- Email: [email protected]
Molecular Cloning and Characterization of LR3, a Novel LDL Receptor Family Protein with Mitogenic Activity
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS ARTICLE NO.
251, 784 –790 (1998)
RC989545
Molecular Cloning and Characterization of LR3, a Novel LDL Receptor Family Protein with Mitogenic Activity Yu Dong,1,2 William Lathrop,1 Daniel Weaver, Qingqing Qiu,3 John Cini, Donald Bertolini,2 and David Chen4 Bayer Research Center, Pharmaceutical Division, Bayer Corporation, West Haven, Connecticut 06516-4175
Received September 16, 1998
We report molecular cloning and initial functional characterization of a novel member of the low density lipoprotein receptor (LDLR) gene family. The cDNA was isolated from a human osteoblast cDNA library and encoded a 1,615 amino acids protein designated as LR3. It has, in the extracellular region, a cluster of three LDLR ligand binding repeats at a juxtamembrane position and four EGF precursor homology domains separated by YWTD spacer repeats. The entire ectodomain shares the same modular organization with the middle portion of the extracellular regions of two LDLR family members, LDLR-related protein (LRP), and gp330/megalin. LR3 mRNA was expressed in most of the adult and fetal tissues examined. The highest expression level was seen in aorta. In human osteosarcoma cells examined, LR3 mRNA was highly enriched in TE85 cells, moderately expressed in MG63 cells and primary human osteoblasts, and undetectable in SaOS-2 cells. NIH 3T3 cells transfected with either full length LR3 or its ectodomain showed significantly increased proliferation, whereas transfection of intracellular domain had no proliferative effect. We predict that LR3 is a multi-functional protein with potential mitogenic activity. © 1998 Academic Press
The low density lipoprotein receptor (LDLR) family is comprised of a group of related glycoprotein receptors which interact with a diverse array of ligands. Since the isolation of human LDLR cDNA, the first member of the family, in 1984 (1), genes for 5 additional human receptors have been identified. These 1
Equal contribution was made by these two authors. Current address: Pharmaceutical Research Institute, BristolMyers Squibb, P O Box 4000, Princeton, NJ 08543-4000. 3 Current address: Genome Therapeutics Co., 100 Beaver Street, Waltham, MA 02154. 4 To whom correspondence should be addressed at Bayer Research Center, Pharmaceutical Division, Bayer Corporation, 400 Morgan Lane, West Haven, CT 06516-4175. Fax: (203) 937-6923. E-mail: [email protected]. 2
0006-291X/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
include closely related VLDLR (2) and apolipoprotein E receptor 2 (apoER2)/LR7B (3, 4), as well as three larger and much more complex relatives, LR11/Sorl-1 (5, 6), LDLR-related protein (LRP)/a2-macroglobulin receptor (7), and glycoprotein 330 (gp330)/megalin (8). Homologues of these proteins and similar proteins such as vitellogenin receptor have been identified in species ranging from Caenorhabditis elegans and insects to birds and other mammals (9 –13). The overall homology among these receptors is, in general, quite moderate. However, all of them share the following common structural features: (i) one or several clusters of LDLR ligand-binding domains; (ii) cysteine-rich EGF-like domains separated by (F/Y)WTD spacer repeats; (iii) a single membrane-spanning region; (iv) one or several FXNPXY coated-pit internalization signal in the cytoplasmic tail. The functions of the LDLR in endocytosis of plasma lipoprotein and cholesterol homeostasis have been elegantly elucidated largely by Michael Brown and Joseph Goldstein in their classical work of defining the molecular defect of familial hypercholesterolemia (14). Many structurally diverse macromolecules, such as lipoprotein lipase, tissue-type plasminogen activator, urokinase plasminogen activator, Pseudomonas exotoxin A, and lactoferrin, have later been identified as ligands for LRP, gp330 and other receptors, and roles of these receptors in LDL transport and lipid metabolism have been investigated in a variety of species (for reviews, see Ref. 15, 16). Other functions of these receptors, such as signal transduction and Ca21 sensing activities, however, remain less well defined. In an effort to identify novel genes from human osteoblasts, the bone formation cells, we obtained several cDNA clones with unique sequences by a PCR based approach (unpublished data). One of them, HOB-28, showed high homology to LRP, gp330 and other members of the LDLR family, suggesting it might encode a new family member. Here, we report the cloning and sequence analysis of the full-length cDNA of this pro-
784
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
tein, designated as LR3, and initial characterization of its gene expression and cellular functions. MATERIALS AND METHODS Screening of human osteoblast cDNA library and sequencing analysis. By homology RT-PCR experiments and sequencing of randomly picked clones, we obtained clone HOB-28 from human osteoblast RNA (unpublished data). HOB-28 is 310 bp in length and has an open reading frame (ORF). This fragment was labeled with a32PdCTP (3000 Ci/mmol, Amersham Life Science) by random priming and used as a probe in both cDNA library screening and RNA blot hybridization. Approximately two million plaques from a lambda ZAP II human osteoblast cDNA library (a generous gift from Dr. Marian Young, National Institutes of Health) were screened according to the standard protocol (17). Positive clones were obtained by three rounds of screening and plaque purification. The phagemid DNA was excised from the Lambda DNA using the Rapid Excision kit (Stratagene). All clones were subjected to partial sequencing analysis using an ABI 377 DNA sequencer (Perkin-Elmer). Sequence comparison and homology searches in nucleotide and protein databases were performed with the University of Wisconsin Genetics Computer Group programs. Expression constructs. The full length LR3 cDNA and the sequences encoding two truncated forms were subcloned into mammalian expression vectors. LR3F was constructed by inserting a cDNA fragment with the entire ORF into pcDNA3.1/V5-His (Invitrogen). The original stop codon was mutated so that the V5 epitope and polyhistidine tag would be fused to the carboxyl terminus. LR3N, which contained the coding region for the LR3 extracellular region (amino acids 1 to 1388, Fig. 1), was obtained by releasing an ApaI fragment from LR3F and inserting the fragment into the ApaI site in pcDNA3.1/V5-His. The third expression construct, LR3C, encoded the fourth EGF-like domain, the transmembrane segment and the entire carboxyl terminus (amino acids 1079 to 1615, Fig. 1). This construct was made by cutting LR3F with EcoRI and KpnI, and cloning the released fragment into the same sites in pSecTag (Invitrogen). These constructs were confirmed by DNA sequencing. Cell culture, transfection, and measurement of cell proliferation. NIH3T3 cells and three human osteoblast-like osteosarcoma cell lines, MG63, TE85, and SaOS-2, were obtained from the American Type Culture Collection and maintained in DMEM medium plus 2 mM glutamine and 10 % fetal bovine serum (FBS). Normal human fibroblasts were obtained from Clonetics and cultured as suggested by the supplier. The expression constructs and pcDNA3.1/V5-His were introduced into NIH3T3 cells by Lipofectamine reagent following the manufacturer’s instruction (GIBCO-BRL). Stable transfectants were established by selecting the cells with either 500 mg/ml G418 (LR3F, LR3N and pcDNA3), or 200 mg/ml Zeocin (LR3C). To measure the rate if cell proliferation, each transfectant and untransfected NIH3T3 cells were seeded at 3,000 cells/well in two 96 well plates in DMEM supplemented with 10% FBS. The medium was replaced with DMEM plus 0.5% charcoal-Dextran treated FBS the next day and the plates were used in 72 hours. The cell proliferative rate was measured by BrdU incorporation in DNA using an ELISA kit under the conditions recommended by manufacturer (Boehringer Mannheim). RNA isolation and northern/dot blot analyses. Total RNA was isolated from subconfluent MG63, TE85, SaOS-2 cells, human fibroblasts and osteoblasts using TRIzol Reagent (GIBCO-BRL). Ten mg of total RNA from each cell line and from human kidney and placenta (Clontech) were separated on 1.2 % agarose- formaldehyde gels and transferred onto Hybond-N filters. The filters were prehybridized in 6XSSC, 5X Denhardt’s, 50 % formamide, 100 mg/ml salmon sperm DNA at 42°C for 4 hours and then hybridized to labeled HOB-28 at 42°C overnight. The Human RNA Master Blot and a human multiple
FIG. 1. Amino acid sequence of human LR3. The amino acid sequence is deduced from the cDNA sequence (GenBank accession number AF077820). The signal sequence, four class B.2 EGF-like domains, three LDLR ligand binding domains, the transmembrane segment, and five potential SH-3 binding motifs are indicated above the corresponding sequences. The YWTD tetrapeptides in the spacer repeats are underlined. Notably, three of the last five YWTD are less conserved.
