Bone-targeted Nanoparticle Drug Delivery System-mediated Macrophage Modulation for Enhanced Fracture Healing

Abstract

Despite decades of progress, developing minimally invasive bone-specific drug delivery systems (DDS) to improve fracture healing remains a significant clinical challenge. To address this critical therapeutic need, we previously developed nanoparticle (NP) DDS comprised of poly(styrene-alt-maleic anhydride)-b-poly(styrene) (PSMA-b-PS) functionalized with a peptide that targets tartrate-resistant acid phosphatase (TRAP) and achieves preferential fracture accumulation. The delivery of AR28, a glycogen synthase kinase-3 beta (GSK3β) inhibitor, via the TRAP binding peptide-NP (TBP-NP) expedites fracture healing. Upon further investigation, our data reveals that NPs are predominantly taken up by fracture-associated macrophages. Together with the evolving role of macrophages in fracture healing, the underlying mechanism of healing via TBP-NP was comprehensively investigated herein. TBP-NP_AR28 promotes M2 macrophage polarization and enhances osteogenesis in preosteoblast-macrophage co-cultures in vitro. Longitudinal analysis of TBP-NP_AR28-mediated fracture healing revealed distinct spatial distributions of M2 macrophages, an increased M2/M1 ratio, and upregulation of anti-inflammatory and downregulated pro-inflammatory genes compared to controls. Taken together, this work demonstrates the underlying therapeutic mechanism of bone-targeted NP DDS, which leverages macrophages as druggable targets and modulates M2 macrophage polarization to enhance fracture healing, highlighting the therapeutic benefit of this approach for fractures and bone-associated diseases.

Graphical Abstract

This study provides a comprehensive investigation of the impact and underlying mechanism of nanoparticle drug delivery system on macrophage-mediated fracture healing using a bone-targeting nanoparticle platform. M2/M1 ratio was found to govern healing outcomes and this ratio was upregulated by promoting M2 macrophage polarization and inhibiting M1 macrophage modulation. This highlights the therapeutic benefit of this approach for fractures and bone-associated diseases.

INTRODUCTION

Fractures are a significant clinical challenge, with more than 8.9 million fractures occurring annually, or a fracture every 3 seconds.^1-5 Despite major advances in surgical procedures, such as bone realignment and immobilization, a high percentage (~10%) of fractures, known as nonunion fractures, do not heal. Healthcare costs associated with nonunion fractures are ~ $9.2 billion/year in the US. By 2050, these costs are expected to double.^6-8 Therefore, novel, minimally invasive therapeutic approaches are urgently needed to prevent and/or treat debilitating and costly nonunion fractures. The field of nanotechnology, specifically nanoparticle (NP)-based drug delivery systems (DDS) in bone regeneration, has advanced significantly in recent years, due to DDS-mediated improvements in drug solubility, stability, bioavailability, and efficiency. In addition, NP-based DDS can control drug release kinetics and enable functionalization with ligands for tissue-specific delivery, ultimately increasing therapeutic efficacy.⁹

To this end, we previously established a bone-targeted NP DDS, where active bone targeting is achieved by incorporation of tartrate-resistant acid phosphatase (TRAP) binding peptide (TBP) to poly(styrene-alt-maleic anhydride)-b-poly(styrene) (PSMA-b-PS) to form TBP-NP.^10-18 We showed ~3-fold greater fracture accumulation of TBP-NP versus untargeted NP, and by entrapping AR28, a glycogen synthase kinase-3 beta (GSK3β) inhibitor, fracture healing was enhanced versus untreated controls. Initially, the overarching conclusion of this work was that AR28 release at the fracture site upregulated Wnt-β-catenin signaling and enhanced osteogenesis of osteoprogenitor cell types, such as mesenchymal stem cells. However, our recent work indicates the impact of anionic NP, including PSMA-b-PS NP used here, on macrophage polarization, motivating the reevaluation of the mechanism of healing via TBP-NP_AR28.¹⁹

There is growing recognition of the critical role of immune cells, especially macrophages, in fracture healing, as they are one of the first immune cells to respond to fractures, infiltrate the hematoma during bone healing, and remain active throughout the healing process. Macrophages have been shown to regulate the differentiation and function of osteoblasts and osteoclasts.^20-27 Specifically, macrophages with various phenotypes, such as classically activated (M1) and alternatively activated (M2) macrophages, exert differential effects in the healing process. Delayed fracture healing or even failed healing is linked to pathological upregulation of M1 macrophages, which secrete pro-inflammatory cytokines, and downregulation or less effective M2 macrophage-mediated anti-inflammatory response. Therefore, promoting M2 macrophage polarization and/or inhibiting M1 macrophage differentiation may enhance fracture healing.

Despite efforts in developing NP-based DDS, most NPs administered systemically are inevitably taken up and eliminated by the mononuclear phagocyte system (MPS),²⁸ one of the significant barriers to translating DDS into effective therapies. Therefore, macrophages, as a crucial component of MPS, may provide an interesting and relevant druggable target to enhance DDS delivery, which may benefit fracture healing. However, few studies have leveraged macrophage targeting for enhanced bone regeneration via systemic delivery. Gold, silica, and silica-iron composite NPs induced M2 macrophage polarization and improved osteogenesis^{21,22, 23} but these studies did not include active bone targeting, which may exacerbate off-target effects and toxicity.^{29, 30} Furthermore, the underlying mechanism of macrophage-associated osteoimmunomodulation via NP DDS is inconclusive. Altogether, the growing recognition of the role of macrophages in fracture healing, and the urgent need for developing systemic bone-targeted DDS motivated investigation of the underlying mechanism of healing via TBP-NP delivery of AR28.

This work comprehensively investigated the impact of TBP-NP_AR28 on macrophage modulation and fracture healing. First, cell types that take up TBP-NPs at the fracture site were evaluated via flow cytometry. Mouse femur fracture healing outcomes of unloaded TBP-NP and TBP-NP_AR28 were verified by microcomputed tomography (μCT) analysis and biomechanical testing. The impact of AR28, TBP-NP, and TBP-NP_AR28 on in vitro macrophage polarization was measured via gene expression and cell surface markers to explore the underlying mechanism of NP-mediated fracture healing. In vitro osteogenic differentiation was then studied in co-cultures of macrophages and preosteoblasts, as assessed by alkaline phosphatase and alizarin red staining. Histology, flow cytometry, and bulk RNA sequencing (RNA-seq) were performed to further investigate NP-mediated macrophage polarization on fracture healing by characterizing the spatial distribution, M2/M1 ratios, and gene expression. This study sought to understand the effect of bone-targeted NP DDS-mediated macrophage polarization on fracture healing and its underlying mechanism.

RESULTS AND DISCUSSION

Previously, we showed that introducing a targeting peptide for TRAP resulted in ~3-fold greater fracture accumulation of targeted versus untargeted NP. The drug, AR28, a GSK3β inhibitor, entrapped within TBP -functionalized PSMA-b-PS NP or TBP-NP, resulted in expedited healing versus untreated controls.¹⁶ Previous data suggested that the underlying healing mechanism was enhanced osteogenesis due to upregulated Wnt-β-catenin signaling in regenerative cells, such as mesenchymal stem cells. However, our recent work implicating M1-to-M2 macrophage polarization of anionic PSMA-b-PS NPs used here¹⁹ combined with the growing understanding of the role of macrophages in fracture healing, motivated us to revisit underlying healing mechanism of TBP-NP delivery of AR28.

Fracture healing was first evaluated to investigate the impact of TBP-NP compared with TBP-NP_AR28 on bone regeneration. TBP-NP were prepared from diblock copolymers, PSMA-b-PS, which were synthesized using a 1-step reversible addition-fragmentation chain transfer (RAFT) polymerization followed by conjugated of the bone-targeting peptide, TBP, via anhydride ring opening (Schematic 1).¹⁰ Self-assembled TBP-NPs were formed from TBP-functionalized PSMA-PS polymers by gradually adding water to dimethylformamide (DMF) solvated polymers, during which AR28 was loaded into TBP-NP to form TBP-NP_AR28. Drug loading and release kinetics are shown in Figure S1, respectively. Drug release was analyzed at pH 4.5, consistent with the pH experienced within endolysosomal compartments, to understand drug release during intracellular trafficking.^{16, 17} The characterization of NPs and polymers were listed in Table 1, Table 2, and Figure S2, respectively.

