An unexpected structure in an avascular tissue
Cartilage presents a remarkable physiological challenge. Unlike most tissues, it contains no blood vessels. Chondrocytes therefore live in an environment in which oxygen must diffuse over considerable distances before reaching them.
While studying how these cells survive chronic hypoxia, we noticed unusual eosinophilic structures in the cytoplasm of chondrocytes. At first, their identity was unclear. Through laser microdissection, mass spectrometry, imaging, biochemical analyses and genetic models, we eventually arrived at an unexpected answer: chondrocytes themselves were producing large amounts of hemoglobin.
Even more surprisingly, this hemoglobin was not simply dispersed throughout the cytoplasm. It assembled into intracellular condensates that we termed hemoglobin bodies, or Hedys. These structures acted as local oxygen reservoirs, helping chondrocytes maintain oxygen homeostasis and survive in the naturally hypoxic environment of cartilage.
The finding challenged a deeply ingrained assumption. Hemoglobin did not necessarily belong exclusively to erythroid cells, and outside red blood cells it could acquire a function adapted to the needs of a completely different cell type.
That realization led us to a new question: if normal cells can repurpose hemoglobin to survive physiological hypoxia, might diseased cells do something similar under pathological stress?
A suspicious signal in liver cancer
The connection to cancer did not begin with a large-scale screen or a carefully designed hypothesis. It began with another visual observation.
While examining pathological sections of hepatocellular carcinoma, corresponding author Qiang Sun noticed hemoglobin staining inside a subset of tumor cells. The observation was intriguing, but it immediately raised an obvious concern: liver tumors are highly vascularized, and red blood cells contain enormous amounts of hemoglobin. Could the apparent signal in cancer cells simply reflect erythrocyte contamination, nonspecific staining, or hemoglobin taken up from the surrounding microenvironment?
This skepticism followed the work into peer review. One concern was particularly intuitive: even if hemoglobin could be detected in tumor cells, its abundance appeared far below the extraordinary concentration found in erythrocytes. Could such a comparatively modest amount really matter biologically?
We therefore went back to the tissue.
When we examined hepatocellular carcinoma tissue microarrays more systematically, an important feature emerged. Hemoglobin expression was heterogeneous, but in a subset of tumors the staining intensity within malignant cells approached that seen in nearby red blood cells. Together with molecular and genetic evidence, these observations argued that the signal could not be explained simply by blood contamination. At least some hepatocellular carcinoma cells were actively expressing substantial amounts of hemoglobin themselves.
The question then shifted from “Is hemoglobin really there?” to “Why would a cancer cell make it?”
The same molecule, but a different problem to solve
Our cartilage work had suggested one obvious possibility: oxygen.
Hepatocellular carcinoma, like many solid tumors, contains regions of severe hypoxia. We found that hypoxia activates a HIF1α–KLF4–hemoglobin axis, inducing hemoglobin expression in tumor cells.
But the function of hemoglobin in cancer turned out to be different from its role in cartilage.
In chondrocytes, hemoglobin helps manage limited oxygen availability. In hepatocellular carcinoma cells, its most striking effect was on oxidative stress. When hemoglobin expression was reduced, intracellular reactive oxygen species increased substantially.
This result initially seemed disproportionate to what might be expected from hemoglobin itself.
Hemoglobin is not generally regarded as a classical cellular antioxidant enzyme. Biochemical studies have suggested that its redox-associated catalytic activities are relatively modest compared with dedicated antioxidant enzymes. We therefore asked a simple functional question: how important is hemoglobin for cellular ROS control when compared directly with the canonical antioxidant machinery?
We individually depleted hemoglobin subunits and several classical antioxidant enzymes—including catalase (CAT), GPX1, GPX4 and members of the peroxiredoxin family (PRDX)—in Hep3B cells, and quantified intracellular ROS using DCF-DA flow cytometry.
The result surprised us.
