In total, 44 CD-1 mice (26 male and 18 female mice) with a body weight of 40 ± 7 g and an age of 130 ± 5 days were used. The animals were bred at the Institute for Clinical and Experimental Surgery (Saarland University, Homburg, Germany), kept at a regular 12 h (h) light and dark cycle and had free access to tap water and standard pellet food (Altromin, Lage, Germany).

The study was conducted in accordance with the German legislation on protection of animals and the NIH Guidelines for the Care and Use of Laboratory Animals and was approved was approved by the local authorities (permission number: 35/2020; State Office for Consumer Protection, Saarbrücken, Germany).

For the present study a well-established ischemic murine fracture model was used, as described previously in detail (Menger et al., 2022). For the surgical procedure, the mice were anesthetized by an intraperitoneal injection of ketamine (75 mg/kg body weight; Pharmacia, Erlangen, Germany) and xylazine (25 mg/kg body weight; Bayer, Leverkusen, Germany). Briefly, a 6-mm incision was performed medial parapatellar at the right knee in the direction of the femoral artery and vein (Figure 1A). Moderate hind limb ischemia was induced by ligation of the right deep femoral artery at the thigh. The artery was ligated using a 6–0 suture (black silk 6-0, non-absorbable; Ethicon, Raritan, United States). The knee joint capsule was then opened by incision medial to the patella, and the femoral condyles were exposed by lateralization of the patella. After drilling a hole (0.5 mm in diameter) into the intracondylar notch, an injection needle with a diameter of 0.4 mm was drilled into the intramedullary canal. Subsequently, a tungsten guidewire (0.2 mm in diameter) was inserted through the needle into the intramedullary canal. After removal of the needle, the femur was fractured by a three-point bending device. An intramedullary titanium screw (diameter: 0.5 mm; MouseScrew™, RISystem AG, Davos, Switzerland) was implanted over the guidewire to stabilize the fracture. Fracture and implant position were confirmed by radiography.

Animals were randomly assigned to one of two study groups. Animals of the metamizole group (n = 23) received 50 mg/kg metamizole (Novaminsulfon; Winthrop Arzneimittel, Mülheim-Kärlich, Germany) daily per os from the day of surgery. Animals of the control group (n = 21) received an equivalent volume of the vehicle (NaCl 0.9%; Braun, Melsungen, Germany).

Animals were sacrificed by cervical dislocation after 2 weeks (n = 13 in the control group; n = 15 in the metamizole group) or 5 weeks (n = 8 in each group) postoperatively. Directly prior to sacrifice, X-rays of the operated femurs were taken to exclude secondary dislocation of the metallic implants (Figure 1B). Femurs were harvested and used for further analyses. During outcome assessments, blinding was employed for unbiased analyses.

To detect whether metamizole concentrations in mice are comparable to those in humans, serum levels were analyzed using high-performance liquid chromatography (HPLC). For this purpose, blood samples of operated metamizole-treated animals (n = 16) were taken under general anesthesia 30 min (min), 60 min and 90 min after the last oral administration of metamizole. The analytical methodology was carried out as described previously (Domínguez-Ramírez et al., 2012; Giorgi et al., 2015; Giorgi et al., 2018). Briefly, a 15 mL polypropylene sample bottle was used for sample extraction. A volume of 0.5 mL plasma was supplemented with 100 µL internal standard solution (Sigma Aldrich, St. Louis, United States) and mixed thoroughly for 30 s by vortexing. Subsequently, 0.1 mL sodium hydroxide solution 1 N (Titrisol, Merck, Darmstadt, Germany) was added and mixed again. The sample was supplemented with 4 mL ethyl acetate:methylene chloride (3:7 v/v; both Merck). Mixing was repeated for 30 s, followed by swirling (60 oscillations/min for 10 min) and then the mixture was centrifuged for 10 min at 10°C at 10.956 × g in a centrifuge with a 5 cm rotor radius. The organic layer was removed in 3 mL and placed in another 15 mL bottle. It was then dried at 40°C under nitrogen vaporization and redissolved with 100 µL of the mobile phase. Finally, 50 µL of this solution was analysed by HPLC. For both active metabolites of metamizole, 4-MAA and 4-AA, the detection limit was 0.03 μg/mL and the limit of quantification was 0.05 μg/mL. The concentrations for curve calibration were set at 0.05–0.1–0.25–0.5–1–5 and 10 μg/mL for both substances. The range of optimal linearity (regression lines) was defined for 4-MAA in the width of y = 0.1749x - 0.0547 (r ² = 0.9989) and for 4-AA in the range of y = 0.3291x - 0.0293 (r ² = 0.9995). The recovery rate for 4-MAA ranged from 90.1%–95.2% and for 4-AA from 93.9%–97.1%. The intraday and interday precision (CV%) were lower than 8.1% and 6.8%, and 9.9% and 9.4% for 4-AA and 4-MAA, respectively.