Schematic 1.

A) Reversible addition–fragmentation chain transfer (RAFT) synthesis of poly(styrene-alt-maleic anhydride)-b-poly(styrene) (PSMA-b-PS) using 4-cyano-4-dodecylsulfanyltrithiocarbonyl sulfanyl pentanoic acid (DCT) as the RAFT Chain Transfer Agent; B) Functionalization of PSMA-b-PS with peptide (TBP-Alloc) targeting moieties via MA ring opening; C) Removal of Alloc protecting group from Alloc-protected Lysine; Self-Assembly of peptide-conjugated PSMA-b-PS diblock copolymers into micelle NPs (D) or drug-loaded NP (E) via solvent exchange.

Table 1.

PSMA-b-PS diblock copolymer characterization

Name	Block 1 (PSMA)	Block 2 (PS)	Diblock Copolymer		Peptide-conjugated diblock Copolymer
	Mn (kDa)	Mn (kDa)	Mn (kDa)	PDI	Peptides/polymer chain	Mn (kDa)
TBP-NP	21	24	45	1.2	27 ± 3	79
TBP-NPAR28	21	24	45	1.0	27 ± 3	79

Characterization polymer via NMR (proton nuclear magnetic resonance spectroscopy), GPC (gel permeation chromatography for molecular weights (Mn) and polydispersity index (PDI)). Values are mean ± Standard Deviation from three independent experiments.

Table 2.

Characterization of Nanoparticle (NP) Physicochemical Properties

NP	Peptides/NP	NP size (nm)	PDI	Zeta Potential (mV)*	Drug Loading
					Efficiency	Capacity
TBP-NP	80 x 103	49 ± 5	0.17	−27 ± 2	N/A	N/A
TBP-NPAR28	80 x 103	67 ± 7	0.30	−24 ± 1	99%	25%

NP characterization via DLS using a Malvern Zetasizer Nano ZS, Nanosight, high-performance liquid chromatography (HPLC) for peptides/NP, size, polydispersity index (PDI)), charge, and loading properties. Values are mean ± Standard Deviation from three independent experiments.

A murine femur fracture model was used to investigate fracture healing. Three days post-fracture, mice were treated with saline, free AR28, TBP-NP, and TBP-NP_AR28. Treatment was introduced at day 3 due to the inflammatory timeline of fracture healing. Three days post-fracture marks a pivotal juncture wherein the inflammatory response begins to abate.^{20, 31, 32} Concurrently, the regenerative stage takes over, which is vital for attenuating inflammation. Successful transition is crucial for bone repair.^{20, 31, 32} Bone healing was examined via microcomputed tomography (μCT) to evaluate mineralization 4 weeks after treatment. Femur diaphyseal fractures treated with TBP-NP_AR28 exhibited a more mature anatomical shape of cortical bone with better union and more organized woven bone at the bridging callus than other groups (Figure 1A). Quantitatively, TBP-NP_AR28 showed 1.8-, 1.5-, and 1.3-fold increased bone volume compared to saline, free AR28, and TBP-NP, respectively, as well as 1.9-, 1.8-, and 1.4-fold increased bone volume fraction (BV/TV) of fractured femurs, suggesting significantly enhanced bone mineralization (Figure 1B). TBP-NP_AR28 also showed increased trabecular number (Tb-N) and decreased trabecular separation (Tb-Sp), compared with saline controls, further indicating improved bone regeneration (Figure S3). Biomechanical analyses of fractured femurs, as the ultimate outcome of bone healing, were evaluated via torsion testing at week 4. Fractured femurs treated with TBP-NP_AR28 exhibited a significant increase in maximum torque, yield torque, and torsional rigidity compared with saline, free AR28, and TBP-NP. The observed increase in bone volume and strength suggested that TBP-NP_AR28 improved bone healing, which aligns with prior findings demonstrating the osteogenic potential of TBP-NP_AR28.

Figure 1. TBP-NP_AR28 treatments expedite fracture healing.

Fractured mouse femurs were scanned for microcomputed tomography (μCT) at 4 weeks post-treatment to evaluate in vivo bone callus formation. Quantification of μCT demonstrated that TBP-NP_AR28 significantly increased both bone volume (B(i)) and bone volume fraction (B(ii)). Scale bar = 1 mm. Biomechanical strength of fractured femurs was also assessed 4 weeks after fractures, where TBP-NP_AR28 significantly enhanced maximum torque (C), yield torque (D), and torsional rigidity (E) of the fractured femurs compared with all the other groups. In addition, TBP-NP also exhibited improved healing via increasing bone volume and torsional rigidity, compared with saline. Data represented as mean ± standard deviation, n = 6. Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparison test. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

The underlying cellular targets of NP at fractures were characterized by isolating the heterogeneous mixture of fracture-associated cells. Briefly, TBP-NP loaded with IR780 dye was introduced systemically to fractured mice at day 3 post-surgery, followed by cell isolation 24 hours later. Fracture-associated cells, including endothelial cells (ECs), mesenchymal stem cells (MSCs), osteoblasts (OBs), macrophages, and neutrophils, were evaluated using flow cytometry. Data suggest that > 99% of TBP-NP were taken up by macrophages and neutrophils rather than cell types typically considered regenerative (e.g., MSCs and OBs) (Figure 2). Macrophages and neutrophils exhibited the greatest NP uptake, consistent with their role in the mononuclear phagocyte system (MPS) (Figure S4).^{12, 28} 20% more TBP-NP positive cells were identified as macrophages than neutrophils, possibly due to greater macrophage lifespan than neutrophils, resulting in a subsequent focus on the effect of TBP-NP on macrophage modulation.

Figure 2. Macrophages and neutrophils take up fracture-associated TBP-NPs.

TBP-NP were injected into fractured mice at day 3 post-surgery. 24 hours later, cells from fractures were harvested and evaluated for TBP-NP uptake, where macrophages and neutrophils took up the majority of TBP-NP. Specifically, flow cytometry analysis of TBP-NP % positive cells from each cell type isolated from bone marrow and bone tissue at the fracture. Live cells are identified by Ghost Dye Violet 450 negative gate, and cell types are identified by cell-specific markers: endothelial cells (CD45−/Ter119−/CD31+), osteoblasts (CD45−/Ter119−/CD31−/Sca-1−/CD51+), macrophages (CD45+/F4/80+/Gr-1−), neutrophil (CD45+/F4/80−/Gr-1+), and mesenchymal stem cells (MSCs) (CD45−/Ter119−/CD31−/Sca-1+/CD51+). N = 5, mean ± standard deviation. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 indicates significance evaluated using two-way ANOVA with Tukey’s multiple comparisons.

To explore the underlying cellular mechanisms of NP-mediated fracture healing, the response of macrophages treated with free AR28, unloaded TBP-NP, and TBP-NP_AR28 was evaluated. Specifically, different macrophage phenotypes, including M1 and M2 phenotypes, were differentiated from mouse bone marrow-derived monocytes (BMDMs) and treated with AR28, TBP-NP, and TBP-NP_AR28. The cytocompatibility of all treatment groups shows no reduction in cell viability, indicating the excellent biocompatibility of this drug delivery system (Figure S5). According to macrophage surface marker expression data shown in Figure 3, all treatment groups show inhibition of M1 marker expression and upregulation of M2 marker expression across different phenotypes versus controls. Specifically, there was 1.4-fold, 1.5-fold, and 3-fold downregulation in M1 markers (MHCII, and CD38), and 1.5-fold, 4-fold, and 12-fold upregulation in M2 markers (CD206, and CD163) after M1 macrophages were treated with AR28, TBP-NP, and TBP-NP_AR28, respectively, indicating AR28 promotes M1-to-M2 macrophage polarization to orchestrate fracture healing^{20, 21, 25, 33, 34}. Data suggest that unloaded TBP-NP also promotes M1-to-M2 macrophage polarization, albeit to a far less extent (4.0-fold versus 12-fold) than TBP-NP_AR28, consistent with our previous findings that specific proteins within the adsorbed corona varied based on NP surface charge and correlated with dramatic differences in macrophage polarization.¹⁹ Nevertheless, M2 phenotypic enhancement was significantly greater in TBP-NP_AR28 treated groups (Fig. 3A, C). This trend was also supported via gene expression analysis, where there was significant downregulation of M1-related genes and upregulation of M2-related genes after TBP-NP_AR28 treatment, compared with free AR28 or TBP-NP treatment. The increase in TBP-NP_AR28-mediated macrophage modulation is consistent with our previous work and literature showing enhanced drug effectiveness after NP loading, which could be associated with increased drug stability, controlled drug release, and/or improved cellular uptake, compared with free AR28. For example, doxorubicin (DOX)-loaded PSMA-b-PS micelles were more cytotoxic than free DOX, attributed to superior intracellular retention of DOX-loaded NPs.¹⁷ In addition, in vitro efficacy of doxorubicin-loaded mesoporous silica nanoparticles (MSN-Dox) and curcumin-loaded chitosan nanoparticles (C-CNPs) in treating cancer cells exhibited sustained drug release, improved cellular uptake, and enhanced anticancer efficacy compared to free drugs.^{35, 36} Another study found that dexamethasone-loaded poly(lactic acid-co-glycolic acid) PLGA NP showed improved anti-inflammatory modulation, demonstrated by reduced nitric oxide levels in macrophages.³⁷ Therefore, TBP-NP DDSs may be a promising approach to impact macrophage polarization and that anionic PSMA-b-PS NP, in particular, may be advantageous to improve fracture healing due to increased M2/M1 ratios.

