Encapsulation of rhBMP-2 as a Strategy for Dose Shielding Whilst Preserving Structural Integrity, Bioactivity, and Osteogenic Potential

Abstract

Recombinant human bone morphogenetic protein-2 (rhBMP-2) is widely used to promote bone regeneration. However, conventional surface-attached delivery on absorbable collagen sponges causes a rapid burst release, excessive inflammation, and suboptimal healing. To overcome these limitations, we developed a thermally controlled Poly(DL-lactide-co-glycolide) (P_DLLGA) encapsulation system, designed to stabilize rhBMP-2 and enable controlled release. rhBMP-2 was incorporated in P_DLLGA pellets using the hot-melt extrusion of a lyophilized mixture containing poloxamer 407 and hydroxypropyl-β-cyclodextrin, and terminal sterilization (X-ray irradiation). The released rhBMP-2 maintained its molecular integrity after sterilization and remained stable for up to 732 days in storage, as confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and capillary electrophoresis (CE). Further, high-affinity binding between released rhBMP-2 and BMPR-IA was confirmed by bio-layer interferometry (BLI), and the released protein induced a robust in vitro ALP response, confirming preserved osteogenic activity. Our encapsulation approach for rhBMP-2 using P_DLLGA, including the combination product with β-TCP (LDGraft; Locate Bio, Nottingham, UK), provides a stable and bioactive rhBMP-2 delivery strategy with inherent dose-shielding properties, supporting safe, controlled, and effective bone regeneration therapies.

1. Introduction

Recombinant human bone morphogenetic protein-2 (rhBMP-2) is a potent osteoinductive growth factor [1], widely used to stimulate bone repair [2]. Its clinical use when delivered on absorbable collagen sponges has revealed pharmacokinetic limitations that can impair optimal bone healing [3]. Despite its widespread clinical use, delivery of rhBMP-2 on absorbable collagen sponges suffers from pharmacokinetic limitations, including rapid burst release, low local retention, and the mechanical fragility of the carrier matrix [4]. In vivo studies have shown that approximately 39.9% of rhBMP-2 remains at the implantation site one day after administration, decreasing to 0.5% within three weeks, indicating rapid diffusion and clearance [5]. Released rhBMP-2 can further distribute beyond the target site, potentially contributing to ectopic bone formation or local inflammation. Proteolytic degradation and systemic elimination lead to a short biological half-life, while the initial burst phase results in transiently high local concentrations that may cause adverse tissue reactions [4]. Bone regeneration proceeds through the following overlapping phases: an early inflammatory phase (first the few days post-injury), a reparative phase (weeks of new tissue formation), followed by a remodeling phase [6]. Relative to smaller mammalian species, such as rodents, rabbits, and sheep, which exhibit faster bone healing and remodeling rates [7], human bone repair is characterized by a delayed release of endogenous BMP-2, with a peak expression 7–14 days post-fracture [8,9]. This availability of growth factors in specific phases of tissue repair is tightly regulated and naturally modulated by the extracellular matrix (ECM) and its inhibitors. Acting as a reservoir, the ECM protects and controls the spatio-temporal activity of growth factors by releasing them in response to matrix degradation [10,11]. The initial inflammatory phase is crucial for triggering tissue healing, but an excessive or prolonged inflammatory response can be detrimental [6,12]. When rhBMP-2 is surface-attached to an absorbable collagen sponge carrier, it results in a burst and uncontrolled release coinciding with the inflammatory phase of bone healing [5]. This supraphysiologic rhBMP-2 bioavailability, immediately post-implantation, interferes with the inflammatory phase of bone healing, and is associated with adverse inflammation and neurologic complications that impair bone regeneration [13,14,15,16,17]. In preclinical studies, an initial burst of rhBMP-2 has been shown to trigger the recruitment of large numbers of macrophages and provoke an exaggerated inflammatory reaction (often manifesting as granuloma, seroma, or hematoma formation) around the implant [18]. Clinically, such early inflammation can translate to complications like painful swelling, neurological deficits, or even airway compression in cervical [19,20].

It has been shown that delaying the delivery of rhBMP-2, such that less is presented during the inflammatory phase and more in the repair phase [18], or until the bone healing phase, leads to improved outcomes [18,21,22]. This delayed peak corresponds to the natural spatio-temporal release pattern of endogenous BMP-2 in human fracture healing [2].

Whilst various carriers for rhBMP-2 surface attachment have attempted to overcome retention challenges, the collective evidence indicates a fundamental limitation of surface rhBMP-2, which is its inherent immediate bioavailability and subsequent receptor activation [23,24,25]. The mechanistic constraint of surface-bound rhBMP-2 arises from its presentation mode, whereby the entire protein fraction is immediately exposed to cell surface receptors, inducing rapid activation and signal attenuation. In contrast, matrix-bound BMP-2 alters the spatio-temporal presentation of the ligand, sustaining receptor engagement over time and promoting prolonged SMAD phosphorylation and signaling [24]. Thus, surface-attached delivery strategies fail to address the sequential needs of bone healing.

