Mesenchymal Stem Cell Alterations in Bone Marrow Lesions in Patients With Hip Osteoarthritis

Abstract

Objective: In patients with osteoarthritis (OA), bone marrow lesions (BMLs) are intimately linked to disease progression. We hypothesized that aberrant multipotential stromal cell (also known as mesenchymal stem cell [MSC]) responses within bone tissue contributes to BML pathophysiology. The aim of this study was to investigate BML and non-BML native subchondral bone MSCs for numeric, topographic, in vitro functional, and gene expression differences.

Methods: Ex vivo 3T magnetic resonance imaging (MRI) of the femoral heads of 20 patients with hip OA was performed. MRI-determined BML and non-BML regions were excised and enzymatically treated to extract cells and quantify MSCs using flow cytometry and colony-forming unit-fibroblast (CFU-F) assay. Immunohistochemical analysis was performed to determine in vivo CD271+ MSC distribution. Culture-expanded CD271+ cells were analyzed for tripotentiality and gene expression.

Results: BML regions were associated with greater trabecular bone area and cartilage damage compared with non-BML regions. The proportion of CD45-CD271+ MSCs was higher in BML regions compared with non-BML regions (median difference 5.6-fold; P < 0.001); the CFU-F assay showed a similar trend (median difference 4.3-fold; P = 0.013). Immunohistochemistry revealed CD271+ cell accumulation in bone adjacent to cartilage defects and areas of osteochondral angiogenesis. BML MSCs had lower proliferation and mineralization capacities in vitro and altered expression of TNFSF11/RANKL and CXCR4/stromal cell-derived factor 1 receptor. OA MSCs showed up-regulated transcripts for CXCR1 and CCR6 compared with MSCs derived from healthy or osteoporotic bone.

Conclusion: This study is the first to show numeric and topographic alterations in native MSCs in the diseased bone of patients with hip OA. Given the associated functional perturbation of MSCs, these data suggest that subchondral bone MSC manipulation may be an OA treatment target.

Figure 1

Segregation of bone marrow lesion (BML) and non‐BML regions and downstream processing. A, Femoral head in plastic bracket. B, Femoral head/bracket in a phosphate buffered saline–containing polypropylene jar. C, Proton density–weighted sequence magnetic resonance images in 3 planes, with measurements for cuts (top and bottom left) and corresponding T1‐weighted sequence image obtained in the “coronal” plane (bottom right). D, Bone segregation apparatus. E, BML bone fragments used for collagenase treatment after mincing with a rongeur.

Figure 2

Histologic appearance of bone and cartilage in bone marrow lesion (BML) and non‐BML specimens. A and B, Gross histologic appearance of an excised BML specimen with a subchondral cyst (arrow) (A) and corresponding non‐BML specimen from the same femoral head (B). C, Comparison of trabecular bone area (BML versus non‐BML) as a percentage of the total area (n = 14 pairs). ∗ = P = 0.001. D and E, Excised BML specimen showing some cartilage abnormalities above the articular end plate (D) and non‐BML fragment from the same femoral head showing intact cartilage above the articular end plate (E). F and G, Photomicrographs of Safranin O–stained paired BML (F) and non‐BML (G) specimens after decalcification in EDTA, showing a greater trabecular area in the BML region. Bars = 3.8 mm (A and B), 500 μM (D and E), and 600 μM (F and G).

Figure 3

Mesenchymal stromal cell (MSC) enumeration in fractions of cells released from bone marrow lesion (BML) and non‐BML regions, following collagenase treatment. A, Representative flow cytometry plots for CD45−CD271+ MSC populations showing a rectangular selection for MSC enumeration. B, Paired‐sample line graphs showing CD45−CD271+ MSCs as a percentage of total live cells (left) and CD90+ cells as a percentage of CD45−CD271+ cells (right) in BML versus non‐BML cell fractions (n = 20 each). C, Left, Dot plots from a BML sample with gating on the CD45+ lymphocyte population. Right, Paired‐sample line graphs showing lymphocytes as a percentage of total live cells in BML versus non‐BML cell fractions (n = 20 each). D, Colony‐forming unit–fibroblast (CFU‐F) assay. Left, Representative 25‐cm² flasks. Right, Paired‐sample line graph showing the number of CFU‐Fs per 10⁶ plated cells in BML and non‐BML regions (n = 14 each). In all line graphs, bars show the median. ∗ = P < 0.05; ∗∗∗ = P < 0.001.

