Introduction

Patients with multiple injuries frequently suffer from bone fractures and are at high risk to develop fracture healing complications, including non-unions (Bhandari et al. 2003; Karladani et al. 2001; Zura et al. 2). Relative gene expression was calculated using the ΔΔCt method with PCR efficiency correction using LinReg PCR 2015.3 (Academic Medical Centre, Amsterdam, Netherlands) (Ramakers et al. 2003). Cycle threshold (Ct) values obtained for each sample were normalized to those of the housekee** gene Gapdh and the control group with isolated fracture.

Table 2 Primer sequences
Full size table

Flow cytometry

Immune cell populations in the fracture hematoma were determined by flow cytometry. Hematoma samples were harvested and homogenized by passing them through a 70-μm cell strainer (Corning Inc., Durham, NC). The resulting cell suspension was stained for 30 min on ice with the following antibodies against the indicated surface markers: anti-Ly-6G-V450 antibody (No. 560603 BD Biosciences), anti-CD11b-Alexa Fluor 700 (No. 56-0112 eBioscience), anti-F4/80-FITC (No. 11-4801 eBioscience), anti-CD3e-PE-Cyanine7 (No. 25-0031 eBioscience), and anti-CD19-PE antibody (No. 12-0193 eBioscience). Corresponding isotype-matched controls from the respective manufacturers served as negative controls. Dead cells were excluded using 7-aminoactinomycin D (7AAD) staining (Sigma Aldrich, Taufkirchen, Germany). Live cells were gated for the following cell populations: neutrophil granulocytes (CD11b+, Ly-6G+), macrophages (CD11b+, Ly-6G−, F4/80+), B cells (CD3−, CD19+), and T cells (CD3+, CD19−). The samples were analyzed using a LSR II flow cytometer (BD Biosciences) and FlowJo software (10.0.8r1, FlowJo, Ashland, USA).

Histomorphometry and immunohistochemistry

Lungs (3 h and 1 day) were harvested and fixed in 4% buffered formalin solution (Otto Fischar GmbH & Co. KG, Saarbruecken, Germany). They were embedded in paraffin and stained with hematoxylin and eosin (Mayer’s hemalum solution, Merck KGaA®, Darmstadt, Germany and Eosin Y, Applichem, Darmstadt, Germany) for morphological investigations. Neutrophil granulocytes were identified using a Ly-6G-antibody (1:300 LEAF™, No. 127632 BioLegend, Fell, Germany).

Fractured femurs (days 1 and 10) were fixed in 4% buffered formalin solution, decalcified in 20% ethylenediaminetetraacetic acid (EDTA) for 10–12 days, and embedded in paraffin for immunohistochemistry. Femur samples collected 21 days after surgery were embedded in methyl methacrylate (MMA) without decalcification. For tissue quantification, femur sections were stained with either Safranin-O (paraffin sections; Merck Chemicals GmbH, Darmstadt, Germany), which stains mainly cartilage, or Giemsa (MMA-embedded samples; AppliChem). The relative amounts of osseous, cartilage, and fibrous tissues were evaluated in the callus between the inner two pinholes using image analysis software (MMAF Version 1.4.0 MetaMorph®, Leica, Heerbrugg, Switzerland). For immunostaining, we used the following antibodies and dilutions: neutrophil granulocytes 1:300 LEAF™ anti-mouse Ly-6G antibody (No. 127632 BioLegend), macrophages 1:500 rat anti-mouse F4/80 antibody (No. MCA497GA AbD Serotec, Puchheim, Germany), and collagen X 1:200 rabbit anti-mouse collagen X antibody (No. ABIN1077945 Antibodies-Online, Atlanta, USA). Secondary antibodies and dilutions: 1:200 goat anti-rabbit IgG secondary antibody (No. B2770 Life Technologies, Carlsbad, USA) and 1:200 goat anti-rat IgG secondary antibody (No. A10517 Life Technologies). Species-specific IgG subtype mixtures obtained from the respective manufactures were used as negative controls. For signal detection, Vectastain Elite ABC kit and Vector NovaRED substrate (both Vector laboratories Inc., Burlingame, USA) were applied according to the manufacturer’s protocols. Sections were counterstained with hematoxylin (Waldeck, Münster, Germany) and analyzed by light microscopy (Leica DMI6000B, Leica). The relative proportion of collagen X-positive stained cartilage was evaluated in the fracture callus between the inner two pinholes using the image analysis software described above.

