Comparison of 5 BMPs for their chondrogenic potentials and microfracture-mediated cartilage repair using heparin/PEAD coacervate sustained release polymer

Abstract

Cartilage defect repair and osteoarthritis treatments remain clinical challenges. Microfracture is a commonly used surgical procedure for the treatment of cartilage defects but often leads to fibrocartilage repair. The aim of this study is to compare the effects of 5 bone morphogenetic proteins (BMPs) on chondrogenic differentiation of human bone marrow mesenchymal stem cells, as well as to investigate the use of the heparin/poly (ethylene arginine aspartate diglyceride (PEAD) coacervate sustained release system to deliver these BMPs for microfracture-mediated cartilage repair. Our results indicate that all 5 human BMPs significantly enhance the chondrogenic differentiation of human bone marrow mesenchymal stem cells (hBMMSCs) with BMPs 2,4 and 9 being more potent than BMP6 or BMP7, as revealed by Alcian blue, SO staining, and immunohistochemistry of COL2. Coacervate-BMPs are biocompatible for both hBMMSCs and rat muscle-derived stem cells (MDSCs) and promote their proliferation. In vivo, sustained release of human BMPs 2,4,6,7,9 with heparin/PEAD coacervate significantly enhances microfracture-mediated cartilage repair in a rat osteochondral defect model, as demonstrated by ICRS macroscopic score, Seller's histology score, and COL2 staining. These effects are mediated by increasing SOX9 expression in the regenerated cartilage. In conclusion, BMPs 2,4,9 are the most potent BMPS to promote chondrogenic differentiation, while all BMPs enhanced microfracture-mediated cartilage repair when delivered with heparin/PEAD coacervate without a significant difference between the different BMPs.

Graphical abstract

Highlights

• •Comparison of 5 BMPs for promoting chondrogenic differentiation of human bone marrow MSCs.
• •Heparin/PEAD polymer for sustained-release 5 different BMPs.
• •Heparin/PEAD for sustained-release 5 BMPs for microfracture-mediated cartilage repair.

1. Introduction

Microfracture is a commonly used surgical technique for the treatment of cartilage injury that releases endogenous stem cells from subchondral bone to repair cartilage defects [[1], [2], [3]]. However, microfracture treatment alone is known to produce primarily fibrocartilage [4,5]. Previously, we demonstrated that oral or intra-articular administration of losartan improved microfracture-mediated cartilage repair by increasing the quantity of hyaline cartilage [6,7] through a reduction of fibrosis. Intra-articular injection of a vascular endothelial growth factor (VEGF) antibody also improved microfracture-mediated cartilage repair by blocking angiogenesis [8]. However, these strategies detailed above focused on blocking deleterious factors of cartilage regeneration without providing chondrogenic factors. Therefore, developing novel strategies that involve providing chondrogenic factors such as BMPs are also attractive to further promote the benefit of microfracture.

Coacervate is a polymer of synthesized poly (ethylene arginine aspartate diglyceride) (PEAD) and heparin that can sustain-release growth factors and has been used to repair a variety of tissues, including skin wounds as well as heart, bone, and cartilage defects, among others [[9], [10], [11], [12], [13]]. Our research team has previously shown that coacervate sustained release bone morphogenetic protein 2 (BMP2)/soluble fms-like tyrosine kinase 1 (sFLT1) can promote cartilage repair to the same extent as lenti-BMP2 transduced human muscle-derived stem cells using a monoiodoacetate (MIA)-induced osteoarthritis model in rats [14]. Our previous study also compared 5 BMPs for bone regeneration and revealed that BMP2 is the most potent BMP to promote bone regeneration using the coacervate sustained release system [15]. Since BMPs are also chondrogenic, in this study, we tested whether coacervate sustained release of these BMPs can also promote microfracture-mediated cartilage repair.

BMPs were first discovered by Marshall R Urist [16]. They belong to the transforming growth factor (TGF) superfamily. Approximately 20 BMPs have been identified to date, each with distinct and versatile functions [17,18]. Two pioneering studies compared 14 BMPs for their bone formation efficiency and found that BMPs 2,6,7, and 9 promoted bone regeneration, while BMP3 inhibited bone formation [19,20]. In addition to promoting bone regeneration, BMPs are also extensively studied for chondrogenic differentiation and cartilage repair. BMP2 has been shown to promote chondrogenic differentiation in vitro and cartilage repair in vivo when delivered with different scaffolds (collagen sponge, PLGA microspheres dispersed in a Pluronic F-127 hydrogel) or with adenoviral or lentiviral vectors [[21], [22], [23], [24], [25], [26], [27]]. BMP4 has been shown to promote chondrogenic differentiation of stem cells in vitro and cartilage repair using various cartilage defects or osteoarthritis models when delivered with retrovirus, adenovirus transduced stem cells or scaffold such as collagen, alginate gel, hydrogel [[28], [29], [30], [31], [32], [33], [34], [35], [36], [37]]. BMP6 (500 ng/ml) has also been shown to induce stem cells chondrogenic differentiation in vitro [38,39], but no in vivo study has been reported. Human BMP7 has been used to repair cartilage defects in different animal models including rabbits, dogs, and goats using different scaffolds such as [40], N,N-dicarboxymethyl chitosan [41], autologous coagulated blood [42,43], collagen 1 scaffolds [44], PLGA scaffolds [45], porous tantalum (PT) [46], and a polyhedron delivery system (PODS) [47]. BMP7 enhanced microfracture-mediated cartilage repair when delivered with a collagen sponge [48]. Delivery of BMP7 via gene modification of chondrocytes or stem cells enhanced auricular cartilage defect healing in a rabbit cartilage defect model [[49], [50], [51]], while adenoviral-transduced chondrocytes to express BMP7, accelerated cartilage repair in horses [52].

BMP9 has been shown to promote stem cell chondrogenic differentiation in vitro [22,53]by activation of the pSMAD1/5 and P38 signaling pathways and is not inhibited by noggin [54]. Intra-articular injection of BMP9-transfected adipose-derived stem cells (ADSCs) promotes cartilage repair in a murine osteoarthritis model, via activation of the Notch/Jagged 1 signaling pathway [55]. Delivery of BMP9 to an amputated mice limb promoted new joint structure formation, including a skeletal element lined with articular cartilage and a synovial cavity, through activation of proteoglycan 4 (PRG4) [56]. However, most of the previous studies used high doses of BMPs. In vivo viral vector therapy or ex vivo viral vectors transduced with BMPs have not been translated into clinical treatments due to safety concerns. Therefore, developing sustained release biomaterials to deliver small doses of BMPs needs to be further studied. It is still unknown which BMP is most potent for promoting cartilage repair.

This study aims to compare the potency of 5 BMPs for enhancing the chondrogenic differentiation potential of human bone marrow mesenchymal stem cells (BMMSCs) in vitro and to use the coacervate sustained release platform to sustain release 5 different BMPs at a lower dose to promote microfracture-mediated cartilage repair in vivo in a rat osteochondral defect model.

2. Materials and methods

2.1. Human BMMSCs isolation

Human BMMSCs were isolated from bone marrow obtained from the femoral heads of 3 patients who had undergone total hip arthroplasty (an 81 year old female (81F), a 66 year old female (66F), and a 52 year old male (52M)) using a protocol previously reported by Dr. Rocky Tuan's lab [57,58]. Briefly, trabecular bone was cored out using a curette or rongeur and then flushed with a medium containing α-MEM (12571-063, Gibco, ThermoFisher Scientific),1 % antibiotic-antimycotic (Invitrogen, CA, USA) using 18-gauge hypodermic needles. The bone chips were then cut with scissors. The flushed medium was passed through a 40 μm mesh cell strainer to remove debris and centrifuged for 5 minutes (min) at 300G. Pellets were washed twice with medium and resuspended in growth medium (GM) containing α-MEM, supplemented with 10 % fetal bovine serum (FBS, Invitrogen, CA, USA), 1 % antibiotic-antimycotic (Invitrogen), and 1 ng/ml FGF2 (233-FB-010,R&D system), and plated in two 150 cm² tissue culture flasks. On day 4, the cells were washed with PBS and fresh GM was added. GM was changed every 3–4 days. Once the cell cultures reached 70–80 % confluency, the cells were detached with 0.25 % trypsin containing 1 mM EDTA (Invitrogen, CA, USA), frozen, and stored in liquid nitrogen for further analyses.

