Osteonecrosis of the femoral head is a post-translational disease that nobody can yet model in human tissue

What the work claims

This is a narrative review, not a primary experiment, so it should be weighted as a synthesis of other people's data. Its claim is that post-translational modifications (PTMs), the reversible chemical marks that switch proteins on or off after they are made, are the molecular switches that convert external insults, chiefly glucocorticoids and alcohol, into the cell death and lineage corruption that collapse the femoral head¹. The review organizes the evidence by cell type: bone marrow mesenchymal stem cells (BMSCs), osteoblasts, osteocytes, osteoclasts and vascular endothelial cells, and by modification type: phosphorylation, acetylation, methylation and ubiquitination.

The stakes are stated plainly. Non-traumatic osteonecrosis of the femoral head (ONFH) is one of the leading causes of hip disability in adults aged 20 to 50, and the review cites an approximately 70 percent rate of progression to total hip arthroplasty. There is no approved disease-modifying drug; the practical ceiling of current care is catching the disease early enough to decompress or rotate the bone mechanically. A mechanism map that pointed at druggable nodes would matter clinically, not just academically.

How it works

The review's core content is a dense catalog of specific, checkable mechanisms, and several are worth holding onto because they are concrete enough to be wrong. In alcohol-induced ONFH, ethanol impairs recruitment of Akt to the plasma membrane and suppresses its activating phosphorylation at Ser473, crippling the Akt/GSK-3beta/beta-catenin cascade and dropping the osteogenic transcription factors RUNX2 and osteocalcin in BMSCs. Dexamethasone works partly through an epigenetic relay: it inhibits the PI3K/Akt-UTX/EZH2 axis, lowering the histone demethylase UTX and raising the methyltransferase EZH2, which enriches the silencing mark H3K27me3 over osteogenic genes. Steroids also activate GSK-3beta by increasing its Tyr216 phosphorylation, driving beta-catenin into proteasomal degradation and biasing BMSCs toward adipogenesis over osteogenesis¹.

Two findings stand out for their specificity. A short peptide called RIP, encoded by the long non-coding RNA DGCR5, binds the N-terminal motif of RAC1, lowers RAC1-GTP and suppresses RAC1/PAK1, cutting beta-catenin Ser675 phosphorylation and pushing BMSCs toward fat; it is reported as highly expressed in BMSCs and femoral head tissue from patients with steroid-induced ONFH. And alcohol leaves an epigenetic memory: ethanol-induced HOTAIR suppresses miR-122, elevating PPARgamma, which recruits HDAC3 and SUV39H1 to the miR-122 promoter so the adipogenic program persists after alcohol withdrawal¹.

The cell-death chapters are equally specific. In osteoblasts, dexamethasone suppresses JAK2 and STAT3 phosphorylation, stripping the anti-ferroptotic defenses GPX4 and SOD2 and raising lipid peroxidation markers, so osteoblasts die by ferroptosis, an iron-dependent form of regulated death, alongside classical mitochondrial apoptosis. In endothelial cells, glucocorticoid receptor signaling activates c-Abl, phosphorylates caveolin-1 and suppresses ERK1/2 and eNOS, while a separate PERK/eIF2alpha/CHOP cascade executes endothelial apoptosis; this is the microvascular failure half of the disease. Ubiquitination supplies quality control failures: oxidative-stress-induced p53 inhibits Parkin-mediated mitophagy, and glucocorticoid promoter hypermethylation silences the E3 ligase NEDD4, starving bone marrow endothelial cells of VEGF and mTOR signaling¹.

Finally, the review tabulates sixteen small-molecule modulators, from icariin and salidroside to pravastatin, fludarabine, irisin and lithium chloride, nearly all acting by nudging phosphorylation nodes such as Akt, AMPK/mTOR or JAK2/STAT3 back toward survival. It makes an honest and important caveat: none of these compounds directly inhibits or activates a PTM-writing enzyme; they modulate signaling around the marks, and most were tested in rodents¹.

Where a skeptic should push

The single most load-bearing assumption is that the modification directions observed in rat steroid models and single-cell-type cultures hold in human femoral head tissue at the same nodes. The review's own summary table contains the tell: Akt phosphorylation is suppressed by glucocorticoids in BMSCs, osteoblasts and endothelial cells, yet reported as elevated, and possibly pro-survival-turned-maladaptive, in osteocytes of the same disease. A pathway that points in opposite directions in neighboring cell types cannot be validated in any model that studies one cell type at a time, and almost everything cited here was measured one cell type at a time, in dexamethasone-treated rats or cultured cells.

Second, the therapeutic table is natural-product pharmacology. Icariin, crocin, astragaloside IV and dragon's blood resin are plausible leads with weak target-engagement evidence; demonstrating that they shift a phosphorylation readout in a rat is a long way from a drug that reaches the femoral head at exposure. Third, the review itself concedes what is missing: cell type-specific PTM codes under ONFH-inducing conditions have never been mapped comprehensively, PTM-targeting drug development has largely happened only in oncology, and emerging metabolic marks such as lactylation and succinylation, attractive precisely because alcohol and steroids are metabolic insults, are unexplored here. This is a competent map of a territory surveyed almost entirely through animal lenses.

What bone organoids could test that rodents cannot

The non-obvious implication is that ONFH is a nearly ideal organoid program disguised as an orthopedic backwater. It has a defined, human-relevant toxicant class (glucocorticoids are among the most prescribed drugs on earth, and osteonecrosis is their quiet dose-limiting toxicity), five interacting human cell types, a sharply druggable mechanism map, and a regulatory vacuum where a disease-modifying therapy would face little competition. The review ends by naming the missing tool: patient-derived femoral head organoid models for high-throughput screening and mechanism work. Nobody has one. The first credible perfused bone-marrow-vascular assembloid that reproduces the steroid-induced osteogenic-to-adipogenic lineage flip, the JAK2/STAT3 ferroptosis module and the endothelial caveolin-1 failure in parallel would instantly become the reference system for this literature.

There is a genuine threat underneath the opportunity, and it is methodological. A bone organoid without a vascular compartment models only half the disease: the review's mechanism is explicitly two-legged, ischemia plus cell death, and the ischemia leg lives in endothelial cells that avascular spheroid cultures do not contain. Worse, the osteocyte result warns that cell-type-averaged readouts will mislead; Akt-centric screening in a homogeneous BMSC organoid would call the osteocyte response wrong. A model that collapses the five cell types into one averaged phosphorylation signature would reproduce the exact error this review accidentally documents. For drug discovery built on organoids, ONFH is a case study in why the next credibility hurdle is multi-lineage fidelity, not throughput: the field does not need another 384-well assay of one cell type, it needs a small human bone ecosystem in which the contradictions between cell types can actually be observed.

The bottom line

As a mechanism catalog, this review is useful and unusually specific. As evidence, it is a reminder that the entire PTM framework of a common human disease rests on rodent steroid models and isolated cells. What would confirm the framework is the experiment the review itself requests: patient-derived femoral head organoids in which glucocorticoid exposure reproduces the cell-type-resolved modification pattern, ideally with a perfused endothelium. What would break it is a human tissue study finding that the modification directions in actual necrotic femoral heads do not match the rat map, which remains entirely possible for the osteocyte branch. Until one of those happens, treat the sixteen-compound table as leads for a human model, not as a pipeline.