Preclinical biological and physicochemical evaluation of two-photon engineered 3D biomimetic copolymer scaffolds for bone healing

A major challenge in orthopedics is the repair of large non-union bone fractures. A promising therapy for this indication is the use of biodegradable bioinspired biomaterials that stabilize the fracture site, relieve pain and initiate bone formation and healing. This study uses a multidisciplinary evaluation strategy to assess immunogenicity, allergenicity, bone responses and physicochemical properties of a novel biomaterial scaffold. Two-photon stereolithography generated personalized custom-built scaffolds with a repeating 3D structure of Schwarz Primitive minimal surface unit cell with a specific pore size of ~400 µm from three different methacrylated poly(D,L-lactide-co-ε-caprolactone) copolymers with lactide to caprolactone monomer ratios of 16:4, 18:2 and 9:1. Using in vitro and in vivo assays for bone responses, immunological reactions and degradation dynamics, we found that copolymer composition influenced the scaffold physicochemical and biological properties. The scaffolds with the fastest degradation rate correlated with adverse cellular effects and mechanical stiffness correlated with in vitro osteoblast mineralization. The physicochemical properties also correlated with in vivo bone healing and immune responses. Overall these observations provide compelling support for these scaffolds for bone repair and illustrate the effectiveness of a promising multidisciplinary strategy which could be widely used for the preclinical evaluation of biomaterials. period which is sufficient for physiological bone replacement. This study illustrates an important new biomaterial with a strategy for pre-clinical evaluation of biomaterials developed for bone repair.


Introduction
The repair of large non-union bone lesions may be a major complication of traumatic injury, metabolic bone disease like osteoporosis and tumours. 1, 2 Synthetic biomaterials have been developed to fill bone defects and accelerate bone healing 1, 3 , because autologous, allogeneic and cadaver bone grafts, which are the gold standard for bone repair, are associated with a myriad of complications and limited availability. 1, 2, 4, 5 The limitations of autologous and allogeneic bone grafts can be overcome by using bone substitutes or synthetic grafts. 3 A large range of osteoconductive and osteoinductive biomaterials have been considered for bone repair which have potential 2 including bioengineered biodegradable materials that provide structure within bone lesion and would be replaced by newly formed bone during bone regeneration.
The biology of bone fracture healing is a multistage process that follows a specific cascade resulting in unscarred restoration of bone tissue. 6 Local and systemic growth and differentiation factors, hormones, cytokines and extracellular matrix proteins interact with cells in the circulation and fracture site 7 and begin with an inflammatory phase, followed by repair and bone remodeling. 8 Transient inflammation after injury is essential 6,9,10 to produce factors that initiate osteoblast (OB)-and osteoclast (OC)-mediated bone repair. Bone biomaterials should not induce chronic inflammation or fibrosis, but should promote mild inflammation, followed by osteoconduction or induction, provide mechanical support and degrade over time. 11 In the event that biomaterials induce excessive inflammation and/or immune and allergic responses, there might be delayed and/or inhibition of bone repair. 12,13 A porous three-dimensional (3D) scaffold for bone repair ideally should mimic an autologous bone graft 14 , be biodegradable and either induce minimal inflammation or no inflammation, as long as they are introduced with the factors that promote bone repair and should not induce fibrosis or scarring. 15 Examples of biodegradable polymers that are resorbed over time frequently used in regenerative medicine [16][17][18] include polyglycolic acid (PGA), polylactide (PLA), polylactic-co-glycolic acid (PLGA), poly-ε-caprolactone (PCL), poly(D,L-lactide) (PDLLA) and additional copolymers and composites. 19,20 Although scaffold degradation is beneficial, undesired effects caused by degradation products via the formation of small chain carboxylic acids may change the local pH and cause inflammation 21,22 or in some cases as occurs within PLA, acidic degradation products may be toxic. 17 Methods that have been developed for the production of these porous polymeric materials include electrospinning, phase separation and porogen leaching. However, these techniques do not enable the manufacture of complex 3D structures with tunable micro-scale features.
Computer-aided design (CAD) based technologies like 3Dprinting and stereolithography are capable of manufacturing scaffolds for tissue engineering, but to obtain true biomimetic or well-defined structures sufficient resolution is required. 23 Twophoton polymerization (TPP) is an example of stereolithography technology and is capable of fabricating well-defined 3D microstructures with a resolution of 100 nm. 24,25 For bone filling scaffolds, the optimum pore size ranges from 200 to 400 µm, which is similar to the average size of human osteon (~223 µm). 26 Moreover, it is possible to design and produce customizable scaffolds that are compressible and tailored to fit into the fracture site. One limitation in the past for TPP has been the slow scan speeds but as the technology improves exciting capability is now available to deliver high resolution scaffolds in commercially attractive timescales.
In this study, methacrylated poly(D,L-lactide-co-εcaprolactone) (PLCL) copolymers of different compositions and chain lengths (LCM3 (16:4), LCM4 (18:2) and LCM6.1 (9:1)), where numbers represent monomer units of lactide (LA) and caprolactone (CL) were used to generate biomimetic 3D Schwarz Primitive (P) minimal surface scaffold structures with a pore size of 400 µm. 27,28 Here, we combine materials science, bone biology and immunology to establish a strategy for evaluating these developed TPP-engineered copolymer scaffolds for eventual use in patients. Our approach includes 1) an in vitro evaluation of osteoblast and osteoclast function and an in vivo mouse calvarial defect model to assess bone healing in the context of the scaffolds, 2) in vitro degradation and mechanical studies (compressive strength and modulus), and 3) in vitro and in vivo evaluation of inflammatory, immune and allergic reactions to biomaterials including an in vitro assay providing information on antigen crossreactivity and cytotoxicity; a rapid high-throughput intraperitoneal assay providing data on the inflammation and the cytokine milieu; and a sub-chronic subcutaneous long-term assay providing information on the longevity of the inflammatory response and the capacity of the biomaterial to induce fibrosis.

