THE ANATOMICAL RECORD 238:23-30 (1994) Intramembranous Bone Matrix Is Osteoinductive CAROLINE KELLY SCOTT, STEVEN DOUGLAS BAIN, AND JAMES ANDERSON HIGHTOWER Department of Cell Biology and Neuroscience, School of Medicine, University of South Carolina, Columbia, South Carolina (C.K.S., J.A.H.);and Zymogenetics, Znc., Seattle, Washington (S.DB.) ABSTRACT All known bone-derived osteoinductive factors have been isolated from endochondral (EC) bones and all initiate bone induction via EC ossification. However, to date no attempt has been made to isolate comparable factors from bones which form initially and completely via intramembranous (IM) ossification. The purpose of this work was to isolate osteoinductive proteins from I M bones. To accomplish this, we extracted proteins from bovine frontal bone matrix (intramembranous origin) using methods previously described for endochondral (EC) bone matrix (i.e., femur). Bone powder (<1 mm) was decalcified and proteins extracted with 4 M guanidine hydrochloride. Ultrafiltration was used to isolate and concentrate a 10-100 kilodalton (kDa) fraction, upon which heparin-Sepharose (HS) affinity chromatography was performed. HS-binding (HS-B)and nonbinding proteins (HS-NB)were lyophilized with bovine type I collagen (Vitrogen) to form pellets which were implanted subcutaneously in rats. Radiology as well as brightfield, fluorescent, and polarizing microscopy were used to assess the formation of ectopic bone at the site of pellet implantation. In this report we demonstrate that a heparin-Sepharose binding, osteoinductive factor can be extracted and partially purified from bovine intramembranous bone matrix. This factor has a different sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) banding pattern than a comparable osteoinductive/chondroinductivefactor isolated from EC bone. o 1994 Wiley-Liss, Inc. Key words: Growth factors, Morphogenesis, Bone development, Bone matrix, Sprague Dawley rats, Heterotopic ossification, Affinity chromatography, Polyacrylamide gel electrophoresis Several osteoinductive factors have been isolated and purified (Urist e t al., 1984; Bentz et al., 1989; Luyten et al., 1989; Sampath et al., 1987; Wozney et al., 1988; Wang et al., 1988; Sampath et al., 1990), which, when combined with a collagenous carrier, can be used to repair bony defects by the induction of endochondral (EC) ossification (bone replacing a cartilage precursor). All known bone-derived osteoinductive factors have been isolated from EC bone, and all initiate bone via EC ossification (Urist et al., 1984; Bentz et al., 1989; Luyten et al., 1989; Sampath et al., 1987; Wozney e t al., 1988; Wang et al., 1988; Sampath et al., 1990). No attempt has been made to isolate comparable factors from bones which form initially and completely via intramembranous (IM) ossification. IM bones include those which form the rostra1 parts of the skull such as the frontal bones. These are dermal, exoskeletal, of neural crest origin and form directly from connective tissue membranes without a cartilaginous intermediate stage (Smith and Hall, 1990; Hall B.K., Personal Communication, LeLievre, 1978). It is well known that intramuscular or subcutaneous implantation of demineralized EC bone matrix induces de novo bone via the EC sequence (Reddi and Huggins, 0 1994 WILEY-LISS, INC. 1972). Although Reddi and Huggins (1972) suggested that IM matrix also induces the EC cascade, it was unclear to us why EC and IM bone powders should possess identical osteoinductive properties. In a n earlier study, we compared the inductive capacities of EC and IM demineralized bone matrix (Scott and Hightower, 1991). Morphological and radiolabeling techniques demonstrated that implants of EC bone matrix induce bone formation via EC ossification in contrast to implants of IM bone matrix which induce only IM ossification. Our results suggested that IM bone matrix contains osteoinductive factors which induce direct bone formation. This finding, which had not been described previously, coupled with the fact that no osteoinductive factors have been isolated from IM bone matrix, stimulated us to search for such a morphogen. We report here the extraction and partial purification of a n osteoinductive factor from bovine IM bone matrix using ultrafiltration and heparin affinity chromatogReceived April 23, 1993; accepted July 26, 1993 Address reprint requests to Dr. James A. Hightower, Department of Cell Biology and Neuroscience, School of Medicine, University of South Carolina, Columbia, SC 29208. 