tissue northern blot (Clontech) were hybridized with the same probe in ExpressHyb solution (Clontech) at 60°C overnight. After sequential washes, the filters were autoradiographed at -70°C on Kodak X-OMAT films with intensifying screens. The RNA level in each sample was normalized to b-actin mRNA. Density of the dot blot spots was measured and quantified on an imaging densitometer (Bio-Rad). In vitro transcription-translation and western blot analysis. The TNT rabbit reticulocyte lysate system (Promega) was used to synthesize LR3 protein in vitro. pcDNA3 with or without the LR3 ORF (1 mg) was incubated with the lysate in the presence of 35Smethionine (1000 Ci/mmol, Amersham Life Science) at 37°C for 1 hour. The translated products were then analyzed on SDS-PAGE and visualized by autoradiograph. To detect the expressed LR3 and its truncated forms in transfected NIH3T3 cells, subconfluent transfectants were lysed and dounce homogenized at 4°C in Tris buffered saline, pH 8.0 (TBS) with 1% Tween 20/1% NP40 and a protease inhibitor cocktail (Boehringer
785
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 2. Comparison of the sequences of the EGF-like domains, ligand binding domains and SH-3 binding domain in LR3 with their consensus sequences. (A) EGF-like domains. The consensus sequence is derived from class B.2 domains in LDLR and LRP (7; also see Ref. 18). The 6 absolutely conserved cysteines are in bold in the consensus sequence and the dashes represent variable amino acids. Gaps are introduced to optimize the alignment. (B) ligand binding domains. The consensus sequences is from Ullman et al. (19) and Sappington et al. (20). The amino acids that are conserved in more than 90% of the 194 repeats from which the consensus is derived are in bold. The rest of the indicated amino acids are conserved in at least 50% of those repeats. The Ser-Asp-Glu (SDE) motifs are underlined. (C), SH3-binding motifs. The consensus sequence is from Yu et al. (25). Two conserved prolines are in bold. Lower case p indicates that a proline is preferred at the position. F represents any hydrophobic amino acids and X can be any amino acids.
Mannheim). The homogenates were centrifuged at 20,000x g for 20 minutes. The supernatants were passed through Talon metal affinity resin columns (Clontech) and washed with TBS plus 1 mM imidazole. The proteins were then eluted off the columns with the same buffer containing 100 mM imidazole. Both the cell lysates and the eluted fractions were analyzed on 10% Tricine polyacrylamide precast gels (Novex). The proteins were transferred onto nitrocellulose membranes and blotted with either Anti-V5 or Anti-myc mouse antibodies (Invitrogen) at 4°C. To detect whether LR3N was present in the culture medium, 1 ml of conditioned medium was mixed with 25 ml of Talon metal affinity resin and incubated at 4°C for 2 hours. The resin was washed three times with TSB plus 1% Tween 20, boiled at 98°C in a sample buffer and pelleted by low speed centrifugation. The supernatant was analyzed as described above.
RESULTS Isolation of LR3 from a human osteoblast cDNA library. Using radiolabeled HOB-28 as a probe, we obtained 15 positive clones from approximately two million phages of a human osteoblast cDNA library. One clone with an insert of .5 kb was subjected to complete sequence analysis by primer walking in both directions. A dbEST search found that more than 25 human ESTs from libraries of several different tissues and cell types had at least 98% identity to this sequence. The cDNA clone has an ORF of 4,845 bp, a 59 UTR of 74 bp, and a 39 UTR of 187 bp including a polyadenylation signal, AATAAA (GenBank Accession No. AF077820). The sequence of 1,615 amino acids deduced from the ORF is shown in Fig. 1. The protein has a calculated molecular weight of 178 kDa and an isoelectric point of 5.1. The ORF is confirmed by analysis of the protein
synthesized in rabbit reticulocyte lysate. A prominent band migrating at approximately 180 kD was revealed by SDS-PAGE (data not shown). The primary structure predicts a type I transmembrane protein consisting of a signal peptide of 24 amino acids, a 1,359 amino acid extracellular region, a single membrane-spanning region of 30 amino acids, and a 202 amino acid intracellular domain (Fig. 1). Database searching indicated that this sequence is most homologous to the LDLR family members and contains the same type of structural modules. These peptide modules, approximately 40 amino acid residues in length, consist of 6 conserved cysteines with distinctive spacing between the cysteines in each type (18 –20). There are three type A LDLR ligand binding domains in one cluster and four EGF-like domains separated by YWTD-spacer repeats N-terminal to the ligand binding domains (Fig. 1, Fig. 2A, 2B). In concordance with the proposed nomenclature system (20, 21), this protein was designated LR3 to reflect the total number of ligand binding domains. Alignment of the EGF-like domains in LR3 with a consensus sequence derived from those in LDLR and LRP (7) indicated that all four domains belong to class B.2 (Fig. 2A). Comparing the ligand binding domains with a consensus sequence derived from 194 type A repeats (19) showed that those in LR3 have all 11 amino acids that are conserved in over 90% of the repeats, and 16 of the 19 amino acids that are conserved in at least 50% of the repeats (Fig. 2B). The Ser-Asp-Glu motif between the fifth and sixth
786
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
cysteines are also perfectly conserved in all three LR3 ligand binding repeats (Fig. 2B). This negatively charged tripeptide is implicated for high-affinity binding of positively charged sequences in ligands such as LDL and b-VLDL (22, 23). The YWTD-spacers (about 50 amino acids apart) are organized in tandem in a group of five, identical to the pattern seen in other LDLR proteins and the EGF precursor (Fig. 1, Fig. 5). The cytoplasmic tail of LR3 does not have a conserved NPXY motif, a common internalization signal sequence present in other members of the LDLR gene family for coated pit-mediated endocytosis (24). On the other hand, up to five potential Src-homology 3 (SH3) binding sites, including two that match perfectly with the consensus sequence XpfPpXP, are present in this region (Fig. 2C; 25). LR3 mRNA expression and tissue distribution. We first analyzed the expression of LR3 mRNA in a variety of adult and fetal tissues by dot blot analysis. Fig. 3A shows the result obtained by hybridizing a multiple tissue poly A1 RNA dot blot with HOB-28 probe. The LR3 mRNA could be detected in most of the tissues on the blot, except testis, bone marrow, peripheral leukocyte and various regions of the brain. The highest expression levels were found in aorta. In addition, the tissues of the female reproductive system, including uterus, mammary gland, ovary and placenta all showed relatively abundant LR3 mRNA (Fig. 3A). Differential expression levels of LR3 mRNA in fetal and adult tissues were noticeable in spleen, lung and kidney, with higher expression levels seen in fetal tissues (Fig. 3A). The different expression levels between adult and fetal heart could result from the inclusion of aorta tissue in the fetal heart. We then examined the LR3 mRNA level in selected tissues and bone cell lines using northern blot analysis. Fig. 3B shows that the HOB-28 detected a single transcript of about 5.4 kb in all tissues except brain. Heart, liver, and lung showed higher expression levels comparing with other tissues. It is possible that the high mRNA level in heart was caused by the aorta tissue attached. A striking difference in the expression levels was observed in three human osteosarcoma cell lines
FIG. 3. Expression of LR3 in tissues and osteosarcoma cell lines. (A) dot blot analysis of the expression of LR3 in different tissues.
