP II cDNA OHO library with agene-specific primer derived from the 59 end ofpOHO11.1 (GSP2) and a vector-specific primer (T3-pol)extended the sequence by 334 bp and 95 amino acidresidues (see Materials and Methods). The extendedoverlapping sequence contains a good consensus startcodon (methionine) as deduced using Kozak criteria(Kozak, 1996). The methionine start codon is posi-tioned 48 bases downstream from the start of the 59cDNA sequence (Fig. 1). Following the start codon, ashort signal peptide of 17 hydrophobic residues termi-nating with 2 alanine residues occurs (Fig. 1). Thepredicted size of the protein with signal peptide (pro-MEPE) is MR 58.4 kDa and without signal peptide isMR 56.3 kDa. The protein contains two N-glycosylationmotifs, NNST and NNSR, toward the 39 end (477–481),a glycosaminoglycan attachment site at residues 256–259 (SGDG), and an RGD cell attachment motif atresidues 247–249. The RGD motif is in a region ofpredicted turn (Garnier prediction Antheprot (Delea-gue, 1997); see Fig. 3) and is flanked by two regions of
hgmp.mrc.ac.uk for computer software details); secondary structures a central line with curves indicating regions of predicted turn.
diamonds, respectively, and the RGD motif is highlighted with atalks (C-terminus), and an alpha helix is indicated by undulatinga checkered box and coincides with a hydrophobic region at theand the shared DSPP/OPN/DMP1 repeat motif (ASARM-motif) is
predicted sheet structure is in turn flanked by tworegions of extended a-helix (216 to 227 and 291 to 296)and contains a thrombin cleavage site. The generalstructural context, with a predicted turn presence of anRGD cell attachment sequence and an associatedthrombin cleavage site, is similar to that found in os-teopontin (Denhardt and Guo, 1993; Katagiri andUede, 1998; Uede et al., 1997; Smith and Giachelli,998). The protein also has a number of predictedhosphorylation motifs for protein kinase C, caseininase II, tyrosine kinase, and cAMP–cGMP-depen-ent protein kinase. MEPE was also found to have aarge number of N-myristoylation sites, and these sitesppear to be a feature of RGD-containing phospho-lycoproteins (osteopontin, vitronectin, collagen, h-in-egrin binding sialoprotein (bone sialoprotein II), den-in sialophosphoprotein, dentin phosphoryn, dentinatrix protein 1, and fibronectin). There is an unusu-
lly high content of aspartate, serine, and glutamateesidues (26%), as in osteopontin (37%). Of particularnterest is the presence of a putative neprilysin (NEP)leavage site at residue 46 (KDNIG. . .FHHL), propro-ein convertase I and II cleavage sites (PCs) at residue3 (GKR. . .INQ), and a nardilysin site at residue 67VQE. . .RKK). These cleavage site motifs are con-erved and occur infrequently within MEPE and otherroteins (Rawlings and Barrett, 1999) (Fig. 3). TheEP site could also be a target site for PHEX given
hat both PHEX and NEP belong to the same M13 zincetalloendopeptidase family (Turner and Tanzawa,
997). Also, there are only two cysteines found in thentire molecule, and they occur upstream (toward the-terminus) relative to the protease cleavage sites.ne of the cysteines is located within the signal pep-
ide, and the other is found at residue position 31.equence homology to the dentin phosphoryn (DPP) or-terminal portion of dentin sialophosphoprotein
DSPP) was found after screening the TREMBL data-ase with the MEPE sequence. A region at the C-erminus of MEPE has a sequence of aspartate anderine residues (residues 509–522) that is almost iden-ical (80% homology) to a recurring motif found in DPPnd DSPP (Fig. 4A). Physiochemical comparison of theEPE motif (DDSSESSDSGSSSESD) with the DSPP/PP motif (SDSSDSSDSSSSSDSS) increases the ho-ology to 93%. The MEPE motif (acidic serine–aspar-
ate-rich MEPE-associated motif; ASARM motif)ccurs once at the C-terminus in MEPE (residues 509o 522), whereas the DSPP homologue is repeated atSPP residue positions 686 to 699, 636 to 646, and 663
o 677. Moreover, two related sequences, DSSDSSD-NSSSDS and DSSDSSDSSNSSDS, also with 80% ho-ology to the MEPE ASARM motif, are found in DSPP
t positions 576 to 589 and 800 to 813, respectively. Aimilar motif with 60% homology (DDSHQSDESHHS-ESD) is also found in osteopontin (residues 101 to16), and a casein kinase II phosphorylation site is
ontained within the region of homology (Fig. 4C). Ad-
ditional sequence homology to the C-terminal MEPEmotif is also found in DMP1 at residues 408 to 429(SSRRRDDSSESSDSGSSSESDG). A graphical pre-sentation of the regional sequence homology of theASARM motif in DSPP, dentin matrix protein 1(DMP1), and OPN (HGMW-approved symbol SPP1) ispresented in Fig. 4 as a llanview statistical plot, andTable 1 presents the sequence similarities in align-ment. Of interest is the repetitive occurrence of themotif at the C-terminal region of DSPP or the dentinphosphoryn portion.
