Calcium: The Messenger Kept at Arm's Length

In 1883 a frog heart kept beating in salt water mixed with London tap water, and stopped within about twenty minutes when the same recipe was mixed with distilled water. Sydney Ringer traced the difference to the hard pipe water: it carried dissolved calcium. Sixty-four years later Lewis Heilbrunn and Floyd Wiercinski injected a calcium salt straight into a frog muscle fibre, with no nerve attached, and the fibre shortened. A heart needs calcium, and calcium by itself can make muscle contract. So how does a bare ion, the same atom that sits in limestone, carry a message inside a cell?

The idea: a messenger the cell cannot make or destroy

Every other second messenger you have met is built and then broken. One enzyme makes cAMP and another destroys it. Calcium is an element. No enzyme can make it and none can break it down, so the only thing a cell can do with calcium is move it.

The cell moves it out, and keeps it out. A resting cell holds free calcium in its cytosol near 100 nanomolar. Outside the cell the concentration is between 1 and 2 millimolar, which is ten to twenty thousand times higher. Inside the endoplasmic reticulum (ER), the folded internal membrane compartment that serves as the cell’s calcium store, free calcium is several hundred micromolar. Muscle’s version of the ER is called the sarcoplasmic reticulum.

Think of a dam. The reservoir is held high at a steady cost, and opening one gate gives a flood at once. To send a calcium signal the cell synthesises nothing. It opens a channel, and ions already pressed against the far side come through within milliseconds. Because the starting level is so low, a modest number of ions raises cytosolic calcium tenfold, to about 1 micromolar, and higher still right beside the open channel.

Where the channel is

plasma membrane (calcium from outside) on the store (e.g. an IP3 receptor)

Channel: closed (click to open) Pumps and buffers: on (click to switch off)

Calcium is an element: the cell cannot make or break it, only move it, and an open channel only opens a path while the gradient does the work. Time is in relative units and the picture is a sketch of the logic, not a measurement. The store bar's height is drawn at 500 µM only for the picture; the lesson says several hundred micromolar, and the outside bar spans 1 to 2 mM.

Ca2+ “on” and “off” mechanisms: every channel has a pump against it

Sort the machinery by which way it moves calcium: “on” mechanisms let it into the cytosol, and “off” mechanisms take it away again.

	Where	What it does
On	Plasma membrane	Channels opened by voltage, by a ligand or by an empty store admit calcium from outside
On	Store membrane	The IP3 receptor and the ryanodine receptor are channels that release stored calcium
Off	Store membrane	The SERCA pump moves 2 Ca2+ into the store for each ATP
Off	Plasma membrane	The PMCA pump expels 1 Ca2+ for each ATP; the Na+/Ca2+ exchanger trades 3 Na+ in for 1 Ca2+ out
Off	Cytosol	Buffer proteins such as parvalbumin bind calcium and take it out of play

Everything on the “on” side is a channel. A channel only opens a path, and the gradient does the work. You can already place one: the IP3 made in the previous lesson lands on the IP3 receptor, on the ER. Everything on the “off” side either spends energy or soaks calcium up.

The evidence that a quiet cytosol is paid for came from muscle. By late 1961 Wilhelm Hasselbach and Madoka Makinose, in Heidelberg, had collected vesicles of sarcoplasmic reticulum from ground muscle, given them ATP, and watched them pull calcium out of the surrounding fluid. Within two years their measurements put the cost at about two calcium ions for every ATP split. A relaxed muscle is spending ATP to stay relaxed. In Tokyo, Setsuro Ebashi supplied the other half. Calcium acts on the filaments of skeletal muscle through a separate protein, which he and Ayako Kodama identified in 1965 and named troponin in 1966.

An empty store asks for a refill

A store that empties once and never refills is good for one signal. In 1986 James Putney, reassessing earlier findings on calcium stores, proposed that the emptying is itself the instruction: a depleted store tells the plasma membrane to open calcium channels. The idea is now called store-operated entry.

It took twenty years to name the parts. STIM1 crosses the ER membrane once, with a calcium-binding site facing into the store. In 2005 two groups, one including Michael Cahalan’s laboratory and one from Tobias Meyer’s, independently reported it as the sensor: when the store runs down, STIM1 gathers in clusters just beneath the plasma membrane. There it opens the pore, Orai1, which Stefan Feske and colleagues identified in 2006 from a family whose immune cells admitted no calcium.

”How do proteins sense calcium?”

An ion has no shape for a protein to recognise. What calcium has is a double positive charge and a liking for oxygen atoms. A protein senses calcium by offering a pocket lined with oxygens that sits empty at 100 nanomolar and fills at around a micromolar, which is the Kd idea from the lesson on binding and dose-response.

