Construction of Recombinant His6-Tagged Bovine spp24 Isoforms
N-terminal Met(His)6-tagged full-length (amino acid residues 24 to 203) and truncated secretory spp24 (amino acid residues 24 to 176) were made with use of recombinant DNA technology, as described previously8. Briefly, the proteins were expressed in the pET-20b expression vector (EMD Chemicals, Gibbstown, New Jersey) with use of Escherichia coli cell strain BL21(DE3). Single colonies were inoculated with use of Overnight Express Instant TB Medium plus 1% glycerol (Invitrogen, Carlsbad, California) and incubated overnight at 37°C in an orbital shaker. Inclusion bodies containing the protein of interest were isolated with use of BugBuster extraction reagents according to the manufacturer's instructions (EMD Chemicals), with the addition of His6 protease inhibitors (Sigma-Aldrich, St. Louis, Missouri). The His6-tagged secretory isoforms were isolated from inclusion bodies with use of HiTrap immobilized metal ion affinity chromatography (IMAC) FF columns (Amersham Biosciences, Uppsala, Sweden) and a BioLogic protein purification workstation (Bio-Rad Laboratories, Hercules, California).
Implant Preparation
Sterile absorbable collagen sponges (Helistat, Integra Life Sciences, Plainsboro, New Jersey) were cut into 10-mm × 5-mm × 5-mm pieces with use of sterile scalpels and scissors. Sponges with either truncated spp18 (TR-spp18) or full-length spp24 (FL-spp24) were prepared in the following manner. The recombinant protein (0.1 mg to 2.5 mg) was placed in 50 µL or 100 µL of 60% acetonitrile in high-performance liquid chromatography-grade (HPLC-grade) water and 0.1% trifluoroacetic acid (Buffer B)8. The 0.1-mg and 0.5-mg doses were dissolved in 50 µL of Buffer B, and the 1-mg and 2.5-mg doses were dissolved in 100 µL of Buffer B. The protein solution or Buffer B alone was added drop-wise to the sponges and allowed to air dry under a sterile hood. The sponges were lyophilized overnight and sterilized with chloroform vapor for four hours in a desiccator. Separately, dilutions of rhBMP-2 (INFUSE, Medtronic, Minneapolis, Minnesota) were prepared in 50-µL aliquots of sterile saline solution and stored at —20°C until use. Before implantation, the collagen sponges were soaked in rhBMP-2 solutions while control sponges were soaked in 50-µL of sterile saline solution.
Rat Spinal Fusion
Animal protocols were approved by the University of California Los Angeles (UCLA) Chancellor's Animal Research Committee prior to experimentation. Male Lewis rats were anesthetized with isoflurane inhalation and monitored continuously for the duration of the procedure. Using a previously described surgical procedure9, a posterior midline incision was made over the lumbar spine. Two subsequent fascial paramedian incisions, 3 mm to 4 mm from the midline, exposed the transverse processes. A high-speed burr was used to decorticate the transverse processes of L4 and L5. The surgical site was irrigated with sterile saline solution, and implants were placed between the transverse processes bilaterally in the paraspinal muscle bed. Both the fascia and skin incisions were closed with 4-0 absorbable sutures. Following surgery, rodents were housed in separate cages, allowed to eat and drink ad libitum, and monitored daily throughout the course of the study.
Study Groups
A table in the Appendix outlines the experimental groups used in this study. In total, 145 male Lewis rats, aged six to eight weeks, were used. One rat in a control group (0.5-mg TR-spp18 alone) died during surgery. All other animals tolerated the entire study well.
Radiographic Analysis
Radiographs were obtained at four, six, and eight weeks after surgery. Animals were anesthetized with isoflurane inhalation at the four and six-week time points, while eight-week radiographs were made after the animal was killed and the spine was harvested. Bone area was measured using ImageJ10 analysis software. Prior to analysis, radiographic measurements were calibrated. The region of interest between L4 and L5 for each radiograph was circled and the area determined on the basis of the initial measurement calibration for that group of animals. An area of approximately 0.5 cm2 was the area measured for animals with no new bone formation, while areas of as much as 4 cm2 were calculated for animals with large bone-mass formation.
