Research ArticleBONE HEALING

Multifunctional scaffolds for facile implantation, spontaneous fixation, and accelerated long bone regeneration in rodents

See allHide authors and affiliations

Science Translational Medicine  24 Jul 2019:
Vol. 11, Issue 502, eaau7411
DOI: 10.1126/scitranslmed.aau7411

Mending with memory

Grafts or osteoconductive scaffolds need to fill large defect volumes to heal critical-size bone defects, but manufacturing and implanting materials to fit snugly within defects can be challenging. Zhang et al. designed compressible macroporous synthetic scaffolds that swelled to their original shape when hydrated (shape-memory). Cylindrical scaffolds implanted into femoral defects in rats delivered a low dose of bone morphogenetic protein while degrading over time to enhance bone formation. Results demonstrate how osteoconductive scaffolds can be engineered to stabilize graft fixation with facile surgical handing, warranting testing in larger animal models.


Graft-guided regenerative repair of critical long bone defects achieving facile surgical delivery, stable graft fixation, and timely restoration of biomechanical integrity without excessive biotherapeutics remains challenging. Here, we engineered hydration-induced swelling/stiffening and thermal-responsive shape-memory properties into scalable, three-dimensional–printed amphiphilic degradable polymer-osteoconductive mineral composites as macroporous, non–load-bearing, resorbable synthetic grafts. The distinct physical properties of the grafts enabled straightforward surgical insertion into critical-size rat femoral segmental defects. Grafts rapidly recovered their precompressed shape, stiffening and swelling upon warm saline rinse to result in 100% stable graft fixation. The osteoconductive macroporous grafts guided bone formation throughout the defect as early as 4 weeks after implantation; new bone remodeling correlated with rates of scaffold composition-dependent degradation. A single dose of 400-ng recombinant human bone morphogenetic protein-2/7 heterodimer delivered via the graft accelerated bone regeneration bridging throughout the entire defect by 4 weeks after delivery. Full restoration of torsional integrity and complete scaffold resorption were achieved by 12 to 16 weeks after surgery. This biomaterial platform enables personalized bone regeneration with improved surgical handling, in vivo efficacy and safety.


In recent decades, tremendous progress has been made toward the design of functional biomaterials for biomedical applications (14), such as shape-memory polymers (SMPs) (5, 6) capable of switching between a preprogrammed, permanent shape and temporary shapes in response to external stimuli, and dynamic three-dimensional (3D) scaffolds capable of environmentally responsive temporal changes (7, 8). However, successful clinical translations of these biomaterials/technologies have been rare because of overly sophisticated material designs, high manufacturing cost, inadequate in vivo efficacy and safety profiles, and the inability to simultaneously address multiple translational challenges in a single scalable system. For instance, despite advancements in programming SMPs [multistage shape changes (9) and multistimuli-responsive shape-memory (10)], achieving facile shape-memory programming and efficient shape recovery under physiologically safe conditions (11) while ensuring in vivo efficacy and safety (5) remains challenging. Meanwhile, implementation of functional biomaterial-based thermoplastic filaments or bioinks has yet to catch up with the fast-evolving 3D printing technologies for guided tissue regeneration applications (12, 13). Here, we aim to demonstrate the feasibility of engineering multiple functional features into a single scalable 3D-printed SMP scaffold to enhance the outcome of scaffold-guided tissue regeneration, specifically targeting the challenge of volumetric long bone defect reconstructions.

Bone graft–mediated regenerative reconstruction of critical-size long bone defects remains challenging because of the difficulty in obtaining adequate volumes of autograft from the patient’s own skeleton (14, 15). Meanwhile, devitalized allografts, the leading choice in arthroplasty, spinal surgery, and craniofacial reconstruction because of their attractive osteoconductive structural frameworks, are limited by inadequate fixation, poor graft incorporation, and consequently notoriously high failure rates for long bone repair (1619). Although collagen sponges loaded with osteogenic recombinant human bone morphogenetic proteins-2 (rhBMP-2) and rhBMP-7 have been clinically used for augmenting spinal fusions and tibial fractures (2024), they are not well suited for volumetric long bone defects due to the tendency of the soft sponges to deform or collapse during and after surgical insertion, which could exacerbate ectopic bone formation and nonunion. Hydrogel-based rhBMP-2 delivery systems have the same limitations due to their inadequate structural or mechanical integrity (2527). Current practices use high loading doses of bone morphogenetic proteins (BMPs) [micrograms to tens of micrograms in rodents (23, 28) and up to tens of milligrams in humans] with collagen carriers to promote union; this approach is expensive and can also lead to local and systemic adverse effects due to the high doses of BMPs (2931). Hard composite synthetic bone graft substitutes composed of degradable synthetic polymers and osteoconductive bioceramics have potential to support cellular attachment and osteogenic differentiation of stem and progenitor cells for guided bone regeneration (3237). However, poor structural integration of osteoconductive minerals with hydrophobic conventional degradable polymers often results in inferior surgical handling characteristics and inconsistent in vivo performance.

Amphiphilic degradable block copolymers provide an excellent opportunity to improve interfacial affinity with hydrophilic bone minerals (36) and to engineer environmentally responsive properties such as thermal-responsive shape-memory behavior. These copolymers undergo polymer chain segment rearrangements or phase separation in response to external perturbations (38, 39). Here, we report 3D-printed macroporous degradable scaffolds composed of 25% by weight (wt %) hydroxyapatite (HA), and 75 wt % poly(lactic-co-glycolic acid)-b-poly(ethylene glycol)-b-poly(lactic-co-glycolic acid) (PELGA) as non–load-bearing, shape-memory bone grafts for guided regeneration of critical-size femoral segmental defects. Combined with a single low dose of recombinant bone morphogenetic protein-2/7 heterodimer (rhBMP-2/7), these grafts facilitated facile surgical handling, self-fitting in vivo fixation, and timely graft resorption and osteointegration, resulting in expedited long bone regeneration with full biomechanical restoration in rodents.


