Advances in Bone-Targeting Drug Delivery: Emerging Strategies Using Adeno-Associated Virus

Bisphosphonates

Due to a high affinity to HA, bisphosphonates (BPs) are the most extensively utilized bone-targeting ligands for the treatment of skeletal diseases. They have been in use since the 1960s, and since then, many BP products have been developed. These molecules were initially developed as water softeners that could remove calcium from hard water. Nitrogen-containing BPs inhibit farnesyl pyrophosphate (FPPS) synthase in osteoclasts, effectively halting bone turnover. The structure of BPs is derived from a stabilized pyrophosphate (Fig. 2). Pyrophosphate naturally regulates bone mineralization through an unstable, rapidly cleaved anhydride bond. The carbon–phosphate part of BP is highly stable, allowing BPs to reside in bone up to 20 years. ⁸ The deprotonated hydroxyls in the two phosphates of a BP molecule are separated by approximately 2.9 to 3.1 Å, which is similar to a native spacing of oxygen atoms in HA. ⁹ The hydroxyl group that extends from the geminal BP carbon also interacts with HA, further enhancing its bone affinity. ⁹

OPEN IN VIEWER

Figure 2.

Chemical structures of bone-targeting molecules. In the field of ligand-based targeting, various compounds have been explored, including pyrophosphate (A), bisphosphonates (B–M), tetracycline (N), acidic peptides (O), (DSS)n (P), and novel recombinant adeno-associated viruses (rAAVs) (Q) specifically engineered for bone targeting with high affinity for hydroxyapatite (HA). An innovative approach involves the modification of the AAV9-VP2 capsid protein by grafting bone-homing peptide motifs, resulting in a bone-targeting capsid. This modification effectively reduces the distribution of rAAV9 to nonskeletal tissues, emphasizing its potential for targeted delivery to bone structures. GOI, gene of interest. DSS, tripeptide repeats of aspartate–serine–serine.

BPs have been widely used to treat cancer-associated skeletal lesions, nonbacterial osteomyelitis, osteoporosis, dental diseases, and other bone-related diseases. One study successfully completed a phase I clinical trial for delivering cytarabine to cancer-induced bone lesions using a BP-based carrier. The improved targeting resulted in a favorable safety profile, with reduced activity in over half of the treated bone lesions. ¹⁰ Another study developed a bortezomib-BP conjugate with a special boron linker designed to remain stable in the bloodstream to treat multiple myeloma. Compared with untargeted bortezomib, the BP-conjugated bortezomib with the molar equivalent dose exhibited an increase in the ability to reduce tumor burden in multiple myeloma and while reducing the occurrence of the neuropathy and thrombocytopenia associated with bortezomib. ¹¹ A significant effort was made to utilize a BP that would not prevent resorption through FPPS inhibition once the bortezomib was released. However, the pyrophosphate-like structure of BPs has been shown to inhibit mineralization without any interaction with osteoclasts. ¹² It remains unclear whether these nonactive BPs would be effective for targeting bone fractures, as the effect of the BP targeting moiety to inhibit mineralization can itself have undesirable effects.

BPs have also been employed to deliver radionuclides such as targeted radiotherapy of bone metastases. BPs enabled highly selective delivery of radionuclides to the metastases, minimizing soft tissue damage during radiotherapy. ^13,14 Other research has focused on developing nanoparticles and micelles with chemotherapeutics to target bone cancers. ^{15 –19} BPs have proven effective in improving the pharmacokinetics of and antibiotics in treating osteomyelitis. ^{20 –22} More relevant to fracture treatment is the use of BPs to target anabolic agents for osteoporosis. ²³ Young et al. has developed BP-based carriers that deliver prostaglandin E2 (PGE2) agonists through a releasable ester linkage for the treatment of osteoporosis. Upon cleavage and hydrolysis, the bone conjugate releases both an antiresorptive BP and an anabolic PGE2 agonist and effectively reverses ovariectomy-induced osteoporosis and improves the overall mechanical properties of the bone. ^24,25 The targeted combination therapy proved more effective than its constituents delivered simultaneously but not tethered together, highlighting the potential of bone-targeting approaches. ²³ Notably, identification of the appropriate linkage is crucial for the success of these conjugates. Other osteoporosis treatments have focused on localizing estrogens ²⁶ or anabolic agents to the bone. ^27,28 BPs have also been used to deliver anabolic agents to treat periodontal diseases. ^29,30 Despite limited success, BPs have been used to deliver proteins and cells to bone surfaces. ^31,32