Various carrier materials were designed to effectively delivery rhBMP-2, including polymeric microspheres, hydrogels, and nanoclay-based composites [26,27,28]. Hydrogel-based drug delivery methods have emerged as potential solutions for bone regeneration due to their biocompatibility, tunable mechanical properties, and ability to support cell growth and differentiation [19,20]. However, hydrogels have several limitations for bone therapies, such as their poor mechanical strength for load-bearing applications, high swelling ratio, rapid degradation, potential immune responses, and leakage due to needle injections [20,29,30]. Similarly, nanoclay–composite hydrogels emerged recently as promising carriers, due to their binding affinity for biomolecules, dose-controlled release strategy, and osteogenic potential [27,31]. Despite promising results, their clinical translation remains limited by challenges with cytotoxicity, biodistribution, mechanical stability, immune reactions, and uniform manufacturing [31,32]. Conventional polymer-based encapsulation systems, such as Poly(lactic-co-glycolic acid) (PLGA) microspheres, or films produced via solvent-based double-emulsion or microfluidic techniques, have improved sustained release but often suffer from solvent-induced protein instability, high burst release, and poor scalability for manufacturing [33]. These limitations have motivated the development of solvent-free thermally controlled systems that ensure greater formulation stability and clinical translatability. Calcium phosphate ceramics remain the gold standard material in bone regeneration, owing to their excellent biocompatibility and osteoconductive properties [34]. In particular, β-tricalcium phosphate (β-TCP) supports bone remodeling through osteoclast-mediated resorption that proceeds in parallel with new bone ingrowth, ultimately resulting in the complete replacement by native bone [35,36]. Moreover, β-TCP is a clinically approved and widely used material for bone-grafting applications [36]. Combining β-TCP with Poly(DL-lactide-co-glycolide) (P_DLLGA) provides a novel strategy to unite the osteoinductive BMP-2 with the osteoconductive properties of the ceramic, using a controlled release profile provided by the polymer [37,38]. Among polymer-based carriers, P_DLLGA was chosen over Poly(L-lactic acid) (PLLA) due to its amorphous structure, which enables uniform degradation within weeks to a month, a timeframe that aligns with the sequential needs of bone regeneration and its controlled release system [38]. Moreover, its faster resorption rate, reduced acidic accumulation, and improved protein stability make it an ideal carrier for temporary BMP-2 delivery compared with the more crystalline and slowly degrading PLLA [39].

An ideal delivery would mimic the natural BMP-2 patterns during fracture healing, shield the majority of the implanted rhBMP-2 from initial receptor binding, and provide a controlled long-term release coinciding with the repair phase. The encapsulation of rhBMP-2 in biodegradable carriers has been explored as a potential strategy to achieve these goals [40,41]. The initial bioavailability can be tempered by providing “dose shielding” of the protein from the local tissue. This encapsulation strategy has been shown to reduce the inflammatory response relative to collagen sponge delivery [40]. To date, no clinically available rhBMP-2 product employs an encapsulated delivery strategy that offers a safer and more controlled alternative for bone regeneration. P_DLLGA (50:50) is a biocompatible polymer used in various drugs approved by the US Food and Drug Administration (FDA) and cleared for medical device products. It is known for its modifiable biodegradation profile [42], making it well-suited as a biomaterial to match the various phases of bone repair. Encapsulating rhBMP-2 in P_DLLGA pellets should provide “dose shielding” of the growth factor during the acute inflammatory phase, but the molecular consequences of incorporating and processing the rhBMP-2 within a clinical-grade polymer matrix is unclear.

In the present preclinical study, we develop clinical-grade rhBMP-2-loaded P_DLLGA (50:50) pellets as a potential encapsulated delivery system and assess the quality of the released protein even after long-term shelf storage, to evaluate its bioactivity and therapeutic relevance. The protein quality of the released rhBMP-2 was assessed by determining the size, charge, and purity, using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and capillary electrophoresis (CE). The binding kinetics of the released rhBMP-2 to the BMP receptor type IA (BMPR-IA) were analyzed using bio-layer interferometry (BLI). Finally, we analyzed the osteogenic potential of the clinical-grade product (LDGraft) that compromises P_DLLGA encapsulation of rhBMP-2 combined with β-TCP granules within a poloxamer 407 carrier.