2.2. Comparison of the chondrogenic potential of 5 BMPs using hBMMSCs

3D chondrogenic pellet cultures were performed for the three populations of hBMMSCs isolated as described above, using StemPro complete chondrogenic medium (StemPro™ Chondrogenesis Differentiation Kit, A1007101, Thermo Scientific) supplemented with 50 ng/ml Recombinant Human/Mouse/Rat BMP-2 Protein (355-BM-050/CF R&D system), Recombinant Human BMP-4 (314-BP-050/CF,R&D system), Recombinant Human BMP-6 Protein (507-BP-020/CF, R&D system), Recombinant Human BMP-7 Protein (354-BP-010/CF, R&D system) and Recombinant Human BMP-9 Protein (3209-BP-010/CF, R&D system). The results were compared to the complete chondrogenic medium without BMP as control (CTL) (N = 4 pellets each treatment). The pellets were cultured for 24 days with medium change every 3 days, harvested 24 days after differentiation, fixed with 10 % neutral buffered formalin (NBF), and washed with PBS. The pellets were imaged for diameter measurement using a ruler as a reference. The pellets were then embedded in NEG-50 freezing medium, and 8 μm cryosections were cut for further histological analysis. All in vitro experiments were repeated three times. All three populations of cells generated similar results. The results for only one population of hBMMSCs are presented.

2.3. Histology analysis and immunohistochemistry staining of chondrogenic pellets

Cryosections of chondrogenic pellets were used for different staining. Alcian blue staining was performed using IHC world online protocol (https://ihcworld.com/2024/01/26/alcian-blue-staining-protocol/). Cartilage hyaluronic acid and acid mucin were stained in blue, the nucleus was stained in red, and the cytoplasm was stained in pale red. Safranin O staining was also performed using IHC world protocol (https://ihcworld.com/2024/01/26/safranin-o-staining-protocol-for-cartilage/#google_vignette). Glycosaminoglycan-rich tissue stained red-orange and other connective tissue stained in green while the nuclei stained in black. Most of the reagents were purchased from Sigma. Wiegert Hematoxylin (SKU: 26044-06 and 26044-16) was purchased from the Electron microscope company.

Immunohistochemistry staining of collagen type 2 (COL2), collagen 10 (COL10), and PRG4 was performed as previously described [14]. Rat distal femur cartilage was used as positive control. Briefly, cryosections of chondrogenic pellets were dried for 10 min at room temperature and then refixed in NBF for 8 min and washed three times with PBS. The pellet sections were then subjected to antigen retrieval using 2 % hyaluronidase (H3506-1G) in PBS for 30 min at room temperature (RT) and then washed three times with PBS. Sections were further blocked with 5 % donkey serum in PBS for 1 hour (h) at room temperature, followed by incubation with primary antibodies at 4 °C overnight. Primary antibodies were diluted in 5 % donkey serum. The primary antibodies’ dilutions were mouse anti-COL2 (MA1-37493, Thermo Fisher Scientific, 1:200 dilution), mouse anti-COL10 (14-9771-82, Thermo Fisher Scientific, 1:50 dilution) and mouse anti-PRG4 (MABT400, EMD Millipore, 1:500 dilution). The following day, the sections were treated with 0.5 % hydrogen peroxide (H2O2) in PBS for 30 min at room temperature, washed in PBS, and then incubated with biotinylated horse anti-mouse secondary antibody (BA 2000, Vector Laboratories, Burlingame, CA, USA, 1:300 dilution) for 2 h at room temperature. After three washes, each slide was incubated with ABC reagent (PK 6100, VECTASTAIN® Elite® ABC-HRP Kit, Peroxidase (Standard), Vector Laboratories) for 2 h at room temperature. After three washes with PBS, diaminobenzidine (DAB) color reaction kit (SK-4100, Vector Laboratories) was used to visualize the COL2-,COL10 or PRG4-positive cells. Hematoxylin (H-3404, Vector laboratories) counterstaining was performed following the DAB color reaction. Histology and immunohistochemistry images were captured using a NIKON-Ti microscope. The positive Alcian blue, Safranin O, and COL2 and COL10 matrix were quantified using Nikon NIS software, by defining views of the interest of the pellets and then picking up blue, orange-red, or brown pixels and expressing these as percentages of total pellet area, respectively.

Immunofluorescence staining for COL2 and collagen 1 (COL1) was also performed. The slides were treated the same as IHC above including antigen retrieval with 2 % hyaluronidase in PBS and blocking. The mouse anti-COL2 (7048, Chondrex, 1:250) and rat anti-COL1 (7086, Chondrex, 1:250) were prepared in 5 % donkey serum, and the slides were incubated overnight at 4 °C. On second day, the slides were washed with PBS for 3 times and then incubated with donkey anti-mouse-488 (715-545-150, Jackson ImmunoResearch Laboratory, 1:200 dilution) and donkey anti-rat (IgM + IgG)-594 (112-585-068, 1:200 dilution) secondary antibodies at room temperature for 2 h. Slides were then washed with PBS for 3 times and counterstained with 4′,6-diamidino-2-phenylindole (DAPI, 1ug/ml,Thermo Fisher Scientific) for 10 min in PBS. Slides were then washed with PBS 3 times and DI water 3 times, and coverslip with Aqueous Mount Medium (H-5501-60). Fluorescent images were taken with a Nikon Eclipse NI upright microscope using NIS software. A blank control was used to set up exposure times to differentiate signal versus noise.

2.4. In vitro biocompatibility testing of coacervate and BMPs on human BMMSCs

Human BMMSCs were seeded in 96-well plates at 2000 cells/well density with 5 replicates in each group. 24 hrs after cells seeding, PBS or the different coacervates in 13.2 μl were added to 4 ml of medium to make a final concentration of 50 ng/ml for the BMPs. Treatments included adding 13.2 μl of the following: PBS group, add 13.2 μl cell culture PBS; CTL, add 4 μl 4 mM HCL +2 μl heparin+7.2 μl PEAD; coacervate-BMP2, add BMP2 in 4 mM HCL 4 μl (200 ng)+2 μl heparin+7.2 μl PEAD; coacervate-BMP4, add BMP4 in 4 mM HCL 4 μl (200 ng)+2 μl heparin+7.2 μl PEAD; coacervate-BMP6, add BMP6 in 4 mM HCL 4 μl (200 ng)+2 μl heparin+7.2 μl PEAD; coacervate-BMP7, add BMP7 in 4 mM HCL 4 μl (200 ng)+2 μl heparin +7.2ul PEAD; and coacervate-BMP9, add BMP9 in 4 mM HCL 4 μl (200 ng)+2 μl heparin+7.2 μl PEAD to 4 ml MSC complete medium. Cells were cultured for 48 or 72 h after the different treatments in 0.1 ml volume. Then, 20 μl of CellTiter 96® AQueous One Solution (Cat#: G3580, Promega Inc.) was added to each well and incubated at 37C for 2 h, then the absorbance at 490 nm was measured with a Tecan Plate Reader. A490 absorbance represents cell proliferation/viability.

2.5. Immunofluorescence staining of cell proliferation marker Ki67

In addition, the same treatment groups as above were used for 24-well plate seeding at 1 × 10⁴ cells/well density. After 24 h the cell culture medium was switched to a medium contains coacervate or coacervate BMPs (7 groups of treatment as above) for 72 h. Cells were then fixed with 4 % paraformaldehyde for 8 min and washed with PBS. After 3 washes using PBS, cells were permeabilized with 0.1 % Triton X 100 in PBS for 30 min and blocked with 5 % donkey serum (Jackson ImmunoResearch) in 0.1 % Triton X 100 in PBS for 1 h. Cells were then incubated in Ki67 primary antibody made in 0.1 % Triton X 100 in PBS (1:100 dilution, 14-5698-82, Invitrogen, Themo Fisher Scientific). The blank control was incubated in 5 % donkey serum in 0.1 %Triton X 100 without primary antibody. After overnight incubation, 24 well plates were washed with 0.1 % Tween 20 containing PBS (PBST) for 5 min and then incubated with donkey anti-rat-488 (1:200 dilution, 712-545-153, Jackson ImmunoResearch Laboratories) for 2 h at room temperature. Following 3 washes in PBST, nuclei were counterstained with DAPI for 30 min in PBST. Cells were further washed three times with PBST and once in PBS, and 1 ml PBS was added to each well for imaging. Fluorescence images were captured using a NIKON-inverted microscope. Positive cell percentages were quantified using Image J and Ki67⁺ cell percentage was used to reveal cell proliferation.