MEPE Fusion Protein from pCAL-n-EK VectorSubcloning
The cDNA coding sequence comprising amino acids95–525 was subcloned into pCAL-n-EK as describedunder Materials and Methods (Fig. 1). This constructlacked the first 95 residues of MEPE (N-terminal cys-teine-containing fragment and signal peptide) and con-tained a valine residue between the CBP-tag and rec-MEPE. Validation of the fusion construct generated byIPTG induction of the E. coli host BL21 (DE3) wasachieved by screening Western blots with pre-abs-anti-op antisera and also with calmodulin conjugatedto alkaline phosphatase as described under Materialsand Methods (Fig. 2). The fusion protein with a CBP-tag (calmodulin binding peptide of 4.5 kDa), containingcalmodulin peptide, an enterokinase site, and a throm-bin site, was 56 kDa as deduced by SDS–PAGE (Fig. 2).This is in approximate agreement with the expectedmolecular size (;48 kDa without the CBP-tag). Purifi-cation of recombinant protein was achieved by calmod-ulin affinity chromatography as described under Ma-terials and Methods (Fig. 2). Preincubation of pre-abs-anti-op with the purified fusion construct resulted in adiminution of the 55- to 57-kDa TCM Western blotsignal, but not the 200-kDa band. The failure to reducecompletely the 55- to 57-kDa signal was presumed tobe due to specific recognition of the highly antigenicglycosylation moiety present in the nascent MEPE pro-tein (TCM), but absent in the microbial fusion con-struct of rec-MEPE. The fusion protein was soluble inaqueous Tris buffers, and detergents were not requiredat any stage of the purification process.
Tissue Expression (RT-PCR and Northern Analysis)
Clontech MTN blots I and III Northern blots contain-ing poly(A)1 RNA from 16 separate tissues werescreened with MEPE cDNA as described under Mate-rials and Methods, and no hybridization was detected.Using a total of 18 separate human tissue cDNAs astemplates for MEPE primer PCR, expression was de-tected in bone marrow and brain, with trace expressionin lung, kidney, and human placenta (Fig. 5). Sequenc-ing of the MEPE primer-amplified bands revealed com-plete identity to MEPE cDNA, and Southern screening
of the amplified bands with MEPE cDNA confirmed the
C
62 ROWE ET AL.
sequencing results (Fig. 6B). Also, a panel of normal-ized Clontech cDNAs derived from eight non-OHO tu-mors (Clontech human tumor panel K1422-1) were allnegative to MEPE PCR, except for very-low-level ex-pression in one case of colon adenocarcinoma, ovariancarcinoma, and prostatic carcinoma, respectively (de-tected after Southern screening of RT-PCR-amplifiedproducts) (data not shown). In sharp contrast, RT-PCRusing MEPE primers amplified poly(A)1 RNA, fromOHO tumors, from four separate patients, BD, DM,EM, and DS, indicating high levels of expression (nor-malized against glyceraldehyde 3-phosphate dehydro-genase and transferrin; Figs. 5 and 6). Poly(A)1 RNAfrom nonphosphaturic tumors (patients LA and LB)and control tissues from OHO patients (skin and ma-terial adjacent to tumors), a CL8 human renal cell line,and human primary osteoblast cells as well as poly(A)1
RNA extracted from a presumed tumor polyp from apatient with linear sebaceous nevus syndrome did not
FIG. 4. Sequence similarity analysis using “sim” and llanview mathe gap open penalty was set to 12, and the gap extension penalty wC (see Duret et al., 1996; Huang et al., 1990, 1992). The similarity scoshown on each protein scheme represent sequence homologies of .8there are five homology blocks in DSPP of .80% sequence similarithomology is also apparent for DMP1 and OPN versus MEPE (B and
amplify using MEPE-specific primers (Fig. 6). Low-
level expression was detected in osteosarcoma cell linesHTB96 and SaOS2 (data not shown).
OHO template poly(A)1 RNA from all OHO patientsconsistently amplified an expected band of 480 bp anda lower band of 190 bp. The upper band was confirmedby sequencing and Southern autoradiography as beingcompletely identical to MEPE sequence, and the lowerband was confirmed as a MEPE derivative by Southernanalysis (Fig. 6). The lower band did not appear in thelow-level expression found in normal tissues or non-OHO tumors. This finding indicates that alternativesplicing may play a role in the tumor-derived MEPE.Northern analysis of poly(A)1 RNA from OHO tumorpatient BD revealed a strong signal for a single band ofapproximately 2–2.1 kb (Fig. 7B).
