Role of the Latent Transforming Growth Factor β–Binding Protein 1 in Fibrillin-Containing Microfibrils in Bone Cells In Vitro and In Vivo
Sarah L. Dallas,
Department of Medicine, Division of Endocrinology and Metabolism, University of Texas Health Science Center at San Antonio, San Antonio, Texas, U.S.A.
Address reprint requests to: Sarah L. Dallas, Ph.D. Division of Endocrinology Department of Medicine University of Texas Health Science Center at San Antonio 7703 Floyd Curl Drive San Antonio, TX 78284, U.S.A.
Latent transforming growth factor β–binding proteins (LTBPs) are extracellular matrix (ECM) proteins that bind latent transforming growth factor β (TGF-β) and influence its availability in bone and other connective tissues. LTBPs have homology with fibrillins and may have related functions as microfibrillar proteins. However, at present little is known about their structural arrangement in the ECM. By using antibodies against purified LTBP1, against a short peptide in LTBP1, and against epitope-tagged LTBP1 constructs, we have shown colocalization of LTBP1 and fibrillin 1 in microfibrillar structures in the ECM of cultured primary osteoblasts. Immunoelectron microscopy confirmed localization of LTBP1 to 10- to 12-nm microfibrils and suggested an ordered aggregation of LTBP1 into these structures. Early colocalization of LTBP1 with fibronectin suggested a role for fibronectin in the initial assembly of LTBP1 into the matrix; however, in more differentiated osteoblast cultures, LTBP1 and fibronectin 1 were found in distinct fibrillar networks. Overexpression of LTBP1 deletion constructs in osteoblast-like cells showed that N-terminal amino acids 67–467 were sufficient for incorporation into fibrillin-containing microfibrils and suggested that LTBP1 can be produced by cells distant from the site of fibril formation. In embryonic long bones in vivo, LTBP1 and fibrillin 1 colocalized at the surface of newly forming osteoid and bone. However, LTBP1-positive fibrils, which did not contain fibrillin 1, were present in cartilage matrix. These studies show that in addition to regulating TGFβ1, LTBP1 may function as a structural component of connective tissue microfibrils. LTBP1 may therefore be a candidate gene for Marfan-related connective tissue disorders in which linkage to fibrillins has been excluded.
Recent evidence suggests that interactions between growth factors and structural components of the extracellular matrix may play fundamental roles in tissue morphogenesis and repair and that the binding of growth factors to the extracellular matrix (ECM) may be a major mechanism for regulation of growth factor activity (reviewed in Ref. (1). It has long been thought that growth factors released from bone matrix play important roles in the coupling of bone resorption to bone formation and in repair processes such as fracture healing. However, at present little is known about the mechanisms by which growth factors are stored in the ECM, which ECM proteins they interact with, or the mechanisms by which they are released.
The latent transforming growth factor β–binding protein 1 (LTBP1) is an example of a bone matrix protein that is important in growth factor regulation. This protein is a member of an emerging superfamily of ECM proteins, including the fibrillins 1 and 2,(2,3) LTBPs 1, 2, and 3,(4–9) and the recently identified LTBP4.(10,11) At least three of the LTBPs (LTBPs 1, 3, and 4) have been shown to form complexes with latent transforming growth factor β (TGF-β).(4,5,9,11,12) Recent data have suggested that LTBPs may play an intricate role in regulating the activity of this multipotent growth factor, which is produced in a latent form and, when activated, exerts profound effects in bone and other cell types (reviewed in Refs. (13–17).
LTBP1 is the prototype member of the LTBP family. This 190-kDa protein was originally identified as a component of the high-molecular-weight latent TGFβ1 complex, in which LTBP1 is disulfide-linked to a single chain of the TGFβ1 precursor homodimer (also known as the latency-associated peptide or LAP).(4,5) LTBP1 is not an absolute requirement for latency of the TGFβ1 complex, because a low-molecular-weight latent TGFβ complex that lacks LTBP1 has been described as a recombinant protein(18) and as a naturally occurring complex in bone cells.(19,20) Current data suggest that LTBP1 is a multifunctional protein. It may facilitate secretion of latent TGFβ(21) and may also modulate activation of latent TGFβ.(22) More recently it has been described as a stable component of the extracellular matrix that is important in storage and release of latent TGFβ.(23–26)
At present very little is known about how LTBPs function as matrix proteins. Like the fibrillins, their primary structure consists predominantly of cysteine-rich repeats of two types: (1) 6-cysteine (EGF-like) repeats, similar to motifs found in epidermal growth factor (EGF) precursor and a number of other proteins,(27–29) and (2) 8-cysteine repeats unique to the LTBPs and fibrillins. The patterning of these repeats is highly conserved between the fibrillins and LTBPs, and the major differences between these proteins are in the numbers of these two types of repeats. This domain structure suggests a structural role for LTBP1 similar to the fibrillins.
Fibrillins are the major components of the 10- to 12-nm microfibrils found in many types of connective tissue, including bone, and which, in association with elastin, can form the larger structures known as elastic fibers (reviewed in Refs. (30–32). LTBP1 has been shown to localize to fibrillar structures in the extracellular matrix of bone cells, fibroblasts, chondrocytes, and cardiac cushion tissue.(24–26,33) In fibroblast cultures, these fibrils resemble connective tissue microfibrils when viewed by immunogold labeling and immunoelectron microscopy(24). LTBPs 2 and 3 were cloned on the basis of homology to fibrillins,(7,9) and LTBP2 has been definitively shown to colocalize with fibrillin 1 in 10- to 12-nm microfibrils(7). Taken together, these data suggest that LTBPs, like the fibrillins, may be integral components of connective tissue microfibrils.
