Literature searches were conducted in PubMed (https://pubmed.ncbi.nlm.nih.gov/) and Web of Science (https://www.webofscience.com/) from database inception through to August 15, 2026. Search terms combined ‘mitochondria-associated endoplasmic reticulum membrane(s)’, ‘MAM(s)’, ‘ER-mitochondrial contact(s)’ and ‘MERC(s)’ with ‘osteoporosis’, ‘intervertebral disc degeneration’, ‘osteoarthritis’, ‘sarcopenia’ and ‘skeletal muscle atrophy’. Peer-reviewed studies examining MAM structure, MAM-associated molecules or ER-mitochondrial communication in relevant cellular, animal or human tissue models were prioritized. Studies outside the musculoskeletal scope, duplicate reports and articles with only peripheral relevance to ER-mitochondrial communication were excluded. Reference lists of key articles were also screened for additional relevant studies.

MAMs are dynamic ER membrane domains located at sites of close apposition to mitochondria. Their extent, spacing and molecular composition vary across cell types and physiological states (25–30). ER-mitochondrial contacts are organized by multiple tethering and regulatory systems, including mitofusin 2 (MFN2), vesicle-associated membrane protein-associated protein B-protein tyrosine phosphatase-interacting protein 51, mitochondrial fission 1 protein-B-cell receptor-associated protein 31 and PDZ domain-containing protein 8 (31–37), while the inositol 1,4,5-trisphosphate receptor (IP3R)-glucose-regulated protein 75 (GRP75)-voltage-dependent anion channel 1 (VDAC1) complex provides a major route for Ca²⁺ transfer from the ER to mitochondria (38,39).

Ca²⁺ transfer at these interfaces couples intracellular signaling to mitochondrial metabolism and stress responses (19,38,39). Excessive transfer can produce mitochondrial Ca²⁺ overload, oxidative stress, loss of membrane potential and cell injury (19,40). MAMs also support non-vesicular phospholipid exchange between the two organelles. Phosphatidylserine synthesized in the ER is transferred to mitochondria for conversion to phosphatidylethanolamine, while oxysterol-binding protein-related proteins 5 and 8 and extended synaptotagmin 1 participate in lipid transfer and the maintenance of mitochondrial membrane composition and respiration (41–45). Direct evidence linking this lipid-transfer machinery to musculoskeletal degeneration remains sparse. The major structural and signaling modules of MAMs are summarized in Fig. 1.

ER-mitochondrial contacts also participate in mitochondrial fission, mitophagy and stress signaling. Dynamin-related protein 1 (DRP1), phosphofurin acidic cluster sorting protein 2 (PACS-2)/receptor accessory protein 5 and FUN14 domain-containing protein 1 link contact-site organization to mitochondrial dynamics and quality control (46–50), whereas Sigma-1 receptor (Sig-1R), inositol-requiring enzyme 1α (IRE1α)/mitochondrial ubiquitin ligase and protein kinase RNA-like endoplasmic reticulum kinase (PERK)/MFN2 connect the interface to ER stress responses (51–53). Autophagosome formation can initiate at ER-mitochondrial contacts, and disturbances at these sites have also been linked to apoptosis and ferroptosis (54–56).

ER-mitochondrial stress is also coupled to inflammasome activation. During NLR family pyrin domain containing 3 (NLRP3) activation, NLRP3 and the adaptor apoptosis-associated speck-like protein containing a CARD redistribute toward ER-mitochondrial regions, while mitochondrial reactive oxygen species (ROS) and Ca²⁺ signaling contribute to inflammasome assembly and activation (57,58). Direct evidence linking MAMs to NLRP3 activation in degenerative musculoskeletal tissues remains limited. The core MAM-associated modules relevant to musculoskeletal degeneration are summarized in Table I (59–62).

Evidence for MAM involvement varies depending on how directly ER-mitochondrial contacts are examined. Structural or biochemical assessment of contact sites by ultrastructural imaging, MAM fractionation or contact-site proteomics provides the most direct evidence, particularly when combined with perturbation of tethering or contact-site regulators and corresponding phenotypic changes (13,14,18,25,63). Studies of MAM-resident or tethering proteins also provide mechanistic evidence, but changes in these proteins do not establish contact-site remodeling unless MAM architecture is measured. Other studies infer MAM involvement from ER-to-mitochondrial Ca²⁺ transfer, mitochondrial Ca²⁺ overload, ER stress or mitochondrial dysfunction (15,19–21). These processes are closely linked to the ER-mitochondrial interface but are not specific measures of MAM structure.