2. Materials and Methods

2.1. rhBMP-2 Encapsulation and Feedstock Preparation

Pellets containing rhBMP-2 were manufactured using a 9 mm twin screw extruder fitted with a bespoke hollow die, conveyor, and granulator (Three-Tec, Seon, Switzerland). A feedstock was prepared by first lyophilizing (SP VirTis AdVantage Pro Freeze Dryer, SP Scientific, Warminster, PA, USA) a solution of Escherichia coli (E. coli), derived rhBMP-2 (Locate Bio, Nottingham, UK) and hydroxypropyl-beta-cyclodextrin (cyclodextrin, Merck, Darmstadt, Germany) in 1% acetic acid. The resulting lyocake was disrupted, sieved, and geometrically mixed with milled P_DLLGA (50:50) (Evonik, Essen, Germany) and poloxamer 407 (P407, BASF, Ludwigshafen, Germany) to prepare the final feedstock which was delivered to the hot-melt extruder (as described elsewhere in WO2022238718A1). The residence time during extrusion was approximately 6–8 min, with processing temperatures ranging from 60 to 80 °C; quantitative data on shear stress were not available. The hollow extrudate was pelletized to form short tubes of less than 1.3 mm by the longest dimension. To minimize the risk of rhBMP-2 inactivation during hot-melt extrusion, hydroxypropyl-β-cyclodextrin was included as a stabilizing excipient to shield the protein from thermal and shear stress. Additional conditioning and packaging steps were applied before and during irradiation to maintain protein integrity. The rhBMP-2 pellets were subsequently irradiated using 25–32.5 kGy X-ray irradiation (Steris, Biel, Switzerland). X-ray irradiation was employed for terminal sterilization, as it provides a shorter exposure time and a lower product temperature than gamma irradiation, thereby reducing the risk of thermal and radiolytic damage to rhBMP-2. The method also enables greater process scalability and dose uniformity compared with E-beam irradiation. As a control material, empty pellets (EP) were manufactured as above, without the inclusion of rhBMP-2. The SDS-PAGE and BLI assays utilized rhBMP-2 pellets alone, and the ALP assay combined the rhBMP-2 pellets with β-TCP, poloxamer 407, and saline to form a putty (LDGraft), as described in patent WO2022238718A1. After manufacturing, pellets/LDGraft were frozen at −20 °C, and subsequent usage was considered as time zero (T0). For shelf-life evaluation, separate batches of pellets were stored at room temperature (RT) (25 ± 2 °C, 60% relative humidity (RH)) for defined periods prior to testing to assess long-term stability and protein release (Tn with n = the number of dates stored at RT).

2.2. Molecular Weight Determination of Released rhBMP-2 from Pellets by SDS-PAGE

Pellets (T0) were thawed and rhBMP-2 (0.5 µg/well) was released from pellets utilizing previously described methodology [43]. Extracted samples were added to Laemmli sample buffer. Samples were boiled at 95 °C for 5 min using a dry bath system (Starlab International; Hamburg, Germany) before loading. SDS-PAGE was performed by loading 0.5 µg of rhBMP-2 per lane under non-reducing conditions and run alongside a pre-stained ladder (PageRuler™ Plus, 10 to 250 kDa, Thermo Fisher Scientific, Waltham, MA, USA) using Novex™ Tris-Glycine Gels (10–20%, WedgeWell™ format; Invitrogen, Waltham, MA, USA). Gel electrophoresis was performed using a mini gel tank (Invitrogen) powered by Fisherbrand Mini 300V Plus, (Thermo Fisher Scientific), at 125 V for 40 min and then 50 V for 20 min. Pierce^TM Silver Stain Kit (Thermo Fisher Scientific) was used as per manufacturer instructions for the staining of the gels. Analysis of the gels was performed manually after taking images using FastGene B/G GelPic Box (GeneFlow, Lichfield, UK).

2.3. Receptor Binding Analysis of rhBMP-2 to rhBMPR-IA/ALK-3 by BLI

rhBMP-2 formulated within P_DLLGA pellets and EP as reference, were thawed (T0) and protein released into the release buffer (1× Kinetics Buffer (KB); 1× Phosphate-Buffered Saline (PBS), 0.01% Bovine Serum Albumin (BSA), 0.002% Tween20, pH 7.4) through a controlled incubation process. Specifically, 60 ± 0.5 mg of pellets were incubated in 1 mL of 1× KB at 37 ± 2 °C for 24 ± 2 h with agitation at 350 revolutions per minute (rpm). Following incubation, the samples were centrifuged at 146× g for 5 min, and the supernatant was collected. The supernatant was then gently mixed by inversion (20 times), aliquoted, and stored at −75 ± 15 °C for subsequent analysis. The exact concentration of rhBMP-2 in the release buffer was not determined; however, all samples were treated and diluted equally prior to sensor loading, enabling relative comparison of binding affinities. On the day of the binding assay, the supernatant was buffer exchanged using spin desalting columns (Zeba™ Spin Desalting Columns, 7K MWCO, Thermo Fisher Scientific), according to manufacturer’s instructions. The binding interaction between the released rhBMP-2 and rhBMPR-IA/Activin Receptor-like Kinase 3 (ALK-3) (Fc Chimera Protein, CF, Catalog #: 2406-BR, R&D systems, Minneapolis, MN, USA) was evaluated using BLI on an OctetRED96 system (Sartorius, Göttingen, Germany) with nickel–nitrilotriacetic acid (Ni-NTA) biosensors (Sartorius). The rhBMP-2 lacks a histidine tag; however, it was successfully immobilized on Ni–NTA biosensors, likely via native histidine or metal-coordinating amino acids. To ensure that the subsequent binding of BMPR-IA was specific and not due to non-specific protein adsorption or background interference, control experiments were performed (Appendix A). To verify non-specific surface loading and selective interaction with BMPR-IA, the following BLI studies were assessed: (i) validation of rhBMP-2 immobilization and structural integrity of released rhBMP-2 (Scheme A1) and (ii) selective interaction or rhBMP-2 with BMPR-IA (Scheme A2). The study included the following four control conditions: (1) EP (P_DLLGA pellets without rhBMP-2); (2) force-degraded rhBMP-2 released from P_DLLGA pellets; (3) native released rhBMP-2 from P_DLLGA pellets (untreated); and (4) standard rhBMP-2 (drug substance, non-encapsulated).