Mutations in fibrillin 1 are associated with the inherited connective tissue disorders Marfan syndrome and familial ectopia lentis, and mutations in fibrillin 2 have been linked to congenital contractural arachnodactyly (reviewed in Refs. (34) and (35). These disorders affect tissues rich in microfibrils, including the skeletal, cardiovascular, and/or ocular systems. Because LTBPs have homology to fibrillins and appear to have related functions, a further understanding of the relationship between LTBPs and the fibrillins will be the key to determining their role in normal physiology and their potential role in connective tissue diseases. In the present study, we sought to further characterize the nature of the LTBP1-positive fibrillar structures present in the extracellular matrix of bone cells in vitro and in vivo, to determine whether a relationship exists between LTBP1 and fibrillin 1, and to identify the bone matrix binding region in LTBP1.
MATERIALS AND METHODS
Unless stated otherwise, all tissue culture reagents were purchased from Life Technologies Inc. (Gaithersburg, MD, U.S.A.) or JRH Biosciences (Lenexa, KS, U.S.A.). MG63 osteosarcoma cells were purchased from the American Type Tissue Culture Collection (ATCC; Manassas, VA, U.S.A.) and were maintained in Eagle's minimum essential medium supplemented with 10% fetal bovine serum (FBS), 2 mM l-glutamine, and 100 U/ml penicillin/streptomycin (P/S). Human 293 embryonic kidney epithelial cells were purchased from ATCC and maintained in Dulbecco's modified Eagle medium supplemented with 10% FBS, 2 mM l-glutamine, and 100 U/ml P/S. The 2T3 osteoblast-like cell line was derived from the calvarial cells of a transgenic mouse in which the T antigen was expressed under control of the BMP-2 promoter, thus directing expression of the immortalizing T antigen to the osteoblast lineage.(36) This cell line has been extensively characterized and shown to express osteoblast-specific markers and to form mineralized bonelike nodules in vitro. 2T3 cells were maintained in minimal essential medium, alpha modification (α-MEM) supplemented with 10% FBS, 2 mM l-glutamine, and 100 U/ml P/S. Primary cultures of fetal rat calvarial osteoblasts were isolated as described previously.(25) Briefly, frontal and parietal bones from ICR-Swiss rat fetuses (19 days' gestation) were digested in 0.2% collagenase/0.05% trypsin in Hank's balanced salt solution at 37°C to obtain six sequential 20-minute digests. The third through sixth digests were combined and grown to confluence in α-MEM supplemented with 10% FBS, 2 mM l-glutamine, and 100 U/ml P/S. At confluence, the cells were trypsinized and used for experiments. For immunostaining experiments, MG63 cells or fetal rat calvarial osteoblasts were plated in Lab-Tek chamber slides (Nalge Nunc International, Naperville, IL, U.S.A.) at 10,000 cells/cm2 growth area in media as described above. At confluence, the medium was changed to α-MEM supplemented as above, but with the addition of 100 μg/ml ascorbic acid and 5 mM β-glycerophosphate and the reduction of the serum concentration to 5%. Thereafter, the culture medium was changed every 3 days, and the cultures were maintained for the indicated periods.
Antibody against LTBP1 was a rabbit polyclonal serum (Ab39) raised against purified human platelet LTBP1, which cross-reacts with mouse and rat LTBP1 (kindly supplied by K. Miyazono, Japanese Foundation for Cancer Research, Tokyo). The specificity of this antibody has been described elsewhere.(4,24,25) Ab39 was used at a dilution of 1:2000 for immunofluorescence, 1:500 for Western blotting, and 1:5 for immunoelectron microscopy. A second antibody against LTBP1 was raised in chicken and in rabbit using the peptide CPGGMGYTVSGIHRRRPIHQHIGK as an immunogen (residues 721–744 of the rat LTBP1 sequence published by Tsuji et al. 1990; Genbank accession number M55431). This sequence, which we have termed the “LTBP1 hinge peptide,” occurs in the so-called hinge region of LTBP1, which is rich in proline and basic amino acids and is not conserved between LTBP family members. This peptide sequence is unique to LTBP1 according to the PIR and SwissProt databases. Peptide synthesis and coupling to carrier were performed by the Institutional Protein Core Facility at the University of Texas Health Science Center at San Antonio. Antibodies against this peptide were produced commercially by Alpha Diagnostics International (San Antonio, TX, U.S.A.). The antibodies were affinity-purified from rabbit serum or from purified chicken immunoglobulin (IgY) with a peptide immunoaffinity column prepared by using the SulfoLink kit according to the manufacturer's instructions (Pierce, Rockford, IL, U.S.A.). The characterization of these affinity-purified antibodies is shown in Fig. 1. These antibodies were used at a dilution of 1:200 for immunofluorescence, 1:100 for Western blotting, and 1:5 for immunoelectron microscopy.
Antibodies against fibrillin 1 were a mouse monoclonal IgG against human fibrillin 1 (MAb201; 1:100 dilution for immunofluorescence) or a rabbit polyclonal antiserum against human fibrillin 1 (PAb9543), which cross-reacts with rat and mouse fibrillin 1 (1:1000 dilution for immunofluorescence, 1:5 dilution for immunoelectron microscopy). The specificities of these antibodies have been described elsewhere.(2,37) Antibody against fibronectin was a commercially available mouse monoclonal IgM against cellular fibronectin (1:1000 dilution for immunofluorescence; Sigma Chemical Co., St. Louis, MO, U.S.A.). Antibody against type I collagen was a commercially available rabbit polyclonal antiserum against rat type-I collagen (1:400 dilution for immunofluorescence; Biodesign, Kennebunk, ME, U.S.A.). Antibodies against the hemagglutinin epitope tag (HA) were a commercially available rabbit polyclonal antiserum and a mouse monoclonal IgG (Berkley Antibody Corporation [BabCo], Richmond, CA, U.S.A.). These antibodies were used at a dilution of 1:1000 for immunofluorescence and 1:500 for Western blotting.