2.6. RNA extraction, cDNA synthesis and quantitative polymerase chain reaction (Q-PCR) analysis for cell proliferation markers

Furthermore, cells were cultured in 24-well plates for 72 h treatment. The cells from the 7 groups were lysed with Trizol (Invitrogen) and RNA was extracted following the manufacturer's protocol. Complementary first strand DNA (cDNA) synthesis was performed using iScript™ Reverse Transcription Supermix for RT-qPCR (Cat#:1708841, BioRad) following the manufacturer's protocol using a 20 μl reaction. Q-PCR was performed using SsoAdvanced Universal SYBR® Green Supermix, 2500 × 20 μl rxns, 25 ml (5 × 5 ml) (Cat#:1725274, BioRad). Primers were designed using the Primer 3 Input [[59], [60], [61]]. Primer sequences are provided in Supplemental Table 1.

2.7. Comparison of 5 BMPs for their chondrogenic potential to promote rat muscle-derived stem cells (rMDSCs) in vitro

To test if human BMPs are also functional in promoting rat cell chondrogenic differentiation, we isolated rMDSCs using preplate-techniques as previously described [62]. Rat MDSCs were cultured in DMEM high glucose (Cat#:11995-073, Gibco) supplemented with 20 % FBS (A5670402, Gibco, Thermo Fisher Scientific), 1 % chicken Embry extract (Accurate Chemical Co) and 1 % P/S (Cat#:15140122, Gibco, Themo Fisher Scientific) as previously described [63]. Chondrogenic differentiation was conducted using the same protocol as stated in 2.2 for human BMMSCs. BMPs dissolved in 4 mM HCL were added to the complete Stem-Pro chondrogenic medium to result in a final concentration of 50 ng/ml. For CTL group, 4 mM HCL was used. After 24 days of chondrogenic differentiation, pellets were fixed in NBF for 1 h and embedded in NEG freezing medium. Cryosections were cut into 8 μm and histology staining of the chondrogenic pellets were performed as 2. 3. Pellet size was measured using the maximum section diameter of each pellet as the overall pellets are much smaller than hBMMSCs pellets. The COL 2 antibody used in the rMDSCs pellets immunohistochemistry was mouse anti-collagen 2 diluted at 1:500 (7048, Chondrex Inc).

2.8. Test biocompatibility of coacervate BMPs for rMDSCs

Similar in vitro experiments were performed as 2.4 to 2.6. Rat MDSCs were cultured in 24-well plates for 48 h before harvest for Ki67 immunofluorescent staining and RNA extraction for Q-PCR. Ki67 staining was performed using the same protocol as 2.5. The Q-PCR genes tested were bone morphogenetic protein receptor 2 (BMPR2) and sirtuin 1 (SIRT1). The primer sequences are listed in Supplemental Table 1.

2.9. Animal use

The use of animal for in vivo experiments was approved by the Institutional Animal Use and Care Committee of Colorado State University (Protocol#4839). Sprague Dawley (SD) rats at 12 weeks old were purchased from Taconic Biosciences and housed in room 106 in the Bay Facility of Colorado State University with 12 h light/dark cycles.

2.10. Synthesis of PEAD and preparation of coacervate BMPs

PEAD was synthesized in Dr. Yadong Wang's lab as previously described [64]. Heparin for making coacervate was obtained from Scientific Protein Labs and is medical USP grade. Recombinant Human/Mouse/Rat BMP-2 Protein, CF(355-BM-050/CF), Recombinant Human BMP-4 Protein, CF (314-BP-050/CF), Recombinant Human BMP-6 Protein, CF (507-BP-020/CF, R&D system), Recombinant Human BMP-7 Protein, CF(354-BP-010/CF), Recombinant Human BMP-9 Protein, CF(3209-BP-010/CF, R&D system) were first dissolved in 4 mM HCL within a week of surgery and then diluted with PBS to the final concentration of 0.4 μg/μl and aliquoted in each tube that was enough for 8 rats and stored in −80 °C until surgery. Each rat required 1.25 μl, which is equal to 500ng/defect.

2.11. In vitro coacervate/Fibrin sealant gel formation and feasibility in rat osteochondral defect model

To test the feasibility of coacervate BMPs in a fibrin sealant scaffold, an in vitro assay was performed in a PCR test tube to test if heparin in coacervate affects fibrin gel formation. Twice the volume intended for use in vivo was used to facilitate visualization. BMP2 in 2.5 μl (1 μg) was added to 4 μl Heparin, allowed to bind for 10 min at room temperature, then 14.4 μl PEAD solution was added and solution become turbid, then 15 μl thrombin (Tisseel Fibrin Sealant Kit) was added to the coacervate, followed by adding 15 μl fibrinogen and pipetting up and down a few times, the solution formed a gel. We also performed pilot study and created osteochondral defect (1.5 mm diameter × 1.5 mm depth) with 1 microfracture in 3 rats, then we added coacervate-BMPs and fibrinogen mixture to the defect, followed by adding thrombin and observed the hydrogel formation. The rats were sacrificed at day 1, and the osteochondral defect was dissected, and the presence of coacervate-fibrin sealant hydrogel was observed. Micro-CT and H&E staining were performed to view the defect and coacervate-BMPs hydrogel morphology.

2.12. Osteochondral defect and microfracture surgery and group assignment

48 Sprague Dawley (SD) rats were divided into 6 groups (N = 8 with 4 males and 4 females in each group): (1) PBS + coacervate (PBS); (2) 500 ng BMP2+coacervate (BMP2); (3) 500 ng BMP4+coacervate (BMP4); (4) 500 ng BMP6+coacervate (BMP6); (5) 500 ng BMP7+coacervate (BMP7); (6) 500 ng BMP9+coacervate (BMP9). Tisseel fibrin sealant 2 ml kit, including recombinant human thrombin and fibrinogen (NDC 00338-4210-02, SKU TIS821002, Baxter Healthcare Corporation) was used as a scaffold for all groups. The thrombin was dissolved in calcium chloride, and fibrinogen was dissolved in a protease inhibitor (aprotinin) at 37 °C on the day of surgery following the manufacturer's instructions and kept at room temperature without loss of activity. Creation of 1.5 mm osteochondral defects and microfracture was performed using similar techniques as previously reported in rabbits [7,8]. Briefly, rats were anesthetized with 3 % isoflurane, bupivacaine was injected at the incision sites, buprenorphine sustained release (SR) was injected in the neck area by subcutaneous injection; then, a 2 cm incision was made in the medial side of the left knee, the knee capsule was opened with a #15 scalpel, the patella was flipped to the lateral side to open the knee joint and expose the trochlear groove cartilage, and a 1.8 diameter trephine drill was used to create a 1.5 mm diameter and 1.5 mm depth osteochondral defect. The cartilage was then removed using 1.4 mm burr. Saline irrigation was performed during the drilling process. Next, one microfracture hole was created in the middle of the defect with a 0.7 mm burr, at approximately 1 mm depth, to release subchondral bone marrow in the defect. Further, coacervate BMPs were added to the osteochondral defect using a pipet tip ring to maintain the coacervate in the defect area. The preparation of coacervate-BMPs was as follows: BMPs, prepared as stated above in 1.25 μl, were added to 2 μl heparin (10 mg/ml) and allowed to bind for at least 10 min before being placed on ice, right after the creation of the osteochondral defect and microfracture. Then, 7.2 μl PEAD (10 mg/ml) was added to each tube to form coacervate (solution turned turbid). 7.5 μl thrombin was mixed with coacervate-BMP complex and added to the osteochondral defect. Immediately, 7.5 μl fibrinogen was added, and the coacervate-BMPs complex was allowed to gel in the defect area. After coacervate-BMPs were applied, the patella was relocated to its anatomic position, the capsule was sutured with 6-0 Vicryl suture, and the skin was closed with 4-0 prolene sutures. Following closure, the rats were ear-tagged and recovered in warm chambers before being returned to their home cages. Eight weeks after the surgery, the rats were euthanized using 3 % CO2 and the injured distal femurs were dissected. Gross images were taken for the ICRS macroscopic scoring and then the femurs were fixed in NBF for 4 days before micro-CT scanning and histology.