All samples were equally diluted and loaded under identical conditions, and BLI signal during loading was used to assess binding to the Ni–NTA surface. The primary BLI experiment was designed to evaluate the ability of encapsulated rhBMP-2 subsequently released from pellets, to bind BMPR-IA under physiological conditions. Ni–NTA biosensors were loaded with BMP-2-containing release material and subsequently exposed to BMPR-IA to assess binding kinetics and interaction specificity.

Force-degraded rhBMP-2 was used as negative control protein. Degradation was achieved by incubating the pellets at 60 °C for 7 d, and the protein was subsequently released following the procedure described above. As a positive control, rhBMP-2 (the drug substance, non-encapsulated) was immobilized on Ni–NTA biosensors and exposed to BMPR-IA to confirm specific receptor binding. A no-load control (sensor without immobilized rhBMP-2) was included to verify the absence of non-specific surface binding.

For the primary binding assay, biosensors were prewetted in 1× KB for 10 min. A baseline was established in 1× KB, followed by a second baseline using released material from EP. Biosensors were loaded with rhBMP-2-containing released material from P_DLLGA pellets for 120 s, equilibrated in 1× KB, and exposed to five non-zero concentrations of BMPR-IA (625, 312.5, 156.3, 78.1, and 39.1 nM), along with a buffer-only control, to assess the binding association. Dissociation was monitored by transferring the biosensors back into 1× KB. To account for assay drift, sensorgrams were reference-subtracted using data from the buffer-only control, and binding curves were aligned at 5 s post-baseline before global fitting to a 1:1 Langmuir binding model using Octet Data Analysis software (Version: 9.0.0.14, Sartorius, Göttingen, Germany), to obtain kinetic constants k_on, k_off and K_D (K_D = k_off/k_on). The assay’s repeatability was assessed by a single analyst under same experimental conditions, while intermediate precision was evaluated through replicate experiments performed by two independent analysts on different days.

2.4. Molecular Stability Assessment Using Capillary Electrophoresis

CE was performed to analyze the molecular stability of released rhBMP-2 from pellets. P_DLLGA pellets containing rhBMP-2 were stored at −20 °C to arrest aging and subsequently provide a day 0 sample (T0), or at 25 ± 2 °C, 60% RH for 129, 227, 551, and 732 d before testing (T129, T227, T551, and T732). Duplicate rhBMP-2 samples (30 ± 0.5 mg each) were extracted with 8 mL of pre-chilled 100% acetone, vortexed, and incubated at −20 °C for 1 h. Following centrifugation at 7142× g for 10 min, 6 mL of the supernatant was discarded. The remaining pellet and residual acetone were transferred to fresh tubes. To maximize protein recovery, the original tubes were rinsed with 2 mL of pre-chilled acetone, vortexed, and transferred to new tubes. After a second centrifugation (7142× g, 10 min), supernatants were removed, and pellets were air-dried. Pellets were resuspended in 100 µL of SDS sample buffer. For alkylation, 1 µL of 20% SDS and 10 µL of 0.5 mol/L iodoacetamide were added and incubated at RT for 10 min. Samples were then denatured under reducing conditions using β-mercaptoethanol and SDS, followed by incubation at 70 °C for 10 min. After cooling, the samples were measured by CE-SDS analysis and a molecular weight sizing standard (Sciex, SDS-MW Analysis Kit, Cat No. 39095, Framingham, MA, USA), a 10 kDa blank, and a reference standard of the non-encapsulated rhBMP-2 (3 mg/mL) were included to evaluate the changes over time by monitoring molecular weight profiles through a sieving mechanism.

CE-SDS was performed on a PA 800 Plus system (Sciex, Framingham, MA, USA) equipped with Empower 3 software (Waters Corporation, Milford, MA, USA). Separations were carried out using bare fused silica capillaries (30.2 cm total, 20 cm effective length) under reducing conditions with β-mercaptoethanol and SDS. Samples were incubated at 70 °C prior to injection. Protein migration was detected by UV absorbance at 220 nm. Injection sequences and instrument settings followed validated standard operating procedures. Data analysis was performed using Empower 3 software (Waters Corporation, Milford, MA, USA, version 3.3). with integrated molecular weight calibration. Purity was assessed as the percentage of main peak (rhBMP-2 homodimer), low molecular weight species (LMWS), and high molecular weight species (HMWS).