Unless stated otherwise, all secondary antibodies for immunofluorescent staining were purchased from Jackson Immunoresearch Laboratories Incorporated (West Grove, PA, U.S.A.) and were preadsorbed to give minimal cross-reactivity against other species. Various combinations of fluorochrome-conjugated secondary antibodies were used as appropriate for each combination of primary antibodies; these combinations are given in the figure legends. These antibodies included a lissamine-rhodamine (LRSC)–conjugated donkey anti-rabbit IgG (red), a fluorescein isothiocyanate (FITC)–conjugated donkey anti-rabbit IgG (green), an FITC-conjugated donkey anti-mouse IgG (green), and a cy3-conjugated donkey anti-chicken (red). For detection of the mouse IgM used for fibronectin staining, a biotinylated goat anti-mouse IgM was used (Vector laboratories, Burlingame, CA, U.S.A.), followed by FITC-conjugated streptavidin (green; Pierce). In addition, a cy3-conjugated donkey anti-rabbit antibody (red) was used for single-labeling studies. Secondary antibodies were diluted according to the manufacturer's recommendations.
Detection reagents for Western blotting were a peroxidase-conjugated donkey anti-chicken (1:20,000 dilution; Jackson Immunoresearch Laboratories) and peroxidase-conjugated protein A (1:2500 dilution; Kirkegaard & Perry Labs Inc., Gaithersburg, MD, U.S.A.).
Proteins in concentrated conditioned media samples were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) using 4–20% gradient polyacrylamide mini gels and then transferred onto a nitrocellulose membrane by electroblotting as described previously.(19,20) Membranes were blocked in 5% nonfat milk in Tris-buffered saline (TBS) buffer (50 mM Tris-HCl, 150 mM NaCl, pH 7.4) for 2 h at room temperature and then incubated with appropriate primary antibodies diluted in TBS + 1% bovine serum albumin for 2 h at room temperature. The membranes were washed five times in TBS + 0.05% Tween-20 for 5 minutes and two times in TBS for 5 minutes. The membranes were then incubated in the appropriate detection antibodies diluted in TBS +5% nonfat milk for 1 h at room temperature. After washing as described above, the immunostained bands were visualized with an enhanced chemiluminescence (ECL) detection system according to the manufacturer's instructions (Amersham Corporation, Arlington Heights, IL, U.S.A.).
Colocalization of LTBP1, fibrillin 1, fibronectin, and type I collagen was performed by using double-staining indirect immunofluorescence techniques. For experiments using cultured cells, the cells were grown in Lab-Tek chamber slides or on collagen-coated glass coverslips, for the specified time periods. The culture medium was then removed, the cells were washed twice in phosphate-buffered saline (PBS), and fixation was performed in 95% ethanol for 10 minutes.
The antigenic determinants recognized by the anti-LTBP1 antibodies were extremely sensitive to most fixatives, which limited the possibilities for tissue processing. Therefore, for localization of LTBP1 in vivo, a protocol was developed using human tissues obtained from therapeutic abortions in accordance with policies established by the University of Washington's institutional review board. Fresh specimens were dissected in PBS, placed in Tissue-Tek OCT embedding medium (Miles Laboratories, Inc., Naperville IL, U.S.A.), frozen in liquid nitrogen, and stored at −20°C. Frozen sections (6 μm thick) were cut from these tissues, placed on gelatin-coated glass slides, air-dried, and stored at −20°C in tight boxes until used. The sections were then brought to room temperature, air-dried and fixed in chilled acetone for 10 minutes.
After fixation, the procedure for immunostaining was essentially identical for cell culture preparations and tissue sections. The specimens were washed twice in PBS and background blocking was performed using 1% goat serum (GS) or 1% horse serum (HS) in PBS for 3 h at room temperature. Samples were then incubated with primary antibodies and diluted in PBS + 1% GS (or HS) for 2 h at room temperature. The samples were washed six times with PBS and incubated for 1 h at room temperature with the appropriate secondary antibodies (see above) and diluted in PBS + 1 % GS (or HS). The specimens were washed six times in PBS and then mounted under glass coverslips in a solution of 9:1 glycerol/PBS containing 5% N-propyl gallate to reduce photobleaching. The specimens were viewed and photographed with an Olympus BX40 microscope with fluorescence attachments (Olympus America, Inc., Melville, NY). For double-labeling experiments, controls were performed in which each primary antibody was tested individually using both secondary antibodies to confirm that the pattern of staining observed was not due to cross-species reactivity of the second antibodies.