For the biomechanical analysis, the fractured right and healthy left femora of both groups were resected at 2 weeks (n = 8 in the control group; n = 10 in the metamizole group) and 5 weeks (n = 8 in each group) and freed from soft tissue. After removing the implants, callus stiffness was measured with a non-destructive test using a three-point bending device (Mini-Zwick Z 2.5; Zwick, Ulm, Germany), as described previously (Orth et al., 2022). Loading was stopped individually in every case when the actual load–displacement curve deviated more than 1% from linearity. To ensure the reproducibility of the procedure, the bone was positioned with the ventral side facing downwards, the femoral head aligned to the right, and the contact stamps centered at a distance of 6 mm from each other. The bending stiffness [N/mm] was calculated from the linear elastic part of the load-displacement diagram after applying a gradually increasing bending force with 1 mm/min. The unfractured left femora were also analyzed, serving as an internal control to account for differences in bone stiffness of metamizole-treated animals when compared to controls. All values of the fractured femora are given as absolute values and in percent of the corresponding unfractured femora. Using this non-destructive approach for biomechanical analyses, the femurs could also be used for subsequent micro-computed tomography (µCT) and histological investigations, resulting in a marked reduction of required animals according to the 3R principle.

X-rays (MX-20 Faxitron; X-ray Corporation, Wheeling, IL, United States) of the fractured femora were performed 2 weeks (n = 8 in the control group; n = 10 in the metamizole group) and 5 weeks (n = 8 in each group) after surgery. Fracture healing was analyzed according to the classification of Goldberg, with stage 0 indicating radiological non-union, stage 1 indicating possible union and stage 2 indicating radiological union, as described previously (Goldberg et al., 1985).

Moreover, µCT of the fractured femurs was performed 2 weeks (n = 8 in the control group; n = 10 in the metamizole group) and 5 weeks (n = 8 in each group) after surgery. Scanning was performed at a spatial resolution of 9 μm with a standardized setup (Skyscan 1,176; Bruker, Billerica, United States), as described previously (Orth et al., 2022; Orth et al., 2017). To express grey values as mineral content (bone mineral density; BMD), calcium hydroxyapatite (CaHA) phantom rods with known BMD values were used for calibration. On each transversal slide the region of interest (ROI) was contoured manually defining exclusively novel bone and excluding original cortical bone. The ROI was processed with a threshold procedure (CTAnalyzer, Bruker), which allowed for differentiation between bone and soft tissue. The thresholds to distinguish between bone and soft tissue were based on visual inspection of the images, qualitative comparison with histological sections and previous studies investigating bone repair and callus tissue by µCT (Orth et al., 2022; Morgan et al., 2009; Isaksson et al., 2009). A BMD with more than 0.410 g/cm³, resulting in grey values of 68–255 was defined as total mineralized bone. The following µCT parameters were calculated from the callus ROI for each specimen: bone volume (BV; mm³), tissue volume (TV; [mm³]), ratio of BV/TV (%) and trabecular parameters, such as trabecular number (TbN; [1/mm]), trabecular separation (TbSp; [mm]) and trabecular thickness (TbTh [mm]).

