Poster N15 - Shelf-Stable Lattice-Reinforced Affibody-Conjugated Hydrogels for Tunable Protein Delivery In Vivo

Introduction: Bone morphogenetic protein-2 (BMP-2) is a potent osteogenic protein used clinically as a bone graft substitute. Current delivery approaches have a limited ability to localize BMP-2 within bone injuries, resulting in off-target effects. Recently, we identified and characterized BMP-2-specific affinity-binding proteins called affibodies and conjugated them into hydrogels to tune BMP-2 release. BMP-2 release rates were slowed by stronger affibody-BMP-2 binding interactions. The affibodies were predicted to bind to BMP-2 at cell receptor sites, which was supported by their inhibition of BMP-2-induced alkaline phosphatase activity in vitro.
With these promising initial results, the goal of this work is to develop affibody-conjugated hydrogels for tunable BMP-2 release in vivo. We identified three barriers to in vivo application: 1) the affibodies were synthesized in E. coli, resulting in endotoxin contamination; 2) the hydrogels had an inconsistent shape and lacked the mechanical integrity necessary to be handled using surgical tools and remain stable under physiological mechanical stresses; and 3) the affibody-conjugated hydrogels needed to be prepared fresh for each experiment, which is technically complex and not clinically feasible.
To address these challenges, affibodies were prepared using solid-phase peptide synthesis to reduce endotoxin contamination, the hydrogels were reinforced using melt electrowritten (MEW) polycaprolactone diamond lattice scaffolds, and the reinforced hydrogels were lyophilized for long-term storage. The effects of each modification on protein release and hydrogel physicochemical properties were measured and compared to the original formulation. BMP-2 retention and mineralization of reinforced synthetic affibody-conjugated hydrogels were measured in vivo in a subcutaneous rodent implant model.

Materials and

Methods: High- and low-affinity BMP-2-specific affibodies were synthesized using Fmoc solid-phase peptide synthesis (“synthetic”) or recombinantly expressed in BL21(DE3) E. coli (“recombinant”). The affibodies were characterized using circular dichroism, matrix-assisted laser desorption and ionization mass spectrometry (MALDI), and biolayer interferometry (BLI) to characterize secondary structure, size, and affibody-BMP-2 binding interactions, respectively. Endotoxins were quantified by chromogenic LAL assay.
4-arm poly(ethylene glycol) maleimide (PEG-Mal) was mixed with high- or low-affinity affibodies and dithiothreitol in phosphate-buffered saline (PBS) to form 20 µM affibody-conjugated PEG-Mal hydrogels. 100 ng of BMP-2 was loaded into synthetic and recombinant affibody-conjugated hydrogels, and protein encapsulation and release were quantified over two weeks using enzyme-linked immunosorbent assays (ELISA). The hydrogels were weighed in relaxed, swollen, and lyophilized states to quantify mesh size.
Hydrogels were reinforced by crosslinking inside 5mm diameter polycaprolactone (PCL) cylindrical scaffolds with a 0.5mm2 diamond pore design printed via high-precision melt electrowriting. The hydrogels were compressed to assess the effects of scaffold reinforcement on mechanical properties, and protein encapsulation and release were quantified as described.
Hydrogels with/without affibodies were synthesized, lyophilized, and stored at 4°C for one week. Hydrogels were rehydrated with 100 ng of BMP-2 in 120 µL of PBS, and their encapsulation and release were quantified by ELISA.
BMP-2 was loaded into the MEW-reinforced affibody-conjugated hydrogels and implanted subcutaneously in rats. Fluorescently labeled BMP-2 retention was measured using an in vivo imaging system (IVIS) over three weeks, and mineralization will be measured using micro-computed tomography (micro-CT) and x-ray radiography after 50 days.

Results, Conclusions, and Discussions: MALDI revealed a single peak associated with the expected masses of each affibody (~6.6 kDa for synthetic, and ~7.4 kDa for recombinant). There were no differences in BMP-2-affibody binding interactions between synthetic and recombinant affibodies, although there was a difference between the high- and low-affinity affibodies, as expected. Affibodies displayed characteristic alpha-helical spectra (troughs at 222 and 208 nm, and a peak at 193 nm), with synthetic affibodies depicting tighter folding attributed to fewer endotoxins. Indeed, synthetic affibodies had fewer endotoxins than the FDA limit for medical devices, whereas recombinant affibodies exceeded this limit. There was a difference in BMP-2 release from affibody-conjugated hydrogels when comparing the affibodies of different affinities but not affibodies of different sources.
Hydrogel mesh size was not affected by the inclusion of affibodies. Hydrogels spontaneously crosslinked and filled the void of the MEW scaffold. The compressive moduli of the reinforced and non-reinforced hydrogels were comparable, suggesting the scaffold did not affect the crosslinking or mesh size of the hydrogels. However, the reinforced hydrogels had a higher stress at 80% compression (3.2 ± 0.17 kPa vs 0.86 ± 0.08 kPa) and could withstand greater compression before yielding. Although cumulative BMP-2 release was higher from the lyophilized hydrogels compared to the non-lyophilized hydrogels, release rates measured by Fickian diffusion slopes were similar.
After 3 weeks, affibody-conjugated hydrogels retained more fluorescent BMP-2 than hydrogels with low-affinity affibodies and without affibodies. We plan to quantify subcutaneous ectopic mineralization in these formulations from ongoing studies.
Reinforcement, switching to synthetic affibodies, and lyophilization didn't affect the overall BMP-2 release, increasing the clinical translatability of the system.
Future work will involve independently tuning the release of multiple proteins from dual affibody-conjugated hydrogels and assessing the effects of protein co-delivery using BMP-2 and the immunomodulatory protein interleukin-4 in vivo in a femoral rat defect model. Taken together, these results demonstrate significant progress towards the development of a shelf-stable and clinically-relevant bone graft substitute that may more effectively mimic the presentation of proteins in the healing cascade by using affibodies, withstand physiologically relevant stresses experienced in the bone injury environment, and minimize endotoxin contamination.

Acknowledgements (Optional):