Biomechanical testing

To assess the bending stiffness of the fractured femurs explanted on day 21, a non-destructive three-point bending test was performed (Röntgen et al. 2010). Briefly, after removal of the external fixator, an axial load with a maximum of 2 N was applied to the top of the cranio-lateral callus side using a materials testing machine (1454, Zwick GmbH & Co KG, Ulm, Germany). The bending stiffness was calculated from the slope of the load-deflection curve (Röntgen et al. 2010).

Micro-computed tomography

After biomechanical testing, the fractured femurs were scanned using a micro-computed tomography (μCT) scanning device (Skyscan 1172; Bruker, Kontich, Belgium) operating at a voxel resolution of 8 μm (50 kV, 200 mA) to evaluate bone formation and structural parameters of the fracture callus. Phantoms with a defined hydroxyapatite density (250 and 750 mg/cm3) were used to calibrate and assess the bone mineral density. The volume of interest comprised the periosteal callus between the two inner pinholes and the fracture gap. A global threshold of 642 mg hydroxyapatite/cm3 was applied to discriminate between mineralized and non-mineralized tissues (Morgan et al. 2009) according to the American Society for Bone and Mineral Research guidelines for μCT (Bouxsein et al. 2010).

Statistical analysis

All data are presented as the mean ± standard deviation. Statistical analysis was performed using GraphPad Prism 6 (GraphPad Software, La Jolla, USA). Data were tested for normal distribution with Shapiro-Wilk test and then compared by either Kruskall-Wallis and Dunn’s post hoc test or by one-way analysis of variation and Fishers LSD post hoc test. The level of significance was set at p ≤ 0.05. The main outcome parameter of flexural rigidity of the fractured femur (power: 80%, α = 0.05) obtained from previous studies (Kovtun et al. 2016) was used to calculate sample size, which is indicated in the figure legends.

Results

Global IL-6 inhibition does not influence compromised bone repair induced by combined fracture and thoracic trauma

Confirming our previous studies (Bergdolt et al. 2017; Kemmler et al. 2015; Kovtun et al. 2016), we found that the combined fracture and thoracic trauma (Fx + TxT) induced systemic and pulmonary inflammation and disturbed fracture healing. The plasma levels of IL-6, MCP-1, and CXCL1 were significantly increased 3 h after combined trauma compared to the isolated fracture (Fx) group indicating a systemic immune response post trauma (Fig. 1a, c, d). The sIL-6R concentration was significantly elevated in the combined trauma group at day 1 suggesting increased shedding of the mIL6R (Fig. 1b). All other measured circulating inflammatory mediators were not significantly affected compared to mice with isolated fracture. Furthermore, the combined trauma slightly increased Crp and Saa expression in the liver (Fig. 2a, b). In the lung, the combined trauma caused tissue damage and inflammation as confirmed by the presence of blood clots, alveolar wall thickening (Fig. 3a, b), increased neutrophil numbers (Fig. 3c, d), and elevated IL-6 and CXCL1 levels (Fig. 3e, f). In the fracture hematoma, the measured inflammatory mediators and immune cell recruitment to the fracture site were not significantly affected by the additional thoracic trauma (Fig. 4). However, fracture healing was disturbed by the trauma as indicated by a reduced bone fraction in the develo** callus at day 10 (Fig. 5a, c), poor bony bridging of the fracture gap (Fig. 6a, b) and decreased mechanical properties of the fractured bone (Fig. 6c) in the late healing phase at day 21 compared to mice with isolated fracture. The relative amounts of bone and cartilage were not significantly altered at day 21 (Fig. 6d, e).