Shape-memory and hydration-induced stiffening and swelling of 3D-printed HA-PELGA scaffolds

Amphiphilic triblock polymers [block compositions depicted in Fig. 1A, weight-average molecular weight (Mw) of ~110 to 120 kDa] of PELGA with varying ratios of poly(glycolic acid) (PGA) to poly(lactic acid) (PLA) repeating units were prepared by ring-opening polymerization with a polydispersity index (PDI) of 1.5. We previously showed that changing the PLA-to-PGA repeating unit ratio within the PLGA blocks from 8/1 to 2/1 accelerated the in vitro hydrolytic degradation of PELGA and HA-PELGA from 35 to 60 wt % and 30 to 51 wt % mass loss by 12 weeks, respectively (40). Macroporous square prisms [1 cm by 1 cm by 4 mm, computer-aided design (CAD); Fig. 1B, left] of the composite scaffolds of PELGA and 25 wt % HA were 3D printed in staggered orthogonal layers (line width, 400 μm; line spacing, 800 μm). The scaffolds have a macroporosity of about 65% and pore sizes greater than 300 μm, both within the recommended range for promoting vascularization and attachment of osteoprogenitor cells (41). von Kossa staining of the HA-PELGA filaments supported uniform distribution of calcium apatite mineral throughout the scaffolds (fig. S1). To examine the shape-memory capacity of 3D-printed HA-PELGA scaffolds, including temporary shape programming for straightforward surgical insertion under ambient conditions and efficient shape recovery by a safe triggering temperature in vivo, a macroporous HA-PELGA(8/1) square prism was compressed into a dense structure at room temperature (RT) and stably fixed in this shape at 4°C. Upon immersion in 50°C water, the compressed scaffold instantaneously recovered to its as-printed macroporous shape (Fig. 1B). Cyclic thermal mechanical testing confirmed a high shape fixing ratio (Rf) of ~99% at 0°C and a good shape recovery ratio (Rr) of ~88% at 40°C (Fig. 1C). 3D-printed HA-PELGA(2/1) scaffolds also exhibited similar shape-memory performance (Rf, 99%; Rr, 91%; fig. S2). The thermal transitions of the amphiphilic blocks (PLGA glass transition of ~10°C and peak poly(ethylene glycol) PEG melt transition of ~50°C; fig. S3) were exploited to enable facile temporary shape programming at RT and to trigger rapid shape recovery near body temperature (Fig. 1A, left).

Fig. 1 3D printing and characterization of the physical and thermomechanical properties of HA-PELGA composites.

(A) Depiction of thermal-responsive shape-memory and hydration-induced stiffening and swelling behaviors of a representative amphiphilic 25% HA-PELGA composite. (B) CAD model of a 3D scaffold with staggered macroporosity (left) and photographs (right) of the temporary shape programming and permanent shape recovery of a 3D-printed 25% HA-PELGA(8/1) scaffold triggered by 50°C water. (C) Stress-based cyclic thermal mechanical test of 25% HA-PELGA(8/1). Rf and Rr were calculated based on the second cycle. Rf, strain fixing ratio; Rr, recovery ratio. (D) Compressive moduli of 25% HA-PELGA scaffolds (n = 5) before (BH) and after hydration (AH). 2/1, 25% HA-PELGA(2/1); 8/1, 25% HA-PELGA(8/1); BH, as-printed/before hydration; AH at 37°C in water for 2 hours. Data are presented as means ± SEM. *P < 0.05, ***P < 0.001, ****P < 0.0001 (one-way ANOVA). (E) Differential scanning calorimetry traces of 25% HA-PELGA scaffolds BH and AH at 37°C in water for 2 hours. (F) Volume swelling ratios of 25% HA-PELGA(8/1) and 25% HA-PELGA(2/1) scaffolds (n = 3) AH at 37°C in water for 2 hours. Data are presented as means ± SEM. n.s., no significant difference, P > 0.05; Student’s t test. RT, room temperature; DI, deionized.

As-prepared scaffolds of HA-PELGA(8/1) and HA-PELGA(2/1) before hydration (BH) were compliant at RT, having compressive moduli of 70 and 38 kPa, respectively, but significantly stiffened to 181 kPa (P < 0.0001) and 126 kPa (P < 0.0001), respectively, after hydration (AH) for 2 hours in 37°C water (Fig. 1D). Although these macroporous scaffolds are designed as resorbable templates to guide bone regeneration rather than as weight-bearing dense permanent implants, maintaining adequate stiffness upon hydration and in vivo equilibration is desired from surgical handling and graft fixation perspectives. The stronger mechanical strength observed with HA-PELGA(8/1) versus HA-PELGA(2/1) likely resulted from its higher molecular weight and PLA content within the PLGA segments, which is consistent with the trend observed in a previous study (40). We previously investigated the underlying mechanism for hydration-induced stiffening in PEG-containing amphiphilic polymers and identified microphase separation and PEG crystallization as major contributing events (40, 42). Differential scanning calorimetry (DSC) of lyophilized 3D-printed HA-PELGA with prior hydration revealed an obvious endothermic peak at ~50°C during the heating process, corresponding to the melting of PEG crystals (AH traces; Fig. 1E). No obvious melting process was detected in as-printed scaffolds BH. PELGA(2/1) polymer chains within the 3D-printed amphiphilic composites had higher flexibility and hydrophilicity than did PELGA(8/1) due to lower PLA content and were thus more sensitive to hydration-induced molecular rearrangements, facilitating more pronounced PEG crystallization upon hydration (6.8 versus 3.2 J/g for the respective endothermic events). Last, both HA-PELGA(2/1) and HA-PELGA(8/1) scaffolds swelled upon hydration with volumetric swelling ratios of about 1.65 (Fig. 1F), supporting good aqueous wettability of the scaffolds. Together, the combination of shape recovery and hydration-induced swelling and stiffening behaviors of these macroporous scaffolds is expected to contribute to their stable fixation within a confined tissue defect upon equilibration.

Facile implantation and stable fixation of HA-PELGA grafts within critical-size femoral defects

To prepare suitable HA-PELGA grafts for precise and stable fitting within 5-mm rat femoral defects in vivo, we punched 3-mm-diameter cylindrical macroporous grafts from 3D-printed square prisms of 25 wt % HA-PELGA that were 5.6 mm in height (14 layers). This method was advantageous to directly printing the small grafts, which would result in peripheral line defects (edge effect) using a low-resolution consumer grader 3D printer. However, because of the relatively soft nature of as-printed macroporous HA-PELGA scaffolds, coring out the cylindrical grafts by ACU-PUNCH (3 mm in diameter, Acuderm Inc.) resulted in collapsed macroporosity (fig. S4). To minimize distortion during coring, dense square prisms of HA-PELGA filled with sacrificial material [poly(vinyl alcohol) (PVA)] were 3D printed. The dense scaffolds better withstood the punching pressure, resulting in preservation of the original shape and macroporosity of the grafts upon washing away the water-soluble PVA (Fig. 2A). No statistically significant difference (P > 0.05) was observed in the weight of the macroporous scaffold prepared by direct printing versus by printing with sacrificial PVA followed by aqueous washes (fig. S5), supporting adequate removal of PVA. The macroporous grafts were sterilized with 70% ethanol and ultraviolet (UV) light irradiation followed by air-drying and then readily pressed into a shorter cylinder at RT, for convenient placement within the 5-mm femoral segmental defect. The compressed grafts could be properly positioned within the defect in seconds and underwent rapid shape recovery, swelling, and stiffening upon warm (37°C) saline rinse (Fig. 2B) or contact with physiological fluid, achieving adequate fixation within minutes.