Figure 3. TBP-NP loading enhances the impact of AR28 on macrophage polarization.

TBP-NP_AR28 inhibits M1 and promotes M2 polarization more efficiently than free AR28, quantified by flow cytometry (A, C) and qPCR (B, D). MHCII and CD38 are used as M1 markers, while CD206 and CD163 are M2 markers. Data represented as mean ± standard deviation, n= 3. * Represent the comparison to untreated controls. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, and ^#p<0.05, ^##p<0.01, ^###p<0.001, ^####p<0.0001 using two-way ANOVA followed by Dunnett’s multiple comparisons.

A cell coculture system was used to further investigated how TBP-NP_AR28-mediated macrophage modulation affects osteogenic differentiation. M1 or M2 macrophages differentiated from BMDM were co-cultured with MC3T3-E1 preosteoblasts in transwell plates. Subsequent assessment of osteogenesis via alkaline phosphatase (ALP) and Alizarin Red staining (ARS) on days 7 and 21, respectively, was performed. As shown in Figure 4, both ALP and mineral deposition were reduced in MC3T3-E1/M1 coculture systems (Figure 4B), while both ALP and mineral deposition increased by 1.3 and 1.4-fold in the MC3T3-E1/M2 cocultures (Figure 4C), compared with the MC3T3-E1 controls, indicating that M2 macrophages enhance in vitro osteogenesis, which is consistent with other studies^{21, 26, 27, 33, 38}. Furthermore, in the MC3T3-E1/M2 coculture system, the TBP-NP_AR28 treated group exhibited greater ALP and Alizarin Red staining than the control group, demonstrating enhanced osteogenesis in TBP-NP_AR28 group may be due to M2 macrophage polarization (Figure 4C).

Figure 4. TBP-NP_AR28 significantly enhanced in vitro osteogenic differentiation, associated to upregulation of M2 macrophage polarization.

Pre-osteoblasts, MC3T3-E1, were cocultured with M1 or M2 macrophages in a transwell (membrane pore size = 0.4 μm) plates. Alkaline phosphatase (ALP) (on day 7) and Alizarin Red staining (ARS) (on day 21) were tested, imaged (A-C), and quantified (D: mineralization, E: ALP assay), where TBP-NP_AR28 mediated upregulation of M2 polarization promote osteogenic differentiation, compared with other groups. Data represented as mean ± standard deviation, n= 3. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 using two-way ANOVA followed by Dunnett’s multiple comparisons

M1 macrophages play key roles in the initial inflammatory phase of fracture healing by recruiting various regenerative cells to the injury site.^{34, 39, 40} However, excessive, or prolonged M1 presence will inhibit osteogenesis and delay overall healing. The dual functionalities of M1 macrophages in regulating bone regeneration have led to further investigation in vitro and in vivo but the impact of M1 macrophages on osteogenesis is inconsistent.^{27, 34, 41-43} For example, coculture of MC3T3E1 and M1 macrophage inhibited ALP activity but increased mineral deposition²⁷, while small extracellular vesicles (sEVs) derived from M1 macrophages (M1-sEVs) inhibited both ALP and Alizarin red staining of mouse mesenchymal stem cells (mMSCs)²⁵. This inconsistency highlights the need for a comprehensive and fundamental understanding of the impact of M1 macrophages on osteogenesis. In our study, ALP staining in the MC3T3E1/M1 cocultured group was significantly reduced. Still, there were no changes in AR28, TBP-NP, and TBP-NP_AR28 treated MC3T3E1/M1 cocultures, which may be due to alleviation of M1-induced inhibition of osteoblastic differentiation by the drug, TBP-NP, and the combination. In our in vitro system, M1 macrophages negatively influenced osteogenesis. While further investigation into this discrepancy is important, our primary objective was determining the influence of the bone-targeted nanoparticle drug delivery systems on macrophage polarization and overall fracture healing. Thus, an exhaustive investigation of the relationship between M1 macrophage and osteogenesis was not performed. Furthermore, no differences in MC-3T3E1 osteogenesis were observed without co-cultured macrophages (Figure S7). Therefore, the robust impact of TBP-NP _AR28 in promoting M2 macrophage polarization and osteogenesis, together with preferential NP macrophage uptake at fractures, motivated further exploration of the underlying mechanism of TBP-NP_AR28 -mediated fracture healing with the focus on macrophage-associated osteoimmunomodulation in fracture healing.

While in vitro studies provide valuable insights into the impact of AR28 and/or TBP-NP_AR28 on macrophage polarization and osteogenic differentiation, complexities of the in vivo fracture environment are not fully recapitulated. Therefore, to further investigate the osteoimmunomodulatory role of macrophages in fracture healing, histological analysis was performed to characterize the spatial localization and polarization of macrophages at fractures on day 10 post-injury (Figure 5A). F4/80+ cells are generally localized to regions adjacent to fracture sites, with the majority at the medullary cavity (Figure 5B). F4/80+ cells in the TBP-NP_AR28 group exhibit greater co-localization with fractures, while in the other groups, F4/80+ cells are distributed more distally to the fracture region. This may be linked to a faster transition from the inflammatory to bone repair microenvironment due to TBP-NP_AR28 treatment since the early inflammatory stage is accompanied by an influx of F4/80+ cells⁴⁴. Based on colocalization with the F4/80 marker, M1- and M2-like macrophages were labeled by inducible nitric oxide synthase (iNOS) and the mannose receptor (CD206), respectively. Quantitative analysis showed a high abundance of M1 and M2 macrophages in the AR28 groups, with increased M2 macrophages in the TBP-NP_AR28 group ((Figure 5B, C, Figure S8). This increase in M2 macrophage population in TBP-NP_AR28 group resulted in a substantially higher M2/M1 ratio (Figure 5D), associated with expedited healing with TBP-NP_AR28 treatment.

Figure 5. TBP-NP_AR28 exhibited increased M2/M1 ratio and distinct spatial distribution of macrophages at fractures.

A) Schematic illustration of experiment design and groups for histology, bulk RNAseq and flow cytometry. Mice were administered saline, AR28, TBP-NP, and TBP-NP_AR28 on day 3 post-fractures, and then fracture-associated cells were collected at day 1, day 4, and day 7 after injection for further analysis; B) Immunofluorescence staining of F4/80, iNOS, and CD206, where M1 macrophages were determined as F4/80+iNOS+ and F4/80+CD206+ for M2 subtype. Scale bars = 1000 μm. C, D) Quantitative analysis of M1, M2, and M2/M1 ratio via Imaris software, where M1 or M2 macrophages amounts were calculated by colocalized area of F4/80+ and iNOS+ or F4/80+ and CD206+, normalized by total femur area. Data represented as mean ± standard deviation (SD), N=5, n=3. E-G) Flow cytometry analysis of M1 and M2 macrophage phenotypes, gated as double positive for MHCII/CD38 (M1 macrophage) and CD206/CD163 (M2 macrophage), at day 10 post-fracture; M1 and M2 macrophage percentages within all macrophages at fracture site over time. Data represented as mean ± standard deviation (SD), n=4. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, as determined by two-way ANOVA and Tukey’s multiple comparison test.