Fig. 1
figure 1

Inflammatory mediators in the blood 3 h and 1 day after fracture (Fx) and combined fracture and thoracic trauma (Fx + TxT) in vehicle-, anti-IL-6 antibody-, and sgp130Fc-treated mice. Data are displayed as means ± standard deviation. a n = 5–10; b n = 5–6; c, d n = 6–10; e n = 6. *p ≤ 0.05; **p ≤ 0.01; ****p ≤ 0.0001. Data of untreated animals are presented in Supplemental Table 1

Full size image
Fig. 2
figure 2

Hepatic acute-phase reaction 1 day after fracture (Fx) and combined fracture and thoracic trauma (Fx + TxT) in vehicle-, anti-IL-6- antibody-, and sgp130Fc-treated mice. Relative gene expression of a CRP C-reactive protein, b SAA serum amyloid A, and c CXCL1 chemokine (C-X-C motif) ligand 1 in the liver. Data are displayed as means ± standard deviation. n = 4–5. *p ≤ 0.05; **p ≤ 0.01

Full size image
Fig. 3
figure 3

Pulmonary inflammation 3 h and 1 day after fracture (Fx) and combined fracture and thoracic trauma (Fx + TxT) in vehicle-, anti-IL-6 antibody-, and sgp130Fc-treated mice. a Representative images of hematoxylin and eosin (H&E) stained lungs of vehicle-treated mice after 3 h and b 1 day. c Representative images of lungs stained for neutrophils (Ly-6G+); arrowheads indicate positively stained cells. d Neutrophil (Ly-6G+) number in lung tissue. e IL-6 and f CXCL1 chemokine (C-X-C motif) ligand 1 concentrations in the broncho-alveolar lavage fluid after 3 h. Data are displayed as means ± standard deviation. D n = 5–6; E, F n = 8–9. *p ≤ 0.05; **p ≤ 0.01. Data of untreated animals are presented in Supplemental Table 1

Full size image
Fig. 4
figure 4

Inflammatory mediators and immune cells in the fracture hematoma 3 h and 1 day after fracture (Fx) and combined fracture and thoracic trauma (Fx + TxT) in vehicle-, anti-IL-6 antibody-, and sgp130Fc-treated mice. Data are displayed as means ± standard deviation. a IL-6, b MCP-1 monocyte chemotactic protein 1, and c CXCL1 chemokine (C-X-C motif) ligand 1 concentrations after 3 h. d Proportion of neutrophils (CD11b+, Ly6G+), e macrophages (CD11b+, F4/80+), f B cells (CD19+), and g T cells (CD3+). a–c n = 6–7, d n = 6–8, e–g n = 7–8. **p ≤ 0.01, ***p ≤ 0.001

Full size image
Fig. 5
figure 5

Histomorphometrical analyses of the fracture callus on day 10 after fracture (Fx) and combined fracture and thoracic trauma (Fx + TxT) in vehicle-, anti-IL-6 antibody-, and sgp130Fc-treated mice. a Representative histological images of the fracture callus stained with Safranin-O: Ct cortex, FG fracture gap. Boxed areas in a indicate the location of the higher magnified images in b. b Immunostaining of collagen type X. c Relative amount of bone and d cartilage in the fracture callus. e Proportion of collagen type X (ColX)-positive stained cartilage of the total cartilage determined by immunohistochemistry. Data are displayed as mean ± standard deviation. c, d n = 6; (e) n = 4–5. *p ≤ 0.05

Full size image
Fig. 6
figure 6

Micro-computer tomography, histomorphometrical, and biomechanical analyses of the fracture callus on day 21 after fracture (Fx) and combined fracture and thoracic trauma (Fx + TxT) in vehicle-, anti-IL-6 antibody-, and sgp130Fc-treated mice. a Representative μCT three-dimensional reconstructions of the fracture callus. b Representative Giemsa-stained histological images of the fracture callus. c Bending stiffness of fractured femurs. Relative amount of d bone and e cartilage determined by histomorphometrical analyses. Data are displayed as mean ± standard deviation. n = 7–9 (c–e). **p ≤ 0.01