For immunoelectron microscopy studies, fetal rat calvarial osteoblasts were cultured on Thermanox coverslips (Nalge Nunc International no. 174950) until confluence in α-MEM supplemented with 10% FBS. At confluence, the medium was changed to α-MEM supplemented with 5% FBS, 100 μg/ml ascorbic acid, and 5 mM β-glycerophosphate. Medium was changed every 3 days, and the cultures were harvested between days 7 and 10, when calcified nodules became visible by eye. After a brief rinse in PBS, the coverslips were floated for 15 h at 4°C, cell side down, in 100 μl of affinity-purified rabbit anti-LTBP1 hinge peptide antibody, rabbit anti-LTBP1 (Ab39) or rabbit anti–fibrillin 1 (PAb 9543), each diluted 1:5 in PBS. After several changes of PBS over 45 minutes, the cultures were floated for 2 h at room temperature, cell side down, on 100 μl of goat anti-rabbit secondary antibody-5 nm gold conjugate (Amersham Corporation, Arlington Heights, IL) diluted 1:3 in 0.1% BSA/150 mM Tris buffer (pH 8.0). The cultures were then rinsed in PBS over 30 minutes, rinsed in 0.1 M sodium cacodylate buffer (pH 7.4), and then fixed for 60 minutes in 1.5% paraformaldehyde/1.5% glutaraldehyde containing 0.05% tannic acid. The samples were then fixed in 1% buffered OsO4, dehydrated in a graded series of ethanol to 100%, rinsed in propylene oxide, and embedded in Spurr's epoxy. Ultrathin sections were cut parallel to the growth substrate and examined on a Philips 410 LS transmission electron microscope (Philips, Santa Clara, CA, U.S.A.).
LTBP1 expression constructs
Constructs for human LTBP1 were produced using the mammalian expression vector pcDNA3, as described previously.(38) Construct pJS-1 consisted of the full length human LTBP1 coding sequence cloned into pcDNA3. Hemagglutinin (HA) epitope-tagged deletion constructs were also used, as described previously.(38) Construct pJS-2 encoded amino acids 507 to 1359, construct pJS-9 encoded amino acids 1 to 467, construct pJS-10 encoded amino acids 1 to 994, construct pJS-11 encoded amino acids 67–467. In addition an HA-tagged full length LTBP1 construct was made by ligating the 3.4-kb Bsu36I-PvuI fragment of pJS-1 to the 6.2-kb Bsu36I-PvuI fragment of pJS-2. This HA-tagged full-length human LTBP1 construct was termed pJS-1T (amino acids 1–1359). Schematic diagrams showing the domain structures encoded by these LTBP1 deletion constructs are given in Fig. 5.
Transient expression of LTBP1 constructs in 2T3 osteoblast-like cells and 293 cells
To examine incorporation of HA-tagged LTBP1 deletion constructs into fibrillin-containing microfibrils, the mouse 2T3 osteoblast-like cell line was used. This cell line was selected for these studies for the following reasons: (1) 2T3 cells are easier to transfect than primary osteoblasts and many of the available osteosarcoma cell lines, which give very low transfection efficiencies (data not shown); (2) they express LTBP1 and fibrillin 1 at similar levels to primary osteoblasts (S. Dallas, unpublished observations); and (3) they more closely resemble primary osteoblasts in terms of in vitro responses to various stimuli than many of the available osteosarcoma cell lines.(36) An additional advantage of the 2T3 osteoblast cell line over primary cells is that it is possible to produce stably overexpressing cell lines from these immortal cells after selection with antibiotics.
For immunocytochemical staining experiments, 2T3 cells were plated on collagen-coated 18-mm-diameter glass coverslips in 12-well plates at 0.8 × 105 cells/well in 2 ml of α-MEM + 10% FBS. After overnight adherence, the cells were transfected with 0.5 μg of supercoiled DNA using the “lipofectamine plus” transfection system according to manufacturer's instructions (Life Technologies, Gaithersburg, MD, U.S.A.). Control cells were transfected with empty vector DNA (i.e., without LTBP1 insert). Transfection conditions were optimized for the 2T3 cell line using a human tartrate-resistant acid phosphatase construct as a reporter gene as described elsewhere.(39) Under optimal conditions, transfection efficiencies of 70–90% were obtained using this reporter construct (data not shown). To assess incorporation of LTBP1 deletion constructs into fibrillin 1–containing microfibrils, double-labeled immunofluorescent staining was performed as described above using a mouse monoclonal anti-HA antibody in conjunction with a rabbit polyclonal anti–fibrillin 1 antibody (PAb 9543).
For Western blotting analysis of transient transfection experiments, 2T3 or 293 cells were plated in 90-mm Petri dishes at 106 cells/well in 10 ml of α-MEM + 10% FBS. The cells were transfected with 5 μg of supercoiled DNA as described above. To examine expression of the constructs, 6 ml of serum-free conditioned medium was collected after a 48-h incubation in the presence of protease inhibitors (20 μg/ml aprotinin, 1 μM pepstatin A, and 10 μM leupeptin). This conditioned medium was concentrated 30-fold by using a Centriplus 30-kDa–cutoff concentrator according to the manufacturer's instructions (Amicon, Beverly, MA, U.S.A.). Concentrated samples of medium were then analyzed by Western blotting as described below.
Characterization of anti-LTBP1 hinge peptide antibodies
To determine whether LTBP1 colocalized with well-known ECM proteins such as type I collagen, fibronectin, and fibrillin 1, it was first necessary to characterize the new anti-LTBP1 antibodies generated in rabbit and in chicken. Immunoaffinity-purified antibodies against the rat LTBP1 hinge peptide were characterized by enzyme-linked immunosorbent assay (ELISA), Western blotting, immunoprecipitation, and immunocytochemistry. By Western blotting analysis under reducing conditions, both the rabbit and chicken antibodies recognized hemagglutinin-tagged human LTBP1, which was stably overexpressed in Chinese hamster ovary cells (Fig. 1). Both antibodies recognized a triplet of bands at 190–220 kDa, which were also recognized by an anti-HA polyclonal antibody. These LTBP1-specific bands were not detected in control empty vector–transfected cells. Specificity was further confirmed by preincubating the antibodies with 50 μg/ml of LTBP1 hinge peptide, which abolished the immunoreactive bands. Both the rabbit and chicken antibodies also recognized human LTBP1 by Western blotting under nonreducing conditions. However, these antibodies did not recognize human LTBP1 by ELISA, immunoprecipitation, or immunocytochemistry (data not shown), suggesting that they did not recognize human LTBP1 under native conditions. Both the rabbit and chicken antibodies recognized murine and rat LTBP1 by immunocytochemistry and immunoelectron microscopy (Figs. 2, 3, 4). Furthermore, the rabbit antibody was able to immunoprecipitate rat LTBP1 from the conditioned medium of fetal rat calvarial cell cultures (data not shown). Taken together, these data indicate that both the anti-LTBP1 hinge antibodies recognize rat and murine but not human LTBP1 under native conditions and recognize denatured human LTBP1 by SDS-PAGE and Western blotting analysis. The human and rat sequences differ by 3 amino acids in the region represented by the hinge peptide. These are a valine substitution for isoleucine at position 732 of the rat sequence, a histidine substitution for glutamine at position 740, and a valine substitution for isoleucine at position 742.