Unlike other groups with F4/80+CD206+ cells localized to the medullary area, these cells were found not only in the medulla and periosteal region (Figure 5B, yellow arrows) but also on the cortical bone surface (Figure 5B, red framed inserts) in the TBP-NP_AR28 treatment group. F4/80+ cells at the cortical bone surface are known as osteal macrophages, or OsteoMacs, a special subtype of macrophages, that also express F4/80 and CD206 markers, found in bony tissue, and are essential in bone healing. Therefore, the localization of F4/80+CD206+ OsteoMacs on the bone surface suggests active involvement of these cells in bone repair, further indicating expedited resolution of inflammation and transition to bone repair after treatment with TBP-NP_AR28. In summary, achieving the appropriate temporal balance between M1 and M2 macrophages during fracture healing is crucial for optimal bone regeneration; the M2/M1 ratio is a useful measure of this balance.

To directly and quantitatively evaluate M2/M1 ratios, flow cytometry was performed to identify M1 and M2 macrophage phenotypes and M2/M1 ratios over time. Cells isolated from fractures were first gated for macrophages (F4/80+CD45+Ly6G−Ly6c−), then for M1 and M2 phenotypes by double gating MHCII+/CD38+ and CD206+/CD163+ cells. The results showed no significant differences in M1 and M2 macrophage percentages across all groups at days 4 and 7 after fracture. At day 10 post-fracture, there was a slight increase in M2 macrophages present with AR28 and TBP-NP_AR28 treatment, compared with saline controls. In contrast, the upregulation of M1 macrophages was substantial in AR28 compared with all the other groups (Figure 5E, Figure S9). Fractures treated with TBP-NP_AR28 exhibited similar M2 macrophage prevalence over time with significantly decreased M1s at day 10, compared with AR28, leading to gradual increases in M2/M1 ratios after treatment that becomes significant at day 10 (Figure 5G). Compared with saline and TBP-NP, AR28 treatments exhibited excessive and prolonged M1 macrophages, which are known to compromise M2-mediated bone repair. These data suggest that TBP-NP_AR28 modulates and rebalances M2/M1 ratio resulting from AR28, which is associated with enhanced fracture healing in both bone formation volume and mechanical properties after TBP-NP_AR28 treatment.

For the TBP-NP group, no differences in M1 and M2 macrophages were observed. However, the M1 phenotype percentage slightly decreased, and the M2 macrophage percentage increased, resulting in a significant increase in M2/M1 ratio at day 10, compared with saline and AR28, which could underpin increased bone volume and torsional rigidity at week 4. The physicochemical properties of NP impact macrophage modulation; previous data¹⁹ showed that anionic PSMA-b-PS NP promotes M2 and inhibits M1 macrophage polarization in vitro, which is further associated with increased M2/M1 ratio during fracture healing to enhance healing from TBP-NP alone in vivo. In summary, improved fracture healing of TBP-NP_AR28 is correlated with increased M2/M1 ratios during bone repair.

To explore the underlying mechanism of enhanced fracture healing via delivery of TBP-NP_AR28 at a gene expression level, bulk RNA sequencing (RNA-seq) analysis was performed. Specifically, mice were treated with saline, AR28, TBP-NP, and TBP-NP_AR28 on day 3 post-fracture. Fracture-associated cells were collected at days 1, 4, and 7 after treatment for bulk RNA-seq analysis. The top 2000 most variable genes of all groups and time points were clustered and plotted into a heatmap (Figure 6A), after which enrichment analysis was performed for each cluster based on the Gene Ontology (GO) Biological Processes and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. According to GO enrichment analysis, Cluster C and D are highly enriched with biological processes associated with fracture healing, such as extracellular matrix organization, endochondral bone morphogenesis, bone development, ossification, etc. (Figure 6B, Figure S8). Within these two clusters, gene expression levels in the TBP-NP and TBP-NP_AR28 treated groups are overall higher than saline controls on day 10 after surgery and significantly differentially expressed genes (DEGs) at day 10 were also enriched in the bone regeneration-associated biological processes (Figure 6C), which may explain the enhanced biomechanical properties of fractures by TBP-NP and TBP-NP_AR28 groups.

Figure 6. Bulk RNAseq analysis indicated that TBP-NP_AR28 enhanced fracture healing via increasing M2/anti-inflammatory and downregulating M1/pro-inflammatory markers.

A) Clustering for the top 2000 most variable genes in all groups at all time points; B) GO analysis of bone regeneration related clusters D and E in panel A; C) Enriched biological processes by GO analysis of differential genes in AR28, TBP-NP, and TBP-NP_AR28, compared with saline at specific time points (p < 0.05 and log2-fold >1); D) Signaling pathway analysis by KEGG of differential genes in TBP-NP_AR28 compared with AR28 at day 10 post-surgery (p < 0.05 and log2-fold >1); E) Network of top 30 differential genes and their associated inflammatory response, plotted by Cytoscape; F) Summary of the relationship between M1/pro-inflammation and M2/anti-inflammation markers and fracture healing.

To further investigate how TBP-NP_AR28 alters fracture healing, gene expression of fracture-associated cells was analyzed. Based on the comparison with free AR28, significantly differentially expressed genes (p < 0.05 and log2-fold >1) at day 10 were plotted into a volcano plot and heatmap, as shown in Figure S11A, B. The biological processes enriched within the DEGs were investigated by GO analysis (Figure. S11C). Results showed that the upregulated DEGs in TBP-NP_AR28 were enriched for extracellular matrix organization, chondrocyte differentiation, bone development, and skeletal system development. More specifically, biomarkers related to chondrogenic differentiation (such as COMP, SOX6, SOX5, COL2A1) and osteogenic differentiation (such as IBSP, RUNX2, IHH, RUNX3, BMP2) were upregulated, further suggesting enhanced fracture healing related to these gene expression patterns. In addition, KEGG pathways analysis further uncovered that osteoclast differentiation-related genes were significantly downregulated in the TBP-NP_AR28 group compared with the AR28 group (Figure 5D), specifically by upregulating OPG gene expression while downregulating RANKL expression, leading to a decrease in RANKL/OPG ratio (Fig. S12). As healing progresses, the RANKL/OPG ratio is dynamically regulated to promote bone formation and remodeling. During bone repair and formation, decreased RANKL/OPG ratio has been suggested to promote osteoblast-mediated bone formation.^45-47 Furthermore, upregulated DEGs were enriched in genes related to vasculature development, including angiogenesis and arteriogenesis, which are essential for oxygen and nutrient exchange during fracture healing and provide a pathway for the recruitment of cells involved in healing^48-51. Therefore, enhanced fracture healing outcomes with TBP-NP_AR28 treatment compared with the AR28 group are associated with elevated chondrogenesis, osteogenesis, vascular formation, and reduced osteoclast differentiation. Gene expression analysis at the later stages of fracture healing further captures the dynamics of gene expression patterns based on different treatments.

Since NPs that accumulate at fractures are largely taken up by macrophages, we hypothesized that enhanced fracture healing is due to altered macrophage polarization, specifically by promoting M2 polarization and reducing M1 presence, resulting in a less inflammatory microenvironment. In fact, GO and KEGG pathway analysis of upregulated DEGs in the AR28 group compared with saline at day 10 suggested that the top 5 highest enriched biological processes involve regulation of response to stimulus, bacterium, external biotic stimulus, other organisms, and inflammation. Similarly, the top 5 enriched pathways are cytokine-cytokine receptor interactions, NF-kappa B signaling pathway, IL-17 signaling pathway, TNF signaling pathway, and lipid and atherosclerosis, most of which are actively involved in pro-inflammatory responses (Figure 5C). Pro-inflammatory cytokines such as IL-17 and TNF-α have been shown to inhibit the differentiation and proliferation of osteoblasts, as well as stimulate the differentiation and activity of osteoclasts, resulting in an imbalance between osteoblasts and osteoclasts and impaired bone healing. Thus, even with significant upregulation of bone regeneration-related genes for the AR28 group at early time points, no DEGs associated with fracture healing were identified at day 10, and bone formation was unaffected compared to saline at week 4. Nevertheless, no DEGs related to inflammation were identified for the TBP-NP_AR28 treated group at any time point compared to saline controls (Fig. 5C). In addition, KEGG pathway analysis of DEGs in TBP-NP_AR28 compared with AR28 indicated that downregulated DEGs were enriched in pro-inflammation-related signaling pathways, such as Th17 cell differentiation, NF-kappa B signaling pathway, TNF signaling pathway, IL-17 signaling pathway, and Toll-like receptor signaling pathway (Figure 5D). These data suggest that the bone-targeted NP DDS, TBP-NP_AR28 downregulated the excessive and prolonged inflammatory response associated with AR28, further enhancing bone regeneration after the resolution of inflammation.

