Research ArticleStem Cells

Thy-1 (CD90) promotes bone formation and protects against obesity

See allHide authors and affiliations

Science Translational Medicine  08 Aug 2018:
Vol. 10, Issue 453, eaao6806
DOI: 10.1126/scitranslmed.aao6806

Stem cells’ balancing act

Mesenchymal stem cells (MSCs) differentiate into multiple cell types. Picke et al. investigated how MSC differentiation is regulated to maintain homeostasis between the bone and fat lineages. Genetic deletion of Thy-1, a protein expressed on multiple cell types including MSCs, prevented MSC differentiation into osteoblasts but promoted differentiation into adipocytes. Thy-1–deficient mice had increased body fat and decreased bone mass. High-fat diet induced obesity in wild-type mice and concurrent reduction in bone formation, which was associated with decreased Thy-1 expression on MSCs. Obese human subjects and subjects with osteoporosis showed reductions in serum soluble Thy-1. This study suggests that Thy-1 regulates the balance between bone and fat lineages, with possible implications for bone and metabolic disorders.


Osteoporosis and obesity result from disturbed osteogenic and adipogenic differentiation and present emerging challenges for our aging society. Because of the regulatory role of Thy-1 in mesenchyme-derived fibroblasts, we investigated the impact of Thy-1 expression on mesenchymal stem cell (MSC) fate between osteogenic and adipogenic differentiation and consequences for bone formation and adipose tissue development in vivo. MSCs from Thy-1–deficient mice have decreased osteoblast differentiation and increased adipogenic differentiation compared to MSCs from wild-type mice. Consistently, Thy-1–deficient mice exhibited decreased bone volume and bone formation rate with elevated cortical porosity, resulting in lower bone strength. In parallel, body weight, subcutaneous/epigonadal fat mass, and bone fat volume were increased. Thy-1 deficiency was accompanied by reduced expression of specific Wnt ligands with simultaneous increase of the Wnt inhibitors sclerostin and dickkopf-1 and an altered responsiveness to Wnt. We demonstrated that disturbed bone remodeling in osteoporosis and dysregulated adipose tissue accumulation in patients with obesity were mirrored by reduced serum Thy-1 concentrations. Our findings provide new insights into the mutual regulation of bone formation and obesity and open new perspectives to monitor and to interfere with the dysregulated balance of adipogenesis and osteogenesis in obesity and osteoporosis.


Mesenchymal stem cells (MSCs), present in the stroma of virtually all mammalian organs, are multipotent cells capable of differentiating into various cell lineages including osteoblasts and adipocytes (1, 2). MSC differentiation and fate decision are regulated by specific transcription factors: Runt-related transcription factor-2 (Runx2) determines osteoblast and chondrocyte differentiation, whereas peroxisome proliferator–activated receptor γ2 (Pparγ) is considered the master regulator of adipogenesis (1). Differentiated osteoblasts and bone are absent in Runx2−/− mice, and the affected mice die shortly after birth (3). Cells derived from Pparγ+/− mice have reduced ability to differentiate into adipocytes (1). Consistent with these findings, Pparγ+/− mice are protected from high-fat diet (HFD)–induced obesity (4).

In bone, tissue integrity is maintained by the controlled actions of bone-forming osteoblasts and bone-resorbing osteoclasts. During aging or osteoporosis, the number of osteoblasts declines, accompanied by a decrease in transcripts for osteoprogenitor markers and an increase in Wnt inhibitors such as sclerostin (Sost) and dickkopf-1 (Dkk-1), resulting in a reduction in bone mass (5). Osteoblasts and adipocytes originate from MSCs through alternative activation of reciprocal transcriptional programs (1, 2, 6). Thus, in parallel to decreased fat mass, Pparγ+/− mice develop increased bone mass due to increased osteoblastogenesis (1). The physiological importance of these two pathways can be appreciated upon disruption of this balance as observed in several conditions in humans including aging, hypercortisolism, or treatment with antidiabetic glitazones (7). Treatment with glucocorticoids or the antidiabetic drug class of glitazones increases bone loss, risk of fractures, and bone marrow adipogenesis (810, 10). On the other hand, patients with progressive osseous hyperplasia develop heterotopic bone within their adipose tissue (AT) (11, 12). Thus, balanced differentiation of MSCs to osteoblasts and adipocytes is essential for bone and fat homeostasis. Consequently, the identification of factors regulating the balance between osteogenic and adipogenic differentiation of MSCs is important to identify new targets to enhance bone formation in osteoporosis and to control AT development.

Thy-1, also known as cluster of differentiation 90 (CD90), is a glycosylphosphatidyl-anchored protein of the immunoglobulin superfamily expressed on the surface of MSCs, fibroblasts, activated microvascular endothelial cells, neurons, hematopoietic stem cells, and mouse T cells (1317). Integrins such as β2, β3, β5, CD97, and syndecan-4 were described as ligands/counter-receptors for Thy-1 (1721). In mesenchymal fibroblasts, Thy-1 controls the balance between proliferation and apoptosis as well as the differentiation of fibroblasts to myofibroblasts (15, 22). In healthy human lung, most of the fibroblasts express Thy-1, whereas in fibroblastic foci of idiopathic pulmonary fibrosis, Thy-1 expression is lost (23). Mice deficient in Thy-1 develop more severe lung fibrosis, suggesting that persistent Thy-1 loss is pathophysiologically relevant to disease progression (24). Primary dermal fibroblasts from Thy-1–deficient mice displayed a higher proliferative capacity and a lower rate of apoptosis compared to wild-type (WT) fibroblasts. Thy-1–mediated effects were strongly dependent on the interaction with β3 integrins (15). In rat lungs, Thy-1–deficient fibroblasts were described as profibrotic cells, whereas in dermal fibroblasts, Thy-1 maintains a differentiated phenotype reflected by enhanced spreading, higher expression of typical myofibroblast markers, and augmented contractile forces (15, 25).

Given this regulatory role of Thy-1 in the differentiation of mesenchyme-derived fibroblasts, we investigated the impact of Thy-1 on control of MSC differentiation into osteoblasts and adipocytes and the consequences for bone formation and AT development in vivo using Thy-1–deficient mice [knockout (KO)]. Our data identified Thy-1 as a critical molecule on MSCs that promotes osteogenesis and thus bone formation while inhibiting adipogenesis and obesity. Mechanistically, Thy-1 deficiency was accompanied by decreased expression of pro-osteogenic factors with concurrently increased expression of the Wnt inhibitors Sost and Dkk-1. Disturbed bone remodeling manifested in osteoporosis or dysregulated AT accumulation in patients with obesity was mirrored by reduced soluble Thy-1 (sThy-1) concentrations in human serum.


