In vivo validation of spray-dried mesoporous bioactive glass microspheres acting as prolonged local release systems for BMP-2 to induce bone regeneration

Despite years of diligent research in fracture healing, an unmet clinical need for safe and effective pharmacological treatments to improve bone regeneration persists with 10 – 20 % of fracture cases exhibiting impaired healing. Bone morphogenetic protein-2 (BMP-2) is a known key mediator of physiological bone regeneration and is clinically approved for selected musculoskeletal interventions. Yet, broad usage of this growth factor is impeded due to side effects that are majorly evoked by high dosages and burst release kinetics. In this study, mesoporous bioactive glass microspheres (MBGs) produced by an aerosol assisted spray-drying, scalable process were found to be biocompatible and to induce a pro-osteogenic response on human MSCs in vitro. Loading of the MBGs with BMP-2 resulted in prolonged, low-dose BMP-2 release without affecting the material features. In a pre-clinical rodent model, BMP-2 loaded MBGs significantly enhanced bone formation and influenced the microarchitecture of newly formed bone. The MBG carriers alone performed equal to the untreated (empty) control in most parameters tested, while additionally exerting mild pro-angiogenic effects. Using MBGs as a biocompatible, pro-regenerative carrier for local and sustained low dose BMP-2 release could limit side effects, thus enabling a safer usage of BMP-2 as a potent pro-osteogenic growth factor.


21
Bone is among the few tissues in the adult mammalian organism that can fully regenerate and restore its 22 physiological function (restitutio ad integrum) [1]. However, in the clinical routine, 10-20 % of fracture 23 bone defects [20,21]. Both, the ACS serving as a carrier as well as the supraphysiological dose of BMP-2, 1 natively occurring in cortical bone in the range of 1-2 µg/kg [22], can be considered suboptimal [15]. 2 Strikingly, the ACS itself has been found to significantly affect bone healing in a pre-clinical study, tested 3 with the commercially available Helistat® (Xemax Surgical Products, Napa, CA) that is used for the 4 administration of rhBMP-2. The ACS was found to cause strong immune responses as well as to impact the 5 osteogenic potential and viability of human mesenchymal stromal cells (hMSCs) [23]. In the same study, 6 another commercially available ACS (Lyostypt®, B. Braun, Germany) that yielded less severe in vitro 7 immune and hMSC responses was observed to impair callus mineralization upon implantation in a 0.7 mm 8 osteotomy gap in mice. Applying a collagen sponge in a pre-clinical model of impaired fracture healing by 9 creating a critical-sized defect, the authors of the present study confirmed this finding (Supplemental figure 10 S1). Moreover, it is known that collagen sponges possess rather unfavorable release kinetics, including low 11 drug retention and high burst release [24,25]. The poor release characteristics result in supraphysiological 12 dosages that have to be applied to render BMP-2 available in sufficient amount at the implantation site 13 despite rapid diffusion from implants and short half-life caused by proteolytic degradation [24,25]. Thus, 14 following the famous quote from Paracelsus "All things are poison, and nothing is without poison, the 15 dosage alone makes it so a thing is not a poison" [26], it was postulated that the supraphysiological BMP-2 16 dosages are key contributors to the side effects [17]. Therefore, it has been suggested to realize strategies to 17 apply minimal effective BMP-2 dosages [15]. 18 Taken together, a reduction in dosage and/or an improvement of release kinetics can potentially dampen the 19 side effects [25,27] and lay the foundation for a broader and safer use of BMP-2. Accordingly, a carrier 20 that features low burst and prolonged release, leading to lower effective dosages per time interval, as well 21 as possesses inherent pro-regenerative properties, low immunogenicity and appropriate degradation kinetics 22 would be an ideal candidate [18,28]. 