Purification and partial amino acid sequence of osteogenin, a protein initiating bone differentiation.

Osteogenin was purified from bovine bone matrix and its activity monitored by an in vivo bone induction assay. The purification method utilized extraction of the bone-inducing activity with 6 M urea, followed by chromatography on heparin-Sepharose, hydroxyapatite, and Sephacryl S-200. Active fractions were further purified by preparative sodium dodecyl sulfate gel electrophoresis without reduction. Osteogenin activity was localized in a zone between 30 and 40 kDa. The amino acid sequences of a number of tryptic peptides of the gel-eluted material were determined. Reduction and alkylation of purified osteogenin in 7 M guanidine hydrochloride resulted in the total loss of biological activity. Sodium dodecyl sulfate gel electrophoresis under reducing conditions revealed a broad band with an apparent molecular mass of 22 kDa.

Osteogenin was purified from bovine bone matrix and its activity monitored by an in vivo bone induction assay. The purification method utilized extraction of the bone-inducing activity with 6 M urea, followed by chromatography on heparin-Sepharose, hydroxyapatite, and Sephacryl 5-200. Active fractions were further purified by preparative sodium dodecyl sulfate gel electrophoresis without reduction. Osteogenin activity was localized in a zone between 30 and 40 kDa. The amino acid sequences of a number of tryptic peptides of the gel-eluted material were determined. Reduction and alkylation of purified osteogenin in 7 M guanidine hydrochloride resulted in the total loss of biological activity. Sodium dodecyl sulfate gel electrophoresis under reducing conditions revealed a broad band with an apparent molecular mass of 22 kDa.
It is well known that bone has a remarkable potentia1 for repair. However, the biochemical and cellular mechanisms underlying bone repair are not understood. The presence of factors in bone which initiate endochondral bone formation has been amply demonstrated by implantation of demineralized bone matrix in extraskeletal sites (1-4). The sequential developmental cascade in response to implantation of demineralized matrix consists of the following major steps: 1) chemotaxis and attachment of mesenchymal cells to the matrix; 2)proliferation of progenitor cells; and 3) differentiation resulting in the formation of cartilage, bone, and hematopoietic marrow (4, 5). The bone-inductive protein that initiates this cascade, osteogenin, was recently isolated from bovine bone matrix by heparin affinity chromatography (6). This report describes an improved purification method for osteogenin as well as the amino acid sequence of a number of tryptic peptides obtained from this protein.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. TTo whom correspondence should be addressed: Bldg. 30, Rm. 211,

MATERIALS AND METHODS
Partial Purification-For the purification of osteogenin, 5-to 10kg lots of dehydrated diaphyseal bovine bone matrix powder (particle size 74-420 pm, American Biomaterials) were demineralized at room temperature in 0.5 N HCl (seven extractions of 4 volumes each) (2). The acid-demineralized matrix was extracted with 20 volumes of 6 M urea, 50 mM Tris-HC1, pH 7.4, containing 1 M NaCl, 100 mM camino-n-caproic acid, 5 mM benzamidine HCI, and 0.5 mM phenylmethanesulfonyl fluoride at room temperature for 16 h (7). The extract was concentrated, exchanged with 6 M urea to reduce the salt concentration, and loaded onto a 2-liter hydroxyapatite (Pharmacia LKB Biotechnology Inc.) column. The column was washed and eluted as described (6). The 100 mM sodium phosphate eluate was loaded directly onto a 0.5-liter heparin-Sepharose (Pharmacia LKB) column, which was washed and eluted as described (6). The 0.5 M NaCl eluate was concentrated and loaded onto tandem Sephacryl S-200 gel filtration columns (2.6 X 100 cm each), equilibrated with 4 M guanidinium chloride (GdmCl),' 50 mM Tris-HCI, pH 7.4. The material was eluted with the same buffer at a flow rate of 36 ml/h; 20-ml fractions were collected and assayed for biological activity.
Gel Elution-Active fractions (25, 26, and 27) from the S-200 column were concentrated and equilibrated first with 6 M urea, 50 mM Tris-HCI, pH 7.4, and then with SDS sample buffer (0.05 M Tris-HCl, pH 6.8, 10% glycerol, 1% SDS with 6 M urea). Fraction 25 was then applied to a 12.5% acrylamide gel (8 cm X 7 cm X 1.5 mm) prepared according to Laemmli (8). The separating gel was cast 1 day in advance and subjected to pre-electrophoresis for 30 min at 100 V/ gel in order to remove charged impurities (9). Prestained molecular weight standards (Bethesda Research Laboratories) were included in lanes at either end of the gel. After electrophoresis, a nitrocellulose replica of the gel was made by soaking the gel briefly in phosphatebuffered saline (PBS, calcium and magnesium free, GIBCO) at room temperature, placing a nitrocellulose membrane (presoaked in PBS, Schleicher & Schuell, BA 85) over the gel, and putting two pieces of Whatman 3" paper (cut to the same size as the gel and presoaked in PBS), a stack of dry paper towels, and a weight over this assembly. Transfer by capillary action was allowed to proceed for 10 min, after which the blot was washed in PBS supplemented with 0.3% Tween 20 at 37 "C for 15 min with three changes of buffer, washed extensively with distilled water, and incubated for 1 h in Aurodye Forte (Janssen, Life Sciences Products) with constant agitation. The molecular weight markers on the stained blot were then aligned with the prestained standards on the unstained gel and used as a template for gel slicing. The 2 X 2-mm slices were electroeluted at room temperature in 50 mM ammonium bicarbonate, 0.1% SDS for 5 h using a Bio-Rad (model 422) electroeluter at a constant current of 8 mA/ glass tube. At the end of the electroelution, the polarity of the electrodes was reversed for 1 min in order to minimize losses of protein on the dialysis membrane. The eluates were filtered through an Acrodisc 13 (Gelman) filter and bioassayed for bone-inductive activity. Fractions 26 and 27 were separated on 16 cm X 20 cm X 0.75-mm SDS gels and electroeluted using radiolabeled gel-eluted material from fraction 25 as a marker (10).
Amino Acid Sequence Analysis-The gel-eluted fractions with in uiuo biological activity were pooled, acetone-precipitated, and digested with trypsin (Worthington) in 0.1 ml of ammonium bicarbonate. Two 0.5-rg aliquots of trypsin were added during the 18-h incubation at 37 "C. Digestion was terminated by the addition of 0.1 ml of 6 M GdmC1,20 mM DTT, 50 mM Tris-HCI, pH 7.5. Tryptic peptides were separated by reverse-phase chromatography, and the sequence of each peptide was determined using an automated gas-phase sequenator (ABI, 470A).
Reduction of Osteogenin-To determine if the in vivo biological activity was sensitive to reduction, partially purified samples after S-200 filtration were dialyzed under nitrogen overnight at 4 "C against 7 M GdrnCl in 50 mM Tris-HCI, pH 8.5, with or without 10 mM DTT.