Thy-1 promotes differentiation of MSCs into osteoblasts while inhibiting differentiation into adipocytes

The tightly controlled lineage commitment of MSCs plays a critical role in the maintenance of bone homeostasis and lean body composition. Recent studies indicate a crucial role of Thy-1 in the control of differentiation of mesenchyme-derived fibroblasts (15, 22). Therefore, we investigated whether Thy-1 acts as a molecular regulator balancing adipogenic and osteogenic differentiation of MSCs. MSCs isolated from bone marrow of WT and Thy-1–deficient mice were characterized by the presence of MSC markers such as CD73, CD140a, and Sca-1 and the absence of hematopoietic markers (CD45, CD11b, and CD117; fig. S1).

We investigated the differentiation potential of WT and Thy-1–deficient (KO) MSCs. To exclude effects of proliferation, MSCs were seeded in high density, and differentiation was induced after 24 hours. KO MSCs had a decreased potential to differentiate into osteoblasts compared to MSCs from WT mice as shown by diminished ALP (alkaline phosphatase) activity and mineralization capacity detected by Alizarin red staining (Fig. 1A), as well as reduced expression of Runx2 and its target genes Oc (Osteocalcin), Osx (Osterix), and Tnalp1 (ALP; Fig. 1B). In parallel, Thy-1–deficient MSCs differentiated more rapidly into adipocytes as detected by Oil red staining (Fig. 1, C and D). Expression of master regulators of adipocyte differentiation, Pparγ and Cebpα (CCAAT enhancer binding protein), as well as Pparγ target genes such as AdipoQ (adiponectin) and Fabp (fatty acid binding protein), was increased in Thy-1–deficient adipocytes (Fig. 1E), thus reflecting the reciprocal regulation of adipogenic and osteoblast differentiation.

Fig. 1 Thy-1 supports osteogenic differentiation of MSC while concurrently inhibiting adipogenic differentiation.

MSCs were isolated from WT and Thy-1–deficient mice (KO). (A) Calcification was detected 14 days after initiating osteogenic differentiation. Alizarin red staining (red, mineralized bone matrix) and ALP activity (purple) by Fast blue RR solution (representative images; scale bars, 25 μm). (B) Gene expression of osteogenesis-related genes detected after 3 (Runx2) or 14 days (Oc, Tnalp1, and Osx) in culture by reverse transcription polymerase chain reaction (RT-PCR). mRNA values were normalized to the reference gene Rpl26, and arbitrary units (AUs) are indicated. (C) Oil red staining of lipid accumulation visualized 14 days after initiation of adipogenic differentiation (representative images; scale bars,100 μm). (D) Solubilization of Oil red absorption measured by enzyme-linked immunosorbent assay (ELISA). (E) Adipogenesis-related genes detected after 3 (Pparγ and Cebpα) or 14 days (AdipoQ and Fabp4) by RT-PCR. The mRNA values were normalized to the reference gene Rpl26, and AUs are indicated. (F and G) WT and KO MSC were differentiated into (F) osteoblasts (representative images; scale bars, 25 μm) and (G) adipocytes in the presence of function-blocking antibody against integrin β3 (aCD61) or integrin β1 (aCD29). Osteoblast differentiation was detected after 14 days by Alizarin red staining and adipocyte differentiation by Oil red staining and subsequent solubilization. (H and I) WT and KO MSC seeded on immobilized rThy-1 or control protein (Fc) and differentiated into osteoblasts or adipocytes. Osteoblast differentiation was detected after 14 days by Alizarin red staining (H) and adipocyte differentiation by Oil red staining (I) and subsequent solubilization. Data represent means ± SD of MSCs from at least four different mice per genotype. *P < 0.05, **P < 0.01, ***P < 0.001 [Student’s t test (B and E) and one-way analysis of variance (ANOVA) (G to I)].

Because integrin β3 (CD61) was previously described as a counter-receptor for Thy-1 (20, 21, 26), we studied whether Thy-1–mediated effects depend on interaction with integrin β3. Blocking integrin β3 with a specific antibody (aCD61) reduced osteogenic differentiation of WT MSCs, whereas differentiation of Thy-1–deficient MSCs was not affected (Fig. 1F). In parallel, adipogenic differentiation was increased by aCD61 treatment in WT MSCs but not in Thy-1–deficient MSCs (Fig. 1G). Blocking β1 integrins by aCD29 antibody did not alter osteogenic or adipogenic differentiation of MSCs from WT or KO mice (Fig. 1, F and G). To distinguish whether Thy-1 or integrin β3 signaling is responsible for the control of MSC differentiation, Thy-1–deficient MSCs were seeded on immobilized recombinant Thy-1 (rThy-1) fused to an Fc domain (15, 22). Thy-1 deficiency did not alter the adhesion of MSCs to rThy-1 or Fc (fig. S2). As expected, Thy-1–deficient MSCs displayed lower osteogenic and higher adipogenic differentiation potential on control Fc protein compared to WT MSCs (Fig. 1, H and I). Plating Thy-1–deficient MSCs on rThy-1 compensated for the lack of Thy-1 and resulted in osteogenic differentiation similar to WT MSCs (Fig. 1H). Consistently, adipogenic differentiation of Thy-1–deficient MSC was significantly reduced on rThy-1 (P < 0.001; Fig. 1I). These data indicate that integrin β3 signaling is involved in the fate decision of MSCs upon interaction with Thy-1.

Thy-1 decreases body fat mass while supporting bone formation

Because Thy-1 regulated MSC fate decision in vitro, we investigated the role of Thy-1 in osteogenesis and adipogenesis in vivo. Thy-1–deficient male and female mice had higher body weights (Fig. 2A) and an elevated percentage of subcutaneous and epigonadal fat mass compared to WT mice. Moreover, an increased fat volume was detected in the medullary cavity of bone from Thy-1–deficient mice (Fig. 2A). Food intake and activity were controlled and did not differ between the genotypes (fig. S3).

Fig. 2 Thy-1 decreases body fat mass.

(A) Body weights of male and female WT and Thy-1–deficient (KO) mice and ratio of subcutaneous (sc) AT and epigonadal fat to body weight of WT and KO mice. Fat volume per bone volume (FV/BV) in male mice was detected by osmium tetroxide staining (representative images, right). (B) Body weight, percentage of fat mass, and ratio of epigonadal and subcutaneous fat weight to body weight for male WT and KO mice fed an HFD for 12 weeks. (C) Gene expression of AT-related genes in epigonodal AT in (B) analyzed by RT-PCR. mRNA values were normalized to the reference gene Rpl26, and AUs are indicated. (D) Skin of WT and KO mice in (B) stained by hematoxylin and eosin (representative images; scale bars, 100 μm). Thickness of white dermal AT (WdAT) and size of adipocytes were measured using Keyence Analyzer software. Each point represents one mouse. *P < 0.05, **P < 0.01, ***P < 0.001 (Student’s t test).