23 In the current study, we analyzed mesoporous bioactive glasses, produced in the form of microspheres by 24 aerosol assisted spray-drying method (SD-MBGs) [29], as an alternative promising carrier for BMP- 2 25 instead of the currently used ACS. We selected SD-MBGs due to their intrinsic excellent bioactive behavior 1 and related pro-regenerative potential, along with the high exposed surface area (approx. 200 m 2 /g) and 2 regular nanopores (8-10 nm), which allow the storage and the release of active agents, such as drugs [30] or 3 biomolecules [31]. In addition, the scalable and reproducible production route of SD-MBGs [29], and the 4 possibility to impart multi-functionality by enriching the composition through the incorporation of selected 5 therapeutic elements (e.g. strontium [32,33]), enables a wide range of applications and can facilitate the 6 clinical translation of the proposed carrier. Accordingly, we hypothesized that the SD-MBGs could act as a 7 suitable carrier for BMP-2 in the bone healing context, exhibiting superior properties than clinically used 8 carrier systems. We found that SD-MBG loading of rhBMP-2 did not alter the spherical morphology and 9 the framework composition of SD-MBG. The release experiment of BMP-2 showed a low burst and a 10 sustained release profile over the entire test interval. Overall, this release kinetics led to low dosages of 11 solute growth factor in the range of 0.5 to 1 µg released BMP-2 over 14 days, which translates into 1-2 % 12 of the clinically applied BMP-2 dosage (in humans 12 mg per application, based on comparison of 13 dosage/kg). After validation of the carriers' biocompatibility and pro-osteogenic effect on hMSCs and 14 human blood, the carrier with and without BMP-2 load was embedded in an autologous blood clot, acting 15 as a place-keeper for the MBG microspheres without delaying the progression of healing [18], and placed 16 into a 2 mm osteotomy gap of a pre-clinical aged rat model of compromised healing [34]. Bone healing 17 outcome was investigated radiologically, histologically and immunohistochemically, showing superior 18 healing in the BMP group, while proving the suitability of the SD-MBG carriers alone to be utilized for 19 bone regeneration purposes. With this approach we aim to provide evidence for the effectiveness and 20 biocompatibility of bioactive glass-based BMP-2 carriers, potentially allowing more patients suffering from 21 different fracture cases to benefit from the vast pro-osteogenic potential of this growth factor. 2. hMSCs respond with higher osteogenic potential to SD-MBGs ionic extracts 13 After confirming the high biocompatibility of SD-MBGs on primary human MSCs, we aimed to validate 14 the pro-osteogenic potential of the material on the same cell source. For this, osteogenesis was induced in 15 vitro by the addition of pro-osteogenic additives to the medium. Similar to the biocompatibility tests, SD-16 MBGs were added to transwells and the hMSCs were exposed to ionic extracts from SD-MBGs over the 1 entire course of the experiments. Using the same initial concentrations of SD-MBGs as in the 2 biocompatibility experiments (1.5 and 5 mg/mL), matrix mineralization was studied by Alizarin Red 3 staining after 10 and 14 days (Figure 2A) while collecting the supernatant before each medium change to 4 quantify the amount of free phosphate released into the medium ( Figure 2B). Matrix mineralization relative 5 to the cell number was significantly increased when cells were exposed to the higher concentration (c2) of 6 SD-MBGs, while the amount of free phosphate did not vary strongly across groups. hMSCs from three 7 donors, previously tested and selected for their reduced osteogenic differentiation capacity (Supplementary 8 figure S3), were exposed to the higher concentration of SD-MBGs. At day 14, hydroxyapatite deposits and 9 cell nuclei were stained using OsteoImage and DAPI ( Figure 2C). The intrinsically slow mineralization 10 potential of the selected hMSCs was confirmed by the low amount of OsteoImage-positive areas (staining 11 hydroxyapatite, green dots, marked with arrows). The mineralization-boosting effect of SD-MBGs in turn 12 was evidenced by the substantially