JOURNAL OF BONE AND MINERAL RESEARCH Volume 8, Number 11, 1993 Mary Ann Liebert, Inc., Publishers

Characterization of Radioiodinated Recombinant Human TGF-& Binding to Bone Matrix Within

Rabbit Skull Defects

LOUISE RICHARDSON,' THOMAS F. ZIONCHECK,' EDWARD P. AMENT0,2 LEO DEGUZMAN,2 WYNE P. LEE,' YVETTE XU,' and L. STEVEN BECK'

ABSTRACT

Bone healing is regulated in part by the local production of TGF-0' and other growth factors produced by cells at the site of injury. The single application of recombinant human TGF-8, (rhTGF-6,) to calvarial de- fects in rabbits induces an accelerated recruitment and proliferation of osteoblasts within 3 days. This ulti- mately results in the formation of new bone and the complete closure of the defect within 28 days. The per- sistence and localization of ['251]rhTGF-& within an osseous defect was investigated after applying a single dose of [1251]rhTGF-P, formulated in a 3% methylcellulose vehicle. Normal bone encompassing the defect site, the periosteum, and the gel film covering the dura were harvested at 0, 4, 8, and 24 h and 3, 7, and 16 days after [ 1251]rhTGF-/3, application. The defect site-associated radioactivity was quantitated, visualized by autoradiography, and characterized by TCA precipitation and SDS-PAGE. Radioactivity was observed in autoradiographs of gross specimens, histologic sections of the bone matrix, and periosteal tissue surround- ing the defect. There was a time-dependent decrease in TCA-precipitable radioactivity; however, radioactiv- ity was still associated with the bone matrix 16 days after application of [1251]rhTGF-P,. SDS-PAGE and autoradiography of the radioactivity in homogenized bone and periosteal samples revealed a 25 kD band, suggesting that the radioactivity remaining at the defect site represented intact [1251JrhTGF-P,. Results of this study indicate that rhTGF-Pl may bind tightly to bone matrix in its active form, and this binding may be as- sociated with the observed increase in osteoblast number and bone matrix within the calvarial defects.

INTRODUCTION

ONE REPAIR is an ordered but complex biologic process B that involves inflammation, angiogenesis, mesenchy- mal cell proliferation, extracellular matrix synthesis, and remodeling. Transforming growth factor PI (TGF-PI), a 25 kD homodimeric protein, has been isolated in relatively high concentrations from bone.") It has been reported to enhance bone formation in a number of animal models, presumably through acceleration of the normal cellular and biochemical mechanisms of repair. We previously demonstrated that a single application of recombinant hu- man (rh) TGF-/3, formulated in 3% methylcellulose induces

an osseous union within 28 days of treatment in defects of the skull that normally heal by fibrous union.(4) Time-de- pendent evaluations of the histologic effects of a single ap- plication of rhTGF-P, to osseous defects in the skull indi- cate an early recruitment and proliferation of osteoblasts at the margin of the defect site, with matrix deposition ob- served within 3

In this report we describe the distribution, histologic lo- calization, and incorporation into osseous matrix of [la51]-

rhTGF-& at various times after application to defects in the rabbit skull. These data provide useful information about the in vivo binding of exogenously administered rhTGF-& to bone and the rate at which rhTGF-P, is incor-

~~ ~

'Metabolism Group, Department of Safety Evaluation, Genentech, Inc., South San Francisco, California. 'Inflammation, Bone and Connective Tissue Research, Department of Endocrinology, Genentech, Inc., South San Francisco, Cali-

fornia.

1407

1408 RICHARDSON ET AL.

porated into osseous matrix. Characterization of TGF-0, binding to defects of the bone may allow the rational de- sign of dosing regimens and formulations for delivery and may provide insight to the concentration of TGF-0, that induces optimal osteoblast recruitment, matrix deposition, and bone remodeling.