Next, male WT and Thy-1–deficient mice were fed an HFD for 12 weeks, and AT was analyzed. Again, lack of Thy-1 resulted in a significantly higher body weight (P < 0.01) and percentage of body (P < 0.01), epigonadal (P < 0.01), and subcutaneous (P < 0.001) fat mass (Fig. 2B). Epigonadal AT from Thy-1–deficient mice displayed an enhanced expression of Pparγ, Fabp4, and AdipoQ (Fig. 2C). At the same time, lack of Thy-1 did not alter fasting blood glucose, glucose tolerance, food intake, or activity (fig. S4). Analysis of skin revealed a significantly higher width of the white dermal AT associated with a larger adipocyte size in Thy-1–deficient mice (P < 0.001; Fig. 2D). Thus, lack of Thy-1 promotes AT accumulation resulting in an obese phenotype.

Because of the reciprocal regulation of adipogenesis and osteogenesis, the impact of Thy-1 on bone mass and stability was investigated in male and female WT and Thy-1–deficient mice on normal diet (chow). Femoral trabecular (Fig. 3, A and I) and cortical (Fig. 3, B and J) bone mass and trabecular (Fig. 3C) and cortical (Fig. 3D) thickness were decreased in Thy-1–deficient compared to WT mice. Bone mineral density (BMD) and tissue mineral density (TMD) were reduced in Thy-1–deficient mice (Fig. 3, E and F). In parallel, markers of bone quality were negatively affected as reflected by an increased cortical porosity (Fig. 3G). Moreover, an increased crystallinity, a marker of advanced bone mineral age reflecting increased crystal size and perfection (27), was observed in Thy-1–deficient mice (Fig. 3H). Like male mice, female Thy-1–deficient mice also displayed lower femoral trabecular and cortical bone mass (fig. S5).

Fig. 3 Thy-1 increases bone mass and stability.

(A to G) Bone volume per total volume (BV/TV) of the trabecular (A) and cortical (B) bone compartments, trabecular (Tb.Th; C) and cortical (Ct.Th; D) thickness, cortical BMD and TMD (E and F), and cortical porosity (Ct.Po; G) detected by microcomputed tomography (μCT) in femora from WT and Thy-1–deficient mice (KO) under standard control diet. (H) Fourier transform infrared (FTIR) spectroscopy evaluation of crystallinity of the tibia cortical bone compartment. (I and J) Representative images of trabecular (I) and cortical (J) bone areas by μCT analysis. (K) Maximum force (Fmax) and elasticity modulus (Emod) analyzed using three-point bending test. (L) First cycle indentation distance (CID 1st) and total indentation distance (TID) evaluated by reference-point analysis. Each point represents one mouse. *P < 0.05 and **P < 0.01 (Student’s t test).

Biomechanical test revealed that less force was needed to fracture the femora of Thy-1–deficient mice, and the elastic modulus was also decreased (Fig. 3K) because of the increased cortical porosity (Fig. 3G), even though the femur length was reduced and the shaft width was increased (fig. S6A). Reference-point indentation showed increased first cycle and total indentation distances compared to WT mice (Fig. 3L), indicating lower material stability of bone in Thy-1–deficient mice without altered femur geometry parameters (Tt.Ar, Ct.Ar, and Ma.Ar; fig. S6B).

By histological analysis, bone mass (Fig. 4, A and C), osteoid surface per bone perimeter (Fig. 4, B and C), and bone formation (Fig. 4D) were decreased in tibiae with Thy-1 deficiency because of a reduced ratio of mineral surface to bone surface (Fig. 4, E and F). However, the number of osteoclasts was not altered in Thy-1–deficient mice (Fig. 4, G to I). Moreover, we did not detect alterations in the expression of Trap, an osteoclast marker, or Rankl, a central inducer of osteoclast differentiation, in bone of WT and Thy-1–deficient mice (fig. S7A). Bone marrow cells isolated from WT and KO mice differentiated similarly into osteoclasts in the presence of macrophage colony-stimulating factor (M-CSF) and receptor activator of NF-κB ligand (RANKL) in vitro (fig. S7B). In line with the histological parameters, serum concentrations of the bone formation marker P1NP (procollagen type 1 propeptide) were decreased in Thy-1–deficient mice (Fig. 4J), whereas the bone resorption marker CTX (carboxy-terminal collagen cross-links) was not altered (Fig. 4K). Again, female Thy-1–deficient mice displayed a similar phenotype as detected in male mice (fig. S5). We observed an increase in proliferation and cell growth in Thy-1–deficient MSCs, whereas apoptosis showed a decreasing trend in these MSCs (fig. S8). Thus, altered proliferation in Thy-1–deficient MSC does not explain the loss of bone in Thy-1 KO mice.

Fig. 4 Thy-1 positively influences bone formation.

Histological and histomorphometrical analysis of tibiae from male WT and Thy-1–deficient mice (KO). (A) Bone volume per total volume (BV/TV) and (B) osteoid surface per bone perimeter (Osteoid.Srf/B.Pm) by (C) von Kossa staining and van Giesson counterstaining (black, mineralized bone matrix; red, unmineralized osteoid/collagen; scale bars, 50 μm). (D and E) Bone formation rate per bone surface (BFR/BS; D) and mineral surface per bone surface (MS/BS; E) analyzed by double calcein labeling [green, calcein labels (F); scale bars, 50 μm]. (G to I) Number of osteoclasts per bone perimeter (N.Oc/B.Pm; G) and osteoclast surface per bone surface (OC.S/BS; H) detected via staining for tartrate-resistant acid phosphatase [TRAP; red, osteoclasts (I); scale bars, 50 μm]. (J and K) Serum concentrations of P1NP and CTX detected in WT and KO mice by ELISA. (L to N) WT and KO MSCs encapsulated in alginate beads and subcutaneously transplanted into WT mice. (L) Calcification (bone/total volume, BV/TV) detected after 3 weeks by μCT analysis. (M) Representative images of WT and KO MSCs in alginate beads 3 weeks after implantation. (N) Gene expression of Tnalp1 and Oc detected by RT-PCR. The mRNA values were normalized to the reference gene Rpl26, and AUs are indicated. Each point represents one mouse. *P < 0.05, **P < 0.01, and ***P < 0.0001 for effect of Thy-1 deficiency (Student’s t test).

To exclude background-related effects on MSC differentiation, WT and Thy-1–deficient MSCs were encapsulated in alginate beads and transplanted subcutaneously into host WT mice. Analysis of calcification by μCT revealed significantly increased bone formation in beads containing WT MSCs compared to Thy-1–deficient MSCs (P < 0.05; Fig. 4, L and M). Reduced expression of Tnalp1 and Oc in beads with encapsulated Thy-1–deficient MSC substantiated these findings (Fig. 4N).