24 C.K. SCOTT ET AL. raphy. Partial characterization of the extract was performed via SDS-PAGE. IM bone extract has a n SDSPAGE banding pattern which differs from that of a comparable extract isolated from EC bone. MATERIALS AND METHODS Extraction of Bone Proteins Each cow skull is composed, in part, of a pair of bilaterally symmetrical frontal bones (intramembranous) (IM). Each frontal bone, in turn, lies adjacent to three IM bones: the nasal bone anteriorly, the parietal bone posteriorly, and the opposite frontal bone medially. However, contamination of frontal bone preparations can occur with EC bone (Moore, 1988) if the sphenoid bone is incompletely separated from the frontal bone during dissection. Special care was taken to avoid all sutures so as not to contaminate IM bone with adjacent EC bone. Bovine skulls, 18 to 36 months of age, were obtained from local abattoirs. Frontal bones were dissected free of all adjacent bones and adjacent tissues and then cleaned, defatted, and pulverized to particle size <1mm in a SPEX liquid nitrogen freezer mill. Resulting bone powder was demineralized in 0.5 M HC1 as described previously (Reddi and Huggins, 1972) yielding demineralized bone powder. Proteins were extracted from demineralized IM bone powder using methods previously described for EC bone powder (Luyten et al., 1989; Sampath et al., 1987). To summarize, powder was extracted for 20 hours a t 25°C in 4 M guanidine hydrochloride (GuHC1)/50 mM Tris-HCU0.1 M NaCl, pH7.0, containing the protease inhibitors N-ethylmaleimide (NEM) and phenylmethylsulfonyl fluoride (PMSF). The GuHC1soluble extract was suction-filtered. Concentration of 10-100 kilodalton (kD) proteins was accomplished by ultrafiltration with a 100 kD followed by a 10 kD cutoff membrane (Pellicon system, Millipore). The 10-100 kD fraction was dialyzed (Spectrapor, Spectrum Medical Industries) against cold deionized water and lyophilized. This 10-100 kD fraction is referred to below as pre-heparin-Sepharose (PRE-HS). A portion of PRE-HS material was reserved for the biological assay. Heparin-Sepharose (HS) Chromatography Lyophilized proteins (PRE-HS) were suspended (12.5 mglml) in 6 M urea/50 mM Tris-HCl/O.15 M NaC1, pH 7.4, containing NEM and PMSF and loaded onto a preequilibrated 40 ml HS column (Pharmacia). HeparinSepharose non-binding (HS-NB) proteins were collected and the column was washed with several volumes of buffered urea. Elution of heparin-Sepharose binding (HS-B) proteins was performed with buffered urea containing 0.5 M NaC1. HS-NB and HS-B fractions were dialyzed against cold deionized water, lyophilized, and stored at -80°C until ready for biological assay or biochemical analysis. Biological Assay Implant preparation All studies were performed in accordance with the PHS policy on Laboratory Animal Welfare, the American Veterinary Medical Association guidelines concerning animal welfare, and the Animal Review Committee, University of South Carolina Medical School. Fig. 1. The arrow indicates an implant site 39 days after 4 mg of HS-B bovine frontal bone extract (IMBE) was subcutaneously implanted into the ventral thorax of a 12-week-oldSprague Dawley rat. x 5. Putative osteoinductive proteins were extracted from bovine frontal bones and lyophilized at three stages of partial purification (PRE-HS, HS-B, and HS-NB). A portion (1-20 mg) of each of the three fractions was combined with 4 ml (3 mg/ml) of telopeptide-free bovine type I dermal collagen (Vitrogen, Collagen Cop.), stirred for -6 hours a t 4"C, and lyophilized to form pellets which were subcutaneously implanted into the ventral thorax of male Sprague Dawley rats (12-15 weeks). Negative controls contained Vitrogen only. Radiological examination Rats were examined radiologically utilizing a Transworld 325 x-ray machine. Radiographs (300 mA, 40 kVp, 2 sec) were taken on days 11,19,28,34, and 39 in order to assess the presence of osseous tissue at the site of pellet implantation. Fluorochrome labeling of de