MATERIALS AND METHODS Source and preparation of pI]rhTGF-p,

Recombinant human TGF-0, (rhTGF-0,) was purified to homogeneity from the conditioned media of transfected CHO cells.(6) The iodination was based on a procedure originally described by Frolick et al.(') Briefly, 7.5 pg rhTGF-0, in 20 mM sodium acetate buffer, pH 5, was iodinated with 2 mCi of NaIa51 (New England Nuclear) in a total volume of 60 pl. The reaction was initiated by adding 15 pl aliquots of a 50 pg/ml of chloramine-T (Sigma, St. Louis, MO) solution at 0, 2, and 3.5 minutes. The reaction was quenched at 4.5 minutes by adding 20 pl of 50 mM acetyltyrosine (Sigma) and 75 pl of 100 mM KI. To en- hance recovery, 200 p1 of 8 M ultrapure urea (Bio-Rad, Richmond, CA), pH 3.2, was added to the reaction tube and vortexed. Radioiodinated rhTGF-0, was separated from free iodine using a PD-10 column (Pharmacia, Inc., Piscataway, NJ) preequilibrated in 75 mM NaCI, 0.1% crystalline bovine serum albumin (cBSA; Calbiochem Inc., San Diego, CA), and 4 mM HCl, pH 2.3. The peak frac- tions containing [1a51]rhTGF-01 were pooled and adjusted to 1 .O% in cBSA before freezing at -70°C. The resulting [1a51]rhTGF-/3, tracer preparation was greater than 98% TCA precipitable at a concentration of 2.5 pg/ml, with a specific activity of approximately 0.2 mCi/pg.

The final formulation was prepared by mixing 1.2 ml of [L251]rhTGF-/3, in 75 mM NaCI, 1.0% cBSA, and 4 mM HCI, pH 2.3, with 3.6 ml sterile 4% methylcellulose and 20 mM sodium acetate, pH 5, using two 5 ml syringes con- nected by a fluid-dispensing connector (Burron Medical, Inc.). The resultant formulation was distributed to individ- ual 1 ml tuberculin syringes for dosing. The radioactivity present in each syringe was determined to be 127 f 4 pCi/ml (approximately 0.6 pg [1a51]rhTGF-/3,/ml) using a Capintec Radioisotope Calibrator CRC-12 (Capintec, Inc., Ramsey, NJ).

Animal surgery and treatment

All studies were done according to the American Associ- ation for the Accreditation of Laboratory Animal Care guidelines. A complete description of the surgical proce- dure, including preoperative and postoperative animal care, was previously described.") Briefly, a 12 mm skull defect was made in male New Zealand white rabbits (2.8- 3.2 kg; Elkhorn Rabbitry, Watsonville, CA) using a sterile trephine attached to an electric drill. Gel film (Upjohn, Kalamazoo, MI), a biodegradable membrane, was inserted through the defect overlying the dura to function as a bar- rier between the dura and the edges of bone. Sterile vehicle (3% methylcellulose) with [1z51]rhTGF-/3, (12.5 pCi) was

applied at a constant volume (0.1 ml, 50-100 nghabbit), filling the defect. The defect was subsequently covered by closure of the periosteal and skin flaps made at the initia- tion of surgery, as previously described.")

Tissue harvesting Groups of rabbits were euthanized at 0, 4, or 8 h or 1, 3,

7, or 16 days after treatment. Bone from the defect site (0.52 f 0.04 g), periosteum (0.44 f 0.30 g), and gel film (0.07 f 0.02 g) were harvested from the rabbits to quanti- tate the relative distribution of bound radioactivity remain- ing at and around the defect site. The defect sites with ad- jacent normal bone and periosteum were removed from the skull and evaluated as follows.

Homogenization: The samples of bone and periosteum were rinsed with physiologic saline, cut into small sections, and placed into 50 ml Falcon tubes containing 10 ml ho- mogenization buffer (1 Vo BSA and 20 mM Na acetate, pH 5.0) at 0-5°C. Each sample was washed free of any loosely associated radioactivity and tissue debris by vigorous vor- texing of the immersed bone for five 10 s intervals. The washed bone chips were then weighed and transferred to new Falcon tubes containing 4 ml fresh homogenization buffer (1% BSA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM EDTA, 10 pg/ml of leupeptin and 20 mM Na acetate, pH 5) per 0.5 g bone. Samples were homoge- nized (Tekmar Tissue Mizer Type SCT-1810) on ice using a fine probe (Type 10-0102-000). After homogenization, ali- quots of pelleted material were analyzed by trichloroacetic acid (TCA) precipitation and gel electrophoresis.