Sequence of Osteogenin
Iodoacetamide was added to a final concentration of 50 mM, and the pH was readjusted to 8.5. The samples were again purged with nitrogen and kept for 2 h in the dark at. 4 "C. Following incubation aliquots of the nonreduced control and the reduced/alkylated sample were reconstituted directly for bioassay. The bulk of the sample was run on a Superose 12 gel filtration column (HR 10/30, Pharmacia LKB), equilibrated in 6 M GdmCI, 50 mM Tris-HC1, pH 8.5, to remove DTT and iodoacetamide from the samples, and aliquots of individual fractions were tested in the bioassay.
I n Vivo Bioassay-Fractions obtained from various steps of the purification were tested for bone induction by reconstituting a portion of the sample with 25 mg of guanidine-insoluble collagenous residue of rat demineralized bone matrix (11). Typically, either 2-20 mg of the urea extract, 0.2-2.0 mg of the hydroxyapatite fraction, 20-200 pg of the heparin-Sepharose fraction, 2-20 pg of the S-200 fraction, or 0.1-1 pg of the final gel-eluted fraction were used in each reconstitution. One milligram of chondroitin 6-sulfate, sodium salt (Seikagaku Kogyo Co., Japan), and 500 pg of acid-soluble type I rat tail tendon collagen were added to each sample as carriers (11). The samples were mixed and left for 1 h a t room temperature before the proteins were precipitated with absolute ethanol overnight. The samples were then centrifuged a t 12,000 rpm (Beckman Microfuge) for 15 min. The supernatants were discarded, and the pellets were washed three times with 85% ethanol, dried, and implanted subcutaneously into male Long-Evans rats (28-35 days old) a t bilateral sites located over the ventral thorax. Each animal received two implants, and fractions were assayed in quadruplicate. The day of implantation was designated as day 0, and the implants were removed on day 10. They were cleared of adherent tissue, weighed, and homogenized in 2 ml of ice-cold 3 mM sodium bicarbonate containing 0.15 M NaCI. The homogenate was centrifuged a t 4,500 X g for 30 min. Alkaline phosphatase activity of the supernatant and calcium content of the acidsoluble fraction of the pellet were used as quantitative parameters for new bone formation (12). Implants were also examined by histology. The specific activity of osteogenin was expressed as units of alkaline phosphatase or as micrograms of calcium/mg of protein used for reconstitution in the bioassay.