The commonest such pocket was first seen at the University of Virginia. In 1973 Robert Kretsinger and C. E. Nockolds published the full crystal structure of parvalbumin, a calcium-binding protein from carp muscle. Its chain folds into six helices, lettered A to F. One calcium ion sat in a loop between helix E and helix F. Hold up your right hand with the forefinger pointing and the thumb out: the forefinger is helix E, the thumb is helix F, and the curled middle finger is the loop that holds the ion. Kretsinger named the unit the EF hand. The loop is twelve residues long, and its aspartate and glutamate side chains wrap the ion in oxygen. The same helix-loop-helix has since been found by sequence in a large family of proteins, troponin and the myosin light chains among them.

Calmodulin: the ultimate in modularity. Parvalbumin binds calcium and that is the end of it. It is a buffer. Calmodulin binds calcium and changes shape, and the new shape does work. It is a sensor. Calmodulin is a small protein of 148 amino acids with four EF hands, two in each of two lobes, and a flexible linker between the lobes. When calcium binds, each lobe opens and exposes a hydrophobic (water-avoiding) patch. The two lobes then close around a short helix on a target protein, like two hands gripping a rope.

The modularity is this. Calmodulin has no catalytic activity of its own, and the one molecule serves many targets, because each target carries its own calmodulin-binding helix. Take one enzyme, CaM kinase II (CaMKII). At rest a segment of the kinase lies across its own active site and blocks it. Calcium-loaded calmodulin grips the region beside that segment and pulls it clear, and the kinase is free to phosphorylate its substrates. A rise in an ion has become a phosphate on a protein, which the lesson on the phosphorylation switch showed you how to read.

Where and how often: place and timing carry the message

Calcium is the same ion in every cell that uses it, so the information has to lie in where it rises and when.

Place first. Cytosolic buffers capture a free calcium ion quickly, so calcium does not travel far from the channel it came through. One small cluster of channels opening gives a brief, local signal. In a heart muscle cell such an event from ryanodine receptors is called a spark. A protein anchored beside the channel sees a high concentration, and a protein across the cell sees nothing. What calcium does therefore depends on which protein is standing close enough to catch it.

A local signal can become a global one because the release channels are themselves stimulated by calcium. Calcium from one cluster triggers the next, which is the positive feedback you met in the lesson on signaling circuits, and a wave crosses the cell. In a heart cell, many sparks set off together sum into the whole-cell rise behind each beat.

Now timing. Under a steady stimulus many cells do not hold calcium high. They spike repeatedly. In 1986 Woods, Cuthbertson and Cobbold recorded single rat liver cells injected with a light-emitting calcium reporter and found that the more hormone they gave, the more often the spikes came. The dose was written in frequency, the way FM radio carries sound.

Something has to count the spikes, and CaMKII can. Its twelve subunits are assembled around a central hub. When two neighbours hold calmodulin at the same moment, one phosphorylates the other, and the phosphorylated subunit stays active after calcium falls. Spikes that come close together build on each other, because subunits are still loaded from the last one. Spikes far apart each start from nothing. In 1998 De Koninck and Schulman showed with purified enzyme that the lasting activity of CaMKII depends on the frequency of calcium pulses.

Seeing calcium: from a jellyfish to a designed protein

None of this could be studied until calcium could be watched in a living cell. The first tool came from the jellyfish Aequorea. Osamu Shimomura, with Frank Johnson and Saiga, purified from it a protein that gives off blue light when calcium binds, and described it in 1962 under the name aequorin. Injected into a cell, as in the liver cells above, it reports each calcium rise in that one cell as a flash.

Dyes followed. Fura-2, from Roger Tsien’s laboratory in 1985, is a small fluorescent molecule whose best excitation wavelength shifts when it binds calcium, so the ratio of the signals at two wavelengths gives the concentration whatever amount of dye a cell took up.

The third tool is a protein the cell makes from a gene, and designing one is your problem.

Design problem

How could we use calmodulin’s shape change to make a calcium sensor? Design a protein that reports calcium by light, and say what control shows that it reports calcium and nothing else.

One way to do it

Calmodulin’s shape change is a change in distance: stretched out when alone, wrapped up when it holds a target helix. Build one chain in this order: a cyan fluorescent protein, calmodulin, a calmodulin-binding peptide, and a yellow fluorescent protein. With calcium low, the two fluorescent proteins are far apart, and exciting the cyan one gives cyan light. When calcium rises, calmodulin grips the peptide in its own chain, the chain folds up, and the two ends come within a few nanometres. At that range energy passes directly from the excited cyan protein to the yellow one, a process called FRET, and the yellow-to-cyan ratio rises. This is the “cameleon” design published by Miyawaki, Tsien and colleagues in 1997.

The ratio could also move with pH or with bleaching. The control is a calcium-blind twin: the same protein with the glutamate at the end of each EF-hand loop changed to glutamine, so that calmodulin can no longer hold calcium at cellular concentrations. It should show no change on stimulation, and any change it does show is not calcium. Then calibrate the working sensor inside the cell. An ionophore, a molecule that carries calcium across membranes, gives the maximum ratio in high calcium, and a chelator, a molecule that grips calcium tightly, gives the minimum.

One caution remains: the sensor is itself a calcium buffer, so use as little as you can read.