2.5. Cell-Based ALP Assay to Assess the Osteogenic Potential of Released rhBMP-2

To assess the osteoinductive potential of released rhBMP-2 from the clinical graft product, LDGraft, Alkaline phosphatase (ALP) activity was quantified in C2C12 cells. LDGraft (1× 5cc pack) was stored at −20 °C until thawed, for use in the assay (T0). C2C12 cells were cultured in medium (DMEM, low glucose, GlutaMAX™ Supplement, Pyruvate, Sigma Aldrich, St. Louis, MO, USA) 10% heat-inactivated Fetal Calf Serum (Thermo Fisher Scientific) and 1% penicillin/streptavidin (Sigma Aldrich), at 37 °C in a humidified atmosphere with 5% CO₂. A total of 40,000 cells in a volume of 1 mL/well was plated into a 12-well plate for 24 h. LDGraft (mean sample weight 0.1453 g) was exposed to cells in polyethylene terephthalate (PET) membrane inserts (Avidity Science, Waterford, WI, USA, formerly Triple Red), covered with 2 mL of fresh medium at the basolateral chamber and 1ml to the insert. The relative positions of each sample in the 12-well plate were maintained throughout the 28-day study. Liquid rhBMP-2 (0.2 μg/mL) was used as a positive control, while culture medium served as the negative control. After 72 h or 96 h incubation, the inserts were transferred onto freshly plated cells (prepared the previous day), along with fresh medium and controls. Harvested cells were lysed (CelLytic M, Sigma Aldrich), the ALP levels measured (SIGMAFAST p-Nitrophenyl phosphate tablets, Sigma Aldrich) according to the manufacturer’s protocol, and results were normalized to protein content of the cell lysate as measured by the bicinchoninic acid assay, (BCA Protein Assay Kit, Pierce, Appleton, WI, USA).

The PET membrane insert transfer/ALP assay cycle was repeated up to 40 d after exposure. Relative ALP activity was normalized to the response of the time-matched positive control of known rhBMP-2 input. This approach ensured valid comparisons across all experiments with varying incubation duration. ALP activity was measured for three independent manufacturing batches (n = 3), with each sample tested in triplicate (technical replicates). Data were expressed as mean ± SD. Statistical significance was evaluated using an unpaired, two-tailed Student’s t-test, with p < 0.05 considered statistically significant.

3. Results

3.1. Structural Analysis of Released rhBMP-2 by SDS-PAGE

The protein structure of released rhBMP-2 was analyzed by SDS-PAGE under non-reducing conditions to assess potential structural alterations. Figure 1 shows SDS-PAGE analysis of rhBMP-2 extracted from three different batches following irradiation, compared to non-irradiated released protein and native rhBMP-2. All samples showed a predominant band within the 15–25 kDa range, corresponding to the expected dimeric form of rhBMP-2, with no evidence of high molecular weight aggregates. Under non-reducing conditions, rhBMP-2 migrated at approximately 15–25 kDa on SDS-PAGE, consistent with the known anomalous migration of cystine-linked dimers (theoretical MW ≈ 26 kDa) caused by their compact, disulfide-stabilized structure comprising seven intrachain and one interchain disulfide bond [44,45]. These results demonstrate that the native disulfide-linked structure of rhBMP-2 remained intact throughout encapsulation, terminal sterilization, and the subsequent release from the pellets.

3.2. Molecular Stability Assessment of Released rhBMP-2 by Capillary Electrophoresis

To assess the structural integrity and molecular weight species distribution of released rhBMP-2 from P_DLLGA after defined long-term storage intervals, CE-SDS, under denaturing conditions, was employed. This technique enables the precise separation of protein species according to their molecular size. Samples from multiple time points (duplicates) were analyzed to monitor fragmentation, aggregation, or degradation over time. T0 served as the reference, reflecting the initial release sample with initial protein integrity prior to storage-induced changes. The main peak corresponding to intact rhBMP-2 homodimer accounted for 89.240% of the total signal, while LMW and HMW species were present at 6.300% and 4.450%, respectively (

Table 1. CE-SDS analysis of released rhBMP-2 from pellets following long-term storage. Relative peak areas (%) of LMW, main peak, and HMW measured for two samples (n = 2) at each time point (T0, T129, T227, T551 and T732) during the long-term stability study of rhBMP-2 containing P_DLLGA pellets. Values expressed as mean (%) and standard deviation (SD). Data shown correspond to those visualized in Figure 2.

Time Point	LMW		Main Peak		HMW
	Mean (%)	SD	Mean (%)	SD	Mean (%)	SD
0	6.300	0.028	89.240	0.453	4.450	0.481
129	8.400	0.438	89.750	0.453	1.845	0.021
227	9.135	0.700	85.705	1.803	5.160	1.103
551	7.585	0.177	82.185	0.884	10.230	0.707
732	8.805	0.120	85.200	0.283	5.990	0.410

). The results provide insight into the purity, stability, and consistency of RT-stored rhBMP-2 in P_DLLGA pellets across defined endpoints (T129, T227, T551, T732 days). Throughout the storage period, slight variations were observed in the main peak as well as in the LMW and HMW species. However, no consistent trend indicating accumulation of either the HMW or LMW species was detected across time points, supporting the overall stability of the protein. The lowest dimer content (82.2%) was detected at T551, which corresponds to the peak in HMW species (10.23%). Both main peak and HMW species returned to their T227 levels at the final storage endpoint, T732. These fluctuations may represent background assay variability, or a slight increase in pH for the sample at T551, which is known to increase rhBMP-2 aggregation [44]. Despite minor shifts, rhBMP-2 remained structurally stable until the study endpoint, T732, with the main peak consistently dominant. The relative distribution of the LMW, main peak, and HMW species across time points is shown in Figure 2, with quantitative values in Table 1.