To understand inflammation-related genes underpinning the advantages of TBP-NP_AR28 over AR28 in fracture healing, the top 30 differentially expressed genes were selected, plotted into a network based on the fold change and adjusted p-value via Cytoscape (Figure 5E). According to the network, 70% of the downregulated genes were related to M1 or pro-inflammatory activity. More upregulated genes were associated with M2 or anti-inflammatory response, such as PTPRS, METAP1D, MTX3, and FN1, while only two (FIRRE and WNK4) have pro-inflammatory impact. Therefore, M2/M1 ratios in TBP-NP_AR28 groups compared with AR28 treatment led to enhanced fracture healing from TBP-NP_AR28 treatment. Studies have shown that increased M2/M1 macrophage ratio during fracture healing is associated with improved bone formation and increased bone mineral density.

Conversely, a decrease in M2/M1 macrophage ratio is normally correlated to delayed healing and increased risk of complications such as non-union or infection³³, further supporting the importance of M2/M1 ratio during healing. Interestingly, and similar to our findings in vitro, TBP-NP treatment also impacted macrophage polarization in vivo. Although downregulation of both M1 and M2 macrophage-related genes was observed with TBP-NP treatments at day 10, M2 macrophage-related genes were significantly upregulated at day 4, leading to an increase in M2/M1 ratio during the inflammation-resolving stage, which may be associated with upregulation of bone regeneration related genes and ultimately enhanced fracture healing.

Interestingly, DEGs in TBP-NP_AR28 also suggests the involvement of neurogenesis and nerve recruitment, compared with saline controls. In addition, pathways of neurodegeneration were downregulated in TBP- NP_AR28 compared with AR28, suggesting neurogenesis-associated enhancements in fracture healing. Neurogenesis has been shown to impact fracture healing positively, for example, nerve growth factor (NGF) can stimulate osteoblast differentiation and mineralization^{52, 53}. Furthermore, the M2 macrophage phenotype has been associated with central nerve system (CNS) repair, such as neurogenesis, axonal regeneration, etc. Therefore, the increased M2/M1 ratio in TBP-NPAR28 may be correlated to the upregulation of neurogenesis and bone repair. Future studies could include single-cell RNA sequencing (scRNAseq) to directly capture dynamic gene expression patterns of all cells (e.g., osteoblast, MSC, endothelial cells, neurons, osteocytes, and osteoclasts) to draw more comprehensive conclusions regarding bone-targeted NP-mediated macrophage polarization on fracture healing.

Compared with saline, the summary of pro-inflammatory M1 macrophage and anti-inflammatory M2 macrophage phenotypes from AR28, TBP-NP, and TBP-NP_AR28 and the correlation to fracture healing is organized in Figure 5G. Enhanced fracture healing observed with TBP-NP_AR28 is associated with upregulation of M2 macrophage (anti-inflammation) genes and downregulation of M1 macrophage (pro-inflammation) genes, altogether resulting in elevated M2/M1 ratios. Therefore, M2/M1 is an important indicator of the immune response and fracture healing efficacy, providing insights into the prognosis and management of fractures.

Our data (Figure 3) shows that, in vitro, free AR28 inhibits M1 and promotes M2 macrophage polarization in vitro. However, in vivo data revealed that free AR28 enhances M1 macrophage polarization, compared with other groups. This may be due to temporal effects or relative concentration of AR28 at the fracture site. Upon injection of free AR28, far less is expected to reach bone versus TBP-NP_AR28 due to lack of targeting. Furthermore, accumulation and clearance of free AR28 will be rapid, which may result in transient suppression of the pro-inflammatory response by inhibiting M1 macrophage polarization. The transient exposure may result in a replenishment/surge in the pro-inflammatory response. In contrast, TBP-NP_AR28, characterized by a prolonged circulating time and fracture persistence, can steadily inhibit M1 and promote M2 macrophage polarization, therefore, ultimately increasing M2/M1 ratio.

Based on our previous work, AR28 enhanced the Wnt/β-Catenin signaling pathway, which promotes osteogenesis and fracture healing,¹⁶ similar to previous findings.⁵⁴ RNAseq data, as presented in Figure S13, supports upregulated Wnt/β-Catenin signaling, as most genes in the Wnt signaling pathway exhibit heightened expression in the TBP-NP_AR28 treated group compared to unloaded TBP-NP. Furthermore, our observations show that most nanoparticles are taken up by macrophages at the fracture site. Given that the Wnt signaling is known to promote M2 macrophage polarization⁵⁵, our latest findings showing the impact of NP on macrophage polarization¹⁹, as well as the emerging recognition of the crucial role macrophages play in bone regeneration^{20, 27, 33, 56-58}, we posit that the modulation of macrophage polarization by TBP-NP_AR28 is intrinsically linked to the activation of Wnt signaling, leading to expedited fracture healing.

Leveraging macrophage functions has a much more profound impact and opportunity than targeting regenerative cells. This finding is particularly significant and promising in nanoparticle drug delivery systems, as macrophages preferentially take up nanoparticles despite efforts to overcome macrophage uptake and the presumably negative consequences of that uptake, including clearance. Therefore, by targeting macrophages, the issue of unintended macrophage uptake can be exploited, ultimately improving delivery efficiency, and enhancing therapeutic efficacy. To this end, our study is the first to leveraging fracture-associated macrophage as a druggable target due to its preferentially nanoparticle uptake and subsequent regenerative impact.

Several studies have investigated nanoparticle uptake by circulatory cells in the context of cancer therapy.^1-3 In this work, it is possible that NP uptake by circulating monocytes/macrophages may play a role in fracture localization. However, if this were a predominant mechanism of transport, it is unlikely that such a stark difference in fracture-targeting of TBP-NP versus controls (NP and SCP-NP) would be observed, as targeting is modulated by the TBP-TRAP interaction consistent with osteoclast secreted TRAP at fractures. Irrespective of circulatory versus fracture-associated cells, the overall conclusions of the work are unaltered. Nevertheless, further investigation is warranted to tease out the potential roles of circulatory cells.

The flexibility in timing of NP fracture targeting is also an advantage of this technology. Specifically, preferential nanoparticle uptake was observed at the early stages of healing and leveraged NP fate to enhance healing and demonstrated that macrophages dominate nanoparticle uptake at later healing stages (Day 21, Figure S14). This presents promising possibilities for fracture treatment, particularly for managing nonunion complications during later stages of bone healing and providing insights into developing drug delivery system to promote therapeutic outcomes at various times corresponding to the predominant cellular type.

In summary, our study demonstrates that following macrophage uptake, femur fracture healing was expedited in mice treated with bone-targeted nanoparticles delivering AR28, compared with control groups. Additionally, in vitro analysis demonstrated the role of bone-targeted nanoparticle drug delivery system in promoting M2 macrophage polarization and enhancing osteogenesis by modulating preosteoblast-macrophage crosstalk. Furthermore, longitudinal analysis of fracture healing employing histology, flow cytometry, and RNAseq, revealed distinct spatial distributions of M2 macrophages, an increased M2/M1 ratio, and upregulation of anti-inflammatory and downregulated pro-inflammatory genes in our bone-targeted nanoparticle drug delivery system compared to controls.

