Study Design and Overview

For in vitro characterization of the effect of the devitalization process on the engineered callus-mimetic spheroids, human MSCs were embedded in a collagen gel and chondrogenically differentiated or stimulated towards a hypertrophic state prior to devitalization. The ECM of the devitalized constructs was characterized and compared with the one of vital control samples.

For the in vivo studies, the same four groups (vital chondrogenic or hypertrophic; devitalized chondrogenic or hypertrophic) were used but with constructs derived from rat MSCs. Allogeneic rat MSCs were encapsulated in a collagen gel, differentiated, and implanted in a rat of a different strain, having a fully functional immune system. The constructs were implanted subcutaneously (n = 6 for each group) and in a critical size femur defect in rats. For the subcutaneous implantation, a carrier material (collagen) group was included. For the femur defect, n = 8 was used for the experimental conditions (devitalized chondrogenic and hypertrophic) and n = 4 was used for the control conditions (vital chondrogenic and hypertrophic). The regeneration induced by collagen carrier control in the orthotopic defect was already evaluated in a previous study and was proven to be very limited (≈5%).[11] For these reasons, this group was not included in the present study. A summary of previous findings for the collagen control can be found in Figure S4, Supporting Information, or in Ref. [11]. The overall experimental outline is depicted in Figure 1.

Isolation and Expansion of Human and Rat Bone Marrow-Derived MSCs

Human MSCs were isolated from bone marrow aspirates of three patients (donor 1: 20-year old, female; donor 2: 60-year old, female; donor 3: 20-year old, female) after informed consent, in accordance to a protocol approved by the local Medical Ethics Committee (TCBio-08-001-K University Medical Center Utrecht). Ficoll-Paque (GE Healthcare, Little Chalfont, UK) was used to separate the mononuclear fraction, which was further selected based on plastic adherence as previously described.[29] Adherent cells were cultured at 37 °C under humidified conditions and 5% carbon dioxide (CO2) in MSC expansion medium consisting of α-MEM (22 561, Invitrogen, Carlsbad, USA) supplemented with 10% heat-inactivated fetal bovine serum (S14068S1810, Biowest, Nuaillé – France), 0.2 mm L-ascorbic acid 2-phosphate (A8960, Sigma-Aldrich, St. Louis, USA), 100 U mL−1 penicillin with 100 mg mL−1 streptomycin (15 140, Invitrogen), and 1 ng mL−1 basic fibroblast growth factor (233-FB; R&D Systems, Minneapolis, USA).

Rat MSCs were isolated from 4-week old Dark Agouti rats (Envigo, Indianapolis, USA) with the approval of the Central Authority for Scientific Procedures on Animals (CCD, no. AVD1150020172465) and the animal ethical committee of the University Medical Center Utrecht. Briefly, the rats were euthanized through CO2 asphyxiation. After removal of the epiphysis, bone-marrow was obtained by flushing through the diaphysis with MSC expansion medium supplemented with 0.025% ethylenediaminetetraacetic acid (EDTA). Cells were allowed to adhere in a Petri dish overnight. Afterwards, StemX Vivo medium (CCM004, R&D Systems) was used for sub-culturing.

Both rat and human MSCs were passaged at 80% confluency until passage 4.