Manual Assessment and Gross Evaluation of Bone Formation
Eight weeks after surgery, the animals were killed and the spines were explanted for manual assessment and gross evaluation of bone formation. Spines were manually tested for bone formation and intersegmental motion and graded on a seven-point scale: (1) no bone mass; (2) stiff, but no bone mass; (3) unilateral bone mass, movement at the level of the intervertebral disc; (4) bilateral bone mass at one level, movement at the intervertebral disc; (5) bilateral bone mass at two levels, movement at the intervertebral disc; (6) spinal fusion at one level; and (7) spinal fusion at two levels. Each specimen was blindly graded by three experienced observers (C.S., Y.M., and M.M.). The mean for the three values was used in subsequent statistical analyses.
Microquantitative Computed Tomography
Following gross evaluation, specimens were fixed in 10% formalin for at least one week. Representative samples were subjected to high-resolution microquantitative computed tomography (µCT40; Scanco USA, Southeastern, Pennsylvania) as previously described11. High-resolution images, from 9 µm to 20 µm, were acquired, and three-dimensional reconstruction allowed for visualization in all directions. Sagittal cross-sections at the level of the transverse processes were used to analyze continuous bone formation.
Histology
Once the spines were fixed in 10% formalin, specimens were then decalcified for one week with use of 10% decalcifying hydrogen chloride solution (Cal-Ex, Fisher Scientific, Fair Lawn, New Jersey); the HCl was replaced with new solution at days one, three, five, and seven. Following decalcification, specimens were stored in 70% ethanol solution until use. Spines were cut along the midline and embedded in paraffin, and serial sagittal sections were made at the level of the transverse processes. Sections were mounted on glass slides and stained with hematoxylin and eosin. A Nikon Super CoolScan 5000 ED Film Scanner (Nikon, Melville, New York) was used to capture images of entire sections.
Statistical Analysis
GraphPad Prism (GraphPad Software, La Jolla, California) was used for statistical analysis. One-way and two-way analyses of variance were used where appropriate, along with Tukey's multiple comparison test for post hoc analysis. Sample size for each group was selected a priori on the basis of previous studies that showed significant differences7.
Source of Funding
This study was partially funded by the Department of Veterans Affairs.
Bone Area Determined by Radiographic Analysis
A summary of the radiographic analysis of bone formation resulting from varying amounts of rhBMP-2 over time is shown in Figure 1. Control animals implanted with an absorbable collagen sponge alone had bone areas of approximately 0.5 cm2 at all time points, indicating no new bone formation. At postoperative weeks six and eight, animals implanted with 10 µg of rhBMP-2 had a significantly larger bone area on radiographs as compared with animals implanted with lower rhBMP-2 doses or an absorbable collagen sponge alone. No significant differences were observed between implantation doses four weeks postoperatively. Two-way analysis of variance revealed a main effect of dose (F(4) = 43.55, p < 0.001) and time (F(2) = 12.26, p < 0.001), and a main effect of interaction between dose and time (F(8) = 8.498, p < 0.001). Radiographs of the 10-µg rhBMP-2 groups revealed significant bone formation and multiple-level fusions.
Bar graph showing the radiographic evidence of bone formation at the L4-L5 fusion site over time and according to the various rhBMP-2 doses given. An asterisk (*) indicates a p value of <0.001 for post hoc pairwise comparisons. ACS = absorbable collagen sponge. Error bars indicate standard error of the mean.
Radiographic analyses for animals implanted with both rhBMP-2 and either FL-spp24 or TR-spp18 are shown in Figure 2. Animals that received 1 µg of rhBMP-2 with incremental doses of either TR-spp18 or FL-spp24 showed no significant differences on two-way analysis of variance (Fig. 2, a). All groups exhibited bone areas similar to those observed in negative control groups (approximately 0.5 cm2; Fig. 1). Figure 2, b displays results from animals that received 10 µg of rhBMP-2 with incremental doses of TR-spp18 or FL-spp24. Higher doses of TR-spp18 (0.5 mg and 2.5 mg) in combination with 10 µg of rhBMP-2 resulted in a significant decrease in bone-area formation as compared with the results obtained with the lower dose (0.1 mg) of TR-spp18 combined with 10 µg of rhBMP-2. Two-way analysis of variance revealed a main effect from the treatment (F(1) = 31.03, p < 0.001) only. Additionally, it is noted that, compared with the positive control (10 µg of rhBMP-2 alone in Fig. 1), the low-dose TR-spp18 resulted in a mild inhibitory effect on rhBMP-2 bone-formation activity. rhBMP-2-mediated bone-area formation was inhibited in animals implanted with all doses of FL-spp24 and 10 µg of rhBMP-2, with no significant difference between doses.