Ethics statement
This study was carried out in strict accordance with the guidelines for the Care and Use of Laboratory Animals of the Austrian Ministry of Science. The protocol was approved by the Committee on the Ethics of the Austrian Ministry of Science (N o : 66:006/00 12-11/3b/2012). All painful procedures were performed under anesthesia, and all efforts were made to minimize suffering.

Mice
Female 8-12 week old BALB/c and C57BL/6 (B6) mice (Charles River Laboratories, Sulzfeld, Germany) were used for in vivo experiments. Neonatal mice were used for the isolation of OBs. The mice were housed in the mouse facility at the Department of Pharmacology and Toxicology, University of Vienna, Vienna, Austria. Mice were provided food (Ssniff Spezialdiäten GmbH, Soest, Germany) and tap water ad libitum.
2D copolymer disk and 3D scaffold production LCM scaffolds were produced and provided by the Institute for Bioprocessing and Analytical Measurement Techniques (iba, Heiligenstadt, Germany) and the detailed description of the methodology can be found in. 27 Briefly, methacrylated PLCL copolymers were synthesized by ring opening polymerization. Three different LA/CL ratios were used and coded LCM3 (16:4), LCM4 (18:2) and LCM6.1 (9:1). Both LCM4 and LCM6.1 have the same molar ratio (LA/CL) of 18:2 and 9:1 respectively compared to LCM3 which is 16:4. However LCM6.1 has lower molecular weight (1059 g mol -1 ) than LCM4 (1822 g mol -1 ) and LCM3 (1742 g mol -1 ). 29 For disk manufacturing, precursor-LCMs and photoinitiator (Irgacure369 (0.1 wt.%), BA740 (0.2 wt.%)) were filled into a silicon mold (Ø= 5 mm, 0.5 mm height) and solidified by 9 min of UV irradiation (Vacuum-UV-Exposure Box2, proMa). The polymer disks were developed in acetone for 7 days before drying under vacuum and sterilized using gamma irradiation at the dose of 25 ± 2.5 kGy and kept in sterilization pouches prior to use. For 3D scaffold manufacturing, photoinitiator and LCM precursors underwent polymerization with a TPP apparatus (M3DL, LZH Hannover, Germany) with a femtosecond laser source (140 fs, 80 MHz, 800 nm) to produce LCM scaffolds followed by washing in acetone to remove the unpolymerized precursors. The produced repeating array of Schwarz P unit cell (with 400 µm size) scaffolds were sterilized using gamma irradiation as described above.

Osteoblasts
Primary mouse OBs (6.2 x 10 4 cells/cm 2 ) were seeded in αMEM, 10% FBS and 1% penicillin/streptomycin onto 96-well culture plates containing the 2D copolymer disks (Ø = 5 mm) or onto tissue culture plastic at 37°C and 5% CO2. To induce OB differentiation, culture medium was supplemented with 50 µg/ml ascorbic acid (Sigma-Aldrich) and 5 mM βglycerophosphate (Sigma-Aldrich) 24 h post seeding. OBs cultured in medium alone served as control group. OB cell proliferation was evaluated by "Presto Blue cell viability assay" (Molecular Probes, Thermo Fisher Scientific, MAN0003232) according to the manufacturer`s instructions. Briefly, cells were incubated with Presto Blue reagent and the reagent was reduced by metabolically active cells. This serves as an indicator for cell viability and proliferation. The reduction product was detected at 560/590 nm. At day 7, alkaline phosphatase (ALP) activity was measured as a marker for OB differentiation from cell lysates with a fluorogenic substrate (6,8-difluoro-4-methylumbelliferyl phosphate, DIFMUP, Molecular Probes, Thermo Fisher Scientific) at 388/455 nm as previously described. 32 In addition, ALP activity was detected using SIGMAFAST™ BCIP ® /NBT substrate stain (Sigma-Aldrich) according to the application note of PromoCell (Heidelberg, Germany). Calcium deposition of mature OBs was visualized by Alizarin Red S (ARS) and von Kossa staining at day 14. For ARS staining, cells were fixed in 10% neutral buffered formalin (Sigma-Aldrich) and stained with 40 mM ARS (pH 4.2) (Sigma-Aldrich). Quantification of ARS-stained cultures was performed by a cetylpyridinium chloride extraction method and extracts were read at 520 nm. For von Kossa staining, cells were fixed in 10% neutral buffered formalin (Sigma-Aldrich) and incubated with 5% silver nitrate solution (Sigma-Aldrich) under UV light before neutralization with 5% sodium thiosulfate (Sigma-Aldrich). All stained plates were scanned on a high-resolution flatbed scanner (Epson Perfection 1200Photo).