In summary, high-level expression of MEPE (asmeasured by mRNA levels) was found only in OHOtumor samples (transcript size of 2–2.1 kb). Normaltissue expression was found in bone marrow and brain.
ematical and software tools (Duret et al., 1996). In each computation4. Comparison matrix for A was PAM40 and BLOSUM62 for B andthreshold was 70% in A and 40% in B and C. The highlighted blocksin A and .62% in B and C. Note that in MEPE versus DSPP (A),a single motif in MEPE (DSSESSDSGSSSES). A similar sequence
), and the MEPE ASARM motif is a feature of all three proteins.
thasre0%
y to
Evidence for very-low-level expression (possibly ec-
topic) was found in lung, kidney, placenta, and 3 of 11non-OHO tumors. Eight of 11 non-OHO tumors werenegative for MEPE mRNA expression (RT-PCR). Allresults were standardized against GA3PDH and trans-ferrin and confirmed by sequencing and or Southernanalysis of amplified bands.
Primer Extension and Northern Analysis
Primer extension using poly(A)1 RNA extracted frompatient BD, with primer (PEP) highlighted in Fig. 1and described under Materials and Methods, resultedin a 151-bp product (Fig. 7A). The expected size fromsequence already isolated and cloned was 146 bp (seeFig. 1). Thus, the sequence described in the legend toFig. 1 is essentially complete with approximately 5–6
TABLE 1
The Acidic Serine–Aspartate-Rich MEPE-AssociatedMotif (ASARM motif)
pancreatic adenocarcinoma. Bone marrow not included in selection sho
bp of 59 untranslated region missing. The entire codingequence is accounted for, and the 1989 bp is in closegreement with the Northern data (Fig. 7B).
apping of MEPE to Chromosome 4q21.1
To identify the human chromosome that containsEPE, the NIGMS somatic cell hybrid panel and the
uman-specific MEPE1 primer pair were used. Gellectrophoresis of the PCR-amplified hybrid DNA re-ealed a single 249-bp product corresponding to thehromosome 4-containing hybrid (not shown). To pro-ide finer localization of MEPE, radiation hybrid map-ing was performed by PCR analysis of the GeneBridgepanel with the MEPE1 primer pair. Individual PCRsere gel-electrophoresed and scored for the presence orbsence of a 249-bp band, and a data vector was com-iled. Analysis of the mapping results with the RHonsortium Chromosome 4 Gene Map indicated thatEPE localized between two ESTs, D4S2785
WI-6336) and D4S2844 (WI-3770), on human chromo-ome 4q21.1 (Fig. 8).
DISCUSSION
Appropriate renal modulation of phosphate uptakeand of vitamin D metabolism is essential for normalskeletal mineralization, and profound disturbances inthese processes occur in both OHO and familial rickets(HYP) (Rowe, 1997, 1998a, 1998b). Using expressionscreening of a tumor cDNA library, we have isolated acandidate tumor factor (MEPE) that may play a director an indirect role in the regulation of skeletal miner-alization, renal phosphate handling, and vitamin Dmetabolism. MEPE has a number of distinct features
that support this functional role: (1) MEPE is ex-pressed at high levels only in those tumors resectedfrom patients with OHO; (2) nonphosphaturic tumors,tissues removed from areas adjacent to OHO tumors,and control skin tissues are negative for MEPE mRNAexpression; (3) MEPE expression in bone marrow sug-gests that the gene product may play an active role inbone biochemistry or physiology; (4) a number of bone–dentin mineral matrix proteins (OPN, DMP1, DSSP,
FIG. 6. RT-PCR using MEPE and GA3PDH primers on poly(A)1
ene). Lanes marked M contain molecular weight markers. (A) Lauman-bone osteoblast enriched (H-OST), obtained from Clonetics;T-PCR-amplified samples (MEPE primers) screened with 32P-radi
control; 2, human renal cell line CL8; 3, water blank; 4, poly(A)1 RNsing GA3PDH and MEPE primers on poly(A)1 RNA from control
following: 1, nontumor tissue from OHO patient ND (peripheral to tupatient EM; 4, nonphosphaturic tumor polyp from linear sebaceous
FIG. 7. Size of MEPE mRNA in OHO tumor tissue deduced byrimer extension (A) and Northern analysis (B). (A) Key lanes of aingle gel subjected to different autoradiographic exposure times.ane C contains a fX174 HinfI digest end-labeled with polynucle-
otide kinase ([g-32P]ATP). The size calibrations (in bases) are indi-cated on the right-hand side of the figure. The PE arrow shows thata single product (;150 bases) was detectable in lane A after a 24-hexposure when using patient ND mRNA as template. This productwas absent in lane B, the water blank (negative control) primerextension reaction. Therefore the 59 end of the MEPE mRNA ex-pressed in the OHO tumor was determined to be approximately 126bases upstream of the 59 extremity of the PEP-1 primer binding siteon MEPE mRNA. (B) A single transcript of ;2–2.1 kb detected afterscreening OHO tumor poly(A)1 RNA (patient BD) with MEPE 32P-
adiolabeled cDNA.
BMP3, bone sialoprotein II (IBSP), etc.) map to thesame chromosomal location of MEPE (4q21.1). Thisintriguing cluster of genes that code for proteins withsimilar molecular–structural features and functionssuggests a shared evolutionary origin (Fig. 8).