Generation of MSC Callus-Mimetic Spheroids

At passage 4, human or rat MSCs were chondrogenically differentiated. Human MSCs were used for the ECM characterization, whereas Dark Agouti rat MSCs were used for the in vivo experiments. Briefly, collagen spheroids were created by encapsulating MSCs (20 × 106 mL−1) in 50 µL collagen type I gel droplets (4 mg mL−1) (354 249, Corning, New York, USA), according to the manufacturer's instructions. After gelation, the samples were cultured in serum-free chondrogenic medium consisting of high glucose DMEM (31 966, Invitrogen) with 1% insulin-transferrin-selenium (ITS) + premix (354 352; Corning), 10−7m dexamethasone (D8893; Sigma-Aldrich), 0.2 mm L-ascorbic acid 2-phosphate (A8960, Sigma-Aldrich), 100 U mL−1 penicillin, and 100 mg mL−1 streptomycin (15 140, Invitrogen). To differentiate human MSCs, the medium was supplement with 10 ng mL−1 TGF-β1 (Peprotech, New Jersey, USA). For rat MSCs, also 100 ng mL−1 BMP-2 (InductOS, Wyeth/Pfizer, New York, USA) was added. Medium was refreshed daily for the first 4 days and afterwards three times per week. After 21 days of chondrogenic differentiation, half of the number of spheroids was subjected to hypertrophic medium, consisting of DMEM (31 966, Invitrogen), 1% ITS + premix, 100 U mL−1 penicillin with 100 mg mL−1 streptomycin, 0.2 mm L-ascorbic acid-2-phosphate, 1 nm dexamethasone, 10 mm β-glycerophosphate (G9891; Sigma-Aldrich), and 1 nm 3,3′,5-triiodo-L-thyronine (T2877; Sigma-Aldrich). Differentiation in chondrogenic or hypertrophic medium proceeded for 10 additional days till day 31.

Devitalization Procedure of the Spheroids and Viability Analyses

At 31 days, samples were harvested and devitalized by a mild procedure including lyophilization (European Patent Application no. 20 195 800.6). To confirm devitalization, bioreduction of resazurin sodium salt (R7017; Sigma-Aldrich) was assessed. Briefly, the vital and devitalized chondrogenic and hypertrophic constructs were incubated for 18 h at 37 °C in the dark with 500 µL of 10% resazurin sodium salt in chondrogenic medium without TGF-β1. Absorbance was measured on a spectrophotometer at 570 and at 600 nm for background correction (Versamax; Molecular Devices, Sunnyvale, USA). Data are presented as percentage, considering the resazurin reduction of the vital chondrogenic and hypertrophic groups as 100%. The values obtained from empty collagen controls were subtracted.

To further confirm the absence of viable cells, constructs were digested using a 3 mg mL−1 collagenase type II (LS004177, Worthington; Lakewood, NJ, USA) in phosphate-buffered saline (PBS) digest solution for a minimum of 2 h at 37 °C. The extracted cells were stained with 0.5 µg mL−1 Calcein-AM (Molecular Probes, Thermo Fisher Scientific, Massachusetts, USA) for 30 min at 37 °C. Samples were excited at 495 nm and emission was registered at 515 nm (ASCENT Fluoroskan plate reader; Labsystem). For quantitative analysis, the signal was calibrated with known numbers of living MSCs to produce a standard curve. Images were acquired using an Olympus IX53 inverted fluorescence microscope. The spheroid digests were then re-plated in a 96-well plate and incubated with MSC expansion medium for 2 days to check for any remaining cell viability and the capacity to adhere to tissue culture plastic. Wells were washed with PBS, fixed in 10% neutral buffered formalin, and stained with methylene blue (341088-1G, Sigma-Aldrich) for 5 min. Images of the monolayers were taken with an Olympus IX53 inverted microscope. At least three constructs per condition for each donor were used.

Histological Analysis of Vital and Devitalized Human MSC-Derived Cartilage Constructs

After fixation, samples were dehydrated in a series of increasing ethanol solutions (70–100%) and cleared in xylene. Subsequently, the samples were embedded in paraffin and sliced into 5 µm thick sections (Microm HM340E; Thermo Fischer Scientific). Prior to staining, tissue sections were deparaffinized with xylene and gradually rehydrated through decreasing ethanol solutions (100–70%).

To identify cell nuclei, collagenous fibers, and glycosaminoglycans (GAGs), sections were triple stained with Weigert's hematoxylin (640 490; Klinipath BV), fast green (FN1066522; Merck), and Safranin-O (FN1164048213; Merck). To detect mineralization, von Kossa staining was performed by incubating the sections with 1% silver nitrate (209 139, Sigma-Aldrich) directly under a light bulb (Philips Master TL5HO 54W 830, 1 m distance), for 1 h. The samples were subsequently washed with 5% sodium thiosulfate (A17629, Alta Aesar, Haverhill, USA) and counterstained with haematoxylin.