Disease-specific causal evidence is most developed in intervertebral disc degeneration (IVDD), where disrupted MAM integrity and phenotypic rescue have been linked to synaptojanin 2 binding protein (SYNJ2BP) and PACS-2 (14,18). In osteoporosis, the strongest data concern MFN2- and Sig-1R-dependent regulation of osteoclast function, with additional evidence from osteoblast and vascular models (15,64–67). OA studies now include direct assessment of abnormal ER-mitochondrial contacts (17,68), although the number of mechanistic studies remains small. Aging skeletal muscle has been examined using ultrastructural and proteomic approaches (13,69,70); evidence specific to clinical sarcopenia, however, still largely comes from aging, hereditary myopathy and metabolic models (21–24,71–74).

Osteoporosis is a systemic skeletal disorder characterized by low bone mass and deterioration of bone microarchitecture, resulting in increased bone fragility and fracture risk (8). Current MAM-related evidence is concentrated in osteoclast biology, with additional findings from osteoblast-lineage cells and the bone vascular microenvironment.

Jung et al (15) showed that MFN2 promotes receptor activator of nuclear factor-κB ligand (RANKL)-induced osteoclast differentiation through the Ca²⁺-calcineurin-nuclear factor of activated T cells 1 (NFATc1) axis. Ballard et al (65) further showed that its tethering function is required for this effect: Mfn2 deficiency impaired store-operated Ca²⁺ entry, reduced NFATc1 activation and increased bone mass. Wei et al (64) identified a related mechanism involving the MAM-localized Sig-1R, which suppresses osteoclastogenesis by promoting ER-associated degradation of sarco/endoplasmic reticulum Ca²⁺-ATPase 2 (SERCA2); deficiency of the gene encoding Sig-1R (Sigmar1) aggravated bone loss in ovariectomized mice. PTEN-induced kinase 1 (PINK1) has also been linked to MAM-dependent Ca²⁺ signaling in osteoclasts, although these data were obtained in periodontitis-associated bone loss rather than osteoporosis (75).

In osteolineage cells, Mfn2 deletion produced a different skeletal phenotype. Osteolineage-specific Mfn2 depletion increased cortical bone formation in female mice and enhanced oxygen consumption and mineralization during early osteogenic differentiation (16). The contrasting effects of MFN2 in osteoclasts and osteolineage cells argue against a uniform skeletal role across bone-cell compartments. In diabetic osteopenia, BK channel activation preserved osteoblast function and reduced bone loss through regulation of mitochondrial Ca²⁺ and solute carrier family 25 member 5/adenine nucleotide translocator 2-PINK1-parkin RBR E3 ubiquitin protein ligase-mediated mitophagy (66). ER-mitochondrial contact remodeling, however, was not directly examined in this study.

MAM-associated regulation also extends to the bone vascular compartment. Wang et al (67) reported that eldecalcitol (ED-71) preserved mitochondrial Ca²⁺ homeostasis in type H vascular endothelial cells, reduced glucocorticoid-induced senescence and improved angiogenesis-osteogenesis coupling. Mechanistic support in osteoporosis is currently strongest in osteoclasts. Direct structural evidence of MAM remodeling remains sparse and several supporting studies derive from periodontitis-, diabetes- or glucocorticoid-associated bone-loss models.

IVDD is characterized by progressive loss of nucleus pulposus cell homeostasis, extracellular matrix breakdown, cellular senescence and cell death. Evidence for MAM involvement includes both altered ER-mitochondrial ion transfer and direct changes in contact-site integrity.

In nucleus pulposus cells exposed to excessive mechanical compression, Lin et al (20) found increased ER stress and ER-mitochondrial Ca²⁺ transfer. Inhibition of ER stress or blockade of the IP3R-GRP75-VDAC1 axis reduced mitochondrial Ca²⁺ overload and poly(ADP-ribose) polymerase-apoptosis-inducing factor-associated programmed necrosis. Zheng et al (76) showed that site-1 protease deficiency caused ER distension and increased ER-mitochondrial contacts, while disrupting ER-to-mitochondrial Ca²⁺ transfer and accelerating nucleus pulposus cell senescence and disc degeneration.