Knockdown of hemoglobin produced an increase in intracellular ROS that was comparable in magnitude to that caused by depletion of several well-established antioxidant enzymes.
For us, this was an important turning point in the project. A protein best known as an oxygen carrier—and whose intrinsic antioxidant activity might appear modest in a biochemical assay—was exerting a quantitatively substantial effect on redox homeostasis inside a cancer cell.
It reminded us that the biological importance of a molecule cannot always be inferred from the catalytic efficiency of the purified protein alone. Cellular abundance, localization, interaction partners, substrate availability and the physicochemical environment can collectively transform what looks like a weak biochemical activity into a meaningful physiological function.
When a survival mechanism becomes a drug-resistance mechanism
This observation became particularly relevant when we considered therapy.
Sorafenib imposes considerable metabolic and oxidative stress on hepatocellular carcinoma cells. If tumor-cell hemoglobin buffered ROS, we reasoned that it might also help cancer cells survive treatment.
That is what we observed.
Reducing hemoglobin expression led to greater ROS accumulation during sorafenib treatment, increased apoptosis and enhanced drug sensitivity. Conversely, the presence of hemoglobin allowed tumor cells to tolerate oxidative stress more effectively.
What had initially looked like an unusual histological observation had therefore led us to a potential mechanism of therapeutic resistance.
Cancer cells appeared to have appropriated a molecule with ancient roles in oxygen and redox biology and incorporated it into their own stress-survival machinery.
From physiological adaptation to pathological exploitation
Looking back, the cartilage and cancer studies tell two related but distinct stories.
In cartilage, extra-erythrocyte hemoglobin is part of a physiological adaptation. It helps normal cells survive in an avascular, chronically hypoxic tissue.
In cancer, the same molecular resource becomes a pathological adaptation. Hypoxic tumor cells induce hemoglobin to restrain excessive oxidative stress, increasing their capacity to survive and withstand therapy.
The distinction is important. Evolution does not provide cancer cells with entirely new biochemical tools. More often, tumors exploit existing cellular programs—stress responses, developmental pathways, metabolic adaptations and survival mechanisms—that originally evolved for normal physiology.
In this sense, hemoglobin offered us an unusually clear example of molecular repurposing.
Following observations that do not fit
There is also a more personal lesson from this work.
Neither part of the story began with the expectation that hemoglobin would become central to the project. In cartilage, it started with an unexplained eosinophilic structure. In liver cancer, it started with an unexpected staining pattern that could easily have been dismissed as contamination.
At several points, the most straightforward interpretation was also the least interesting one: perhaps the signal came from red blood cells; perhaps the amount of hemoglobin in tumor cells was too small to matter; perhaps a molecule lacking the catalytic power of a canonical antioxidant enzyme could not make a meaningful contribution to ROS control.
Each of those possibilities was reasonable. But testing them rather than assuming them allowed the story to move forward.
The finding that some tumor cells contain hemoglobin at levels approaching those of neighboring erythrocytes made the histological observation difficult to ignore. The finding that hemoglobin depletion perturbed cellular ROS to an extent comparable with depletion of classical antioxidant enzymes made its functional significance equally difficult to dismiss.
Those experiments changed the way we thought about the molecule. Hemoglobin was no longer simply an erythroid protein appearing in an unusual place. It had become part of the stress-adaptation machinery of non-erythroid cells.
A familiar molecule with an unfamiliar life
Hemoglobin has been studied for more than a century, yet its biology is apparently not exhausted by what happens inside red blood cells.
Its journey from cartilage to liver cancer illustrates how the same molecule can acquire very different meanings depending on cellular context: an oxygen reservoir in one tissue, a redox buffer in another; a physiological adaptation in normal cells, but a therapeutic liability when hijacked by cancer.
More broadly, our work reinforces a principle that continues to shape how we approach biology: familiar molecules can still have unfamiliar lives.
Sometimes the most productive experiments begin with an observation that seems out of place-and with the decision not to dismiss it.