In vivo intracalvarial defect model
The LCM scaffolds (4 mm diameter) were incubated overnight in sterile PBS. Orthovita Vitoss™ Foam (Vitoss, Malvern, USA) was used in the experiments as a positive control. Vitoss was cut into 4 mm diameter and then soaked in 50 µl PBS before implantation. Non-healing critical, full-thickness size defects were created in anesthetized (100 mg/kg ketamine (Ketamidor ® , Richter Pharma, Wels, Austria) and 5 mg/kg xylazine (Rompun ® , Richter Pharma)) 12-week old female BALB/c mice (n=5-6) in the right parietal bone using a dental trephine with 4 mm diameter under constant irrigation while preserving the underlying dura mater. LCM scaffolds or Vitoss were implanted into the defect or left empty and then the skin was sutured. The animals were monitored postop and treated with 0.1 mg/kg buprenorphine (Bupaq ® , Richter Pharma) for analgesia every 10-12 h for 3 days. At 12 weeks post operation, the animals were sacrificed by cervical dislocation, skulls were harvested and fixed in 4.5% buffered formalin (Carl Roth). For histological examination, light-curing resin (Technovit 7200 VLC + BPO, Heraeus Kulzer, Wehrheim, Germany) embedded undecalcified sections (80-100 µm) were prepared parallel to the sagittal suture with an Exakt Cutting and Grinding Equipment (Exakt, Norderstedt, Germany) and stained with Levai Laczko dye according to a standardized method. 22 Immunological assessment In vivo immunological assessment of scaffolds The LCM scaffolds (2 mm 3 ) were incubated overnight in sterile PBS. Vitoss was used in all immunological experiments as a positive control because it contains type I bovine collagen, which is recognized as a foreign protein and generates an immune response in mice. Vitoss was cut into 2 mm 3 and then soaked in 50 µl PBS before implantation. Scaffolds and Vitoss were implanted either intraperitoneal (i.p.) or subcutaneous (s.c.) into 8-week old female BALB/c and B6 mice (n=5-8) and compared to age-matched sham controls. For implantation, mice were anesthetized with 100 mg/kg ketamine (Ketanest ® , Pfizer Corporation Austria GmbH., Vienna, Austria) and 6 mg/kg xylazine (Rompun ® , Bayer AG, Leverkusen, Germany).
For the high throughput model, we used the implantation approach described in 33 and. 34 Abdominal fur was shaved and cleaned with 70% alcohol and polyvidone iodine (Betadine ® ). Using aseptic techniques, an 8 mm midline incision was made along the linea alba followed by an incision in the peritoneum. LCM scaffolds or Vitoss were placed in the peritoneal cavity. The abdominal muscles and skin were sutured with absorbable 4-0 Vicryl suture (Ethicon Inc., Somerville, USA) and 4-0 nylon suture (Ethicon), respectively. The same incision and suture were used without implanting materials for the sham controls. The peritoneal cavity was lavaged 7 days after implantation with 3 ml PBS and peritoneal cells were counted and peritoneal fluid was stored at -20°C until thawed for the measurement of IL-1β (eBioscience Inc., San Diego, USA), IL-2 and IL-4 (ELISA MAX TM Standard, Biolegend, San Diego, USA) cytokines.
For the subchronic model, we used the implantation approach described in 33 and. 35 An 8 mm midline abdominal incision of the skin along the linea alba was made under aseptic conditions. LCM scaffold or Vitoss of the same size (2 mm 3 ) or no materials were inserted s.c. and then the skin was closed with 4-0 nylon suture. The mice were monitored until recovery from anesthesia. At 12 weeks, the implantation site with surrounding tissue (10 mm 2 ) from each mouse was excised and fixed in 4% formaldehyde overnight and then embedded in paraffin. Tissue sections (4 μm) were prepared and stained with Hematoxylin and Eosin (H&E) and Masson's trichrome (Sigma-Aldrich).
In vitro splenocyte responses to scaffolds Please do not adjust margins Please do not adjust margins LCM copolymer disks (Ø= 5 mm) were incubated with BALB/c and B6 mouse splenocytes. Spleens were minced and passed through a sterile 40 µm cell strainer (Corning Life Sciences, Durham, USA) in cold sterile PBS. The cells were centrifuged and RBCs were lysed with lysis buffer (BD Pharm Lyse™, BD Bioscience, New Jersey, USA). Titrated numbers of splenocytes were suspended in RPMI with 10% FBS, 1% penicillin/streptomycin, 0.1% gentamicin (Gibco), 0.2% ßmercaptoethanol 50 mM (Gibco) and 1% (100x) non-essential amino acids (Gibco) and were incubated with the copolymer disks, Vitoss, 10 µg/ml Concanavalin A (ConA, Amersham Pharmacia Biotech, Piscataway, USA) or a combination of ConA and biomaterials in a 96-well plate at 37°C at 5% CO2 for 72 h. Cell proliferation was measured using a cell proliferation ELISA BrdU kit (Roche Diagnostics, Vienna, Austria) by adding BrdU at 48 h and measuring absorbance at 450 nm at 72 h. In addition, supernatants were removed at 72 h and stored at -20 o C until thawed for the ELISA measurements of IL-1β, IL-2, IL-4 and IFNγ (ELISA MAX TM Standard, Biolegend, San Diego, USA).