Full size image

To block IL-6 signaling globally during the inflammatory phase, we administered an anti-IL-6 antibody 30 min and 2 days after combined trauma. The anti-IL-6 antibody considerably reduced IL-6 plasma levels compared to vehicle-treated mice 3 h after injury indicating efficient IL-6 inhibition. IL-6 plasma levels also remained low after 1 day (Fig. 1a). The trauma-induced increase of circulating sIL-6R was also significantly diminished by the anti-IL-6 antibody, suggesting that the shedding of the mIL-6R may be mediated by IL-6 (Fig. 1b). Whereas the trauma-induced increase of MCP-1 was not significantly affected after global IL-6 inhibition, CXCL1 and CRP plasma levels were significantly reduced 3 h and 1 day after combined injury, respectively (Fig. 1c–e). The hepatic acute-phase response was strongly diminished by the anti-IL-6 antibody treatment as indicated by significantly reduced Crp, Saa, and Cxcl1 gene expression (Fig. 2a–c). Lung damage (data not shown) and neutrophil invasion into the lung tissue (Fig. 3c, d) after combined injury were not influenced by global IL-6 inhibition; however, IL-6 in the BAL fluid was significantly reduced compared to vehicle-treated mice (Fig. 3e).

In the fracture hematoma, IL-6 was also significantly reduced after global IL-6 inhibition compared to vehicle-treated mice (Fig. 4a); however, other measured inflammatory mediators were unaffected. FACS analysis revealed a reduced recruitment of neutrophils to the fracture site in mice with combined fracture and thoracic trauma that received the anti-IL-6 antibody (Fig. 4d). Histological and μCT evaluation demonstrated that the amount of newly formed bone and cartilage in the fracture callus were unaffected in the combined trauma group after global IL-6 inhibition at days 10 (Fig. 5) and 21 (Fig. 6). Additionally, the mechanical properties of the fracture callus were not significantly influenced (Fig. 6c).

In summary, global IL-6 inhibition reduced circulating IL-6, sIL-6R, and CXCL1, the hepatic acute-phase response, and neutrophil numbers in the fracture hematoma but did not influence the healing outcome after severe trauma.

Inhibition of IL-6 trans-signaling improves compromised bone repair induced by combined fracture and thoracic trauma

To selectively block IL-6 trans-signaling, we treated mice with combined fracture and thoracic trauma with the artificial fusion protein sgp130Fc in the early posttraumatic phase 30 min and 2 days after injury. IL-6 trans-signaling inhibition significantly reduced circulating IL-6 3 h after combined trauma, but significantly increased it after 1 day compared to vehicle-treated mice (Fig. 1a). The trauma-induced increase of sIL-6R observed after 1 day was also significantly diminished (Fig. 1b). Other systemic inflammatory mediators were not significantly affected by sgp130Fc administration compared to vehicle-treated mice. In the liver, expression of Crp was significantly diminished after IL-6 trans-signaling inhibition (Fig. 2a). Saa and Cxcl1 expression were also slightly reduced, although not significantly (Fig. 2b, c). In the lungs, sgp130Fc treatment affected neither lung damage (data not shown) nor the inflammatory response induced by the thoracic trauma (Fig. 3). Additionally, the early inflammation at the fracture site was not significantly influenced after IL-6 trans-signaling inhibition (Fig. 4). The bone and cartilage fractions in the develo** fracture callus were unaltered after the blockade of IL-6 trans-signaling (Fig. 5a, c, d). However, the amount of collagen type X expressing hypertrophic cartilage was significantly increased indicating accelerated cartilage-to-bone transformation (Fig. 5b, e). Confirming this, 21 days after trauma, bony bridging of the fracture gap (Fig. 6a, b) and the bending stiffness (Fig. 6c) of the fracture callus were significantly elevated in mice treated with sgp130Fc, suggesting that the selective blockade of IL-6 trans-signaling significantly improved the fracture healing outcome after severe trauma.

Discussion

Here, we investigated the hypothesis that IL-6 trans-signaling is involved in the pathomechanisms of trauma-induced compromised fracture healing. Using a mouse model of severe injury, we demonstrated that the transient blockade of IL-6 trans-signaling in the early posttraumatic phase with sgp130Fc significantly improved bone repair. By contrast, healing was not improved by an anti-IL-6 antibody, which blocks both IL-6 classic and trans-signaling, suggesting that the classic pathway rather exerts beneficial effects of augmenting bone repair under conditions of severe trauma, as it similarly does in uncomplicated fracture healing (Prystaz et al. 2017) (Fig. 7).