Song et al (14) observed disrupted MAM integrity in degenerative disc tissue and tert-butyl hydroperoxide-treated nucleus pulposus cells. SYNJ2BP deficiency was associated with mitochondrial Zn²⁺ overload and cellular senescence, whereas SYNJ2BP overexpression promoted MAM formation, stabilized the NLR family member X1 (NLRX1)-solute carrier family 39 member 7 (SLC39A7) complex, restored mitochondrial Zn²⁺ homeostasis and attenuated disc degeneration.

Kang et al (18) reported impaired MAM integrity and reduced PACS-2 expression in degenerative human and rat disc tissues and in nucleus pulposus-derived stem cells exposed to an acidic microenvironment. PACS-2 preserved ER-mitochondrial contacts through the specificity protein 1 (SP1)/leucine-rich repeat kinase 2 (LRRK2)/MFN2 axis, reduced ER stress and mitochondrial dysfunction and limited stem-cell apoptosis. Loss of PACS-2 aggravated disc degeneration, whereas its restoration improved the reparative effect of transplanted cells. Human degenerative disc tissues support the presence of MAM disruption, while causal rescue has so far been demonstrated mainly in experimental models.

OA is characterized by progressive articular cartilage degeneration and impaired chondrocyte homeostasis. Studies in OA have identified MAM-related transcriptional signatures and experimentally linked ER-mitochondrial communication to chondrocyte dysfunction (17,68,77,78).

Li et al (77) identified MAM-related genes associated with OA and cellular senescence, including protein tyrosine phosphatase non-receptor type 1 and inositol 1,4,5-trisphosphate receptor type 1. ER-mitochondrial contacts were not directly assessed in this study. Hou et al (17) showed that loss of MFN2 impaired mitochondrial function and cartilage matrix metabolism, with fewer mitochondrial-ER junctions also observed in damaged human cartilage. Sirtuin 3 stabilized MFN2, while MFN2-mediated mitochondrial-ER junctions supported Ca²⁺ homeostasis and reduced chondrocyte senescence. Intra-articular delivery of MFN2 mRNA attenuated cartilage degeneration in experimental OA models.

ER-mitochondrial Ca²⁺ transfer has also been linked to chondrocyte apoptosis. In temporomandibular joint OA, optineurin deficiency enhanced Ca²⁺ transfer through the IP3R-GRP75-VDAC1 complex, increased mitochondrial Ca²⁺ loading and promoted apoptosis (78). Pharmacological inhibition of this pathway reduced Ca²⁺ overload and apoptosis.

Song et al (68) showed that inflammatory stress increased aberrant ER-mitochondrial contacts in chondrocytes. Circular RNA ZFP609 (cZFP609) stabilized oligomeric binding immunoglobulin protein (BiP), attenuated IRE1α-associated ER stress, reduced aberrant ER-mitochondrial contacts and suppressed lipid peroxidation and ferroptosis. Intra-articular cZFP609 delivery also reduced cartilage degeneration in an experimental OA model. Human cartilage provides structural evidence of altered ER-mitochondrial contacts, whereas causal rescue and therapeutic testing remain predominantly preclinical.

Sarcopenia is an age-related skeletal muscle disorder characterized by progressive loss of muscle mass and function. Most structural evidence concerning the ER-mitochondrial interface comes from aging skeletal muscle rather than clinically defined sarcopenia.

Lu et al (13) identified age-related changes in MAM ultrastructure and protein composition in striated muscle using electron microscopy and proteomic profiling. Allen et al (69) likewise detected early alterations in ER-mitochondrial contacts and MAM protein composition in aging skeletal muscle, changes that were prevented by exercise. Unten et al (70) reported increased mitochondria-ER contact sites (MERCs) in myoblasts from older human donors. MERCs denote the physical contacts between the two organelles, whereas MAMs more specifically refer to the membrane domains and molecular components associated with these interfaces.

Ca²⁺ transfer at these contacts has also been linked to age-related muscle atrophy. Shi et al (21) showed that age-related reduction of thyroid hormone receptor α increased IP3R type 1-mediated Ca²⁺ transfer and MAM formation, leading to mitochondrial Ca²⁺ overload, mitochondrial dysfunction and skeletal muscle atrophy in mice. Separately, Grepper et al (71) localized BCL2-like 13 to ER-mitochondrial contact sites and showed that its loss altered Ca²⁺ dynamics and impaired skeletal muscle function, without changing the contact-site number.

Additional mechanisms have been identified in hereditary and metabolic muscle models. Selenoprotein N-related myopathy is associated with defective ER-mitochondrial contacts and impaired bioenergetics (72), whereas OPA1 mitochondrial dynamin-like GTPase deficiency increases ER-mitochondrial tethering through an activating transcription factor 4-dependent response (22). Metabolic studies have reported both reduced ER-mitochondrial coupling and excessive MAM formation in insulin-resistant skeletal muscle (23,24,74). Clinical validation in well-characterized sarcopenia populations remains limited.