LTBP1 colocalizes with fibrillin 1 in osteoblast-like cells
LTBP1 was found to colocalize with fibrillin 1 in a network of discrete fibrils present in the extracellular matrix of osteoblast-like cells in culture. This was found by using both primary cultures of fetal rat calvarial osteoblasts and the human osteosarcoma cell line MG63 (Figs. 2A–2F). These observations were made by using two independent sets of antibodies; for the fetal rat calvarial cells a chicken polyclonal antibody against rat LTBP1 hinge peptide was used in conjunction with a rabbit polyclonal against fibrillin, and for the MG63 osteosarcoma cells a rabbit polyclonal against LTBP1 (Ab39) was used in conjunction with a mouse monoclonal antibody against fibrillin 1. Fibrillar staining for LTBP1 using the chicken polyclonal antibody was abolished by preincubation with the immunizing peptide (Fig. 2N). Similar staining patterns were obtained when each of the primary antibodies was tested individually in single-labeling experiments in the presence of both secondary antibodies, confirming that the observed staining patterns were not a result of species cross-reactivity of the secondary antibodies (data not shown).
LTBP1 and fibrillin 1 were found to colocalize at all stages of maturation in both MG63 and fetal rat calvarial cell cultures, from the initial appearance of LTBP1-positive fibrils when the cells had just reached confluence (Figs. 2A–2C) to the appearance of a well-organized network of fibrils by days 7–10 after confluence, as in the examples shown in Figs. 2D–2F. This is in contrast to the localization pattern found with fibronectin, as described below.
Localization of LTBP1 and fibronectin in osteoblast-like cells
Figures 2G–2M show double-staining immunofluorescence for LTBP1 and fibronectin in primary cultures of fetal rat calvarial cells and MG63 osteosarcoma cells. LTBP1-positive fibrils appeared in the matrix when the cells had just reached confluence (Fig. 2G). At this stage, when the LTBP1-positive fibrils were initially being formed, LTBP1 colocalized with fibronectin in both FRC and MG63 cells (Figs. 2H and 2I). Note that at this stage of culture, the fibronectin network was already well formed and appeared more extensive than the LTBP1 fibrillar network. In contrast, in day 7–10–postconfluent cultures, LTBP1 was localized in large, parallel, threadlike fibrils (Fig. 2K) that were clearly distinct from the fibronectin network (Figs. 2L and 2M). A similar pattern of colocalization was observed between fibrillin 1 and fibronectin, whereby they were found to colocalize at an early stage when the fibrillin 1–positive fibrils were initially being formed, but they did not colocalize at later stages of culture (data not shown).
LTBP1 does not colocalize with type 1 collagen in osteoblast-like cells
Figure 3 shows immunofluorescent staining for LTBP1 and type I collagen in day 7–postconfluent cultures of fetal rat calvarial cells (Figs. 3A–3C) and MG63 osteosarcoma cells (Figs. 3D–3F). In fetal rat calvarial cell cultures, the type I collagen–positive fibrils were clearly distinct from those that were positive for LTBP1 and appeared finer and more diffuse. Time course experiments revealed that by 2 days postconfluence, the LTBP1 fibrillar network was well established. In contrast at this time, type I collagen staining was still predominantly cytoplasmic and collagen fibers were not abundant in the extracellular matrix until 5–6 days postconfluence (data not shown). Similarly, in day 7–postconfluent cultures of MG63 osteosarcoma cells, a well-formed network of LTBP1-positive fibrils was observed, whereas at this stage of culture type I collagen staining was predominantly cytoplasmic, with only a small amount of fibrillar staining (Figs. 3D–3F).
Electron microscopic immunolocalization of LTBP1
To further examine the nature of the fibrils to which LTBP1 localized, immunoelectron microscopy was performed on day 10–postconfluent cultures of fetal rat calvarial cells. LTBP1 was found to localize to ≃10-nm microfibrillar structures in the ECM of these cells with the use of both a polyclonal antiserum against the native protein (Ab39; Fig. 4B) and a rabbit polyclonal antibody against the LTBP1 hinge peptide (Fig. 4C). These fibrils were identical to those that stained with a polyclonal antibody to fibrillin 1 (Ab9543; Fig. 4A). Banded collagen fibrils, which were also present in the ECM, were not labeled by any of these antibodies. In Fig. 4C (inset), using the rabbit anti-LTBP1 hinge peptide antibody, there is a suggestion that the epitope recognized by this antibody occurs periodically along the microfibrils, as has been shown to occur with various monoclonal antibodies against fibrillin 1.(2,40)
The N terminus of LTBP1 is required for incorporation into fibrillin-containing fibrils.
