Obesity hinders the efficacy of adipose-derived stem cells for knee osteoarthritis by reducing the proportion of DPP4+ stem cells

Abstract

Osteoarthritis (OA) is the most prevalent and disabling joint disease, while adipose-derived stem cells (ASCs) have emerged as a promising therapeutic option in pre-clinical studies. However, the therapeutic efficacy of ASCs may be influenced by the source of these cells, especially in obese patients. This study compared the effects of intra-articular injections of ASCs from wild-type (WT) and ob/ob (OB) mice. Behavioral and histological analyses demonstrated that WT-ASCs significantly alleviated OA symptoms, restoring paw withdrawal thresholds and improving gait parameters while reducing cartilage degradation. In contrast, OB-ASCs only partially improved gait and did not significantly affect cartilage degeneration. Single-cell RNA sequencing of stromal vascular fractions from subcutaneous adipose tissue revealed distinct ASC subpopulations, with DPP4+ cells being notably reduced in obese mice. In vitro, OB-ASCs and high-fat-diet (HFD)-ASCs exhibited impaired proliferation and chondrogenesis but HFD-ASCs retained anti-inflammatory properties. Further investigation using fluorescence-activated cell sorting (FACS) isolated DPP4+ and DPP4− ASCs from WT mice, demonstrating that DPP4+cells had superior chondrogenic potential and reduced OA pain more effectively than DPP4− cells. These findings suggest that obesity impairs the therapeutic potential of ASCs in OA, primarily due to reduced proliferation and chondrogenesis, and highlight DPP4+ ASCs as a promising candidate for cell therapy in OA.

Graphical Abstract

Graphical Abstract.

Significance statement.

In lean iWAT, ASCs possess potential application in osteoarthritis regeneration due to chondrocyte differential ability and immune regulation. Moderate obesity decreases the ability of chondrogensis, while retaining the anti-inflammatory with an unchanged ratio of DPP4+ ASCs. Severe obesity impairs chondrogenesis potential, and increases the expression of pro-inflammatory genes with a low proportion of DPP4+ cells. Therefore, ASCs with a low ratio of DPP4+ ASCs cannot be used for osteoarthritis treatment.

Introduction

Osteoarthritis (OA) is a prevalent musculoskeletal disorder that primarily afflicts the elderly, characterized by cartilage degeneration, bone remodeling, synovial hypertrophy, osteophyte formation, and ligament dysfunction.¹ The global burden of OA is substantial, affecting approximately 240-250 million individuals and ranking among the leading causes of pain and disability.² Alarmingly, OA onset is occurring at progressively younger ages, with around 30% prevalence among individuals aged over 45, posing a significant societal challenge.³ Current standard treatments for OA, which include medication and surgery, mainly target symptom management and inflammation reduction but do not effectively halt the disease’s progression.⁴ Hence, there is an urgent need to explore novel approaches for OA prevention and treatment.

Mesenchymal stem cell (MSC) therapy has emerged as a promising avenue in regenerative medicine for OA.^5,6 MSCs, which can be found in various tissues such as bone marrow (BM), adipose tissue, synovium, and umbilical cord blood, possess the remarkable ability to self-renew and differentiate into diverse cell lineages, including chondrocytes, osteoblasts, and adipocytes.⁷ These cells exhibit essential properties, including proliferation, multipotency, and immunosuppressive functions, which are crucial for cartilage regeneration in OA. Furthermore, MSCs have demonstrated immunomodulatory and anti-inflammatory effects by augmenting regulatory T-cell function, promoting M2 macrophage polarization, and inhibiting effector lymphoid cell activity. MSCs also possess paracrine anti-inflammatory properties, as evidenced by their ability to down-regulate inflammatory factors like IL-1β, IL-6, and IL-8 when cocultured with chondrocytes. Inflammatory environments have been shown to activate autophagy and inhibit catabolism in chondrocytes, highlighting MSCs’ protective effects.

Single-cell RNA sequencing revealed that adipose-derived stem cells (ASCs) have lower transcriptomic heterogeneity and higher immunosuppressive potential compared to BM-MSCs.⁸ A meta-analysis of clinical trials indicated that ASCs are more efficient and safer for OA treatment than BM-MSCs.⁹ Additionally, ASCs have garnered attention as a promising source for OA treatment due to their accessibility and robust immunosuppressive properties.¹⁰ ASC therapy can be categorized as autologous or allogeneic, with the latter offering advantages such as non-invasive cell collection process and quicker cell expansion.¹¹ Numerous studies have substantiated the effectiveness of ASC injections in mitigating cartilage degradation and improving clinical and radiographic OA symptoms, such as pain reduction, prevention of joint space narrowing and osteophyte formation.^12,13 Nevertheless, the heterogeneity of ASCs, influenced by donor tissue source, age, and metabolic status, may impact their therapeutic potential.^14,15 Notably, subcutaneous ASCs have demonstrated superior anti-inflammatory and chondrogenic potential compared to visceral ASCs, making them a suitable cell source for OA treatment.¹⁶ Age and obesity are significant risk factors for OA and can adversely affect the quality of ASCs.¹⁷ Clinical data indicates that favorable outcomes are more likely for patients below the age of 60 and with a BMI below 27.5 kg/m².¹⁸

Obesity is a prevalent health concern, projected to affect nearly 20% of the global population by 2030, and is closely linked to the effectiveness of ASCs.¹⁹ ASCs are heterogeneous, comprising distinct subpopulations with varying biological functions and therapeutic effects.^20,21 Single-cell RNA sequence analyses have identified that cells marker by dipeptidyl peptidase-4(DPP4)/CD26 expression are highly proliferative, multipotent progenitors.²² Obesity further impacts this heterogeneity, with stromal vascular fraction (SVF) from nonobese individuals containing a higher number of CD90+ cells compared to SVF from obese patients.²³ Studies have shown that ASCs from obese individuals exhibit increased adipogenic and osteogenic differentiation but decreased chondrogenic capabilities.²⁴ Moreover, obesity impairs the differentiation potential and proangiogenic capacities of ASCs.²³ However, the precise influence of obesity on the therapeutic potential of ASCs in OA remains to be fully elucidated.