Recurrent abnormalities in Ca²⁺ signaling, mitochondrial function and ER stress occur across the disorders discussed above, but they do not imply a uniform MAM defect. ER-to-mitochondrial Ca²⁺ transfer has been directly interrogated in several disease models (20,21,78), whereas mitochondrial dysfunction or ER stress alone cannot establish altered contact-site organization. The common molecular pattern therefore reflects convergence on organelle stress pathways more clearly than a uniform structural defect at the ER-mitochondrial interface.

The direction of MAM remodeling also varies by cellular context. MFN2 promotes osteoclast differentiation through Ca²⁺-dependent signaling, whereas osteolineage-specific Mfn2 depletion enhances cortical bone formation (15,16,65). In skeletal muscle, both loss of contact integrity and stress-associated increases in ER-mitochondrial tethering have been reported (13,22–24), while endurance exercise can reduce ER-mitochondrial contacts (79). Contact abundance alone is therefore an inadequate marker of dysfunction. These observations point to cell type, metabolic state and disease stage as key determinants of whether contact remodeling is adaptive or maladaptive.

Translational evidence remains uneven across disorders. Direct structural and causal evidence is concentrated in IVDD and selected OA models, whereas osteoporosis and skeletal muscle studies rely more heavily on molecular or functional readouts and preclinical systems. Human interventional validation remains limited, particularly in clinically defined sarcopenia.

Taken together, these observations support a disease-oriented framework in which distinct MAM-associated processes converge on disrupted MAM homeostasis and shared downstream consequences, while giving rise to tissue-specific outcomes across osteoporosis, intervertebral disc degeneration, OA and sarcopenia/skeletal muscle atrophy (Fig. 2). A comparative summary of disease-specific evidence strength, directness and translational status is provided in Table II.

MAM organization can be modified through contact-site regulators, membrane composition and MAM-resident proteins. Genetic restoration of contact-site regulators has been examined in IVDD models. SYNJ2BP overexpression promoted MAM formation and NLRX1-SLC39A7 complex assembly, restored mitochondrial Zn²⁺ homeostasis and reduced nucleus pulposus cell senescence and disc degeneration (14). PACS-2 restoration similarly preserved MAM integrity through the SP1/LRRK2/MFN2 pathway, reduced apoptosis of nucleus pulposus-derived stem cells and improved the reparative efficacy of transplanted cells (18).

Contact organization can also be altered without directly manipulating classical tethering proteins. In OA models, cZFP609 stabilized oligomeric BiP, attenuated IRE1α-associated ER stress, reduced aberrant ER-mitochondrial contacts, and suppressed chondrocyte ferroptosis and cartilage degeneration (68). In palmitic acid-exposed nucleus pulposus cells, phosphatidylcholine increased ER-mitochondrial interactions, restored mitochondrial phosphatidylcholine content and fatty-acid oxidation and alleviated lipotoxic injury (80).

Pharmacological modulation of a MAM-resident protein has been demonstrated in experimental bone-loss models. Sig-1R activation suppressed osteoclastogenesis by promoting ER-associated degradation of SERCA2, whereas Sigmar1 deficiency aggravated bone loss (64). Local Sigmar1 overexpression and treatment with the Sig-1R agonist dimemorfan preserved bone mass in experimental models.

Beyond direct manipulation of contact-site organization, pharmacological studies have largely targeted functional processes coordinated at the ER-mitochondrial interface, particularly Ca²⁺ handling and mitochondrial homeostasis. Ruthenium Red suppressed RANKL-induced ROS production and NFATc1 activation, inhibited osteoclastogenesis and reduced bone loss in ovariectomized mice (81). Mitochondria-targeted Ca²⁺-regulating nanoparticles reduced mitochondrial Ca²⁺ overload and pro-inflammatory macrophage polarization in experimental OA (82). In glucocorticoid-induced bone loss, ED-71 reduced GRP75-dependent MAM-mediated mitochondrial Ca²⁺ overload in type H vascular endothelial cells and attenuated cellular senescence (67).