For histomorphometric analyses, bones were fixed in 4% phosphate-buffered formalin for 24 h and decalcified in ethylenediaminetetraacetic acid (EDTA) solution for 14 days. Dehydration was carried out in an ascending alcohol series. After embedding decalcified bones in paraffin, longitudinal sections with a thickness of 5 µm were stained with Safranin-O (at 2 weeks: n = 8 in the control group; n = 10 in the metamizole group; at 5 weeks: n = 8 in each group). At a magnification of 12.5 × (Olympus BX60 Microscope; Olympus, Shinjuku, Japan; Zeiss Axio Cam and Axio Vision 3.1, Zeiss, Jena, Germany) structural indices were calculated based on recommendations as described elsewhere (Gerstenfeld et al., 2005). For histomorphometric evaluation the following parameters were measured: (i) total periosteal callus area, (ii) bone callus area, (iii) ratio of bone tissue area/total callus area. The total periosteal callus area was defined as all osseous, cartilaginous and fibrous callus tissue outside of the cortices. Pre-existing cortical bone and endosteal callus formation were excluded. Each area was marked and calculated using the ImageJ Analysis System (NIH, Bethesda, United States).

Protein expression within the callus tissue was determined by Western blot analyses, including the expression of the angiogenic markers Cyr61, CD31, the osteoclast markers osteoprotegerin (OPG) and receptor activator of NF-κB ligand (RANKL), the osteogenic marker Runt-related transcription factor 2 (RUNX2) and the proliferation marker proliferating cell nuclear antigen (PCNA). After harvesting callus tissue 2 weeks after surgery (n = 5 in each group), tissue samples were transferred in lysis buffer and stored at −80°C. After saving the whole protein fraction, proteins were separated and transferred to membranes by standard protocols and probed using anti-Cyr61 (AF4055; R&D Systems, Minneapolis, United States), anti-CD31 (77,699; Cell Signaling Technology Europe, Frankfurt, Germany), anti-OPG (MAB4591; R&D Systems), anti-RANKL (ab62516; Abcam, Cambridge, United Kingdom), anti-RUNX2 (EPR22858-106; Abcam) and anti-PCNA (HRP-60097; Proteintech, Planegg-Martinsried, Germany) antibodies. All antibodies were incubated overnight at 4°C and afterwards for 4 h at room temperature. The appropriate peroxidase-conjugated anti-IgG antibodies served as secondary antibodies (R&D Systems and Dako Agilent, California, United States). Protein expression was visualized by means of luminol-enhanced chemiluminescence after exposure of the membrane to the Intas ECL Chemocam Imager (Intas Science Imaging Instrument GmbH, Göttingen, Germany). To correct for unequal loading, signals were normalized to β-actin signals (Santa Cruz Biotechnology, Heidelberg, Germany).

All data are given as means ± standard error of the mean (SEM). The needed sample size was initially computed. For parametric data (Shapiro-Wilk test), the comparison between two groups was carried out using Student’s t-test after determining the equality of variance (Brown-Forsyte test), while analyses of three groups (serum analysis) were performed by one-way ANOVA, including the correction of the α-error according to Bonferroni probabilities to compensate for multiple comparisons. For non-parametric data, the comparison between two groups was carried out using the Mann-Whitney U-test, while analyses of three groups (serum analysis) were performed by one-way ANOVA on Ranks, followed by a Dunn’s test for all pairwise comparisons. The statistical analyses were performed using the SigmaPlot software 13.0 (Systat Software GmbH, Erkrath, Germany). A p-value <0.05 was considered to indicate significant differences.

