Macrophages deliver mitochondria to MSCs in the bone marrow
To verify mitochondrial transfer from macrophages to MSCs, we labeled MSCs with MitoTracker Green and co-cultured them with macrophages (labeled with MitoTracker Deep Red) for 0, 3, 6, and 24 h. Confocal microscopy and flow cytometry confirmed the presence of macrophage-derived mitochondria in MSCs (Fig. 1a, d, and e; Supplementary Video 1). Previous studies found that mitochondrial transfer relies on membrane carriers such as tunneling nanotubes (TNTs) and microvesicles (MVs) in most cases23,24. Moreover, free mitochondria or mitochondrial components can also be extruded or internalized into recipient cells. To investigate the mechanism of mitochondrial transfer and acceptance, we analyzed mitochondrial transfer from macrophages (labeled with MitoTracker Deep Red) to MSCs using a transwell (TW) culture system (0.4 µm pore size) to isolate MSCs from macrophages. Mitochondrial transfer from macrophages to MSCs was partially inhibited in TW culture (Fig. 1b–e), indicating that free mitochondria or mitochondria-containing MVs can be internalized by MSCs. A transmission electron microscopy (TEM) analysis demonstrated that macrophages deliver mitochondria to MSCs in various ways, including free mitochondria, MVs, and TNTs. Further, these structures were internalized by MSCs (Supplementary Fig. 1a, b). MSC internalization was inhibited by dynasore, an endocytosis inhibitor that acts on dynamin-dependent clathrin activity but had no effect on MSC viability (Supplementary Fig. 1c-e).
To test whether intercellular mitochondrial transfer occurs in vivo, we extracted bone marrow-derived macrophages (BMDMs) for cell transplantation experiments. BMDMs were labeled with the mitochondrial marker protein Tom20 and used in subsequent experiments after flow cytometry (Supplementary Fig. 1f-h). BMDMs expressing Tom20-GFP were re-injected into irradiated mice via the tail vein. We analyzed the MSC phenotype using flow cytometry (Supplementary Fig. 1i) and found that green fluorescence appeared in MSCs, indicating that mitochondrial delivery from macrophages to MSCs occurs in vivo (Fig. 1f-i).
Mitochondrial function in macrophages is altered in osteoporosis
Macrophages are essential for regulating bone homeostasis5,6. Moreover, immune cell activation and function are closely related to cellular metabolism25,26. To investigate the metabolic profile of macrophages in osteoporosis, we used an ovariectomy (OVX) -operated mouse osteoporosis model. Using the flow cytometric analysis of F4/80 macrophages from sham and OVX mice, we assessed mitochondrial reactive oxygen species (mtROS) production by staining cells with 2,7-Dichlorodi -hydrofluorescein diacetate (DCFH-DA). The mtROS levels of BMDMs from OVX mice were higher than those in sham mice (Fig. 2a, b). We also analyzed the mitochondrial membrane potential (MMP) of the BMDMs. The flow cytometry of tetramethylrhodamine methyl ester (TMRM) showed that the MMP was elevated in BMDMs from OVX mice (Fig. 2c, d).
Next, we explored whether the expression of specific gene sets was induced under osteoporotic conditions to identify stage-specific metabolic profiles. We compared the macrophage transcriptomes in sham and OVX mice using RNA-seq. Principal component analysis (PCA) (Supplementary Fig. 2a) and hierarchical clustering (HC) (p ≤ 0.05) (Fig. 2e) of the analyzed genes revealed profound transcriptional differences between macrophages between the sham and OVX groups. We found 1908 dysregulated genes with at least a 2-fold change (1220 upregulated and 688 downregulated) (Fig. 2f) in BMDMs treated with OVX compared with Sham-treated controls. We then performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses based on the RNA-seq data (Fig. 2g; Supplementary Fig. 2b). The GO analysis revealed that the enrichments occurring in OVX rats mainly included the regulation of immune effector processes, responses to oxidative stress, mitochondrial organization, and mitochondrial ATP synthesis coupled with electron transport, carbohydrate transport, and glucose transmembrane transport (Fig. 2g). Metabolism-related pathways, including HIF-1 and PI3K-Akt signaling, were enriched in osteoporosis (Supplementary Fig. 2b). Furthermore, compared to those of sham mice, BMDMs from OVX mice showed enrichment of specific mitochondrial processes, including Complex I, Complex III, reactive oxygen species, glutathione metabolism, carbohydrate metabolism, and the tricarboxylic acid (TCA) cycle (Fig. 2h).