novo bone formation Rat hosts received a series of tetracycline injections in order to assess the rate and quantity of de novo bone formation at each implantation site. Oxytetracycline 100 (Vedco Co.) was injected (i.p., 30 mg/kg) on days 25-28. Fluorescent microscopy was performed with a n Olympus Model BH-RFL System using two BG-12 exciter filters and a 0-515 barrier filter. INTRAMEMBRANOUS BONE MATRIX 25 Fig. 2.Osseous tissue from implant site observed in Figure 1. This tissue is decalcified, embedded in paraffin, and stained with hematoxylin and eosin. Note the trabeculae of woven bone (b). Most osteocytes (0)are haphazardly arranged in the matrix. Endosteum lines the marrow cavities (m) and periosteum (p) envelops the osseous tissue. Bar, 10 pm. Implant processing All implants were extirpated, embedded, sectioned, and stained. Histological examination of radiologically opaque implants that were decalcified, embedded in paraffin, and stained with H&E revealed a significant amount of spongy, woven (primary) bone and active bone marrow, but no hyaline cartilage, in two of the three IM HS-B implants (Fig. 2). Neither concentric, interstitial, nor circumferential lamellae were obSodium Dodecyl Sulfate Polyacrylamide Gel served. Nonradiologically opaque implants were devoid Electrophoresis (SDS-PAGE). of both hyaline cartilage and bone. Undecalcified plastic embedded sections of radiologSDS-PAGE was used to characterize protein extracts a t various stages of purification. Protein was solubilized ically opaque implant sites also revealed numerous train a 6 M urea sample buffer and run under reducing beculae/spicules of woven (primary) bone and well deconditions with dithiothreitol on 15% gels, which were veloped, highly proliferative red bone marrow, but no hyaline cartilage (Fig. 3a). Hence, 39 days after imfixed and silver stained (Gerton and Millette, 1986). plantation of the HS-B fraction of frontal bone, both RESULTS decalcified and undecalcified tissue sections demonIdentification of Osseous Tissue strate that radiologically opaque implant sites consist, Radiological screening on the 11th day following im- in large part, of a central region composed of woven plantation showed opacities a t selected implant sites. (primary) spongy bone and red bone marrow devoid of The sites contained 4 mg of the heparin-Sepharose hyaline cartilage separated from surrounding skeletal binding (HS-B) fraction of frontal bone. The radiologi- muscle and connective tissue of the anterior thoracic cal opacities became progressively more dense until wall by a thick capsule of dense connective tissue. The day 39 (Fig. 1)when the implants were removed. Opac- bone marrow is typical of that observed during fetal ities were not observed in any of the implant sites con- development (i.e., hyperplastic with a few unilocular taining the heparin-Sepharose nonbinding (HS-NB) adipocytes randomly scattered throughout the hefraction of frontal bone or the pre-heparin-Sepharose matopoietic tissue). (PRE-HS) fraction of frontal bone. Undecalcified plastic embedded tissue sections of ra- The implants were excised on day 39 and prepared for routine histologic examination. Undecalcified, methylmethacrylate-embedded tissue was stained with Von Kossa and toluidine blue. Decalcified, paraffin-embedded tissue was stained with either hematoxylin and eosin, or alcian blue for cartilage. 26 C.K. SCOTT ET AL. Fig. 3. Osseous tissue from the implant site observed in Figure 1. This tissue which is adjacent to that observed in Figure 2 is undeculcifzed, embedded in methylmethacrylate and stained with toluidine blue and von Kossa. a: This low power view demonstrates numerous trabeculae (t) of spongy bone separated from one another by highly proliferative bone marrow (m). Enclosed regions [bl, [cl, [d], and [el are enlarged in b, c, d, and e. [b]: Region of osteogenesis: Note osteoid (os), calcified osseous matrix (b), and osteoblasts (ob). [cl: Region of bone resorption: Note multinucleated osteoclast (oc) in Howship’s lacunae. [d]: Putative de novo region of IM ossification: Note osteoblasts (ob) and tissue stained positively by von Kossa (vk) (i.e., bone or calcified osseous matrix). [el: Highly proliferative bone marrow: Note numerous mononuclear cells (mc) and blood vessels (bv). Bar, 10 pm. In summary, a significant amount of viable osseous diologically opaque implants reveal additional details that are not observed in decalcified sections. They in- tissue is observed via brightfield microscopy in radioclude 1) numerous areas of osteogenesis judging from logically opaque implants of IM bone. Polarizing and the thin layers of osteoid immediately adjacent to the brightfield microscopy suggests that most of this tissue calcified osseous matrix. Next to the osteoid, in turn, is is wovedprimary bone. However, fluorescent microsa single layer of ovallround osteoblasts with abundant, copy demonstrates the presence of both woven and basophilic cytoplasm (often with negative Golgi im- lamellarlsecondary bone. Bone is only observed in IM ages) and rounded, eccentrically placed, euchromatic implant sites which contain HS-B proteins. These sites nuclei (Fig. 3b). 2 ) Less common than regions of osteo- are devoid of hyaline cartilage 39 days after implantagenesis are areas of bone resorption which are sug- tion. gested histologically by large, multinucleated osteoSDS-PAGE Characterization clasts located in Howship’s lacunae (Fig. 3c). 3) In The IM fraction which possesses osteoinductive acaddition to the spicules and trabeculae of bone, there are also putative de novo regions of IM ossification as tivity (HS-B) contains two distinct bands below 20 kD suggested by aggregations of small, black granules (Fig. 4,lanes 3 and 4, bands a and b; Fig. 5, lanes 1, 2 , (Fig. 3d). 4) Highly proliferative, well-vascularized and 3, bands b and c) which are absentlor present in much lower amounts in comparable EC fractions (Fig. bone marrow is observed (Fig. 3e). Microscopic analysis of radiologically opaque im- 5, lanes 4 , 5 and 6, regions d and e). Although a distinct plant sites with polarizing filters reveals no birefrin- 30 kD band is typically present in osteoinductive EC gence. Fluorescent microscopic analysis demonstrates HS-B preparations (Fig. 5, lanes 4, 5 and 6, band a; discrete, intensely stained bands as well as diffuse re- Wozney, 19891, no 30 kD band is apparent in our IM gions of staining in the peripheral portion of the radio- HS-B preparation (Fig. 5, lanes 1,2, and 3; Fig. 4, lanes logically opaque implants. Large trabeculae of bone are 3 and 4, region c). However, a polypeptide doublet is devoid of fluorescence in the more central portions of observed at 30 kD in the nonosteoinductive IM PRE-HS fraction (Fig. 4,lanes 1 and 2, band d) which the implant sites. INTRAMEMBRANOUS BONE MATRIX 27 Fig. 3 h . appears to be enriched in the nonosteoinductive IM HS-NB fraction (Fig. 4, lanes 5 and 6, band e; Fig. 6, lanes 1, 2, and 3, band a). No such band was readily apparent in the comparable nonosteoinductive EC HS-NB fraction (Fig. 6, lanes 4, 5, and 6, region b). Extracts of frontal bone, run under reducing and nonreducing conditions, were also compared in order to determine whether a 30 kD polypeptide in the HS-B fraction may have been reduced. No 30 kD band was observed in the HS-B nonreduced gel. DISCUSSION In this report, we demonstrate that a heparinSepharose binding osteoinductive factor can be ex- tracted and partially purified from bovine frontal bone matrix, a n example of what is classically termed intramembranous (IM) bone. Until now all known bone-derived osteoinductive factors have been isolated from endochondral (EC) bone and all initiate bone induction via EC ossification (Urist et al., 1984; Bentz et al., 1989; Luyten et al., 1989; Sampath et al., 1987; Wozney et al., 1988; Wang et al., 1988). No attempt has been made thus far to isolate comparable factors from bones which form initially and completely via IM ossification. This is the first demonstration that osteoinductive factors can also be extracted from osseous tissue which forms initially and completely via IM ossification, and thus indicates a new 28 C.K. SCOTT ET AL. PRE-HS Std kD 1 2 HS-B HS-NB 3 5 4 6 Std kD 97.4 66.2 45 97.4 66.2 45 31 31 21.5 21.5 Fig. 4. Silver-stained 15% SDS-PAGE of bovine frontal bone samples. Lanes 1, 3, and 5, 15 pg; lanes 2, 4, and 6, 30 pg: Pre-heparin Sepharose (PRE-HS); heparin-Sepharose bound (HS-B), and heparinSepharose unbound (HS-NB). HEPARIN-SEPHAROSE