MACF1 condensates as molecular rheostats in bone mechanotransduction

How bone cells decode mechanical cues

Bone maintains its mass partly by sensing mechanical load and converting it into cellular responses, a process called mechanotransduction. The paper asks a specific gap question: how do bone cells turn a mechanical cue into a sustained adaptive response rather than a transient signal? Its answer centers on MACF1, a cytoskeletal crosslinker that is also a core component of stress granules (SGs), the RNA-protein assemblies cells form under stress.

The authors propose that MACF1 does not merely pass force along the cytoskeleton. Instead, it condenses into liquid-like droplets through a disordered segment, IDR2, and uses those droplets to hold signaling proteins in a protected state. The two conditions used to mimic reduced mechanical input are mechanical unloading (MU) and culture on a soft matrix (Soft), each compared with its own reference: control (Ctrl) for MU and a stiff matrix (Stiff) for Soft.

IDR2 is the segment that drives condensation

The C-terminal region of MACF1 contains a GAR domain and an intrinsically disordered region (IDR) split into IDR1, IDR2 and IDR3. Disordered regions lack a fixed structure, which lets them engage in many weak interactions and, in some proteins, separate into a dense phase distinct from the surrounding solution. The paper tested this directly with purified mCherry-tagged MACF1-GAR and MACF1-IDR: phase diagrams across protein and NaCl concentrations showed the IDR forming droplets, and fluorescence recovery after photobleaching (FRAP) confirmed that the droplets exchange material, the hallmark of a liquid-like condensate.

Specificity came from subregion tests. In molecular dynamics simulations, the IDR2-G3BP1 complex stabilized at a radius of gyration of about 1.8 nm, whereas IDR1 failed to form a stable complex. IDR2 also showed stronger electrostatic attraction to G3BP1, with a Coulombic interaction energy near -500 kJ/mol versus about -300 kJ/mol for IDR1 and IDR3, and formed roughly 8 interfacial hydrogen bonds versus about 4 each for the other subregions. Functionally, reintroducing IDR2 into MACF1ΔIDR cells restored SG assembly under MU or Soft, while IDR1 and IDR3 did not.

Condensates recruit mechanosensitive cargo into stress granules

To connect condensation to signaling, the authors isolated SG cores from wild-type and MACF1ΔIDR cells under Ctrl, MU, Stiff and Soft, then analyzed their contents by mass spectrometry. In wild-type cells, MU versus Ctrl and Soft versus Stiff changed the SG-associated protein set and enriched focal-adhesion and Wnt-pathway categories. Comparing wild type with MACF1ΔIDR under MU or Soft identified proteins whose SG recruitment depended on the IDR.

Imaging and interaction assays narrowed this to three effectors: the mechanosensitive channel PIEZO1, the focal-adhesion protein paxillin (PXN), and β-catenin (CTNNB1). In wild-type cells these proteins colocalized with the SG marker G3BP1 under MU or Soft, and the proportion of protein-positive SGs fell in MACF1ΔIDR cells. The reported wild-type versus MACF1ΔIDR comparisons were P = 0.0005 (MU) and 0.04 (Soft) for PIEZO1, P = 0.01 and 0.001 for PXN, and P = 0.005 and 0.001 for CTNNB1, with 60 cells quantified per condition across three biologically independent experiments. Co-immunoprecipitation and proximity ligation assays further supported MACF1 association with G3BP1 and with these cargo proteins under unloading.

Condensation, not binding alone, protects the cargo

A central test separated protein binding from droplet formation. The IDR2-Mut3 variant (R19K/R32K/F52L) retained proximity to G3BP1, PIEZO1, CTNNB1 and PXN but failed to form condensates, enrich these proteins from cell lysates, or restore their SG localization. That result argues that physical association is insufficient; the condensed state itself is required.

The paper then traced what protection means. In the MU time course, Piezo1, Pxn and Ctnnb1 mRNA levels were unchanged from 0 to 8 h but fell significantly at 16 h, while cycloheximide chase assays showed earlier protein destabilization at 8 h in MACF1ΔIDR cells and after pharmacological inhibition of SG assembly. Because transcripts were still stable at 8 h, the authors assessed ubiquitination and proteasome dependence at that time point, avoiding transcriptional confounding. Inhibiting SG formation with ISRIB likewise reduced PIEZO1, PXN and CTNNB1 abundance and lowered osteogenic outputs, including RUNX2 and COL1A1 expression and ALP staining.

A disease-linked variant adds a second line of evidence: R5353H, an arginine substitution associated with short stature, disrupted stable IDR2-G3BP1 complex formation in simulations and impaired IDR2 condensation and SG localization in cells.

Bone-targeted IDR2 reduces unloading-induced bone loss

The functional test used hindlimb unloading (HLU) in 3-month-old male mice, with a bone-targeting AAV vector encoding MACF1-IDR-FL or IDR2 delivered by intra-articular injection. GFP signal confirmed bone targeting with no detectable signal in major organs. Treated mice showed denser, more aligned trabeculae, increased osteoid deposition, and restored mineral apposition rate, which had been reduced by HLU.

Micro-CT showed significantly higher trabecular bone mineral density and cortical thickness in both femur and tibia, along with higher trabecular number and thickness and a reduced Structure Model Index. Three-point bending tests showed increased maximum load, stiffness and elastic modulus in treated tibiae. At the molecular level, unloading reduced PIEZO1, PXN, CTNNB1 and RUNX2, and IDR2 or IDR-FL treatment substantially rescued these levels, supported by immunoblotting of femoral lysates.

Where the rheostat model still needs testing

The authors describe MACF1 condensates as molecular rheostats that buffer and modulate stress signals, and they place this alongside membrane-proximal views of bone mechanotransduction centered on integrin activation and cytoskeletal tension. The evidence for the core claim is strongest in osteoblast-lineage cells, with supporting observations in osteocytes and chondrocytes under MU, Soft or fluid shear stress.

Several boundaries matter for interpretation. The proteomics comparison used n = 1 biological sample per genotype-condition combination, so those protein-set differences are hypothesis-generating rather than statistically replicated. Ubiquitination was not directly measured at 16 h, and the authors note that mRNA declines at that time would confound protein-turnover interpretation, which is why the mechanistic assays focused on 8 h. The in vivo protection was shown in male mice only. The authors also report that IDR deletion impaired osteogenic differentiation even under basal conditions without prominent SG assembly, and they speculate that MACF1-IDR may additionally support microtubule stability; that SG-independent layer remains untested here.

Why the separation-of-function mutant matters most

The most decision-relevant result is not that MACF1 binds PIEZO1, PXN and CTNNB1, but that the IDR2-Mut3 mutant keeps those interactions while losing condensation and fails to restore SG localization. That comparison makes condensate formation, rather than protein-protein contact, the variable that predicts whether mechanosensitive cargo is protected. For readers evaluating condensate-based mechanisms, the practical threshold is therefore whether a candidate construct forms droplets and enriches cargo, not merely whether it co-immunoprecipitates.

The in vivo result follows the same logic: delivering the IDR2 segment alone was sufficient to improve bone structure and strength after unloading, tying the protective effect to the condensation-competent region. The open question the paper leaves is how mechanical input is converted into MACF1 condensation in the first place; the authors state that the proximal regulators coupling mechanical cues to MACF1 condensation remain to be identified.

Sources used in this answer