3.3. Binding Affinity of Released rhBMP-2 to BMPR-IA Measured by Bio-Layer Interferometry

Previously, studies using live cell membranes and fluorescence recovery after photobleaching (FRAP), have shown that BMP-2-induced heteromeric receptor complexes form and disassemble quickly, implying the ligand frequently unbinds and rebinds in a “vibration-like” interaction pattern [46]. Binding affinity to BMPR-IA by rhBMP-2 released from encapsulated P_DLLGA pellets was assessed by BLI. Sensorgrams revealed a characteristic binding profile, with robust association and dissociation kinetics observed across defined BMPR-IA concentrations (Figure 3).

Corresponding kinetic analysis revealed a high-affinity interaction for released rhBMP-2, with a dissociation constant (K_D) in the low nanomolar range, consistent with previously reported values for native BMP-2 [47]. The observed kinetic parameters were highly reproducible across experiments performed on separate days by different analysts. Intermediate precision analysis yielded an average K_D of 4.89 × 10⁻⁸ M and a coefficient of variation (CV) of 24%, reflecting acceptable reproducibility across independently performed assays (

Table 2. Intermediate precision of binding method to assess binding affinity of BMPR-IA to released rhBMP-2 immobilized on the Ni-NTA biosensors. Data represent repeated measurements across independent runs (1–4) and analysts (A and B) to assess assay reproducibility and robustness. Kinetic analysis was performed using a 1:1 Langmuir binding model to determine the association and dissociation rates.

Run Number	Analyst	KD (M)	kon (M−1 s−1)	koff (s−1)
1	A	5.61 × 10−8	1.85 × 104	1.04 × 10−3
2	B	5.98 × 10−8	2.09 × 104	1.25 × 10−3
3	A	4.66 × 10−8	2.15 × 104	1.00 × 10−3
4	B	3.33 × 10−8	1.96 × 104	6.52 × 10−4
Average		4.89 × 10−8	2.01 × 104	9.85 × 10−4
CV (%)		24.1	6.6	25.1

), with low inter-run variability in association (k_on) and dissociation (k_off) rate constants. The inter-assay variability was moderate (CV 24.1%), yet all K_D values remained within the expected nanomolar range, indicating consistent receptor-binding affinity across replicates. Control experiments confirmed the specificity of the BLI assay and surface-bound interaction. No binding of BMPR-IA was observed to unloaded Ni-NTA sensors, degraded rhBMP-2, or buffer from EP. Receptor interaction occurred exclusively in the presence of structurally intact, surface-immobilized rhBMP-2, confirming the specificity of both protein and surface-mediated binding (Scheme A1 and Scheme A2).

The repeatability of the binding method by a single analyst confirmed the consistency of the assay, with a K_D of 5.13 × 10⁻⁸ M and CV of 13% across two independent runs indicating minimal analytical drift or artifact (Figure 4 and

Table 3. Repeatability of binding method to assess binding affinity of BMPR-IA to released rhBMP-2 immobilized on the Ni-NTA biosensors. Results show intra-assay variability across technical replicates, demonstrating assay consistency.

Run Number	Analyst	KD (M)	kon (M−1 s−1)	koff (s−1)
1	A	5.61 × 10−8	1.85 × 104	1.04 × 10−3
3	A	4.66 × 10−8	2.15 × 104	1.00 × 10−3
Average		5.13 ×10−8	2.00 × 104	1.02 × 10−3
CV (%)		13.0	10.4	2.6

).

3.4. ALP Activity Induced by Released rhBMP-2 in Cell-Based Assay

ALP response was measured to evaluate the osteogenic potential of the released rhBMP-2 over a period of 40 days. The release of bioactive rhBMP-2 was determined by assaying the ALP activity in whole cell lysates of C2C12 cells exposed to LDGraft. A strong ALP response was observed at each time point throughout the assay period, indicating the bioactive potential of rhBMP-2 post-encapsulation (Figure 5). ALP response was highest at days 22–26 and 33–36, compared to the remaining time points. All values were normalized to the corresponding positive control to allow consistent comparison across experiments. Minimal ALP activity was detected in negative controls, confirming that the observed effect was specific to the presence of released rhBMP-2.