BOUND POLYPEPTIDES IM EC kD Std 31 1 2 3 - 4 5 6 Std kD - 31 -21-5 -14-4 Fig. 5. Heparin-Sepharose bound polypeptides extracted from IM and EC osseous tissues. Lanes 1and 4 , 3 0 pg; lanes 2 and 5, 15 pg; lanes 3 and 6,20 pg. Standards (Std) are in the far left and far right hand lanes. source of osseous tissue from which to isolate such factors. Histological examination of numerous implant sites revealed osseous tissue with and without bone marrow cavities. This suggests that IM bone-derived factors induce a cascade of events which are initiated by the appearance of osseous tissue. Bone, in turn, creates the microenvironment which is required for hematopoiesis to occur. There may be clinical settings in which i t would be more appropriate to replace traumatized or diseased osseous tissue with IM rather than EC bone (Scott and Hightower, 1991).Since IM bone grafts are often superior to EC bone grafts (Smith and Abramson, 1974; Zins and Whitaker, 1983), we reasoned that a factor which could be extracted from IM bone could be potentially more beneficial in the clinical repair of bony defects than a factor extracted from EC bone. From this hy- 29 INTRAMEMBRANOUS BONE MATRIX HEPARINSEPHAROSE UNBOUND POLYPEPTIDES IM EC kD Std 1 2 3 - 31 4 5 kD Std - 31 Fig. 6. Silver-stained 15%SDS-PAGE of heparin-Sepharose unbound polypeptides extracted from IM and EC osseous tissues. Lanes 1 and 4, 30 pg; lanes 2 and 5, 15 kg; lanes 3 and 6, 20 pg. pothesis arose our interest in isolating a factor from IM bone matrix which could induce direct bone formation. We proposed that such a factor would most likely be present in IM bones since they form initially and completely via IM ossification in vivo and since IM demineralized bone powder induces IM ossification (Scott and Hightower, 1991). Our results are consistent with the hypothesis that the factor isolated from IM bone induces direct bone formation. However, we have not proven the hypothesis because we have not excluded the possibility that at some time during bone formation, cartilage may have been present, even though it is not at day 39. This is explained in more detail below. In our protocol, intramembranous bone extract (IMBE) was not combined with the typical carrier, guanidine hydrochloride insoluble rat bone matrix, which contains undefined growth factors. As a n alternative, we used Vitrogen, a 3 mg/ml solution of bovine dermal type I collagen (99.9% pure). Use of Vitrogen eliminates the possibility that transforming growth factor-beta (TGF-P), insulin-like growth factor I (IGF-I) and insulin-like growth factor I1 (IGF-11) or other osteoinductive helper factors may be coincidentally implanted with the typical DBM carriers that are commonly used in this assay. Thus, our carrier was nonbiased because it was derived from EC or IM extract rather than the demineralized bone matrix. In order to determine which specific protein is responsible for the osteoinductive activity of frontal bone extract, further purification of IMBE is necessary. We plan to use reverse-phase HPLC (microbore or standard) to purify putative osteoinductive factors. We possess circumstantial evidence that IMBE induces ossification via the IM pathway. This evidence includes 1) our observation of osseous tissue, but no cartilage, in decalcified, paraffin-embedded, H&Estained IMBE implant sites at day 39; 2) a n absence of tissue which stains with alcian blue in those same IMBE implant sites; and 3) a n absence of cartilage in subcutaneous implant sites of demineralized IM bone matrix (Scott and Hightower, 1991).Regarding the last point, since demineralized frontal bone matrix is the parent substratum for IMBE, we think it is unlikely that chondrogenesis would precede osteogenesis in IMBE implant sites. However, we have not demonstrated this fact unequivocally because we have not examined histologically protein implant sites a t early and intermediate stages of bone development. Radiological data was obtained a t early, intermediate, and late time frames; however, a complete histological analysis has only been done on day 39 implant sites. Therefore, despite our impression that IMBE induces direct bone formation, we can not exclude the possibility of complete resorption of cartilage occurring in the