The intensity of each dot was measured by an imaging densitometer and normalized by the amount of RNA in each dot (100 to 500 ng/dot). The expression levels in fetal and adult tissues are represented by open and filled bars, respectively. The expression levels are relative to that in aorta. (B) tissue distribution of LR3 analyzed by northern blot. Two micrograms of poly A1 RNA from the indicted tissues were probed with HOB-28. Hybridizing the same blot with a b-actin probe indicated that significantly greater amount of RNA from skeletal muscle and pancreas was blotted on the filter (not shown). (C) expression of LR3 in human osteoblastic cells. Ten micrograms of total RNA prepared from each indicated tissue or cell line were probed with HOB-28. Hybridizing the same filter with a b-actin probe estimated a roughly equal loading (not shown).
787
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
the cells could be nonspecific or via binding to a receptor on the plasma membrane.
TABLE 1
DNA Synthesis Rates in NIT3T3 Cells Transfected with LR3 Expression Constructs
DISCUSSION
BrdU incorporation (%) Cell type
Experiment I
Experiment II
pcDNA3.1 LR3F LR3N LR3C NIH3T3 (.5%) NIH3T3 (1%) NIH3T3 (10%)
100 161.8 6 18.8 192.3 6 6.8 96.1 6 14.7 127.5 6 7.3 155.8 6 16.2 231.4 6 15.2
100 130.6 6 8.7 136.6 6 10.2 80.1 6 6.8 103.1 6 5.5 107.1 6 3.1 148.8 6 9.1
Note. Untransfected NIH3T3 cells cultured in DMEM supplemented with 1% and 10% FBS were included as positive controls. NIH3T3 cells in DMEM with 0.5% charcoal-Dextran treated FBS were served as serum starved controls. Results are from two independent experiments.
(Fig. 3C). LR3 was highly expressed in TE85 cells, while undetectable in SaOS-2 cells. Since the same amount of RNA from these cells and human tissues was blotted on the same filter, it was estimated that the LR3 mRNA level in TE85 cells was about 10-fold higher than that in liver, and the expression levels in MG63 cells and kidney were comparable. The expression of LR3 in STRO-1 positive human primary osteoblasts was at a level similar to that of MG63 (data not shown), but the expression was undetectable in normal human fibroblast cells (Fig. 3C). Cell proliferative effect of LR3. When an expression construct LR3F which contains the full length LR3 ORF was transfected into murine NIH3T3 cells, an increased cell proliferation was noticed. Measurement of BrdU incorporation rate also indicated a significant increase in DNA synthesis. Constructs that encoded either the amino or carboxyl terminus truncated LR3 were designed to map the structure(s) that mediated the mitogenic activity. Two independent experiments showed that NIH3T3 cells transfected with LR3N, which encoded the entire LR3 extracellular region, had a BrdU incorporation rate similar to that with 10% FBS stimulation (Table 1). In contrast, cells transfected with LR3C, which encoded an N-terminal deleted LR3 (missing the first three EGF-like domains), showed a slightly lower DNA synthesis rate (Table 1). These preliminary results suggest that LR3 had a mitogenic effect on NIH3T3 cells and the three amino terminal EGF-like domains were necessary for this activity. The proliferative activity, however, is not dependent on the cytoplasmic region. The expression of LR3 and its truncated forms were confirmed by western blot using antibodies against their tag epitopes (Fig. 4). LR3N, which did not have a membrane anchor sequence, could be detected in the cell lysate as well as in the conditioned medium (Fig. 4). Its association with
Six mammalian members in the LDLR family have been identified since 1984. They can be grouped into three subfamilies based on their sizes and modular structures. The prototype of the family, LDLR, and its two close relatives, VLDLR and LR7B/apoER2, all have one cluster of either 7 or 8 ligand binding repeats (1– 4). The giant glycoproteins, LRP and gp330, on the other hand, have 4 clusters of more than 30 ligand binding repeats (7, 8). A newly identified protein, LR11/Sorl-1, has a unique combination of motifs of fibronectin III and VPS10, a yeast receptor for vacuolar protein sorting, and one cluster of 11 ligand binding repeats in the middle portion of the extracellular domain (5, 6). A schematic structure comparison of LR3 with these receptors is illustrated in Fig. 5. The structural organization of the entire LR3 extracellular segment closely resembles the middle portion of LRP and gp330 (Fig. 5). However, there are major differences which distinguish LR3 from these two proteins and other LDLR family proteins (Fig. 5). (i) LR3 is only one third of the size of LRP and gp330. (ii) No class B.1 domain, a subclass of EGF-like domains (7), was found in LR3 (LRP and gp330 have 6 and 1 copies of class B.1 domain, respectively). (iii) The ligand binding domains in LR3 are situated at a juxtamembrane position, whereas in other receptors, there are always other structural motifs between the ligand binding domains and the membrane-scanning region. (iv) LRP and gp330 have 31 and 36 copies of the ligand binding domain, respectively. LDLR, the smallest protein in the family, has 7 ligand binding repeats. In contrast, LR3 has only 3 copies of the ligand binding domain. (v) The cytoplasmic tail of LR3 does not have a conserved NPXY motif for coated pit-mediated endocytosis. The
FIG. 4. Western blot analysis of LR3 expression in NIH 3T3 cells. Samples were prepared as described in Materials and Methods. Lane 1, pcDNA3.1; lane 2, LR3F; lane 3, LR3C, lane 4 and 5, LR3N from cell lysate and conditioned medium, respectively. The bands are indicated by arrowheads.
788
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 5. Schematic representation of the protein structure of all known mammalian LDLR family members. Plasma membrane is indicated by the long vertical line. VLDLR and LR7B have very similar architectural structures, so they are illustrated in one diagram. The similarity of LR3 ectodomain and the middle portion of the LRP and gp330 extracellular regions is illustrated. A similar structure comparison which includes the C. elegans LRP-like (9) and Drosophila yorkless (12) proteins can be found in a review by Schneider (13).