Bar graph showing the radiographic evidence of bone formation in samples treated with both rhBMP-2 and the recombinant isoform of spp24. a: Bone area for animals implanted with 1 µg of rhBMP-2 plus varying doses of truncated spp18 (TR-spp18) (left) or full-length spp24 (FL-spp24) (right). b: Bone area for animals implanted with 10 µg of rhBMP-2 plus varying doses of TR-spp18 (left) or FL-spp24 (right). An asterisk (*) indicates a p value of <0.05 and a number sign (#) indicates a p value of <0.01, as determined by two-way analysis of variance. Low dose = 0.1 mg, medium dose = 0.5 mg, and high dose = 2.5 mg. Error bars indicate standard error of the mean.
Manual Assessment and Gross Evaluation of Bone Formation
Manual assessment and gross evaluation of bone formation, conducted immediately after the spines were harvested, showed a dose-dependent response to increasing concentrations of rhBMP-2 (Fig. 3, a). The manual assessment scores confirmed results from radiographic analysis in which a dose-dependent response was evident in rhBMP-2-mediated spine fusion. In particular, 10 µg of rhBMP-2 resulted in a significantly higher manual palpation score as compared with the negative control (absorbable collagen sponge alone) and the lower rhBMP-2 doses. Significant effects were also noted when rhBMP-2 was combined with the two different spp24 isoforms (Fig. 3, b). The 10-µg dose of rhBMP-2, which normally leads to multiple-level fusion in the rodent model, combined with a small amount of FL-spp24, resulted in little to no bone formation. These manual assessment and gross evaluation data were consistent with the results of radiographic evaluation.
a: Dose response of rhBMP-2-mediated bone formation, as determined by manual assessment of bone formation. Increasing the rhBMP-2 dose resulted in higher manual assessment scores; a dose of 10 µg of rhBMP2 was associated with a significantly higher score than were the lower doses or no dose (control). b: All doses of truncated spp18 (TR-spp18) and full-length spp24 (FL-spp24) were grouped to show the significant inhibitory effect on rhBMP-2-induced bone formation in this analysis. FL-spp24 significantly inhibits rhBMP-2 mediated spine fusion, particularly when combined with a high dose of BMP-2. The manual assessment scores of samples treated with 10 µg of rhBMP-2 plus FL-spp24 were consistently lower than those of samples treated with rhBMP-2 alone or rhBMP-2 in combination with TR-spp18. An asterisk (*) indicates a p value of <0.001 and a number sign (#) indicates a p value of <0.01, as determined by one-way analysis of variance. Error bars indicate standard error of the mean.
Microquantitative Computed Tomography
Figure 4 shows representative anterior and lateral microquantitative computed tomography images for samples that received either 1 µg of rhBMP-2 (Fig. 4, a through d) or 10 µg of rhBMP-2 (Fig. 4, e through h). Animals that received 1 µg of rhBMP-2 exhibited little bone formation at four, six, and eight weeks, whereas animals that received 10 µg of rhBMP-2 had robust bone formation at all time points. Sagittal sections at the levels of the transverse processes are shown in Figure 4, d for animals that received 1 µg of rhBMP-2 and in Figure 4, h for animals that received 10 µg of rhBMP-2. Although the specimen in Figure 4, c appears to have some bone formation at eight weeks, the cross-sectional image shows gaps between the transverse processes, indicating incomplete union. Conversely, animals implanted with 10 µg of rhBMP-2 exhibited large bone masses, with no gaps between the transverse processes (Fig. 4, h).
Representative microquantitative computed tomography images of spines treated with rhBMP-2 and harvested at varying time points. Panels a, b, and c show an anterior view of a specimen implanted with 1 µg of rhBMP-2 for four weeks (a), six weeks (b), and eight weeks (c). Panel d shows a lateral view and sagittal cross section through the transverse process of the specimen shown in panel c. Panels e, f, and g show an anterior view of a specimen that was implanted with 10 µg of rhBMP-2 for four weeks (e), six weeks (f), and eight weeks (g). Panel h shows a lateral view and sagittal cross-section through the transverse processes of the specimen shown in panel g.