TCA Precipitation Assays: Each homogenized sample (100 pl) was transferred in triplicate to silated polyprop- ylene tubes on ice before the addition of an equal volume of chilled 20% TCA. Samples were incubated on ice for 60 minutes before centrifugation for 3 minutes at 12,000 x g. The supernatant was collected and transferred to a new tube. Radioactivity in the pellet and supernatant fractions was determined using a gamma counter. The percentage TCA-precipitable radioactivity in the pellet was deter- mined by dividing the radioactivity associated with the pel- let fraction by the total radioactivity (i.e., supernatant and pellet fractions).

Gel Electrophoresis: Bone homogenate (200 pl) was pel- leted by centrifugation. Little or no radioactivity was pres- ent in the supernatant; therefore the supernatant fraction was not processed for electrophoresis. The bone pellet was resuspended in 200 pl sodium dodecyl sulfate (SDS) sample buffer and boiled for 5 minutes. Residual particulates were pelleted by centrifugation, and 100 pl supernatant fraction was loaded per lane. Periosteum homogenate (20 pl) was mixed with 40 pl SDS sample buffer. These samples were boiled for 5 minutes and loaded directly. Samples of pro- cessed bone homogenate were subjected to electrophoresis on 10-20% gradient polyacrylamide gels (PAGE) using a Laemmli buffer system.(8) Radioactive bands were visual- ized by autoradiography using Kodak X-Omat AR film and enhancing screens at -70°C.

TGF-01 BINDING WITHIN SKULL DEFECTS 1409

Autoradiography: Additional defects with adjacent nor- mal bone were harvested for whole-sample autoradiogra- phy or histoautoradiography to examine the local distribu- tion of radioactivity at various times after dosing. Skull samples that encompassed the entire defect site (see Fig. 3A) were washed as described and the dorsoventral aspect exposed directly to Kodak X-Omat AR film for autoradi- ography or phosphor imaging of 12sI using a Phosphor- Imager (Molecular Dynamics, Sunnyvale, CA). ( 9 ) Sagittal sections from the skull samples were then sliced and placed on the film such that the cortical surface of each section was exposed to film or phosphor imaging as described. Fragments of bone isolated from the defect with adjacent normal bone were decalcified using an ion-exchange resin. Sections (4 pm) were mounted on histologic slides and dipped in Ilford KSD emulsion. After exposure for 2 weeks the slides were developed in Kodak D19 for 1.5 minutes and then fixed in Kodak GBX and stained with hernatoxy- lin and eosin.

RESULTS Distribution of radioactivity after application of r2sIJrhTGF-@l to a skull defect

Total recovery of radioactivity from bone, periosteum, and gel film was 68.5% at the first sampling time, 25-29070 over the first 24 h, 5-1 1070 at 3 and 7 days, and less than 1 % at 16 days (Table 1) . Initially, the radioactivity was predominantly associated with the gel film that covered the dural surface. The highest percentage of radioactivity at subsequent sampling times was associated with the perios- teum. Approximately 20% of the dose was associated with the periosteal membrane at 4, 8, and 24 h after applica- tion. The periosteally associated radioactivity diminished to less than 8% in both the 3 and 7 day samples and to less than 1 % after 16 days. Less than 5% of the radioactivity

was associated with either the bone or underlying gel film covering the dura 4 h after dosing (Fig. 1) . The relatively small percentage of radioactivity associated with the bone persisted, however, and was detectable by autoradiography 16 days after dosing (data not shown).