2.13. MicroCT scanning and analysis

After fixation in NBF, the entire distal femurs with the osteochondral defects were scanned using 31.2 μm voxel size, 70kVP, 113 μA with Vivo-CT 80. The 3D reconstruction was performed using Gauss support = 0.8, Gauss sigma = 1, threshold = 212 utilizing built-in software using program 1 (1 solid object). 3D micro-CT images were used for determining subchondral bone healing of the osteochondral defect. To quantify new bone formation in the osteochondral defect area, 50 slices of defect region were contoured (31.2μm × 50 = 1560 μm) and bone microarchitecture parameters were analyzed using same thresholds as above using bone morph program 5 in the Scanco Micro-CT scanner Built-in software. Bone volume (BV), bone volume/total volume (BV/TV) and trabecular number (Tb.N) were reported.

2.14. Osteochondral defect tissue histology

Following micro-CT scanning, the tissues were decalcified with 5 % formic acid made in H₂O for 2 weeks. The tissues were then processed via dehydration in gradient alcohol, cleared with xylene, paraffin immersion, and paraffin embedded using a HeiDi Tissue Processor and Embedder station, and sectioned with a Shandon microtome. Alcian blue and Safranin O staining were performed to reveal the cartilage matrix, as stated in section 2.3. Seller's histology scoring was performed, as reported in the literature [21]. Immunohistochemistry staining of COL2 and SOX9 was performed, as described in section 2.4, after deparaffinization and rehydration with water. The SOX9 antibody (PA5-81966, Invitrogen, Thermo Fisher Scientific) dilution was 1:500 in 5 % donkey serum. Quantification of SOX9⁺ cells was performed by taking 200x magnification images of the entire regenerated osteochondral defect and counting SOX9⁺cells and total nuclei in the cartilage area, as well as calculating the SOX9⁺ cell percentage.

2.15. Statistical analysis

Animal numbers for each group were determined using power analysis set alpha = 0.05, 1-β = 0.8 to detect the main outcome, cartilage repair Seller's histology, we need N = 8 rats. All data analysis was performed using Graphpad Prism 10 using ANOVA followed by Tukey's Post-hoc multiple comparisons. P < 0.05 was considered statistically significant.

3. Results

3.1. BMPs dramatically increased hBMMSC pellet size and cartilage matrix deposition

We performed 3D pellet culture of hBMMSCs using 50 ng/ml of BMPs 2,4,6,7, and 9 for 24 days. Gross images of the pellets demonstrated that BMP-treated groups showed larger and more transparent morphology than the control (CTL) group. Quantification showed that all BMP-treated groups had significantly larger pellet diameters than the CTL group (P < 0.0001 for all comparisons). The pellet size of the BMP2,4,6,7, and 9 groups increased an average of 71.7 %, 68.7 %, 37.2 %, 31.6 % and 49.6 %, respectively, compared to the CTL group for all the three MSCs populations tested including the population presented in Fig. 1D. The BMP 2,4 & 9 groups also had larger pellet sizes than the BMP 6 and 7 groups (All P < 0.0001, Fig. 1A–D). Alcian blue staining and quantification revealed that all BMP groups had significantly larger blue stained matrix areas than the CTL group. Further, the BMP 2,4 & 9 groups revealed entirely blue stained pellets, while the CTL group, as well as the BMP 6 and 7 groups, showed a small central portion of the pellets not stained blue. All of the BMP groups demonstrated significantly higher percentages of blue cartilage matrix compared to the CTL group, while no statistically significant differences were found between the different BMP groups (Fig. 1B–E). We further performed Safranin O staining to reveal glycosaminoglycans (GAG). The majority area of the pellets of all BMP groups stained in orange-red color, compared to relatively large negatively stained portions of the pellets in the CTL group. High magnification revealed partial peripheral staining of the pellets in an orange-red color in CTL group, while the BMP groups showed a completely orange-red stained matrix at the periphery of the pellets. Quantification of the percentage of orange-red matrix demonstrated significantly higher percentages of GAG in all BMPs groups compared to the CTL group (P < 0.0001 for all comparisons). The BMP4 group also showed a significantly higher percentage of the orange-red matrix than the BMP7 group (Fig. 1C–F).

Fig. 1.

Comparison of the effect of 5 BMPs on the chondrogenic differentiation of hBMMSCs. (A) Gross images of pellets of each group. All BMP groups showed more transparent morphology than the CTL group. (D) Pellet diameter. The BMP 2,4,9 groups have larger pellet sizes than control as well as the BMP6 and 7 groups. All BMP treated groups showed significantly larger pellet diameters than the CTL group. (B) Alcian blue staining. Entire pellets were stained blue with BMP 2,4 & 9, while small portions of the pellets were not stained blue in the CTL, BMP6 and 7 groups. (E)Alcian blue matrix percentage. All BMP groups showed significantly higher percentages of blue stained matrix area than the CTL group. (C) Safranin O staining. The majority part of the pellets in the BMP groups are stained in orange red compared to relatively large portions of negatively stained area of the pellets in the CTL group. High magnification (200×) showed partial, peripheral staining of the pellets in the CTL group, while the BMP groups showed full orange-red stained matrix at the periphery of the pellets. (F)Safranin O orange-red matrix percentage by group. All BMP groups demonstrated significantly higher percentages of orange-red matrix compared to the CTL group. The percentage of orange-red matrix in the BMP4 group was also significantly higher than in the BMP7 group. Scale bars = 250 μm for 40X and 50 μm for 200X. ∗P < 0.05. ∗∗∗∗P < 0.0001.

3.2. BMP treatment significantly increased COL2 expression in the chondrogenic pellets with BMP4 being the most potent

We performed immunohistochemistry of COL2 to detect cartilage specific matrices. At low magnification, a thin layer of COL2-positive brown matrix was revealed at the periphery of the pellets in the CTL group. In contrast, all BMP groups showed a thick brown COL2 matrix at the periphery of the pellets. At 200X magnification, COL2 could be seen to be partially expressed at the periphery of the pellets in the CTL group and stained a less intense brown. In contrast, all BMPs groups had strong expression of brown COL2 matrix at the entire periphery of the pellets (Fig. 2A). The percentage of COL2 matrix was drastically higher in all of the BMP groups when compared to the CTL group (Fig. 2B). Compared to CTL, adding BMP2, BMP4, BMP6, BMP7 and BMP9 increased COL2 expression by 139.6 %, 191.04 %, 131.19 %, 176.58 %, and 172.77 %, respectively. Immunohistochemistry for COL10, a hypertrophic cartilage marker, was then performed. Our positive control rat patella groove cartilage showed obvious COL10 positive cells in the deep zone as well as highly expressed in the growth plate (Supplemental Fig. 1). We found some extent of expression of COL10 in all groups of pellets. The COL10 expression was mainly located in the middle zone between the peripheral 1/3 and core of the pellets BMP2,4,9 groups compared to CTL and BMP6 and 7 in the area outside of core region. Quantification of the COL10⁺ area percentage indicated that the BMP2,4,7 groups showed no significant differences compared to CTL group. BMP6 group demonstrated a significantly higher COL10⁺ area percentage than CTL group while BMP9 group demonstrated a significantly lower COL10⁺ area percentage than CTL group. The COL10⁺ area percentage in BMP6 group was also significantly higher than BMP2 and BMP9 groups. BMP9 group also showed significantly lower COL10⁺ area percentage than BMP7 group (Fig. 2C and D). Double immunofluorescence staining of COL2 and COL1 was further performed, which revealed that the BMP2, 4, 6, 7, and 9 groups expressed strong COL2 (green) compared to the CTL group. Only a few red stained cells (COL1) were detected in the core of the pellets of the BMP9 group (Fig. 2E, Insets highlighted red COL1⁺cells). Immunohistochemistry staining of PRG4 was also performed. At 40X magnification, PRG4 was not expressed in the periphery (differentiated cells) of the pellets, but rather, in the center of the pellets. At 200X magnification, PRG4 positive cells were located in the central area of the pellets and there was staining in the cell cytoplasm of some of the cells, while no significant differences were found between the BMPs and the CTL group. Interestingly, extracellular expression of PRG4 was observed in the BMP6 and 7 groups (Fig. 2F). To investigate if the core of the pellets underwent mineralization, Von-Kossa staining was performed and revealed minimal mineralization in the periphery of the CTL group and the core of the BMP group. This is likely because the basal medium is osteocyte/chondrocyte medium, some cells may undergo mineralization (Supplemental Fig. 2). This finding may be beneficial in vivo, as in osteochondral defect repair, the subchondral bone healing is very important for cartilage repair.