To determine which portion of the LTBP1 protein is important for incorporation of LTBP1 into fibrillin-containing microfibrils, deletion constructs of LTBP1 were transiently transfected into the 2T3 mouse osteoblast-like cell line. Figure 5A shows schematic diagrams of the domain structures of the HA epitope-tagged full-length and deleted LTBP1 constructs that were transiently overexpressed in these cells. Incorporation of these overexpressed constructs into the matrix was monitored by immunofluorescent staining with anti-HA antibodies. 2T3 cells transfected with empty vector showed no fibrillar staining in the matrix using anti-HA antibodies (Fig. 5B), although a well-formed network of fibrillin 1–positive fibrils was clearly present (Fig. 5C). In contrast, in cells transiently transfected with HA-tagged full-length LTBP1 (construct pJS1T), anti-HA staining was observed in small discrete areas adjacent to overexpressing cells (Fig. 5D) and was localized to fibrillar structures that also stained positive for fibrillin 1 (Fig. 5E). When cells were transfected with construct pJS-2 (amino acids 507–1359), no evidence of fibrillar staining with anti-HA was found, although there were many examples of cells that exhibited positive staining in the cytoplasm (Fig. 5F, arrows) and a well-formed network of fibrillin 1 fibrils was clearly seen (Fig. 5G). This construct lacks the 4-cysteine motif at the N terminus of the protein, the first two EGF-like repeats, the hybrid 8-cysteine motif, and the first 8-cysteine repeat, suggesting that one or more of these functional domains are important for incorporation into fibrillin-containing microfibrils. In support of this, when cells were transfected with deletion constructs pJS-9 and pJS-10, which both included at least the N-terminal 467 amino acids, fibrillar staining was observed with anti-HA, which colocalized with fibrillin 1 (Figs. 5H–5K). Cells transfected with construct pJS-11, which was further deleted within this N-terminal region, showed that only amino acids 67–467 were required for incorporation into fibrillin-containing fibrils (Figs. 5L and 5M).
Construct pJS2, which lacked the N-terminal 507 amino acids, was the only LTBP1 construct that did not localize to fibrillin 1–containing fibrils but gave a cytoplasmic staining pattern. This suggests that sequences deleted from this construct are important for incorporation of LTBP1 into fibrillin-containing fibrils. However, an alternative explanation for this result could be that this mutated construct was improperly processed and therefore remained trapped in the endoplasmic reticulum. To determine whether construct pJS2 was efficiently secreted, Western blotting analysis was performed using conditioned media from both 293 and 2T3 cells transiently transfected with the LTBP1-deletion constructs. 293 cells, which readily take up DNA and express recombinant proteins at high levels, secreted construct pJS2 into the conditioned media at levels similar to those found with the other LTBP1-deletion constructs, as determined by immunoblotting with anti-HA (Fig. 6A). 2T3 osteoblast-like cells transfected with the constructs gave specific bands corresponding to the expected sizes of all the constructs (Fig. 6B, arrowheads). Two nonspecific bands were also observed as a result of the lower uptake of DNA and the lower expression level in 2T3 cells compared with 293 cells, which resulted in the need for longer exposure times and thus a higher level of background staining. Again, in 2T3 cells, construct pJS2 was secreted into the culture medium as efficiently or more efficiently than several of the other constructs that gave fibrillar staining, suggesting that the lack of incorporation of this construct into fibrils was not a result of inefficient secretion.
At present, it is not known whether incorporation of LTBP1 into fibrillin-containing microfibrils is a cell surface-mediated phenomenon or whether microfibrils are capable of self-assembly from monomers under appropriate conditions. To address this question in part, we examined whether exogenously added LTBP1 could be incorporated into the microfibrils produced by bone cells. Conditioned medium harvested from 293 cells that overexpressed the HA-tagged N-terminal LTBP1 construct pJS-9 was incubated for 6 days with fetal rat calvarial cell cultures. In this way, exogenously added LTBP1 could be distinguished from the endogenous protein by its immunoreactivity with anti-HA. Double-staining immunofluorescence revealed that the exogenously added LTBP1 N-terminal construct was assembled into fibrillin-positive microfibrils (Fig. 7).
Localization of LTBP1, fibrillin 1, and fibronectin in bone in vivo
To confirm the results obtained with in vitro culture systems and to examine the distribution of LTBP1 in vivo, double-staining immunofluorescence was also performed on tissue sections of developing long bones. Figure 8 shows double-staining immunofluorescence for LTBP1, fibrillin 1 and fibronectin in a 72-day-old fetal long bone. A schematic representation of a developing long bone is shown in Fig. 8A to indicate the locations of the areas depicted in Figs. 8B–8M. Skeletal structures initially develop as a condensation of cartilaginous mesenchyme. In long bone development, this cartilage anlagen is surrounded by a “collar” of mesenchymal cells that gives rise to a bilayered periosteum. The inner layer of this periosteum secretes osteoid matrix, first at the mid-diaphysis, which is then mineralized. Hypertrophy of the chondrocytes in the cartilage core follows, and the osteoid collar advances toward both epiphyses in a wave of cellular differentiation.(41) The fields depicted in Figs. 8B and 8C were taken from the advancing osteoid front, and those depicted in Figs. 8D–8K were taken from the mineralized bone collar, which at this stage of development existed as a single layer. At the advancing osteoid front, LTBP1 staining was observed in fibrils in the outer periosteum that ran parallel to the long axis of the bone (Figs. 8B and 8C). In addition, the first layer of osteoblasts, which was directly adjacent to the newly forming bone and osteoid, was also positive. Interestingly, a line of fibrillar LTBP1 staining could be observed (Fig. 8C, arrowheads) that appeared to demarcate the boundary between the cartilage and newly forming bone. Individual fibrils could be seen to extend out beyond the advancing edge of the newly forming osteoid in a manner that appeared to delineate where new bone would subsequently form.