CONCLUSION

This study provides a comprehensive investigation of the impact and underlying mechanism of NP DDS on macrophage-mediated fracture healing using a bone-targeting NP platform. Specifically, data suggested that macrophages took up most fracture-associated TBP-NP. Mice treated with TBP-NP_AR28 exhibited enhanced healing via increased bone volume formation and mechanical properties, associated with macrophage modulation due to anionic NP’s effect on M2 macrophage polarization. To verify this, in vitro macrophage modulation study revealed the robust impact of AR28 on promoting M1-to-M2 macrophage polarization. Bone-targeted TBP-NP loaded with AR28 (TBP-NP_AR28) demonstrated greater influence on macrophage modulation leading to greater osteogenic differentiation of preosteoblasts in vitro. In vivo, this effect was associated with distinct spatial distribution of bone-associated macrophages and increased M2/M1 ratio qualified and quantified by histology and flow cytometry. At the gene expression level, bulk RNA-seq analysis demonstrated increased expression of anti-inflammatory-related genes and downregulating pro-inflammatory biological processes in TBP-NP_AR28 group. Therefore, M2/M1 ratio was found to govern healing outcomes, and TBP-NP_AR28 upregulated this ratio by promoting M2 macrophage polarization and inhibiting M1 macrophage modulation. Furthermore, quantification of fracture-localized AR28 and TBP-NP_AR28 to further understand the pharmacokinetics and pharmacodynamics of this bone-targeted NP system is warranted. In addition, the impact of NP-mediated healing in more clinically-relevant models, including advanced age, rheumatoid arthritis, and diabetes, which are highly associated with nonunion fractures, is worthwhile. Altogether, enhanced fracture healing via delivery of TBP-NP_AR28 increased M2/M1 ratio. The importance of macrophage phenotype in various diseases, preferential accumulation at fractures, and regulation of macrophage polarization, highlight a potent therapeutic benefit of bone-targeted NP DDS.

METHODS

Poly(styrene-alt-maleic anhydride)-b-poly(styrene) (PSMA-b-PS) Synthesis and Characterization

Diblock polymers were synthesized as previously described.^{10, 12, 13, 16, 17} In brief, to create block copolymers with alternating maleic anhydride (MA) and styrene (STY) in the first block (PSMA) and polystyrene (PS) in the second block, STY was added in excess of MA (4:1 [STY]/[MA]) in the presence of the chain transfer agent (CTA) 4-Cyano-4-dodecylsulfanyltrithiocarbonyl sulfanyl pentanoic acid (DCT) (100:1 [MA]/[CTA]). The radical initiator 2,2′-Azo-bis(isobutylnitrile) (AIBN) was used (10:1 [CTA]/[initiator]) in dioxane (128% w/w). The reaction was purged with nitrogen for 45 minutes before being placed in a 60 °C oil bath for 72 hours. Polymer solutions were exposed to air and dissolved in acetone before precipitation in petroleum ether. Precipitated polymers were dried in vacuum overnight prior to characterization of their number average molecular weight (Mn), weight average molecular weight (Mw), and dispersity (Ð) via gel permeation chromatography (GPC).

Peptide Synthesis and Characterization

Peptides synthesis was performed according to previously described.¹⁰ Briefly, tartrate-resistant acid phosphatase (TRAP) binding peptide (TBP_Alloc) (sequence: TPLSYLK^AllocGLVTVG) was synthesized using Fluorenylmethyloxycarbonyl chloride (FMOC)-protected amino acids, Fmoc-Gly-Wang resin (Millipore, MA), by a Liberty1 synthesizer (CEM Corp). Coupling was accomplished with an activator mix of 0.5 M O-(benzotriazole-1-yl)-N, N, N’, N’-tetramethyluronium hexafluorophosphate (HBTU) in dimethylformamide (DMF) and activator base mix of 2 M N, N – Diisopropylethylamine (DIEA) in 1-methyl-2-pyrrolidinone (NMP). Synthesized peptides were then deprotected with 5% piperazine in DMF, followed by cleavage using 92.5% trifluoroacetic acid (TFA), 2.5% H2O, 2.5% 3, 6-dioxa-1,8-ocatanedithiol (DODT), and 2.5% triisopropylsilane (TIPS) mixture for 2.5 hours. Products were then precipitated in ice-cold diethyl ether to remove DMF and byproducts of the cleavage mixture, followed by three times of washing cycle to remove trace impurities. The products were purified by dialysis (Spectra/Por, MWCO = 500 D) against water, freeze-dried, and analyzed for purity using high-performance liquid chromatography (HPLC) (Shimadzu LC-20AD HPLC system, SPD-20AV UV-Vis detector) and molecular weight using matrix-assisted laser desorption ionization time-of-light mass spectrometry (MALDI-TOF) (Brüker Autoflex III), as described in our previous publications.^{10, 12, 13, 16, 17} Peptide purity was ≥ 95% for peptides used in this study.

Polymer Functionalization and NP Self-Assembly

TBP-Alloc peptide was conjugated to PSMA-b-PS diblock copolymers via MA ring opening nucleophilic addition-elimination conjugation. TBP-Alloc was added at 10 mol% of maleic anhydride within the PSMA alternating copolymer block, which was 104 repeat units for the diblock used herein (see Table 1).¹⁰ The peptide-polymer mixture was stirred at room temperature overnight.¹⁰ Water was added to quench the reaction, which was subsequently dialyzed (MWCO = 6-8 kDa) for 3 days to remove any unreacted peptide and DMF. The resulting peptide-polymer, TBP_Alloc-PSMA-b-PS, was freeze-dried and analyzed by GPC to determine the number of conjugated peptides on each polymer, as described in our previous work¹⁰. The Alloc group was then removed using tetrakis(triphenylphosphine)palladium [Pd(PPh₃)₄] in the presence of phenylsilane⁵⁹. The final product TBP-PSMA-b-PS was then verified by nuclear magnetic resonance (NMR). To form TBP-NP, ddH₂O was added at a rate of 24 μL/min via syringe pump to TBP-PSMA-b-PS dissolved in 30 mL DMF (200 mg/30 mL), which was then dialyzed against water for 3 days (MWCO 6-8 kDa) and filtered using 0.2 μm sterile cellulose acetate filters.^{10-12, 16, 17, 19} Dynamic light scattering (DLS, Malvern Instrument, Worcestershire, UK) was used to determine the size and surface charge of TBP-NP. A Malvern Nanosight analysis further determined the number of peptides per NP to measure the number of TBP-NP.¹⁶

Mouse Mid-Diaphyseal Femur Fracture Model

Mouse studies were approved by the Institutional Animal Care and Use Committee at the University of Rochester School of Medicine and Dentistry (UCAR). Female Balb/c mice aged 6-8 weeks (The Jackson Laboratory) were used for in vivo studies following a previously established mid-diaphyseal femur fracture model.^{12, 14, 16} Briefly, slow-released Buprenorphine (0.05 mg/kg) was intraperitoneally injected into mice before surgery. During the surgery, mice were anesthetized via isoflurane inhalation (1-2.5%) after one week of acclimation. An 8 mm-length skin incision was made on the femur, followed by a blunt muscle dissection to expose the femur's midshaft. Mid-diaphyseal femur fractures were then generated using a rotary Dremel attached with a diamond blade. Femur was stabilized by a 25-gauge needle inserting into the medullary canal of the femur from the distal end and sterile 5-0 nylon surgical sutures were used for closing surgical incisions.

In Vivo Cellular Uptake by Flow Cytometry

Flow cytometry was performed to evaluate NP cellular uptake at fractures.¹⁶ On day 3 or day 21 after the fracture, mice were given injections of IR780-loaded TBP-NP at a dose of 5 mg/kg (NP basis) retroorbitally. IR780 loading was performed as previously reported.⁶⁰ Briefly, 0.8 mL of 0.75 mg/mL IR780 in acetone was added dropwise to 10 mL of stirring NP solution in water at 0.8 mg/mL. The IR780-loaded NP were then purified using centrifugal filtration (MWCO = 10 kDa, Milipore). 24 hours after injection, bone and marrow tissue samples were collected from the fracture sites for cell isolation. The tissue was treated with red blood cell lysis buffer to isolate bone marrow cells and then digested with collagenase type IV, dispase, and DNase. Bone tissue was crushed into small pieces and digested with collagenase type I. Single cell suspensions of both bone marrow and bone-associated cells were obtained by passing the cells through 100 μm cell strainers. The cells were then stained with Ghost Dye Violet 450 (VWR, #10140-978) for cell viability for 30 minutes on ice in the dark, which was then resuspended in a flow buffer containing 2% FBS, and 0.01% trypan blue at a concentration of 1 × 10⁶ cells/ml. To evaluate cell markers, the cells were stained and incubated with rat anti-mouse antibodies (PE-Cy5-labeled Ter119, Alexa Fluor 488-labeled CD45, PerCP-Cy5.5 labeled CD31, Alexa Fluor 647-labeled Ly-6A/Ly-6E, PE-Cy7 labeled F4/80, Brilliant Violet 605 labeled Ly-6C, Brilliant Vilot 785 labeled Ly-6G) for 30 minutes on ice in the dark. Finally, the cells were analyzed using a BD LSR II Flow Cytometer.