4. Discussion

The current study demonstrates that encapsulating rhBMP-2 in P_DLLGA (50:50) pellets via hot-melt extrusion and irradiation effectively preserves the molecular integrity and biological activity of the protein. Specifically, we demonstrated that the protein structure remained intact post-encapsulation, as shown by SDS-PAGE and CE-SDS analyses (Figure 1 and Figure 2). The accumulation of protein agglomerates, in addition to the expected structure of the protein, may been seen when they are processed, and care must be taken to avoid this [42,48,49]. The comparison with the non-released, non-irradiated reference protein confirmed that the encapsulation, sterilization, and release process did not markedly compromise the structural integrity of rhBMP-2, as evidenced by the preserved band pattern in SDS-PAGE. Analysis showed no evidence of high molecular weight aggregates or degradation products on SDS-PAGE, and all three batches displayed a predominant band corresponding to the expected dimeric form of rhBMP-2 (Figure 1). Although the theoretical molecular weight of the rhBMP-2 dimer is ~26 kDA, its migration at 15–25 kDA under non-reducing conditions is well documented and reflects the anomalous behavior of compact, disulfide-linked dimers rather than discrepancy in molecular mass [44,45]. Furthermore, CE-SDS analysis provided insights into the molecular stability of the released rhBMP-2 under long-term storage conditions. Throughout the storage period, CE-SDS analysis revealed only minor changes in the distribution of the LMW and HMW species. The proportion of the HMW species increased slightly, from 4.450% at T0 to 10.230% at T551, while the LMW species showed a moderate rise, from 6.300% at T0 to 8.805% at T732. These small variations (2.5 percentage points for the LMW, and 5.78 percentage points for the HMW) are within the expected analytical variability and indicate acceptable levels of aggregation and degradation over time. Overall, the consistent main peak (89.240% at T0 to 85.200% at T732) and stable electrophoretic profile across all time points demonstrate that the encapsulated rhBMP-2 retained its structural integrity and exhibited no evidence of progressive degradation during long-term storage (Figure 2 and Table 1). Additionally, the encapsulation polymer (P_DLLGA) demonstrated high stability, without any marked changes in molecular weight or polydispersity until 732 days of storage, measured by Gel Permeation Chromatography (See Appendix A,

Table A1. Stability analysis of P_DLLGA under controlled conditions (25 °C/60% RH). Average molecular weight (NLT 30 kDa) and polydispersity index were determined by gel permeation chromatography (GPC). Data from multiple long-term storage time points confirmed compliance with specifications.

Time Point (d)	Molecular Weight (kDA)	Polydispersity
0	47	1.579
129	47	1.572
227	47	1.615
551	47	1.517
732	50	1.610

). Taken together, these results demonstrate that the encapsulation and sterilization process preserved rhBMP-2 integrity, despite prior reports that proteins are susceptible to thermal or mechanical stresses [49,50,51,52,53], and to terminal sterilization [54,55,56], which can induce misfolding or aggregation, and ultimately compromise therapeutic efficacy.

Critically, the released rhBMP-2 retained its molecular function, as demonstrated by its receptor-binding affinity to BMPR-IA, following both processing and terminal sterilization, as confirmed through rigorous kinetic characterization using BLI. The receptor-binding assay focused on BMPR-IA (type I receptor), which mediates the initial high-affinity interaction with BMP-2, followed by the recruitment of the lower-affinity type II receptor (BMPR-II). BMPR-IA thus provides a sensitive and well-established model for assessing ligand integrity and receptor-binding competence, while BMPR-II’s transient interaction makes it less suitable for kinetic BLI analysis [57,58]. Bioactivity, defined here as functional receptor engagement with BMPR-IA, was confirmed using BLI, with K_D in the low nanomolar range (average K_D ~4.9 × 10⁻⁸ M) (Figure 3). This is comparable to the low single-digit K_D values reported in the literature [47,59,60], suggesting minimal loss of binding competence through encapsulation and terminal sterilization. Although the inter-assay coefficient of variation (CV) was moderate (24.1%), this level of variability is acceptable for surface-based ligand–receptor binding assays, such as BLI, where small differences in receptor immobilization density, ligand orientation, and mass transport can affect apparent kinetics. Importantly, all K_D values remained within the expected nanomolar range, demonstrating that the observed variation reflects methodological sensitivity rather than reduced protein stability. Overall, the BLI assay demonstrated consistent performance, confirming that the bioactivity of rhBMP-2 post-encapsulation is both reproducible and intrinsic to the released protein.

The preserved receptor-binding profile further supports that the tertiary structure required for effective BMPR engagement was maintained. In vitro, upon the binding of BMP-2 to BMPR type I, BMP-2 subsequently recruits type II receptors with lower affinity. The overall binding kinetics critically influence the strength and duration of downstream signaling events, including ALP induction [47,57,58]. To test this hypothesis, we have performed cell-based studies with released rhBMP-2 from our clinical-grade product, LDGraft. Results showed a strong ALP response for released rhBMP-2 at all time points (period of 40 d) confirming its osteogenic potential after encapsulation, and long-term storage and release (Figure 5). Since ALP induction requires active downstream signaling, BMP-2 must bind both type I (e.g., BMPR-IA) and type II (e.g., BMPR-II) receptors [46]. This implies that released rhBMP-2 from LDGraft is fully bioactive and facilitates the formation of a heteromeric receptor complex essential for downstream osteogenic signaling. The elevations in ALP response at days 22–26 and 33–36 may indicate a transient increase in rhBMP-2 release rate in vitro, potentially associated with the degradation and surface erosion phases of the LDGraft composite matrix (β-TCP/P_DLLGA), or influenced by serum proteins and residual enzymatic activity. However, this remains a hypothesis that would require further investigation to confirm. Although this study does not seek to establish a detailed release profile, it focuses on demonstrating that the released rhBMP-2 retains its osteogenic potential, as evidence by sustained ALP activity in vitro. As previously shown, β-TCP-based ceramics primarily act as osteoconductive scaffolds and do not independently initiate osteogenic differentiation in C2C12 cells. Therefore, the observed ALP activity confirms the bioactivity of released rhBMP-2 rather than the effect of the LDGraft itself [61].