more pronounced OsteoImage signal. Altogether, these findings 13 underline the pro-regenerative/ pro-osteogenic capacity of SD-MBG, making it an attractive candidate as a SD-MBGs were successfully loaded with BMP-2 by incipient wetness impregnation method, as confirmed 10 by differential thermal analysis (DTA) ( Figure 3A). While no peaks were observed for the SD-MBG 11 thermogram, SD-MBG loaded with BMP-2 (SD-MBG + BMP-2) showed a characteristic endothermic peak 12 ascribed to the protein denaturation [35]. Additional proof for successful incorporation of BMP-2 was 13 provided by thermogravimetric analysis (TGA) ( Figure 3B). A negligible weight loss in the range of  150 °C ascribed to the release of the surface adsorbed water was observed for SD-MBG, validating the 15 absence of any residual organic compounds. In comparison, a significantly higher weight loss in the same 16 temperature range was detected for the SD-MBGs + BMP-2, due to the release of water bound to the BMP-17 2 protein. The additional weight loss in the temperature range between 200 and 400°C, observed exclusively 18 for the BMP-2 loaded material, can be assigned to the decomposition of BMP-2. Moreover, zeta-potential 19 analysis showed a negative surface charge (-24.5 ± 1.7) for SD-MBG particles alone when suspended in 20 water, which resulted to be more negative after BMP-2 loading (-32.1 ± 1.9 for BMP-2).  figure S4), in fair agreement to the data previously published [29]. Energy-2 dispersive X-ray spectroscopy (EDS) spectra of both particles ( Figure 3D, dry particles pre-soaking in SBF)) 3 revealed a Si/Ca molar ratio in good agreement with the nominal one. The FE-SEM observations and EDS 4 analysis of dry material powders ( Figure 3C-D) evidenced that the BMP-2 loading into the mesopores does 5 not alter the morphological features and the chemical composition of the MBG microparticles. In particular, 6 the Si/Ca molar ratio revealed by EDS before and after the BMP-2 incorporation ( Figure 3D) remained 7 unaffected, indicating that the loading procedure did not induce substantial ion release. 8  in vitro bioactivity test, values that allow osteoblasts to maintain their physiological activity [40]. Taken 20 together, BMP-2 was found to be successfully loaded into the mesopores of the SD-MBGs without altering 21 the morphology, chemical composition or the intrinsic bioactivity of the carrier.  1 The release kinetics of BMP-2 from the SD-MBG microspheres was investigated by soaking the BMP-2 2 loaded SD-MBG in either PBS or Tris-HCl with a physiological pH of 7.4 at 37 °C for 14 days. While the 3 phosphate ions contained in PBS allow hydroxyapatite deposition, Tris-HCl medium prevents 4 hydroxyapatite formation due to a lack of phosphates. Hence, an effect of hydroxyapatite formation on 5 BMP-2 release kinetics can be clearly identified. The supernatant was collected repeatedly, serving as 6 samples for BMP-2 quantification by means of enzyme-linked immunosorbent assay (ELISA) ( Figure 4A). 7

SD-MBGs enable prolonged BMP-2 release
Overall, irrespective of the elution buffer utilized, a prolonged and sustained release of low amounts of 8 BMP-2 was observed over the entire testing interval of 14 days, without an initial burst release.  blinded fashion, asking them to categorize the fracture gaps showing "no bridging", "partial bridging" or 22 "complete bridging", a scoring previously described [43]. Within the BMP-2 group, complete bridging 23 occurred in 50% of cases, in the other 50%, at least partial bridging was observed. For the vast majority of 24 all other animals, no bridging was seen (Table 1). 