Fig. 2 Preparation and facile surgical fitting of 3D macroporous 25% HA-PELGA grafts.

(A) CAD illustration and photographs of a 3D macroporous HA-PELGA graft fabricated by coprinting a dense HA-PELGA/PVA composite square prism, coring, and subsequent removal of sacrificial PVA material. (B) Photograph of placement of a cylindrical-compressed HA-PELGA graft into a 5-mm rat femoral segmental defect and the graft fixation driven by shape recovery, swelling, and stiffening of the graft upon 37°C saline rinse. (C) Peak forces required to pull HA-PELGA(8/1) grafts, HA-PELGA(2/1) grafts, or collagen sponges (n = 3) from a customized specimen holder simulating a 5-mm rat femoral segmental defect upon hydration as measured by a mechanical test machine (MTS Bionix 370, MTS Systems Corporation). Hydrated collagen sponges were dislodged with negligible force (not detectable, <10 mN). All specimens were prepared in a cylindrical shape (diameter, 3 mm; length, 5 mm) and were hydrated in 37°C DI water for 1 min before the pull-out test. Data are presented as means ± SEM. *P < 0.05.

Stable graft fixation was accomplished in 100% of the subsequent in vivo studies; no HA-PELGA graft was dislodged from the defect throughout 16 weeks of post-op monitoring. By contrast, collagen sponges lost their shape and structure upon contact with saline or physiological fluid, making their surgical handling and placement tedious. Multiple attempts were often needed for proper positioning; sponges were easily dislodged once fully swelled, and sponge placement prolonged surgery time. A customized pull-out test measuring the force required to dislodge the hydrated HA-PELGA graft or collagen sponge from a simulated 5-mm femoral defect (fig. S6) revealed better fixation of HA-PELGA grafts than that of the collagen sponge (Fig. 2C). Whereas force of 100 to 300 mN was required to dislodge the hydrated HA-PELGA grafts, the soft collagen sponges could barely remain within the holder upon hydration and were dislodged with undetectable force (<10 mN). These observations demonstrate the advantage of HA-PELGA grafts over collagen sponges in terms of handling characteristics and support that the combination of shape-memory, hydration-induced swelling and stiffening behaviors of HA-PELGA scaffolds translated into more convenient implant placement and stable long-term graft fixation, both critical for successful clinical translation of this graft technology.

Bone regeneration within critical-size femoral defects guided by HA-PELGA grafts alone

To ensure the safety of HA-PELGA grafts for in vivo implantation, we first examined the cytocompatibility of the degradation products of HA-PELGA obtained upon accelerated in vitro hydrolytic degradation. Gel permeation chromatography (GPC) analysis identified the degradation products as low–molecular weight oligomers of lactic acid and glycolic acid (Mn, 1150 or less; fig. S7). When supplemented at concentrations of 2.5 to 7.5 μg/ml to the culture of rat bone marrow stromal cells (BMSCs) in expansion media, the degradation products did not negatively affect cell proliferation (fig. S8A). Higher concentration of degradation products (12.5 μg/ml) in culture media slowed the proliferation of BMSCs by day 5 (P = 0.0004 versus untreated culture); however, the ability of BMSCs to undergo robust osteogenesis was not compromised even when the high-concentration degradation products were supplemented two to three times a week for 2 weeks (fig. S8B).

To examine the ability of macroporous HA-PELGA grafts to guide bone regeneration and to determine the impact of scaffold degradation rate on bone regeneration outcomes, HA-PELGA(8/1) and HA-PELGA(2/1) grafts were implanted into 5-mm rat femoral segmental defects (n = 10 each; fig. S9A). The longitudinal guided bone regeneration over the course of 16 weeks was quantified by monthly in vivo micro-computed tomography (μCT) scans; bone remodeling and maturation was examined by femoral histology at 4, 12, and 16 weeks (Figs. 3 and 4). Reconstructed 3D images and 2D bone mineral density (BMD) maps at the region of interest (ROI) revealed substantial new bone formation as early as 4 weeks after implantation for both HA-PELGA(8/1)– and HA-PELGA(2/1)–treated groups, and the overall volume and maturity of new bone continued to increase over time (Fig. 3A), as supported by bone volume (BV) and BMD quantifications (Fig. 3B). Overall, aside from a difference in BMD at 8 weeks (P = 0.0011) and 12 weeks (P = 0.0019; fig. S9B), there were no obvious significant differences (P > 0.05) in BV or BMD between the two graft treatment groups. However, a close examination of the distribution of new bone over time revealed differences between the two graft treatments. Analysis of the center of the scaffold-filled ROI (cROI) revealed that although BV steadily increased over 12 weeks in HA-PELGA(8/1)–filled defects, it started to steadily decrease within the cROI in HA-PELGA(2/1)–filled defects after 8 weeks (Fig. 3, A and C). We hypothesize that dissolution of the more amorphous, lower-density new bone minerals (43) by the acidic local environment, partly due to scaffold degradation (44) [noting that crystallinity of calcium phosphate bone mineral inversely correlates with its solubility in an acidic environment (45, 46)] could contribute to the earlier drop in BV observed with the global thresholding applied in the faster-degrading HA-PELGA(2/1)–treated group. Trichome straining of the 16-week explants (Fig. 3D) revealed that collagen, the organic matrix material of new bone, indeed formed throughout the ROI in both groups despite the calcified mineral volume drop revealed by μCT, supporting that the scaffold-guided bone regeneration indeed occurred throughout the defect, including within cROI.

Fig. 3 Graft-guided bone regeneration within 5-mm rat femoral segmental defects in the absence of rhBMP-2/7.

(A) 3D μCT images and 2D bone mineral density (BMD) color maps (center sagittal slice) of 25% HA-PELGA(8/1)–filled versus 25% HA-PELGA(2/1)–filled defects over time. Global thresholding was applied to include minimum densities of 248.2 mg HA/cm3 (left column) or 518.2 mg HA/cm3 (right two columns) within the defect. HA-PELGA scaffolds were invisible with either thresholding. Red and blue represent higher and lower mineral densities in the color mapping, respectively. (B to C) Longitudinal μCT quantification of bone volume (BV) and BMD (n ≥ 6) within the entire defect [region of interest (ROI)] and the center of scaffold-filled defect (cROI) as indicated by the black and red boxes in (A), respectively. All data are presented as means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Two-way ANOVA for longitudinal comparisons within each graft composition over time. Global lower threshold of 518.2 mg HA/cm3 was applied for all quantifications. (D) Trichrome staining of longitudinal sections of 16-week explants showing new bone formation within ROI (boxed), with the cortical bone flanking the defect shown on the left and right sides in the lower-resolution images. Scale bars, 1.2 mm (25× magnification) and 300 μm (100× magnification).