Fig. 2.

Immunohistochemistry of chondrogenic pellets to detect cartilage specific matrix. (A,B) Immunohistochemistry of COL2 and quantification. COL2⁺ cells and matrix are stained brown color. At low magnification, a thin layer of COL2 at the periphery of the pellets was observed in the CTL group, while all BMP groups showed a thick, brown COL2 matrix at the periphery of the pellets. At 200X, part of the periphery of the pellets in CTL group was stained in a less intense brown, in contrast, all BMPs groups demonstrated a strong brown COL2 matrix in the entire periphery of the pellets. The percentage of COL2 matrix was markedly higher in all BMP groups than in the CTL group. ∗P < 0.05. ∗∗∗∗P < 0.0001. (C) Immunohistochemistry staining of COL10. The COL10 expression was detected in all groups. The COL10⁺ area was mainly located in the middle zone of the BMP2,4 and 9 groups while its expression in CTL, BMP6 and 7 groups was located in the area outside of the core region. (D) Quantification of COL10⁺ area percentage. ∗P < 0.05, ∗∗P < 0.01, ∗∗P < 0.001. (E) Double immunofluorescence of COL2 and COL1. BMPs have stronger COL2 expression (green) than CTL group as demonstrated in A and B. The pellets are mostly negative for COL1, while only a few COL1⁺ cells were found in the core of the BMP9 pellets. (F) Immunohistochemistry staining of PRG4 at 40X magnification and 200X. PRG4 is expressed in the cytoplasm of cells in the center of the pellets. Scale bars = 250 μm for 40X and 50 μm for 200X and 100 μm for 100X.

3.3. Coacervate-BMPs promoted human BMMSC cells proliferation

To test the biocompatibility of coacervate, in vitro cell proliferation assays were performed using the CellTiter 96® AQ_ueous One solution method. The results indicated that coacervate and coacervate-BMPs all significantly increased A490 (high cell number) at 48 h (Fig. 3A). At 72 h, all wells showed higher A490 compared to 48 h. The coacervate and the coacervate-BMP groups all significantly increased A490 compared to the PBS group. No significant differences between the coacervate-BMP groups were present (Fig. 3B). Immunofluorescence staining of Ki67 was also performed. Coacervate and coacervate-BMPs all significantly increase Ki67⁺cell percentage compared to the PBS group (Fig. 4C and D). Furthermore, Q-PCR results of Ki67 indicated that coacervate and coacervate-BMP2 significantly increased Ki67 mRNA compared to the PBS group. BMP4 and BMP6 also showed a trend of increased Ki67 mRNA (P = 0.063 and 0.052, respectively). There were no significant differences between the coacervate-BMP7, coacervate-BMP9, and PBS groups (Fig. 3E).

Fig. 3.

Effect of coacervate on cell proliferation of hBMMSCs. (A–B) Absorbance at 48 and 72 h after cell culture and 2 h after adding CellTiter 96® AQueous One Solution. Coacervate alone and coacervate BMPs all significantly increased A490, which indicates higher cell number. (C) Immunofluorescent staining of Ki67. Ki67 was stained green in the nuclei of all stages of proliferating cells. All nuclei stained blue. All cells that expressed Ki67 in the entire nuclei or part of the nuclei are counted as positive. Scale bars = 100 μm. (D) Quantification of Ki67⁺ cell percentage. Coacervate and coacervate-BMPs groups have significantly increased Ki67⁺ cells percentage compared to the PBS groups. E Q-PCR analysis of Ki67 mRNA expression at 72 h after cell treatment. Coacervate and coacervate-BMP2 significantly increased Ki67 mRNA expression compared to the PBS group. Coacervate-BMP4 and -BMP6 also showed a trend of increased Ki67 mRNA expression. No significant differences were seen between the BMP7 and BMP9 groups. Exact P values are indicated between group bars.

Fig. 4.

Effect of Coacervate and coacervate BMPs on cell proliferation of rMDSCs. (A) Cell proliferation at 48hrs after cells culture. (B) Cell proliferation at 72hrs after cell culture. (C) Representative immunofluorescent images of Ki67 staining. Ki67+cells were stained green. All nuclei are stained blue. Scale bars = 100 μm. (D) Ki67⁺ cells percentage after being treated for 48hrs. (E) Q-PCR analysis of BMPR2 mRNA. Coacervate and coacervate BMPs groups significantly increased BMPR2 expression compared to PBS group. Coacervate BMP9 group expressed significantly higher BMPR2 mRNA than any other groups. (F) Q-PCR results of SIRT1 mRNA. Coacervate and coacervate BMPs all significantly increased SIRT1 compared to PBS group. Coacervate BMP9 also significantly increased SIRT1 mRNA expression than any other groups. For all histograms, Exact P values are indicated between group bars.

3.4. Effects of coacervate-BMPs on the rMDSCs proliferation/viability and gene expression

To test if coacervate and coacervate BMPs affect rat cells proliferation when used in vivo, we performed cell proliferation assay using CellTiter 96® AQ_ueous One Solution. We found that at 48 h after rMDSCs cell culture, coacervate BMP4 and BMP9 groups significantly increased A490 compared to PBS group (Fig. 4A). At 72 h culture, coacervate BMP2,4,6 significantly increased cell proliferation compared to PBS group while BMP7 and BMP9 groups did not significantly increase A490 compared to PBS group. Coacervate-BMP2 and BMP6 groups also demonstrated significantly higher A490 compared to coacervate-BMP9 group (Fig. 4B). Immunofluorescence staining of Ki67 indicated coacervate BMP2,4 groups significantly increased Ki67⁺cell percentage compared to PBS group. Coacervate-BMP2 group also showed significantly higher Ki67⁺ cells percentage than coacervate-BMP7 and BMP9 groups. Coacervate-BMP4 group also demonstrated significantly higher Ki67⁺cells percentage than coacervate-BMP9 group (Fig. 4C and D). Q-PCR results revealed coacervate and all coacervate-BMPs significantly increased BMPR2 expression. Furthermore, coacervate BMP9 significantly increased BMPR2 mRNA compared to coacervate and any other coacervate-BMPs groups (Fig. 4E). Additionally, coacervate, coacervate-BMPs all significantly increased SIRT1 mRNA expression compared to PBS groups. Coacervate-BMP9 group also significantly increased SIRT1 mRNA compared to any other groups (Fig. 4F).

3.5. Comparison of 5 BMPs on chondrogenic differentiation of rMDSCs

To investigate if human BMPs are also functional in promoting rat cells chondrogenic differentiation. We tested the effects of 5 human BMPs on chondrogenic differentiation of rMDSCs. We found rMDSCs pellets partially fell apart likely due to these cells growing very fast, cells can still proliferate during the first few days of chondrogenic differentiation. But the pellets from BMP2,4 9 were obviously bigger. As the pellets were very tiny and ran the risk of loss of pellets by taking out from the tubes for gross imaging, we did not measure pellet sizes using gross images, instead, we measured pellet sizes using the maximum diameter of sections of Alcian blue staining. Our results demonstrated that all BMPs formed significantly larger pellets than CTL group. BMP2,4,9 also formed significantly larger pellets than BMP6,7 groups (Supplemental Fig. 3A and E). Furthermore, Alican blue staining indicated the CTL group showed minimum light blue cartilage matrix while BMP groups demonstrated obvious light blue matrix especially in BMP2,4,9 groups. The percentage of Alcian blue⁺ matrix of the entire pellets was significantly higher in all BMPs groups compared to CTL group. In addition, BMP2,4,9 also demonstrated significantly higher Alcian blue⁺ matrix percentage compared to BMP6 and 7 groups (Supplemental Fig. 3A and F). Safranin O staining revealed no orange-red staining in all pellets which indicated chondrogenic differentiation was still immature. However, the CTL group showed intense green (cytoplasm) which indicated undifferentiated cells, while BMP2,4,9 groups cells have become larger and partially negative for green staining but not yet expressed glycosaminoglycan (GAG) as demonstrated light green staining (Supplemental Fig. 3B). Immunohistochemistry demonstrated that CTL group, BMP6,7 had no COL2⁺cells emerged yet while BMP2,4,9 showed few positive cells which indicated initial chondrogenic differentiation (Supplemental Fig. 3C and insets). Lastly, immunohistochemistry of COL10 revealed no expression in any groups (Supplemental Fig. 3D). Overall, the significantly less intense chondrogenic differentiation of rMDSCs stimulated by human BMPs compared to hBMSCs likely attributed to the cells’ properties, which are different, and the chondrogenic medium is for human cells. But we found the same tendency of BMPs effects on pellet size and Alcian blue staining of chondrogenic pellets of the rMDSCs as we found for hBMMSCs, which indicated human BMPs are functional in rat cells due to high homology.