Figures 5 and 6 show representative microquantitative computed tomography images of specimens from animals that received both rhBMP-2 and a recombinant spp24 isoform. Results show that TR-spp18 (Fig. 5) mildly to moderately inhibited rhBMP-2-mediated bone formation, whereas FL-spp24 (Fig. 6) completely prohibited rhBMP-2-mediated bone formation. More specifically, when rhBMP-2 was implanted in combination with TR-spp18 (Fig. 5, e, f, k, l, q, and r) or FL-spp24 (Fig. 6, e, f, k, l, q, and r), only single-level fusion or no fusion was observed. This is in contrast to animals treated with only 10 µg of rhBMP-2, which consistently exhibited multiple level fusions, even as early as four weeks (Fig. 4, e). In addition, the fusion rates for groups that were implanted with 1 µg of rhBMP-2 were not significantly affected by TR-spp18, regardless of the dose. Typically, 1 µg of rhBMP-2 is associated with a fusion rate of approximately 40%7, a rate considered to be unsuccessful in the clinical setting. Groups implanted with TR-spp18 in combination with 10 µg of rhBMP-2 displayed single-level fusions (Fig. 5, e and f, k and l, and q and r) as opposed to the two-level fusions seen in animals implanted with 10 µg of rhBMP-2 alone (Fig. 4, g and h).
Representative anterior and lateral microquantitative computed tomography images of specimens treated with truncated spp18 (TR-spp18) with or without rhBMP-2. Specimens a through f were treated with 0.1 mg of TR-spp18. Specimens g through l were treated with 0.5 mg of TR-spp18. Specimens m through r were treated with 2.5 mg of TR-spp18. Specimens a, g, and m received 0.1, 0.5, and 2.5 mg of TR-spp18 alone, respectively. Specimens b, h, and n display the respective sagittal cross sections through the transverse processes of a, g, and m. Specimens c, i, and o received 1 µg of rhBMP-2 in addition to 0.1, 0.5, and 2.5 mg of TR-spp18, respectively. Specimens d, j, and p are the sagittal cross sections of c, i, and o, respectively. Specimens e, k, and q received 10 µg of rhBMP-2 in addition to 0.1, 0.5, and 2.5 mg of TR-spp18, respectively. Specimens f, l, and r are the sagittal cross sections of e, k, and q, respectively.
Representative anterior and lateral microquantitative computed tomography images of specimens treated with full-length spp24 (FL-spp24) with or without rhBMP-2. Specimens a through f were treated with 0.1 mg of FL-spp24. Specimens g through l were treated with 0.5 mg of FL-spp24. Specimen m was treated with 2.5 mg of FL-spp24. Specimens n through r were treated with 1 mg of FL-spp24. Specimens a, g, and m received 0.1, 0.5, and 2.5 mg of FL-spp24 alone, respectively. Figures b, h, and n are the respective sagittal cross sections through the transverse processes of a, g, and m. Specimens c, i, and o received 1 µg of rhBMP-2 in addition to 0.1, 0.5, and 1 mg of FL-spp24, respectively. Specimens d, j, and p are the sagittal cross sections of c, i, and o, respectively. Specimens e, k, and q received 10 µg of rhBMP-2 in addition to 0.1, 0.5, and 1 mg of FL-spp24, respectively. Figures f, l, and r are the sagittal cross sections of e, k, and q, respectively.
Histology
The results of histological staining with hematoxylin and eosin confirmed the findings from radiographic and manual assessments. Standard hematoxylin and eosin staining helped to illustrate bone formation between transverse processes. Figures 7, 8 and 9 display representative hematoxylin-and-eosin-stained sagittal sections of the lumbar spine, with the transverse processes of L4 and L5 labeled in the low-magnification images. Figure 7 shows representative sagittal sections of animals that received control implants. Consistent with previous findings, samples that received either absorbable collagen sponge alone or 1 µg of rhBMP-2 showed little new bone growth between the transverse processes (Fig. 7, a and f, respectively). Higher magnification images are shown in Figure 7, b through e (absorbable collagen sponge alone) and Figure 7, g, h, and i (1 µg of rhBMP-2). These images clearly show muscle between the transverse processes for both specimens. There was occasional evidence of new bone formation either originating from the decorticated transverse process or from normal remodeling (Fig. 7, h and j), but this new bone formation did not bridge the gap between the two transverse processes and was not considered as fusion. The surface of the transverse processes of both L4 and L5 of the specimen receiving 1 µg of rhBMP-2 was irregular and exhibited a "wavy" texture. These features are characteristic of immature, woven bone.