Fig. 4 Histology of graft-guided bone regeneration, remodeling, and scaffold degradation within 5-mm rat femoral segmental defects over time.

(A) Longitudinal sectioning diagram of explanted femurs. (B) Micrographs of hematoxylin and eosin (H&E), alkaline phosphatase (ALP; blue) and tartrate-resistant acid phosphatase (TRAP; red), and toluidine blue (Tol blue)–stained sections from bone treated with 25% HA-PELGA(8/1) versus 25% HA-PELGA(2/1). Boxed regions are shown at higher magnification in bottom rows. CB, cortical bone; HC, healing callus; S, scaffold; NB, new bone; BM, bone marrow. Scale bars, 1.2 mm (25× magnification) and 300 μm (100× magnification).

Because μCT quantification does not provide insights into the remodeling of the organic matrices of guided bone regeneration, we performed femoral histology analyses at 4, 12, and 16 weeks after implantation. Hematoxylin and eosin (H&E) staining, alkaline phosphatase (ALP) and tartrate-resistant acid phosphatase (TRAP) staining, toluidine blue (Tol blue) staining, and polarized light microscopy were carried out to reveal cellularity, osteoclast/osteoblast-mediated bone remodeling, ossification mechanism, and degree of collagen fibril orientation, respectively. H&E staining of the longitudinal sections of decalcified graft-filled femurs (Fig. 4A) confirmed new bone formation throughout the ROI in both graft treatment groups at 4 weeks (Fig. 4B, left columns). The regenerated bone appeared well integrated with adjacent cortical bone, without detectable disruption by fibrotic tissues. Positive (purple) Tol blue staining supported an endochondral ossification mechanism (47) for the macroporous graft–guided bone regeneration. Pronounced osteoclast activity (denoted by red TRAP staining; Fig. 4B, middle columns) was detected at 4 weeks, most intensely surrounding the remaining scaffold material in the faster-degrading HA-PELGA(2/1)–treated group, but subsided by 12 weeks. Note that graft polymer dissolved during histology processing, leaving behind a void in place of the scaffold. By contrast, in the slower-degrading HA-PELGA(8/1)–treated group, substantial TRAP staining was detected at 12 weeks rather than at the earlier 4-week time point. These observations suggest an association between the extent of HA-PELGA scaffold degradation [more substantial for HA-PELGA(2/1) at 4 weeks] and the activated resorption activity of osteoclasts.

Osteoclast activities involve not only the secretion of degradative enzymes but also the reduction in local pH, thereby dissolving less-crystalline new bone minerals. The latter was consistent with the trended lower BV detected by μCT at the cROI after 8 and 12 weeks in HA-PELGA(2/1) and HA-PELGA(8/1) graft–treated defects, respectively (Fig. 3C). The remodeling of new bone coordinated by osteoclastic and osteoblastic activities were also detected across the bridging bony callus over time. New bone maturation was evidenced by recanalization of new bone revealed by μCT and histology [detection of bone marrow within bony callus; Figs. 3(A and D) and 4B] and stronger birefringence of polarized light microscopy, consistent with more aligned collagen fibrils (fig. S10A) within the ROI at 16 weeks. By 16 weeks, the ROIs in both treatment groups were characterized with smaller unfilled macroporous gaps and decreased TRAP/ALP staining, consistent with near-complete degradation of synthetic grafts and replacement with new bone (Fig. 4B). These histological analyses confirmed that new bone filled the defect and underwent remodeling into more mature bone matrix over time.

Functional regeneration of critical-size femoral segments by HA-PELGA grafts with minimal rhBMP-2/7

To examine whether a single low dose of rhBMP-2/7 may accelerate the bridging of the critical long bone defect by mature new bone and fully restore the biomechanical integrity of the defect, we preabsorbed 400-ng rhBMP-2/7 to the HA-PELGA grafts before surgical insertion (experimental design shown in fig. S11A). As early as 4 weeks after implantation, the HA-PELGA(8/1) and rhBMP-2/7 graft–filled defects showed complete bridging by bony callus templated by the scaffold, which was quickly remodeled into high BMD bone and became increasingly recanalized by 8 to 16 weeks as revealed by longitudinal BMD color mapping (Fig. 5A). Statistically significant increases in BV and BMD over time (P < 0.0001 versus 4 weeks) were confirmed by quantitative μCT analyses (Fig. 5B). Note that by 4 weeks, with the addition of 400-ng rhBMP-2/7, the BV at the ROI had already surpassed values observed at 16 weeks in the HA-PELGA graft (without rhBMP-2/7) groups (Fig. 3B). Femoral histology also revealed new bone remodeling by coordinated osteoclast/osteoblast activities occurred earlier in the presence of rhBMP-2/7, as demonstrated by intense and tightly coupled TRAP/ALP staining at 4 weeks rather than at later time points (Fig. 5C). Complete disappearance of synthetic scaffolds and substantial bone marrow penetration throughout the healing callus and the intramedullary canal were observed by 12 weeks (Fig. 5C). A much stronger birefringence in polarized microscopy, consistent with high degree of bone remodeling, was also observed with the single low dose of rhBMP-2/7 (fig. S10A), accompanied by a greater degree of recanalization as revealed by trichrome staining (fig. S10B). Full functional restoration of the torsional strength of the regenerated femurs was achieved, with six of seven femurs showing values comparable to or exceeding those of intact femur controls in weight-matched rats by 16 weeks (Fig. 5D).

Fig. 5 Accelerated healing of 5-mm rat femoral segmental defects by 25% HA-PELGA(8/1) grafts preabsorbed with 400-ng rhBMP-2/7.