25 The analysis of ex vivo µCT data ( Figure 5), which allows higher resolution imaging than the in vivo µCT, 6 confirmed the significantly higher bone and tissue formation for the SD-MBG + BMP-2 compared to all 7 other groups ( Figure 5A). The ratio of bone volume within the callus volume (BV/TV) was not significantly 8 affected by the BMP-2 treatment, since both BV and TV were elevated in this group. Bone mineral density 9 (BMD) for the BMP-2 group was found to be decreased ( Figure 5A) in the volume of interest (VOI), which 10 also includes cortical bone fragments. This can be explained by the larger callus volume that was formed in 11 the BMP-2 group, resulting in more newly formed bone when compared to the other groups with less callus 12 volume, where the cortical bone represents a higher percentage of the total volume. However, upon 13 excluding the cortical bone fragments, no significant change in callus BMD could be observed between 14 groups (in vivo µCT data, Supplementary figure S5C). Despite similar ratios of bone volume within the 15 callus volume (BV/TV) in all groups, the mineral content within the callus volume, also called bone mineral 16 content (BMC), is significantly increased upon BMP-2 treatment, indicating that a higher net amount of 17 deposited hydroxyapatite can be found in the formed callus ( Figure 5A). Comparing the SD-MBG groups 18 with and without BMP-2 in respect to their microarchitecture, BMP-2 release into the fracture area 1 decreased the trabecular thickness (Tb.Th.) and increased the trabecular number (Tb.N.), thus inducing a 2 callus with a finer microarchitecture, but increased trabecular branching ( Figure 5B). The polar minimal 3 moment of inertia (MMI (polar)), a 3D computational calculation of torsional stability, showed a 4 significantly higher value for the BMP-2 group, indicated a higher torsional stiffness as a consequence of 5 the progressed healing under BMP-2 influence ( Figure 5C). Examples for reconstructed µCT images are 6 depicted in Figure 5D. 7 Upon comparing the BC alone and in combination with SD-MBGs, no radiological difference could be 16 determined ( Figure 5). Similarly, the comparison of either the BC alone or BC + SD-MBG with the empty 17 group, did not yield significant differences. Thus, it can be assumed that neither the BC nor the addition of 18 SD-MBG impair the healing outcome, rendering the tested hybrid system a suitable drug release platform.

12
In this study, we validated spray-dried mesoporous bioactive glass microspheres (SD-MBG) [29], as a 13 suitable carrier for prolonged, low-dose BMP-2 release, exerting beneficial effects on the bone healing 14 outcome. In contrast to the conventional ACS as BMP carrier [23], we report for SD-MBG an excellent 1 biocompatibility without a negative impact on bone formation when placed in the bone osteotomy gap. 2 Moreover, cell vitality was not affected by SD-MBG when tested on primary human MSCs and the pro-3 inflammatory response from human whole blood was negligible. Ionic dissolution products of SD-MBG 4 were revealed to even amplify the osteogenic differentiation of hMSCs in vitro, indicating the potential of 5 being pro-regenerative in the context of bone healing. Loading SD-MBG with BMP-2 did not induce any 6 effect on the material morphology and composition. SD-MBG showed no initial burst release while retaining 7 a sustained low-dose release in the range of 1-2 % of clinically applied BMP-2 over the entire testing interval 8 of 14 days (approximately 0.5 -1 µg compared to 50 µg in vivo dosage in a rat osteotomy model [41,42]). 9 When applied in the fracture gap of a pre-clinical animal model of compromised healing [34], pure SD-10 MBG did not impair the healing progress, but rather exerted mild pro-angiogenic effects. The additional 11 BMP-2 load was found to improve the healing outcome in all tested bone healing parameters, indicating 12 that SD-MBG microspheres represent a suitable carrier and BMP-2 release platform for impaired bone 13 healing scenarios, with the carrier possessing an intrinsic pro-regenerative potential ( Table 2). 