HPLC revealed high levels of 4-MAA at all investigated time points after administration of metamizole (Figure 2A). 4-AA was found to be low at 30 and 60 min and significantly higher concentrated at 90 min after administration (Figure 2B).

Femurs of the metamizole and control group presented with a low bending stiffness at 2 weeks after surgery and a higher stiffness at 5 weeks after surgery (Figure 3). Intragroup comparisons revealed a significant increase of bending stiffness at 5 weeks after surgery compared to results at 2 weeks (Figures 3A,B). No significant differences could be observed between both groups for fractured femurs at 2 weeks (Figure 3A) and 5 weeks (Figure 3B) and for healthy controls (Figures 3C,D). The ratio of biomechanical stiffness between fractured and unfractured bones was very low at 2 weeks and indicated an incomplete osseous stability of fractured femurs at 5 weeks after surgery in both groups (Figures 3E,F). This shows a delayed bone healing in the herein used ischemic fracture model. These results are in line with a previous study using this animal model (Menger et al., 2022).

X-rays of animals of the control and metamizole group showed signs of ongoing healing throughout the study period (Figures 4A–D). Corresponding to the X-rays, the mean Goldberg score at 2 weeks after fracture was 0.80 ± 0.13 for metamizole-treated animals and 0.75 ± 0.16 for control animals, whereas at 5 weeks the mean score for metamizole-treated animals was 1.88 ± 0.12 and for control animals 2.0 ± 0.0.

µCT analyses showed no significant differences for BV and TV at both observation time points between the two groups (Figures 4E–H). Accordingly, the BV/TV ratio did also not differ between metamizole-treated and control mice (Figures 4I,J). However, the ratio showed a significant increase of bone tissue as a fraction of TV between week 2 and 5 after surgery for each group as an indicator for ongoing bone healing. µCT analysis of the trabecular structures exhibited no significant differences between both study groups for TbN (2 weeks: control: 1.91 ± 0.33; metamizole: 2.31 ± 0.27; 5 weeks: control: 2.59 ± 0.33; metamizole: 3.47 ± 0.17), TbSp (2 weeks: control: 0.55 ± 0.05; metamizole: 0.49 ± 0.06; 5 weeks: control: 0.25 ± 0.02; metamizole: 0.22 ± 0.01) and TbTh (2 weeks: control: 0.09 ± 0.01; metamizole: 0.10 ± 0.01; 5 weeks: control: 0.17 ± 0.02; metamizole: 0.15 ± 0.02).

The histomorphometric analysis 2 weeks after surgery demonstrated a large callus area with mostly lack of osseous bridging in both groups (Figures 5A,C). In contrast, 5 weeks after surgery bone tissue bridged the initial fracture site in both groups (Figures 5B,D). The total callus area did not differ between the two groups (Figures 5E,F). The total callus area was smaller after 5 weeks than after 2 weeks in each group as a typical sign for remodeling of bone. Accordingly, the fraction of bone tissue of the total callus area increased over time in both groups (Figures 5G,H), whereas intergroup comparisons of this parameter revealed no differences.

The Western blot analysis revealed a significantly higher expression of the pro-angiogenic factor Cyr61 in the callus tissue of metamizole-treated animals when compared to controls at 2 weeks after surgery (Figure 6A). The expression of CD31 did not differ between the two groups (Figure 6B). The osteoclastic marker RANKL exhibited an increased expression in the callus tissue of metamizole-treated animals, whereas expression of OPG did not show significant differences (Figures 6C,D). The expression of RUNX2 as an indicator of osteoblasts was significantly higher in the callus of metamizole-treated animals than in the callus of control animals. The expression of the proliferation marker PCNA did not differ between the two groups at this early time point (Figure 6F). These findings indicate a slightly different expression profile in metamizole-treated animals when compared to controls, indicating a trend towards a higher angiogenic activity and accelerated bone turnover at 2 weeks after surgery.