For collagen type II (0.6 µg mL−1, II-II6B3, Developmental Studies Hybridoma Bank) and collagen type X (10 µg mL−1, 1-CO097-05, clone X53, Quartett, Germany) endogenous peroxidase activity was blocked by incubating samples for 15 min with 0.3% H2O2. For collagen type II staining, antigen retrieval was done by a sequential treatment of 1 mg mL−1 pronase (Sigma-Aldrich) and 10 mg mL−1 hyaluronidase (Sigma-Aldrich) for 30 min each at 37 °C. For collagen type X staining, antigens were retrieved by sequential incubation with 1 mg mL−1 pepsin (Sigma-Aldrich) at pH 2.0 for 2 h and 10 mg mL−1 hyaluronidase for 30 min, both at 37 °C. Prior to primary antibody incubation, samples were blocked with 5% BSA/PBS for 30 min at room temperature. Samples were incubated with the primary antibody overnight at 4 °C. After 30 min of incubation with the secondary BrightVision antibody (VWRKDPVM110HRP, BrightVision poly HRP-anti-mouse IgG, VWR, Radnor, USA), the labels were visualized by 3,3′-diaminobenzidine oxidation. Sections were then counterstained with haematoxylin, washed, dehydrated, and mounted with Depex mounting medium. Mouse isotypes (X0931, Dako, Santa Clara, USA) were used as negative controls at the same concentration as the primary antibodies.

Images were taken with an Olympus BX51 microscope (Olympus DP73 camera, Olympus, Hamburg, Germany). Histology of empty collagen control can be found in Figure S5, Supporting Information.

Biochemical Analysis

For total protein quantification, samples were digested with 0.5 mg mL−1 collagenase II for 5 h at 37 °C. Protein concentration was determined using the Pierce BCA protein assay kit (23 225, Thermo Fisher Scientific) according to manufacturer's instructions. Known concentrations of bovine serum albumin were used to create a standard curve. Absorbance was measured at 562 nm.

Samples for GAG and collagen analysis were digested overnight at 60 °C in papain digestion buffer (250 µg mL−1 papain, 0.2 M NaH2PO4, 0.1 EDTA and 0.01 m DL-cysteine hydrochloride; all from Sigma-Aldrich). The total amount of GAGs was determined using the 1,9-dimethyl-methylene blue (DMMB pH 3.0; Sigma-Aldrich) assay.[30] Known concentrations of shark chondroitin sulfate C (Sigma-Aldrich) were used as standard. Absorbance values were detected at 525 and 595 nm.

To measure hydroxyproline content, 50 µL of the papain digests of all the samples were freeze-dried overnight. Afterwards, samples were hydrolyzed by sequential incubation with 0.4 m NaOH at 108 °C and 1.4 m citric acid. Hydroxyproline contents were measured using a colorimetric method (extinction 570 nm), with chloramine-T and dimethylaminobenzaldehyde as reagents as previously described.[31] Hydroxyproline (Merck) was used as a standard.

Alkaline phosphatase (ALP) activity was measured by using the p-nitrophenyl phosphate (pNPP) substrate system (N2765; Sigma). Different concentrations of ALP with a known activity (U per milliliter) were used as standard curve. The constructs and the standard series were incubated with the pNPP substrate at 37 °C for 8 min. Absorbance was measured at 405 nm with 655 nm as a reference wavelength.

The DNA content was quantified using a Quant-iT Picogreen dsDNA assay (P11496, Thermo Fisher Scientific) according to the manufacturer's instructions.

Four constructs were used per condition for each donor in all analyses. Three MSC donors were used for the analysis of GAGs, collagen, ALP, and DNA. Due to inferior proliferation capacity and shortage of primary cells obtained from one donor, only two out of three MSC donors were used for the total protein quantification.

Evaluation of the ECM Porosity and Susceptibility to Degradation

Fixed samples were dehydrated using a critical point dryer (CPD 030, Bal-Tec) for SEM. After gold sputtering (JEOL, JFC-1300, JEOL Ltd, Tokyo, Japan), samples were imaged using a SEM (JEOL JSM-5600, JEOL Ltd).