MEPE is structurally characteristic of a phospho-glycoprotein with an RGD cell attachment sequencethat is situated in the center of the molecule. Thestructural RGD context in MEPE is similar to theRGD-containing bone mineral matrix glycoprotein os-teopontin, is flanked by b-sheet and a-helix, in a regionof a predicted turn, and contains a thrombin cleavagesite. There are a number of other structural similari-ties (glycosylation status; casein kinase, protein ki-nase, and tyrosine kinase phosphorylation motifs; anda size of approximately 60–75 kDa). The arginine–glycine–asparagine (RGD) tripeptide sequence is foundin a distinct number of extracellular matrix proteins(DSPP, DMP1, IBSP, DPP, fibronectin, vitronectin, os-teopontin, human integrin binding protein, and colla-gen) and coagulation factors (fibrinogen, throm-bospondin, and von Willebrand factor) (Yamada, 1991)and is prerequisite for binding to a number of integrinclasses. Adhesion–migration of different cell types ismediated through av-containing integrins (Weber etal., 1997; Nasu et al., 1995), and osteopontin has beenshown to interact with integrin classes avb3, avb5, avb1,and a4b1 and CD44 receptor. In X-linked rickets (HYP),casein kinase II phosphorylation of osteopontin is de-fective (Rifas et al., 1997), and it is noteworthy thatMEPE also has a number of potential casein kinase IIphosphorylation motifs (two associated with the C-ter-minal ASARM motif).
The mapping of MEPE to chromosome 4q21.1 in a
A from control cell lines and OHO patient ND tumor (used to clonecontain the following: 1, human kidney cell line CL8; 2, primaryater blank; 4, OHO patient BD poly(A)1 RNA. (B) Southern blot of
beled MEPE cDNA; lanes contain the following: 1, MEPE plasmidrom OHO patient ND (MEPE cloned from this source). (C) RT-PCRn-OHO tumors and OHO patients BD and EM; lanes contain ther); 2, tumor tissue from OHO patient DS; 3, tumor tissue from OHOus syndrome patient LS; 5, water blank.
protein) (DSP), DPP), BMP3, and IBSP is of interestgiven the molecular similarities of MEPE to dentin–bone extracellular matrix RGD containing phospho-glycoproteins (DSPP, DMP1, OPN, IBSP, DPP; Fig.8). A key molecular–structural feature is the pres-ence of an aspartate–serine-rich motif in the C-ter-minal portion of MEPE that has considerable se-quence homology to a repeated motif in DSPP,DMP1, and OPN (MacDougall et al., 1997). The as-
artate–serine motif (. . . DSSDSSDSSSSSDS . . .),is highly repeated in the DPP portion of DSPP, and asingle localized region of sequence homology occursin MEPE (C-terminus). Recently DPP and DSP wereshown to be cleavage products expressed from a sin-gle gene transcript encoding the larger parent mol-ecule DSPP (MacDougall et al., 1997). DPP is pro-posed to play a key role in dentin mineralization.Given that OPN and DMP1 also have similar motifsand are thought to be integral for dentin– bone min-
FIG. 8. Chromosomal location of MEPE. MEPE maps to the longCodes are translated as follows: 1, BMP3, bone morphogenetic protDMP-1 dentin matrix protein 1; 5, ANX, annexin; 6, IBSP, integrin biprotein 1B; 8, MEPE, matrix extracellular phosphoglycoprotein. Theat the Human Genome Mapping Centre (HGMP-MRC) at Web sitment) and the National Center for Biotechnology Information Web siMapview was used to construct and download maps from the HGMPthe chromosome 4 cytogenetic map.
eralization, it is possible to speculate on a comple-
mentary or similar role for MEPE and/or derivativepeptides. Recently, dentinogenesis imperfecta typeII and type III (DGI-II and DGI-III) have been fine-mapped to 4q21, with dentin sialoprotein and dentinmatrix protein proposed as possible candidate genes(MacDougall et al., 1999; Aplin et al., 1999). MEPEcan now also be considered a DG candidate giventhat it maps to the same region.
Relevant to the presence of a presumptive zinc met-alloendopeptidase cleavage site in the N-terminalMEPE region for PHEX/NEP is the presence of a single“putative” site for proprotein convertases I and II (Fig.3). The regulation of BMP activity and a number ofother hormones (pro-melanin, profibrillin, prothyro-tropin-releasing hormone, pro-alpha1(V) collagen) byproprotein convertases (PCs) via endoproteolytic cleav-age of inactive precursor proteins has been demon-strated and extensively documented (Constam andRobertson, 1999; Schaner et al., 1997; Raghunath et
of chromosome 4 (4q21.1) between markers D4S1534 and D4S3381.3; 2, DSPP, dentin sialo phospho protein; 3, OPN, osteopontin; 4,
ng sialo protein/bone sialoprotein II; 7, bone morphogenetic receptorromosome 4 map was obtained using data from and maps depositedttp://www.hgmp.mrc.ac.uk/gdb-bin/genera/genera/hgd/GenomicSeg-t http://www.ncbi.nlm.nih.gov/Entrez/Genome/main_genomes.html.