Our findings provide valuable insights into enhancing the therapeutic delivery of rhBMP-2 in bone regeneration. Numerous reports have highlighted the drawbacks of the burst-release kinetics associated with surface-attached delivery materials, which result in supraphysiologic local concentrations of BMP-2 in the immediate post-implantation period [62]. The leakage and uncontrolled delivery of rhBMP-2 have been linked to pronounced inflammatory responses, ectopic ossification, and neurological complications, particularly in the cervical spine [63,64,65].

Our encapsulation approach within P_DLLGA offers a potential strategy to decouple rhBMP-2 bioavailability from the inflammatory phase, promoting a more sustained release aligned with the reparative phase of bone healing—a temporal window where endogenous BMP-2 signaling naturally peaks.

This study also extends prior work on polymeric encapsulation systems for rhBMP-2. Our approach uses P_DLLGA (50:50), a biodegradable polymer with a well-characterized safety profile, and incorporates formulation modifications such as cyclodextrin and poloxamer 407, which likely act as protein stabilizers during lyophilization, extrusion, and terminal sterilization. The terminal sterilization via X-ray irradiation, as opposed to ethylene oxide or gamma irradiation, appear not to compromise protein integrity. A key strength of this study is the integration of both structural and functional assays, which collectively support the long-term molecular stability and bioactivity of encapsulated rhBMP-2 under long-term storage conditions. This is a critical parameter when considering the translation into therapeutic products, where stability during long-term storage must be ensured without compromising efficacy and safety. The results of this study highlight the strong clinical translation potential of the LDGraft compared to surface-attached delivery systems, such as the collagen sponge carrier. While collagen sponges typically exhibit rapid, uncontrolled burst release, the encapsulation strategy presented here enables a more controlled and sustained release of rhBMP-2, potentially reducing the effective dose required and critically minimizing adverse events. These features make the delivery platform particularly attractive for clinical translation, as they address the key safety and efficiency challenges with currently approved rhBMP-2 delivery platforms. Nonetheless, this study has limitations. The in vitro nature of the receptor-binding assays, while informative, cannot fully predict the in vivo osteoinductive potential. While the molecular integrity, receptor-mediated activity, and downstream osteogenic potential of released rhBMP-2 were assessed under accelerated conditions, the relationship between their storage duration, release kinetics, and osteogenic potential are yet to be fully established. Quantitative release kinetics were not measured in this study, as the analytical design focused on confirming protein stability and bioactivity under accelerated conditions and long-term storage conditions, rather than establishing a detailed release profile. The absence of direct release data limits the ability to directly correlate release kinetics with biological performance. Future studies will aim to characterize release kinetics both in vitro and in vivo, and establish a clear link between release, activity, dose distribution, and efficacy. While P_DLLGA is a well-characterized, biocompatible, and FDA-approved polymer, its degradation produces acidic by-products that modulate the pro-inflammatory response of macrophages [66]; these effects should be further evaluated in future in vivo studies.

The in vitro and analytical assays presented here demonstrate that rhBMP-2 retains its integrity post-encapsulation; however, the in vivo performance needs to be elucidated. Future studies will aim to further validate the translational relevance of these findings in vivo, focusing on the therapeutic efficacy and safety of the release and dose-shielding system in preclinical animal models. In particular, the comparison between burst-release and controlled-release systems will be conducted to evaluate differences in bone regeneration, local inflammation, and systemic exposure. Comprehensive histological, molecular, and biochemical analyses will be undertaken to confirm the biological activity of the released growth factor and assess potential adverse effects such as ectopic bone formation and inflammatory reactions. Ultimately, in vivo studies will determine whether our rhBMP-2 encapsulation and dose-shielding approach translates into enhanced healing outcomes, reduced inflammatory responses, or lower effective doses required for bone regeneration. Further in vivo preclinical and clinical evaluations of the LDGraft have already commenced, marking an important step toward its translation into patient care. Given that some authors are affiliated with the manufacturer of the tested biomaterial, the study design and data interpretation were conducted independently to ensure transparency, scientific rigor, and objectivity.

5. Conclusions

This study established that rhBMP-2 encapsulated in P_DLLGA retains its structural integrity and biological activity following sterilization, extended storage, and its subsequent release. The clinical-grade product LDGraft further demonstrates preserved osteogenic potency despite long-term storage, underscoring the translational relevance of this delivery approach. These insights are highly relevant to therapeutic safety and efficacy, demonstrating that the protein remains stable and biologically active under the stringent conditions required for clinical-grade manufacturing. Our encapsulation approach offers a reliable dose-shielding strategy for rhBMP-2 that aligns with temporal sequence of bone regeneration, while mitigating the burst release associated with surface-attached carriers. Future studies will validate its therapeutic efficiency and safety, supporting its clinical application in bone repair and spine fusion therapies.