Scaffold degradation studies
The degradation study was performed for LCM scaffolds of dimensions 2 mm 3 according to the standard BS EN ISO 10993-13:2010 36 . The high reproducibility of the method due to the TPP production allows individual scaffolds to be followed through their degradation pathway 27 . Scaffolds were placed individually into glass vials containing 30 ml PBS (pH = 7.4 ± 0.2) solution and maintained in an oven at 37 o C. At various time points the specimens were extracted with tweezers, placed into empty vials and dried in a vacuum oven (Medline Scientific, UK) at 50°C for 60 min. Afterwards, the dry weight of the scaffolds was recorded and samples were then returned back to vials containing fresh PBS. The percentage mass loss (ML) at each time point was calculated using the following equation; where md is the mass of the degraded scaffold after drying at 50°C in the vacuum oven for 60 min and mi is the initial dry mass of the scaffold.
Optical images of dried LCM scaffolds were captured at each degradation time point using a Nikon digital camera (Dxm1200F, Japan) attached to Nikon microscope (Japan). The images were processed using image analysis software (Nikon ACT-1 v. 2.62, LEAD Technologies, USA).

Mechanical test
The compressive strength and modulus were determined using Hounsfield tester and the calculations were done according to the standard ASTM 1621-10: 2010. 37 Scaffolds were inserted vertically between two flat plates of the testing machine where the load was applied on the cross section of the scaffold. A crosshead speed of 0.5 mm/min and a 5 N load cell was used and the test was carried out up to 20% strain (below the yield strain of the specimens) to obviate permanent deformation within the scaffolds. At various time points, the specimens were removed from PBS and dried in a vacuum oven at 50°C for 60 min before testing. The measurements were applied on dry scaffolds and carried out in three successive cycles with a time interval of 10 min to permit scaffold recovery. The compressive strength is taken as the maximum stress at 20% strain, while compressive modulus was determined as the gradient of the linear portion of stress-strain curve. The samples were then placed back in the medium until the next time point.

Statistical analysis
Statistical analysis of biological assays was done using one-way analysis of variance (ANOVA) with Tukey´s post hoc test for multiple comparison (GraphPad Prism v 5.0, San Diego, USA). p values were considered significant at < 0.05.

In vitro assessment of OC and OB differentiation and maturation
To determine the effect of scaffolds on OC differentiation, OBs and bone marrow OC precursors were co-cultured in the presence of LCM3, 4 and 6.1 scaffolds and the medium alone (M) control group (Fig. 1A). The OC morphology differed between the M control and LCM3, 4 and 6.1 scaffolds with a high number of differentiated mature multinucleated TRAP+ OCs (≥ 3 nuclei) in the M controls whereas the LCM scaffolds induced fewer, smaller and less spread out, irregularly-shaped mature OCs. TRAP+ multinucleated OCs for the M group was 8.34 ± 0.87 /mm 2 and for the scaffolds were LCM3: 7.01 ± 1.39 /mm 2 , LCM4: 9.47 ± 0.90 /mm 2 and LCM6.1: 3.74 ± 1.17 /mm 2 (Fig. 1B). LCM3 and M controls induced a similar number of differentiated TRAP+ OC compared to LCM4, though only LCM4 was statistically significantly higher compared to LCM6.1 (p<0.01). These data demonstrate that LCM6.1 significantly reduced OC differentiation compared to the other tested groups.
To determine the effect of LCMs on OBs, the OB viability and proliferation was evaluated using the Presto Blue assay. The OB proliferation peaked at day 3 in the presence of the scaffolds ( Fig.  2A) and was significantly higher for LCM3, 4 and 6.1 compared with osteogenic medium (OM) controls. The OB proliferation increased minimally and remained stable for the OM controls over the 14-day culture period. In contrast, LCM3 induced proliferation from day 1 to day 3 that remained stable until day 14, whereas LCM4 and LCM6.1 increased proliferation initially followed by decreased proliferation after day 7 with a more prominent reduction for LCM6.1. These data demonstrate that there were more proliferating OBs in the presence of LCM3 compared with LCM4 and LCM6.1. To analyze OB differentiation, cells (Fig. 2B) were stained and the ALP activity was measured (Fig. 2C). The OBs differentiated normally into more mature OBs and had higher ALP activity in the OM compared to M control alone (Fig. 2B, 2C). In Fig. 2B, the number of stained ALP+ cells was increased in all groups compared to controls without mineralization medium. However, the staining pattern exhibited when the cells were on LCM disks differed from the OM controls. Nevertheless, when we quantified the ALP activity in OBs, no statistical difference in ALP activity between OM and LCM groups was observed.
To assess OB mineralization, cultured OBs were stained with ARS and von Kossa. LCM6.1 and LCM4 induced higher mineralization compared with LCM3 and OM samples as shown in the tissue culture plate wells (Fig. 2D) and by absorbance of the extracted ARS calcium complex measurements (Fig. 2E). These data show that LCM4 and LCM6.1 disks supported higher matrix mineralization than LCM3.