RC Web site. Maps used included the GDB comprehensive map and
armeinndiche hte a-M
al., 1999; Viale et al., 1999; Imamura et al., 1998).
66 ROWE ET AL.
PACE4, SPC1, and Furin, in particular, are PCs thathave been shown to initiate BMP activity by directendoproteolytic cleavage (Constam and Robertson,1999). Also, the expression pattern of specific pro-con-vertases during embryogenesis is dynamic, and colo-calization with BMPs occurs (Tsuji et al., 1999). Thus,it is possible that cleavage of the N-terminal cysteine-containing peptide of MEPE by PCs or PHEX mayinitiate activation or inactivation of MEPE function.
In addition to the isolation of six separate MEPEcDNA clones, screening of the cDNA library derivedfrom patient ND resulted in the parallel isolation ofthree cDNA clones that were identical to human clus-terin. Clusterin is produced by a wide variety of tissues(Jenne and Tschopp, 1992) and is constitutively ex-pressed in a number of tumor cell lines (Kyprianou andIsaacs, 1989; Dvergsten et al., 1994; Danik et al., 1991;French et al., 1994). One role proposed for clusterin isthe packaging and stabilization of peptide hormones orthe initiation of exocytosis and endocrine secretion.The presence of clusterin in patient ND may thereforeaffect the pathophysiology of the disease by interactingwith tumor-secreted MEPE.
A number of models have predicted that a new cir-culating hormone factor plays a major role in bonemetabolism, phosphate handling, and vitamin D me-tabolism (Rowe, 1997, 1998a; Econs and Drezner,1994; Nesbitt et al., 1999). Moreover, abnormal pro-cessing of this factor is thought to be responsible for thepathophysiology seen in OHO and HYP. The evidencepresented in this study suggests that MEPE is a goodcandidate for this novel bone–renal phosphate regula-tor. The presumed polyfunctional nature of this novelfactor (MEPE) is likely to be mediated by posttransla-tional processing. The normal processing mechanismsare envisaged to include specific sequential cleavageand perhaps phosphorylation of derivative MEPE pep-tides. Derivative MEPE peptides would then be des-tined for distinct biochemical and physiological func-tions. Also, the processing may be exquisitelycontrolled by other factors impacting signal transduc-tion pathways and gene expression (phosphate, cal-cium, calcitriol, etc.). In OHO tumors, it is possible tospeculate that MEPE processing may not be appropri-ate, resulting in a larger molecule (aberrant cleavage)or an abnormally processed form (abnormal phosphor-ylation). Once secreted into the circulation, processingof tumor MEPE is proposed not to occur (occurs onlywithin a target cell). The abnormal tumor-secretedform is then predicted to interfere with normal physi-ology by competing directly for renal receptors and/orbone receptors with normally processed MEPE. Alter-natively, tumor MEPE may activate specific receptorsdirectly or indirectly, resulting in abnormal renal phos-phate handling, bone mineralization, and vitamin Dmetabolism. PHEX may play a role in the processing ofthis factor directly or indirectly. If this were the case,
then defective PHEX in familial rickets would be pre-
dicted to cause inappropriately processed MEPE (al-tered cleavage and/or phosphorylation). The resultingabnormal MEPE in HYP may therefore be similar (butperhaps not identical) to tumor MEPE. Although theabove schemes are speculative, a new reagent (MEPE)now exists to facilitate further exploration of the com-plex processes controlling bone–renal physiology andOHO tumor pathophysiology.
In conclusion, a new factor (MEPE) has been clonedfrom an OHO tumor that is an RGD-containing extra-cellular phospho-glycoprotein. This factor is specifi-cally up-regulated in OHO tumors, has structural–molecular similarities to a number of extracellularmatrix bone–tooth proteins (OPN, DMP1, DSPP, andIBSP), and maps to the same region of chromosome 4q.Expression in nonphosphaturic tumors is very low ornonexistent, with normal tissue expression in bonemarrow and brain. Trace expression of MEPE wasfound in, kidney, lung, and placenta. Further experi-ments are needed to characterize fully the function ofthis new bone–dentin extracellular matrix-like proteinin vivo and in vitro.
ACKNOWLEDGMENTS
The authors acknowledge with gratitude the support of the Med-ical Research Council of the United Kingdom (Senior FellowshipPSNR), the European Commission (Gene CT930027), and NIHGrants R01AR4228, K24AR02095, and AR08550. We extend ourthanks to all the clinicians who have contributed OHO tumors andclinical samples (Dr. Trevor Stamp, Mr. Briggs, Mr. Crockard, Pro-fessor Michelle Garabedian, Professor Ewa Pronicka, Dr. FrederickGilkey, Dr. Mark A. Goodman, Mr. Mayo, Dr. Mike Dillon, and Dr.Albert Ong). Also, we are grateful for the support and help of Pro-fessor J.L.H. O’Riordan, Dr. Luce Condamine, and Dr. EwaPopowska.