implant site prior to the time that it was examined. The majority of known molecules associated with chondro- and osteo-induction have apparent molecular weights of 30 kD or less (Luyten et al., 1989; Wozney et al., 1988; Wang et al., 1988; Sampath et al., 1990; Joyce et al., 1989). Several groups have described 16, 18, and 30 kD polypeptides associated with osteoinductive activity (Luyten et al., 1989; Wang et al., 1988; Sampath et al., 1990; Wang et al., 1990). Recombinant human Bone Morphogenetic Protein-2 (rhBMP-21, the only molecule known to possess complete cartilage and bone inductive activity (Wozney, 19891, has a molecular weight of approximately 30 kD and is composed of two disulfide-linked 16 to 18 kD subunits (Wang et al., 1990). Purification of bovine osteogenic protein (OP) has shown that it also migrates a t a n apparent molecular weight of 30 kD and upon reduction yields two subunits that migrate a t molecular weights of 16 and 18 kD (Sampath et al., 1990). In contrast t o the above work, our SDS-PAGES demonstrate that no 30 kD molecules are present in either reduced or nonreduced gels of IM HS-B fractions. This, in turn, suggests that the mechanism by which bone formation is induced by IM and EC extracts may also differ. How does the biological activity of IMBE compare to that of other bone growth factors previously described 30 C.K. SCOTT ET AL. such as IGF-I, IGF-11, TGF-P, and the BMPs? Recently TGF-p, a 25 kD homodimeric peptide, has been implicated in playing a major role in bone induction (Noda and Camilliere, 1989; Mackie and Trechsel, 1990; Beck et al., 1991; Joyce et al., 1989, 1990). Repeated subperiosteal injections of TGF-P in the frontal and parietal bones [IM bones] of the r a t induce IM bone formation (Noda and Camilliere, 1989; Mackie and Trechsel, 1990). In addition, a single application of TGF-P1 to a 12 mm skull defect in rabbits induces bony closure of the defect without evidence of a cartilage intermediate (Beck et al., 1991). In contrast, repeated injections of TGF-fi under the periosteum of the femur stimulate IM and EC ossification (Joyce et al., 1989,1990).However, Sampath et al. (1987) have shown that a subcutaneous TGF-P injection does not induce bone formation. Instead it causes the formation of granulation tissue. Therefore, it appears from the work of Joyce et al. (1989, 1990) that, although TGF-P stimulates the terminal differentiation of osteoblast precursors, the molecule lacks the ability to induce differentiation of undifferentiated cell types into osteoblasts (Sampath e t al., 1987). The cellular response to TGF-P appears directly related to the committed phenotype at the site of TGF-p administration (Beck et al., 1991). Although TGF-p plays a significant role in the process of osteoinduction, we have two pieces of evidence which suggest that i t does not play a n osteoinductive role in our model system. First, TGF-p does not have affinity for heparin (George-Nascimento and Fedor, 1990) and, therefore, should not be present in the HS-B preparation that we implanted. Second, a s mentioned previously, IMBE is combined with Vitrogen rather than guanidine HC1 insoluble rat bone matrix, a potential source of TGF-p and other osteoinductive agents. Although the mechanism by which IMBE induces bone formation is not understood, we do know that the electrophoretic pattern of a n osteoinductive extract of bovine frontal bones differs in several ways from a comparable femoral bone extract. Hence, the mechanism by which bone formation is induced by these respective extracts may also differ. ACKNOWLEDGMENTS We greatly appreciate the help and advice of the following individuals: Dr. Brian Genge, Dr. Roy Wuthier, and Dr. Yoshinori Ishikawa (Department of Chemistry, USC); Dr. Duncan Howe (Department of Radiology); Dr. Larry Lamb (Department of Biology, USC); Dr. Clarke Millette and Dr. Sean Newton, Mr. Don Shenenberger, Ms. Denise Evering, and Ms. Neda Osterman (Department of Cell Biology and Neuroscience, USC); Dr. Ann Prewett (Osteotech, Shrewsbury, NJ); and Ms. Megan Lantry (Zymogenetics, Seattle, WA). A portion of this work was published in abstract form in the Journal of Bone and Mineral Research and in Connective Tissue Research. LITERATURE CITED