Hematoxylin-and-eosin-stained sagittal sections of control spines. Panels a through e show the control spine that was implanted with carrier alone. Panels f through j show the control spine that was implanted with 1 µg of rhBMP-2 alone. Panels k through o show the control spine that was implanted with 10 µg of rhBMP-2 alone. Boxed areas are shown at a higher magnification in the images to the right of the boxes. Note the gaps between the transverse processes of L4 and L5 in panels a through j. L4 and L5 cannot be distinguished in the spine that was implanted with 10 µg of rhBMP-2 (see panel k). Panels n and o detail endochondral bone formation and the presence of cartilage cells. Scale bar is approximately 100 µm for panels b, d, g, i, l, and n and approximately 50 µm for panels c, e, h, j, m, and o.
Hematoxylin-and-eosin-stained sagittal sections of spines implanted with truncated spp18 (TR-spp18) and rhBMP-2. Panels a through e show representative images of a specimen that was implanted with 1 µg of rhBMP-2 and TR-spp18. Panels f through j show representative images of a specimen implanted with 10 µg of rhBMP-2 and TR-spp18. Scale bar is approximately 100 µm for panels b, d, g, and i, and approximately 50 µm for panels c, e, h, and j.
Hematoxylin-and-eosin-stained sagittal sections of spines implanted with full-length spp24 (FL-spp24) and rhBMP-2. Panels a, b, and c show representative images of a specimen that was implanted with 1 µg of rhBMP-2 and FL-spp24. Panels d, e, and f show representative images of a specimen implanted with 10 µg of rhBMP-2 and FL-spp24. Scale bar is approximately 100 µm for panels b and e and approximately 50 µm for panels c and f.
Figure 7, k through o show images of a representative specimen implanted with 10 µg of rhBMP-2. All animals implanted with 10 µg rhBMP-2 contained bone masses so large that the interface between the transverse processes was indistinguishable. Figure 7, m and o represent the higher-magnification images of Figure 7, l and n, respectively. The trabeculae within the bone mass for these specimens were thicker and demonstrated more connectivity, characteristics substantiating that this bone mass is more mature. Figure 7, n and o reveal an area of the bone mass that appears to be in the middle of endochondral ossification and shows the presence of cartilage cells, some of which exhibited hypertrophy. Additionally, the presence of cartilage cells in a columnar pattern (Fig. 7, o) indicates that the large bone mass was formed via endochondral ossification.
Animals that received implants with the recombinant isoforms alone displayed similar histology results as the control animals that received the absorbable collagen sponge (data not shown). Although normal remodeling was noted, no substantial new bone formation was apparent.
Figure 8 shows hematoxylin-and-eosin-stained specimens implanted with TR-spp18 in combination with 1 µg of rhBMP-2 (Fig. 8, a through e) or 10 µg of rhBMP-2 (Fig. 8, f through j). Specimens from animals that received 1 µg of rhBMP-2 and TR-spp18 did not exhibit any new bone formation at the implantation site. The higher-magnification images of the interface between the transverse processes show mature trabecular bone, with some areas of normal remodeling. A specimen implanted with 10 µg rhBMP-2 in combination with TR-spp18 is shown in Figure 8, f through j. Some areas of normal remodeling were seen, as indicated by the wavy, woven bone appearance and a large area of new bone formed between L4 and L5, although not as large as the bone mass seen with implantation of 10 µg of rhBMP-2 alone.
Figure 9 displays hematoxylin-and-eosin-stained specimens implanted with FL-spp24 in combination with either 1 µg of rhBMP-2 (Fig. 9, a, b, and c) or 10 µg of rhBMP-2 (Fig. 9, d, e, and f). L4 and L5 are clearly distinguishable in both groups, confirming that FL-spp24 is inhibitory to rhBMP-2-mediated bone formation. The trabeculae of the bone are healthy and viable, indicted by the thickness and apparent nuclei present within the bone. This also suggests that FL-spp24 is not toxic to existing bone and only inhibits new bone formation.
Foremost among the materials used to enhance bone healing, particularly in spinal fusions, are BMPs, which are extremely efficacious but may be associated with dose-dependent side effects. Thus, delivery systems such as those incorporating BMP-binding proteins and peptides that exploit the full therapeutic potential of BMPs yet allow for control of the side effects are needed.
As mentioned, spp24 is a bone-matrix protein that was first isolated from bovine bone4. Subsequent research demonstrated that this protein contained a BMP-binding region12. It has become apparent, however, that earlier investigators recognized the osteogenic potential of spp24 without completely identifying the protein. Urist et al. Identified an 18.5-kDa glycoprotein with an acidic isoelectric point that eluted from hydroxyapatite at 180 mM of phosphate and co-purified with the active fraction of his BMP/NCP.5 Following Urist's protocols, we previously isolated the 18.5-kDa protein and identified it as a proteolytic fragment of spp2412. Furthermore, in 1987, Sen et al. described an osteogenic protein that was likely also a proteolytic fragment of spp24 because the amino terminal sequence of their protein revealed twelve of fourteen residues identical to the sequence that was later delineated for spp246. Murray et al. later defined three primary proteolytic products of spp248. Because of differences in post-translational modification in different systems, the exact molecular weights of these primary fragments vary somewhat but have been described as approximately 24 kDa (full length), 18 kDa, 16 kDa, and 14 kDa8.