REFERENCES
Aplin, H. M., Hirst, K. L., and Dixon, M. J. (1999). Refinement of thedentinogenesis imperfecta type II locus to an interval of less than2 centimorgans at chromosome 4q21 and the creation of a yeastartificial chromosome contig of the critical region. J. Dent. Res. 78:1270–1276.
Aschinberg, L. C., Solomon, L. M., Zeis, P. M., Justice, P., andRosenthal, I. M. (1977). Vitamin D resistant rickets associatedwith epidermal nevus syndrome: Demonstration of a phosphaturicsubstance in the dermal lesions. J. Paediatr. 91(1): 56–60.
Cai, Q., Hodgson, S. F., Kao, P. C., Lennon, V. A., Klee, G. G.,Zinsmiester, A. R., and Kumar, R. (1994). Brief report: Inhibitionof renal phosphate transport by a tumor product in a patient withoncogenic osteomalacia. N. Engl. J. Med. 330: 1645–1649. [Seecomments]
Constam, D. B., and Robertson, E. J. (1999). Regulation of bonemorphogenetic protein activity by pro domains and proproteinconvertases. J. Cell Biol. 144: 139–149.
Danik, M., Chabot, J. G., Mercier, C., Benabid, A. L., Chauvin, C.,Quirion, R., and Suh, M. (1991). Human gliomas and epileptic fociexpress high levels of a mRNA related to rat testicular sulfatedglycoprotein 2, a purported marker of cell death. Proc. Natl. Acad.Sci. USA 88: 8577–8581.
David, K., Revesz, T., Kratimenos, G., Krausz, T., and Crockard,
H. A. (1996). Oncogenic osteomalacia associated with a meningeal
phosphaturic mesenchymal tumor. J. Neurosurg. 84: 288–292.[Case report]
Deleague, G. (1997). Software for protein analysis: AntheroplotV2.5e. Microsoft group. 7 Passage du Vercours 69–367, VercorsLyon Cedex 07.
Denhardt, D. T., and Guo, X. (1993). Osteopontin: A protein withdiverse functions. FASEB J. 7: 1475–1482.
Drezner, M. K. (1990). Tumour-associated rickets and osteomalacia.In “Primer on Metabolic Bone Diseases and Disorders of MineralMetabolism” (M. J. Favus, Ed.), pp. 184–188. Am. Soc. Bone andMin. Res., Kelseyville, CA.
Drwinga, H. L., Toji, L. H., Kim, C. H., Greene, A. E., and Mulivor,R. A. (1993). NIGMS human/rodent somatic cell hybrid mappingpanels 1 and 2. Genomics 16: 311–314.
Duret, L., Gasteiger, E., and Perriere, G. (1996). LALNVIEW: Agraphical viewer for pairwise sequence alignments. Comput. Appl.Biosci. 12: 507–510.
Dvergsten, J., Manivel, J. C., Correa-Rotter, R., and Rosenberg,M. E. (1994). Expression of clusterin in human renal diseases.Kidney Int. 45: 828–835.
Econs, M. J., and Drezner, M. K. (1994). Tumour induced osteoma-lacia—Unveiling a new hormone. N. Engl. J. Med. 330(23): 1679–1681.
Francis, F., Strom, T. M., Hennig, S., Boeddrich, A., Lorenz, B.,Brandau, O., Mohnike, K. L., Cagnoli, M., Steffens, C., Klages, S.,Borzym, K., Pohl, T., Oudet, C., Econs, M. J., Rowe, P. S. N.,Reinhardt, R., Meitinger, T., and Lehrach, H. (1997). Genomicorganisation of the human PEX gene mutated in X-linked domi-nant hypophosphataemic rickets. Genome Res. 7(6): 573–585.
Francis, R. M., and Selby, P. L. (1997). Osteomalacia. Baillieres Clin.Endocrinol. Metab. 11: 145–163.
French, L. E., Wohlwend, A., Sappino, A. P., Tschopp, J., andSchifferli, J. A. (1994). Human clusterin gene expression is con-fined to surviving cells during in vitro programmed cell death.J. Clin. Invest. 93: 877–884.
Gyapay, G., Schmitt, K., Fizames, C., Jones, H., Vega-Czarny, N.,Spillett, D., Muselet, D., Prud’Homme, J. F., Dib, C., Auffray, C.,Morissette, J., Weissenbach, J., and Goodfellow, P. N. (1996). Aradiation hybrid map of the human genome. Hum. Mol. Genet. 5:339–346.uang, X., Miller, W., Schwartz, S., and Hardison, R. C. (1992).Parallelization of a local similarity algorithm. Comput. Appl. Bio-sci. 8: 155–165.uang, X. Q., Hardison, R. C., and Miller, W. (1990). A space-efficient algorithm for local similarities. Comput. Appl. Biosci. 6:373–381.YP Consortium, Francis, F., Hennig, S., Korn, B., Reinhardt, R., deJong, D., Poustka, A., Lehrach, H., Rowe, P. S. N., Goulding, J. N.,Summerfield, T., Mountford, R. C., Read, A. P., Popowska, E.,Pronicka, E., Davies, K. E., O’Riordan, J. L. H., Econs, M. J.,Nesbitt, T., Drezner, M. K., Oudet, C., Pannetier, S., Hanauer, A.,Strom, T. M., Meindl, A., Lorenz, B., Cagnoli, M., Mohnike, K. L.,Murken, J., and Meitinger, T. (1995). A gene (PEX) with homolo-gies to endopeptidases is mutated in patients with X-linked hy-pophosphatemic rickets. The HYP Consortium. Nat. Genet. 11:130–136.