At the mineralizing bone collar, LTBP1 staining was again observed in parallel fibrils in the outer periosteum, but in addition the layer of osteoblasts adjacent to the bone surface was also positive for LTBP1 (Figs. 8D and 8H). The line of fibrillar LTBP1 staining that demarcated the boundary between cartilage and osteoid appeared to be absent once the osteoid had become mineralized. Double staining for fibrillin 1 revealed that it colocalized with LTBP1 in the fibrils in the outer periosteum (Figs. 8E and 8F). Fibrillin 1 also colocalized with LTBP1 in osteoblasts adjacent to the bone surface. Double staining with fibronectin revealed that the LTBP1-positive fibrils in the outer periosteum also contained fibronectin (Figs. 8H–8J). However, in contrast to LTBP1, fibronectin staining was also observed in the inner layer of the periosteum. In addition, LTBP1 and fibronectin were present in distinctly different cell layers at the bone surface (Fig. 8J).
A different localization pattern was observed in cartilage compared with bone (Fig. 9). LTBP1-positive fibrils were observed throughout the cartilage matrix, as shown in Fig. 9C. However, these fibrils did not appear to contain fibrillin 1 (Fig. 9D). In the perichondrium, LTBP1 and fibrillin 1 were colocalized in parallel fibrils, although the LTBP1 staining appeared to extend further into the cartilage matrix than the fibrillin 1 staining (Figs. 9E–9H). Neither the fibrils in the perichondrium nor those in the cartilage matrix stained positively for fibronectin (data not shown).
LTBPs have a striking degree of domain homology with the fibrillins, suggesting that they may be extracellular matrix proteins with related functions. In the present study we have shown, by using both an antibody against purified native LTBP1 and antibodies to a short peptide sequence specific for LTBP1, that LTBP1 and fibrillin 1 are colocalized in an organized network of fibrillar structures in the extracellular matrix of bone cells in vitro and in vivo. By immunoelectron microscopy we have confirmed that these LTBP1-positive fibrillar structures are true microfibrils that are identical to those that stain with fibrillin-1 antibodies. Furthermore, immunogold localization of LTBP1 using antibodies directed against a short peptide sequence showed evidence of periodicity, as has been shown with various monoclonal antibodies against fibrillin 1.(2,40) This is suggestive of an ordered incorporation of LTBP1 into microfibrils and microfibrillar bundles.
Fibrils positive for LTBP1 and fibrillin 1 first appeared in the matrix of fetal rat calvarial osteoblasts and MG63 osteosarcoma cells at the time the cells reached confluence and initially colocalized with fibronectin. However, in day 7–10–postconfluent cultures the fibrils that stained for LTBP1 and fibrillin 1 were clearly distinct from the fibronectin network. Thus it appeared that LTBP1 and fibrillin 1 were initially incorporated into the matrix in association with fibronectin but then became organized into a distinct microfibrillar network, which appeared later in culture and did not contain fibronectin. This initial colocalization of LTBP1 and fibrillin 1 with fibronectin may reflect a matrix assembly process in which newly synthesized extracellular matrix is deposited onto a fibronectin template as in the model described by Mosher and McKeown-Longo.(42) Interestingly, it has been shown that fibronectin assembly is required for the incorporation of fibulin 1 (a fibrillin-related protein) into the extracellular matrix of fibroblasts.(43) Thus it is possible that fibronectin assembly may similarly be required for the initial incorporation of LTBP1 and fibrillin 1 into the matrix. In support of this, colocalization of LTBP1 and fibronectin has been reported in fibroblasts, and a direct interaction between LTBP1 and fibronectin has been shown with ligand blotting assays.(24) Consistent with these findings, we have also observed that when LTBP1 is stably overexpressed in a rat osteosarcoma cell line (UMR-106) that does not constitutively express LTBP1 or fibrillin 1, it is incorporated into the extracellular matrix in association with fibronectin (S. Dallas, unpublished observations).
LTBP1/fibrillin 1–positive fibrils appeared in the matrix of bone cell cultures several days before the appearance of banded collagen fibrils, and at no time were LTBP1 or fibrillin 1 found to colocalize with type I collagen, the major component of bone extracellular matrices. Although no direct relationship was found to exist between LTBP1/fibrillin-1 fibrils and banded collagen fibrils, it remains to be determined whether incorporation of LTBP1/fibrillin 1 into the matrix is required for efficient incorporation of type I collagen.
At present, it is not clear whether microfibril assembly is cell mediated or whether microfibrils are capable of self-assembly from monomers under appropriate conditions. Our data show that exogenously added N-terminal LTBP1 can be incorporated into fibrillin-containing microfibrillar structures in the matrix of bone cells. Although endogenous production of LTBP1 or other microfibrillar proteins may still be required for assembly of the microfibrils, these data suggest that soluble LTBP1 can be efficiently incorporated into the matrix. This allows for the possibility that LTBP1 could be produced by cells at a site distant from the site at which fibril formation occurs.
Recent studies have begun to examine the mechanism by which LTBP1 is incorporated into the extracellular matrix. The covalent association between LTBP1 and the extracellular matrix has been shown to be transglutaminase dependent.(44) Furthermore, it has been reported that the amino-terminal region of LTBP1 is important in matrix incorporation because (1) antibodies to the amino terminus of LTBP1 can block transglutaminase-dependent incorporation of LTBP1 into the extracellular matrix(44) and (2) deletion of 400 amino acids from the N terminus of LTBP1 prevents efficient incorporation into the matrix in COS cells.(38) By using transient expression of HA epitope–tagged deletion constructs for LTBP1, we have confirmed that the N-terminal portion of the protein is required for incorporation into the ECM of bone cells in vitro and extended these observations specifically to show that the N-terminal portion is required for incorporation into fibrillin-containing fibrils. This immunolocalization data using antibodies directed against an epitope tag that is unrelated to any of the fibrillins or LTBPs provides independent corroboration that LTBP1 and fibrillin 1 are colocalized in bone ECM. The minimum LTBP1 sequence tested that showed binding to fibrillin-containing fibrils consisted of amino acids 67–467. This portion of the protein includes the first two EGF-like repeats, the hybrid 8-cysteine motif and the first 8-cysteine repeat. Future studies will further define the binding region within this portion of the protein.