Lack of Thy-1 is associated with altered Wnt signaling

Wnt signaling plays a central role in the balance between adipogenic and osteogenic differentiation (28). Wnt signaling is also controlled by inhibitor molecules such as sclerostin and Dkk-1 that prevent binding of Wnt ligands to its receptor (29, 30). Thy-1 deficiency was accompanied by a reduced expression of the Wnt ligands Wnt5a, Wnt11, and Wnt16 in bone (Fig. 5A), whereas the expression of the Wnt inhibitor Sost was increased (Fig. 5B). In the serum of Thy-1–deficient mice, both Dkk-1 and sclerostin were elevated (Fig. 5, C and D). Among these Wnt ligands, Wnt5a and Wnt16 expression was reduced in MSCs generated from Thy-1–deficient mice (Fig. 5E). Interaction of Thy-1–deficient MSCs with immobilized rThy-1 increased Wnt5a and Wnt16 expression, indicating that Thy-1 directly promotes the expression of these ligands (Fig. 5F). Blocking of Thy-1/β3 integrin interaction via blocking antibody reversed the Thy-1–mediated effect on Wnt5a and Wnt16 expression (Fig. 5F). When testing the responsiveness of WT and Thy-1–deficient MSC to stimulation of canonical Wnt signaling, we found that stimulation with Wnt3a resulted in attenuated induction of target genes such as Lef1 and Cyclin D1 in Thy-1–deficient MSCs (Fig. 5G). Together, Thy-1 promotes expression of Wnt ligands while inhibiting the expression of their inhibitors. Moreover, expression of Thy-1 on MSCs supports canonical Wnt signaling that might contribute to osteogenic differentiation in WT MSCs.

Fig. 5 Thy-1 deficiency is associated with altered Wnt pathway.

(A and B) Gene expression of Wnt ligands (A) and Wnt inhibitors Sost and Dkk-1 (B) in bone detected by RT-PCR. The mRNA values were normalized to the reference gene Gapdh and x-fold expression in Thy-1–deficient mice (KO) compared to WT mice. nd, not detectable. (C and D) Serum concentrations of Wnt inhibitors Dkk-1 (C) and sclerostin (D) were detected by ELISA. Each point represents one mouse. (E) MSCs were isolated from WT and Thy-1–deficient (KO) mice, and gene expression of Wnt ligands was detected by RT-PCR. The mRNA values were normalized to the reference gene Rpl26. X-fold expression in KO mice compared to WT mice is indicated (MSCs of ≥11 different mice per genotype). (F) KO-MSCs cultured for 24 hours on immobilized rThy-1 (KO-rThy-1) or control protein (Fc). Thy-1/β3 integrin interaction was blocked by a function-blocking antibody against integrin β3 (aCD61), and expression of Wnt5a and Wnt16 was detected by RT-PCR. The mRNA values were normalized to the reference gene Rpl26, and x-fold expression is indicated (MSCs of six different mice per genotype). (G) WT and KO MSCs were stimulated with Wnt3a (50 ng/ml) or vehicle for 24 hours. Gene expression of Cyclin D1 and Lef1 was detected by RT-PCR. The mRNA values were normalized to the reference gene Rpl26, and AUs are indicated (MSCs of three different mice per genotype). *P < 0.05, **P < 0.01, ***P < 0.001 [Student’s t test (A to E) and one-way ANOVA (F and G)].

Disturbed bone formation and AT accumulation are associated with decreased Thy-1 expression

Our data indicate that lack of Thy-1 results in increased AT accumulation and concurrent decreased formation of bone mass. Next, we investigated whether disturbed bone formation and AT accumulation are associated with altered Thy-1 expression in WT mice. We used a mouse model of HFD-induced obesity in which WT mice were fed for 20 weeks with HFD, resulting in marked AT accumulation (Fig. 6, A to C) and concurrent reduction of bone volume, bone formation, and gene expression of Runx2, Osx, and Oc (Fig. 6, D to F). Mice fed a chow diet (low fat) were lean and exhibited normal bone mass and bone formation (Fig. 6, A to F).

Fig. 6 Disturbed bone formation and AT accumulation in obesity are associated with decreased Thy-1 expression.

(A to C) Body weight (A), percentage of fat mass (B), and fat volume per bone volume (FV/BV; C) in male WT mice fed control chow or HFD for 12 weeks. FV/BV detected by osmium tetroxide staining (C; representative images, right). (D) Bone volume per total volume (BV/TV) analyzed by μCT (representative images, right). (E) Bone formation rate per bone surface (BFR/BS) analyzed by double calcein labeling. (F) Gene expression of Osx, Oc, and Runx2 in bone of control and HFD-fed mice. (G) The number of CFU per 100,000 cells detected from digested bone from chow and HFD mice. (H) Number of CD45/Sca-1+/CD73+ cells per 1 × 106 living cells in bone marrow of WT mice under chow and HFD analyzed by flow cytometry. (I and J) Expression of Pdgfrα (I) and Thy-1 (J) in bone detected by RT-PCR. The mRNA values were normalized to the reference gene Gapdh and AUs are indicated. (K) Expression of Thy-1 on CD45/Sca-1+ cells in bone detected by flow cytometry. Dead cells were excluded by Zombie staining. Mean fluorescence intensity of Thy-1 was detected on CD45/Sca-1+ cells of chow- and HFD-fed mice. Each point represents one mouse. *P < 0.05, **P < 0.01, ***P < 0.001 (Student’s t test).

The impact of obesity on the number of MSCs was investigated by detecting the number of colony-forming units (CFUs), flow cytometric analyses, and gene expression analyses to exclude an effect of obesity on MSC number. No differences in the number of CFUs between bone marrow cells from lean and obese mice were observed (Fig. 6G). Flow cytometric analyses documented similar numbers of CD45/Sca-1+/CD73+ cells in bone marrow of WT mice fed chow or HFD (Fig. 6H). Expression of Pdgfrα, a typical marker of MSCs, was not altered in bone of HFD-fed mice (Fig. 6I). Obese mice with reduced bone formation showed significantly reduced Thy-1 gene expression in bone compared to lean controls (P < 0.05; Fig. 6J). These data were further supported by flow cytometric analysis of Thy-1 protein expression on MSCs. To address this issue, we characterized stromal cells by the absence of CD45 in parallel with the expression of Sca-1 by flow cytometry. Expression of Thy-1 on CD45/Sca-1+ stromal cells was reduced under obese conditions (Fig. 6K). Thus, obesity reduced the expression of Thy-1 on MSCs, whereas the total number of MSCs was not altered.