It is our hypothesis that the 18-kDa protein examined here (TR-spp18) corresponds to Urist's 18.5-kDa protein and Sen's "osteogenic protein." Furthermore, it is our hypothesis that the different size variants of spp24 bind BMPs and related proteins differently and, through this mechanism, regulate the availability of various bone-growth factors in normal bone. The goal of the current studies was to compare how TR-spp18 and FL-spp24 affect rhBMP-2 activity in a prototypic model of bone-healing. This provides information on how the two proteins may function differently in the regulatory environment of normal bone. It also provides information that can be used to engineer therapeutic proteins for different clinical applications. A nineteen-amino-acid peptide (BBP) based on the sequence of the BMP-binding site of spp24 has demonstrated its ability to enhance rhBMP-induced bone-healing in animal models of spine fusion7 and long-bone healing13. While all of the spp24 size variants contain the BMP-binding region, there are other differences between these proteins and important differences between the native proteins and the proteins used in our studies.
A detailed analysis of the kinetics of the interaction between rhBMP-2 and the spp24 size variants has been presented elsewhere14. In general, the "affinity" for the interaction between rhBMP-2 and TR-spp18 is somewhat greater than that of the interaction between rhBMP-2 and spp24, as reflected in a smaller KD for the BMP/spp18 interaction (4.79 nM) than for the BMP/spp24 interaction (17.7 nM). These differences are not great, and other features of the molecules probably also influence their function. The kinetic studies and the studies presented here were conducted with use of proteins produced in bacterial expression systems that do not have the same post-translational modifications, such as phosphorylation, that are found in the native proteins synthesized in mammals. Phosphorylation of proteins expressed in bacterial systems occurs in a less specific manner than in mammals. That our materials are phosphorylated to a certain degree can be inferred by observing the "smear" of the proteins along the isoelectric point axis of a two-dimensional gel, as is seen in a previous report8. However, it is currently unclear what effect phosphorylation would have on the interaction of BMPs and spp24. There is one serine residue in the 19-amino-acid BMP-binding region that is highly conserved between species. Hu et al. examined the degree of phosphorylation of serines in native bovine spp24, but their study included only those serine residues in the C-terminal portion of the molecule and not the serine residue in the BMP-binding site (which was undefined at the time)4. The presence of serine residues in the C-terminal region of the molecule is highly variable between mammalian species, making it difficult to hypothesize an essential function. The hypothesis that mammalian phosphorylation might make spp24 pro-osteogenic is not supported by the observation of reduced bone mineral density in transgenic mice as expressed with spp24 compared with that of wild-type controls15.
The results of this investigation confirm findings from a previous study in which full-length spp24 inhibited rhBMP-2 activity in a mouse ectopic bone-forming assay15. In the rodent bone-healing model presented here, the recombinant isoforms of spp24 affect rhBMP-2 activity differently. While the full-length form markedly inhibited bone-healing, the shorter form did not. Although these proteins were exogenously added to the site of bone formation, we hypothesize that their respective activities provide insight into their physiological functions in bone. Specifically, we hypothesize that proteolytic processing of spp24 regulates the effective availability of rhBMP-2 by affecting a number of chemical and physical properties of the parental protein. These may include affinity for rhBMP-2, solubility of the binding protein, aggregate formation of spp/BMP complexes, interactions with other growth factors of the TGF-ß family, post-translational modifications of spp24, and other currently unrecognized parameters. The diminished inhibition of healing by the truncated form represents an effective net increase in rhBMP-2 activity. TR-spp18 was engineered to reflect our best approximation of the structure of the 18.5-kDa protein that appeared to enhance osteogenic potential in Urist's "BMP/NCP."5 Thus, it seems possible that proteolytic processing of FL-spp24 to TR-spp18 is a physiological regulatory step that results in a net increase of BMP-2 availability for bone formation. The microenvironment of the bone-healing site is complex, and much work needs to be done to define the many active participants and their interactions. Further progress in this area will allow the development of improved orthopaedic treatments.