Heme Ligand-Switching: How Biological Carbon Monoxide Sensors Work

Every biological carbon monoxide sensor found so far uses the same trick: a **regulatory heme** whose iron is clamped by the protein's own amino acid side chains, which CO displaces. The displacement is the signal. ## The mechanism A heme is an iron atom held in a flat porphyrin ring, with two remaining coordination positions above and below. In a *transport* heme like haemoglobin's, one is held by a histidine and the other is left open for oxygen. In a *sensor* heme, **both** axial positions are occupied by the protein's own residues. The iron has no free site. When CO arrives it must **displace** one of those internal clamps and take its place — and because that clamp is part of the protein chain, knocking it off pulls on the backbone, propagating a conformational change to the protein's functional end. Binding energy becomes mechanical motion becomes changed output. Requiring displacement rather than mere binding is also a **selectivity filter**: a higher bar than simply arriving. ## CooA — the cleanest example The bacterial CO sensor **CooA** is a transcription factor with a heme-sensing domain and a DNA-binding domain, functioning as a dimer. In the resting state the heme iron is pinned by two internal ligands. One is the protein's own **N-terminal proline** — and remarkably, it reaches across from the *other* subunit of the dimer to clamp its partner's heme. The second ligand switches with redox state: a cysteine thiolate when the iron is Fe³⁺, a histidine when it is Fe²⁺. CO binds only the reduced Fe²⁺ form, displaces the proline, and the resulting motion repositions the DNA-binding domain so it can clamp onto the promoter — switching on **CO dehydrogenase**, the machinery that lets the organism metabolise CO for carbon and energy. Two safeties prevent false alarms: the sensor only arms in the correct redox state, and CO must win a ligand-displacement contest. **RcoM** is the aerobic cousin, using histidine/methionine ligation with CO displacing the methionine. ## Soluble guanylate cyclase — the same heme, two different answers **sGC** carries a heme with a deliberately **weak** iron–histidine bond — weak on purpose, so that it ignores the oxygen it is permanently bathed in. **NO** binds and *snaps* that iron–histidine bond, leaving a five-coordinate Fe–NO complex. The enzyme's activity rises several hundred-fold, producing cGMP and relaxing blood vessels. **CO** binds the same site but **keeps** the histidine bond intact, giving a six-coordinate complex — and only a few-fold activation. So sGC is really an NO sensor that CO can nudge weakly. Same heme, same gas-binding site, different resulting geometry, wildly different output. Pharmacology exploits this directly: sGC stimulator drugs work partly by helping CO break that histidine bond, and constitute a real cardiovascular drug class. ## NPAS2 — CO wired into the body clock **NPAS2** is a transcription factor that pairs with BMAL1 to drive the core circadian genes. Its PAS domains bind CO in the low-micromolar range, and CO binding **shuts off** its DNA binding. Since CO is produced continuously as cells recycle heme, this wires metabolic state into the circadian clock. There is a CO-responsive transcription factor in essentially every cell — tuning when you get sleepy, not whether you are being poisoned. ## The selectivity problem A heme binds O₂, CO and NO. Discriminating between them is genuinely hard, and the sensors cheat in combination: arming only in the right redox state, requiring displacement of an internal ligand, shaping the distal pocket to favour one gas geometrically, and — as sGC shows — producing different downstream conformations from the same binding event. It is an active research area, and nature's discrimination here is subtle and imperfect. See Electrochemical CO Detectors and the UL 2034 Time-Delay Curve for how the engineered solution sidesteps the problem entirely.