In vivo evaluation of the bone response to LCM scaffolds
To evaluate the ability of the scaffolds to integrate into the calvarial bone defects, the three individual LCM scaffolds were inserted into the empty calvaria and the scaffold structure and surrounding tissue histologically (Fig. 3) were examined after 12 weeks. In the sham (empty defect) control, a thin layer of fibrous tissue was observed over the entire defect but with an absence of bone or foreign materials. In contrast, when Vitoss (βtricalcium-phosphate) was implanted, there were agglomerates and remnants of Vitoss bridging the defect area (visible in black). On high power field, it appears that Vitoss induced a dense fibrous tissue with some blood vessels and few inflammatory cells without evidence of new bone formation. Post implantation of 3D LCM scaffolds, no new bone formation was observed, but a significant difference in the structure of the scaffolds (scaffolds are blue in the bone sections). The LCM3 scaffold has an intact porous structure with loose connective tissue within the pores containing inflammatory cells and blood vessels. LCM4 also contains loose connective tissue, inflammatory cells and blood vessels, but is smaller and thinner than LCM3 despite having the exact same size at implantation. In contrast, LCM6.1 appears to be separated into 2 layers and contains less connective tissue, inflammatory cells and blood vessels within the pores compared with LCM3 and 4 scaffolds. Because there was no bone formation observed, no quantitative bone evaluation was done. Thus, the raw results are shown in histological sections. To assess the immunological responses to the biomaterials, both in vivo and in vitro responses were tested. Firstly, a high throughput in vivo model was used for rapid screening of immune responses by implanting LCM scaffolds i.p. and measuring the cellular response and production of IL-1β, IL-2 and IL-4 (Fig. 4). In the negative sham control, less than 2 x 10 6 cells were observed and were predominantly peritoneal macrophages, with few eosinophils, lymphocytes, or neutrophils. In contrast, the positive Vitoss control had more than 4 x 10 6 cells which included a significant increase in macrophages, eosinophils and lymphocytes. All the LCM scaffolds generated minimal inflammatory responses that were similar to the sham control (Fig. 3A). These results were mirrored by the cytokine production with baseline values for IL-1β, IL-2 and IL-4 in the peritoneal fluid (Fig. 4B) whereas Vitoss significantly increased IL-1β and IL-4 compared to the sham controls in B6 and BALB/c mice, respectively.
To assess whether the LCMs induced inflammatory or fibrotic changes upon implantation, a subchronic model was used to examine histological sections for evidence of inflammation, foreign body giant cells (FBGCs) and fibrosis 12 weeks after implantation. The LCM scaffolds were compared to both the minimal inflammation in the sham control and the moderate response with FBGCs in response to Vitoss (Fig. 5). LCM3 and LCM4 induced mild inflammation without FBGCs compared to a more intense response observed with the LCM6.1 scaffolds. Figure 5 illustrates that Vitoss induced more collagen deposition compared with LCM3 and LCM4, which had collagen fibers around degraded scaffold filaments and more collagen was observed to be present in the LCM6.1 implantation site. Figure 4: Scaffold evaluation in a rapid high throughput mouse model. Female BALB/c and B6 mice were implanted i.p. with either LCM scaffolds, Vitoss, or no materials (sham). Seven days later, mice underwent peritoneal lavage. The lavage fluid was analyzed (A) for type and number of inflammatory cells (data are presented as mean cell counts ± SEM) and (B) for cytokine concentrations (data are presented as mean cytokine concentrations pg/ml ± SEM). These data are representative of two independent experiments (n=5). *p<0.05, **p<0.01 and ***p<0.001 is considered significant compared to sham.
To determine whether the in vivo responses could be predicted by an in vitro assay, BALB/c and B6 splenocytes were incubated with LCM3, LCM4 and LCM6.1 scaffolds or Vitoss and then IL-1β, IL-2, IL-4 and IFNγ production and cell proliferation (Fig. 6) were measured. When the splenocytes were incubated with medium alone or in the presence of LCM scaffolds and Vitoss, cytokine secretion and cell proliferation were similar. However, Figure 5: Evaluation of scaffolds in a subchronic mouse model. Female BALB/c mice were implanted s.c. with either LCM scaffolds, Vitoss, or no materials (sham). At 12 weeks after implantation, skin from the implantation sites were incised and stained with H&E and Masson's Trichrome to evaluate wound healing and fibrosis. Photomicrographs of stained skin sections are shown at 10x and are representative of two independent experiments (n=5). Arrows indicate FBGCs around the implant. *denotes degraded scaffolds.
in the presence of the mitogen, ConA, there is an expected increase in cell proliferation and cytokines for medium, Vitoss, LCM3, LCM4 and LCM6.1. There were minor differences in the magnitude of the responses between mouse strains, with the most evident difference being a higher BALB/c vs. B6 IL-4 response. Thus, the in vitro studies were predictive of the in vivo assay, though unable to provide information about chronicity of inflammation and capacity to induce fibrosis.