Proteins controlling Ca²⁺ transfer and mitochondrial permeability have also been targeted in dystrophin-deficient muscle. MKT077 inhibited GRP75-dependent Ca²⁺ transfer from the sarcoplasmic reticulum to mitochondria, reduced mitochondrial Ca²⁺ overload and alleviated muscle pathology in mdx mice (83). Alisporivir improved mitochondrial Ca²⁺ handling and mitochondrial homeostasis in dystrophin-deficient muscle (84), while VBIT-4, a voltage-dependent anion channel inhibitor, reduced mitochondrial Ca²⁺ overload and partially improved muscle pathology in a severe Duchenne muscular dystrophy model (85). The relevance of these findings to sarcopenia remains indirect because the evidence derives from dystrophin-deficient muscle models.

Pharmacological evidence for MAM-linked ER stress signaling is less developed in degenerative musculoskeletal disease. PERK and IRE1α signaling pathways are closely associated with ER-mitochondrial stress communication, but most musculoskeletal studies have examined ER stress without directly assessing changes in MAM organization (86–88).

Beyond molecular and pharmacological interventions, ER-mitochondrial communication is also responsive to physiological and metabolic cues. Exercise provides the clearest example in skeletal muscle. Endurance exercise alters ER-mitochondrial contacts in a demand-dependent manner (79), and exercise prevented early age-related abnormalities in these contacts in aging muscle (69). In diabetic skeletal muscle, swimming also modified MAM-related ER stress and mitochondrial dysfunction (73). Exercise-related changes in ER-mitochondrial contacts therefore appear to form part of the broader metabolic adaptation of skeletal muscle.

Metabolic state itself can reshape the ER-mitochondrial interface. Reduced MAM integrity has been associated with skeletal muscle insulin resistance (23), whereas obesity-related activation of pyruvate dehydrogenase kinase 4 can increase ER-mitochondrial contact and impair insulin signaling (24). Muscle-specific p53 deletion likewise modifies MAM organization in diet-induced insulin resistance (74). A smaller number of bioactive interventions have been linked more directly to MAM-related functions. Polygonatum sibiricum polysaccharide improved skeletal muscle aging phenotypes by regulating MAM-mediated Ca²⁺ homeostasis (89). Representative experimental interventions targeting MAM structure or MAM-related functions are summarized in Table III.

A major barrier to the field is the lack of a standardized approach for defining MAM remodeling. Transmission electron microscopy and electron tomography provide ultrastructural information on contact distance, extent and three-dimensional organization, whereas proximity ligation assays assess molecular proximity and genetically encoded contact reporters enable dynamic monitoring of ER-mitochondrial interactions. MAM fractionation characterizes the biochemical composition of contact-site fractions, while live-cell Ca²⁺ imaging measures functional ER-to-mitochondrial signaling. These approaches interrogate different features of the interface and are not interchangeable. Changes in contact distance, abundance of a MAM-associated protein or Ca²⁺ transfer alone cannot define the structural and functional state of the contact site. Combining structural and functional measurements within the same experimental system would improve comparison across tissues, models and disease stages (25,63,90).

Molecular profiling can add disease- and cell-specific context to these measurements. MAM-enriched and proximity-labeling proteomics can resolve changes in contact-site composition, whereas single-cell and spatial transcriptomic approaches can localize MAM-associated pathways to defined cell populations and tissue regions. Public transcriptomic datasets from aging and diseased skeletal muscle may also help prioritize candidate regulators and examine their associations with age, disease severity or treatment response. Because transcriptomic signatures do not demonstrate physical ER-mitochondrial contacts, candidates identified through these datasets require structural or functional validation (13,63,77,90).

Human validation remains limited. Most causal studies rely on cultured cells and animal models, whereas human evidence is derived largely from cross-sectional tissue analyses (14,18,23,70). Well-phenotyped patient-derived tissues and primary cells, together with stage-stratified and, where feasible, longitudinal sampling, could help determine whether MAM alterations precede tissue degeneration or arise during disease progression. Integration with clinically relevant phenotypes will also be necessary to establish the disease relevance of contact-site abnormalities across bone, cartilage, intervertebral disc and skeletal muscle.

Therapeutic translation is further constrained by target specificity. Many interventions discussed above modify Ca²⁺ handling, mitochondrial permeability, ER stress or systemic metabolism without acting selectively at MAMs (67,81–85,89). Preclinical studies therefore need to demonstrate target engagement at the ER-mitochondrial interface rather than infer MAM correction solely from downstream mitochondrial or stress responses. Candidate interventions should link contact-site modulation to disease-relevant functional outcomes and ultimately be validated in human tissues or patient-derived systems.