only other members in the LDLR superfamily sharing the same feature are insect vitellogenin receptors (11, 12). (vi) Several potential SH3 binding sites are present in the LR3 intracellular region. The only other mammalian LDLR protein that has SH3 motifs is gp330 (8). These distinctive structural features suggest that LR3 may represent a new branch in the LDLR family and have unique functions. The LR3 mRNA is detectable in a wide variety of human tissues (Fig 3A, 3B). The most intriguing observation in LR3 gene’s expression profile is its high mRNA levels in aorta and an osteoblastic cell line, TE85. An association between decreased bone mineral density (osteoporosis) and arterial calcification has been documented in patients (26) and more recently, in mice deficient in either osteoprotegerin or matrix GLA protein (27, 28). Our observation may further substantiates the interconnection of these two seemingly unrelated tissues. It is also interesting to note that LR3 mRNA was present at a much greater level in TE85 cells in comparison with those in MG63 and SaOS-2 cells. These three lines are osteoblast-like osteosarcomas with distinct differentiation phenotypes based on biochemical markers (29). The MG63 cell line is at the earliest stage of differentiation, followed by TE85 and
SaOS-2. The variation of LR3 mRNA level in these cell lines indicated that LR3 gene expression is probably regulated during osteoblast differentiation. The mitogenic activity of LR3 is, to the best of our knowledge, the first time such activity observed in any LDLR family members. Although our preliminary mapping effort showed that the first three EGF-like domains were necessary to confer this activity, the LR3 ectodomain-dependent cell proliferative effects may not be explained by known growth factor activity due to the following considerations: (i) only class B.1 EGF-like domains have been shown to have mitogenic activity (7), but all four copies in LR3 belong to class B.2; (ii) NIH3T3 cells contain undetectable amount of endogenous EGF receptors; (iii) although we did not detect whether there was any EGF-like peptide released from LR3, no processed LR3 forms were seen with tag epitope antibody in LR3F and LR3N transfectants. Further investigation is needed to define the underlying molecular mechanisms. Neither the ligands for other LDLR family receptors, nor the 39 kDa receptor-associated protein, which inhibits the ligand-receptor interaction when binds to the LDLR family receptors (30, 31), has been tested for their binding affinity to LR3. Kim et al. (4) recently
789
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
demonstrated that a variant of human LR7B/apoER2 with only three ligand binding repeats (resulted from alternative splicing) could bind b-VLDL with the same affinity as that of its normal form of 8 ligand binding repeats. This suggests that three repeats are sufficient to confer the binding activity to at least some ligands. However, the unique position of the ligand binding repeats in LR3, and the lack of internalization signal and the Ca21 binding activity (not shown) all lead us to speculate that LR3, instead of being a new receptor mediating endocytosis of lipoproteins, may interact with ligand(s) of different biochemical entity. ACKNOWLEDGMENTS We are grateful to Dr. Joseph Catino for his support for this work, and to Dr. Hong-wei Sun and John Hart for their help in bioinformatics support.
REFERENCES 1. Yamamoto, T., Davis, C. G., Brown, M. S., Schneider, W. J., Casey, M. L., Goldstein, J. L., and Russell, D. W. (1984) Cell 39, 27–38. 2. Sakai, J., Hoshino, A., Takahashi, S., Miura, Y., Ishii, H., Suzuki, H., Kawarabayasi, Y., and Yamamoto, T. (1994) J. Biol. Chem. 269, 2173–2182. 3. Kim, D. H., Iijima, H., Goto, K., Sakai, J., Ishii, H., Kim, H. J., Suzuki, H., Kondo, H., Saeki, S., and Yamamoto, T. (1996) J. Biol. Chem. 271, 8373– 8380. 4. Kim, D. H., Magoori, K., Inoue, T. R., Mao, C. C., Kim, H. J., Suzuki, H., Fujita, T., Endo, Y., Saeki, S., and Yamamoto, T. T. (1997) J. Biol. Chem. 272, 8498 – 8504. 5. Jacobsen, L., Madsen, P., Moestrup, S. K., Lund, A. H., Tommerup, N., Nykjaer, A., Sottrup-Jensen, L., Gliemann, J., and Petersen, C. M. (1996) J. Biol. Chem. 271, 31379 –31383. 6. Morwald, S., Yamazaki, H., Bujo, H., Kusunoki, J., Kanaki, T., Seimiya, K., Morisaki, N., Nimpf, J., Schneider, W. J., and Saito, Y. (1997) Arterioscler. Thromb. Vasc. Biol. 17, 996 –1002. 7. Herz, J., Hamann, U., Rogne, S., Myklebost, O., Gausepohl, H., and Stanley, K. K. (1988) EMBO J. 7, 4119 – 4127. 8. Hjalm, G., Murray, E., Crumley, G., Harazim, W., Lundgreen, S., Onyango, I., Ek, B., Larsson, M., Juhlin, C., Davis, H., Akerstrom, G., Rask, L., and Morse, B. (1996) Eur. J. Biochem. 239, 132–137.
9. Yochem, J., and Greenwald, I. (1993) Proc. Natl. Acad. Sci. USA. 90, 4572– 4576. 10. Nimpf. J., and Schneider. W. J. (1994) Ann. NY Acad. Sci. 737, 145–153. 11. Sappington, T. W., Kokoza, V. A., Cho, W. L., and Raikhel, A. S. (1996) Proc. Natl. Acad. Sci. USA 93, 8934 – 8939. 12. Schonbaum, C. P., Lee, S., and Mahowald, A. P. (1996) Proc. Natl. Acad. Sci. USA 92, 1485–1489. 13. Schneider, W. J. (1996) Int. Rev. Cytol. 166, 103–137. 14. Brown, M. S., and Goldstein, J. L. (1986) Science 232, 34 – 47. 15. Krieger, M., and Herz, J. (1994) Ann. Rev. Biochem. 63, 601– 637. 16. Moestrup, S. K. (1994) Biochim. Biophys. Acta 1197, 197–213. 17. Sambrook, W. J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: a Laboratory Manual, 2nd Ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 18. Campbell, I. D., and Bork, P. (1993) Curr. Opin. Struct. Biol. 3, 385–392. 19. Ullman, C. G., Haris, P. I., Smith, K. F., Sim, R. B., Emery, V. C., and Perkins, S. J. (1995) FEBS Lett. 371, 199 –203. 20. Sappington, T. W., and Raikhel, A. S. (1998) J. Mol. Evol. 46, 476 – 487. 21. Bujo, H., Lindstedt, K. A., Hermann, M., Dalmau, L. M., Nimpf, J., and Schneider, W. J. (1995) J. Biol. Chem. 270, 23546 –23551. 22. Chan, L. (1992) J. Biol. Chem. 267, 25621–25624. 23. Mahley, R. W. (1988) Science 240, 622– 630. 24. Chen, W. J., Goldstein, J. L., and Brown, M. S. (1990) J. Biol. Chem. 265, 3116 –3123 25. Yu, H., Chen, J. K., Feng, S., Dalgarno, D. C., Brauer, A. W., and Schreiber, S. L. (1994) Cell 76, 933–945 26. Parhami, F., and Demer, L. L. (1997) Curr. Opin. Lipidol. 8, 312–314. 27. Luo, G., Ducy, P., McKee, M. D., Pinero, G. J., Loyer, E., Behringer, R. R., and Karsenty, G. (1997) Nature 386, 78 – 81. 28. Bucay, N., Sarosi, I., Dunstan, C. R., Morony, S., Tarpley, J., Capparelli, C., Scully, S., Tan, H. L., Xu, W., Lacey, D. L., Boyle, W. J., and Simonet, W. S. (1998) Genes Dev. 12, 1260- 1268. 29. MacNamara, P., O’Shaughnessy, C., Manduca, P., and Loughery, H. C. (1995) Calif. Tissue Int. 57, 436 – 441. 30. Battey, F. D., Gafvels, M. E., FitzGerald, D. J., Argraves, W. S., Chappell, D. A., Strauss, J. F. 3rd, and Strickland, D. K. (1994) J. Biol. Chem. 269, 23268 –23273. 31. Herz, J., Goldstein, J. L., Strickland, D. K., Ho, Y. K., and Brown, M. S. (1991) J. Biol. Chem. 266, 21232–21238.