Drug Loading and Release from NP

AR28 was loaded into TBP-NP with methods akin to NP self-assembly. Specifically, 15 mL of ddH₂O was added into a mixture of AR28 (25 mg) and TBP-polymer (100 mg) in 15 mL of DMF using a syringe pump at the speed of 24.4 μL/min in the dark, as AR28 is light-sensitive. The resulting product, TBP-NP_AR28, was dialyzed against water to remove DMF. Drug purification and release were then performed based on our previous approach¹⁶. Specifically, drug-loaded NP was centrifuged twice at 4000 rpm for 10 min to remove insoluble drug precipitates and then further purified twice using centrifugal filters (MWCO = 10 kDa, Milipore) at 2000 rpm for 10 min to remove free drug and polymers. Drug loading efficiency and capability was quantified by HPLC at flow rate of 0.5 mL/min, from 90% to 30% A (0.1% TFA in HPLC-grade water) and 10% to 70% B (HPLC-grade acetonitrile) over 6 min. The retention time was 3 min at 302 nm.¹⁶ Drug loading efficiency and capacity were calculated as loaded drug/total drug × 100% and loaded drug mass/NP mass × 100%, respectively. Drug release from TBP-NP_AR28 were performed by dialyzing in a dialysis membrane (MWCO 6-8 kDa) at physiological pH (pH 7.4) and acidic pH (pH 4.5) with changing release media twice daily, and samples were collected over 9 days. Drug release profile was also quantified by HPLC as aforementioned description.

Mouse Bone Marrow-derived Macrophages (BMDM) and Differentiation

Bone marrow-derived macrophages (BMDMs) were collected from 6-8-week-old female Balb/c mice (Jackson Laboratory), followed previous method¹⁹. Briefly, femurs and tibias were collected, cleaned with cold PBS, and marrow flushes were filtered through a 70-micron cell strainer. After centrifugation, cell pellets underwent red blood cell lysis using lysis buffer (5 mL for 1 minute on ice), which was then neutralized with 15 mL of high-glucose Dulbecco's Modified Eagle Medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (FBS) and 100 units/mL penicillin-streptomycin (Gibco). The cells were centrifuged again at 1100 rpm for 5 minutes. Once resuspended in 15 mL of DMEM containing macrophage colony-stimulating factor (M-CSF, 25 ng/mL), cells were placed in a T-75 flask and grown for 24 hours and non-adherent cells were transferred to a new plate or flask for further differentiation. By day 7, monocytes were differentiated to M0 macrophages and prepared for M1 polarization using 50 ng/mL of lipopolysaccharide (LPS) and 25 ng/mL of interferon-γ (INF-γ), and M2 polarization using 20 ng/mL of interleukin-4 (IL-4).

Drug In Vitro Cytotoxicity Assay

BMDM were plated to 96-well plates and cultured for 24 hours, followed by drug treatment at various concentrations with nontreated, 10% DMSO, and 0.1% DMSO as negative, positive, and vehicle controls, respectively. 24 hours later, alamarBlue reagent (Thermo Fisher) was added to each well and allowed for 4 hours incubation at 37 °C. Fluorescence (Excitation/Emission: 560/590 nm) from each well was measured by Cytation 5 Plate Reader (BioTek) to calculate cell viability.

In Vitro Macrophage Polarization, Flow Cytometry, and Quantitative Real-Time PCR

M0, M1, and M2-polarized macrophages were exposed to various groups, including saline, AR28 (0.45 μM, based on the dose-dependent study, as shown in Figure S4), TBP-NP, and TBP-NP_AR28. At day 1, 3, and 6, cells were harvested for flow cytometry to determine cell surface marker. The gating strategy for flow cytometry was based on our previous description.¹⁹ Specifically, to obtain a relatively pure population, macrophages were first gated for F4/80+CD45+Ly6C−Ly6G− cells, from which M1 and M2 phenotypes were then identified as MHCII+CD38+ and CD206+CD163+, respectively. The influence of compounds on M1 and M2 polarization was determined by the proportions of M1 and M2 markers, normalized by untreated groups. Quantitative Real-Time PCR (qPCR) was also performed to evaluate the impacts on macrophage polarization at gene expression level. Specifically, cells were collected at day 1 after treatment and lysed with TRK lysis buffer (Bio-Rad) containing 1% v/v 2-mercaptoethanol. After evaluating the quantity and quality of the extracted RNA using Cytation 5 (Biotek) by the A260/280 ratio, cDNA was generated using the iScript cDNA Synthesis Kit (Bio-Rad). PCR was performed using PowerUp SYBR Green Master Mix Assay (Applied Biosystems) to determine the expression of M1 and M2 related genes, with primer sequences listed in Table S2.

In Vitro Osteogenesis via Coculture of Macrophages and Preosteoblasts

BMDM were harvested, plated at transwell insert (12-well plate, pore size of 0.4 μm, Corning), and differentiated as aforementioned procedures. MC3T3-E1 cells were seeded at the bottom chamber in osteogenic media (minimum essential medium containing 10 mM β-glycerophosphate, 0.05 mM ascorbic acid, and 100 nM dexamethasone). After polarization, M1, and M2 macrophages were separately cocultured with MC3T3-E1, specifically MC3T3-E1/M1 group and MC3T3-E1/M2 group, immediately followed by drug treatment with free AR28, unloaded TBP-NP, and TBP-NP_AR28, prepared in a mixed medium (2:1 ratio of osteogenic media and DMEM). Media was changed every other day. While we recognize the significance of studying M0 macrophages' influence on osteogenesis, practical considerations led us to focus our research on M1 and M2 macrophage phenotypes, which are most relevant to the fracture milieu. By studying them, we aim to capture a comprehensive understanding of macrophages' overall effects on osteogenesis.M1 macrophages play a pivotal role in fracture healing, particularly in cell recruitment. Nevertheless, literature on its role in in vitro osteogenesis remains varied.

Osteogenic Staining

For staining, MC3T3-E1 were rinsed in PBS, fixed with neutral buffered formalin (NBF), and rinsed in ddH₂O. For NBT/BCIP (alkaline phosphatase, ALP) staining, reagent was applied to cells for 30 min at 25 °C in the dark at day 7. For Alizarin Red (calcium), cells were stained with 0.2 μm-filtered Alizarin Red Staining Solution (Sigma) for 30 min at 25 °C. Cells were rinsed with ddH2O to preclude non-positive staining and imaged with an Epson Perfection V700 photo scanner and Cytation 5 Imager. To quantify the ALP staining, we dissolved stained cells in for 10 min, and the absorbance of the extracted solution was measured at 562 nm. To assay ALP activity, unstained MC3T3 were rinsed with PBS and lysed with 1% Triton in PBS. Lysates were combined 1:1 with a solution of 5 mg/mL pNPP (p-Nitrophenyl Phosphate, Disodium Salt, Thermo Fisher) in 1 M diethanolamine buffer, pH 9.8, containing 0.5 mM MgCl₂. Absorbance at 405 nm was measured every minute for 5 min normalized to cellular DNA and analyzed as described previously.⁶¹

Mouse Mid-Diaphyseal Femur Fracture Healing Study

Mouse mid-diaphyseal femur fractures were generated surgically, as previously described.¹⁶ Day 3 post-surgery, fractured mice were treated with saline, free AR28 (5 mg/kg), unloaded TBP-NP (8 mg/kg), and TBP-NP_AR28 (5 mg/kg of AR28, 8 mg/kg of polymer) via retro-orbitally injection. The 5 mg/kg dose of AR28 is based on our previous reference.¹⁶ At week 4, mice were euthanized in a CO₂ chamber, followed by cervical dislocation, and femurs were harvested and kept in PBS for microcomputed Tomography (μCT) and biomechanical testing as described previously¹⁶. Specifically, samples were scanned using a Scanco Medical VivaCT 40 at high resolution (10.5 μm voxel size), with X-ray energy of 55 kVp and intensity of 145 μamps, integration time 300 ms. Scanco’s proprietary evaluation software was used for analysis. Contours were drawn around the external callus and along the edge of the cortical bone for each sample. To measure new bone callus volume, fracture contouring was applied to exclude cortical bone from all bone spaces. Parameters, including bone volume (BV), mineralization density, trabecular thickness, trabecular number, and trabecular bone volume fraction (bone volume (BV)/total volume (TV)) were analyzed within a volume of interest (VOI) that included 1 mm of the proximal and distal regions of the femur fracture.^{62, 63} After μCT scanning, fractured femurs were tested biomechanical properties. Torsional loading was employed to evaluate the strength of healed bones under nearly pure shear conditions.⁶⁴ The proximal and distal ends of the harvested femur specimens were secured in 6.35 mm² aluminum tubes using bone cement prepared according to the manufacturer's specifications (DePuy Endurance; Warsaw, IN). After potting, specimens were soaked in PBS at room temperature for 2 hours to allow for tissue rehydration and bone cement hardening. Samples were then mounted on an EnduraTec TestBench system (200 N mm torque cell; Bose Corp., Minnetonka, MN) and subjected to torsion until failure at a rate of 1°/s to determine ultimate torques.⁶³

