How One Protein Learned to Count: Cooperative Oxygen Binding in Hemoglobin

Every red blood cell carries roughly 270 million copies of a protein that solves a problem most molecules cannot. It has to grab oxygen tightly in the lungs and then let almost all of it go a few seconds later in a working muscle, even though the difference in oxygen pressure between those two places is not very large. A molecule that simply bound oxygen in proportion to how much was around would do a mediocre job at both ends. Hemoglobin does well at both because its four subunits talk to each other. That conversation is the subject of this essay.

A curve with a kink in it

The cleanest way to see the problem is to compare hemoglobin with its cousin myoglobin, the oxygen store inside muscle. Plot how saturated each protein is against the partial pressure of oxygen, and myoglobin gives a hyperbola: it loads up quickly and stays nearly full across the whole physiological range. Useful for storage, useless for delivery. Hemoglobin instead gives an S-shaped, or sigmoidal, curve. At the low oxygen pressures found in active tissue it dumps a large fraction of its cargo, and at the high pressures in the lungs it fills right back up.

The shape of that curve is a fingerprint of cooperativity. Binding the first oxygen makes the second easier, the second makes the third easier, and so on. Archibald Hill tried to capture this in 1910 with an equation whose exponent, now called the Hill coefficient, measures how cooperative the binding is. A value of 1 means each site acts on its own; a value of 4 would mean all four sites of hemoglobin act in perfect lockstep. The measured value sits around 2.8 under normal conditions, which tells you the coupling is strong but not absolute (Hill, 1910).

The two states

Hemoglobin is a tetramer: two alpha and two beta chains, each wrapped around a flat iron-containing ring called a heme. The iron, in its ferrous (Fe2+) form, is where oxygen actually binds. Max Perutz spent decades working out, by X-ray crystallography, what happens to the structure when it does, and the picture he assembled in 1970 still anchors the textbook account (Perutz, 1970).

The protein flips between two quaternary arrangements. The deoxygenated form is the T state, for tense. It is held together by a web of salt bridges and hydrogen bonds between the subunits, and it binds oxygen poorly. The oxygenated form is the R state, for relaxed, which binds oxygen well. The trigger that links one heme to the whole assembly is almost mechanical. In the empty T state the iron sits slightly out of the plane of the heme ring, pulled to one side. When oxygen binds, the iron slides into the plane, dragging the histidine residue attached to it, and that small motion of a fraction of an angstrom is transmitted through the protein. Once enough subunits have shifted, the salt bridges of the T state break and the whole tetramer snaps into R. Each binding event raises the odds that the rest will follow.

Two models compete to describe this, and both are still taught because each captures part of the truth. The concerted model of Monod, Wyman, and Changeux (1965) treats the tetramer as switching all at once between a pure T and a pure R state, with oxygen simply shifting the balance. The sequential model of Koshland, Némethy, and Filmer (1966) lets each subunit change one at a time, with binding inducing a local fit. Real hemoglobin behaves like a blend of the two, but the MWC framework remains the more popular shorthand because it explains the sigmoidal curve with so few assumptions.

The protein listens to its surroundings

Cooperativity alone would already make hemoglobin a good delivery system. What makes it a responsive one is that several small molecules tune the T-to-R balance from outside the oxygen site. This is allostery in the literal sense: regulation from another place.

The first of these is the Bohr effect, named for Christian Bohr, who described it in 1904. Working tissue produces carbon dioxide and lactic acid, which lowers the local pH. Hemoglobin senses that acidity and releases oxygen more readily, exactly where the oxygen is needed. The mechanism runs partly through a specific histidine near the end of each beta chain, His146, whose tendency to hold a proton changes depending on whether the salt bridge that stabilizes the T state is intact. More protons stabilize T, so more protons mean more oxygen let go. Carbon dioxide reinforces the effect by reacting with the protein’s terminal amino groups to form carbamates, which also favor the T state. The result is a feedback loop with no moving parts beyond the protein itself.

The second regulator took longer to find. In 1967 Reinhold and Ruth Benesch showed that a small molecule abundant in red cells, 2,3-bisphosphoglycerate, binds hemoglobin and lowers its oxygen affinity (Benesch and Benesch, 1967). It does this by wedging into the central cavity of the deoxygenated tetramer, a pocket lined with positive charges from the two beta chains, and clamping the T state shut. Without 2,3-BPG, hemoglobin would bind oxygen so tightly that it would barely release any in the tissues. With it, the pressure at which hemoglobin is half saturated, the P50, sits near 27 mmHg, right in the range that makes delivery efficient.

This is not a laboratory curiosity. People who travel to high altitude raise their 2,3-BPG levels within a day or two, which nudges the curve so that more oxygen is unloaded despite the thinner air. Fetal hemoglobin offers a different twist on the same theme. Its gamma chains bind 2,3-BPG weakly, so fetal blood holds oxygen more tightly than the mother’s and can pull it across the placenta. The same regulator, used to opposite ends, by changing one binding pocket.

When the mechanism is the target

Because the T-to-R equilibrium is a switch, drugs can be designed to push it one way. Sickle cell disease offers a hard case. The mutated hemoglobin, HbS, polymerizes when it is in the deoxygenated T state, stiffening the red cell into the sickle shape that blocks small vessels. One therapeutic idea is to hold more of the hemoglobin in the R state, which does not polymerize. Voxelotor, approved by the FDA in 2019, did exactly that by binding the protein and raising its oxygen affinity. The chemistry worked as intended. The clinical story did not end well: in 2024 the manufacturer withdrew the drug worldwide after trial data showed more painful crises and deaths in treated patients than expected (Pfizer, 2024). The episode is a reminder that shifting a 270-million-copy equilibrium in every red cell has consequences that a binding curve cannot fully predict.

Still, the basic lesson holds. Hemoglobin is not a passive container. It is a small machine that reads oxygen pressure, acidity, carbon dioxide, and phosphate levels, and adjusts its grip accordingly. Four subunits, one shared decision, repeated billions of times a second across the body. For a molecule with no brain and no moving parts to speak of, it manages something close to arithmetic.