mamura, Y., Steiglitz, B. M., and Greenspan, D. S. (1998). Bonemorphogenetic protein-1 processes the NH2-terminal propeptide,and a furin-like proprotein convertase processes the COOH-termi-nal propeptide of pro-alpha1(V) collagen. J. Biol. Chem. 273:27511–27517.
Ioakimidis, D., Dendrinos, G. K., Frangia, K. B., Babiolakis, D. N.,Chilas, G. I., and Lyberopoulos, K. L. A.-K. (1994). Tumour in-
duced osteomalacia. J. Rheumatol. 21(6): 1162–1164.
Jenne, D. E., and Tschopp, J. (1992). Clusterin: The intriguing guisesof a widely expressed glycoprotein. Trends Biochem. Sci. 17: 154–159.
Katagiri, Y. U., and Uede, T. (1998). [Functional domain and itsreceptor of osteopontin]. Seikagaku 70: 253–264.
Kozak, M. (1996). Interpreting cDNA sequences: Some insights fromstudies on translation. Mamm. Genome 7: 563–574.
Kyprianou, N., and Isaacs, J. T. (1989). “Thymineless” death inandrogen-independent prostatic cancer cells. Biochem. Biophys.Res. Commun. 165: 73–81.
Lyles, K. W., Berry, W. R., Haussler, M. R., Harrelson, J. M., andDrezner, M. K. (1980). Hypophosphataemic osteomalacia: Associ-ation with prostrate carcinoma. Ann. Intern. Med. 93: 275–278.
MacDougall, M., Jeffords, L. G., Gu, T. T., Knight, C. B., Frei, G.,Reus, B. E., Otterud, B., Leppert, M., and Leach, R. J. (1999).Genetic linkage of the dentinogenesis imperfecta type III locus tochromosome 4q. J. Dent. Res. 78: 1277–1282.
MacDougall, M., Simmons, D., Luan, X., Nydegger, J., Feng, J., andGu, T. T. (1997). Dentin phosphoprotein and dentin sialoproteinare cleavage products expressed from a single transcript coded bya gene on human chromosome 4. Dentin phosphoprotein DNAsequence determination. J. Biol. Chem. 272: 835–842.iyauchi, A., Fukase, M., Tsutsumi, M., and Fujita, T. (1988). He-mangiopericytoma induced osteomalacia: Tumour transplantationin nude mice causes hypophosphataemia and tumour extractsinhibit renal 25-hydroxyvitamin D a1-hydroxylase activity.J. Clin. Endocrinol. Metab. 67(1): 46–53.asu, K., Ishida, T., Setoguchi, M., Higuchi, Y., Akizuki, S., andYamamoto, S. (1995). Expression of wild-type and mutated rabbitosteopontin in Escherichia coli, and their effects on adhesion andmigration of P388D1 cells. Biochem. J. 307: 257–265.esbitt, T., Fujiwara, I., Thomas, R., Xiao, Z. S., Quarles, L. D., andDrezner, M. K. (1999). Coordinated maturational regulation ofPHEX and renal phosphate transport inhibitory activity: Evidencefor the pathophysiological role of PHEX in X-linked hypophos-phatemia. J. Bone Miner. Res. 14: 2027–2035.
opovtzer, M. M. (1981). Tumour induced hypophosphataemic osteo-malacia: Evidence for a phosphaturic cyclic AMP-independent ac-tion of tumour extract. Clin. Res. 29: 418A [Abstract]
aghunath, M., Putnam, E. A., Ritty, T., Hamstra, D., Park, E. S.,Tschodrich-Rotter, M., Peters, R., Rehemtulla, A., and Milewicz,D. M. (1999). Carboxy-terminal conversion of profibrillin to fibril-lin at a basic site by PACE/furin-like activity required for incor-poration in the matrix. J. Cell Sci. 112: 1093–1100.awlings, N. D., and Barrett, A. J. (1999). MEROPS: The peptidasedatabase. Nucleic Acids Res. 27: 325–331.ice, P. (1995). Programme Manual for the EGCG Package. Univer-sity of Cambridge, Cambridge, UK.ifas, L., Cheng, S., Halstead, L. R., Gupta, A., Hruska, K. A., andAvioli, L. V. (1997). Skeletal casein kinase activity defect in theHYP mouse. Calcif. Tissue Int. 61: 256–259.owe, P. S. (1997). The PEX gene: Its role in X-linked rickets,osteomalacia, and bone mineral metabolism. Exp. Nephrol. 5: 355–363.owe, P. S. N. (1994). Molecular biology of hypophosphataemic rick-ets and oncotenic osteomalacia. Hum. Genet. 94(5): 457–467.owe, P. S. N. (1998a). The role of the PHEX gene (PEX), in familieswith X-linked hypophosphataemic rickets. Curr. Opin. Nephrol.Hypertension 7(4): 367–376.owe, P. S. N. (1998b). X-linked rickets and tumour osteomalacia:PHEX and the missing link. Clin. Exp. Nephrol. 2(3): 183–193.owe, P. S. N., Francis, F., and Goulding, J. (1994a). Rapid isolationof DNA sequences flanking microsatellite repeats. Nucleic Acids
Res. 22(23): 5135–5136.