For the degradation study, samples were incubated with 10 U mL−1 collagenase II (Worthington) in plain DMEM at 37 °C. Medium was collected and completely replaced after 1, 2, 4, 12, 18, 24, 36, 48, 60, 72, and 80 h. The collected medium was processed as described above to measure hydroxyproline content.

To indirectly measure the porosity of the constructs, samples were immobilized at the bottom of a custom-made mold of 3% agarose gel. The top of the spheroid was exposed to Visipaque solution (iodixanol, GE Healthcare) and microCT images (Quantum FX; PerkinElmer, Waltham, USA) were taken at different time points (20 µm resolution, voltage 90 kV, current 180 mA, field of view = 10 mm). For each spheroid, the diameter was measured and a ROI of 0.1 × 1.8 mm was selected at the center of the spheroid. The changes in average pixel intensity within the ROI due to the inward diffusion of the contrast agent were monitored over time using the image processing software Image-J (Java, Redwood Shores, USA).

One MSC donor was used to evaluate construct porosity and susceptibility to degradation. A triplicate was used for the quantitative measures whereas one sample per group was used for qualitative images.

Construct Preparation for In Vivo Implantation

Chondrogenic differentiation and metabolic activity of the Dark Agouti MSCs was verified prior to in vivo implantation (Figures S6 and S7, Supporting Information). For subcutaneous implantation, two chondrogenic spheroids per group (vital chondrogenic and hypertrophic; and devitalized chondrogenic and hypertrophic) were embedded in collagen (4 mg mL−1) and cast in custom-made square cuboid molds (3 mm x 3 mm x 2 mm). Gelation was allowed for 45 min at 37 °C according to manufacturer's instructions. Empty collagen controls were included as controls. For the orthotopic defects, eight chondrogenic spheroids were encapsulated in collagen gel in 3.5 mm x 3.5 mm x 6 mm custom-made molds, as described above. The constructs were prepared the day before implantation and incubated overnight in a chondrogenic differentiation medium without TGF-β1 and BMP-2.

Animal Experiment and Surgical Procedures

The animal experiments were performed with the approval of the Central Authority for Scientific Procedures on Animals (Dutch national CCD) and of the local animal welfare body (2465-2-01) in accordance with the ARRIVE guidelines for animal experimentation.[32] The power analyses used to determine the number of samples required per group are presented in the Supporting Information. Twenty-four male Brown Norway rats of 11 weeks old (Envigo) were randomly housed in pairs at the Central Laboratory Animal Research Facility of the Utrecht University. Animals received standard food pellets and water ad libitum, under climate-controlled conditions (21 °C; 12 h light/12 h darkness). After 7 days of acclimatization, subcutaneous pockets were created under general anesthesia from 5 mm dorsal incisions and blunt dissection as previously described[9] (1-3.5% isoflurane in oxygen, AST Farma, Oudewater, the Netherlands). In each pocket, one construct of either group (collagen control, vital chondrogenic, vital hypertrophic, devitalized chondrogenic, or devitalized hypertrophic) was implanted (n = 6 per group). The skin was closed transcutaneously with Vicryl Rapide 4-0 sutures (VR 2297; Ethicon). Each animal received a maximum of 2 subcutaneous pockets. For implantation of the construct in a femur defect, a 6-mm critical-size segmental bone defect was created as previously described[33] (n = 8 for the devitalized chondrogenic and hypertrophic experimental groups and n = 4 for the vital chondrogenic and hypertrophic controls). Briefly, the right hind leg was shaved and carefully disinfected. A lateral skin incision was made and soft tissue was dissected in order to expose the right femur. After the periosteum removal, three proximal and three distal screws were used to stabilize a 23 × 3 × 2 mm polyether ether ketone (PEEK) plate to the femur in the anterolateral plane. After fixation, a saw guide and a wire saw (RISystem, Davos, Switzerland) were used to remove a 6-mm cortical bone segment. The collagen constructs were press-fit into the defect and a single dose of antibiotic (Duplocillin LA, 22.000 IE/kg, MSD Animal Health, Boxmeer, the Netherlands) was locally injected intramuscularly. The fascia and skin were sutured in layers using resorbable Vicryl Rapide 4-0 sutures (Ethicon). Subcutaneous injection of pain medication (buprenorphine, 0.05 mg kg−1 bodyweight, AST Farma, Oudewater, the Netherlands) was given pre-operatively and twice a day for the following 3 days. When devitalized constructs were implanted in the orthotopic defect, the rats also received only devitalized constructs subcutaneously. Rats were euthanized 12 weeks after surgery with an overdose of barbiturates (phenobarbital; 200 mg kg−1 body weight, TEVA Pharma, Haarlem, the Netherlands). The femora and the subcutaneous implants were retrieved and processed for histological analysis and micro-computed tomography (microCT) scanning.