14 15 Table 2 Initial clinical studies investigating the effects of BMP-2 administered via an ACS, including the large, 1 multi-centered human "BMP-2 evaluation in surgery for tibial trauma" study (BESTT), showed superior 2 effects of the treatment due to decreased rates of revision surgeries and infections with a 1.5 mg/mL BMP-3 2 dosage [20]. Other studies confirmed the beneficial BMP-2 effects while extending the range of 4 application to other long bone defects [46]. As a result of these early studies, BMP-2 was increasingly 5 applied in the clinics. Subsequently, reports on side effects largely neglected in the initial studies appeared, 6 leading to a reassessment of the treatment's safety and efficacy. Carragee et al. estimated that the treatment 7 risk is 10 to 50 fold higher than originally anticipated [47]. There is vast agreement in the scientific and 8 clinical community that most of the treatment-related risks arise from the supraphysiological dosage that 9 needs to be applied profile [15]. Additionally, it is of outmost importance that a carrier exhibits high retention capacities and 7 localizes the drug at the target site [12]. 8 Bearing these criteria in mind, we employed SD-MBG with SiO2-CaO binary composition, mainly due to 9 its excellent biocompatibility and sustained degradation upon exposure to physiological fluids [38, 52]. 10 Enhancement of the pro-regenerative potential due to ionic dissolution products from bioactive glasses has 11 been demonstrated on various cell types [38, 39, 53] and is highly dependent on the ionic composition, size 12 and interaction mode [39,54]. In this study, we were able to assess the SD-MBG intrinsic pro-osteogenic inflammatory effects for example during their degradation as described for instance for poly(d,l-lactic-co-23 glycolic Acid, PLGA) [59, 60]. In the current study, a low-dose, sustained release profile of BMP-2 without 24 burst effect from SD-MBG mesoporous structure, was found to effectively induce superior bone healing 25 compared to the carrier alone. In vitro, a maximum cumulative release over 14 days of approximately 1 26 µg/mL was measured. In another rat study investigating bone healing, 1 µg was observed to be ineffective 1 in promoting bone healing if administered via a collagen sponge [25], again highlighting the importance of 2 proper release kinetics. The obtained release profiles ( Figure 4A) depend on multiple factors, such as the 3 occurrence of multiple interactions between proteins and internal pore surface, hydroxyapatite deposition 4 partially blocking SD-MBG mesopores, as well as the overall morphological features. In particular, BMP-5 2 loading was driven by adsorption on MBG surface through the engagement of intermolecular interactions 6 (e.g. mainly H-bonding) between surface -OH species and BMP-2 protein functionalities (-NH2, -COOH) 7 as was described for BMP-2 binding to hydroxyapatite [61]. Furthermore, SD-MBG possess a negative 8 surface charge, due to deprotonated silanols, able to bind positively charged protein molecules through 9 electrostatic interactions. This charge-dependent binding was reported for BMP-2 binding the kerateine [62] figure S7). Lastly, the size of the spheres is essential in determining the 22 degradation kinetics [39] since particle and pore size define the total exposed surface and thus the loading 23 capacity. 24 The SD-MBG were embedded in an autologous blood clot for in vivo application and validation. This hybrid 1 formulation can be easily produced even in the clinical context. In a previous study, the importance of the 2 fracture hematoma for a successful fracture healing has been described [64], the autologous blood clot can 3 be considered as an artificial but similar tissue when compared to the initial fracture hematoma. Both tissues 4 are derived from blood, therefore contain similar cellular species, and underwent the process of coagulation. 5 Other biomaterial-based approaches oftentimes spatially limit the formation of a hematoma due to 6 hindrance/blockage of new tissue formation in the fracture area. In this approach however, the hematoma- applying the aerosol spray-drying assisted method under mild aqueous conditions represents a scalable, cost 10 and safety effective approach [29]. The combination of SD-MBG with an autologous blood clot for in vivo 11 application has proved successful and could easily be translated to the clinical routine. These findings 12 revealed a translatable biomaterial-based approach to limit side effects of BMP-2 usage by dampening 13 excessive amount of soluble BMP-2 as observed for clinically employed collagen sponges, with the MBG 14 possessing intrinsic characteristics that are beneficial for bone healing.