790
251, 784 –790 (1998)
RC989545
Molecular Cloning and Characterization of LR3, a Novel LDL Receptor Family Protein with Mitogenic Activity Yu Dong,1,2 William Lathrop,1 Daniel Weaver, Qingqing Qiu,3 John Cini, Donald Bertolini,2 and David Chen4 Bayer Research Center, Pharmaceutical Division, Bayer Corporation, West Haven, Connecticut 06516-4175
Received September 16, 1998
We report molecular cloning and initial functional characterization of a novel member of the low density lipoprotein receptor (LDLR) gene family. The cDNA was isolated from a human osteoblast cDNA library and encoded a 1,615 amino acids protein designated as LR3. It has, in the extracellular region, a cluster of three LDLR ligand binding repeats at a juxtamembrane position and four EGF precursor homology domains separated by YWTD spacer repeats. The entire ectodomain shares the same modular organization with the middle portion of the extracellular regions of two LDLR family members, LDLR-related protein (LRP), and gp330/megalin. LR3 mRNA was expressed in most of the adult and fetal tissues examined. The highest expression level was seen in aorta. In human osteosarcoma cells examined, LR3 mRNA was highly enriched in TE85 cells, moderately expressed in MG63 cells and primary human osteoblasts, and undetectable in SaOS-2 cells. NIH 3T3 cells transfected with either full length LR3 or its ectodomain showed significantly increased proliferation, whereas transfection of intracellular domain had no proliferative effect. We predict that LR3 is a multi-functional protein with potential mitogenic activity. © 1998 Academic Press
The low density lipoprotein receptor (LDLR) family is comprised of a group of related glycoprotein receptors which interact with a diverse array of ligands. Since the isolation of human LDLR cDNA, the first member of the family, in 1984 (1), genes for 5 additional human receptors have been identified. These 1
Equal contribution was made by these two authors. Current address: Pharmaceutical Research Institute, BristolMyers Squibb, P O Box 4000, Princeton, NJ 08543-4000. 3 Current address: Genome Therapeutics Co., 100 Beaver Street, Waltham, MA 02154. 4 To whom correspondence should be addressed at Bayer Research Center, Pharmaceutical Division, Bayer Corporation, 400 Morgan Lane, West Haven, CT 06516-4175. Fax: (203) 937-6923. E-mail: [email protected]. 2
0006-291X/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
include closely related VLDLR (2) and apolipoprotein E receptor 2 (apoER2)/LR7B (3, 4), as well as three larger and much more complex relatives, LR11/Sorl-1 (5, 6), LDLR-related protein (LRP)/a2-macroglobulin receptor (7), and glycoprotein 330 (gp330)/megalin (8). Homologues of these proteins and similar proteins such as vitellogenin receptor have been identified in species ranging from Caenorhabditis elegans and insects to birds and other mammals (9 –13). The overall homology among these receptors is, in general, quite moderate. However, all of them share the following common structural features: (i) one or several clusters of LDLR ligand-binding domains; (ii) cysteine-rich EGF-like domains separated by (F/Y)WTD spacer repeats; (iii) a single membrane-spanning region; (iv) one or several FXNPXY coated-pit internalization signal in the cytoplasmic tail. The functions of the LDLR in endocytosis of plasma lipoprotein and cholesterol homeostasis have been elegantly elucidated largely by Michael Brown and Joseph Goldstein in their classical work of defining the molecular defect of familial hypercholesterolemia (14). Many structurally diverse macromolecules, such as lipoprotein lipase, tissue-type plasminogen activator, urokinase plasminogen activator, Pseudomonas exotoxin A, and lactoferrin, have later been identified as ligands for LRP, gp330 and other receptors, and roles of these receptors in LDL transport and lipid metabolism have been investigated in a variety of species (for reviews, see Ref. 15, 16). Other functions of these receptors, such as signal transduction and Ca21 sensing activities, however, remain less well defined. In an effort to identify novel genes from human osteoblasts, the bone formation cells, we obtained several cDNA clones with unique sequences by a PCR based approach (unpublished data). One of them, HOB-28, showed high homology to LRP, gp330 and other members of the LDLR family, suggesting it might encode a new family member. Here, we report the cloning and sequence analysis of the full-length cDNA of this pro-
784
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
tein, designated as LR3, and initial characterization of its gene expression and cellular functions. MATERIALS AND METHODS Screening of human osteoblast cDNA library and sequencing analysis. By homology RT-PCR experiments and sequencing of randomly picked clones, we obtained clone HOB-28 from human osteoblast RNA (unpublished data). HOB-28 is 310 bp in length and has an open reading frame (ORF). This fragment was labeled with a32PdCTP (3000 Ci/mmol, Amersham Life Science) by random priming and used as a probe in both cDNA library screening and RNA blot hybridization. Approximately two million plaques from a lambda ZAP II human osteoblast cDNA library (a generous gift from Dr. Marian Young, National Institutes of Health) were screened according to the standard protocol (17). Positive clones were obtained by three rounds of screening and plaque purification. The phagemid DNA was excised from the Lambda DNA using the Rapid Excision kit (Stratagene). All clones were subjected to partial sequencing analysis using an ABI 377 DNA sequencer (Perkin-Elmer). Sequence comparison and homology searches in nucleotide and protein databases were performed with the University of Wisconsin Genetics Computer Group programs. Expression constructs. The full length LR3 cDNA and the sequences encoding two truncated forms were subcloned into mammalian expression vectors. LR3F was constructed by inserting a cDNA fragment with the entire ORF into pcDNA3.1/V5-His (Invitrogen). The original stop codon was mutated so that the V5 epitope and polyhistidine tag would be fused to the carboxyl terminus. LR3N, which contained the coding region for the LR3 extracellular region (amino acids 1 to 1388, Fig. 1), was obtained by releasing an ApaI fragment from LR3F and inserting the fragment into the ApaI site in pcDNA3.1/V5-His. The third expression construct, LR3C, encoded the fourth EGF-like domain, the transmembrane segment and the entire carboxyl terminus (amino acids 1079 to 1615, Fig. 1). This construct was made by cutting LR3F with EcoRI and KpnI, and cloning the released fragment into the same sites in pSecTag (Invitrogen). These constructs were confirmed by DNA sequencing. Cell culture, transfection, and measurement of cell proliferation. NIH3T3 cells and three human osteoblast-like osteosarcoma cell lines, MG63, TE85, and SaOS-2, were obtained from the American Type Culture Collection and maintained in DMEM medium plus 2 mM glutamine and 10 % fetal bovine serum (FBS). Normal human fibroblasts were obtained from Clonetics and cultured as suggested by the supplier. The expression constructs and pcDNA3.1/V5-His were introduced into NIH3T3 cells by Lipofectamine reagent following the manufacturer’s instruction (GIBCO-BRL). Stable transfectants were established by selecting the cells with either 500 mg/ml G418 (LR3F, LR3N and pcDNA3), or 200 mg/ml Zeocin (LR3C). To measure the rate if cell proliferation, each transfectant and untransfected NIH3T3 cells were seeded at 3,000 cells/well in two 96 well plates in DMEM supplemented with 10% FBS. The medium was replaced with DMEM plus 0.5% charcoal-Dextran treated FBS the next day and the plates were used in 72 hours. The cell proliferative rate was measured by BrdU incorporation in DNA using an ELISA kit under the conditions recommended by manufacturer (Boehringer Mannheim). RNA isolation and northern/dot blot analyses. Total RNA was isolated from subconfluent MG63, TE85, SaOS-2 cells, human fibroblasts and osteoblasts using TRIzol Reagent (GIBCO-BRL). Ten mg of total RNA from each cell line and from human kidney and placenta (Clontech) were separated on 1.2 % agarose- formaldehyde gels and transferred onto Hybond-N filters. The filters were prehybridized in 6XSSC, 5X Denhardt’s, 50 % formamide, 100 mg/ml salmon sperm DNA at 42°C for 4 hours and then hybridized to labeled HOB-28 at 42°C overnight. The Human RNA Master Blot and a human multiple
FIG. 1. Amino acid sequence of human LR3. The amino acid sequence is deduced from the cDNA sequence (GenBank accession number AF077820). The signal sequence, four class B.2 EGF-like domains, three LDLR ligand binding domains, the transmembrane segment, and five potential SH-3 binding motifs are indicated above the corresponding sequences. The YWTD tetrapeptides in the spacer repeats are underlined. Notably, three of the last five YWTD are less conserved.