While BPs are attractive tools to localize various compounds to HA, they have also been reported to cause several adverse effects, including osteonecrosis of the jaw (ONJ), atypical subtrochanteric fractures in the femur, and esophageal cancer, at relatively high frequencies. ^12,33 Although arguments can be made that concentrations of BPs for targeting are far lower than concentrations that induce side effects, the association with BP-related ONJ has hindered the full development of most BP-based drugs. ⁹ Current research in this field focuses on nonactive forms of BPs as targeting ligands or dual action targeting compounds where the skeletal activity of BP is desired. Sato et al. demonstrated that focal adhesion kinase (FAK) inhibition can replicate the effects of mechanical loading. ³⁴ This finding holds promise for the development of bone-targeting therapeutics for disuse osteoporosis utilizing nonactive BPs.

Tetracycline

Tetracycline is a broad-spectrum antibiotic that inhibits protein synthesis in bacteria. It was originally isolated from the bacterial genus Streptomyces and has been used as a therapeutic agent for decades. ^35,36 However, its strong affinity to calcium resulted in unexpected skeletal side effects, such as discoloring bones and teeth and preventing normal tooth development in children. Tetracycline should therefore be taken without milk to prevent reducing its bioavailability by binding to calcium ions. ³⁷ These same properties responsible for tetracycline’s skeletal side effects also enable its use as a biomarker for dynamic histomorphometry measurements in bone, as it displays intrinsic fluorescence emission at 390 nM. The β-diketone system at positions 1 and 2, the enol system at positions 4 and 6, and the carboxamide group at position 5 are responsible for its chelating behavior. ⁹ Moreover, since the chelation of tetracycline is irreversible, the undesirable side effects are also permanent. ³⁸

The ability of tetracycline to bind and label primarily growing bone is useful to localize anabolic agents to bone surface by conjugating them to tetracycline and tetracycline-derived analogs. Moreover, its oral bioavailability makes it an attractive targeting ligand. However, its toxicity and skeletal side effects have prevented it from being fully developed as a bone-targeting strategy. Tetracyclines have been used to create bone-targeted poly(lactic-co-glycolic) acid (PLGA) nanoparticle systems to deliver the drug simvastatin to osteoporotic bone with modest success. ³⁹ However, given that tetracyclines decrease collagenase expression and osteoblast activity, a previous study attempted to modify the structure of tetracycline to reduce its toxicity. ³⁷ For example, Neale et al. demonstrated that 50% of the full tetracycline bone-binding ability can be retained after the modification of one of the rings in estradiol, improving its safety profile for the treatment of osteoporosis. ⁴⁰ Tetracycline preferentially binds to areas of low crystallinity on bone-forming surface rather than those of high crystallinity on bone-resorption surface. ⁴¹ The osteoblastic toxicities associated with tetracyclines and their strong affinity to newly formed bone make them better targeting ligands for osteosarcoma and bone metastases than for osteoporosis. ⁴⁰

Acidic oligopeptides

Nature harnesses the potential of acidic oligopeptides to facilitate the protein localization to bone. Prominent examples such as bone sialoprotein and osteopontin possess amino acid sequences exhibiting remarkably high acidity, including up to 22% of glutamic acid residues. ⁴² These proteins play pivotal roles in protein localization to bone structures, instigating HA nucleation, and facilitating collagen mineralization. ⁴³ The carboxylic acid side chains within these amino acids are presumed to chelate the calcium component of bone, thereby enhancing protein’s affinity to HA. ^9,44 As the number of repeating units of aspartic acid or glutamic acid increases, affinity toward bone typically increases, plateauing at around 6–8 repeated units. ⁴⁵ Nevertheless, longer chains demonstrate augmented compound delivery and in vivo retention. ^46,47