Synthesis of MBG Samples by Aerosol-Assisted Spray Drying Method 2
Based on the procedure reported by Pontiroli et al. [29], MBG micro-particles with a binary SiO2-CaO 3 composition (molar ratio Si/Ca= 85/15 hereafter named SD-MBG) were synthesized by aerosol-assisted 4 spray drying (SD) method. Briefly, 2.03 g of the non-ionic block copolymer Pluronic P123 were dissolved 5 in 85.0 g of double distilled H2O (ddH2O). In a separate batch, 10.73 g of TEOS were pre-hydrolyzed under 6 acidic conditions using 5.0 g of an aqueous HCl solution at pH= 2 until a transparent solution was obtained. 7 The TEOS solution was then dropped into the Pluronic P-123 solution and kept stirring for 30 minutes. 8 Thereafter, 1.86 g of calcium nitrate tetrahydrate were added. The final solution was stirred for 15 min and 9 finally sprayed with a Mini Spray-Dryer B-290 (Büchi Labortechnik, Flawil, Switzerland) using nitrogen 10 as the atomizing gas (inlet temperature 220 °C, N2 pressure 60 mmHg, feed rate 5 mL/min). The obtained 11 powder was calcined at 600 °C in air for 5 h at a heating rate of 1 °C/min using a furnace (Carbolite 1300 12 CWF 15/5, Carbolite, Hope Valley, UK), in order to remove the templating agent. in humans [20,21]. For this, an average human body weight of 75 kg was assumed, leading to an average 21 dose of BMP-2 of 0.16 mg/kg bodyweight. The body weight of the rats in this study was around 300 g, on powders dispersed on carbon tape by analyzing an area of 75x50 µm. 10 Thermogravimetric and differential thermal analysis and Zeta-potential 11 The successful loading of the protein was assessed by thermogravimetric and differential thermal analysis 12 (TG/DTA) and Zeta-potential measurements. TG/DTA were conducted on a STA 2500 Regulus (Netzsch- Osteogenic differentiation, phosphate and mineralization assays 3 To induce osteogenic differentiation, hMSCs were cultivated in 500 µL osteogenic differentiation medium 4 (OM), which consists of expansion medium supplemented with 100 nM Dexamethasone, 50 µM L-Ascorbic 5 acid 2-phosphate sesquimagnesium salt hydrate, and 10 mM ß-Glycerophosphate disodium salt hydrate (all: 6 Sigma Aldrich, St. Louis, USA). Prior to each medium change, supernatants were collected, centrifuged and 7 cell-/debris-free supernatants were stored at -80°C. The OM (+/-exposure to SD-MBG ionic dissolution 8 products) supernatants were diluted 1:700, the EM controls 1:100 to be used in the phosphate assay (Abcam, 9 Cambridge, United Kingdom). At 10 and 14 days of cultivation in OM, cells were washed with PBS, fixed 10 in 4% neutral buffered formaldehyde (VWR, Darmstadt, Germany), cell number was determined by DAPI 11 staining (Sigma-Aldrich, St. Louis, USA) as mentioned above. 0.5% w/v Alizarin Red S (Sigma-Aldrich, 12 St. Louis, USA, 10 min incubation, RT) in distilled water was used to stain the mineralized extracellular 13 matrix (ECM), thorough washing with distilled water removed unbound Alizarin Red S. For quantification, 14 10% w/v cetylpyridinium chloride (Sigma-Aldrich, St. Louis, USA) in distilled water was added to the air-15 dried wells and kept on an orbital shaker for 45 min, followed by an absorbance measurement of the 16 dissolved stain at 562 nm. As an additional method to visualize deposited hydroxyapatite, an OsteoImage 17 Assay (Lonza, Basel, Switzerland) was carried out. After washing, a DAPI staining to allow detection of 18 cell nuclei was performed as described above. The cell cultures were imaged using a fluorescent microscope 19 (BZ-X810, Keyence, Osaka, Japan) at 10x magnification. 20

21