Characterization of the radioactivity within the skull defect site

Greater than 90% of the radioactivity was TCA pre- cipitable in the bone and periosteal tissue homogenates, suggesting that the radioactivity represented radiolabeled rhTGF-/3, rather than free or low-molecular-weight metabolites. The majority of the extracted radioactivity migrated at 25 kD after periosteum and bone samples was subjected to nonreducing SDS-PAGE. This provided fur- ther evidence that the observed radioactivity represented intact [12SI]rhTGF-@, (Fig. 2). A minimal amount of radioactivity migrated lower than the 14 kD molecular weight marker and likely represented rhTGF-& monomer (12.5 kD). In addition, a radiolabeled band migrating at the interface of the stacking gel and the resolving gel was also detected. This band is presumed to be due to the cova- lent association of [12SI]rhTGF-@l with a high-molecular- weight matrix protein or to aggregation of [12SI]rhTGF-/31 during sample handling.

Autoradiography of rabbit skull defects Skull samples were harvested at selected time points as

described in the schematic (Fig. 3A). The dorsoventral autoradiograms of the excised skull sample indicated the presence of [12SI]rhTGF-/31 at all time points of investiga- tion, with the radioactivity located circumferentially around each defect. The intensity of the signal varied around a given defect for each time point but was greatest on the day of [1asI]rhTGF-/31 administration (0-24 h post-

TABLE 1 . QUANTITATION OF RADIOACTIVITY IN BONE, PERIOSTEUM, AND GEL FILM SAMPLES (n = 2) AT VARIOUS TIMES AFTER APPLYING [12SI]rhTGF-& TO A RABBIT SKULL DEFECT

Time of sample harvest ~

Oh 4 h 8 h 24 h 3 Days 7 Days 16 Days

Bone 9'0 Dose 1 9.5 '70 Dose 2 1.8

Mean 5.7

'70 Dose 1 10.1 070 Dose 2 21.9

Mean 16.0

07'0 Dose 1 44.1 '7'0 Dose 2 49.5

Mean 46.8 Mean total recovery, Qo 68.5

Periosteum

Gel film

3.4 3.4 3.2 5.6 6.9 4.2 4.5 5.2 3.7

15.7 22.4 14.4 23.5 18.0 23.3 19.6 20.2 18.9

3.8 3.6 2.9 4.3 3.1 1.8 4.1 3.4 2.4

28.2 28.8 25.0

2.1 2.2 2.2

6.9 7.5 7.2

1.7 1.8 1.8

11.2

1.8 < 1 NDb

11.0 2.5 2.2

2.7 1 .o 1.9

<5.9

<la < 1 ND

<1 < 1 ND

< 1 < I ND ND

aRadioactivity in sample was too low to quantitate accurately. bNot determined.

1410 RICHARDSON ET AL.

M ~n F - 1 I I d a u 1 I I

0 hr 4 hr 8 hr 24 hr 3 days 7 day$

FIG. 1. Quantitation of [1*SI]rhTGF-/3, bound to bone, periosteum, and gel film. Associated radioactivity was measured at each time point and the amount of [l’sI]- rhTGF-0, determined. Data are presented as Vo of the total dose of [1a51JrhTGF-/3, applied and are representative of two independent experiments.

kRa 200 -

Characterization of Periosteum- Associated Radioactivity

Hours Days 0 4 8 1 3 7 -.)-.--

97 - 66 -

45 -

21 -

dose) and then gradually decreased with time thereafter. Binding of the radioligand appeared to be distributed as much as 1-2 mm away from the edge of the defect site along the periosteal surface and the surface of the skull ad- jacent to the dura (Fig. 3B). Autoradiograms of sagittal sections from the defect sites were done to evaluate further binding of [1‘SI]rhTGF-j3, to bone. Binding of the radio- ligand in the sagittal sections was located along the cortical (cut) surface as well as the periosteal and dural surfaces (Fig. 3B). However, with this method of analysis it was not possible to distinguish the relative intensity of binding in the cortical surface compared to the periosteal or dural surface of the bone samples.