The interaction between acidic oligopeptides and HA manifests as a nonchiral interaction, where both L and D enantiomers of amino acids exhibit similar affinities to HA. However, in vivo observations suggest that the D enantiomer degrades at a slower rate, resulting in greater accumulation. ^{45,47 –49} Acidic oligopeptides stand as promising targeting ligands for bone, given their minimal toxicity compared with BPs and tetracyclines. ¹⁷ They also exhibit shorter half-lives than tetracyclines and BPs. ^33,38,49 Moreover, acidic oligopeptides offer tunability, facilitating the localization of nanoparticles using as few as three aspartic acid molecules or longer polymers to achieve the desired half-life. ⁵⁰ Their versatile application extends to targeting therapies for various bone-related diseases. For example, the clinically approved agent asfotase alfa relies on a chain of 10 aspartic acids (D₁₀) to target tissue-nonspecific alkaline phosphatase (TNALP) to bone for hypophosphatasia treatment. In preclinical contexts, they have shown potential to deliver antibiotics to osteomyelitis lesions, ⁵¹ liposomes, and micelles containing chemotherapy agents to osteosarcoma tumors and bone metastasis lesions, ^52,53 and radionucleotides to bone tumors for imaging and photothermal treatment. ^47,54,55 Broader applications to target bone infections and osteolytic lesions in cancer have also been suggested. Finally, the ability of acidic oligopeptides to deliver EP1 agonists to bone ⁴¹ or small compounds to bone fracture sites improves the therapeutic efficacy and safety of bone anabolic agents. ⁵⁶ For instance, subcutaneous administration of dasatinib-conjugated acidic oligopeptide showed significant tolerability and accelerated recovery of fractured femurs by over 50%. ⁵⁷ However, despite their promising safety profiles and suitable half-lives for acute skeletal damage, acidic oligopeptide requires further investigation to optimize enantiomeric configurations and lengths for effective bone fracture repair pharmacokinetics.

Tripeptide repeats of aspartate–serine–serine

Tripeptide repeats of aspartate–serine–serine (DSS)_n have been originally identified as a HA-binding motif of dentin phosphoprotein (DPP). DPP, a major noncollagenous protein component of the dentin extracellular matrix, plays a pivotal role in HA nucleation during dentin mineralization. Its HA-binding region harbors a plethora of DSS repeats, which are extensively phosphorylated. Notably, unphosphorylated DSS repeats can still localize to HA, indicating that phosphorylation is not essential for binding. Moreover, replacing the serines in the DSS repeats with alanine did not affect binding affinity. Previous studies demonstrated that increasing the number of DSS repeats proportionally enhances targeting capacity up to approximately six repeats. ⁵⁸ DSS peptide exhibits a preferential affinity to bone-forming surfaces over bone resorption pits characterized by low crystallinity HA. ⁵⁹ (DSS)₆ has been shown to favorably bind to mantle dentin, comprising small and randomly oriented crystals, rather than the enamel surface, which consists of elongated and well-oriented HA crystals. These properties make DSS an attractive targeting ligand due to its ability to target newly formed bone.

DSS has emerged as a promising delivery platform for osteoporosis treatment. Saidak et al. demonstrated that (DSS)₆ effectively delivers an osteogenic cyclic peptide ligand for α5β1 integrin to the bone. Priming this integrin stimulates the differentiation of osteolineage progenitors into osteoblasts. The targeted delivery of this ligand restricts its effects to the desired regions and enhances long bone mass and microarchitecture. ⁶⁰ (DSS)₆ has also proven effective for the delivery of siRNA-encapsulating liposomes to bone-forming surfaces by dioleoyl trimethylammonium propane (DOTAP)-based cationic liposomes. ⁶¹ This system holds the potential to expand the use of siRNA to treat bone diseases. While DSS₆ may not possess as high affinity to HA as BPs and can exhibit relatively high liver and renal uptake, it offers the advantages of biocompatibility, low toxicity, and a preference for bone-forming surfaces.

Identification of AAV serotype transducing bone cells

The initial reports of AAV serotype tests for bone targeting were designated against bone fracture or bone loss using AAV2, AAV-DJ, and AAV8 ^{84 –86} listed in Table 1. Lee et al. tested a panel of 18 AAV variants (AAV1, AAV2, AAV3, AAV4, AAV5, AAV6, AAV8, AAV9, AAVrh10, AAV-DJ, AAV-DJ8, AAV2i8, AV-7m8, AAV-LK01, AAV-LK02, AAV-LK03, AAV-LK19, and AAV-Anc80), encompassing both natural and engineered variants, which show tdTomato (Red fluorescent protein, RFP) expression following cre-mediated recombination by Cre-loxP system in bone tissue. ⁸⁷ Subsequently, rAAV vectors were designed to restrict Cre expression to osteoblast-lineage cells using the Sp7 and Col2.3 promoters (rAAV-Sp7-Cre or rAAV-Col2.3-Cre). When intravenously administrated, AAV8 harboring the CAG promoter drove tdTomato expression in bone, liver, heart, spleen, and kidney. On the other hand, Yang et al. performed in vitro screening of 14 distinct AAV capsids (AAV1, AAV2, AAV3, AAV4, AAV5, AAV6, AAV6.2, AAV7, AAV8, AAV9, AAVrh.8, AAVrh.10, AAVrh.39, and AAVrh.43) using the enhanced green fluorescent protein (eGFP) reporter gene (rAAV-eGFP) to identify AAV variants that can transduce skeletal cells. ⁶⁹ In this screening, eight AAV capsids (rAAV1, rAAV4, rAAV5, rAAV6, rAAV7, rAAV9, rAAVrh.10, and rAAVrh.39) exhibited the ability to transduce osteoblasts. Additionally, rAAV1, rAAV4, rAAV5, rAAV6, rAAV7, and rAAV9 vectors demonstrated transduction of osteoclasts, while rAAV1, rAAV6, rAAVrh.10, and rAAVrh.39 vectors transduced chondrocytes. When intravenously administered into mice, only rAAV9 was able to transduce osteoblasts and osteoclasts on the bone surface and osteocytes embedded within the bone matrix. As previously reported, ^87,88 systemically delivered rAAV9 also transduced heart, muscle, and liver while little to no transduction of lung, kidney, or spleen. Notably, rAAV-eGFP vectors were not effective in transducing articular cartilage when administered intra-articularly (i.a.) into mouse knee joints. This discrepancy is potentially attributable to the limited accessibility of rAAV vectors to chondrocytes, which are embedded within the avascular microenvironment of these structures. In this study, among 14 conventional AAV variants, rAAV9 was most effective in targeting osteoblast lineage cells and osteoclasts in mice. ⁸⁹