tissue northern blot (Clontech) were hybridized with the same probe in ExpressHyb solution (Clontech) at 60°C overnight. After sequential washes, the filters were autoradiographed at -70°C on Kodak X-OMAT films with intensifying screens. The RNA level in each sample was normalized to b-actin mRNA. Density of the dot blot spots was measured and quantified on an imaging densitometer (Bio-Rad). In vitro transcription-translation and western blot analysis. The TNT rabbit reticulocyte lysate system (Promega) was used to synthesize LR3 protein in vitro. pcDNA3 with or without the LR3 ORF (1 mg) was incubated with the lysate in the presence of 35Smethionine (1000 Ci/mmol, Amersham Life Science) at 37°C for 1 hour. The translated products were then analyzed on SDS-PAGE and visualized by autoradiograph. To detect the expressed LR3 and its truncated forms in transfected NIH3T3 cells, subconfluent transfectants were lysed and dounce homogenized at 4°C in Tris buffered saline, pH 8.0 (TBS) with 1% Tween 20/1% NP40 and a protease inhibitor cocktail (Boehringer
785
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 2. Comparison of the sequences of the EGF-like domains, ligand binding domains and SH-3 binding domain in LR3 with their consensus sequences. (A) EGF-like domains. The consensus sequence is derived from class B.2 domains in LDLR and LRP (7; also see Ref. 18). The 6 absolutely conserved cysteines are in bold in the consensus sequence and the dashes represent variable amino acids. Gaps are introduced to optimize the alignment. (B) ligand binding domains. The consensus sequences is from Ullman et al. (19) and Sappington et al. (20). The amino acids that are conserved in more than 90% of the 194 repeats from which the consensus is derived are in bold. The rest of the indicated amino acids are conserved in at least 50% of those repeats. The Ser-Asp-Glu (SDE) motifs are underlined. (C), SH3-binding motifs. The consensus sequence is from Yu et al. (25). Two conserved prolines are in bold. Lower case p indicates that a proline is preferred at the position. F represents any hydrophobic amino acids and X can be any amino acids.
Mannheim). The homogenates were centrifuged at 20,000x g for 20 minutes. The supernatants were passed through Talon metal affinity resin columns (Clontech) and washed with TBS plus 1 mM imidazole. The proteins were then eluted off the columns with the same buffer containing 100 mM imidazole. Both the cell lysates and the eluted fractions were analyzed on 10% Tricine polyacrylamide precast gels (Novex). The proteins were transferred onto nitrocellulose membranes and blotted with either Anti-V5 or Anti-myc mouse antibodies (Invitrogen) at 4°C. To detect whether LR3N was present in the culture medium, 1 ml of conditioned medium was mixed with 25 ml of Talon metal affinity resin and incubated at 4°C for 2 hours. The resin was washed three times with TSB plus 1% Tween 20, boiled at 98°C in a sample buffer and pelleted by low speed centrifugation. The supernatant was analyzed as described above.
RESULTS Isolation of LR3 from a human osteoblast cDNA library. Using radiolabeled HOB-28 as a probe, we obtained 15 positive clones from approximately two million phages of a human osteoblast cDNA library. One clone with an insert of .5 kb was subjected to complete sequence analysis by primer walking in both directions. A dbEST search found that more than 25 human ESTs from libraries of several different tissues and cell types had at least 98% identity to this sequence. The cDNA clone has an ORF of 4,845 bp, a 59 UTR of 74 bp, and a 39 UTR of 187 bp including a polyadenylation signal, AATAAA (GenBank Accession No. AF077820). The sequence of 1,615 amino acids deduced from the ORF is shown in Fig. 1. The protein has a calculated molecular weight of 178 kDa and an isoelectric point of 5.1. The ORF is confirmed by analysis of the protein
synthesized in rabbit reticulocyte lysate. A prominent band migrating at approximately 180 kD was revealed by SDS-PAGE (data not shown). The primary structure predicts a type I transmembrane protein consisting of a signal peptide of 24 amino acids, a 1,359 amino acid extracellular region, a single membrane-spanning region of 30 amino acids, and a 202 amino acid intracellular domain (Fig. 1). Database searching indicated that this sequence is most homologous to the LDLR family members and contains the same type of structural modules. These peptide modules, approximately 40 amino acid residues in length, consist of 6 conserved cysteines with distinctive spacing between the cysteines in each type (18 –20). There are three type A LDLR ligand binding domains in one cluster and four EGF-like domains separated by YWTD-spacer repeats N-terminal to the ligand binding domains (Fig. 1, Fig. 2A, 2B). In concordance with the proposed nomenclature system (20, 21), this protein was designated LR3 to reflect the total number of ligand binding domains. Alignment of the EGF-like domains in LR3 with a consensus sequence derived from those in LDLR and LRP (7) indicated that all four domains belong to class B.2 (Fig. 2A). Comparing the ligand binding domains with a consensus sequence derived from 194 type A repeats (19) showed that those in LR3 have all 11 amino acids that are conserved in over 90% of the repeats, and 16 of the 19 amino acids that are conserved in at least 50% of the repeats (Fig. 2B). The Ser-Asp-Glu motif between the fifth and sixth
786
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
cysteines are also perfectly conserved in all three LR3 ligand binding repeats (Fig. 2B). This negatively charged tripeptide is implicated for high-affinity binding of positively charged sequences in ligands such as LDL and b-VLDL (22, 23). The YWTD-spacers (about 50 amino acids apart) are organized in tandem in a group of five, identical to the pattern seen in other LDLR proteins and the EGF precursor (Fig. 1, Fig. 5). The cytoplasmic tail of LR3 does not have a conserved NPXY motif, a common internalization signal sequence present in other members of the LDLR gene family for coated pit-mediated endocytosis (24). On the other hand, up to five potential Src-homology 3 (SH3) binding sites, including two that match perfectly with the consensus sequence XpfPpXP, are present in this region (Fig. 2C; 25). LR3 mRNA expression and tissue distribution. We first analyzed the expression of LR3 mRNA in a variety of adult and fetal tissues by dot blot analysis. Fig. 3A shows the result obtained by hybridizing a multiple tissue poly A1 RNA dot blot with HOB-28 probe. The LR3 mRNA could be detected in most of the tissues on the blot, except testis, bone marrow, peripheral leukocyte and various regions of the brain. The highest expression levels were found in aorta. In addition, the tissues of the female reproductive system, including uterus, mammary gland, ovary and placenta all showed relatively abundant LR3 mRNA (Fig. 3A). Differential expression levels of LR3 mRNA in fetal and adult tissues were noticeable in spleen, lung and kidney, with higher expression levels seen in fetal tissues (Fig. 3A). The different expression levels between adult and fetal heart could result from the inclusion of aorta tissue in the fetal heart. We then examined the LR3 mRNA level in selected tissues and bone cell lines using northern blot analysis. Fig. 3B shows that the HOB-28 detected a single transcript of about 5.4 kb in all tissues except brain. Heart, liver, and lung showed higher expression levels comparing with other tissues. It is possible that the high mRNA level in heart was caused by the aorta tissue attached. A striking difference in the expression levels was observed in three human osteosarcoma cell lines
FIG. 3. Expression of LR3 in tissues and osteosarcoma cell lines. (A) dot blot analysis of the expression of LR3 in different tissues.