Figure 4

CD271 cell distribution in light microscopy images of bone marrow lesion (BML) and non‐BML specimens as assessed by immunohistochemistry. A and B, Perivascular distribution in representative negative control (A) and CD271‐stained (B) non‐BML specimens. C, Representative non‐BML specimen showing CD271 staining in bone lining. D, CD271+ staining near osteochondral junction and surrounding a bone cyst (BC) in a representative BML sample. E, Higher‐magnification view of boxed area in D, showing CD271+ staining within the subarticular end plate immediately beneath a chondral lesion (arrows). F, High‐magnification view of BML sample, showing subchondral CD271+ perivascular staining (arrowheads) and CD271+ staining within subarticular end plate immediately beneath chondral lesion (arrow). G, Non‐BML sample, showing relatively intact cartilage and lacking CD271 expression. Bars = 100 μm (A–C), 3 mm (D), 400 μm (E and F), and 200 μm (G).

Figure 5

Surface phenotype, growth rates, and differentiation capacities of CD271 cell–derived cultures from bone marrow lesions (BMLs) and non‐BMLs. A, Flow cytometry histograms showing positive (CD73, CD90, CD105) and negative (CD45) marker expression in BMLs and non‐BMLs from a representative donor. Dashed lines indicate isotype controls. B, Culture growth rates at different passages (for passages 0–2 [P0–P2], n = 5; for P3, n = 2). Bars show the mean ± SD. C, Adipogenesis in paired BML and non‐BML adipogenic cultures. Left, Percentage of oil red O–positive area versus total cell area in BML and non‐BML cultures (n = 5 each). Right, Representative photomicrographs of paired BML and non‐BML cultures at 40× magnification and 200× magnification (left and right images, respectively). D, Osteogenesis in paired BML and non‐BML osteogenic cultures. Left, Calcium production in BML and non‐BML osteogenic cultures (n = 5 each). Right, Representative alkaline phosphatase (purple) and alizarin red (red) staining of BML and non‐BML osteogenic cultures. E, Chondrogenesis in paired BML and non‐BML chondrogenic cultures. Left, Glycosaminoglycan (GAG) production in BML and non‐BML chondrogenic cultures (n = 5 each). Right, Gross images of wet chondrogenic pellets (bars represent 1‐mm spacing) and light microscopy images of toluidine blue–stained cartilage pellets at 40× magnification. Data in C–E are shown as box plots, representing the 25th, 50th, and 75th percentiles. ∗ = P < 0.05. PD = population doubling.

Figure 6

Relative gene expression in CD271+ cell–derived mesenchymal stem cell cultures from bone marrow lesions (BMLs) and non‐BMLs in the bone of patients with osteoarthritis (OA), healthy controls (HCs), and patients with osteoporosis (OP). A, Mean fold changes in relative gene expression in BML versus non‐BML cultures; n = 7 paired samples. The $2^{-\Delta {\mathrm{C}}_{\mathrm{t}}}$ value was normalized to the value of HPRT. Black bars indicate P < 0.05 by Wilcoxon's test for paired data. B, Top, Validation of CXCR4 differential expression as determined by TaqMan quantitative polymerase chain reaction (qPCR) analysis. Bottom, Validation of RANKL surface protein expression as determined by flow cytometry. C, Validation of genes putatively specific for OA mesenchymal stem cells (MSCs) using additional OA (n = 7), HC (n = 9), and OP (n = 5) MSC cultures. D, Expression of CXCR4 and TNFSF11 (encoding RANKL) in additional OA, HC, and OP MSC cultures. Gene expression that is significantly different in OA MSCs compared with both HC MSCs and OP MSCs is indicative of an OA phenotype. The y‐axis indicates expression relative to HPRT. Kruskal‐Wallis grouped comparison P values are shown in graphs, with paired significance indicated by P < 0.05 (*) and P < 0.01 (**). MFI = mean fluorescence intensity; NS = not significant.