(A) 3D μCT images and BMD color maps (center sagittal and axial slices) of the ROI showing maturing regenerated bone within the defect over time. Global thresholding was applied to exclude bone densities below 518.2 mg HA/cm3 (HA-PELGA graft invisible at this threshold). (B) Longitudinal μCT quantification of BV and BMD (n ≥ 12) within the ROI over time. Statistical significance over the 4-week data or as indicated by brackets indicated as follows: **P < 0.01, ****P < 0.0001. One-way ANOVA with Tukey’s post hoc test. Data are presented as means ± SEM. The global lower threshold of 518.2 mg HA/cm3 was applied for all quantifications. (C) Histological micrographs of H&E-, ALP/TRAP-, and Tol blue–stained sections of explanted graft-filled femurs over time. Scale bars, 1.2 mm (25× magnification) and 300 μm (100× magnification). Boxed regions shown at higher magnification in bottom rows. HC, healing callus; S, scaffold; NB, new bone; BM, bone marrow. (D) Boxplots of failure torque and stiffness of intact (control, Ctl) versus regenerated femur (8/1 + BMP) 16 weeks after being treated with HA-PELGA(8/1) grafts preloaded with 400-ng rhBMP-2/7 (n = 7). *P < 0.05, n.s.: P > 0.05, Wilcoxon–Mann-Whitney rank sum test.

With 400-ng rhBMP-2/7, the faster-degrading HA-PELGA(2/1) grafts also guided more efficient bone regeneration bridging across the entire defect. New bone remodeling and recanalization was observed at earlier time points (figs. S11 and S12), although the BV at the ROI at 16 weeks (P = 0.0095; fig. S11B) and failure torque at 16 weeks (P = 0.0117; fig. S11C) were lower than those in the HA-PELGA(8/1)– and rhBMP-2/7–treated group. The explanted femurs exhibited the same spiral fracture pattern after torsion test across the center of the regenerated femoral segment as observed in the intact control femurs (fig. S11D), supporting well-integrated graft-cortical bone interface. Organ pathology revealed no systemic negative side effects in the treatment group loaded with 400-ng rhBMP-2/7 when compared to normal controls (fig. S13), supporting the safety of the synthetic grafts, their degradation products, and the low dose of the rhBMP-2/7 delivered. In contrast, the bone regeneration guided by collagen sponge controls preabsorbed with 400-ng rhBMP-2/7 was characterized with extensive ectopic bone formations (100%; eight of eight) and some inconsistent bridging (fig. S14).


To address the challenges of regenerative repair of critical-size long bone defects, 3D-printed macroporous scaffolds composed of degradable amphiphilic polymer PELGA and osteoconductive HA were developed. The shape-memory property and hydration-induced stiffening and swelling effects of these HA-PELGA grafts enabled convenient surgical placement and 100% long-term fixation within 5-mm femoral segmental defects in rats, representing improved surgical handling over commonly used collagen sponges. The osteoconductive macroporous HA-PELGA grafts guided substantial new bone formation while they safely degraded and resorbed by 12 to 16 weeks after implantation. When loaded with a single dose of 400-ng rhBMP-2/7, these grafts accelerated bone regeneration: New bone fully bridged the defect by 4 weeks after implantation, and full restoration of the torsional strength was achieved by 16 weeks [HA-PELGA(8/1)]. No ectopic bone formation was detected, unlike the rhBMP-2/7–loaded collagen sponge control condition. These observations are consistent with literature reports of nonunions and ectopic bone formation resulting from BMP therapeutics delivered by collagen carriers (48, 49) including our prior study using the same femoral defect model in combination with collagen carrier preabsorbed with 3-μg rhBMP-2 (50).

Convenient surgical placement and stable graft fixation is difficult to achieve within critical long bone defects using conventional “hard” structural bone allografts (16, 18) or synthetic bone substitutes composed of stiff/brittle bioceramics such as HA (37), tricalcium phosphate (51), and bioglass (52). Conversely, soft polymer meshes (53), hydrogels (26), and collagen sponges/foams (54) are limited by inadequate structural or mechanical integrity for standalone long-term uses. Well-integrated SMP-bioceramic composite grafts that exhibit compliant properties before and during surgical insertion are compressible for facile surgical insertion and exhibit efficient shape recovery, and swelling and stiffening upon equilibration within the defect could provide exciting solutions. The 3D-printed macroporous HA-PELGA scaffolds reported here were designed to simultaneously achieve these multifunctional specifications when used in conjunction with standard load-bearing fixators [polyetheretherketone (PEEK) plate fixator used in this rat femoral segmental defect model]. The improved structural integration and useful surgical handling characteristics were accomplished without overly complicated material designs that could present translational hurdles. HA-PELGA grafts were prepared from components (PLGA, PEG, and HA) already used in U.S. Food and Drug Administration (FDA)–approved implants using the well-established ring opening polymerization process and scalable rapid prototyping.

In addition to distinct surgical handling characteristics, HA-PELGA scaffolds also exhibit compositional and structural properties desired for promoting bone regeneration. When well dispersed within amphiphilic polymer matrices, HA, an osteoconductive mineral, was able to promote osteogenic lineage commitment of BMSCs in the absence of culture induction (36). The surface potential of HA also enables surface adsorption of endogenous protein factors essential for initiating the bone healing cascade (36, 50, 55). Here, we demonstrated that the well-integrated HA-PELGA scaffolds enabled osteointegration across the defect in the absence of exogenous osteogenic factors. Although we previously showed that electrospun amphiphilic HA-PLA-PEG-PLA (HA-PELA) composite meshes could guide new bone formation in vivo, relatively slow degradation and limited microporosity of the meshes impeded fusion of new bone growth across the mesh, preventing timely resorption of the scaffold and restoration of mechanical integrity of the new bone (53). Here, we showed that HA-PELGA scaffolds with interconnected macroporosity facilitated earlier osteointegration and fusion of new bone across the defects, replacing the degraded scaffolds. Literature suggests that mismatch between scaffold degradation rates and tissue growth rates could negatively affect the quality of templated bone regeneration (56). Here, with the incorporation of the faster-degrading PLGA instead of PLA segment, HA-PELGA(8/1) and HA-PELGA(2/1) appeared to be mostly resorbed by 16 and 12 weeks, respectively. The extent of scaffold degradation apparently coincided with osteoclast-mediated resorption: Positive TRAP staining was detected more prominently at earlier time points in faster-degrading graft-treated groups, and the transient resorption of the inorganic component of new bone also occurred earlier in faster-degrading graft-treated defects. Further mechanistic investigation is warranted to understand potential positive feedback among PELGA degradation, local immune response, and osteoclast activity, which could all result in local pH reduction and affect scaffold resorption and new bone remodeling outcome (5759).