Tumor necrosis factor–α and interleukin-1β down-regulate Thy-1 expression of MSCs

To gain insights into the mechanisms responsible for reduced Thy-1 expression in obesity, we stimulated WT MSCs with candidate cytokines, and Thy-1 expression was analyzed. Exogenous tumor necrosis factor–α (TNFα) and interleukin-1β (IL-1β) down-regulated Thy-1 gene and protein expression (Fig. 7, A and B) in mouse and human MSCs (Fig. 7, C and D). Obese mice expressed markedly enhanced Tnfα in bone (Fig. 7E), whereas Il-1β was not altered in bone of chow and HFD-fed mice (Fig. 7E). Moreover, Tnfα expression negatively correlated with Thy-1 expression in bone (Fig. 7F). These data suggest that elevated amounts of TNFα in obesity may contribute to the down-regulation of Thy-1, which in turn might result in reduction of bone formation as seen in Thy-1–deficient mice.

Fig. 7 Regulation of Thy-1 in vitro and in vivo.

(A to F) Regulation of Thy-1 in MSCs in vitro. (A) Thy-1 expression detected by RT-PCR (means ± SD of MSCs from ≥3 different mice) and x-fold expression of murine WT MSCs stimulated with indicated cytokines compared to untreated cells. (B) Representative Western blot of Thy-1 protein expression (n = 4). GAPDH, glyceraldehyde phosphate dehydrogenase. (C and D) Thy-1 gene and protein expression detected by RT-PCR (C) and Western blot (D) from human MSCs stimulated with TNFα or IL-1β. Means ± SD of MSCs from three different donors performed in duplicate. X-fold expression compared to untreated cells is indicated. (E) Gene expression of Tnfα and Il-1β in bone of chow and HFD-fed mice detected by RT-PCR. The mRNA values were normalized to the reference gene Gapdh, and AUs are indicated. Each point represents one mouse. ***P < 0.001 (Student’s t test). (F) Correlation analysis of Tnfα and Thy-1 expression in bone of chow- and HFD-fed mice (Spearman correlation analysis). (G to K) Detection of human sThy-1 by ELISA. (G) Serum sThy-1 in female and male subjects with BMI of <26 and age of <50 or >60. (H) Serum sThy-1 in individuals with normal weight (BMI, <25 kg/m2), overweight (BMI, 25 to 30 kg/m2), and obese subjects (BMI, >30 kg/m2). (I) Serum sThy-1 in healthy (T-score between −1 and +1) and osteoporotic (T-score, ≤2.5) nonobese (BMI, <30) women. (J) Spearman correlation analysis of P1NP concentration and sThy-1 in serum of healthy and osteoporotic nonobese (BMI, <30) women. (K) Serum sThy-1 in patients with polymyalgia rheumatica. Individuals were divided into patient subgroups without (T-score, ≥1) and with osteoporosis (T-score, ≤2.5). Each point represents one patient. *P < 0.05, **P < 0.01, ***P < 0.001 (Student’s t test).

Osteoporosis and obesity are characterized by decreased sThy-1 in human serum

Finally, we aimed to validate our findings in human disease. Detection of sThy-1 in human serum (31, 32) was used to reflect Thy-1 expression under conditions of disturbed bone formation and obesity. We detected no difference in serum sThy-1 concentrations between male and female subjects with a body mass index (BMI) of <26 kg/m2 (cohort 1, Fig. 7G) and could therefore exclude gender-specific effects. To detect an effect of obesity on serum sThy-1 concentrations, we divided patients according to their BMI: lean (BMI, <25 kg/m2), overweight (BMI, 25 to 30 kg/m2), and obese (BMI, >30 kg/m2). Patients with type 2 diabetes mellitus were excluded to eliminate a potential bias of impaired glucose metabolism on Thy-1 expression. Obese patients exhibited significantly reduced serum concentrations of sThy-1 (P < 0.05; cohort 1, Fig. 7H).

Because osteoporosis is characterized by disturbed bone formation and resorption, we assessed serum sThy-1 in lean women with and without osteoporosis. Women with a bone density T-score between −1 and +1 served as a control group, and women with a T-score of ≤2.5 were classified as osteoporotic (table S3). We found significantly reduced sThy-1 serum concentrations in osteoporotic women (P < 0.001; Fig. 7I). Moreover, sThy-1 positively correlated with P1NP, a serum marker for bone formation (Fig. 7J). These data were supported by measurement of sThy-1 in cohort of patients with polymyalgia rheumatica with or without osteoporosis. Polymyalgia rheumatica is an inflammatory condition often associated with osteoporosis (33). All patients displayed a C-reactive protein (CRP) of <9.7 mg/liter and received only low-dose prednisolone (<2 mg/day), excluding severe ongoing systemic inflammation in these patients (table S4). Again, patients with a T-score of ≤2.5, indicative of osteoporosis, displayed significantly lower serum sThy-1 concentrations than patients with a T-score ranging from −1 to +1 (P < 0.001; Fig. 7K). Together, both low bone mass and increased fat accumulation in humans observed in osteoporosis and obesity, respectively, were reflected by decreased serum concentrations of sThy-1.


Disorders such osteoporosis and obesity are major health challenges to our aging society. They are characterized by decreased activity and/or numbers of osteoblasts, whereas the fat-storing function of adipocytes and/or their number increase (34, 35). This reciprocal relationship results from a shift in the differentiation capacity of MSCs, precursors of both osteoblasts and adipocytes. Various signaling molecules affecting MSC fate have been described, but the complex underlying mechanisms are not yet fully understood. Here, we identified Thy-1 as a critical molecule regulating the MSC fate decision. We showed that Thy-1 on MSCs promotes osteogenesis and, thus, bone formation while inhibiting adipogenesis and obesity.

Controversial results regarding the impact of Thy-1 on osteogenesis have been published. Comparison of different subpopulations of AT-derived MSCs or cells from rat dental pulp has shown that Thy-1+ cells efficiently produce bone (36, 37). Higher ALP activity, an increased amount of mineralized matrix, and enhanced osteogenic gene expression profile in Thy-1+ MSCs compared to Thy-1–negative MSCs derived from human AT were observed (36). Mouse embryonic fibroblasts from Thy-1–deficient mice displayed a strongly reduced osteogenic differentiation (38). In contrast, stimulation of osteogenesis was detected after reducing Thy-1 expression in MSCs from different sources in ex vivo cultures (39). Moreover, Paine et al. (38) did not find differences in bone formation in Thy-1–deficient mice. However, our data show both reduced osteogenic differentiation of Thy-1–deficient MSCs in vitro and decreased bone mass, formation, and stiffness in Thy-1–deficient mice, indicating that Thy-1 is a pivotal factor supporting osteogenesis and bone strength in vivo. Increased calcification of WT MSCs encapsulated in alginate beads compared to Thy-1–deficient MSCs upon transplantation into the same mouse substantiated the cell-autonomous effect of Thy-1 on MSC differentiation.