II, 31 -

14 -

Histologic localization of ra51]rh TGF-0, within skull defects

Subsequent evaluation of histoautoradiograms con- firmed the location and distribution of [1’SI]rhTGF-/3, binding to bone along the periosteal, dural, and cortical (cut) surface of the defect sites (Fig. 4). Binding to bone was observed in defect sites exposed to [1*SI]rhTGF-/3, for

200-

97 - 66-

45-

31 -

Characterization of Bone- Associated Radioactivity

Hours Days 0 4 8 1 3 7

14-

FIG. 2. Characterization of associated radioactivity by SDS-PAGE-autoradiography from homogenized samples of periosteum and bone. Samples were processed as described in Materials and Methods using a 10-20% SDS-polyacryl- amide gel. The predominant radioactive band at 25 kD comigrates with intact [1’SI)rhTGF-/3,, and the faint bands at < 14 kD are presumed to be TGF-P, monomer (12.5 kD).

TGF-@, BINDING WITHIN SKULL DEFECTS 1411

less than 5 minutes (data not shown). The intensity of binding to the periosteal and cortical surfaces of the bone appeared to be similar 4 h after in vivo exposure to [ T I rhTGF-P, at the defect sites, with little or no binding to the dural surface (Fig. 4A).

Binding occurred in the defect sites, including the corti- cal, periosteal, and dural surfaces, as well as the superficial osteocytic lacunae along the periosteal surface, 8 h after in vivo exposure to [1151]rhTGF-/3,. In addition, at 8 h the periosteal membrane was clearly demarcated by an intense lining of [1*51]rhTGF-@1, and the subcutaneous tissue over- lying the periosteum appeared swollen, containing [1251] rhTGF-@, in the extracellular matrix (Fig. 4B). Defect sites exposed to [1251]rhTGF-P, for 24 h exhibited binding char- acteristics similar to those of sites exposed for 8 h, with the exception that radioligand was present in osteocytic la- cunae and marrow space at sites remote from the perios- teal, dural, or cut surfaces of the defect. The intensity of binding of [1251]rhTGF-@, to bone at 24 h was similar to that at 8 h (Fig. 4C).

Binding of [1251]rhTGF-@, continued to be present along the periosteal, dural, and cut surfaces of the bone at 3 days. The defects began to exhibit signs of inflammation within the defect space, including swelling and an apparent increase in the periosteal extracellular matrix. The relative intensity of the signal of [1251]rhTGF-P, at the defect site after 3 days was less than that at earlier time points (data not shown). New bone was observed in defect sites 7 days after treatment with [1151]rhTGF-& (Fig. 4D). The amount of new bone observed was consistent with the results from previous dose-response studies. (') A line of radioactivity clearly demarcated the border between the original site of the defect and new bone, with the radioactivity incorpo-

Remove Dslect and Autoradiograph " / r 1

rated into the newly deposited matrix (but not extending beyond the initial site of binding). The radioactivity ap- peared to be bound to matrix rather than to cells within the matrix. Bone matrix-associated [1251]rhTGF-/3, was still present by 16 days along the initial edge of the defect, with an increased amount of bone bridging the defect (data not shown). The amount of the radioactivity by 16 days was greatly diminished compared to earlier time points.

DISCUSSION

Transforming growth factor PI has been isolated from most cell types and has been proposed as a key modulator of extracellular matrix homeostasis. ( lo) Extracellular TGF- PI has been localized in both endochondral and intramem- branous bone during embryogenesis(ll,ll) and has been iso- lated from neonatal(13) as well as adult bone.''.") In vitro, TGF-@, is secreted in a latent form from o ~ t e o b l a s t s ( ~ ~ . ~ ~ ) and, upon activation, stimulates components important to bone repair, such as osteoblast he mot ax is,(^^^^^) prolifera- tion, and extracellular matrix secretion.(19,'0) In addition to inducing a bone repair phenotype, extracellular matrix- associated TGF-@, has been considered important in nor- mal homeostasis of bone and may be the coupling factor for the bone resorption and formation processes in remod-

This study demonstrated the sustained presence in vivo of exogenously applied rhTGF-0, to sites of bone damage. [1151]rhTGF-@, applied to skull defects at the time of injury was detectable at the defect site for the duration of investi- gation (16 days). Gradual incorporation of [1251]rhTGF-/3, into the matrix occurred at the original margin of the de-

eling. (11.11)

B 24 Hours

t--, I

FIG. 3. (B) Phosphor image of bone associated radioactivity surrounding the skull defect 24 h after administration of [1z51]rhTGF-@,. The dark circle represents the dorsoventral view, whereas the bands and dots adjacent to the arrows represent radioactivity from sagittal sections of bone. Sagittal sections from the gross specimen are presented next to the respective phosphor image.