In vitro degradation and mechanical performance
In an attempt to determine whether physical properties of the scaffolds could correlate with the biological findings, the degradation and mechanical performance of the LCM scaffolds in vitro was explored. Figure 7A shows the change in mass loss and pH of LCM3, LCM4 and LCM6.1 scaffolds with respect to time throughout degradation in PBS at 37°C for up to 83 days. Mass loss profiles for all scaffolds showed a two-stage linear increase with mass loss increased initially at a lower rate up to 28 days and followed by a more rapid rate until day 90. At the end of degradation experiment, LCM3 showed lower mass loss (ca. 24%) in comparison with LCM4 and LCM6.1 (ca. 30 and 36% respectively). The pH of the degradation medium, PBS, remained neutral for all specimens at approximately 7.4 ± 0.1 until the end of degradation period at 90 days. The integrity of the scaffolds was traced during degradation by imaging at all time points using 4x magnification (Fig. 7B). LCM3 and 4 scaffolds remained intact until the end of the degradation study at day 90, whereas LCM6.1 collapsed at day 69. The change in the image colour was due to change of the microscope camera during the study.
Change in compressive modulus and strength of LCM3, 4 and 6.1 scaffolds vs. degradation time is shown in Figure 8. Compressive moduli for LCM3, LCM4 and LCM6.1 scaffolds were ca. 0.18, 4.4 and 5.5 MPa, respectively before degradation. Compressive strengths at 20% strain for LCM4 and LCM6.1 scaffolds were more than 15 times higher than LCM3 (0.06MPa). No changes were seen in both strength and modulus for all scaffolds over 85 days of degradation in PBS at 37 o C.