The intensity of each dot was measured by an imaging densitometer and normalized by the amount of RNA in each dot (100 to 500 ng/dot). The expression levels in fetal and adult tissues are represented by open and filled bars, respectively. The expression levels are relative to that in aorta. (B) tissue distribution of LR3 analyzed by northern blot. Two micrograms of poly A1 RNA from the indicted tissues were probed with HOB-28. Hybridizing the same blot with a b-actin probe indicated that significantly greater amount of RNA from skeletal muscle and pancreas was blotted on the filter (not shown). (C) expression of LR3 in human osteoblastic cells. Ten micrograms of total RNA prepared from each indicated tissue or cell line were probed with HOB-28. Hybridizing the same filter with a b-actin probe estimated a roughly equal loading (not shown).
787
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
the cells could be nonspecific or via binding to a receptor on the plasma membrane.
TABLE 1
DNA Synthesis Rates in NIT3T3 Cells Transfected with LR3 Expression Constructs
DISCUSSION
BrdU incorporation (%) Cell type
Experiment I
Experiment II
pcDNA3.1 LR3F LR3N LR3C NIH3T3 (.5%) NIH3T3 (1%) NIH3T3 (10%)
100 161.8 6 18.8 192.3 6 6.8 96.1 6 14.7 127.5 6 7.3 155.8 6 16.2 231.4 6 15.2
100 130.6 6 8.7 136.6 6 10.2 80.1 6 6.8 103.1 6 5.5 107.1 6 3.1 148.8 6 9.1
Note. Untransfected NIH3T3 cells cultured in DMEM supplemented with 1% and 10% FBS were included as positive controls. NIH3T3 cells in DMEM with 0.5% charcoal-Dextran treated FBS were served as serum starved controls. Results are from two independent experiments.
(Fig. 3C). LR3 was highly expressed in TE85 cells, while undetectable in SaOS-2 cells. Since the same amount of RNA from these cells and human tissues was blotted on the same filter, it was estimated that the LR3 mRNA level in TE85 cells was about 10-fold higher than that in liver, and the expression levels in MG63 cells and kidney were comparable. The expression of LR3 in STRO-1 positive human primary osteoblasts was at a level similar to that of MG63 (data not shown), but the expression was undetectable in normal human fibroblast cells (Fig. 3C). Cell proliferative effect of LR3. When an expression construct LR3F which contains the full length LR3 ORF was transfected into murine NIH3T3 cells, an increased cell proliferation was noticed. Measurement of BrdU incorporation rate also indicated a significant increase in DNA synthesis. Constructs that encoded either the amino or carboxyl terminus truncated LR3 were designed to map the structure(s) that mediated the mitogenic activity. Two independent experiments showed that NIH3T3 cells transfected with LR3N, which encoded the entire LR3 extracellular region, had a BrdU incorporation rate similar to that with 10% FBS stimulation (Table 1). In contrast, cells transfected with LR3C, which encoded an N-terminal deleted LR3 (missing the first three EGF-like domains), showed a slightly lower DNA synthesis rate (Table 1). These preliminary results suggest that LR3 had a mitogenic effect on NIH3T3 cells and the three amino terminal EGF-like domains were necessary for this activity. The proliferative activity, however, is not dependent on the cytoplasmic region. The expression of LR3 and its truncated forms were confirmed by western blot using antibodies against their tag epitopes (Fig. 4). LR3N, which did not have a membrane anchor sequence, could be detected in the cell lysate as well as in the conditioned medium (Fig. 4). Its association with
Six mammalian members in the LDLR family have been identified since 1984. They can be grouped into three subfamilies based on their sizes and modular structures. The prototype of the family, LDLR, and its two close relatives, VLDLR and LR7B/apoER2, all have one cluster of either 7 or 8 ligand binding repeats (1– 4). The giant glycoproteins, LRP and gp330, on the other hand, have 4 clusters of more than 30 ligand binding repeats (7, 8). A newly identified protein, LR11/Sorl-1, has a unique combination of motifs of fibronectin III and VPS10, a yeast receptor for vacuolar protein sorting, and one cluster of 11 ligand binding repeats in the middle portion of the extracellular domain (5, 6). A schematic structure comparison of LR3 with these receptors is illustrated in Fig. 5. The structural organization of the entire LR3 extracellular segment closely resembles the middle portion of LRP and gp330 (Fig. 5). However, there are major differences which distinguish LR3 from these two proteins and other LDLR family proteins (Fig. 5). (i) LR3 is only one third of the size of LRP and gp330. (ii) No class B.1 domain, a subclass of EGF-like domains (7), was found in LR3 (LRP and gp330 have 6 and 1 copies of class B.1 domain, respectively). (iii) The ligand binding domains in LR3 are situated at a juxtamembrane position, whereas in other receptors, there are always other structural motifs between the ligand binding domains and the membrane-scanning region. (iv) LRP and gp330 have 31 and 36 copies of the ligand binding domain, respectively. LDLR, the smallest protein in the family, has 7 ligand binding repeats. In contrast, LR3 has only 3 copies of the ligand binding domain. (v) The cytoplasmic tail of LR3 does not have a conserved NPXY motif for coated pit-mediated endocytosis. The
FIG. 4. Western blot analysis of LR3 expression in NIH 3T3 cells. Samples were prepared as described in Materials and Methods. Lane 1, pcDNA3.1; lane 2, LR3F; lane 3, LR3C, lane 4 and 5, LR3N from cell lysate and conditioned medium, respectively. The bands are indicated by arrowheads.
788
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 5. Schematic representation of the protein structure of all known mammalian LDLR family members. Plasma membrane is indicated by the long vertical line. VLDLR and LR7B have very similar architectural structures, so they are illustrated in one diagram. The similarity of LR3 ectodomain and the middle portion of the LRP and gp330 extracellular regions is illustrated. A similar structure comparison which includes the C. elegans LRP-like (9) and Drosophila yorkless (12) proteins can be found in a review by Schneider (13).