(A) Processing for autoradiography.

1412 RICHARDSON ET AL.

FIG. 4. (A) The 4 h exposure of bone to [1251]rhTGF-&. Arrows indicate the cut and periosteal surface coated with radioactivity. (B) The 8 h exposure demonstrates binding of bone to [1251]rhTGF-P, to both the periosteum (closed arrows) and the periosteal, cut, and dural surfaces of bone (arrowheads). [1’SI]rhTGF-/3, is also located in the superficial lacunae (open arrows). (C) The 24 h exposure of the bone defect to [1’51]rhTGF-P,. Closed arrows indicate superficial binding of [1’51]rhTGF-@,; open arrows indicate binding of [1*51]rhTGF-/3, within marrow cavities. (D) The 7 day exposure of bone demonstrates [1251]rhTGF-/3, binding to bone at the original site of the defect (closed arrows) with new bone (open arrows) formation at the advancing edge. The asterisk indicates the gel film.

Histoautoradiograms of selected sagittal sections from skull defects administered (1’51]rhTGF-/3,.

fect. At the early time points, binding was primarily a sur- face phenomenon in which [12SI]rhTGF-P, was observed on the cut surface and the periosteal surface of the bone where the periosteum was elevated free from the bone. At later time points, the surface binding persisted and was ac- companied by the appearance of [1251]rhTGF-&, first within superficial osteocytic lacunae and finally at rela- tively distant sites, such as deeper osteocytic lacunae and small marrow cavities separated from either the cut or peri- osteal surface of the skull. The mechanism by which rhTGF-P, was transported to the osteocytic lacunae is not known but may be related to diffusion through superficial osteocytic canaliculi connecting with deeper lacunae or marrow cavities.

The sustained presence of TGF-P, at the defect site may be related to binding to extracellular matrix proteins, such as P-glycan (type I11 binding site) or decorin. These poly- morphic proteoglycans bind with high affinity to TGF-

and may modulate the activity of the growth fac- tor. Both membrane-bound and soluble forms of 0-glycan have been isolated with the soluble form thought to be im-

portant in local retention, delivery, or clearance of the active form of TGF-0,. This study does not provide direct evidence for the specific binding of [1251]rhTGF-& to re- ceptors or extracellular matrix proteins. However, the presence of osteoblasts and deposition of extracellular ma- trix observed at days 7 and 16 provide indirect evidence for specific binding at the defect site. In addition, the stimula- tion of bone formation observed with [1251]rhTGF-PI was comparable to results from previous studies.(s) The grad- ual decrease in the amount of [1251]rhTGF-/31 present at the site, especially noted after day 3, may be related to clear- ance by soluble binding proteins. Alternatively, the de- crease in [1251]rhTGF-/3, may be related to the high meta- bolic activity, the presence of proteolytic enzymes, and rapid cellular turnover associated with inflammatory pro- cesses within damaged tissue. Concurrent to the gradual clearance of [1251]rhTGF-P,, osteoblast proliferation and matrix synthesis occurred with the incorporation of the re- maining [1’51]rhTGF-/3, into the newly deposited matrix. Matrix incorporation of [1251]rhTGF-& was isolated to the original site of the defect and was not observed in other

TGF-@I BINDING WITHIN SKULL DEFECTS 1413

areas, such as the periosteal surface, osteocytic lacunae, or marrow cavities. Incorporation of [lzsI]rhTGF-@, into the matrix was initially observed at 7 days, with a gradual de- crease of [lzsI]rhTGF-@, within the matrix at subsequent time points. The decrease in the amount of [lzSI]rhTGF-& present at 16 days compared to 7 days may be related to re- modeling from woven to lamellar bone that occurs follow- ing bone healing.