Capsid modifications for bone targeting

While rAAV2, rAAV8, or rAAV9 demonstrated promise in bone-targeting delivery, their non-specific transduction of nonskeletal tissues poses a challenge due to the potential for undesirable off-target effects. Inspired by previous studies demonstrating high affinity of the bone-targeting peptide motif to osteoclast-enriched bone-resorption surfaces, an aspartic acid octapeptide (D8) was fused to the capsid to increase vector affinity to HA, ⁹⁰ Alméciga-Díaz et al. demonstrated that grafting of aspartic acid octapeptide (D8) insertion to the N-terminus of the rAAV2-VP2 capsid protein increased transgene delivery and expression in the bone. On the other hand, Yang et al. applied a rational design approach to incorporate DSS-encoding DNA sequences into the N-terminus of rAAV9-VP2 capsid (AAV9.DSS-Nter, Fig. 2Q). ^91,92 rAAV9.DSS-Nter capsid exhibited robust transduction of osteoblasts and osteoclasts in vitro and were as effective as the rAAV9 capsid in transducing osteoblasts, osteoclasts, and osteocytes in mouse femurs. ⁸⁹ However, compared with the rAAV9 capsid, rAAV9.DSS-Nter capsid exhibited significantly less transduction of the liver (∼50% less) and muscle (∼30% less) with minimal transduction of heart. An (Asp)₁₄ that preferentially binds osteoclast-enriched bone-resorbing surfaces ⁴⁷ was also grafted to the N-terminus of rAAV9-VP2 capsid (AAV9.D14-Nter). Both AAV9.DSS-Nter and AAV9.D14-Nter capsids increased HA-binding affinity without affecting the ability of an rAAV9 vector to transduce osteoclasts and osteoblasts in vitro. In contrast to rAAV9.DSS-Nter capsid, systemic delivery of the AAV9.D14-Nter capsid to mice did not confer bone-homing specificity to rAAV9. These results suggest that incorporation of the DSS6-VP2 capsid protein, but not the D14-VP2 capsid protein, significantly enhanced the bone-homing specificity of rAAV9 vectors by reducing their ability to transduce nonrelevant tissues as an example of a detargeting strategy. ⁸⁹ This modification presents a promising strategy for improving the specificity and safety of rAAV9-mediated gene therapy applications, particularly in the context of bone-related disorders.