only other members in the LDLR superfamily sharing the same feature are insect vitellogenin receptors (11, 12). (vi) Several potential SH3 binding sites are present in the LR3 intracellular region. The only other mammalian LDLR protein that has SH3 motifs is gp330 (8). These distinctive structural features suggest that LR3 may represent a new branch in the LDLR family and have unique functions. The LR3 mRNA is detectable in a wide variety of human tissues (Fig 3A, 3B). The most intriguing observation in LR3 gene’s expression profile is its high mRNA levels in aorta and an osteoblastic cell line, TE85. An association between decreased bone mineral density (osteoporosis) and arterial calcification has been documented in patients (26) and more recently, in mice deficient in either osteoprotegerin or matrix GLA protein (27, 28). Our observation may further substantiates the interconnection of these two seemingly unrelated tissues. It is also interesting to note that LR3 mRNA was present at a much greater level in TE85 cells in comparison with those in MG63 and SaOS-2 cells. These three lines are osteoblast-like osteosarcomas with distinct differentiation phenotypes based on biochemical markers (29). The MG63 cell line is at the earliest stage of differentiation, followed by TE85 and
SaOS-2. The variation of LR3 mRNA level in these cell lines indicated that LR3 gene expression is probably regulated during osteoblast differentiation. The mitogenic activity of LR3 is, to the best of our knowledge, the first time such activity observed in any LDLR family members. Although our preliminary mapping effort showed that the first three EGF-like domains were necessary to confer this activity, the LR3 ectodomain-dependent cell proliferative effects may not be explained by known growth factor activity due to the following considerations: (i) only class B.1 EGF-like domains have been shown to have mitogenic activity (7), but all four copies in LR3 belong to class B.2; (ii) NIH3T3 cells contain undetectable amount of endogenous EGF receptors; (iii) although we did not detect whether there was any EGF-like peptide released from LR3, no processed LR3 forms were seen with tag epitope antibody in LR3F and LR3N transfectants. Further investigation is needed to define the underlying molecular mechanisms. Neither the ligands for other LDLR family receptors, nor the 39 kDa receptor-associated protein, which inhibits the ligand-receptor interaction when binds to the LDLR family receptors (30, 31), has been tested for their binding affinity to LR3. Kim et al. (4) recently
789
Vol. 251, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
demonstrated that a variant of human LR7B/apoER2 with only three ligand binding repeats (resulted from alternative splicing) could bind b-VLDL with the same affinity as that of its normal form of 8 ligand binding repeats. This suggests that three repeats are sufficient to confer the binding activity to at least some ligands. However, the unique position of the ligand binding repeats in LR3, and the lack of internalization signal and the Ca21 binding activity (not shown) all lead us to speculate that LR3, instead of being a new receptor mediating endocytosis of lipoproteins, may interact with ligand(s) of different biochemical entity. ACKNOWLEDGMENTS We are grateful to Dr. Joseph Catino for his support for this work, and to Dr. Hong-wei Sun and John Hart for their help in bioinformatics support.
REFERENCES 1. Yamamoto, T., Davis, C. G., Brown, M. S., Schneider, W. J., Casey, M. L., Goldstein, J. L., and Russell, D. W. (1984) Cell 39, 27–38. 2. Sakai, J., Hoshino, A., Takahashi, S., Miura, Y., Ishii, H., Suzuki, H., Kawarabayasi, Y., and Yamamoto, T. (1994) J. Biol. Chem. 269, 2173–2182. 3. Kim, D. H., Iijima, H., Goto, K., Sakai, J., Ishii, H., Kim, H. J., Suzuki, H., Kondo, H., Saeki, S., and Yamamoto, T. (1996) J. Biol. Chem. 271, 8373– 8380. 4. Kim, D. H., Magoori, K., Inoue, T. R., Mao, C. C., Kim, H. J., Suzuki, H., Fujita, T., Endo, Y., Saeki, S., and Yamamoto, T. T. (1997) J. Biol. Chem. 272, 8498 – 8504. 5. Jacobsen, L., Madsen, P., Moestrup, S. K., Lund, A. H., Tommerup, N., Nykjaer, A., Sottrup-Jensen, L., Gliemann, J., and Petersen, C. M. (1996) J. Biol. Chem. 271, 31379 –31383. 6. Morwald, S., Yamazaki, H., Bujo, H., Kusunoki, J., Kanaki, T., Seimiya, K., Morisaki, N., Nimpf, J., Schneider, W. J., and Saito, Y. (1997) Arterioscler. Thromb. Vasc. Biol. 17, 996 –1002. 7. Herz, J., Hamann, U., Rogne, S., Myklebost, O., Gausepohl, H., and Stanley, K. K. (1988) EMBO J. 7, 4119 – 4127. 8. Hjalm, G., Murray, E., Crumley, G., Harazim, W., Lundgreen, S., Onyango, I., Ek, B., Larsson, M., Juhlin, C., Davis, H., Akerstrom, G., Rask, L., and Morse, B. (1996) Eur. J. Biochem. 239, 132–137.
9. Yochem, J., and Greenwald, I. (1993) Proc. Natl. Acad. Sci. USA. 90, 4572– 4576. 10. Nimpf. J., and Schneider. W. J. (1994) Ann. NY Acad. Sci. 737, 145–153. 11. Sappington, T. W., Kokoza, V. A., Cho, W. L., and Raikhel, A. S. (1996) Proc. Natl. Acad. Sci. USA 93, 8934 – 8939. 12. Schonbaum, C. P., Lee, S., and Mahowald, A. P. (1996) Proc. Natl. Acad. Sci. USA 92, 1485–1489. 13. Schneider, W. J. (1996) Int. Rev. Cytol. 166, 103–137. 14. Brown, M. S., and Goldstein, J. L. (1986) Science 232, 34 – 47. 15. Krieger, M., and Herz, J. (1994) Ann. Rev. Biochem. 63, 601– 637. 16. Moestrup, S. K. (1994) Biochim. Biophys. Acta 1197, 197–213. 17. Sambrook, W. J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: a Laboratory Manual, 2nd Ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 18. Campbell, I. D., and Bork, P. (1993) Curr. Opin. Struct. Biol. 3, 385–392. 19. Ullman, C. G., Haris, P. I., Smith, K. F., Sim, R. B., Emery, V. C., and Perkins, S. J. (1995) FEBS Lett. 371, 199 –203. 20. Sappington, T. W., and Raikhel, A. S. (1998) J. Mol. Evol. 46, 476 – 487. 21. Bujo, H., Lindstedt, K. A., Hermann, M., Dalmau, L. M., Nimpf, J., and Schneider, W. J. (1995) J. Biol. Chem. 270, 23546 –23551. 22. Chan, L. (1992) J. Biol. Chem. 267, 25621–25624. 23. Mahley, R. W. (1988) Science 240, 622– 630. 24. Chen, W. J., Goldstein, J. L., and Brown, M. S. (1990) J. Biol. Chem. 265, 3116 –3123 25. Yu, H., Chen, J. K., Feng, S., Dalgarno, D. C., Brauer, A. W., and Schreiber, S. L. (1994) Cell 76, 933–945 26. Parhami, F., and Demer, L. L. (1997) Curr. Opin. Lipidol. 8, 312–314. 27. Luo, G., Ducy, P., McKee, M. D., Pinero, G. J., Loyer, E., Behringer, R. R., and Karsenty, G. (1997) Nature 386, 78 – 81. 28. Bucay, N., Sarosi, I., Dunstan, C. R., Morony, S., Tarpley, J., Capparelli, C., Scully, S., Tan, H. L., Xu, W., Lacey, D. L., Boyle, W. J., and Simonet, W. S. (1998) Genes Dev. 12, 1260- 1268. 29. MacNamara, P., O’Shaughnessy, C., Manduca, P., and Loughery, H. C. (1995) Calif. Tissue Int. 57, 436 – 441. 30. Battey, F. D., Gafvels, M. E., FitzGerald, D. J., Argraves, W. S., Chappell, D. A., Strauss, J. F. 3rd, and Strickland, D. K. (1994) J. Biol. Chem. 269, 23268 –23273. 31. Herz, J., Goldstein, J. L., Strickland, D. K., Ho, Y. K., and Brown, M. S. (1991) J. Biol. Chem. 266, 21232–21238.
790