R
R
S
S
S
T
T
U
V
W
W
Y
68 ROWE ET AL.
Rowe, P. S. N., Goulding, J., Read, A., Lehrach, H., Francis, F.,Hanauer, A., Oudet, C., Biancalana, V., Kooh, S. W., Davies, K. E.,et al. (1994b). Refining the genetic map for the region flanking theX-linked hypophosphataemic rickets locus (Xp22.1–22.2). Hum.Genet. 93: 291–294.
Rowe, P. S. N., Goulding, J. N., Francis, F., Oudet, C., Econs, M. J.,Hanauer, A., Lehrach, H., Read, A. P., Mountford, R. C., Summer-field, T., Weissenbach, J., Fraser, W., Drezner, M. K., Davies,K. E., and O’Riordan, J. L. (1996a). The gene for X-linked hy-pophosphataemic rickets maps to a 200–300 kb region in Xp22.1,and is located on a single YAC containing a putative vitamin Dresponse element (VDRE). Hum. Genet. 97: 345–352.
owe, P. S. N., Ong, A., Cockerill, F. J., Goulding, J. N., and Hewi-son, M. (1996b). Candidate 56 and 58 kDa protein(s) responsiblefor mediating the renal defects in oncogenic hypophosphataemicosteomalacia. Bone 18(2): 159–169.
owe, P. S. N., Oudet, C., Francis, F., Sinding, C., Pannetier, S.,Econs, M. J., Strom, T. M., Meitinger, T., Garabedian, M., David,A., Macher, M.-A., Questiaux, E., Popowska, E., Pronicka, E.,Read, A. P., Mokrzycki, A., Glorieux, F. H., Drezner, M. K.,Hanauer, A., Lehrach, H., Goulding, J., and O’Riordan, J. L. H.(1997). Distribution of mutations in the PEX gene in families withX-linked hypophosphataemic rickets (HYP). Hum. Mol. Genet. 6:539–549.
chaner, P., Todd, R. B., Seidah, N. G., and Nillni, E. A. (1997).Processing of prothyrotropin-releasing hormone by the family ofprohormone convertases. J. Biol. Chem. 272: 19958–19968.
hane, E., Parisien, M., Henderson, J. E., Dempster, D. W., Feld-man, F., Hardy, M. A., Tohme, J. F., Karaplis, A. C., and Clemens,T. L. (1997). Tumor-induced osteomalacia: Clinical and basic stud-
ies. J. Bone Miner. Res. 12: 1502–1511.
mith, L. L., and Giachelli, C. M. (1998). Structural requirements foralpha 9 beta 1-mediated adhesion and migration to thrombin-cleaved osteopontin. Exp. Cell Res. 242: 351–360.
suji, A., Hashimoto, E., Ikoma, T., Taniguchi, T., Mori, K., Naga-hama, M., and Matsuda, Y. (1999). Inactivation of proprotein con-vertase, PACE4, by alpha1-antitrypsin portland (alpha1-PDX), ablocker of proteolytic activation of bone morphogenetic proteinduring embryogenesis: Evidence that PACE4 is able to form anSDS-stable acyl intermediate with alpha1-PDX. J. Biochem. 126:591–603.
urner, A. J., and Tanzawa, K. (1997). Mammalian membrane met-allopeptidases: NEP, ECE, KELL, and PEX. FASEB J. 11: 355–364.ede, T., Katagiri, Y., Iizuka, J., and Murakami, M. (1997). Os-teopontin, a coordinator of host defense system: A cytokine or anextracellular adhesive protein? Microbiol. Immunol. 41: 641–648.iale, A., Ortola, C., Hervieu, G., Furuta, M., Barbero, P., Steiner,D. F., Seidah, N. G., and Nahon, J. L. (1999). Cellular localizationand role of prohormone convertases in the processing of pro-mel-anin concentrating hormone in mammals. J. Biol. Chem. 274:6536–6545.eber, G. F., Ashkar, S., and Cantor, H. (1997). Interaction betweenCD44 and osteopontin as a potential basis for metastasis forma-tion. Proc. Assoc. Am. Physicians 109: 1–9.eidner, N., and Cruz, D. S. (1987). Phosphaturic mesenchymaltumours: A polymorphous group causing osteomalacia rickets.Cancer 59(8): 1442–1442.
amada, K. M. (1991). Adhesive recognition sequences. J. Biol.