Bone remodeling is realized by the balanced activity of bone-forming osteoblasts and bone-resorbing osteoclasts. Reduced bone formation rate, surface mineralization, amount of osteoid, and diminished serum P1NP concentration detected in Thy-1 deficiency indicate a role of Thy-1 in bone formation. Unaltered numbers of osteoclasts and CTX concentration in serum of Thy-1–deficient mice compared to WT suggest that Thy-1 is not involved in bone resorption.

In parallel to reduced osteogenic differentiation, Thy-1–deficient MSCs showed increased adipogenic differentiation. Consistently, Thy-1–deficient mice displayed a higher body weight accompanied by an increase in AT in the bone, skin, and epigonadal and subcutaneous fat depots under chow diet. Moreover, Thy-1 deficiency promoted fat accumulation under HFD conditions as already described by Woeller et al. (40). Similar food intake and activity of WT and Thy-1–deficient mice excluded that these systemic parameters were responsible for increased adiposity of Thy-1–deficient mice. In line with our findings, another study showed that ectopic expression of Thy-1 in adipocyte-like 3T3-L1 cells inhibits adipogenesis (40). Koumas et al. (41) demonstrated that only Thy-1−/− human myometrial and orbital fibroblast subsets were capable of developing lipid droplets after exposure to adipogenic stimuli. Woeller et al. (40) demonstrated that ectopic expression of Thy-1 in a pre-adipocyte fibroblast cell line inhibits fyn kinase activity resulting in inhibition of Pparγ expression and subsequently diminished adipogenesis. Here, we also observed reduced expression of Pparγ in WT compared to Thy-1–deficient MSCs.

Compensation for the lack of Thy-1 by seeding Thy-1–deficient MSCs on rThy-1 indicated that Thy-1 is also capable of promoting osteogenesis and inhibiting adipogenesis via its interaction partner. MSCs interact with the surrounding microenvironment mainly through integrins (42, 43). Integrins such as β2, β3, and β5 were described as counter-receptors for Thy-1 (1721), and MSCs express β1 and β3 integrins (44). Reduction of osteogenesis, in parallel to the support of adipogenesis upon blocking integrin β3 signaling in WT but not in Thy-1–deficient MSCs, demonstrated the critical role of Thy-1/integrin β3 interactions for the control of adipogenesis versus osteogenesis in bone marrow–derived MSCs. Our findings support data showing the impact of integrin signaling in the control of osteoblast cell fate and functions (45). Silencing of α5 integrin blunts osteoblast differentiation (46). Integrin αvβ3 is critical for CYR61-mediated bone morphogenetic protein 2 (BMP2) expression and subsequent osteoblast differentiation (47). The extracellular matrix-integrin-cytoskeleton axis is a major pathway in integrin-mediated control of osteoblast function. Our data indicate that integrin signaling mediated by cell-cell interaction such as β3/Thy-1 interaction also controls osteoblast cell fate and function.

The Wnt signaling pathway plays a crucial role during MSC differentiation into the osteogenic or adipogenic lineage (28, 48, 49). Our finding of reduced expression of Wnt ligands with concurrent increase of Wnt inhibitors and alleviated response to canonical Wnt stimulation in Thy-1–deficient MSCs suggests that Thy-1 interferes with Wnt signaling in several ways. A link between Wnt signaling and Thy-1 was previously shown in mouse embryonic fibroblasts, where the activation of the Wnt pathway and the expression of the Wnt targets Cyclin D1, Axin2, and BMP4 was higher in Thy-1+ compared to Thy-1 cells (50).

Our data show that lack of Thy-1 reduces bone formation and increases AT. Obesity characterized by marked AT accumulation and concurrent reduction of bone volume and formation exhibited strongly reduced Thy-1 expression. Increased fat mass is detrimental to bone formation and bone mass (34, 5156), although obesity is traditionally viewed as beneficial to bone health because of the well-established positive effect of mechanical loading on bone formation due to increased weight (5759). Consistent with our observations, in mouse models, HFD decreases femoral trabecular bone mass by increasing trabecular separation and reducing trabecular number and connectivity density (58, 60). As expected, we detected elevated expression of TNFα in bone of obese mice. Inverse correlation of TNFα and Thy-1 expression in bone pointed to a critical role of TNFα in the regulation of Thy-1 expression. TNFα strongly reduced Thy-1 expression in murine and human MSCs. Our data suggest that elevated amounts of TNFα, as found in obesity, down-regulate Thy-1, which in turn might result in reduction of bone formation as seen in Thy-1–deficient mice. Consequently, understanding regulatory mechanisms of Thy-1 expression, such as TNFα-mediated inhibition of Thy-1, might create additional therapeutic options to counteract the dysregulated balance of adipogenesis and osteogenesis in obesity and osteoporosis.

On the basis of our animal model data, we hypothesized that Thy-1 expression is decreased in human osteoporosis and obesity. Thy-1 is detectable as a soluble molecule in the serum (31, 32). We previously found elevated serum sThy-1 concentrations in patients with systemic sclerosis, particularly in patients with vascular involvement of the lungs, suggesting the use of serum sThy-1 as a marker for diagnosis of pulmonary arterial hypertension in systemic sclerosis (31). Here, assessment of sThy-1 serum concentrations was used to reflect Thy-1 expression under conditions of disturbed bone remodeling and obesity. Similar amounts of sThy-1 in women and men largely exclude gender-specific effects. Serum sThy-1 concentrations were lower in patients with osteoporosis compared to non-osteoporotic patients. Similarly, obese patients displayed reduced serum sThy-1 concentrations. Thus, serum sThy-1 mirrors dysregulated balance of bone formation and AT accumulation in osteoporosis and obesity.

As summarized in fig. S9, we identified Thy-1 as a critical molecule for MSC differentiation promoting osteogenesis and thus bone formation while inhibiting adipogenesis and obesity. Interference with mechanisms controlling Thy-1 expression or stimulation of osteogenic pathways by Thy-1 might offer new avenues to restore the dysregulated balance of adipogenesis and osteogenesis in obesity and osteoporosis. Moreover, detection of serum sThy-1 might provide a new indicative tool for monitoring bone quality in osteoporosis and obesity.


Study design

This study was performed to evaluate the impact of Thy-1 on the differentiation potential of MSCs and the consequences for bone and AT homeostasis. MSCs were prepared from WT and Thy-1–deficient mice, and the differentiation capacity was analyzed by in vitro differentiations assays. Bone mass and quality were characterized by μCT technique and histology analysis, as well as by mechanical tests in WT and Thy-1–deficient mice. PCR and ELISA were used to investigate the mechanisms of Thy-1–mediated control of bone and fat homeostasis. Primer sequences and antibodies are listed in tables S1 and S2. Using HFD-induced obese mice and in vitro stimulation experiments of MSCs, the regulatory mechanisms of Thy-1 were analyzed. Finally, we measured the amount of sThy-1 in serum of human patients with disturbed bone remodeling (osteoporosis) and AT accumulation. Patient data are summarized in tables S3 and S4. Serum was obtained after written informed consent of all participants. Studies were approved by the ethics committee of the University of Leipzig (approval no. 159-12-21052012) and Technische Universität Dresden (approval nos. EK24508201 and EK 161052010). Individual subject-level data are reported in table S5.