RESULTS AND DISCUSSION
Purification- Table I  After hydroxyapatite and heparin-Sepharose affinity chromatography, the active fraction contained 119 mg of protein.
As is shown in Table I, there is an increase in total activity after hydroxyapatite and heparin-Sepharose chromatography, in comparison to the crude urea extract. This increase has been observed previously and suggests the possible removal of endogenous inhibitors of osteogenic activity (6). Gel filtration of the active heparin-Sepharose fraction on Sephacryl S-200 is shown in Fig. 1. Bone-inductive activity was found in a single peak. The protein profiles of the S-200 fractions were a Specific activity of osteogenin was expressed as specific activity of the alkaline phosphatase in the implant per mg of protein used for bioassay. Alkaline phosphatase activity is a reliable marker for new bone differentiation. Protein was measured by the method of Lowry et al. (16). For the gel-eluted fractions, protein content was deter-

Fraction Number
. analyzed by SDS gel electrophoresis under nonreducing conditions (Fig. 1, bottom). The apparent low recovery of total activity from the S-200 column is due to exclusion of all but the most active fractions for further purification. Final purification of osteogenin was obtained by preparative SDS gel electrophoresis of fractions 25-27 and electroelution of the active material. Activity was only found in the region from 30 to 40 kDa (Fig. 2). In other preparations activity was confined to a narrow region from 28 to 32 kDa (data not shown). The apparent low yields may be due to surface adsorption and possible inactivation of osteogenin activity during electroelution from SDS gels. In more recent experiments, electroendosmotic elution after preparative SDS gel electrophoresis (13) of the bioactive S-200 fractions yielded substantially higher recoveries of osteogenin (7-17%).
Previous work had established the utility of heparin affinity chromatography for the isolation of osteogenin activity (6). In that work, a 22-kDa protein band was found on SDS gel electrophoresis of 12,400-fold purified material. Localization of activity in the gel was not established, however. In the present experiments, employing 5-15-kg lots of starting material, we demonstrate that the major activity was confined to the 30-40-kDa region on nonreduced SDS gels (Fig. 2). The reason for the broad range of molecular mass is unknown but could be due to heterogeneous glycosylations, proteolytic or acid cleavage during isolation, or an SDS gel artifact mined by amino acid analysis.
perhaps related to poor solubility. The major advantage of the current urea extraction method over the previously described GdmCl extract is that it circumvents the time-consuming buffer exchange from guanidine to urea. It is also noteworthy that fewer contaminating bands were observed on SDS gel electrophoresis of material isolated by urea extraction.
SDS gel electrophoresis of this labeled material showed coincidence of radioactivity and protein visualized by silver staining (data not shown). SDS gel electrophoresis under nonreducing and reducing conditions of radiolabeled protein is shown in Fig. 2. Without reduction, the gel showed a broad band at 35 kDa; following reduction with 20 mM DTT, 1% pmercaptoethanol (15) the gel showed a broad band centered at 22 kDa. Incomplete reduction was observed when only 10 or 20 mM DTT was used (Fig. 2). Reduction of osteogenin with 10 mM DTT in 7 M GdmCl followed by alkylation resulted in a total loss of biological activity, while alkylation without reduction or the same reduction and alkylation in 6 M urea had no effect on the biological activity. These findings and the gel results described above demonstrate that stringent conditions are required to fully reduce osteogenin.
Amino Acid Sequencing-Several attempts to obtain amino acid sequences directly from gel-eluted material were unsuccessful (data not shown), suggesting that the amino terminus of osteogenin may be blocked. Therefore, active gel-eluted digest of gel-eluted osteogenin was chromatographed on a Synchropak RP4-4000 column (100 X 2.1 mm). The column was equilibrated in 0.08% trifluoroacetic acid, 8% acetonitrile, and eluted with a linear gradient to 0.093% trifluoroacetic acid, 40% acetonitrile in 60 min. The numbered peaks were sequenced (Table 11). Peaks 25a and 25b were sequenced together. material from Sephacryl S-200 fractions 25-27 was digested with trypsin and the tryptic peptides separated by reversephase HPLC (Fig. 3). Table I1 shows the amino acid sequences that were obtained. Four peaks gave unique sequences and three gave mixtures. The unique sequence from peaks 27 and 28 was used to resolve the mixture sequences. The same sequence was found in different peaks from a number of tryptic fragments, perhaps due to incomplete digestion of the sample. Computer-assisted searches of a protein data base (17) showed that these sequences do not match any known proteins. We do note, however, some homology of peptide 15 with inhibin-a (18) and decapentaplegic gene product (19), two members of the transforming growth factor-@ family. Transforming growth factor-@ either purified from human platelets (6)  osteogenic in this bioassay. In conclusion, osteogenin was purified more than 300,000-fold, and amino acid sequences of tryptic peptides of active highly purified protein were determined. Final proof that the protein we have characterized here is in fact the osteogenin activity will require molecular cloning and expression of sufficient recombinant material to demonstrate bone formation in vivo.