BMPs are usually the go-to for bone formation, but in the aged marrow, they seem to behave unpredictably. While BMP2 and 7 stay osteogenic, BMP4 has been shown to induce p21-dependent senescence in older mesenchymal stem cells (MSCs). That’s a serious problem for therapy. Simply bumping up BMP ligands—a common move in placental-derived MSC transplantation—might accidentally speed up MSC exhaustion if the underlying signaling machinery is rewired.

I suspect the age-related shift from making bone to making fat isn't just about losing BMP activity. It looks more like a stoichiometric drift in how receptors and ligands interact. In this scenario, BMP ligands are sequestered by TGF-β receptors, specifically ALK5. This leads to a pro-senescent p-SMAD3 signal rather than the pro-osteogenic p-SMAD1/5/8 signal we’d expect.

The ALK5 Sequestration Hypothesis

Healthy MSCs use ALK2, 3, or 6 to move the SMAD1/5/8 pathway along. But aging changes the niche. We see upregulated TGF-β/SMAD3 signaling and a loss of SIRT6-mediated H3K27me3 repression. When SIRT6 stops keeping that repression in check, the resulting epigenetic collapse leads to an overabundance of ALK5 (TGFBR1) and its accessory proteins. This creates a signaling sink.

In this model, the lack of SIRT6 allows the ALK5 receptor pool to expand. Because BMP ligands like BMP4 and BMP7 have a moderate affinity for TGF-β type I receptors under certain conditions, the excess ALK5 in aged MSCs acts as a decoy. Instead of forming the usual osteogenic complex, BMP ligands are diverted into ALK5/TGFBR2 complexes. This triggers a "cross-talk" phosphorylation of SMAD3 instead of SMAD1.

This shift explains a few things that haven't quite added up before:

1. Senescence Induction: SMAD3 signaling is a known driver of p21 and p16 expression in senescent MSCs.
2. Lineage Plasticity: Dual activation of SMAD2/3 by BMP ligands could explain why Runx2+ preosteoblasts start expressing adipogenic markers, as SMAD3 can work with C/EBPβ to push lipid droplet formation.
3. Inflammatory Feed-forward: The Adipoq-lineage secretome likely produces ligands that keep ALK5 dominant, effectively locking the niche in a state where osteogenic BMPs are trapped in pro-fat pathways.

Falsifiability and Testing

We can test this by modulating the ALK5/ALK3 ratio in aged MSCs. If this hypothesis holds water, we should see the following:

• Ligand Switching: In aged MSCs, BMP4 treatment will result in a higher p-SMAD3:p-SMAD1 ratio than what we see in young MSCs.
• Rescue via Competitive Inhibition: Small-molecule inhibition of ALK5 (using something like SB431542) or shRNA-mediated knockdown of ALK5 in aged MSCs shouldn't just stop senescence; it should restore BMP4-induced p-SMAD1/5/8 translocation and bone formation.
• Epigenetic Restoration: Overexpressing SIRT6 should specifically downregulate ALK5 surface expression, "unplugging" the SMAD3 sink and restoring canonical BMP/SMAD signaling.

If blocking ALK5 doesn't restore p-SMAD1/5/8 phosphorylation after BMP stimulation, then the sequestration isn't happening at the receptor level. In that case, the block is likely further downstream, perhaps involving Smurf1-mediated degradation or defects in nuclear transport.