3.6. Sustained release BMPs with coacervate improved subchondral healing of osteochondral defects and macroscopic score after microfracture

We first tested the feasibility of coacervate and fibrin sealant to form hydrogel scaffold in vitro and in vivo. We found that coacervate-BMPs could form hydrogel after being mixed with fibrinogen and thrombin in vitro in less than 1 min in test tube (Supplemental Fig. 4A). We also demonstrated that coacervate-BMPs and fibrinogen and thrombin could form hydrogel in rat osteochondral defects with microfracture. The hydrogel was present on day 1 and could be removed (Supplemental Fig. 4B). Micro-CT revealed that the osteochondral defect was obvious (Supplemental Fig. 4C). H&E staining revealed the presence of coacervate-BMP hydrogel and a defect. Many cells had migrated into the hydrogel with coacervate-BMPs (Supplemental Fig. 4D–F).

Based on this pilot study, we further used the heparin/PEAD coacervate sustain release polymer to compare 5 BMPs for their capacities to promote microfracture-mediated cartilage repair. At 8 weeks after surgery, micro-CT results demonstrated that all coacervate-BMP groups had better subchondral bone healing compared to the PBS group, but no significant differences were found between the coacervate-BMPs groups. No heterotopic bone formation was observed in any coacervate-BMP groups (Fig. 5A). Further quantification of subchondral bone healing demonstrated increased bone volume in the regenerated subchondral bone. BMP4 significantly increased bone volume while BMP2, 6, and 7 showed an increasing trend (P = 0.058, 0.066 and 0.080, respectively) (Fig. 5B and C). Coacervate-BMP4 and -BMP6 also significantly increased BV/TV, which reflects the bone quality. BMP2, 7, and 9 showed an increased trend of BV/TV (Fig. 5D). Tb.N also increased but did not reach statistical significance (Fig. 5E). Gross image observation indicated that the coacervate-BMP groups showed a relatively smooth, regenerated cartilage surface, with more complete filling of the osteochondral defect. The ICRS macroscopic scores of all coacervate-BMP groups were significantly higher than that of the PBS group. BMPs 4 and 9 showed relatively higher scores than the other BMP groups, but this difference did not reach statistical significance (Fig. 5F and G).

Fig. 5.

Micro-CT and macroscopic observation of the repaired cartilage at 8 weeks after surgery. (A) Micro-CT 3D images of the osteochondral defect healing. All coacervate-BMP groups showed better subchondral bone healing when compared to the PBS group, but there were no significant differences between the coacervate-BMP groups. No heterotopic bone formation was observed in any of the coacervate-BMP groups. Scale bar = 1 mm. Red arrows point to the defect area. (B) Segmental new regenerated bone in the osteochondral defect. (C) New bone volume in the defect area. All groups showed increased bone volume compared to the coacervate group. (D). BV/TV. All BMPs groups also showed increased BV/TV. (E) Tb.N of the new regenerated bone. BMPs groups showed a trend of increased Tb.N. (F)Gross images of osteochondral defect healing. The coacervate-BMP groups showed a relatively smooth regenerated cartilage surface with more complete defect filling. (G) ICRS macroscopic score quantification. Higher scores indicate better healing. Exact P values are indicated between group bars.

3.7. Sustain release of BMPs with heparin/PEAD coacervate improved histology score of the repaired cartilage

Alcian blue staining was performed to reveal hyaluronic acid and acid mucin. The normal group showed blue matrix in different layers of cartilage, as well as scattered positive staining area in the subchondral bone. A limited positive blue matrix was detected in the PBS group of both the best and worst repair rat. In the BMP2 group, blue matrix was found in the regenerated cartilage layers, but chondrocyte morphology was not the typical hyaline cartilage morphology in the rat with the best repair and almost no blue matrix was found in the rat with the worst cartilage repair. For the BMP4 group, the rats with the best cartilage repair showed an intense blue matrix across all regenerated cartilage, with even a tidemark visible, while the rat with the worst cartilage repair showed a slight blue matrix with immature chondrocyte morphology. For the BMP6 group, the rat with the best cartilage repair showed intense blue matrix staining across all layers containing chondrocytes with round morphology, while the rat with the worst cartilage repair showed no blue matrix in the cartilage defect layer. For the coacervate-BMP7 group, the rat with the best cartilage repair demonstrated a relatively thin layer of cartilage with medium intensity of blue staining. The rat with the worst cartilage repair showed a defect that had not healed and no blue matrix. For the coacervate-BMP9 group, the best repair showed mild blue matrix across all layers of cells in the regenerated cartilage while the worst repair showed negative blue matrix despite a healed defect (Fig. 6A–D).

Fig. 6.

Cartilage matrix staining and histology score evaluation. (A–D) Alcian blue staining. Hyaluronic acid and acid mucin were stained blue. Normal patella groove cartilage was stained blue in chondrocytes across all layers as well as extracellular matrix. Subchondral bone also showed scattered blue staining in the trabecular bone which represents unabsorbed cartilage matrix. (A–B) showed the best cartilage repair of each group at 40X and 100X magnification. (C–D) showed the worst repair of each group at 40X and 100X. (E–H) Safranin O staining. GAGs were stained in orange red. Normal cartilage showed positive orange-red staining in the cartilage layers and scatted distribution in the subchondral bone despite different staining intensities. E-F showed the best repair of each group at 40X and 100X magnification. G-H showed the worst cartilage repair in each group at 40X and 100X magnification. (I)Seller's histology score. Coacervate-BMP 2,4,6,7 and 9 groups demonstrated significantly lower histology scores (better repair) than coacervate-PBS group. No significant differences were detected among the BMP groups. Scale bars = 100 μm for 100X, 250 μm for 40X. Exact P values are indicated between group bars.

For the Safranin O staining, GAGs in the cartilage were stained orange red. Normal cartilage showed positive orange-red staining in the cartilage layers and a scattered distribution in the subchondral bone with different staining intensities. In the coacervate-PBS group, the rats with the best and worst repair showed no orange-red matrix. In the coacervate-BMP2 group, the rat with best cartilage repair showed weak orange-red staining in the regenerated cartilage layer while nearly no orange-red matrix was identified in the cartilage of the worst repair rat. For the coacervate-BMP4 group, the best repair rat showed orange-red staining more intensely in the cartilage near subchondral bone while less intense staining on the cartilage surface. No orange-red matrix was found in the cartilage of rat with the worst repair. For the coacervate-BMP6 group, the rat with the best cartilage repair showed intensive orange-red staining across all cartilage layers with a visible tidemark, while the worst repair showed no orange-red matrix staining. For the coacervate-BMP7 group, the rat with the best repair showed weak orange-red matrix staining in parts of the cartilage surface area, while in the rat with the worst cartilage repair, no cartilage regenerated, and no orange-red matrix was identified. For the BMP9 group, weak orange-red staining was identified in the cartilage area near subchondral bone and microfracture hole area for the rat with the best repair, while no orange-red staining was found in the rat with the worst cartilage repair. However, Safranin O staining was overall not intense as compared to normal control in most samples (Fig. 6E–H). Furthermore, Seller's histology score was used to quantify the cartilage repair. The results revealed that all coacervate-BMP groups had significantly lower histology scores (better repair) than the coacervate-PBS group (Fig. 6I).