This study demonstrates that ASCs obtained from obese ob/ob mice exhibit suboptimal outcomes in terms of cartilage repair and pain relief. Furthermore, OB-ASCs display reduced proliferation, anti-inflammatory properties, and chondroprotective characteristics compared to ASCs from wild-type mice (WT-ASCs). We identify DPP4 as a distinguishing factor between WT and ob/ob mouse subcutaneous ASCs. This research also reveals that high-fat diet-induced obesity impairs proliferation and chondrogenesis. Importantly, our findings validate the therapeutic potential of DPP4+ cells in alleviating OA-related pain. Collectively, this research sheds light on the significant role of obesity in shaping the efficacy and characteristics of ASCs and highlights the positive contribution of DPP4+ cells to ASC-based treatments for osteoarthritis.

Materials and methods

Study design

The objective of this study was to compare the efficacy of adipose-derived stem cells (ASCs) from healthy and obese mice in the context of osteoarthritis. ASCs were obtained through collagenase digestion of subcutaneous adipose tissue from mice. An osteoarthritis surgical model, the DMM model, was executed in 8-week-old C57BL/6 mice. The mice were randomly assigned to four groups: Control group, PBS group, WT-ASCs treatment group, and OB-ASCs treatment group within cages. For stem cell treatment, WT-ASCs or OB-ASCs (5 × 10⁵ cells/mice) were directly injected into the intra-articular space of the knee joints of recipient mice 4 weeks after surgery. The investigators performing intra-articular injections and measuring behaviors were blinded to the treatment. Behavioral experiments were conducted 8 weeks post-operation, and the knee joints of the mice were collected. Cartilage degeneration was assessed through Safranine O-fast green and hematoxylin and eosin (H&E) staining. Pain was evaluated using the CatWalk gait analysis system and Von Frey test. Osteoarthritis severity scoring was performed by two independent assessors. The proliferation ability of ASCs was tested using CCK8 kits. ASCs were treated with chondrogenic differentiation medium, and chondrogenesis potential was assessed through Western blot and qPCR. RAW264.7 cells were treated with conditioned medium from different adipose-derived stem cells, and the expression of inflammatory factors was measured by qPCR. Cell culture experiments were performed in triplicates and repeated at least three times. To investigate whether the heterogeneity of ASCs affects the efficacy of adipose-derived stem cells, single-cell sequencing on WT-SVF and OB-SVF was performed.

Animal

The 8-week-old C57BL/6 and leptin knockout ob/ob mice were procured from Gem Pharmatech. The mice were housed in transparent cages with temperature control set at 23°C, following a 12-hour light-dark cycle. They had unrestricted access to food and water. All animal experiments were conducted in accordance with the ethical guidelines and approved by the Fudan University Shanghai Medical College.

To induce diet-induced obesity (DIO), 8-week-old C57BL/6J mice were provided ad libitum access to a high-fat diet (60% kcal% fat; Cat. D12492, Research Diet, New Brunswick) for a duration of 12 weeks.

Osteoarthritis mice model

In the beginning, we performed DMM (Supplementary Figure 1A) and CIOA (collagenase-induced osteoarthritis) (Supplementary Figure 1B) models to observe the progress of osteoarthritis. The results found that at 4 weeks after DMM, the OARSI score was 3, and at 8 weeks after surgery, the OARSI score was 5. However, cartilage degeneration with OARSI score of 3 was observed at 2 weeks with CIOA and gradually increased to 5 points at 4 weeks. Taken together, the progress of CIOA was faster and appeared to be less stable than the DMM model. In addition, we also used ASC treatment in DMM and CIOA models and found the efficiency in both models (Supplementary Figure 1C). Due to the consistency with normal human OA and the effect being stable, so we chose DMM for the remaining experiments.

Destabilization of the medial meniscus

To induce standardized joint injury in knees, 8-week-old male C57BL/6 mice (n = 24) were anesthetized with pentobarbitone and underwent medial para-patellar arthrotomy under a surgical microscope. The joint was opened by separating the medial margin of the quadriceps from the muscles of the medial compartment and laterally dislocating the patella. The fat pad over the cranial horn of the medial meniscus was excised and the medial meniscotibial ligament was cut to destabilize the medial meniscus. The wound was sutured with 6-0 stitches. Control group underwent sham surgery (N = 5). Four weeks after the DMM surgery, mice received intra-articular injection (Dpp4+ ASCs, Dpp4− ASCs, PBS, separately). Control animals received an equal volume of sterile saline.

Parameters measured in CatWalk gait analysis system

It has been reported that animals with osteoarthritis (OA) tend to minimize contact with the floor and exert less pressure on the painful limb during walking, resulting in decreased weightbearing on the osteoarthritic limb. The Cat Walk test was employed to measure weightbearing on the two hind paws, utilizing signal intensity. Several parameters were assessed: Stance phase is the duration in seconds of contact of a paw with the glass plate. Print area is the surface area of claws in contact with glass. Swing Speed is the speed (distance unit/second) of the paw during Swing. The formula of Swing Speed is: Swing Speed = Stride Length /Swing. Duty Cycle (%) is percentage of standing phase in cycle. The formula of Swing Speed is: Duty Cycle = Stand/(Stand + Swing) × 100%. These parameters provide valuable insights into the weightbearing and locomotor characteristics of animals with osteoarthritis, aiding in the assessment of the impact of the condition on their gait and mobility.

Von Frey test

The mechanical allodynia test was conducted using a calibrated set of von Frey filaments. Mice were placed in a transparent plastic box with a stainless steel wire grid. The mice were acclimated for 30 minutes, and the 50% force withdrawal threshold was determined using the classical up-down iterative method [2]. The procedure involved the following steps: At first, 0.2g von Freys was used to stimulate the paw center of the mice, and then the licking or lifting behavior was observed 5 seconds later, with an interval of 5 seconds between each stimulation. If positive, select a weaker filaments for stimulation, if negative, select a stronger filaments for stimulation, and stop testing four times after changes. This method allowed for the determination of the force threshold at which the mice exhibited a response, providing valuable information about mechanical allodynia.