In vivo bone healing study 22 Housing conditions, osteotomy surgery and study design 23 The bone regeneration potential of the different treatments was studied in a rat osteotomy model of delayed 24 healing as previously described [34]. A total of 18 adult female Sprague-Dawley rats (aged >7 months, ≥300 25 g, Janvier Labs, Le Genest-Saint-Isle, France), that had more than three litters (ex-breeders) were included 26 in this study. Rats were kept in small groups under obligatory hygiene standards and conventional housing 1 conditions with controlled temperature set to 20 ± 2 °C, a light/dark period of 12 h and food and water being 2 available ad libitum. All animal experiments were approved by the local animal protection authorities 3 (Landesamt für Gesundheit und Soziales Berlin, Germany: G0258/18) and performed in accordance with 4 the German Animal Welfare Act, the National Institutes of Health Guide for Care and Use of Laboratory 5 Animals and the ARRIVE guidelines. 6 Before starting the surgery, the rats were anesthetized by inhalation of isoflurane (Forene, Abott, 7 Wiesbaden, Germany) and received a potent analgesic (Bubrenorphine, RB Pharmaceuticals, Berkshire, 8 United Kingdom; 0.1 mg/kg BW), an antibiotic bolus (Clindamycin, Ratiopharm, Ulm, Germany; 45 mg/kg 9 BW) and eye ointment. The osteotomy was carried out under deep anesthesia on a heating plate set to 37°C. 10 The operation area of the left femur was clipped and disinfected, the femur was exposed by a longitudinal 11 skin incision and blunt preparation of the muscles. An external fixator (RatExFix, RISytem, Davos, 12 Switzerland) was mounted on the femur, followed by creation of a 2 mm osteotomy using an oscillating saw 13 (W&H, Bürmoos, Austria) and a saw guide. The wound was closed with sutures and the rats were returned 14 to their cages. As post-operative analgesia, Tramadolhydrochloride (Grünenthal, Aachen, Germany; 0.5 15 mg/mL) was added to the drinking water for three days post-surgery. At the end of the staining, the slides were dehydrated using Xylol (Fisher Chemical, Thermo Fisher 1 Scientific, Waltham, USA) and embedded with Vitroclud (Langenbrink, Emmendingen, Germany). 2 For the immunohistochemical analysis, after deparaffinization and re-hydration, slides were blocked using 3 5 % normal horse serum (Vector Laboratories, Burlingame, USA) for 1 h and 1 % BSA/PBS, followed by 4 overnight incubation at 4 °C with α-α-SMA (1:400, mouse monoclonal, clone 1 A4, DAKO Agilent 5 Technologies, Santa Clara, USA) or α-CD68 (1:2000, mouse monoclonal, clone BM4000, OriGene 6 Technologies, Rockville, USA). An α-mouse, rat adsorbed biotinylated secondary antibody (Vector 7 Laboratories, Burlingame, USA) diluted 1:50 in 2 % normal serum horse and 1 % BSA/PBS was incubated 8 on the slides for 30 min. AB complex (Vector AK 5000, Vector Laboratories, Burlingame, USA) was 9 incubated for 50 min, then the milieu was slightly alkalized by using a chromogen buffer (pH 8.2), followed 10 by the visualization of the staining (Vector SK 5100, Vector Laboratories, Burlingame, USA). As 11 counterstaining, hematoxylin (Mayer's) was chosen and the slides were embedded using Aquatex (Merck, 12 Darmstadt, Germany). Microscopic images of all slides were taken at 10x magnification under bright field 13 (Axioskop 40, Carl Zeiss, Oberkochen, Germany). Histomorphometric analyses were carried out using the 14 MOVAT's pentachrome-stained slides and applying a custom-made macro embedded in FIJI ImageJ 15 Software [69]. Callus area was determined manually. Detection of mineralized tissue and cartilage was 16 performed according to color thresholding to determine the areas of the respective tissues. Blood vessels 17 and osteoclasts were revealed in an analog manner by α-SMA and CD68 staining respectively. Finally, the 18 amount of blood vessels was normalized to the total area of the callus and the length of mineralized callus 19 surface covered by CD68+ cells to was normalized to the total length/ perimeter of the mineralized callus. 20 21

22
The statistical evaluation of the presented data was performed using GraphPad Prism® (GraphPad Software, 23 San Diego, USA). Confidence interval was set to 0.95, p-values for statistical significance were *p < 0.05, 24 **p < 0.01, ***p < 0.001, ****p < 0.0001. Detailed information on all statistical analyses performed, 25 including statistical tests, depicted values and sample size, are mentioned in the figure legends. In general, 26 for the small sample sizes of the in vivo preclinical study, the data can not be considered normally 1 distributed. Accordingly, the statistical test applied was a Mann-Whitney U test. For larger sample sizes that 2 were tested for normal distribution, an ANOVA with Tukey's multiple comparison test was performed.