3.8. Sustained release BMPs with coacervate improved general morphology of the regenerated cartilage

We further performed H&E staining. The normal cartilage group showed nuclei as dark blue, cartilage matrix as bluish, while subchondral bone matrix was stained red. In the coacervate-PBS group, the osteochondral defect area showed no cartilage regeneration, and only fibrosis and subchondral bone were stained. In the coacervate-BMP2 group, the blueish cartilage layer extended to the microfracture hole in the rat with the best repaired osteochondral defect, while mainly fibrotic tissue was found in the osteochondral defect area in the rat with worst repair. In the coacervate-BMP4 group, the regenerated cartilage showed bluish matrix in the rat with the best cartilage repair, while a fissure at the microfracture hole was found in rat with the worst repair with no bluish stained cartilage. For the coacervate-BMP6 group, the rat with the best cartilage repair showed typical cartilage morphology in all the layers with blue matrix, while the osteochondral defect area of the worst repair showed a fiber-like structure. For the coacervate-BMP7 group, a thin layer of cartilage with bluish matrix was found in the cartilage with the best repair, while the cartilage was not healed in the osteochondral defect of rat with the worst repair. For the coacervate-BMP9 group, a typical cartilage layer with bluish matrix was found in the rat with the best repair, while the worst repair showed only a partial cartilage-like structure in the osteochondral defect area (Fig. 7A and B). For Herovici's staining, collagen type 1 (COL1) was stained in a pink-red color and collagen type 3 (COL3) was stained in a dark blue color. In the normal cartilage, COL1 stained in pink-red, COL3 was stained dark blue, and the nuclei stained light blue. The cartilage layer showed mixed blue and pink-red staining, and the subchondral bone was stained an intense pink-red color. In the coacervate-PBS group, the rat with the best repair showed an osteochondral defect area with an intense pink-red color with a fiber structure while surface area was stained light blue. Only subchondral bone was staining in dense intense pink-red in the worst repair rat. For the coacervate-BMP2 group, the osteochondral defect of the best cartilage repair rat showed a mixture of red and blue which extended into the microfracture hole. In the rat with the worst cartilage repair the osteochondral defect was filled with light blue on the tissue surface and intense red below the surface layer. For the coacervate-BMP4 group, the regenerated cartilage in the best repair osteochondral defect showed a mixed red and blue matrix while a fiber-like red and blue matrix was found in the osteochondral defect of the rat with the worst repair. For the coacervate-BMP6 group, the best repair osteochondral defect showed a mixture of red and blue staining while the worst repair osteochondral defect showed a fiber-like pink-red color near subchondral bone with a light blue color on the surface. For the coacervate-BMP7 group, the regenerated cartilage of the rat with the best repair showed a mixture of red and blue, but with a blue color that is relative intense, while in the osteochondral defect in the rat with the worst repair showed a fiber-like red staining. For the coacervate-BMP9 group, the regenerated cartilage in the rat with the best repair showed mixture of red and blue staining that extended to the microfracture hole. However, the osteochondral defect with the worst repair showed red staining with a fiber-like structure with light blue on the tissue surface (Fig. 7C and D).

Fig. 7.

H&E and Herovici's staining of the regenerated cartilage. (A–B) H&E staining to reveal general morphology of regenerated cartilage. The normal cartilage group showed nuclei as dark blue, the cartilage matrix as bluish color, and the subchondral bone was stained red. (A) showed the best repair osteochondral defect at 40X and 100X magnification of each group. (B) showed the worst cartilage repair osteochondral defect of each group at 40X and 100X magnification. (C–D) Herovici's staining. In the normal cartilage, COL1 was stained pink-red, COL3 was stained dark blue, and the nuclei stained light blue. The cartilage layer showed a mixture of blue and pink-red staining, while subchondral bone was stained intense pink red. (C) The best cartilage repair osteochondral defect of each group at 40X and 20X magnification. (D) The worst repair osteochondral defect at 40X and 100X magnification. Scale bar = 100 μm for 100X and 250 μm for 40X.

3.9. Sustained release of BMPs improved microfracture-mediated cartilage repair by increasing expression of COL2 and SOX-9 positive chondrogenic progenitor cells in the cartilage

To further characterize the repaired cartilage, COL2 immunohistochemistry staining was performed. COL2 positive cells and extracellular matrix are stained brown, and the nuclei are stained blue. Normal cartilage showed variation of COL2 staining intensity with lack of staining in the perichondrium while all other cartilage layers are positive for COL2. For the coacervate-PBS group, a patchy COL2-positive area was detected at the osteochondral defect area while the rest of the osteochondral defect area was mostly negative in the rat with the best repair. It was negative for COL2 in the osteochondral defect of the worst repair rat. For the coacervate-BMP2 group, a moderate intensity of COL2 staining was detected in the entire healed defect area for the rat with the best repair, while there were only COL2-positive areas of staining at the surface or near the subchondral bone region for the rat with the worst repair. For the coacervate-BMP4 group, intense COL2 staining was found in the bottom half of the regenerated cartilage and negative on the upper half of the cartilage for the rat with the best cartilage repair. While in the osteochondral defect with the worst repair, COL2 staining was mainly located at the bottom area of the osteochondral defect only. For the coacervate-BMP6 group, the COL2 stained intensely in the entire regenerated cartilage area with negatively stained perichondrium similar to normal cartilage for the rat with the best repair. Only a small portion of the osteochondral defects were positive for COL2 staining in the rat with the worst repair. For the coacervate-BMP7 group strong COL2 staining was found in the entire layers of regenerated cartilage in the rat with the best repair. However, only residual native cartilage was positive in the osteochondral defect of the rat with the worst repair. For the coacervate-BMP9 group, COL2-positive matrix was identified in the regenerated cartilage in the rat with the best cartilage repair with strong COL2 staining located in the bottom of the regenerated cartilage and residual native cartilage on the edge of the defect and weak staining on the surface of the regenerated cartilage. Only weak scattered COL2 staining was found in the osteochondral defect of rat with the worst repair (Fig. 8A-B).

Fig. 8.

Immunohistochemistry staining of COL2 and SOX9. (A–B) Immunohistochemistry of COL2 for the best and worst repair of each group. COL2-positive cells and extracellular matrix stained brown, nuclei stained blue. Normal cartilage showed variations in COL2 staining intensity with perichondrium negative while all other cartilage layers were positive for COL2. (A) COL2 staining of the regenerated cartilage with the best repair for each group at 40X and 100X magnification. (B)COL2 staining of regenerated cartilage with the worst repair for each group at 40X and 100X magnification. (C) Immunohistochemistry staining of SOX9 at 40X and 200X. SOX9⁺ cells stained in brown color in the nuclei. For the normal cartilage, SOX9⁺cells were found in the entire layer of cartilage with around 70 % of were stained positive. More SOX9⁺ cells were found in coacervate-BMP groups. (D) Quantification of SOX9⁺cell percentages. The coacervate-BMP2,4,6,7&9 groups all had significantly higher percentages of SOX9⁺cells in the regenerated cartilage than the PBS group. Exact P values are indicated between group bars. Scale bars = 100 μm for 100X, 250 μm for 40X, and 50 μm for 200X field.

Immunohistochemistry of SOX9 was also performed to detect chondroprogenitor cells. SOX9⁺ positive cells were stained in brown color in the nuclei. For the normal cartilage, SOX9⁺cells were found in the entire layer of cartilage with around 70 % of chondrocytes were positively stained. More SOX9⁺ cells were found in regenerated cartilage in the coacervate-BMP groups (Fig. 8C). Quantification of SOX9⁺ cell percentages demonstrated that the coacervate-BMP2,4,6,7&9 groups all had significantly higher percentages of SOX9⁺ cells in their regenerated cartilage than the PBS group (Fig. 8D). No statistical differences were found among coacervate-BMPs groups.

4. Discussion

This study compared 5 BMPs for promoting chondrogenic differentiation of human BMMSCs and revealed that BMPs2,4,6,7&9 all significantly enhanced human BMMSCs chondrogenic differentiation in vitro, as revealed by 3D chondrogenic pellet size, alcian blue staining, Safranin O staining, and immunohistochemistry of COL2. Furthermore, BMPs 2,4&9 are the most potent BMPs for promoting chondrogenesis. Coacervate and coacervate-BMPs also promoted human BMMSC cell proliferation. Coacervate-BMP2, 4, and 6 also promoted rat MDSCs proliferation while coacervate-BMP7 and 9 did not affect cell proliferation. In vivo, when delivered with the heparin/PEAD-coacervate sustain release polymer, all BMPs significantly improved microfracture-mediated cartilage repair, as demonstrated by ICRS macroscopic scoring and Seller's histology score. The enhancement of microfracture by coacervate sustain released BMPs is mediated by increased chondrogenic differentiation of cells released from subchondral bone via enhanced expression of SOX-9. However, no significant differences were found among the 5 BMPs when delivered with coacervate.