Fig. 7
figure 7

Scheme of IL-6 classic and trans-signaling and the proposed effects of fracture healing. In IL-6 classic signaling, IL-6 binds to its membrane-bound receptor (IL-6R), which then binds to a dimer of transmembrane glycoprotein 130 (gp130), inducing intracellular signal transduction. In IL-6 trans-signaling, IL-6 binds to its soluble receptor (sIL-6R), which is mainly shed by A Disintegrin and Metalloproteinase 17 (ADAM 17). The IL-6/sIL-6r complex then binds to the gp130 dimer. Our previous (Prystaz et al. 2017) and present results indicate that IL-6 classic signaling induces a balanced immune response and pro-regenerative effects on bone repair. In contrast, IL-6 trans-signaling, which is induced after severe injury, negatively affects fracture healing

Full size image

IL-6 classic and trans-signaling differently modulate systemic posttraumatic inflammation

In this study, we used a mouse model of combined fracture and thoracic trauma to elucidate the role of IL-6 in compromised fracture healing after severe injury. As expected (Bergdolt et al. 2017; Kemmler et al. 2015; Kovtun et al. 2016), the combined injury induced a systemic inflammation with increased plasma levels of inflammatory mediators, including IL-6, and sIL-6R, which indicated enhanced shedding of the mIL-6R. This is in agreement with experimental (Kleber et al. 2015) and clinical studies (Beeton et al. 2004) in subjects with fracture and concomitant injury, and confirms that IL-6 trans-signaling is activated in posttraumatic inflammation.

To discriminate between IL-6 actions, we applied an anti-IL-6 antibody, which inhibits IL-6 globally, and sgp130Fc, which selectively blocks IL-6 trans-signaling (Jostock et al. 2001). It is not possible to inhibit IL-6 classic signaling selectively. However, by comparing the effects of global and trans-signaling inhibition, indirect, but valid conclusions can be drawn about the role of IL-6 classic signaling (Barkhausen et al. 2011; Prystaz et al. 2017). The anti-IL-6 antibody efficiently reduced IL-6 levels in the blood and BAL fluid. Global IL-6 inhibition also decreased serum levels of the sIL-6R, suggesting that IL-6 directly and indirectly stimulates the shedding of its membrane-bound receptor during posttraumatic inflammation (Lokau et al. 2017). While sgp130Fc diminished the trauma-induced increase of circulating IL-6 less efficiently after 3 h, it notably enhanced it after 1 day. The initial reduction of IL-6 may result from trap** of the IL-6/sIL-6R complex by sgp130Fc, which favors the restoration of the initial equilibrium of IL-6/sIL-6R complexes and, thus, reduces circulating free IL-6 molecules (Garbers et al. 2011). The later increase of IL-6 may be caused by a delay of IL-6 degradation after interception of IL6/sIL-6R complexes by sgp130Fc and was also found in patients, who were treated with the anti-IL-6R antibody tocilizumab (Nishimoto et al. 2008).

The anti-IL-6 antibody also abolished the trauma-induced increase of CXCL1 in the circulation and its hepatic expression. By contrast, selective inhibition of IL-6 trans-signaling did not significantly reduce circulating CXCL1. This is in agreement with our previous data (Prystaz et al. 2017) indicating that the liver is a major source of CXCL1 after injury, and that hepatic CXCL1 production is regulated by IL-6 classic signaling. The classic IL-6 pathway is known to induce the hepatic acute-phase response (Schmidt-Arras and Rose-John 2016). Confirming this, the posttraumatic expression of CRP and SAA in the liver was significantly reduced by the anti-IL-6 antibody, whereas inhibition of IL-6 trans-signaling provoked only minor effects. Hepatic IL-6 classic signaling is suggested to act pro-regenerative, as it induces the first-line defense against pathogens and limits inflammatory responses (Schmidt-Arras and Rose-John 2016).