ASCs administration

The mice were anesthetized with isoflurane, and iodophor was disinfected. The right knee joint of mice was flexed to fully expose the joint space. WT-ASCs/OB-ASCs were injected into the articular cavity with insulin microsyringe.

ADSC isolation, culture, and differentiation

ADSC was isolated from adult male mice. The mice were anesthetized and the subcutaneous adipose tissue was taken and rinsed in PBS 3 times. The adipose tissue was cut into small pieces and digested at 37°C for 40 minutes with 0.075% Type Ⅷ collagenase. The cell suspension was filtered through a 100 µm filter and centrifuged at 600g for 5 minutes. After red blood cell lysis with 1× lysis buffer, cells were centrifuged and resuspended in a growth medium (DMEM/F12 1:1 containing 1% penicillin and streptomycin and 10% FBS). On day 3, the cells were trypsinized and expanded until day 5. Cell counting, proliferation, and differentiation testing were performed on P2.

For adipocyte differentiation, the cells were allowed to grow to confluence and then were maintained at confluence for 2 days without changing the medium prior to treatment with a differentiation cocktail (5 g/mL insulin, 1 M dexamethasone, 0.5 mM IBMX [a phosphodiesterase inhibitor], 1 M rosiglitazone for 48 hours, then 5 g/mL insulin and 1 M rosiglitazone for 48 hours). After exposure to the differentiation cocktail for 4 days, the cells were maintained in F12 medium with 10% FBS until harvesting on day 8. For staining, the cells were fixed in 4% formaldehyde in PBS for 15 min., rinsed with water, stained with Oil Red O for 1 hour, and then rinsed with water again.

For chondrocyte differentiation, ASCs were centrifuged into cell aggregates and induced along the chondrogenic lineage. Cells were suspended to a concentration of 1 * 10⁷ cells/mL, seeding 5 µL droplets of cell solution in the center of the 6-well plates. After cultivating the micro mass cultures for 2 hours under high humidity conditions, warmed chondrogenesis media were added to the culture vessels and incubated at 37°C with 5% CO₂. Then, the supernatant was aspirated and replaced with chondrogenic inductive medium (CIM) consisting of F12/DMEM (containing L-glutamine and sodium pyruvate), 1% Ab/Am cocktail, 1% ITS + Premix (Life Technologies), 40 µg/mL proline, 100 nM dexamethasone, 50 lg/mL ascorbic acid-2-phosphate (Sigma-Aldrich) and 10 ng/mL rhTGF-b1(Peprotech). Media were changed every other day. After 21 days, micromasses were harvested for Western, qPCR, and Alcian blue staining. For staining, the cells were fixed in 4% formaldehyde in PBS for 15 minutes, rinsed with water, stained with 1% Alcian Blue (Sigma-Aldrich) for 2 hours, and then rinsed with water again.

For osteogenic differentiation, cells were cultured in DMEM with 10% FBS, 50 µM ascorbic acid, and 10 mM β-glycerophosphate for 20 days. For staining, the cells were fixed in 4% formaldehyde in PBS for 15 minutes, rinsed with water, stained with 2% Alizarin Red (Sigma-Aldrich) for 1 hour, and then rinsed with water again.

Cell proliferation assay

Cells were seeded in 96-well plates at 10⁴ cells/well. CCK-8 (Beyotime) assays were performed according to the manufacturer’s instructions. Briefly, cells were seeded at different time points, then add 10 µL of CCK-8 buffer to each well and incubation at 37°for another 1 hour. Then the absorbance at 450 nm was measured using a microplate reader (Thermo Fisher Scientific).

Cartilage harvest

Mice were briefly anesthetized, and mouse knee joints were exposed. Muscle and fascial tissue surrounding the knee joint was removed. The joint capsule was opened to expose the distal femur and tibial plateau. Cartilage tissue was scraped.

Sc-RNA-seq with 10× genomics chromium platform

SVCs were loaded onto a GemCode instrument (10× Genomics) to generate single-cell barcoded droplets according to the manufacturer’s protocol with the 10× single-cell 3′ v2 chemistry. Sequencing was performed with Illumina HiSeq 4000 according to the manufacturer’s instructions (Illumina). The resulting reads were aligned, and gene-level unique molecular identifier (UMI) counts were obtained by using Cell Ranger (Pipeline).

Sc-RNA-seq data processing and analysis

Raw reads were demultiplexed and mapped to the mm10 reference genome by 10× Genomics Cell Ranger pipeline (v6.1.2) using default parameters in Linux. Downstream single-cell analyses were performed using Seurat package(v4.3.0.1) on Rstudio. Cells were first filtered to have >200 detected genes (a gene with expression in more than 3 cells was considered as expressed) and less than 5% of total UMIs mapping to the mitochondrial genome. Then, each individual dataset was log-normalized, scaled, and corrected for dataset-specific batch effects using Harmony(v1.2.0). A union set of top 3000 highly variable genes from each dataset expressed in all samples was used by Seurat multicanonical correlation analysis to further correct for dataset-specific batch effects.³² The informative number of correlated components was examined by the elbow plot. The top 30 aligned correlated components were used as input for uMAP dimension reduction and subsequent clustering analysis. Clusters were identified using Seurat FindClusters function using default parameters plus a resolution of 0.1. Then, Seurat FindAllmarkers function with the Wilcox test was used to identify conserved markers that were specifically highly expressed in each cluster versus all other clusters (logFC ≥ 0.6 plus adjust P-value < .05).

Violin plots, heatmaps, and individual uMAP plots for the given genes were generated by using the Seurat toolkit VlnPlot, DoHeatmap, and FeaturePlot functions, respectively.

Western blot analysis

Western blot analyses were performed as previously described^[3]. Primary antibodies against the following proteins were used: ACTIN(Santa Cruz, 1:1000), SOX9(Cell Signaling Technology, 1:1000), AGGRECAN(Proteintech, 1:1000), COLLAGEN Ⅱ(Abcam, 1:1000).

Total RNA extraction and real-time PCR

Total RNA was extracted from cultured cells or tissue samples using TRIzol (Invitrogen) and stored in diethyl pyrocarbonate (DEPC) H₂O at −20°C. cDNA was synthesized from total RNA using RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific), and 2 μg cDNA was amplified by qPCR. Next, qPCR was performed to analyze the levels of multiple mRNA using the primers shown in table. Complementary DNA synthesized from total RNA was analyzed in a Sequence Detector (Q5; Bio-Rad, Hercules) with specific primers and SYBR Green PCR Master reagents (ABI). Relative mRNA levels were calculated using the comparative cycle threshold method and normalized to 18S ribosomal RNA mRNA. The relative amount of each mRNA was expressed as fold change relative to the control group.

Flow cytometry analysis of the SVF of iWAT.

Stromal vascular fractions were obtained by collagenase digestion and filtration. Red blood cells were removed by red blood cell lysate. The cells were washed with PBS and added 10% sheep serum to seal them for 30 minutes. The antibody diluted with 5% sheep serum was added and incubated at 4°C for 40 minutes under dark conditions. The stained cells were resuspended by PBS in 2% sheep serum and analyzed by BD FACS Verse flow cytometry. FlowJo V. 10.7.0 was used to analyze the data. The following commercial antibodies were used: CD31-PE-CY7 (BD Pharmingen,102427,1:1000),CD45-FITC (biolegend,103107,1:1000), DPP4-CY7 (biolegend,137809,1:200).

Culture of RAW264.7

Mouse mononuclear macrophage RAW264.7 was purchased from ATCC. RAW264.7 cells were cultured in DMEM complete medium containing 10% fetal bovine serum. When the cell density reached 70%-80%, a complete medium containing 1 µg/ mL LPS was added for M1 polarization.

CCK8 assay

Adipose derived stem cells were obtained by collagenase digestion. Cell suspensions were obtained by pancreatin digestion of passage 2 cells. Ninety-six well plates were seeded with 3000 cells per well and counted as 0 hour after 4 hours of seeding. Complete medium containing 10% CCK-8 was added at 0 hour, 24 hours, 48 hours, 72 hours, 96 hours, and incubated for 2 hours. The absorbance of each well at 450 nm was measured with a microplate reader.

Histology

The knee joint was placed in 4% paraformaldehyde for 24 hours at room temperature. Knee joints were decalcified, paraffin-embedded, and sectioned. The sections (5 mm) were dewaxed in xylene and then washed with a series of graded alcohols. The sections were stained with safranin o fast green and HE. In order to determine the degree of cartilage degeneration, the histological scoring method published by the International Society for the Study of Osteoarthritis (OARSI) was used for analysis.

Statistical analysis

All data are expressed as mean ± SD., as indicated in the figure legends. N-values indicated in the figures refer to biological replicates. The means of the two groups were compared using a two-tailed Student’s test. Means of multiple groups were compared using one-way ANOVA followed by Tukey’s post hoc test. Statistical analyses were completed with Prism GraphPad. P < 0.05 was considered statistically significant.

Results

Transplantation of ASCs from ob/ob mice cannot attenuate OA phenotype

To assess the impact of obesity on ASCs in osteoarthritis (OA) treatment, we compared the efficacy of intra-articular injection of WT-ASCs and OB-ASCs in mitigating cartilage damage. An OA model was induced by destabilizing the medial meniscus (DMM) in 8-week-old male C57BL/6 mice, followed by behavioral evaluation and histological analyses conducted at 8 weeks post-DMM surgery (Figure 1A). Intra-articular injection of WT-ASCs in OA mice starting 4 weeks after DMM surgery, Von Frey tests were performed to assess OA pain sensitivity. The PBS-treated group exhibited a significant decrease in paw withdrawal thresholds (PWTs), which were restored in WT-ASCs-treated DMM mice. Conversely, OB-ASCs treatment partially reversed the reduced PWTs compared to the PBS treatment group (Figure 1B). Gait analysis using the CatWalk system showed that in PBS-treated OA mice, various parameters including left hind/right hind (RH/LH) stance phase, RH/LH print area, swing speed, and duty cycle were significantly decreased. All of these parameters were restored in WT-ASCs-treated DMM mice. However, the OB-ASCs-treated group exhibited no statistically significant differences in RH/LH stance phase and swing speed compared to the PBS-treated group 8 weeks after DMM surgery. Notably, RH/LH print area and duty cycle were notably increased in the OB-ASCs-treated group compared to the PBS-treated group, indicating partial improvement in abnormal gait (Figure 1C).

Figure 1.

OB-ASCs limited OA repairing in the mouse DMM model. (A) Experimental flow chart. (B) Mechanical paw withdrawal threshold were evaluated using Von Frey filaments (n = 6-7). (C) Variations in the ipsilateral and contralateral hind limbs of gait parameters, including Stance Phase, Print Area, Swing Speed, and Duty cycle obtained from CatWalk gait analysis system (n = 7). (D) Representative Safranin O- and HE staining images of osteoarthritic knee joints were collected 8 weeks after DMM surgery. (E) The severity of OA-like phenotype was analyzed using the Osteoarthritis Research Society International (OARSI) score system (n = 5). (F)Western blot analysis of Collagen Ⅱ, Sox9, Aggrecan of cartilage tissue. (G) Gene expression of MMP3, MMP13, and ADAMTS-4 from isolated cartilage tissue (n = 3). Data are expressed as means ± SEM, *P < .05, **P < .01,***P < .001,****P < .0001.

WT-ASCs significantly reversed cartilage degradation, as evidenced by increased safranin o staining and a smoother surface, and lower OARSI score compared to the PBS-treated group. Conversely, there was no significant difference in cartilage degeneration between the PBS-treated and OB-ASCs-treated groups. Overall, these results suggest that WT-ASCs delayed OA progression in the mouse DMM model, demonstrating a more effective outcome compared to OB-ASCs (Figure 1D-E). To explore the impact of intra-articular injection of WT-ASCs and OB-ASCs on cartilage anabolism in vivo, articular cartilage was isolated from control and treatment groups, the data indicated a decrease in the expressions of cartilage anabolism proteins (Aggrecan, Sox9, and Collagen II) in the PBS-treated group, which were restored in the WT-ASCs-treated group. In contrast, the expressions of these proteins were not statistically different between the PBS-treated and OB-ASCs-treated groups (Figure 1F). Since matrix metalloproteinases (MMP)-3, -13 and a disintegrin and metallo proteinases (ADAMTS)-4 are major catabolic factors implicated in OA, q-PCR analysis showed that WT-ASCs significantly decreased the expression of MMP-3, MMP-13, and ADAMTS-4, while OB-ASCs increased the expression of these genes (Figure 1G). In summary, WT-ASCs attenuated OA pain more effectively than OB-ASCs and demonstrated a better outcome in the DMM model.

Single-cell sequencing reveals decrease in DPP4 + stem cell subset in the inguinal white adipose tissue (iWAT) of OB mouse

The efficiency of ASCs therapy is attributed to the heterogeneity of ASCs, promoting us to perform a single cell sequencing on stromal vascular fraction of subcutaneous adipose tissue from 1-month-old WT mice and ob/ob mice. Single-cell suspensions were profiled using 10x Genomics Chromium droplet scRNA-seq. The resulting quality-controlled atlas of WAT-SVF included 17 265 and 18 557 cells obtained from WT and OB mice, respectively. These cells were clustered based on differential expression of marker genes and visualized using a uniform manifold approximation and projection (UMAP) plot. Clustering analysis revealed 11 distinct clusters: adipocyte stem cells (ASC), B cells, T cells, natural killer (NK) cells, myeloid cells, endothelial cells, and others (Figure 2A and Supplementary Figure 2A). Since we used P2 cultured ASCs, from which the immune cells has been dropped, we focused on the stem cells. Adipocyte stem cells were divided into 2 groups: ASC-S1 and ASC-S2. ASC-S1 cells were marked by expression of DPP4, Pi16, CD34, and Ly6a1, which are highly expressed in early multipotent stem cells. ASC-S2 cells expressed Pdgfra, Pdgfrb, Col4a1, and ICAM1, suggesting that these cells could represent “committed preadipocytes” (Figure 2B-C, Supplementary Figure 2B). Moreover, gene set enrichment analysis of the common upstream regulators suggested significant pathways related to cartilage development and collagen-containing extracellular matrix in ASC-S1 cluster (Supplementary Figure 2C).

Figure 2.

Single-cell sequencing reveals a diverse heterogeneity of WT-ASCs and OB-ASCs. (A) Unsupervised clustering of 35 822 cells (mean number of genes per cell = 1544) from SVF of subcutaneous adipose tissue from male WT (17 265 cells) and OB (18 557 cells) mice. (B) Heat map of expression of top 19 population markers; n (ASC-S1) = 5714 cells; n (ASC-S2) = 8287 cells. (C) Gene expression distribution of genes associated with adipocyte progenitor. (D) Fraction of cells per cluster to SVF from WT and OB mice. (E) FACS analysis and quantification of cell populations from WT and OB mice; populations were selected first, followed by Lin- (CD45−CD31−), and then stained with CD34 and Sca1 (n = 5). (F) Gene expression of adipocyte progenitor cell markers in WT-SVF and OB-SVF (n = 3). (G) RNA expression level of DPP4 in DPP4 positive cells from WT-ASCs and OB-ASCs. (H) Representative FACS Peak chart and proportion of DPP4+ cells from WT-ASCs and OB-ASCs. Data are expressed as means ± SEM, *P < .05, **P < 0.01,***P < .001,**** P < .0001.

Analysis of the frequency of WT and OB samples in each cluster revealed an increased proportion of monocytes and endothelial cells, which indicated increased inflammation and angiogenesis in obese adipose tissue. In contrast, frequency analysis suggested a decreased proportion of DPP4 + ASC-S1 cells in OB sample, which may be related to the damaged pluripotency ability of stem cells (Figure 2D). During aging,²⁵ scRNA-seq analysis indicated a significant decrease in the proportion of DPP4 + stem cells in iWAT (Supplementary Figure 3A-D), suggesting impaired proliferation and differentiation capacity. To confirm whether the stem cell subset decreased during obesity, we examined the population of Sca1 + CD34 + stem cells in WAT from WT and OB mice. The frequency of Sca1 + CD34 + stem cells decreased in OB mice compared to WT mice (Figure 2E). To further validate the results from single-cell sequencing, we isolated SVF from subcutaneous adipose tissue of WT and OB mice and examined the mRNA levels of mesenchymal markers. We found that the mRNA expression of DPP4, PPARγ, and CD142 in SVF of WT mice was significantly decreased in ob/ob mice compared with WT mice (Figure 2F). Using single-cell sequence analysis, we found that the expression of DPP4 was slightly reduced in each DPP4 + cells from OB-ASCs compared with WT-ASCs (Figure 2G). Flow experiment showed a significant decrease in the number of DPP4 + cells in SVF from OB-ASCs compared with WT-ASCs (Figure 2H). In conclusion, the downregulation of DPP4 in OB-ASCs may result from the decreased expression and number of DPP4+ cells, which is a predictive signal for impaired effectiveness of ASCs cell therapy.

Effects of obesity on adipose-derived stem cells: impaired proliferation and chondrogenesis despite retained anti-inflammatory ability

To further elucidate the mechanism underlying the disparate efficacy of the 2 ASCs in osteoarthritis, we conducted a comparative analysis of the characteristics of WT-ASCs and OB-ASCs in vitro. Microscopic examination revealed that ASCs from both groups exhibited a long spindle shape, with no discernible differences between the WT and OB groups (Figure 3A). However, subcutaneous ASCs isolated from obese mice exhibited decreased proliferation rates, as evidenced by a reduction in the data from CCK-8 assays(Figure 3B), along with impaired chondrogenic differentiation characterized by decreased protein and mRNA expression levels of sox9, aggrecan, and col2a1 (Figure 3C-E). To assess whether obesity affects the anti-inflammatory properties of ASCs in osteoarthritis, we induced RAW264.7 cells to mimic the chronic inflammatory milieu using LPS (1 μg/mL) and incubated them with supernatants from WT-ASCs and OB-ASCs. Our findings revealed that the levels of IL-1β and IL-6 significantly increased following LPS stimulation, a response effectively attenuated by treatment with supernatants from WT-ASCs. However, mRNA expression of proinflammatory genes such as IL-1β and IL-6 did not exhibit statistically significant differences between the LPS-treated group and the OB-ASCs-supernatant-treated group (Figure 3F-G).

Figure 3.

ASCs from obese mice exhibit impaired cell proliferation, chondrogenesis potential but preserved anti-inflammatory ability. (A) Cell morphology of WT-ASCs and OB-ASCs of P2 generation was compared (Scale bar = 100 μm). (B) WT-ASCs and OB-ASCs were subjected to the CCK-8 assay (n = 3). (C) Alcian staining in micromass cultures at different times with chondrogenic media. (D) Western blot analysis of Sox9 in micromass cultures at 21 d with chondrogenic media. (E) Gene expression of Sox9, Collagen Ⅱ, Aggrecan in micromass cultures at 21 d with chondrogenic media (n = 5-6). (F-G) Gene expression of IL-1β, IL-6 in RAW 264.7 cells treated with F12/DMEM or supernatant from mouse WT-ASCs and OB-ASCs (n = 3). (H) Cell morphology of WT-ASCs and HFD-ASCs of P2 generation (Scale bar = 200 μm). (I) ND-ASCs and HFD-ASCs were subjected to the CCK-8 assay (n = 3). (J) Alcian staining and Western blot analysis of Sox9 in micromass cultures at 21 d with chondrogenic media. (K) Gene expression of Sox9, Collagen Ⅱ, Aggrecan in micromass cultures at 21 d with chondrogenic media (n = 3). (L-M) Gene expression of IL-1β, IL-6 in RAW 264.7 cells treated with F12/DMEM or supernatant from mouse WT-ASCs and OB-ASCs (n = 3). Data are expressed as means ± SEM, *P < .05,**P < .01,***P < .001,**** P < .0001.

To further verify the impact of obesity on ASCs, we conducted additional validation using a diet-induced obesity mouse model. Mice were divided into groups fed either a normal diet (ND) or a high-fat diet (HFD, 60 kcal % fat) for 12 weeks. Following consumption of the high-fat diet, mice exhibited a significant increase in body weight gain and fasting blood glucose levels (Supplementary Figure 4A, B). ASCs were isolated from both ND and HFD mice, revealing cells with a consistent long spindle shape across groups, without significant differences (Figure 3H). CCK-8 assays revealed reduced proliferation ability of HFD-ASCs in vitro compared to ND-ASCs (Figure 3I). Analysis via western blotting and qPCR indicated that the chondrogenic potential was notably decreased in HFD-ASCs (Figure 3J-K). However, treatment with the supernatant of both ND-ASCs and HFD-ASCs significantly inhibited the levels of IL-1β and IL-6 induced by LPS (Figure 3L, M). In the DMM-induced OA mice model, both WT-ASCs and HFD-ASCs treatment ameliorated the abnormality of Stance Phase and Duty Cycle using CatWalk system. However, gait pattern damage, cartilage structure deterioration, and proteoglycan loss were observed in the HFD-ASCs treatment group compared with WT-ASCs group (Supplementary Figure 3D, E). Collectively, these findings suggest that obesity influences the properties of ASCs, leading to impaired cell proliferation and chondrogenesis. However, ASCs derived from a high-fat diet-induced obesity model retain their anti-inflammatory capabilities, contrasting with ASCs from leptin-deficient ob/ob mice, which exhibit a loss of anti-inflammatory capacity. Using single-cell RNA sequencing data from the HFD/ND and OB/WT,²⁶ we found that the expression of pro-inflammatory genes was increased in OB-ASCs compared with WT-ASCs, while no difference was observed between HFD-ASCs and ND-ASCs (Supplementary Figure 5A, B).

DPP4+ cells effectively reduce the pain of osteoarthritis

To gain further insight into the role of DPP4 + ASCs in the treatment of osteoarthritis, fluorescence-activated cell sorting (FACS) strategies were employed to isolate DPP4 + [CD45−, CD31−, DPP4+] cells and DPP4− [CD45−, CD31−, DPP4−] cells, purifying two subpopulations of ASCs from WT mice (Figure 4A). DPP4+ and DPP4- cells were cultured in F12 media to passage 2 (P2) in vivo. Subsequently, we evaluated the therapeutic effects of DPP4+ and DPP4− stem cells in the DMM model. Four weeks after DMM surgery, either DPP4+ or DPP4− ASCs were administered to the mice (Figure 4B). Four weeks post-treatment, CatWalk gait analysis (Figure 4C) and Von Frey tests (Figure 4E) were conducted to assess the effect on OA-associated pain. In the PBS-treated group of DMM-induced OA mice, the left hind/right hind (LH/RH) stance phase, print area, swing speed, duty cycle, and paw withdrawal threshold (PWT) were significantly decreased. These deficits were abrogated in DPP4+-treated DMM mice. In contrast, there were no statistically significant differences in the LH/RH stance phase, swing speed, duty cycle, and PWT between the PBS-treated group and the DPP4−-treated group. After these assessments, all mice were sacrificed, and joint samples were collected for fixation, dehydration, and histological evaluation. Representative images of safranin O-green staining and H&E staining indicated that the DPP4+ ASCs group showed minor signs of cartilage degeneration, whereas matrix depletion and surface roughness were clearly observed in the cartilage layers of the DPP4− ASCs group (Figure 4D). The average OARSI score of the DPP4− ASCs group was significantly lower than that of the PBS group but higher than that of the DPP4+ASCs group (Figure 4F). The positive cells of MMP13 decreased and Col2a1 increased in DPP+ ASCs group. TUNEL assay further confirmed the inhibition of cell death in DPP4+ ASCs group (Figure 4G). In conclusion, both DPP4 + and DPP4− cells effectively improve OA development, with DPP4+ cells showing a higher potential as a cell treatment for OA.

Figure 4.

DPP4 + cells can effectively reduce the pain of osteoarthritis. (A) Representative gating strategy for DPP4+ (CD31−CD45−DPP4+) and DPP4− (CD31−CD45−DPP4−). (B) Experimental flow chart. (C) Variations in the ipsilateral and contralateral hind limbs of gait parameters, including Stance Phase, Print Area, Swing Speed, and Duty cycle obtained from CatWalk analysis system (n = 8). (D) Representative Safranin-O and HE staining images of osteoarthritic knee joints that were collected 8 weeks after DMM surgery. (E) Mechanical paw withdrawal threshold were evaluated using Von Frey filaments (n = 6-7). (F) The severity of OA-like phenotype was analyzed using the Osteoarthritis Research Society International (OARSI) score system (n = 6). (G) Immunohistochemical staining of MMP-13, Col2a1, and Tunel staining were performed using murine cartilage. Data are expressed as means ± SEM, *P < .05, **P < .01.

DPP4+ cells have higher chondrogenesis potential and lower adipogenesis ability

Next, we characterized the DPP4− and DPP4+ ASCs, focusing on their anti-inflammatory properties and multipotent differentiation potential. Both DPP4− and DPP4+ ASCs exhibited long shuttle-shaped forms (Figure 5A). DPP4+ ASCs exhibited faster proliferation rates, as evidenced by an increase in the data from CCK-8 assays (Figure 5B). To assess the anti-inflammatory properties of DPP4+and DPP4− ASCs in osteoarthritis, we induced RAW264.7 cells to mimic a chronic inflammatory milieu using LPS (1 μg/mL) and incubated them with supernatants from F12/DMEM, DPP4+ ASCs, and DPP4− ASCs. Our findings revealed that the levels of IL-1β and IL-6 significantly decreased following incubation with supernatants from both DPP4+ and DPP4- ASCs. However, mRNA expression of IL-1β and IL-6 did not show statistically significant differences between the DPP4+ ASCs group and the DPP4− ASCs group (Figure 5C). The characterization of mesenchymal progenitor cells includes a capacity for high proliferation activity and multilineage differentiation. While DPP4− cells differentiated efficiently into adipocytes, DPP4+ cells exhibited a lower adipogenic capacity (Figure 5D-F). However, we found that the DPP4+ cell population displayed enhanced differentiation into chondrocytes, with higher induction of chondrocyte-specific marker proteins and genes (Figure 5G-I). Osteogenesis experiment showed that DPP4+ cells differentiated into more calcium nodules, and higher expression of osteocyte-related marker genes (Figure 5J-K). Together, these data confirm that DPP4+ cells have a high proliferation ability and are multipotent progenitors, possessing many properties that contribute to OA repair.

Figure 5.

DPP4+ cells display enhanced chondrogenesis but decreased adipogenesis differentiation capacity. Cells were isolated by using the following FACS strategy: Lin− (CD45−CD31−) cells were stained with DPP4. Groups were gated as follows: DPP4+ (CD31−CD45−DPP4+) and DPP4− (CD31−CD45−DPP4−). (A) Cell morphology of DPP4+ and DPP4− of P2 generation (Scale bar = 200 μm). (B) DPP4− ASCs and DPP4+ ASCs were subjected to the CCK-8 assay (n = 3). (C) Gene expression of IL-1β, IL-6 in RAW 264.7 cells treated with F12/DMEM or supernatant from mouse DPP4+ ASCs and DPP4− ASCs (n = 3). (D) Oil-Red-O staining of adipocyte in cell cultures from adult mice after exposure to the complete adipogenic differentiation cocktail. (E) Western blot analysis of Fabp4 and PPARγ at 8 d with adipogenesis media. (F) mRNA levels of Fabp4 and PPARγ at 8 d with adipogenesis media. (G) Alcian staining in micromass cultures at 21 d with chondrogenic media. (H) Western blot analysis of Aggrecan, Collagen Ⅱ and Sox9 in micromass cultures at 21 d with chondrogenic media. (I) Alcian staining in micromass cultures of DPP4− ASCs and DPP4+ ASCs with chondrogenic media. (J) Alizarin-Red staining of osteocyte in cell cultures from adult mice after exposure to the complete osteogenic differentiation cocktail. (K) Gene expression of ALPL (alkaline phosphatase), Ocn (Osteocalcin), iBsp (Integrin-binding sialoprotein) and Osx (Osterix) at 21 d with osteogenic cultures (n = 5-6).

Discussion

Our study demonstrates that adipose-derived stem cells (ASCs) from wild-type (WT) mice exhibit superior efficacy in mitigating osteoarthritis (OA) symptoms compared to ASCs from obese (ob/ob) mice. In contrast, ASCs from ob/ob mice only partially reversed OA symptoms, highlighting the negative impact of obesity on the therapeutic potential of ASCs. Single-cell sequencing of stromal vascular fractions (SVF) from WT and ob/ob mice revealed a significant decrease in DPP4+ stem cell subsets in obese mice. These DPP4+ cells were identified as a key subpopulation with high chondrogenic potential and low adipogenic capacity, crucial for effective OA treatment. In conclusion, the therapeutic efficacy of ASCs in OA is significantly influenced by the donor’s metabolic status. WT-ASCs demonstrated superior outcomes in alleviating OA symptoms and promoting cartilage repair compared to OB-ASCs. These results highlight the need for careful consideration of donor characteristics in developing ASC-based therapies for osteoarthritis.

Cartilage degeneration and joint pain are typical symptoms of OA. Several preclinical studies have shown that MSCs can effectively reduce cartilage degradation and joint pain in animal models of OA, including mice, rats, pigs, and dogs.²⁷ Zhu et al. reported significant changes in gait parameters, such as decreased print area, shorter duty cycle, smaller swing speed, and reduced paw withdrawal thresholds (PWTs) in the DMM-induced OA model.²⁸ Our results showed that ASCs treatment improved gait abnormalities and reduced pain sensitivity by reversing changes in gait parameters (stance phase, print area, swing speed, duty cycle) and PWT in DMM mice. This confirms that subcutaneous ASCs from WT mice are more effective in regenerating cartilage lesions and improving OA-related pain behavior. Koh found that a high BMI significantly predicted poor cartilage repair following clinical MSC implantation in OA knee patients.¹⁸ Consistent with our results, subcutaneous ASCs from OB mice and HFD mice were less effective in treating OA compared with WT mice. However, intra-articular injection of subcutaneous ASCs from HFD mice still improved some gait parameters, possibly due to the anti-inflammation effect.