In the field of cartilage biology, transforming factor β3, not BMPs, are usually added to in vitro chondrogenic differentiation medium. However, the current study revealed that adding BMPs 2,4,6,7&9 (50 ng/ml) dramatically increased chondrogenic differentiation of human BMMSCs by increasing pellet size, cartilage specific matrix (Alcian blue and Safranin O positive matrix) and COL2. These results indicate the high efficacy of BMPs in promoting chondrogenic differentiation. BMPs 2,4&9 have also been found to be more potent than BMPs 6 and 7.

Few studies have compared the effects of different BMPs on chondrogenic differentiation and cartilage repair. It has been reported that BMP2 and BMP9 enhanced chondrogenic differentiation by increasing COL2, aggrecan, and cartilage oligomeric protein. BMP2 and BMP9 treatments also partially blocked the inhibitory effects of interleukin 1 (IL-1) on chondrogenic differentiation [22]. BMP4 was shown to be more effective in promoting rabbit MSCs and chondrocytes towards chondrogenic lineage differentiation than BMP7 at a dose of 10 ng/ml [35]. BMP2 was shown to be more potent in promoting umbilical cord blood cell chondrogenic differentiation than BMP6 [39]. Others indicated that BMP9 was significantly more potent in inducing chondrogenic differentiation of mouse C3H10T1/2 and ATDC5 cells compared to BMP2 and BMP6, via activation of pSMAD1/5 while avoiding inhibition by noggin [54]. Adenoviral BMP2- and BMP4-transduced human BMMSCs are both potent in promoting chondrogenic differentiation but showed signs of inducing chondrocyte hypertrophy [31]. In this study, we also detected COL10 expression in CTL and BMP2,4,6,7,9 treated chondrogenic pellets, but except BMP6, BMPs treatment did not significantly increase COL10⁺ area percentage compared to CTL group. BMP9 group had a significantly lower COL10⁺ area percentage compared to CTL group. The current study is the first to compare the chondrogenic differentiation of 5 BMPs and revealed BMPs 2,4&9 are more potent than BMPs 6 and 7 in vitro. Furthermore, coacervate-BMPs are safe and can even promote human BMMSCs and rat MDSC proliferation. Therefore, the coacervate sustained release platform is clinically translatable. Interestingly, all coacervate-BMPs can increase BMPR2 and SIRT1 mRNA expression of rat MDSCs.

Few studies have used different BMPs to enhance microfracture-mediated cartilage repair using different scaffolds. Early studies using a rabbit osteochondral defect model showed that BMP7 alone increased the amount of repair tissue without affecting the quality of repaired cartilage compared to the osteochondral defect control, but the combination of microfracture and BMP7 increased both the quality and quantity of repair tissue compared to BMP7 alone or microfracture alone [48]. Adeno-BMP4 implanted in perforated decalcified cortical bone matrix (DCBM) led to rapid hyaline cartilage repair in an osteochondral defect model with microfracture at 6 weeks and complete cartilage repair at 12 weeks after surgery. The perforated DCBM or DCBM completely repaired the defect with hyaline cartilage at 24 weeks, while the microfracture alone repaired only with fibrotic tissues [33]. Vayas R et al. used injectable BMP2 incorporated into PLGA microspheres dispersed in a Pluronic F-127 solution and BMMSCs for microfracture-mediated cartilage repair. They demonstrated that all the treated groups, except the microfracture group, responded significantly better than the control group (non-treated defect) as revealed by histology. Although no significant differences were found among the treated groups, only the BMP2(12ug/scaffold), MSC-BMP2(12ug), and MFX-BMP2 (3ug/scaffold) groups showed non-significant differences when compared with the normal cartilage [25]. Our previous study showed coacervate sustained release BMP2 promoted MIA-induced osteoarthritis cartilage repair [14]. No studies have used the same delivery system to compare different BMPs for microfracture-mediated cartilage repair. Our results reveal that all BMPs promote microfracture-mediated cartilage repair when delivered with coacervate, but there were no differences between BMPs 2,4,6,7&9 in a microfracture-mediated cartilage repair model.

Furthermore, we demonstrated that all BMPs delivered with coacervate promoted cartilage repair via an increase in SOX9 positive cell percentage in the regenerated cartilage. SOX9 is a master transcription factor of chondrogenesis and chondrocytes, via activation of the chondrocyte-specific genes such as COL2 and aggrecan expression [65]. A previous in vitro study demonstrated that SOX9 promoted production of COL2 and aggrecan, while BMP2 and BMP9 treatments increased SOX9 expression [22]. SOX9 expression is maintained even in osteoarthritis chondrocyte progenitor cells [66,67]. Co-delivery of BMP2/SOX-9 genes promoted chondrocyte gene expression of dedifferentiated chondrocytes and cartilage repair in vivo compared to individual BMP2 and SOX-9 gene deliveries [68]. Delivery of the SOX Trio (SOX5,6&9) either with stem cells or alone resulted in cartilage repair with phenotypically stable chondrocytes by increasing COL2A1 expression without increasing COL10A1 and COL1A1 expression in a cartilage defect model or a subcutaneous implantation model [[69], [70], [71]]. The finding that all BMPs significantly increased SOX9 indicated their similar function in promoting cartilage repair when delivered with the coacervate sustained release polymer.

Finally, it is noteworthy that the repaired cartilage in this rat microfracture model by coacervate/BMPs is still inferior to native hyaline cartilage. The overall lower Safranin O staining intensity in the repaired cartilage might be caused by the coacervate-BMPs complex not being entirely maintained in the defect area due to the small defect size (1.5 mm). This is more likely to have happened in the worst repair animal. Other doses of BMPs such as 1μg/knee may also be further investigated to achieve better cartilage repair using the coacervate sustained release platform or in combination with TGFβ3. In addition, we used human BMPs for rat cartilage repair, as not all rat BMP proteins are available at the time of this study. However, using rat BMPs for rat cartilage repair, it might be more effective when they are available. Further, in the worst repaired cartilage of each group, we still see evidence of fibrosis. Previously, we have shown that oral or intra-articular injection losartan administration improved microfracture-mediated cartilage repair by blocking fibrosis [6,7], therefore, it will likely be more effective when combining coacervate-BMPs and losartan in future investigations.

5. Conclusion

In summary, this study demonstrated that human BMPs 2,4,6,7&9 increased the chondrogenic pellet size of human BMMSCs by an average of 71.7 %, 68.7 %, 37.2 %, 31.6 % and 49.6 %, respectively, compared to CTL. BMP2,4,6,7, and 9 also increased COL2 percentages by 139.6 %, 191.04 %, 131.19 %, 176.58 %, and 172.77 %, respectively, compared to CTL group in human BMMSCs chondrogenic pellets. BMPs 2,4&9 are more potent than BMP6 or BMP7. Coacervate-BMPs are biocompatible for both hBMMSCs and rat MDSCs. Sustained release of human BMPs with heparin/PEAD coacervate significantly improved microfracture-mediated cartilage repair, as revealed by gross image macroscopic scoring and Seller's histology scoring, by increasing subchondral bone healing and SOX9 expression. However, the regenerated cartilage is still inferior to native hyaline cartilage. Further improvements are needed, such as testing different doses coacervate BMPs or combining anti-fibrotic factor with coacervate-BMPs and long-term observation in a larger animal model.

CRediT authorship contribution statement

Xueqin Gao: Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Nathaniel Wright: Methodology, Investigation. Matthieu Huard: Methodology, Investigation. Jian Tan: Methodology, Investigation. Joseph Ruzbarsky: Writing – review & editing, Validation. Aiping Lu: Methodology, Investigation. Laura Chubb: Resources, Methodology, Investigation. Rocky Tuan: Writing – review & editing, Resources. Marc J. Philippon: Writing – review & editing, Supervision, Conceptualization. Yadong Wang: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. Johnny Huard: Writing – review & editing, Validation, Supervision, Project administration, Funding acquisition, Conceptualization.

Ethics approval and consent to participate

Animal experiment in this study was approved by the Institutional Animal Care and Use Committee (IACUC) of the Colorado State University (Protocol# 4389).

Declaration of competing interest

J.J.R: received Consultant for Smith + Nephew. MJP has received support from Smith + Nephew Inc-services other than consulting, consulting fee, travel and lodging, food and beverage, royalty or license, Linvatec Corporation-royalty or license, education, Synthes GMBH- faculty/speaker at an accredited/certified education program, food and beverage, Siemens Medical Solutions USA, Inc-travel and lodging, food and beverage. JH received royalties from Cook Myosites.

All other authors have no conflict of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: