Optimization of the MIRB-labeling of integrin α10-MSCs
Intracellular uptake of the MIRB label in the MSCs was confirmed by fluorescence microscopy (Fig. 3A). The labeling frequency of the MSCs was 100% after both 6 hours and 16 hours MIRB-labeling time as demonstrated by flow cytometry, and the median fluorescence intensity (MFI) of Rhodamine B was 11,613 and 20,545 in 6h-MIRB and 16h-MIRB, respectively. After 3 days of proliferation in culture post-labeling, the MFI had decreased to 1,322 (6h-MIRB) and 2,530 (16h-MIRB). At this time point the frequency of labeled MSCs was 27.86% (6h-MIRB) and 59.57% (16h-MIRB). After additional 2 days of proliferation in culture, the MFI of the 6h-MIRB was decreased to 841 with a labeling frequency of 8.4% (Supplementary Fig. 1, Additional File 1).
MSC viability immediately after labeling was high in both 6h-MIRB (99.69%) and 16h-MIRB (97.38%) assessed by 7-AAD (7-Amino-Actinomycin D, Biolegend) staining and flow cytometry analysis. However, total cell count of labeled integrin α10-MSCs was lower than the cell count of the UL-MSCs count, showing a 30% reduction in cell count in 6h-MIRB and 58% reduction in 16h-MIRB (Supplementary Fig. 1, Additional File 1).
Further, labeled integrin α10-MSCs showed reduced proliferation rate compared to UL-MSCs. After 3 days of proliferation in culture post-labeling, the number of cell doublings was 2.66 for 6h-MIRB and 2.11 for 16h-MIRB, compared to 2.81 cell doublings for UL-MSCs. After additional 2 days in culture, the 6h-MIRB showed 3.26 cell doublings compared to 3.5 cell doublings for UL-MSCs (Supplementary Fig. 1, Additional File 1).
MRI detection of MIRB-labeled integrin α10-MSCs in vitro
In agarose phantoms, all cell-concentrations of 6h-MIRB could be visually detected (lowest concentration of cells = 1000 cells/0.2ml) (Fig. 3B); in agarose phantoms containing the same number of 16h-MIRB, the visualization of the MSCs was better compared to 6h-MIRB. At a concentration of 1x105 cells, remarkable T2* shortening was detected with 6h-MIRB compared to UL-MSCs (16,118ms vs. 27,057ms. (Supplementary Fig. 2, Additional File 2).
Taken together, these results confirmed that 6h-MIRB had retained acceptable quality and that they could be detected by MRI. Thus, 6h-MIRB were used in the following in vivo experiment.
In vitro characterization of MIRB-labeled integrin-α10-MSC
Both UL-MSCs and 6h-MIRB showed a > 99% frequency of cells expressing the stem cell surface markers CD73, CD90 and CD105 (Supplementary Fig. 3, Additional File 1). The frequency of MSCs positive for integrin α10 was higher in the 6h-MIRB (81.3%) than the UL-MSCs (58.8%).
The chondrogenic differentiation assay showed that compared to the non-induced control α10-MSCs chondrogenesis was induced in both UL and the 6h-MIRB α10-MSCs, assessed by gene expression of both COL2, aggrecan and integrin α10, and that the expression levels seemed to be higher in the 6h-MIRB compared to UL (Fig. 4A).
Osteogenic differentiation in both the UL and the 6h-MIRB was confirmed by the staining of extracellular calcium deposits with Alizarin Red, while adipogenic potential was confirmed by the observation of cytoplasmic lipid droplets stained with Oil Red O in both the UL and the 6h-MIRB (Fig. 4B). After one freeze-thaw cycle the viability of the 6h-MIRB was 97.0% and the MFI was 8,654 compared to 11,613 before freezing.
In vivo integrin α10-MSC tracking
The rabbits did not show any lameness or change in spontaneous movements after surgery and did not display any signs of pain during the study. Further, lameness or swelling was not observed after the intra-articular injection of the MIRB-labeled integrin α10-MSCs. At the time of euthanasia all cartilage defects were partially healed in both treated and untreated rabbits (Supplementary Fig. 4, Additional File 3).