The efficacy of bone regeneration was further enhanced by delivering osteogenic growth factor rhBMP-2/7 via the HA-PELGA grafts, in a dose substantially lower than that of rhBMP-2 or rhBMP-7 used with conventional collagen carriers. rhBMP-2 and rhBMP-7 are approved by the FDA for expediting spine fusion, maxillofacial reconstructions, and fibula repair (31, 60). However, complications and side effects, some even life-threatening (30, 60), associated with the doses (typically at milligram scales) of these osteogenic factors applied with commercial carriers such as collagen sponges have raised serious safety concerns. In animal studies, the doses of rhBMP-2 required for healing rat critical long bone defects using collagen sponge or hydrogels as carriers range from several to tens of micrograms [equivalent to submilligrams to milligrams in human when scaled to 60 kg human body weight (61); table S1]. Higher doses of rhBMP-7 (50 to 200 μg in rats) are required for healing the same critical segmental long bone defects, which could lead to abnormal bone structure formation (table S1). rhBMP-2/7 heterodimer exhibits more potent osteoinductivity than either rhBMP-2 or rhBMP-7 (62), which may potentially translate into lower minimal effective loading doses. Although rhBMP-2/7 has not been approved by the FDA for clinical use, it has been used preclinically for treating calvarial defects and spine fusion in rats and minipigs (63, 64).

Here, we showed that a single dose of 400-ng BMP-2/7 [about 13 μg when scaled to a 60-kg human (61)] loaded on the HA-PELGA scaffold accelerated robust bone formation and achieved full functional restoration of the mechanical integrity of the defect. This substantial reduction in minimal effective dose of the rhBMP delivered could greatly enhance the safety and reduce the cost of such therapeutic interventions. When osteoconductive HA is homogeneously distributed within a polymer matrix, it is shown to facilitate not only the adsorption of exogenous proteins but also the enrichment of endogenous cytokines critical to bone healing (55), thereby improving scaffold-guided bone regeneration outcome. HA can also act as a buffering reagent to alleviate the negative impact of acidic degradation products of PLGA to some extent, mitigating inflammatory responses to the degrading scaffold. Meanwhile, local delivery of rhBMP-2/7 via HA-PELGA grafts also expedited the resorption of the synthetic grafts and new bone remodeling. Overall, we did not observe notable local and systemic side effects, highlighting the safety of both the synthetic scaffold and the low dose of BMP therapeutics.

Overall, the shape-memory and hydration-induced stiffening and swelling behavior of 3D-printed HA-PELGA enabled their facile surgical delivery and stable fixation in critical long bone defects in rats. Combined with a low dose of rhBMP, HA-PELGA grafts safely and effectively promoted functional long bone regeneration. This synthetic bone graft platform technology may be applicable to other indications. For instance, grafts may be exploited to facilitate minimally invasive surgical reconstructions of tibial plateau fractures (65), limited intercarpal fusion (66), spine fusion (67, 68), and mandibular defects (69), potentially reducing surgery time, minimizing pain (eliminate donor site morbidity), and improving regenerative reconstruction outcome. From a clinical translation perspective, it would be necessary to examine the efficacy of in vivo performances of HA-PELGA scaffolds in treating these defects in preclinical large animal studies. Tailoring macroporosities/degradability of HA-PELGA scaffolds for the specific anatomical site and using 3D printers with better resolution for recapitulating finer anatomical features may be required. Last, whereas the current rodent study used the potent osteoinductive growth factor rhBMP-2/7 to expedite the guided functional long bone regeneration, this heterodimer is not yet FDA approved for human uses. Thus, future large animal studies using rhBMP-2 homodimer would be advantageous from a regulatory approval perspective.


Study design

3D-printed composites of HA (25 wt %) and amphiphilic block copolymer PELGA with varying degradability [lactic/glycolic ratio within PLGA blocks: 8/1 or 2/1; HA-PELGA(8/1) or HA-PELGA(2/1)] were prepared as macroporous osteoconductive SMP bone grafts to guide the regeneration of 5-mm rat femoral segmental defects. These scaffolds were first characterized for their shape-memory and hydration-induced swelling and stiffening behaviors in vitro (n = 3), which were then exploited for facile and stable fixation into 5-mm segmental defects surgically created in rat femurs. The 5-mm rat femoral segmental defect is a critical-size defect that cannot heal on its own, as demonstrated by us and others (55, 70). This defect model was chosen to rigorously examine the in vivo efficacy of the HA-PELGA(8/1) and HA-PELGA(2/1) grafts (fig. S9A), as well as the HA-PELGA(8/1) and 400-ng BMP-2/7 graft and HA-PELGA(2/1) and 400-ng BMP-2/7 (fig. S11A) graft in guiding the functional regeneration of the long bone. The success of the shape-memory bone graft fixation, the quality and quantity of graft-guided bone regeneration, and the safety of graft degradation products and rhBMP-2/7 delivered were determined by a combination of monthly longitudinal μCT, femoral histology, end-point torsion testing, and systemic organ pathology. Skeletally mature (10 to 12 weeks) male Charles River SASCO Sprague Dawley rats (290 to 300 g) were assigned randomly to the various experimental groups. Each rat received identical graft/BMP treatments on the operated femurs to avoid contralateral effect. Power analysis for the in vivo study was detailed in the “Statistical analysis” section. Bone regeneration was longitudinally monitored by in vivo μCT immediately post-op (to establish baseline) and every 4 weeks until time of sacrifice (n ≥ 6). In one subgroup, rats were sacrificed at 4, 12, or 16 weeks for retrieval of grafted legs for femoral histology (n = 2 to 4 per treatment group per time point; minimum of three sections examined for each specimen and each staining). Another subset of rats was sacrificed at 16 weeks for torsion testing (n = 7 per treatment group) and the retrieval of organs for pathology (minimum of three sections examined for each harvested organ). Unoperated femurs (n = 7) retrieved from weight-matched healthy rats were used as controls. One harvested femur was excluded from torsion test due to accidental breakage during cement potting, but overall sample size did not change. Femoral histology and organ pathology examinations were carried out in a blinded fashion.

Polymer synthesis and characterization

PELGA(8/1) and PELGA(2/1) were synthesized by ring opening polymerization of D,L-lactide (LA) and glycolide (GA) initiated with PEG20000, purified, and characterized by 1H nuclear magnetic resonance (1H NMR) (to determine incorporated LA/GA ratio) and GPC as previously reported (40). Mw of PELGA(8/1) and PELGA(2/1) were 120 ± 10 kDa and 110 ± 10 kDa, respectively, with a PDI of 1.5. Polymers stored at 4°C for no more than 2 months were used for this study, with no obvious change in molecular weight, composition, or PDI due to time in storage as supported by 1H NMR and GPC.