All animal experiments were performed in accordance with institutional and state guidelines and approved by the Committee on Animal Welfare of Saxony (TVV 03/16, T26/16). Mice were kept under a 12-hour light-dark cycle and given food and water ad libitum. Thy-1–deficient (KO) mice on C57BL/6J background were provided as a gift from R. Morris [King’s College London (61)]. Twelve-week-old male Thy-1 deficient (KO) mice and WT C57BL/6J mice were used for bone analysis. Four- to 5-week-old male C57BL/6J mice were fed either an HFD (EF R/M D12331 diet modified by Surwit, ssniff) or a chow diet for 12 to 20 weeks. Body fat percentage was determined using nuclear magnetic resonance technology with an EchoMRI700 instrument (Echo Medical Systems), and fasting glucose was measured after 14 to 16 hours of starvation.

Male and female WT and KO mice were observed via a feeding and infrared activity monitoring system (TSE Systems GmbH). The mice were kept singly with free access to chow or HFD and water. After 1 day of habituation, spontaneous activity and daily food intake were recorded. The average food intake and activity of each animal were calculated by the mean of three subsequent days.

Isolation and culture of MSCs

Femora and tibiae of Thy-1–deficient and WT mice were enzymatically digested using Liberase DL (26 U/ml) (Roche) for 2 hours at 37°C and 5% CO2. Cell suspension was filtered through a cell strainer (70 μm) and cultured in α-minimal essential medium (MEM; Lonza) supplemented with 10% fetal calf serum (FCS; Thermo Fisher Scientific) and 1% penicillin/streptomycin (Biochrom AG) at 37°C and 5% CO2. The medium was replaced every 3 days. CD11b+ cells were removed by magnetic cell separation using anti-CD11b magnetic beads according to the manufacturer’s protocol (Miltenyi Biotec). MSCs were stimulated with TNFα (10 ng/ml), IL-1β (10 ng/ml), transforming growth factor–β (50 ng/ml), IL-6 (50 ng/ml; Miltenyi Biotec) or 1 μM insulin (Sigma).

For culture of MSCs on recombinant mouse Thy-1 (rThy-1), 96-well NUNC MaxiSorp plates (Nunc) were coated with rThy-1 fused to Fc protein (4 μg/ml; Enzo) or as control with Fc protein (Enzo) in a coating buffer containing 0.2 M Na2CO3 and 0.2 M NaHCO3 (pH 9.5) overnight at 4°C. After several washes with the medium, plates were blocked with α-MEM containing 10% FCS.

Cell differentiation

As described for fibroblasts previously (15), we observed that Thy-1–deficient MSCs show higher cell growth and proliferation and reduced apoptosis (fig. S8). To exclude proliferation-associated effects, MSCs were plated at high density (100,000 cells/cm2), and differentiation was induced after 24 hours.

Osteogenic differentiation. Osteoblast differentiation was induced using the StemPro Osteogenesis Diff Kit (Gibco) according to the manufacturer’s protocol. For ALP staining, cells were rinsed with phosphate-buffered saline (PBS), fixed in 60% buffered acetone for 15 min at room temperature (RT), washed with PBS, and stained with Fast Blue RR solution (Sigma) for 30 min at RT. Cells were washed again with PBS and photographed with a BZ-9000Z microscope (Keyence). Calcification was verified by Alizarin red staining. Cells were washed with PBS, fixed in 60% buffered acetone for 15 min at RT, washed with PBS, and stained with Alizarin red solution (Sigma). After several washes with water, the cells were photographed with a BZ-9000Z microscope or dye was solubilized by 100 mM hexadecylpyridinium chloride monohydrate for 30 min, and absorption at 540 nm was measured at a BioTek Synergy HT microplate reader (BioTek Instruments).

Adipogenic differentiation. Adipogenic differentiation was induced by addition of α-MEM containing 10% FCS, insulin (1 μg/ml; Sigma), 1 μM dexamethasone (Sigma), and 0.5 mM 3-Isobutyl-1-methylxanthine (Sigma). The formation of adipocytes was monitored by Oil red staining. Cells were washed with PBS, fixed in 60% buffered acetone for 15 min at RT, washed with PBS, and stained with Oil red solution (Sigma). Unbound dye was removed by thorough rinsing with water, cells either photographed with a BZ-9000Z microscope or the dye was solubilized by addition of 70% isopropanol, and absorption at 540 nm was measured at BioTek Synergy HT microplate reader (BioTek Instruments).

Osteoclast differentiation. Bone marrow cells were cultured for 2 days in α-MEM containing 10% FCS, 1% penicillin/streptomycin, and M-CSF (50 ng/ml; Miltenyi Biotec). After 2 days, RANKL (50 ng/ml; Miltenyi Biotec) was added, and the medium was exchanged every 2 days. After 7 days, the number of osteoclasts was determined by TRAP staining (Sigma) and counting of osteoclasts (TRAP+ cells with more than three nuclei).

Assessment of bone and bone marrow fat volume

Femura were fixed for 48 hours in 4% paraformaldehyde (Carl Roth) and then dehydrated in 80% ethanol. By using the μCT vivaCT40 (isotropic voxel size of 10.5 μm, 70 kVp, 114 μA, 200-ms integration time; ScancoMedical), femora and third lumbar vertebral bodies were analyzed as previously described (34, 62). The analysis of trabecular and cortical bone volume per total volume (BV/TV), BMD, thickness (Tb.Th and Ct.Th for the trabecular and cortical thickness, respectively), and cortical porosity (Ct.Po) was performed using established analysis protocols, and the μCT parameters were reported according to international guidelines (63). After identification of the periosteal and endosteal surfaces at the femoral midshaft, the total area (Tt.Ar), marrow area (Ma.Ar), and cortical bone area (Ct.Ar) were calculated as the area within the periosteal boarder, the area within the endosteal border, and the area between the periosteal and endosteal borders, respectively. To assess the bone marrow fat content, fixed femora were decalcified (Osteosoft) for 7 days. After scanning the femora via μCT to ensure complete decalcification, they were washed with PBS for 5 min, stained for 2 hours with 2% osmium tetroxide (Electron Microscopy Science), diluted in 0.1 M sodium cacodylate buffer (pH 7.4) (5), and transferred into PBS. The complete femur was analyzed to evaluate the fat volume using the established protocols from ScancoMedical. Femora were measured at a resolution (70 kVp, 114 μA, 300-ms integration time, and 10.5-μm isotropic voxel size). Femur length (highest greater trochanter region to distal condyle) and width (femur shaft) were measured using a caliper (MIB Messzeuge GmbH).