MicroCT Scanning

Mineralization in the orthotopic defect area was assessed at 0, 4, 8, and 12 weeks after surgery. While under general anesthesia, the hind leg of the rat was fixed to a custom-made support to allow scanning of the femur with a microCT imaging system (Quantum FX). Three minutes of scan time was required per leg for an isotropic voxel size of 42 µm resolution (voltage 90 kV, current 180 mA, field of view = 21 mm). All scans were oriented in the same fashion using the ImageJ plugin Reorient3 TP (Image-J 2.0.0; Java, Redwood Shores, CA, USA). A volume of interest (VOI) of 6.3 × 5 × 5 mm3 was selected. After euthanasia, subcutaneous implants were also analyzed. Three minutes of scan time was required per subcutaneous implant for an isotropic voxel size of 20 µm resolution (voltage = 90 kV, current = 180 mA, field of view = 10 mm). After segmentation with a global threshold, the mineralized volumes (MV) for both the subcutaneous and femur implants were measured in millimeter cube using the image processing software plugin BoneJ[34] (Image). 3D reconstructions of the femur defect were based on the microCT data and created using ParaView (ParaView 5.3.0, Kitware Inc., USA).

Histological Analysis of the In Vivo Samples

All specimens were fixed in a 10% neutral buffered formalin solution for 1 week and thereafter decalcified for 6 weeks in a 10% EDTA-phosphate buffered saline solution (pH 7.4). After decalcification, samples were additionally fixed for 2 days, dehydrated in a Leica ASP300S tissue processor in graded ethanol solutions (70–100%), cleared in xylene, embedded in paraffin and sliced into 5 µm thick sections (Microm). Before staining, samples were deparaffinized with xylene and gradually rehydrated through decreasing ethanol solutions (100–70%). New bone formation was evaluated using H&E, Masson–Goldner trichrome staining and Safranin-O/fast green staining.

Histomorphometric analysis was performed for both the subcutaneous and orthotopic samples after H&E staining. Briefly, an overview of the whole sample was made by merging images into a panoramic image in Adobe Photoshop C6. For the subcutaneous implants, bone formation throughout the entire construct area was quantified. For the orthotopic implants, a region of interest (ROI) of 6.5 × 5 mm2 was selected in the center of the defect. The titanium screw holes present on each side of the defect were used as reference points in order to ensure an equivalent positioning of ROI in all samples. Three different areas were manually selected for each ROI: bone, hypertrophic cartilage, and bone marrow. The number of pixels for each area was quantified via the function “recording measurement” and expressed as a percentage of the total construct area for the ectopic implants and of the ROI area in the orthotopic ones. The sections were scored independently by two scientists and the results are presented as an average.

Statistics

A randomized block design with Bonferroni's post hoc correction was applied for the in vitro data to accommodate donor variation, including viability and biochemical analyses (protein, GAG, hydroxyproline and DNA content and ALP activity). A linear mixed model followed by a Bonferroni's post hoc correction was used to compare mineralization in the femur defect over time and to evaluate statistical differences in the Visipaque diffusion test (IBM SPSS 22.0, New York, USA). For the histomorphometric measures, when data were normally distributed, a one-way ANOVA test was performed, followed by Tukey post-hoc test (GraphPad Prism 6, San Diego, CA, USA). When the condition of normality was not satisfied, a Kruskal–Wallis test, followed by a Dunn's post hoc test was performed. Differences were considered to be statistically significant for p < 0.05.