Binding of TGF-@, to the periosteum may be important for healing of bone. The periosteum has been proposed as a source of preosteoblasts or progenitor cells of mesency- ma1 origin that can be induced by irritation or other extrin- sic factors to form osteoblasts. ( zs .16 ) Diffusion chambers containing enzymatically isolated periosteal cells induce bone when implanted intraperitoneally into athymic nude mice.(278) Other in vivo studies have demonstrated the in- duction of bone by exogenously administered TGF-@, in- jected under the periosteal membrane of the long bones of

or the skulls of mice.(z.z8) In our studies, approxi- mately 20% of the [lzSI]rhTGF-@, applied was associated with the periosteum in the 4, 8, and 24 h samples (Fig. 1). The presence of TGF-0, bound to the periosteum may be important to induction of bone formation adjacent to pre- existent bone.") As the bone forms from the margin of the defect, the TGF-@, bound to the periosteum may function as a reservoir that stimulates phenotypic transformation of osteoblast precursors to osteoblasts, as well as the chemo- taxis and proliferation of osteoblasts across the newly formed matrix. In addition, the bone formed at the edge of the defect may be a result of upregulation of endoge- nous TGF-0, induced by the presence of the exogenously applied growth factor.(29.30)

These studies demonstrate that exogenously applied rhTGF-@, can bind to bone and periosteum, persists intact at the defect site, and can penetrate to deep osteocytic la- cunae. Additional studies concerning the nature of binding of TGF-P, in vivo, as well as examination of the early ef- fects of this binding, are currently being evaluated.

ACKNOWLEDGMENTS

We thank Stan Hansen for his expert technical assis- tance in histoautoradiography and Tue Nguyen and Mary Cromwell for preparation of rhTGF-@, formulations.

REFERENCES

1. Seyedin SM, Thompson AY, Benta H, Rosen DM, McPher- son JM, Conti A, Siege1 NR, Galluppi GR, Piez KA 1986 Cartilage-inducing factor-A. Apparent identity to transform- ing growth factor 8. J Biol Chem 2615693-5695.

2. Noda M, Carnilliere JJ 1989 In vivo stimulation of bone for- mation by transforming growth factor-@. Endocrinology 124:

3. Joyce ME, Jinguski S, Roberts AB, Sporn MB, Bolander ME 1989 Transforming growth factor-@ initiates cartilage and bone formation in vivo. J Bone Miner Res 4:255-259.

4. Beck LS, DeGuzman L, Lee WP, Xu Y, McFatridge LA, Gillett NA, Amento EP 1991 TGF-beta-1 induces bone clo- sure of skull defects. J Bone Miner Res 6:1257-1265.

2991-2994.

5 . Beck LS, Amento EP, Xu Y, DeGuzman L, Lee WP, Nguyen T, Gillett NA 1993 TGF-@, induces bone closure of skull defects: The temporal dynamics of bone formation in defects exposed to rhTGF-0,. J Bone Miner Res 8:753-761.

6. Derynck R, Jarrett JA, Chen EY, Eaton DH, Bell JR, Assoian RK, Roberts AB, Sporn MB, Goeddel DV 1985 Hu- man transforming growth factor-@ complementary DNA se- quence and expression in normal and transformed cells. Nature 316:701-705.

7. Frolik CA, Dart LL, Meyers CA, Smith DM, Sporn MB 1983 Purification and initial characterization of a type beta transforming growth factor from human placenta. Proc Natl Acad Sci USA 80:3676-3680.

8. Laemmli UK 1970 Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680- 685.

9. Johnston RF, Pickett SC, Barker DL 1990 Autoradiography using storage phosphor technology. Electrophoresis 11:355- 360.

10. Sporn MR, Roberts AB 1990 The multifunctional nature of peptide growth factors. In: Sporn MB, Roberts AB (eds.) Handbook of Experimental Pharmacology: Peptide Growth Factors and Their Receptors I. Springer-Verlag. New York, pp. 3-15.

1. Heine UI, Flanders K, Roberts AB, Munoz EF, Sporn MB 1987 Role of transforming growth factor-@ in the develop- ment of the mouse embryo. J Cell Biol 105:2861-2876.

2 . Carrington JL, Roberts AB, Flanders KC, Roche NS, Reddi HA 1988 Accumulation, localization, and compartmentation of transforming growth factor @ during endochondral bone development. J Cell Biol 107:1%9-1975.

