Metaphyseal chondrodysplasia type Schmid (MCDS), is a rare skeletal disorder caused by mutations in COL10A1, the gene encoding collagen X. This collagen is produced by hypertrophic chondrocytes, a specialized population of cells in the cartilage growth plate that normally helps drive bone growth.
In MCDS, mutant collagen X fails to fold properly and accumulates inside these cells, placing the endoplasmic reticulum, a compartment responsible for folding and processing secreted proteins, under considerable stress (ER-stress). Chronic stress can kill the cells; yet, that is not what happens. Instead, many hypertrophic chondrocytes survive, change their state, and disrupt the normal program of bone elongation. For years, our laboratory has been trying to understand how they do this.
A clue hidden in an earlier dataset
The hypoxia story began well before the current paper. In earlier work using the 13del transgenic MCDS mouse model, different layers of the growth plate were carefully dissected and analyzed by RNA sequencing. Among the many changes associated with ER stress, one signal stood out: genes and pathways related to hypoxia were broadly increased in the hypertrophic zone. It was intriguing, but also puzzling.
Cartilage is a relatively low-oxygenated tissue, so an altered hypoxia signature could have several explanations. In MCDS, the hypertrophic region of the growth plate becomes greatly enlarged. Perhaps the cells were simply farther away from the oxygen supply in the marrow space. Alternatively, the transcriptomic signal might represent a secondary stress response rather than environmental hypoxia. For some time, the RNA-seq hint remained unresolved.
When we returned to the question, we asked: could we visualize a hypoxic response in the mutant hypertrophic chondrocytes?
Seeing the signal directly
We first examined HIF1α and HIF2α, proteins that enable cells to respond and adapt to low oxygen. The staining pattern was striking. In normal growth plates, HIF proteins were largely absent from hypertrophic chondrocytes. In the MCDS mice, both HIF1α and HIF2α were detected throughout the expanded hypertrophic zone, providing in vivo evidence that the hypoxic pathway is biologically relevant. However, HIF accumulation alone was not conclusive, as HIF proteins can sometimes be stabilized by cellular stress without significant oxygen reduction. Therefore, we used EF5, a chemical probe that labels cells experiencing low oxygen in vivo.
Again, the mutant hypertrophic chondrocytes showed a clear signal, marking a key turning point. Now, we were no longer just observing hypoxia pathway activation; we had evidence supporting that the stressed hypertrophic chondrocytes were entering a hypoxic state. This immediately raises a new question: why?
Was hypoxia caused by a change in the tissue environment?
Having confirmed increased hypoxia, we tested several possible environmental explanations. First, the enlarged hypertrophic zone might simply distance cells further from the oxygen supply of the bone marrow. However, at E15.5, EF5 staining was already increased before significant growth plate expansion had developed. Second, we asked whether loss of normal collagen X from the extracellular matrix might impair oxygen diffusion. Yet Col10a1-null mice did not exhibit the same hypoxic response. Lastly, CD31 staining showed no obvious vascular abnormalities surrounding the growth plate.
That left us with a more intriguing possibility; perhaps the stressed hypertrophic chondrocytes were somehow creating an intrinsic hypoxic state.
Putting normal and mutant cells side by side
To test this, we generated chimeric mice containing a mixture of normal GFP-labelled cells and MCDS mutant cells in the same tissue. This was our most informative experiments. Within the same growth plate, and thus the same local environment, EF5 staining followed the mutant chondrocytes. Normal cells located adjacent to mutant cells remained EF5-negative, while neighboring mutant cells were strongly positive.
We observed the same striking pattern in the talus cartilage, where normal and mutant cells were intermixed in a “salt-and-pepper” pattern. This shifted our understanding of the phenotype: the low-oxygen state was not merely a property of an abnormal growth plate, but at least in large part, a cell-autonomous consequence of the stress in mutant chondrocytes.
Our data also linked this response to ATF4 expression, an important component of the unfolded protein response and the integrated stress response (ISR). ATF4 was increased in mutant hypertrophic chondrocytes, and experimentally elevated ATF4 in hypertrophic chondrocytes led to accumulation of HIF proteins, providing a link between ER stress, ATF4 and hypoxia/HIF signaling.
How ER stress creates this intracellular hypoxic state remains an open question. One possibility is that the high burden of folding and handling misfolded collagen changes mitochondrial activity and cellular oxygen consumption. The answer is likely to involve both metabolism and stress signaling pathways, and we hope future studies will clarify these mechanisms.
An experiment that did not make the final paper
One important question remained for us: was HIF activation merely a marker of stressed cells, or did it make a functional contribution to their survival? To address this, we conditionally targeted Hif1a in hypertrophic chondrocytes under ER stress. Strikingly, in the 13del transgenic background, Hif1a targeting was associated with a significant increase in TUNEL-positive hypertrophic chondrocytes, revealing a clear biological phenotype and providing functional support for a pro-survival role of HIF1α in this context.
The main limitation was that we could not demonstrate sufficiently complete loss of HIF1α protein by IHC in the targeted cells. Because IHC is not an absolute quantitative measure of protein abundance, we could not exclude the possibility that HIF1α levels had nevertheless been reduced to a biologically meaningful extent. However, without a more definitive measure of target reduction, we felt that the experiment could not be used as conclusive genetic evidence and therefore chose not to include it in the final paper.
For us, this experiment nevertheless remained an important part of the story. The emergence of a clear cell-death phenotype following Hif1a targeting suggested that HIF1α activation was unlikely to be merely a passive marker of ER-stressed hypertrophic chondrocytes, but may instead contribute functionally to their survival. Although the data were not definitive for inclusion in the final paper, it helped shape the direction of our subsequent questions.
From an unexplained signal to a broader stress-response principle
What began as an unexpected hypoxia signature in an earlier transcriptomic dataset evolved into a broader question about how stressed cells survive. In MCDS hypertrophic chondrocytes, our findings suggest that ER-stress can generate a cell-autonomous hypoxic state and engage HIF signaling as part of an adaptive survival response.
Traditionally, the relationship between hypoxia and ER stress has been viewed largely in the opposite direction: low oxygen disrupts protein folding, triggers the unfolded protein response, and helps cells adapt to hypoxic stress. Our study provides, to our knowledge, the first in vivo evidence that ER stress itself can trigger a hypoxic state.
This raises a broader question beyond MCDS. ER-stress is a central feature of many diseases, particularly those involving protein misfolding, high secretory demand, or chronic cellular stress. If ER stress is an early or primary event in these conditions, could a secondary hypoxic response also be part of the cellular adaptation? Whether this ER stress–hypoxia connection applies to other tissues and diseases remains an open question, one worth exploring.
For us, the HIF story reminds us that unexpected signals can reveal connections between stress pathways that extend beyond a single disease model and offer insight into fundamental cell biology.
Author contribution and AI image statement
This article was developed with valuable scientific input and comments from Professor Danny Chan, whose suggestions contributed to the refinement of the manuscript and interpretation of the work.
The header image was generated with the assistance of artificial intelligence based on the major findings and conceptual framework of this study. The image is intended as a visual summary of the work and does not represent additional experimental data.