RNA Isolation and Sequencing

Mice were euthanized 1 day, 4 days, and 7 days after injection, and both bone and marrow tissues were collected from fracture sites for cell isolation as described previously¹⁶. We selected day 3 post-fracture for drug treatment because this marks the shift in the inflammatory response: the M1/pro-inflammatory response begins to subside while the M2/anti-inflammatory response rises and starts mitigating inflammation.^{20, 31, 32} By day 7, the inflammation is projected to have receded due to the increased M2/anti-inflammatory activity, subsequently transitioning to the bone repair phase. Thus, day 10 emerges as an optimal juncture for assessing the resolution of inflammation and the commencement of osteogenesis. For RNAseq experiment, specifically, bone marrow tissue was treated with red blood cell lysis buffer for 5 minutes on ice at room temperature, and bone tissues were crushed into small fragments and digested in 0.7 mg/mL collagenase type I (Sigma) for 30 minutes. Single cell suspensions were obtained by passing both bone marrow and bone-associated cells through 100 μm cell strainers (Fisher). Fracture-associated cells from bone marrow and bone were centrifuged at 1200 rpm for 5 minutes to form cell pellets which was then placed into 350 μL TRK lysis buffer (Omega Bio-tek) containing β-mercaptoethanol (β-ME, 20 μL per 1mL) and stored at −80 °C, which were then submitted to the University of Rochester (Rochester, NY) Genomics Research Center (GRC) for RNA extraction, cDNA synthesis, and gene sequencing. Briefly, RNAs were extracted using RNeasy Plus Micro Kit (Qiagen, Valencia, CA) and extracted RNA concentration was determined with the NanoDrop 1000 spectrophotometer (NanoDrop, Wilmington, DE), and RNA quality was assessed with the Agilent Bioanalyzer 2100 (Agilent, Santa Clara, CA). The quantity and quality of the converted cDNA was determined using the Qubit Fluorometer (Life Technnologies, Carlsbad, CA) and the Agilent Bioanalyzer 2100 (Agilent, Santa Clara, CA), which was sequenced using an Illumina HiSeq2500v4 high-throughput DNA sequencer with approximately 35 million reads generated for each sample.

RNA Sequencing Data Analysis

Our bulk RNAseq follows a multi-perspective strategy⁶⁵ for Quality Control (QC) of RNAseq experiments (RNA quality, raw read data (FASTQ), alignment, and gene expression). Raw reads generated from the Illumina basecalls were demultiplexed using bcl2fastq version 2.19.1. Quality filtering and adapter removal were performed using FastP version 0.23.1.⁶⁶ Processed/cleaned reads were then mapped to the GRCm39/genecode M31 or GRCh38/genecode38 (Mouse OR Human) reference using STAR_2.7.9a.^{67, 68} Gene-level read quantification was derived using the subread-2.0.1 package (featureCounts) with a GTF annotation file (GRCm39/genecode M31 or GRCh38/gencode42).⁶⁹ Differential expression analysis was performed using DESeq2-1.34.0 with a P-value threshold of 0.05 within R version 3.5.1 (https://www.R-project.org/).⁷⁰ Heatmaps were generated using the heatmap⁷¹ or iDEP.96 (http://bioinformatics.sdstate.edu/idep96/). Package were given the rLog transformed expression values.⁷¹ Gene ontology analyses were performed using the EnrichR package.^72-74 Volcano plots and dot plots were created using ggplot2.⁷⁵

In Vivo Macrophage Polarization via Flow Cytometry

Experiment design with groups, time points, and cell harvesting akin to sample preparation for RNAseq analysis. After cell collection, cells pellets were resuspended in PBS and stained with Ghost Dye Violet 450 for cell viability for 30 minutes on ice in the dark, which was then resuspended in a flow buffer containing 2% FBS, and 0.01% trypan blue at a concentration of 1 × 10⁶ cells/ml. Cell surface markers were determined to evaluate macrophage polarization, Specifically, the cells were stained and incubated with rat anti-mouse antibodies (BV510-labeled CD38, PE-Cy7-labeled CD206, APC-labeled MHCII, BV650-labeled F4/80, Alexa Fluor 488-labeled Ly-6G, BV786-labeled Ly-6C, BUV396-labeled CD45, and PE-Dazzle 594-labeled CD163) for 30 minutes on ice in the dark. Samples were washed three times with 2 mL of flow buffer to remove nonspecific-bind antibodies. Finally, the cells were analyzed using a BD LSR II Flow Cytometer. Data was analyzed by FCS Express 7 and fluorescence minus one (FMO) were used as controls for gating strategy as described previously.¹⁹

Statistical Analysis

Data are presented as mean ± standard deviation, with sample sizes specified in the figure legends. Group differences were analyzed using unpaired Student's t-tests or one-way or two-way ANOVA, accompanied by Tukey's or Dunnett's posthoc testing, as indicated in the figure legends. For in vivo studies, 5-10 mice per condition were utilized based on a priori power analysis and preliminary in vivo data. Statistical significance was defined by a p-value ≤ 0.05. Statistical evaluations were conducted using GraphPad Prism 9 Software.

ABBREVIATIONS

DDS

drug delivery systems

TRAP

tartrate-resistant acid phosphatase

PSMA-b-PS

poly(styrene-alt-maleic anhydride)-poly(styrene)

GSK3β

glycogen synthase kinase-3 beta

NP

nanoparticles

TBP

TRAP binding peptide

MPS

mononuclear phagocyte system

μCT

microcomputed tomography

RNA-seq

RNA sequencing

RAFT

reversible addition–fragmentation chain transfer

DCT

4-cyano-4-dodecylsulfanyltrithiocarbonyl sulfanyl pentanoic acid

MA

maleic anhydride

Sty

styrene

GPC

gel permeation chromatography

NMR

nuclear magnetic resonance

DLS

dynamic light scattering

Mn

molecular weights

PDI

polydispersity index

HPLC

high-performance liquid chromatography

ECs

endothelial cells

BMDMs

bone marrow-derived monocytes

MSCs

mesenchymal stem cells

OBs

osteoblasts

qPCR

quantitative polymerase chain reaction

DOX

doxorubicin

MSN-Dox

doxorubicin-loaded mesoporous silica nanoparticles

C- CNPs

chitosan nanoparticles

PLGA

poly(lactic acid-co-glycolic acid)

ALP

alkaline phosphatase

ARS

Alizarin Red staining

sEVs

small extracellular vesicles

iNos

inducible nitric oxide synthase

CD206

the mannose receptor

OsteoMacs

osteal macrophages

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

DEGs

differentially expressed genes

COMP

cartilage oligomeric matrix protein

SOX6

SRY-box transcription factor 6

SOX5

SRY-box transcription factor 5

COL2A1

collagen type II alpha 1 chain

IBSP

integrin binding slaloprotein

IHH

Indian hedgehog signaling molecule

RUNX2

runt-related transcription factor 2

RUNX3

runt-related transcription factor

BMP2

bone morphogenetic protein 2

OPG

osteoprotegerin

RANKL

receptor activator of nuclear factor kappa-B ligand

IL17

interleukin 17

TNFα

tumor necrosis factor alpha

Th17

T-helper 17

IL11

interleukin 11

Scube1

Signal peptide, CUB domain and EGF like domain containing 1

Gli3

GLI Family Zinc Finger 3

Mir100hg

Mir-100-Let-7a-2-Mir-125b-1 Cluster Host Gene

Notch3

Neurogenic locus notch homolog protein 3

Nos1

neuronal nitric oxide synthase

NGF

nerve growth factor

M-CSF

macrophage colony-stimulating factor

IFNγ

interferon gamma

IL-4

interleukin 4