Mitochondrial dysfunction in macrophages affects the osteogenic differentiation of MSCs
Macrophages are activated and polarized to M1-like (usually pro-inflammatory) or M2-like (usually anti-inflammatory) states in response to environmental factors3,7. We found that in the OVX osteoporosis mice, the percentage of CD80+F4/80+ M1 phenotype macrophages was significantly increased compared to that of sham mice (Fig. 3a). Furthermore, the percentage of CD206+F4/80+ M2 phenotype macrophages was significantly reduced by OVX, indicating a transition to the M1 phenotype (Fig. 3b). To further validate the phenotypic alterations in macrophages, we analyzed pro-inflammatory gene expression in BMDMs using qPCR. Inducible nitric oxide synthase (Inos), IL-1β, TNF-α, and CD80 expression in OVX macrophages was higher than those in sham macrophages (Supplementary Fig. 3a-d). Furthermore, we found that endotoxin levels in the serum of OVX mice were higher than those in serum of sham mice (Supplementary Fig. 3e). These results demonstrate that macrophages are in a state of inflammatory activation when osteoporosis occurs.
We used a lipopolysaccharide (LPS)-induced bone loss model to explore whether higher endotoxin levels activate and polarize macrophages during osteoporosis. A micro-CT analysis confirmed substantial trabecular bone loss in LPS-treated mice compared to that in control mice (Fig. 3c-h). In LPS-treated mice, macrophages exhibited phenotypes similar to those in OVX mice, confirming that endotoxin levels affect macrophage polarization and metabolic status (Supplementary Fig. 3f-i).
Various genetic and chemical stresses trigger increased mitochondrial exchange in various cellular contexts. Inflammatory signaling receptor cells induced by LPS and TNF-α treatment are strong promoters of mitochondrial exchange16,28. The flow cytometric analysis showed that more mitochondria from OVX-derived macrophages were transferred to MSCs in the OVS group than those in the sham group (Fig. 3i, j). More mitochondrial transfer was also found by co-culturing RAW 264.7 cells (untreated or treated with LPS) and MSCs (Fig. 3k, l).
Next, we investigated whether mitochondria transferred from macrophages to MSCs regulate the osteogenic differentiation of MSCs. Indeed, MSCs that received normally functioning mitochondria showed an enhanced osteogenic phenotype compared with the MSC-alone group (Supplementary Fig. 4). In contrast, MSCs receiving dysfunctional mitochondria (OVX or LPS) showed impaired osteogenic capacity, such as decreased alkaline phosphatase (ALP) levels, decreased alizarin red-stained mineralized nodules, and decreased osteogenic gene expression (Supplementary Fig. 4). Then, we clarified how altered osteogenic function is influenced by mitochondrial transfer from macrophages. We extracted intact macrophage mitochondria and analyzed voltage-dependent anion channel (Vdac), cytochrome c oxidase subunit IV isoform 1 (COXIV), and translocase of outer mitochondrial membrane 20 (Tom20) expression (Fig. 3m, n). Mineralization was induced in MSCs alone, MSCs cultured with normal mitochondria, and MSCs cultured with dysfunctional mitochondria (LPS treatment). These results confirm that dysfunctional mitochondria directly affected osteogenesis, showing lower ALP levels, fewer mineralized nodules, and decreased osteogenic gene expression (Fig. 3o, p).
Macrophages and derived mitochondria affect the metabolic status of MSCs
Energy metabolism in MSCs is closely related to osteoporosis occurrence and progression3. We investigated how mitochondrial transport from macrophages to MSCs affects the MSC metabolic state by measuring mtROS and MMP levels in MSCs. Mitochondria from LPS-treated macrophages led to increased ROS levels in MSCs, as measured by MitoSOX and DCFH-DA assays (Fig. 4a, c). Furthermore, the MMP of MSCs was significantly decreased (Fig. 4b, d).
We measured the steady-state ATP levels in MSCs to explore potential changes in bioenergetics (Fig. 4e). Increased ROS production and dysregulated ATP metabolism in MSCs demonstrated impaired mitochondrial function and reduced electron transport chain efficiency. Subsequently, we studied the impact of mitochondrial transfer on MSCs by metabolic profiling using the Seahorse platform. As previously described, we assessed basal respiration, proton leakage, and non-mitochondrial respiration29. Exposure to untreated macrophages enhanced respiration in MSCs. This transfer-enhanced respiration was abolished when macrophages were pretreated with LPS (Fig. 4f, g). Furthermore, we estimated the cellular glycolytic activity by assessing the extracellular acidification rate (ECAR) in MSCs. An increased ECAR level was observed after MSCs were incubated with LPS-treated macrophages (Fig. 4h, i). This metabolic shift to glycolysis was further supported by elevated glucose uptake (Fig. 4j) and increased cellular lactate levels (Fig. 4k). Finally, MSCs were incubated with mitochondria from untreated and LPS-treated macrophages. We observed that LPS-treated mitochondria promoted glycolysis (Fig. 5l, m) and reduced oxygen consumption (Fig. 5n, o).
