Sensing the World: Vision, Smell, Nitric Oxide and cGMP

A rod cell in your retina can register a single photon, and it reports that photon by switching a current off. In the dark the rod runs a steady inward current and releases transmitter without pause. Light reduces both. Every cascade you have met so far made a messenger. This one destroys a messenger that was already there. How can a fall carry a signal? And why does the same molecule, cGMP, mean “light” in the eye and “relax” in the wall of an artery?

The idea: one bond turns, and everything after it is dark

Rhodopsin is a G-protein-coupled receptor (GPCR), the seven-helix architecture from the receptor field guide. Its ligand does not arrive from outside. A small molecule called retinal, which the body makes from vitamin A, is joined by a covalent bond to a lysine in the pocket where another receptor would bind its hormone. The protein without its retinal is called opsin.

In the dark the retinal is bent. This form is 11-cis retinal, kinked at the double bond between carbons 11 and 12. Think of a mousetrap held cocked: opsin holds retinal in its higher-energy shape, and in that shape retinal holds the receptor off.

What does light actually do? In 1958 Ruth Hubbard and Allen Kropf at Harvard chilled rhodopsin so that its bleaching stalled partway, and asked which steps needed light. Only the first did. A photon turns that one double bond, and 11-cis retinal becomes the straight all-trans form. Every later step went forward in the dark as the sample warmed. George Wald put it this way in his Nobel lecture: “Everything else that happens — chemically, physiologically, indeed psychologically — represents ‘dark’ consequences of this one light reaction.”

Straight retinal does not fit a pocket shaped for bent retinal. The protein is pushed into a new conformation, metarhodopsin II, and that is the active receptor. From here on rhodopsin behaves like any GPCR that has just bound an agonist. Because its ligand was already in the pocket, it never has to wait for a molecule to diffuse to it.

A cascade that runs backwards: light makes cGMP fall

Start in the dark. The outer segment, the part of the rod packed with rhodopsin, contains cGMP, which binds directly to cation channels in its membrane and holds some of them open, with no kinase in between. Sodium and calcium flow in through those channels. This is the dark current. When Denis Baylor, Trevor Lamb and King-Wai Yau drew the outer segment of a single toad rod into a glass pipette, they recorded roughly 20 picoamps flowing in darkness and a dip of about 1 picoamp for each rhodopsin that a photon isomerized.

Because of the dark current the rod sits partly depolarized and releases its transmitter, glutamate, continuously. Now add light.

1. Light turns retinal and rhodopsin becomes active.
2. Active rhodopsin works as the exchange factor for the rod’s G protein, transducin. The α subunit of transducin drops GDP and takes up GTP, the switch from the lesson on GTPase switches.
3. The α subunit with GTP bound binds an inhibitory subunit of a cGMP phosphodiesterase called PDE6 and lifts its hold on the catalytic site.
4. PDE6 hydrolyzes cGMP to GMP.
5. cGMP falls, channels close, the inward current shrinks and the membrane hyperpolarizes. Glutamate release drops, and the next cell reads that drop.

light → active rhodopsin → transducin-GTP → PDE6 on → cGMP falls → channels close → less glutamate

Compare this with the adrenaline cascade from the signal transduction overview. There a G protein switched on a cyclase and a messenger appeared. Here a G protein switches on a phosphodiesterase and a messenger disappears. A signal is a change from a baseline, and a cell can read a change in either direction. The cost is that the rod must maintain that baseline all through the dark.

Two steps in this chain multiply: one active rhodopsin switches on many transducins, and one PDE6 destroys many molecules of cGMP. The arithmetic of that gain is worked through in the lesson on signaling circuits.

Recovery: four things have to be undone

A rod that could not shut off would report one flash forever. What has to be reset?

• Rhodopsin. A kinase phosphorylates active rhodopsin, and arrestin then binds it and blocks further contact with transducin. This is the desensitization from the previous lesson.
• Transducin. The α subunit hydrolyzes its GTP, helped by an RGS protein, and releases PDE6. The inhibitory subunit clamps back down.
• cGMP. A guanylyl cyclase remakes cGMP from GTP, and calcium tells it when. In the dark, calcium enters through the open channels and an exchanger carries it back out. When light closes the channels, entry stops while extrusion continues, so calcium falls. Low calcium switches on guanylyl cyclase-activating proteins, cGMP returns and the channels reopen.
• Retinal. All-trans retinal leaves opsin and is turned back into 11-cis retinal by enzymes, most of them in the pigment epithelium next to the rods. Fresh 11-cis retinal then joins opsin with no enzyme needed. This loop takes minutes, which is one reason your eyes adjust slowly to a dark room after bright light.