Discussion
The overall goal of this study was to evaluate the biological and physicochemical characteristics of novel customizable, biodegradable, biomimetic TPP-generated LCM copolymers as support scaffolds for bone fracture healing. We assessed the PLCL LCM3, LCM4 and LCM6.1 copolymers using in vitro and in vivo models and found an evident influence of the copolymer composition in all tests. The LCM6.1 scaffold reduced OC differentiation and OB proliferation compared with LCM3 and 4, while OB differentiation was similar for all scaffolds in vitro. LCM3 was more effective in the early stages of bone healing in the calvarial defect model than LCM4 and LCM6.1. In the in vitro splenocyte assays, all LCMs were immunologically inert. LCM3 and LCM4 scaffolds were well tolerated in the in vivo assays, though LCM6.1 induced more intense inflammation and fibrosis comparably in the subchronic mouse model. In in vitro degradation studies, LCM4 and LCM6.1 compared to LCM3 scaffolds had faster degradation rates and reduced mechanical retention. Our study demonstrates that TPP-generated LCM copolymer scaffolds, especially LCM3 with a lactide to caprolactone monomer ratio of 16:4 performs well in all tests. Overall, our results emphasize that polymer composition Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins determines physical and biological properties and that these characteristics correlate. Our combined bone, immunological and physiochemical tests exemplify the benefit of a multidisciplinary strategy for preclinical evaluation of novel biomaterials for bone regeneration. The innovative TPP microfabrication technology was used in manufacturing LCM scaffolds due to the ability to produce 3D polymeric microstructures of almost any desired complexity on the basis of computer models and form features of approximately 100 nm. The Schwarz P unit cell minimal surfaces have high stability at low weight 38 and larger fluid permeability in comparison to other pore structures geometries. 39 The Schwarz P structure is especially useful because it consists of channels and pores resembling cancellous bone which allows cell migration.
Scaffolds are structures designed to provide stability and to guide new bone formation. To determine the effect of LCM3, 4 and 6.1 scaffolds on bone repair, these materials were tested in a calvarial defect model in mice and it was found that the scaffolds alone did not induce new bone formation. Our findings support previous studies using a rat calvarial model 40 and a mandibular defect repair model in rabbits with PLGA scaffolds. 41 Moreover, we predicted this result because for normal bone healing additional osteoinductive stimuli are necessary. 42 However, the scaffolds, compared to the sham controls, were associated with more early stage angiogenesis and tissue repair suggesting minimal bone repair in response to LCM3, 4 and 6.1 scaffolds.
There were two characteristics differentiating LCM3 and LCM4 from LCM6.1: After 12 weeks, LCM3 and 4 scaffolds were 1) thicker and 2) were associated with more cellular infiltrates than LCM6.1. A possible explanation for these observed differences may be due to degradation (thinning and fracturing of the scaffold of LCM6.1 in histological sections) and potential toxic degradation products leading to reduced cellular infiltration. These results are supported by the findings that there was accelerated degradation of LCM6.1 during in vitro degradation studies and the finding that LCM6.1 was associated with a different immune response compared to the other scaffolds. Taken together, data from in vivo and in vitro biological studies correlate with in vitro degradation studies.
While the LCM scaffolds did not induce bone formation in our calvaria model, it is important to note that Vitoss, a commercially available bone substitute material, did not induce new bone either. In support of this finding, Barbiere et al. implanted Vitoss in combination with Bioglass granules in a canine intraspinous implantation model and showed no bone formation. 43 In contrast, however, in both a mouse ectopic bone formation model and in a rabbit critical defect femur model, Vitoss has been reported to induce bone formation. 44,45 There are several possible explanations for these contradicting data including the animals, e.g., mouse vs. rabbit and differences between the protocols. For example, one possibility is that bone healing rates differ between long bone and calvarial bone. Another possible explanation is that in the ectopic model, the skin provides a more reactive environment, e.g., contains more inflammatory and other mediators promoting bone formation than in the calvarium. It is also possible that based on studies of clinical traumatic fractures, Vitoss works optimally because of the inflammatory environment in the injured site that provides mediators essential for bone repair. 44 This differs substantially from the calvarial implantation site in this study, which was almost free of blood and thus, lacks circulating and local inflammatory mediators that might explain the lack of bone formation with Vitoss.
The in vivo bone repair studies indicated that early stages of repair occurred with differences between the LCM scaffolds tested. To further understand the LCM effects observed in vivo, we assessed the cellular response to the LCMs by investigating in vitro OB and OC responses. All the scaffolds increased OB proliferation compared with OM controls. However, only LCM3 resulted in a sustained increase in proliferation with comparably much less for LCM6.1. In vitro data show that LCM6.1 reduced OB proliferation and OC differentiation, whereas LCM6.1 > LCM4 > LCM3 increased OB mineralization while there was no influence by any scaffold on ALP activity at the time point measured. The difference in mineralization may be ascribed to the huge disparity in the compressive moduli of the scaffolds, LCM6.1 and 4 scaffolds are more than an order of magnitude higher than LCM3 (0.18 MPa). This finding supports the previous study on 3D hydrogel scaffolds with moduli ranged from 0.01 MPa to 0.3 MPa, Please do not adjust margins Please do not adjust margins which concluded that as the modulus increased to ca. 0.23 MPa, a significant increase in OB mineralization was observed. 46 These results also support a previous study with poly(L-lactic acid) that showed changes in OC differentiation. 47 Interestingly, the mechanical stiffness of the scaffolds (LCM6.1 > LCM4 > LCM3) appears to correlate with in vitro findings and supports previous studies showing that stiff-more than soft-substrates influenced OB differentiation and osteogenesis [48][49][50] , suggesting that increased mechanical stiffness modulates cell attachment, proliferation and differentiation. Remarkably, in vitro responses did not predict the outcome of in vivo bone repair which was also observed by other groups summarized in a review 51 and might be because the duration of the in vivo experiments (weeks) was longer than in vitro incubation periods (days), or because the LCMs are hydrophobic and could lead to the OBs taking longer to achieve confluence compared to cultures with osteogenic medium without scaffolds. Additional explanations may be related to the increased mass loss due to degradation of LCM4 and 6.1, compared with LCM3 (Fig. 7), resulting in increased degradation products influencing mineralization, or even the experimental conditions used. For example, in vitro differentiation is done with medium containing ascorbic acid and β-glycerophosphate to induce collagen production and promote differentiation to mature mineralizing OBs, whereas, there is no stimulation of osteogenesis in vivo because the scaffolds are not osteoinductive.