AAV-Mediated gene therapy to treat skeletal diseases

Expression of the N-acetylglucosamine-1-phosphate transferase alpha and beta subunits (GNPTAB) using AAV8-mediated delivery has shown therapeutic effects in a mucolipidosis II (ML-II) model established by the deletion of GNPTAB exons12—20. ML-II is a severe systemic genetic disorder caused by a dysfunctional mannose 6-phosphate recognition system. This impairment disrupts the proper targeting of multiple lysosomal hydrolases, consequently leading to the accumulation of nondegraded materials within lysosomes. ML-II mice display hip dysplasia, scoliosis, rickets, and osteogenesis imperfecta (OI). Systemic delivery of AAV8-expressing GNPTAB alleviated bone loss in ML-II mice by reducing IL-6 production. Mucopolysaccharidosis IVA (MPS-IVA), also known as Morquio A disease, is an autosomal recessive disorder characterized by a deficiency of the lysosomal enzyme N-acetylgalactosamine-6-sulfatase (GALNS). This results in the systemic accumulation of the glycosaminoglycans (GAGs) keratan sulfate and chondroitin-6-sulfate. Skeletal manifestations, such as short stature and skeletal dysplasia, are the primary clinical features, driving the therapeutic focus. Systemic administration of the AAV2-expressing GALNS with bone-targeted capsid modification by acidic oligopeptides (D8) increased GALNS enzyme activity in the skeleton of GALNS-deficient mice. However, long-term studies to evaluate pathological and clinical improvement in skeletal phenotypes have yet to be performed for this strategy. Fibrodysplasia ossificans progressiva (FOP) is a rare genetic disorder characterized by progressive extraskeletal bone formation. ⁹³ AAV gene therapy is an attractive therapeutic approach for FOP in that approximately 97% of FOP patients harbor a recurrent, heterozygous missense ACVR1^R206H mutation (c.617G>A;p.R206H). Treatment with an rAAV9 carrying an ACVR1^R206H -specific silencer and codon-optimized human ACVR1 reduced aberrant activation of BMP-Smad1/5 signaling and the chondrogenic/osteogenic differentiation of Acvr1^R206H skeletal progenitors. Moreover, systemic delivery of this same AAV vector prevented spontaneous heterotopic ossification (HO) in FOP mice while trauma-induced HO was also decreased when this vector was administered transdermally to injured muscle. ⁹⁴ This AAV vector was further advanced by a subsequent strategy to simultaneously suppress Activin A and its receptor ACVR1^R206H while limiting activity in nonskeletal organ, such as liver, through the silencing of transgene expression using liver-expressing miRNA (miR-122)-targeting sequence. ⁹⁵

In addition to rare skeletal diseases, AAV-mediated gene therapy utilized to express human COX2 carried by the AAV-DJ capsid to promote the healing of bone fracture through local administration. ⁸⁴ Using the rAAV9.DSS-Nter capsid, Yang et al. employed AAV-mediated silencing of the bone-forming suppressor Schnurri-3 [SHN3, human immunodeficiency virus type I enhancer-binding protein 3 (HIVEP3)] in osteoblasts to reverse bone loss and enhance bone quality in ovariectomized osteoporotic mice. ⁸⁹ SHN3 is a large zinc finger protein and its deletion results in a progressive increase in bone mass due to augmented osteoblast function. ^{96 –99} Notably, as rAAV9 is also capable of transducing osteoclasts, rAAV9-mediated silencing of osteoclast forming/activating factors such as receptor activator of nuclear factor kappa B (rank) or cathepsin K in osteoclasts resulted in increased bone mass by inhibiting osteoclast differentiation or activity. ¹⁰⁰ These findings highlight that rAAV9 can mediate suppression of osteoclast-mediated bone resorption and promotion of osteoblast-mediated bone formation simultaneously, thereby counteracting bone loss and enhancing bone quality in osteoporosis. In a report from Oh et al., single or dual silencing of WNT antagonists, Sclerostin (Sost) and/or Shn3, through a rAAV9.DSS-Nter-mediated delivery effectively enhanced WNT signaling, osteoblast function, and bone formation. ¹⁰¹ Systemic administration of these rAAVs not only reversed bone loss in both postmenopausal and senile osteoporosis models, but also promoted healing of bone fractures. Additionally, transplantation of AAV-bound allograft bone to osteotomy sites significantly improved critical-sized bone defect repair. ¹⁰¹ Bhat et al. has reported that AAV9-mediated modulation of natural osteoblast and osteoclast-regulating miRNAs, miR-214-3p or miR-39a-5p, effectively reversed bone loss in mouse models of postmenopausal osteoporosis and aging-associated osteoporosis. ¹⁰² Finally, rheumatoid arthritis (RA) is a chronic inflammatory disease that causes systemic and articular bone loss by activating bone resorption and suppressing bone formation. ⁹¹ Despite current therapies, inflammation-induced bone loss in RA remains a significant clinical challenge due to joint deformity and impaired articular and systemic bone repair. Stavre et al. has identified SHN3 as a potential target to prevent bone loss in RA. Silencing of Shn3 expression in RA mouse models using systemic delivery of a rAAV9.DSS-Nter effectively limited articular bone erosion and systemic bone loss. ¹⁰³ These findings underline the transformative potential of AAV-mediated gene therapy to address a wide range of bone disorders, including osteoporosis, bone fractures, critical-sized bone defects, and RA. AAV-mediated gene therapy can provide a solid foundation for further development and clinical translation of these therapeutic strategies.