MRI show homing of MIRB labeled integrin α10-MSC to the osteochondral defect
Subjective visual analysis revealed that the 6h-MIRB labeled MSCs were visible as hypointense “black dots” on the MR images. Immediately after injection they were distributed as single cells or small cell clusters in the synovial fluid, particularly in the anterior part of the knee (Fig. 5), and after 12 hours the MSCs were found lining the synovial membrane and few MSCs were found in the synovial fluid (Fig. 5). At this timepoint the MSCs were also detected in the posterior joint compartments. At 12 hours after injection, labeled integrin MSCs were found in and around the cartilage defect (Fig. 6). The amount of labeled MSCs in the cartilage defects increased up to day 4 and began to decline thereafter. In the synovial membrane, labeled MSCs seemed to decline after 2–4 days, and hypointesity in the synovial membrane was low at day 10 (Table 1). There were no MSCs attaching to healthy cartilage. Some MSCs were detected in the infrapatellar fat pad up to day 2.
Table 1
Visual grading of integrin α10-MSC distribution on magnetic resonance images Distribution of labeled mesenchymal stem cells selected for a high expression of integrin α10β1 (integrin α10-MSCs) after intra-articular injection in the knee of 6 rabbits with a surgically created cartilage defect. The distribution is determined by visual grading of magnetic resonance images. Distribution and cell concentration was graded subjectively as 0 = none; + = mild; ++ = moderate; +++ = marked; grey box = rabbit not scanned at that timepoint. Rabbit #3 was removed from the study because of unintentional peri-articular injection of labeled α10-MSCs.
| 0 hours | 12 hours | 24 hours | 2 days | 4 days | 7 days | 10 days |
Rabbit 1 | Cartilage Defect | | +++ | | | ++ | + | |
| Surrounding Cartilage | | 0 | | | 0 | 0 | |
| Synovial Membrane | | +++ | | | + | + | |
| Infrapatellar Fat Pad | | + | | | 0 | 0 | |
Rabbit 2 | Cartilage Defect | | | + | | + | | 0 |
| Surrounding Cartilage | | | 0 | | 0 | | 0 |
| Synovial Membrane | | | ++ | | + | | + |
| Infrapatellar Fat Pad | | | 0 | | 0 | | 0 |
Rabbit 4 | Cartilage Defect | 0 | | | + | + | | 0 |
| Surrounding Cartilage | 0 | | | 0 | 0 | | 0 |
| Synovial Membrane | ++ | | | ++ | + | | 0 |
| Infrapatellar Fat Pad | 0 | | | 0 | 0 | | 0 |
Rabbit 5 | Cartilage Defect | + | | + | | + | | |
| Surrounding Cartilage | + | | 0 | | 0 | | |
| Synovial Membrane | ++ | | + | | + | | |
| Infrapatellar Fat Pad | + | | 0 | | 0 | | |
Rabbit 6 | Cartilage Defect | | ++ | | ++ | + | + | |
| Surrounding Cartilage | | 0 | | 0 | 0 | 0 | |
| Synovial Membrane | | ++ | | + | + | + | |
| Infrapatellar Fat Pad | | + | | + | 0 | 0 | |
The objective SI measurements confirmed the visual analysis results, demonstrating a drop in SI in the cartilage defect from 12 hours after injection, indicating accumulation of MIRB-labeled MSCs in the defect with a peak on day 2 to 4 (p = 0.036 relative to baseline) after injection. On day 4 the SI was reduced up to 80% compared to baseline, and on day 10 SI had returned to baseline (Fig. 7).
MIRB-labeled integrin α10-MSCs co-localize with aggrecan and collagen type II in the regenerated cartilage tissue
Fluorescence microscopy of sections from the cartilage defect showed MIRB labeled integrin α10-MSCs in all layers of the cartilage repair tissue in all treated rabbits (Fig. 8C), while no labeled MSCs were detected in normal undamaged cartilage (Fig. 8A). No MIRB signal was detected in the untreated rabbits (Fig. 8B). The number of labeled MSCs varied between the treated rabbits (Supplementary Fig. 5, Additional File 3), however, the degree of MIRB labeled MSCs corresponded with the degree of MIRB signal on MRI images (r = 0.94; p = 0.0167) at time of euthanasia (Supplementary Table 1, Additional File 3).
Immunofluorescence analysis of the cartilage specific matrix components aggrecan and COL2 showed co-localization with MIRB labeled MSCs in the treated rabbits (Fig. 8D-E) suggesting differentiation of the MSCs to chondrocyte-like cells. Immunodetection of integrin α10 showed that both the MIRB labeled MSCs and the resident chondrocytes expressed integrin α10 (Fig. 8F).