ASCs are a promising source of cells for OA therapy due to their ability to migrate to injury sites, differentiate into appropriate cell phenotypes, and regulate immune responses and catabolism.²⁹ Previous studies have shown that the proliferation and angiogenic potential of ASCs from patients with obesity and metabolic syndrome are impaired compared to those from healthy individuals.²³ Studies have also reported that ASCs from obese patients are deficient in functionalities, including multilineage potential and immune regulation.³⁰ Consistent with these findings, our data showed that WT-ASCs had superior proliferative capacity compared to OB-ASCs and HFD-ASCs. Furthermore, our results suggested that WT-ASCs had greater chondrogenic potential compared to OB-ASCs and HFD-ASCs. Therefore, WT-ASCs improved cartilage degeneration, as indicated by increased safranin O staining and lower OARSI scores in the DMM-induced OA model, while OB-ASCs and HFD-ASCs had no effect on cartilaginous cells, joint structure, and OARSI scores.

Blanca Oñate found that inflammatory genes were upregulated in ASCs from obese patients, and inflammatory factor expression in the joint cavity of OA patients was increased.³⁰ Proinflammatory cytokines such as IL-1β and IL-6 contribute to OA development by inducing matrix-destructive enzymes like matrix metalloproteinases and ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs), leading to cartilage degeneration. In our vivo experiment, WT-ASCs effectively prevented cartilage destruction progression in an OA model by increasing the expression of anabolic enzymes such as collagen, Sox9, and aggrecan. Additionally, ASCs exert anti-inflammatory effects in a paracrine manner. Our previous study found that macrophages treated with ACM from S-ASCs significantly reduced levels of inflammation-related genes such as IL-1β and IL-6.¹⁶ In this manuscript, we found that compared with OB-ASCs, WT-ASCs exhibited superior anti-inflammatory capacity, as demonstrated by lower expression of IL-6 and IL-1β in an LPS-induced inflammation model. However, WT-ASCs and HFD-ASCs did not differ significantly in anti-inflammatory capacity. Taken together, obesity affects the characteristics of mouse subcutaneous ASCs, influencing the effect of OA disease.

ASCs are a heterogeneous population of cells comprising progenitors and lineage-committed cells used in stem cell therapies.²² Obesity affects ASC heterogeneity, leading to the accumulation of CD142 + ABCG1+ cell subsets, which represent a distinct adipogenic population.²⁰ In obese mice, the subcutaneous adipose tissue-specific subpopulation of CXCL14+ ASCs is significantly increased and closely related to inflammation regulation.³¹ Our research explored the inguinal WAT SVF from WT and ob/ob mice using single-cell sequencing, identifying two ASC populations: ASC-S1 (DPP4+) and ASC-S2 (DPP4−). Our data showed an abundance of DPP4+ cells in the SVF of WT mice. However, no discrepancy was observed regarding the abundance of DPP4+ cells in WAT of diet-induced obese mice. Merrick et al. found that the DPP4+ interstitial progenitor pool in the WAT of obese mice was depleted mainly in visceral WAT, with a tendency to decrease without significant difference in subcutaneous WAT.²² In contrast, our study indicated a significant decrease of DPP4+ progenitor cells in subcutaneous WAT. This difference likely stems from variations in high-fat diet durations and ob/ob mice. In our study, mice on a high-fat diet for 12 weeks had an average body weight of about 45 grams, while the mice in Merrick et al’s study weighed less than 40 grams. In conclusion, obesity affects ASC heterogeneity, resulting in the loss of the subcutaneous WAT DPP4+ progenitor pool. DPP4+ cells, marked by the expression of dipeptidyl peptidase-4 (DPP4), are highly proliferative and pluripotent progenitor cells. The high content of DPP4+ cells and the high proliferative potential of subcutaneous ASCs in WT mice align with this.²² In our study, we observed intra-articular injection of DPP4+ cells reversed these parameter changes, while DPP4- cells had less repairmen effect. Thus, we demonstrated that DPP4+ cells effectively alleviate OA-related pain, but further studies are needed to explore their chondroprotective effects.

There are several limitations to the present study. First, while an 8-week study allows conclusions to be drawn about the early efficacy of one-time ASCs treatment, a longer study would allow interpretation of sustained outcomes following multi-treatment with ASCs. Second, as OB-ASCs had no effect on cartilage repair with a decreased ratio of DPP4 + cells, lower proliferative ability, chondrogenesis potential, and increased pro-inflammatory activity. Which index could be used to measure the quality of ASCs? Third, while recording obesity's influence on ASCs quality through decreased the ratio of DPP4 + cells, could purified DPP4 + ASCs from obese be a candidate for treating musculoskeletal pathology in obese patients? Notwithstanding, a strength of this paper is that in vitro and vivo studies confirm a negative effect of obesity on the quality of ASCs. Therefore, ASCs harvested from obese patients may not provide the optimal biologic profile for treating OA. Further studies, potentially including weight loss for obese mice or patients, should be conducted to improve ASCs function,

Supplementary material

Supplementary material is available at Stem Cells Translational Medicine online.

Author contributions

Yan Tang, Zhi-ying Pang and Feng Yin designed research; Yan Tang, Zhen-yu Xu, Sai-sai Song, Yan-jue Song, Li-jie Yang, Lei Wang (Formal Analysis), Yan Tang, Zhi-ying Pang, Yang Liu, Shu-wen Qian and Qi-qun Tang (Data curation); Yan Tang and Feng Yin. (Writing—original draft).

Funding

This work was sponsored by Medical discipline Construction Project of Pudong Health Committee of Shanghai (PWYts2021-08), Science, Technology and Innovation Action Project of Shanghai Science and Technology Commission (23J11900400) and Peak Disciplines (Type IV) of Institutions of Higher Learning in Shanghai.

Conflicts of interest

The authors declared no conflict of interest.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.