The early phase of bone healing is enhanced with an inflammatory microenvironment without chronic and aggressive inflammation, immune or allergic responses that may lead to reduced bone healing. 13 To assess immunogenicity and allergenicity of the scaffolds, we used in vivo and in vitro mouse models and found that implantation of LCM3, LCM4 and LCM6.1 in the peritoneum (rapid throughput model) or under the skin (subchronic model) induced little inflammation in the peritoneum and at the implantation sites, respectively. LCM6.1 scaffold induced more collagen fibers in the implantation site compared to LCM3 and LCM4, which is likely to be related to its faster degradation. This is consistent with previous studies showing that degradation products might cause undesired effects via local pH changes causing inflammation. 21 Cellular immune assays revealed that LCM3, LCM4 and LCM6.1 were unable to stimulate or inhibit splenocytes upon primary stimulation indicating that these polymers are not toxic. Three immunological assays were performed to assess the immunogenicity and allergenicity of the scaffolds. The in vitro assays provide information on cytotoxicity, the i.p. response provides in vivo data on the inflammation and the cytokine milieu, while the s.c. reaction is long-term and provides information on the longevity of the inflammation and capacity to induce fibrosis. Our results support differences in inflammation caused by the LCMs in the calvarium experiments, with LCM3 and LCM4 but not LCM6.1 associated with inflammatory cell infiltrates at 12 weeks. While our i.p. and s.c. results demonstrate that these polymers are not highly immunologically reactive, they do appear to induce minimal inflammation in the bone, which is necessary for bone healing. 12,13,52,53 To ensure optimal bone repair, additional factors should be used in conjunction with the scaffolds. This is further supported with clinically relevant, Vitoss that is a successful bone graft material, because it contains bovine collagen type I that activates inflammation and OCs and OBs required for bone healing. 54,55 The LCM scaffolds were designed for mechanical support and stability at the tissue site until new bone is fully matured and able to withstand mechanical load. Additionally, LCM scaffolds were intended to degrade and be resorbed at a controlled rate to allow in vivo formation of the new bone tissue. 56 Optimal degradation kinetics of scaffolds used for bone lesion healing remains unknown and requires further research. Ideally, the optimal polymer scaffold material should degrade and resorb at the same rate as tissue growth into the implant so it can be replaced with natural bone. PLA and PCL have been approved by the Food and Drug Administration and are suitable for cartilage and bone implants. 26 Importantly, their degradation and mechanical profiles can be tuned based on monomers ratio within their copolymer. They commonly require 1-3 years for complete degradation and their copolymers degrade faster than the homopolymers, depending on the LA to CL ratio 26 . An approach to achieve scaffolds with controlled degradation rates that fit specific biomedical application demands may be to use copolymers of two polymers with different degradation kinetics. 57 It is advantageous to use an elastic and soft material to fabricate mechano-active scaffolds that can be used under cyclic mechanical strain. Jeong et al. 58,59 synthesized very flexible scaffolds based on copolymers of poly(glycolide-co-ɛcaprolactone) and PLCL. Those scaffolds maintained a recovery of approximately 96% for two weeks under cyclic tensile strain (5 to 20%) in PBS. 59 Degradation in biological systems was validated by a non-biological in vitro assessment of degradation. LCM3 scaffolds revealed lower rate of mass loss compared with LCM4 and LCM6.1 (Fig. 7A), which was attributed to differences in LA/CL ratio between LCM3, 4 and 6.1 scaffold materials. LA/CL ratio for LCM3, LCM4 and LCM6.1 was 16/4, 18/2 and 9/1 respectively. 27 Degradation time, the required time for complete resorption, of PDLLA is approximately half of that for PCL. [60][61][62] Therefore, the increase in LA content would lead to a higher degradation rate. This explains the lower rate of mass loss for LCM3 in comparison with LCM4 and LCM6.1. However, LCM4 and LCM6.1 scaffolds contain the same LA/CL ratio, LCM6.1 degraded faster than LCM4. This was attributed to the difference in their initial molecular weight, ca. 1059 and 1822 g mol -1 for LCM6.1 and LCM4 respectively. 27 The difference in degradation rates among LCM scaffolds was also supported by optical images of the scaffold throughout the degradation period (Fig. 7B). LCM6.1 disintegrated into fragments by day 69, while LCM3 and 4 scaffolds remained intact till the end of the degradation study, which was ascribed to the fast degradation rate of LCM6.1 composition. The scaffolds lost ca. 20 -35% on their initial weight throughout the degradation period, however the pH of the degradation medium remained at neutral levels (7.4 ± 0.1) as a result of the buffering capability of PBS, gradual mass loss of the scaffolds and replacing the medium in a weekly basis.
Compressive properties of these scaffolds are mainly dependent on their compositions, specifically their LA/CL ratio, as all scaffolds have approximately similar porosity and pore sizes. 27 Elastic modulus for PCL is approximately five-fold lower than PDLLA. 60,61 Consequently, it was expected that mechanical 10 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins properties of the scaffolds would decrease as the CL content increased within their copolymer. Bramfeldt et al. 63 demonstrated that rates of degradation and mechanical loss of PLCL copolymer increased gradually as the amount of LA changed from 0 to 70%.
Faster degradation means more inflammation and more tissue damage over the long term. A good correlation was observed between in vitro and in vivo findings. LCM6.1 scaffolds showed faster degradation rate compared to LCM3 and 4 and consequently, revealed the least evidence for early stages of new bone formation. Such validated in vitro assays address the 3Rs research directed towards reducing animal experimentation with predictable in vitro assays and suggest that they are useful for pre-clinical assessment of bone repair biomaterials.

Conclusions
Our study focused on a novel biomimetic, biodegradable twophoton engineered PLCL repeating 3D Schwarz P minimal surface unit cell copolymer scaffold for the orthopedic surgeon to use for pain relief and to promote the repair of large non-union bone fractures. An integrated, multidisciplinary platform combining expertise in bone biology, immunology and materials science is a robust strategy which includes: 1) an in vitro evaluation of OBs and OCs and an in vivo mouse calvarial defect model, 2) in vitro and in vivo evaluation of inflammatory, immune and allergic reactions to biomaterials, and 3) in vitro degradation and mechanical compression and strength studies. Our results indicate that these copolymers will perform well in the clinic for bone repair because they can be tailor-made for exact fit in large bone lesions, the scaffolds with the optimal copolymer composition are immunologically compatible, allow for normal OB and OC function and promote bone healing. Moreover, these scaffolds have mechanical properties designed in consultation with surgeons to have at least 20% recoverable compression for fitting them into bone fracture sites. The degradation studies in vitro agree with in vivo observations and indicate that they degrade in a suitable period which is sufficient for physiological bone replacement. This study illustrates an important new biomaterial with a strategy for pre-clinical evaluation of biomaterials developed for bone repair.

Conflicts of interest
There are no conflicts to declare.