FTIR spectra were measured at a Spotlight 400 (PerkinElmer) attached to a Frontier FTIR spectrometer (PerkinElmer) in attenuated total reflection (ATR) mode. Spectra were acquired in a wave number range from 4000 to 570 cm−1 with a resolution of 4 cm−1. For each pixel, 16 measurements were taken, and the size of each pixel was 6.25 μm. SpectrumIMAGE software R. (PerkinElmer) was used for automatic background subtraction for each pixel spectrum and for automatic ATR correction. Each region of interest was 300 μm × 300 μm in size and included the whole cortical thickness. The spectra were postprocessed using a customized MATLAB routine with poly(methyl methacrylate) subtraction and baseline correction. Crystallinity was calculated by subpeak fitting of the entire phosphate peak (1154 to 900 cm−1) and calculation of the ratio of the 1030 and 1020 cm−1 subpeaks (27). The subpeak of 1020 cm−1 is linked to nonstoichiometric apatite, whereas the subpeak of 1030 cm−1 is linked to stoichiometric apatite (64). An increase of the crystallinity ratio reflects either an increase of stoichiometric apatite or a decrease of nonstoichometric apatite (that is, the crystallinity ratio reflects crystal size and perfection).

Biomechanical testing: Reference-point analysis and three-point bending test

Femora were shock-frozen in liquid nitrogen and thawed from −80°C 5 min before reference-point indentation (BioDent2, ActiveLife Scientific). Bones were stabilized using an ex vivo small bone stage filled with PBS to avoid sample drying. The reference probe was located at the anterior side of the femur shaft, and indentation measurements were performed (2 N, five cycles) in triplicate for each bone sample by lifting up of the measurement head unit and movement of the sample by a minimum. The first cycle indentation distance (CID 1st) and total distance increase (TID) were calculated. Immediately afterward, the bone samples were mechanically tested in three-point bending (zwicki-Line 2.5 kN, Zwick). Load was applied at the anterior site of the femoral shaft until failure, and the maximum load (Fmax; N) and elastic module (Emod; MPa) were recorded, as previously reported (34, 65).

Detection of human sThy-1 in serum

sThy-1 was detected in serum as described previously (31).

For cohort 1, in connection with a study on insulin resistance at the Department of Medicine, University of Leipzig, a total of 88 (men, n = 15; women, n = 73) were consecutively recruited to represent a wide range of obesity. The age ranged from 17 to 80 years, and the BMI range was from 17.1 to 39.9 kg/m2. Patients with type 2 diabetes were excluded. All baseline blood samples were collected between 8:00 a.m. and 10:00 a.m. after an overnight fast. BMI was calculated as weight divided by squared height.

For cohort 2 (table S3), healthy women and women with osteoporosis were recruited at the Department of Medicine, University of Leipzig and Bone Center of the Medical Faculty of the Technische Universität Dresden. Osteoporosis was diagnosed on the basis of dual-energy x-ray absorptiometry (DXA) at the lumbar spine and hip. Women with a BMD of a T-score between −1 and +1 were included into the control group, and women with a T-score below −2.5 were classified as osteoporosis. Individuals with glucocorticoid treatment, type 2 diabetes, and acute and chronic inflammatory diseases were excluded from the study.

For cohort 3 (table S4), serum samples and clinical data were collected from 19 patients with polymyalgia rheumatica at the Division of Rheumatology at the University Clinic Dresden. The T-score was evaluated by DXA measurements of the spine. Only patients with a CRP lower than 9.7 mg/liter were included to exclude effects of systemic inflammation. Moreover, the present dose of prednisolone at the time of blood drawing did not exceed 2 mg/day.

Statistical analysis

Results are presented as means ± SD. Distribution of data was assessed by Shapiro-Wilk test. Depending on the normality of the data, analysis was performed using the Mann-Whitney rank sum test or the two-tailed t test. Multiple parameters were tested using one-way ANOVA test. P < 0.05 was considered significant.


Fig. S1. Characterization of MSCs in Thy-1–deficient and WT mice.

Fig. S2. Thy-1 does not affect adhesion of WT and Thy-1–deficient MSCs.

Fig. S3. Food intake and activity do not differ between WT and Thy-1–deficient mice.

Fig. S4. Blood glucose metabolism, food intake, and activity of WT and Thy-1–deficient mice under HFD.

Fig. S5. Thy-1 increases bone mass in female mice.

Fig. S6. Thy-1 deficiency affects femur size but does not alter femur geometry.

Fig. S7. Osteoclasts in WT and Thy-1–deficient mice.

Fig. S8. Thy-1–deficient MSCs exhibited higher cell growth and proliferation in parallel to a decreased apoptosis rate.

Fig. S9. Thy-1 supports osteogenesis and bone formation while inhibiting adipogenesis.

Table S1. Murine primer sequences for RT-PCR.

Table S2. Antibodies.

Table S3. Characterization of healthy and osteoporosis subjects (cohort 2).

Table S4. Characterization of patients with polymyalgia rheumatica (cohort 3).

Table S5. Individual subject-level data (Excel file).


Acknowledgments: We thank D. Gutknecht, I. Gloe, S. Hippauf, N. Pacyna, and U. Kelp for their excellent technical assistance. Funding: This research study was supported by the Deutsche Forschungsgemeinschaft (SA863/2-3 to A.S.; SFB Transregio 67, project B4 to U.A.; SFB Transregio 67, project B2 to L.C.H.; BU2562/3-1 to B.B.; KR3614/4-1 to U.K.; and SFB 1149, project C02 to J.P.T.). A.-K.P. and J.S.-H. were supported by MeDDrive grants of the medical faculty of the Technische Universität Dresden and a Young Investigator Award of the Dachverband Osteologie e.V. F.N.S. acknowledges the Joachim Herz Stiftung for a PhD Scholarship in cooperation with the PIER Helmholtz Graduate School, University of Hamburg and Deutsches Elektronen-Synchrotron Hamburg. Author contributions: A.S. and A.-K.P. designed the study, performed experiments, analyzed and interpreted data, and wrote the manuscript. J.P.T. discussed data and revised the manuscript. U.K. provided feeding and activity data. U.A. provided expertise and methods for RNA analysis. M.R., J.S.-H., and L.C.H. provided experiences and methods for bone analysis and interpreted the bone data. V.V. was involved in MSC experiments. G.M.C. performed part of the μCT measurements and analyzed data. F.N.S. and B.B. carried out compositional analyses. J.C.S. designed translational studies. M.B., M.R., E.T., M.W., and L.C.H. provided patient material and data. All authors discussed the data and read and edited the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.

Stay Connected to Science Translational Medicine

Navigate This Article