No gross abnormalities in LTBP1 or fibrillin-1 staining were observed in 2T3 cells that transiently expressed the HA-tagged LTBP1 deletion constructs, suggesting that these constructs do not interfere grossly with fibril assembly. However, due to the low transfection efficiency in the 2T3 cells, the HA staining was confined to small discrete areas surrounding isolated islands of overexpressing cells. Thus the overall expression level of the deletion constructs may have been too low relative to expression of endogenous proteins to cause gross abnormalities in microfibril formation. Future studies using stably expressing 2T3 cell lines will examine the potential ability of these deleted LTBP1 constructs to influence microfibril assembly.
In vivo immunolocalization studies in developing long bones showed that LTBP1 and fibrillin 1 were colocalized in longitudinal fibrillar structures in the outer periosteum and in the perichondrium. Fibronectin was present in the fibrils in the periosteum, but was not present in the fibrils in the perichondrium. LTBP1 and fibrillin 1 were also found to colocalize in the layer of osteoblasts adjacent to the surface of newly forming osteoid and bone. In addition, a line of fibrillar LTBP1 staining appeared to delineate the boundary between the newly forming osteoid and the underlying cartilage. Individual fibrils could be seen to extend out beyond the advancing edge of the newly forming osteoid as if demarcating where bone would subsequently form. This distribution suggests a role for LTBP1-containing microfibrils in bone formation, perhaps by directing cell migration and attachment or by acting as a scaffold for matrix formation. In support of a role in bone formation, a previous in vitro study showed that antibodies and antisense oligonucleotides against LTBP1 inhibited bone formation in cultures of fetal rat calvarial osteoblasts.(25) Interestingly, a recent study has demonstrated that Marfan syndrome is associated with a lower bone mineral density and that these patients are at greater risk for osteoporosis compared to the normal population.(45) At present the mechanism for this is unclear, however, it is possible that members of the fibrillin superfamily may play a role in bone formation or in determination of bone mineral density.
Colocalization of LTBP1 and fibrillin 1 to fibrillar structures in tendon and in muscle septae was also observed (data not shown). However, their patterns of staining intensity appeared very different; thus with fibrillin antibodies the fibrils in tendon stained intensely and those in the muscle septae stained weakly. The reverse was true for staining with LTBP1 antibodies. Interestingly, LTBP1-positive fibrils were observed within the cartilage matrix, which were negative for both fibrillin 1 and fibronectin, suggesting a specialized role for LTBP1 in microfibrillar structures in the cartilage matrix. Taken together these data suggest that although LTBP1 and fibrillin 1 are colocalized in some tissues, they also show distinct patterns of distribution, indicating that they may have overlapping but distinct roles in microfibrillar function. This is consistent with other reports on the “heterogeneity of microfibrils.” It is now becoming apparent that microfibrils are composed of a number of constituent proteins and proteoglycans, including; fibrillins 1 and 2,(2,3,46) LTBPs 2 and 3,(7,9) microfibril-associated proteins (MAGPs) 1 and 2,(47,48) fibulins 1 and 2,(49,50) fibronectin,(51) versican,(52) thrombospondin,(53) and emelin.(54) The ratios of these constituents may vary in different tissues to create microfibrils with distinct and perhaps tissue-specific functions.
In summary, LTBPs have a striking degree of domain homology with the fibrillins, and LTBP2 has previously been shown to colocalize with fibrillin 1 in 10- to 12-nm microfibrils. In the present study we have now demonstrated that LTBP1 colocalizes with fibrillin 1 in an organized network of fibrillar structures in the extracellular matrix of bone cells in vitro and in vivo. These findings add LTBP1 to the growing list of proteins that are components of microfibrils and therefore candidate genes for Marfan-related connective tissue disorders where linkage to fibrillin-1 and fibrillin-2 genes have been excluded. An exciting recent finding is the detection of mutations in LTBP2 in three patients with a Marfan-related syndrome that manifests with a predominantly skeletal phenotype (Maurice Godfrey, personal communication, October 1998). Dermal fibroblasts from these patients showed defects in microfibril assembly, suggesting that LTBP2 (and perhaps other members of the LTBP family) are integral components of microfibrils. The additional role of the LTBPs in closely regulating the activity of TGF-β provides an added dimension to their function as extracellular matrix proteins. The further study of this important new class of multifunctional matrix proteins will therefore provide valuable insights into the role of microfibrils in health and disease.
We thank Anthony J. Makusky of the Institutional Protein Core Facility at the University of Texas Health Science center at San Antonio for advice and expert technical assistance with design, synthesis and coupling of LTBP1 peptides. Electron microscopy facilities were provided in part by the Fred Meyer and R. Blaine Bramble Charitable Trust Foundations. We also thank Sara Ford Tufa for her excellent technical assistance, the Central Lab For Human Embryology at the University of Washington for their generous cooperation, and Frances J. Ramirez for her expert secretarial assistance. Supported by a research grant from the National Osteoporosis Foundation, by grant R01AR43775 from the National Institutes of Health, and by the Academy of Finland.