3D macroporous 25 wt % HA-PELGA scaffold fabrication

Polycrystalline HA (3.3 g; Sigma-Aldrich) was sonicated in 20 ml of chloroform for 30 min to break up mineral aggregates, followed by the addition of 10 g of PELGA. This suspension was briefly vortexed and then stirred for 12 hours before it was poured into Teflon molds and left for another 12 hours in a ventilated fume hood to evaporate most solvent. The cast film was then thoroughly dried under vacuum at 60°C and cut into smaller pieces for thermal extrusion. HA-PELGA filaments were extruded at 160°C using an LCR7000 capillary rheometer (Dynisco Instruments) through a die with a 1.75-mm-diameter central hole. A MakerBot Replicator 2X 3D printer (MakerBot Industries) housed at 4°C was used to print 3D macroporous square prisms (10 mm by 10 mm by 4 mm, with a line cross section of 0.4 mm by 0.4 mm and a line spacing of 0.8 mm) according to CAD models designed in 3-matic (Materialise; Fig. 1B). Printing speed was 90 mm/s with nozzle temperatures of 160°C, and the build platform was maintained at 30°C.

Swelling of 3D-printed macroporous scaffolds

3D-printed square prisms (10 mm by 10 mm by 4 mm; n = 3) were hydrated in 37°C deionized (DI) water for 2 hours before the swelled dimensions were measured by a digital caliper and compared with those BH.

von Kossa staining

To examine the calcium distribution within HA-PELGA scaffolds, the extruded HA-PELGA(8/1) and HA-PELGA(2/1) filaments were stained by 1% silver nitrate aqueous solution under UV light for 20 min and then rinsed with DI water and dried in a vacuum oven overnight. Light microscopy images were taken for the filaments with and without von Kossa staining.

GPC analyses and cytocompatibility of the degradation products of HA-PELGA

To generate HA-PELGA degradation products for in vitro evaluation of their impact on the proliferation and osteogenesis of rat BMSCs, 0.5 mg of 3D-printed 25% HA-PELGA(2/1) and (8/1) were each placed in 1 ml of phosphate-buffered saline (PBS) (pH 7.4) at 37°C. The mixtures were heated to 70°C in a metal-heating block for 7 hours daily accompanied with frequent vortexing to accelerate the hydrolytic degradation. The degraded mixtures obtained after 17 days were centrifuged to pellet HA powder before the aqueous supernatant was retrieved, sterile-filtered through a 0.22-μm membrane, and stored at −20°C before GPC analyses or being supplemented in BMSC cultures in expansion and osteogenic media, respectively.

GPC was carried out on a Varian Prostar high-performance liquid chromatography system equipped with two Agilent 5-mm PLGel MiniMIX-D columns, a UV-visible detector and a 1260 Infinity evaporative light-scattering detector (Agilent Technologies). Tetrahydrofuran was used as an eluent at a flow rate of 0.3 ml/min at RT. The number-average molecular weight (Mn) and the PDI were calculated by the Cirrus AIA GPC Software using the narrowly dispersed polystyrenes (ReadyCal kits, PSS Polymer Standards Service, Germany) as the calibration standards. No oligomers with Mn over 3 kDa were detected, suggesting near-complete polymer degradation.

Rat BMSCs, freshly isolated from skeletally mature male SASCO-SD rats (Charles River) and enriched by adhesion culture (71), were seeded at 5000/cm2 in a 48-well culture plate and cultured in mesenchymal stem cell expansion media composed of α–minimum essential medium (GIBCO, no ascorbic acid), 20% fetal bovine serum (Sigma-Aldrich), 1% l-glutamine, and 1% penicillin/streptomycin. Filtered degradation product solutions were added to the cell culture media upon cell attachment and refreshed after each media change every other day, to result in constant concentrations of 2.5, 7.5, and 12.5 μg/ml of degradation product in culture (n = 3). Supplementing PBS served as a normal control. Cell viability was quantified by Cell Counting Kit 8 (CCK8, Dojindo) assay per vendor instructions at days 3 and 5 of the culture. The absorbance was read at 450 nm on a Multiskan FC Microplate Photometer (Thermo Fisher Scientific) as a function of degradation product concentration and culture time.

In parallel, rat BMSCs were plated at 25,000/cm2 and cultured in expansion media for 2 days to reach >70% confluency before being switched to osteogenic media (expansion media with 10-nM dexamethasone, 20-mM β-glycerol phosphate, and 50-μM l-ascorbic acid 2-phosphate) along with the supplementation of degradation products (7.5 or 12.5 μg/ml). Media were changed every 2 days for 2 weeks with the supplementation of the same concentrations of degradation products. After 2 weeks of osteogenic induction, cells were fixed with 10% formalin and stained with Alizarin red S stain for microscopic photodocumentation (100× magnification).

HA-PELGA femoral grafts preparation

To minimize graft distortion during subsequent punching, a CAD model was designed to coprint HA-PELGA and sacrificial PVA into a dense structure. Specifically, PVA was printed in between adjacent lines of HA-PELGA and vice versa (line cross section, 0.4 mm by 0.4 mm; line spacing, 0.8 mm) (Fig. 2A). The square prisms (16 mm by 16 mm by 5.6 mm) were printed using a dual-printing model with nozzle temperatures set at 160° and 200°C for HA-PELGA and PVA, respectively. The dense grafts were punched with a 3-mm-diameter biopsy punch. Grafts were washed with DI water until PVA was completely dissolved. As a control, macroporous HA-PELGA square prisms directly 3D printed without PVA fill were also subjected to coring by a 3-mm-diameter biopsy punch, which resulted in visible distortion/collapsing of the graft macroporosity. The macroporous HA-PELGA grafts obtained using the sacrificial PVA method were sterilized in 70% ethanol for 1 hour, followed by 1-hour UV light irradiation. rhBMP-2/7 (R&D systems) was reconstituted at 40 μg/ml in sterile 4 mM HCl PBS solution containing 0.1% bovine serum albumin. Then, 10 μl of the reconstituted growth factor solution was loaded onto each HA-PELGA graft to achieve an overall loading dose of 400 ng per graft. The loaded grafts were then dried in a ventilated biosafety cabinet for 3 hours and stored overnight at −20°C before surgical use.