Beck, L.J., L. Deguzman, W.P. Lee, Y. Xu, L.A. Mcfatridge, N.A. Gillett, and E.P. Amento 1991 TGF-p, induces bone closure of skull defects. J. Bone Miner. Res., 6t1257-1265. Bentz, H., R.M. Nathan, D.M. Rosen, R.M. Armstrong, A.Y. Thompson, P.R. Segarini, M.C. Mathews, J.R. Dasch, K. A. Piez, and S.M. Seyedin 1989 Purification and characterization of a unique osteoinductive factor from bovine bone. J. Biol. Chem., 264; 20805-20810. George-Nascimento, C., and J . Fedor 1990 Purification of growth factors In: Protein Purification Applications: A Practical Approach. E.L.V. Harris and S. Angal, eds. IRL Press, Oxford, pp. 128-136. Gerton, G.L., and C.F. Millette 1986 Stage-specific synthesis and fucosylation of plasma membrane proteins by mouse pachytene spermatocytes and round spermatids in culture. Biol. Reprod., 35t1025-1035. Joyce, M.E., S. Jinguski, A.B. Roberts, M.B. Sporn, and M.E. Bolander 1989 Transforming growth factor-p initiates cartilage and bone formation in uiuo. J. Bone Miner. Res., 4t255-259. Joyce, M.E., A.B. Roberts, M.B. Sporn, and M.E. Bolander 1990 Transforming growth factor-p and the initiation of chondrogenesis and osteogenesis in the rat femur. J. Cell Biol., 110t21952207. LeLievre, C.S. 1978 Participation of neural crest-derived cells in the genesis of the skull in birds. J. Embryol. Exp. Morphol., 47r1737. Luyten, F.P., N.S. Cunningham, S. Ma, N. Muthukumaran, R.G. Hammonds, W.B. Nevins, W.I. Woods, and A.H. Reddi 1989 Purification and partial amino acid sequence of osteogenin, a protein initiating bone differentiation. J . Biol. Chem., 264t1337713380. Mackie, E.J., and U. Trechsel 1990 Stimulation of bone formation in uiuo by transforming growth factor-p: Remodeling of woven bone and lack of inhibition by indomethacin. Bone, lIr295-330. Moore, K.L. 1988 The Developing . -Human. W.B. Saunders Co., Philadelphia, p. 340. Noda, M., and J.J. Camilliere 1989 In uiuo stimulation of bone formation by transforming growth factor-p. Endocrinology 124.22912294. Reddi, A.H., and C. Huggins 1972 Biochemical sequences in the transformation of normal fibroblasts in adolescent rats. Proc. Natl. Acad. Sci. U.S.A., 69r1601-1605. Sampath, T.K., N. Muthukumaran, and A.H. Reddi 1987 Isolation of osteogenin, a n extracellular matrix-associated, bone-inductive protein, by affinity chromatography. Proc. Natl. Acad. Sci. U.S.A., 84t7109-7113. Sampath, T.K., J.E. Coughlin, R.M. Whetstone, D. Banach, C. Corbett, R.J. Ridge, E. Ozkaynak, H. Oppermann, and D.C. Rueger 1990 Bovine osteogenic protein is composed of dimers of OP-1 and BMP-PA, two members of the transforming growth factor-p superfamily. J. Biol. Chem., 265t13198-13205. Scott, C.K., and J.A. Hightower 1991 The matrix of endochondral bone differs from the matrix of intramembranous bone. Cal Tis. Int., 49t349-354. Smith, J.D., and M. Abramson 1972 Membranous vs. endochondral bone autographs. Arch. Otolaryngol., 99t203-205. Smith, M.M., and B.K. Hall 1990 Development and evolutionary origins of vertebrate skeletogenic and odontogenic tissues. Biol. Rev., 65t277-373. Urist, M.R., Y.K. Huo, A.G. Brownell, W.M. Hohl, J . Buyske, A. Lietze, P. Tempst, M. Hunkapiller, and R.J. DeLange 1984 Purification of bovine bone morphogenetic protein by hydroxyapatite chromatography. Proc. Natl. Acad. Sci. U.S.A., 81 t371-375. Wang, E.A., V. Rosen, P. Cordes, R.M. Hewick, M.J. Kriz, D.P. Luxenberg, B.S. Sibley, and J.M. Wozney 1988 Purification and characterization of other distinct bone-inducing factors. Proc. Natl. Acad. Sci. U.S.A., 85r9484-9488. Wang, E.A., V. Rosen, J.S. DAlessandro, M. Bauduy, P. Cordes, T. Harada, D.I. Israel, R.M. Hewick, K.M. Kerns, P. LaPan, D.P. Luxenberg, D. McQuaid, I.K. Moutsatsos, J. Nove, and J.M. Wozney 1990 Recombinant human bone morphogenetic protein induces bone formation. Proc. Natl. Acad. Sci. U.S.A., 87t22202224. Wozney, J.M., V. Rosen, A.J. Celeste, L.M. Mitsock, M.J. Whitters, R.W. Kriz, R.M. Hewick, and E.A. Wang 1988 Novel regulators of bone formation: Molecular clones and activities. Science 242: 1528-1534. Wozney, J.M. 1989 Bone morphogenetic proteins. Prog. Growth Factor Res., lr267-280. Zins, J.E., and L.A. Whitaker 1983 Membranous versus endochondral bone: implications for craniofacial reconstruction. Plast. Reconst. Surg., 72t778-784.