The calcium loop is negative feedback. Under steady background light it pulls cGMP partway back up, so some channels reopen and the rod can still answer an extra flash. That is adaptation, whose general logic is in the signaling circuits lesson.

Color and smell: the same plan with more receptors

How do you get color from a molecule that light can only straighten? Cones, the photoreceptors you use in daylight, hold the same 11-cis retinal and run the same kind of cascade. What differs is the opsin. The protein around the retinal shifts the wavelength it absorbs best: roughly 420 nm for the blue cone pigment, 530 nm for the green and 560 nm for the red, against about 500 nm for rhodopsin.

In 1986 Jeremy Nathans, with Thomas and Hogness at Stanford, isolated the three human genes by using the bovine rhodopsin gene as a probe. Each cone pigment is about 41 percent identical to rhodopsin. The red and green pigments are 96 percent identical to each other, and their genes lie side by side on the X chromosome, where sequences so alike are easily mispaired when chromosomes exchange segments. That is why red-green color blindness is common in men.

Smell takes the same idea much further. In 1991 Linda Buck and Richard Axel reported a very large family of genes, expressed in the olfactory tissue of the rat, that encode seven-helix receptors, and proposed that these are the odorant receptors. Mice carry about 1,000 such genes and humans about 400 working ones. Each sensory neuron in the nose expresses one of them. An odorant binds its receptor, a G protein switches on adenylyl cyclase, cAMP rises, and cAMP directly opens a cation channel related to the rod’s. The neuron depolarizes and fires.

So the nose runs in the familiar direction, messenger up and channel open, and the eye runs in reverse. One odorant usually activates several kinds of receptor and one receptor answers to several odorants, so the brain identifies a smell from the combination of neurons that fire. In taste, sweet, bitter and umami compounds are detected by GPCRs, while salt and acid act through ion channels.

A gas that relaxes arteries: nitric oxide raises cGMP

Outside the eye, cGMP carries its signal by rising.

In 1980 Robert Furchgott and John Zawadzki mounted rings of rabbit aorta in an organ bath. Some rings kept their endothelium, the single layer of cells that lines the vessel. In others it had been rubbed off. Acetylcholine relaxed the intact rings and failed to relax the rubbed ones. So acetylcholine was acting on the lining, and the lining was sending something to the muscle beneath it.

That something lasted only seconds, which made it hard to identify. Furchgott proposed in 1986 that it is the gas nitric oxide, NO, and Ignarro independently made the same proposal; in 1987 Ignarro’s group at UCLA, Palmer, Ferrige and Moncada in England, and Furchgott’s own laboratory all reported matching evidence. Ten years earlier Ferid Murad’s group at the University of Virginia had found that nitroglycerin and NO gas both switch on guanylyl cyclase, the enzyme that makes cGMP.

1. Acetylcholine binds a GPCR on the endothelial cell, and calcium rises in that cell through the Gq branch.
2. Calcium, bound to the protein calmodulin, switches on nitric oxide synthase, which makes NO from the amino acid arginine.
3. NO is small and uncharged, so it crosses membranes freely and reaches the smooth muscle cell next door. It cannot be stored or aimed, and it reacts away within seconds, so only near neighbors receive it.
4. In the muscle cell NO binds the iron of a heme group in soluble guanylyl cyclase. The enzyme changes shape and makes cGMP from GTP.
5. cGMP switches on protein kinase G. Smooth muscle contracts when calcium is high and its myosin is phosphorylated, and protein kinase G works against both. The muscle relaxes and the vessel widens.
6. A phosphodiesterase, PDE5, hydrolyzes cGMP and ends the signal.

Two drugs act on this pathway. Nitroglycerin, used for the chest pain of angina since the nineteenth century, is converted in the tissue to NO. It supplies from outside the gas that the endothelium makes for itself. Sildenafil inhibits PDE5. It makes no cGMP; it lets the cGMP made in response to NO last longer.

Design problem

A drug blocks PDE5, the phosphodiesterase that destroys cGMP in the smooth muscle of blood vessels. It also blocks PDE6, the related enzyme in the rod, about ten times less well. Predict its effect on blood vessels and its effect on vision, and explain each from the two cascades.