3. Centrella M, Canalis E 1985 Transforming and nontrans- forming growth factors are present in medium conditioned by fetal rat calvariae. Proc Natl Acad Sci USA 82:7335- 7339.

14. Seyedin SM, Thomas TC, Thompson AY, Rosen DM, Piez KA 1985 Purification and characterization of two cartilage- inducing factors from bovine demineralized bone. Proc Natl Acad Sci USA 82:2267-2271,

15. Robey PG, Young MF, Flanders KC, Roche NS, Kondaiah P, Reddi AH, Termine JD, Sporn MB, Roberts AB 1987 Osteoblasts synthesize and respond to TGF-beta in vitro. J Cell Biol 105:457-463.

16. Jennings JC, Mohan S 1990 Heterogeneity of latent trans- forming growth factor-@ isolated from bone matrix proteins. Endocrinology 126:1014-1021.

17. Lucas PA 1989 Chemotactic response of osteoblast-like cells to transforming growth factor beta. Bone 10:459-463.

18. Pfeilschifter J , Wolf 0, Naumann A, Mime HW, Mundy GR, Ziegler R 1990 Chemotactic response of osteoblastlike cells to transforming growth factor @. J Bone Miner Res 5:

19. Centrella M, McCarthy TL, Canalis E 1987 Transforming growth factor @ is a bifunctional regulator of replication and collagen synthesis in osteoblast-enriched cell cultures from fetal rat bones. J Biol Chem 262:2869-2874.

20. Bortell R, Barone LM, Tassinari MS, Lian JB, Stein GS 1990 Gene expression during endochondral bone development: Evidence for coordinate expression of transforming growth factor 6, and collagen type. J Cell Biochem 44:81 91.

21. Pfeilschifter J , Mundy GR 1987 Modulation of type @ trans- forming growth factor activity in bone cultures by osteo- tropic hormones. Proc Natl Acad Sci USA 84:2024-2028.

22. Pfeilschifdter J, Bonewald L, Mundy GR 1990 Characteriza- tion of the latent transforming growth factor p complex in bone. J Bone Miner Res 5:49-58.

825-830.

1414 RICHARDSON ET AL.

23.

24.

25.

26.

27.

28.

29.

Andres JL, Stanley K, Cheifetz S , Massague J 1989 Mem- brane-anchored and soluble forms of betaglycan, a polymor- phic proteoglycan that binds transforming growth factor- beta. J Cell Biol 109:3137-3145. Yamaguchi Y, Manan DM, Ruoslahti E 1990 Negative regu- lation of transforming growth factor-8 by the proteoglycan decorin. Nature 346:281-284. Wlodarski KH 1989 Normal and heterotopic periosteum. CIin Orthop 241:265-277. Caplan A1 1991 Mesenchymal stem cells. J Orthop Res 9:

Nakahara H, Bruder SP, Haynesworth SE, Holecek JJ Baber MA, Goldberg VM, Caplan A1 1990 Bone and carti- lage formation in diffusion chambers by subcultured cells de- rived from the periosteum. Bone 11:181-188. Marcelli C, Yates AJ, Mundy GR 1990 In vivo effects of hu- man recombinant transforming growth factor 0 on bone turnover in normal mice. J Bone Miner Res 91087-1096. Varga J, Rosenbloom J, Jimenez SA 1987 Transforming

641-650.

growth factor 0 (TGFB) causes a persistent increase in steady-state amounts of type I and type I11 collagen and fi- bronectin mRNAs in normal human dermal fibroblasts. Bio- chem J 247597-604.

30. Ishikawa 0, Yamakage A, LeRoy EC, Trojanowska M 1990 Persistent effect of TGF-8, on extracellular matrix gene ex- pression in human dermal fibroblasts. Biochem Biophys Res Commun 169:232-238.

Address reprint requests to: L. Steven Beck

Department of Endocrinology Genentech, Inc.

460 Point San Bruno Boulevard South San Francisco, CA 94080

Received in original form January 8, 1993; in revised form March 17, 1993; accepted April 8, 1993.