IL-6 classic and trans-signaling differently modulate inflammation at the site of fracture

In the fracture hematoma, the anti-IL-6 antibody reduced IL-6 as expected, but neither global nor trans-signaling inhibition affected the concentration of other inflammatory mediators. The proportion of neutrophils was significantly reduced after global but not after trans-signaling inhibition, indicating that IL-6 classic signaling regulates neutrophil recruitment and/or apoptosis directly or indirectly. This corroborates our previous study in the isolated fracture model (Prystaz et al. 2017) and can be explained by the reduced plasma concentrations of the neutrophil chemoattractant CXCL1. Studies about direct IL-6 effects on neutrophil functions are conflicting. In an air-pouch model, neutrophil trafficking was induced by IL-6 trans-signaling (Rabe et al. 2008), whereas in a mouse model of peritoneal inflammation, IL-6-induced STAT3-signaling diminished neutrophil recruitment (Fielding et al. 2008). Others reported that IL-6 did not directly act as a neutrophil chemoattractant or induce apoptosis (Wright et al. 2014), although its therapeutic inhibition by tocilizumab induces neutropenia (Espinoza et al. 2017; Wright et al. 2014). The proportions of other immune cell populations in the fracture hematoma were not affected by global or trans-signaling inhibition. This is in contrast to the isolated fracture model, where monocytes, macrophages, and lymphocytes were significantly reduced after IL-6 inhibition (Prystaz et al. 2017). The reason for the different results in the isolated fracture and combined fracture and thoracic trauma models could be that severe trauma affects the phenotype and function of many immune cells (Flohe et al. 2008; Lord et al. 2014), possibly also their responsiveness to IL-6.

Inhibition of trans-signaling, but not global IL-6 inhibition, ameliorates the deleterious effects of a concomitant injury on bone repair

Our study demonstrated that global IL-6 inhibition did not affect trauma-induced impaired fracture healing. By contrast, inhibition of IL-6 trans-signaling accelerated cartilage-to-bone transformation in the intermediate healing phase, and enhanced bony bridging of the fracture gap and mechanical callus properties in the late stage. This suggests that, under conditions of severe trauma, IL-6 trans-signaling mediates negative effects on bone repair, whereas classic signaling may act rather pro-regenerative, as it similarly does in uncomplicated fracture healing (Prystaz et al. 2017). But how can the positive effects of IL-6 trans-signaling inhibition be explained mechanistically? One difference between both treatment groups was the reduced neutrophil number in the fracture hematoma after global but not after trans-signaling inhibition. Neutrophils are the most abundant immune cell population in the early fracture hematoma (Hoff et al. 2016). They remove pathogens, coordinate the transition to a more sustained population of mononuclear cells, and contribute to the resolution of inflammation (Bastian et al. 2011; Kovtun et al. 2016). Neutrophils may augment bone regeneration, because it is impaired after neutrophil depletion (Chan et al. 2015; Kovtun et al. 2016). However, after severe trauma, neutrophils can become dysfunctional and aggravate tissue damage, for example, by the massive production of reactive oxygen species (ROS) and neutrophil extracellular traps (NETs) (Hazeldine et al. 2014). Therefore, the role of neutrophils in fracture healing might depend on trauma severity. A limitation of the present study is that we did not assess neutrophil activity at the fracture site. Therefore, their role remains unclear. Further work is necessary to elucidate neutrophil functions in fracture healing and how they are regulated by IL-6. Another striking observation, which could possibly explain improved bone repair was the altered kinetics of circulating IL-6 in sgp130Fc-treated mice. After initial reduction, IL-6 plasma levels were moderately, but significantly increased after 1 day, whereas they remained low after anti-IL-6 antibody treatment. As explained above, this could result from delayed IL-6 degradation after sgp130Fc treatment (Rose-John et al. 2007). Possibly, the free IL-6 then provokes rather pro-regenerative effects by activating classic signaling, because trans-signaling might still be inhibited. However, a limitation of the present study is that we did not include enough early investigation time points to unravel the interconnection between the early immune response and regenerative processes. Further investigations are needed to mechanistically explain improved bone regeneration in trauma-induced compromised fracture healing by sgp130Fc.

Conclusions

In summary, the present study demonstrates for the first time that IL-6 trans-signaling is involved in the pathomechanisms of compromised fracture healing after severe injury, whereas IL-6 classic signaling rather mediates pro-regenerative effects augmenting bone regeneration. However, further studies are necessary to elucidate the underlying mechanisms in detail. Nevertheless, our results can help to develop new treatment strategies to reduce fracture healing complications after severe injury.