Surgical procedure

All animal procedures were approved by the University of Massachusetts Medical School Animal Care and Use Committee. Male Charles River SASCO-SD rats (290 to 300 g) were sedated with 5% isoflurane-oxygen and maintained with 2% isoflurane-oxygen throughout surgery. A 5-mm femoral segmental defect was surgically created and stabilized with a PEEK internal fixation plate as previously described (55). Briefly, upon incision, the femur was exposed with a combination of blunt and sharp dissection of the muscles. The periosteum of the entire diaphysis was circumferentially removed to replicate a challenging clinical scenario. The PEEK fixation plate was stabilized with two stainless steel bicortical screws immediately adjacent to each side of the defect and further secured with two stainless steel hex nuts on the far cortex. The 5-mm defect was then created with an oscillating Hall saw under thorough irrigation with sterile saline. Further irrigation of the incision site after drilling ensured that bone debris was rinsed away. Compressed 3D-printed HA-PELGA(8/1) (n = 10; fig. S9A), HA-PELGA(2/1) (n = 10; fig. S9A), HA-PELGA(8/1) and 400-ng BMP-2/7 (n = 16; fig. S11A), or HA-PELGA(2/1) and 400-ng BMP-2/7 (n = 16; fig. S11A) grafts were positioned in the defect site and allowed to rapidly undergo shape recovery, swelling, and stiffening upon contact with physiological fluid to snugly fit within the defect. As clinically relevant controls, sterile absorbable collagen sponge (Integra LifeSciences Corporation) cylinders (3 mm in diameter and 5 mm in length) preabsorbed with 400-ng rhBMP-2/7 (n = 8) were placed into the same-sized defects. The muscle and skin were closed with 4-0 resorbable polydioxanone sutures in layers. The rats were given subcutaneous injections of cefazolin (20 mg/kg, once per day) and buprenorphine (0.08 mg/kg, every 8 hours) immediately post-op and for two more days thereafter. Longitudinal bone healing was monitored for 4 to 16 weeks, comparing to unoperated femurs retrieved from weight-matched healthy rats as controls.

Femoral histology and pathology

Fresh explants of the grafted femurs retrieved at 4, 12, and 16 weeks were fixed in periodate-lysine-paraformaldehyde fixative (72) at 4°C for 48 hours followed by decalcification in 18% aqueous EDTA (pH 8.0) for 4 weeks with frequent changes of fresh decalcification solution. The PEEK fixator and the bicortical screws were removed before the decalcified specimens were subjected to serial dehydration, paraffin embedding, and longitudinal sectioning (6-μm sections). Explanted sections were stained by trichrome for collagen, H&E for cellularity, osteogenic marker ALP (fast blue) and osteoclast lineage marker TRAP (fast red) for bone remodeling, and Tol blue for endochondral ossification. The heart, lung, kidney, liver, pancreas, spleen, and ribs were collected at 16 weeks, fixed and paraffin-sectioned and stained with H&E, and compared with normal controls (without receiving implants). All histology slides were imaged at 25× and 100× magnifications. Polarized light microscopy was also used to assess collagen fibril orientation and bone maturity at 50× magnification.

Statistical analysis

A priori power analysis based on a previous study using the same rat femoral segmental defect model in evaluating synthetic bone graft performances (50) determined that a sample size of 5 was needed to obtain 90% statistic power at the significance value of 0.05 when comparing temporal changes (monthly) in quantitative μCT analyses outcome. All μCT data presented in this study had a minimum sample size of 6. The power analysis also determined that a sample size of 6 was needed to obtain 90% statistic power at the significance value of 0.05 when comparing end-point torsion strengths between a control group and a treatment group. All torsion data presented in this study had a minimal sample size of 7.

All statistical analyses were performed using Prism 7.0 (GraphPad Software Inc.). Shapiro-Wilk normality testing was used to evaluate data distribution. Pair-wise comparisons passing normality test were analyzed with Student’s t test while the Mann-Whitney rank sum test was used for pairwise comparison of nonparametric data. Multiple group comparisons passing normality test were analyzed using one-way analysis of variance (ANOVA) with Tukey’s post hoc, whereas nonparametric multiple group comparisons were analyzed using the Kruskal-Wallis test with Dunn’s post hoc testing. Multivariant comparisons were carried out using two-way ANOVA with Tukey’s post hoc test. P values less than 0.05 were considered significant. All data were presented as means ± SEM, with the exception of data graphed using the boxplots. Individual subject level data are reported in data file S1.


Materials and Methods

Fig. S1. von Kossa staining of HA-PELGA filaments.

Fig. S2. Stress-controlled cyclic thermal mechanical tests of 3D-printed macroporous 25% HA-PELGA(2/1) scaffolds.

Fig. S3. DSC traces of as-prepared 25% HA-PELGA scaffolds during heating.

Fig. S4. HA-PELGA(8/1) graft preparation without sacrificial PVA.

Fig. S5. Weight comparison between HA-PELGA grafts prepared with and without sacrificial PVA.

Fig. S6. Pull-out test of in vitro graft fixation.

Fig. S7. GPC traces of degradation products of HA-PELGA(8/1) and HA-PELGA(2/1).

Fig. S8. Cytocompatibility of degradation products of HA-PELGA grafts.

Fig. S9. Experimental design and μCT quantification of HA-PELGA graft-guided bone regeneration.

Fig. S10. New bone formation and maturation over time templated by HA-PELGA with or without 400-ng rhBMP-2/7 by polarized light microscopy and trichrome staining.

Fig. S11. Experimental design, μCT quantification, and torsion test of HA-PELGA and 400-ng rhBMP-2/7 graft-guided bone regeneration.

Fig. S12. Histological micrographs of H&E-, ALP/TRAP-, and Tol blue–stained femoral sections from HA-PELGA(2/1) and 400-ng rhBMP-2/7–treated group.

Fig. S13. Representative micrographs of H&E-stained sections of organs.

Fig. S14. Healing of 5-mm rat femoral segmental defects by collagen sponge preabsorbed with 400-ng rhBMP-2/7.

Table S1. Literature BMP therapeutics applied to rat femoral segmental defects.

Data file S1. Individual subject-level data.

References (7385)


Acknowledgments: We thank A. Mason-Savas for histology support. Funding: This work was supported by an Alex Lemonade Stand Foundation Innovation Grant (to J.S.). Author contributions: B.Z. conducted polymer synthesis, scaffolds preparation, and physical property characterization of scaffolds. J.D.S. carried out the in vivo experiments, μCT analyses, and torsion test. J.R.M. assisted in polymer synthesis, GPC characterizations, and scaffold preparation. J.S. conceived and designed the project. D.C.A. advised the in vivo study. B.Z., J.D.S., and J.S. analyzed the data and wrote the manuscript. All authors reviewed and revised the manuscript. Competing interests: The authors declare that they have no competing financial interests. J.S. and B.Z are named as inventors on U.S. patent application PCT/US2017/064384 submitted by the University of Massachusetts Medical School that covers the chemical compositions of the amphiphilic SMPs and composites. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.

Stay Connected to Science Translational Medicine

Navigate This Article