Mitochondrial transfer causes abnormal succinate accumulation in MSCs
To investigate the molecular basis of metabolic reprogramming of MSCs by mitochondrial transfer, we analyzed the MSC transcriptome. We found that pro-inflammatory gene expression (including IL-6 and IL-1β) were increased in the dysfunctional mitochondrial transfer group compared with that in the untreated macrophage and MSC groups (Fig. 5a). Interestingly, glycolytic genes were also upregulated (Fig. 5b), consistent with the increased ECAR (Fig. 4H and I). We then performed GO and KEGG enrichment analyses on the RNA-seq data. GO analysis showed that the MSCs in the dysfunctional mitochondrial transfer group showed changes in ROS metabolic process, ATP metabolic process, and mitochondrial organization (Fig. 5c). Furthermore, we found that the genes involved in the regulation of inflammatory responses related to inflammatory activation and osteoclast differentiation in bone remodeling-related systems also changed (Fig. 5c).
Subsequently, we labeled MSCs with 13C6-glucose and used mass spectrometry to trace the fate of glucose isotopes (Fig. 5e). We found that t the flux of glucose metabolism, was significantly altered in MSCs after dysfunctional mitochondrial transfer. Furthermore, in the dysfunctional mitochondrial transfer group, succinate, a TCA cycle intermediate, was elevated (Supplementary Fig. 5a). We observed succinate accumulation in MSCs in the osteoporosis-induced mice compared to that in mice in the sham and PBS groups (Fig. 5f, g). In BMDMs and serum of osteoporosis model mice, the succinate content also increased (Supplementary Fig. 5b-e). Interestingly, macrophages pretreated with succinate (5 mM) also influenced the succinate levels in MSCs (Fig. 5h). This abnormal succinate accumulation in MSCs of the dysfunctional mitochondrial transfer group may be related to the decreased activity of oxidases, such as succinate dehydrogenase (SDH) (Supplementary Fig. 5f).
Hif-1α plays a key role in glucose metabolism and bone resorption, and its activation is regulated by increased succinate levels26,30,31. Moreover, a sustained increase in succinate induces pro-inflammatory cytokine expression (such as IL-1β) via Hif-1α26,32. Accordingly, Hif-1α gene and protein expression in MSCs in the presence of LPS-treated macrophages were higher than those in the MSCs and untreated groups (Fig. 5i-k). MSCs in the dysfunctional mitochondrial transfer group were enriched in inflammatory and metabolism-related pathways such as the TNFα signaling pathway, NF-κB signaling pathway, oxidative phosphorylation, Hif-1α signaling pathway, and Toll-like signaling pathway. In addition, after pretreating macrophages with succinic acid, MSCs showed Hif-1α activation (Fig. 5l-m), upregulated glycolytic genes such as hexokinase 2 (Hk2) and solute carrier family 2 member 1 (Slc2a1), and upregulated pro-inflammatory genes such as IL-1β and TNF-α compared to those in the blank control group, (Supplementary Fig. 5g-p). These data suggest that succinate may directly regulate Hif-1α signaling and proinflammatory cytokine release, thereby affecting the osteogenic differentiation of MSCs.
Mitochondrial transfer maintains bone health in vivo
To examine the effect of dysfunctional mitochondria on skeletal health, we administered PBS and LPS-treated macrophage mitochondria [mito(LPS)] via the tail vein of 8-week-old mice twice a week for 4 weeks and analyzed their skeletal phenotype (Fig. 6a). Consistent with the histological data, a micro-CT analysis revealed lower trabecular volume and higher trabecular separation in the mito (LPS-treated) group than in the PBS group (Fig. 6b-h). We measured osteoclast bone resorption in mito(LPS) mice versus control mice and found that mito(LPS)-treated mice had more tartrate-resistant acid phosphatase (TRAP) + osteoclasts on the trabecular bone surface than in PBS-treated controls (Fig. 6i, g). Taken together, these results suggest that dysfunctional mitochondria disrupt skeletal homeostasis by reducing bone turnover.
Finally, we tested the effects of mitochondrial transfer on osteoporosis in vivo. We treated OVX mice by injecting mitochondria twice a week via the tail vein. von Kossa staining and micro-CT analysis of the femur showed that mice treated with mitochondria displayed higher trabecular bone mass than the OVX group (Fig. 6k-q). Furthermore, the TRAP + osteoclasts were significantly reduced in OVX + mito-treated mice compared to those in OVX-treated mice (Fig. 6r, s).