PLC, IP3, DAG and Protein Kinase C

Open small holes in the surface of a pancreatic cell, add one small molecule cut from a membrane lipid and no hormone at all, and the cell releases calcium from its internal store. That 1983 result showed what the molecule is for. This lesson follows the cut that makes that molecule and a second one, and the kinase that waits for both.

The idea: one cut makes two messengers

“PLC substrate—look familiar??” It should. The substrate is PIP2 (phosphatidylinositol 4,5-bisphosphate), which you met in the lesson on lipid anchors and addresses as a label on the inner face of the plasma membrane that PH domains read. Here the same lipid has a second job. It is a stockpile of signal, waiting to be cut.

PIP2 has two fatty acid tails on a glycerol backbone, buried in the membrane. Its head group sits in the cytosol: an inositol ring, which is a six-carbon ring with a hydroxyl on each carbon, joined to the glycerol through a phosphate and carrying two more phosphates at ring positions 4 and 5.

Phospholipase C (PLC) is the enzyme that cuts between that linking phosphate and the glycerol. The phosphate leaves with the ring, which is why a lipid named for two phosphates gives a product named for three.

PIP2 + H2O → IP3 (inositol 1,4,5-trisphosphate, leaves the membrane) + DAG (diacylglycerol, stays in it)

Think of a ticket torn along its perforation. The stub stays at the door and the other half goes with you, and either half is proof of the same event. IP3 carries three phosphates, so it dissolves in water and diffuses away. DAG is two tails on a glycerol, an oil, so it stays in the inner leaflet where it was made. A second messenger is a small molecule made inside the cell in response to a signal outside it. This one reaction makes two of them in two places: one free in the cytosol, one fixed at the spot where the receptor fired.

Two enzymes for two receptor families

Line up two members of the PLC family by their domains. Three modules of PLC-gamma are left as letters on purpose. Guess what they are before you read on.

PLC-beta: PH · EF hands · X · Y · C2 · long tail

PLC-gamma: PH · EF hands · X · A · B · C · Y · C2

X and Y fold together into one catalytic barrel, so both enzymes do the same chemistry. The PLC-gamma line leaves out one more module of the insert, a split PH domain. They differ in how they are told to start.

PLC-beta answers to G-protein-coupled receptors. A receptor that couples to the G protein Gq loads GTP onto the Gq alpha subunit, the cycle you know from the lesson on GTPase switches. Gq-GTP binds PLC-beta at the membrane and switches it on. PLC-beta also speeds up GTP hydrolysis by Gq, so the enzyme helps reset its own activator, and cutting continues only while receptors keep reloading Gq.

The next question is “why is gamma special??” The lettered modules are the answer. A and B are SH2 domains, and C is an SH3 domain, a module that binds proline-rich stretches of other proteins. An SH2 domain binds a tyrosine that carries a phosphate, within a particular short sequence. When a growth factor pairs up two receptor tyrosine kinases, each phosphorylates tyrosines on the other’s tail. PLC-gamma docks on one of those phosphotyrosines through an SH2 domain. Docking puts the enzyme beside its substrate and beside the receptor kinase, which phosphorylates PLC-gamma and releases an internal brake on the catalytic barrel.

So two receptor architectures with little in common converge on the same cut. Acetylcholine at a Gq-coupled muscarinic receptor uses PLC-beta. Platelet-derived growth factor at its receptor tyrosine kinase uses PLC-gamma. Both yield IP3 and DAG.

IP3, the half that leaves

By 1983 Michael Berridge, working on blowfly salivary glands, had shown that IP3 appears within seconds of the hormone arriving. Eight years earlier Robert Michell had proposed that the breakdown of inositol lipids is how receptors raise calcium. But a correlation leaves a gap. A hormone at its receptor starts many things at once, and IP3 and calcium might both follow from some third event.

Closing the gap takes IP3 inside a cell with no hormone outside it. IP3 is too charged to cross a membrane, so the cell has to be opened. In Irene Schulz’s laboratory in Frankfurt, H. Streb permeabilized pancreatic acinar cells, which means making the plasma membrane leaky to small molecules while the compartments inside stay sealed. IP3 that Robin Irvine had prepared in England from human red blood cells was added to those cells. Calcium came out of an internal store that was not the mitochondria. IP3 was sufficient without a hormone.

Later work identified the store as the endoplasmic reticulum (ER) and the target as the IP3 receptor, a channel in the ER membrane, built of four subunits, that opens when IP3 binds. The ER holds free calcium at a thousand times or more the resting cytosolic level of about 100 nM, so an open channel means a fast rise. How the cell reads that rise and clears it is the next lesson.

DAG, the half that stays, and protein kinase C

DAG found its target first. In Kobe, Yasutomi Nishizuka’s laboratory had isolated a protein kinase from brain, and in 1977 it reported switching the enzyme on by clipping it with a protease. Then calcium plus membrane phospholipid proved enough to activate the intact enzyme, and some lipid preparations worked better than cleaner ones. The group identified the difference instead of purifying it away. By 1980 it had a name: diacylglycerol, which let the kinase work at much lower calcium.

The enzyme is protein kinase C (PKC), which puts phosphates on serines and threonines. A conventional PKC has four parts that matter here:

• a pseudosubstrate, a stretch that resembles a substrate but has alanine where the serine or threonine would be, and that sits in the active site and blocks it;
• a C1 domain, which binds DAG in the membrane;
• a C2 domain, which binds acidic membrane lipids such as phosphatidylserine only when it holds calcium;
• the kinase domain.

At rest the enzyme is folded shut in the cytosol. Calcium lets C2 land on the membrane. There C1 finds DAG. With both domains holding on, the pseudosubstrate is pulled out of the active site and the kinase works. Cut the tether instead and you get a kinase that needs no signal, which is what the protease did in Nishizuka’s first assay. Stress pathways use the same trick: in a cell committed to programmed death, a protease called a caspase clips PKC-delta and frees its kinase domain.

Not every PKC asks for both signals.

Class	Members	Regulatory parts	Needs
Conventional	alpha, beta, gamma	C1; C2 that binds calcium	DAG and calcium
Novel	delta, epsilon, eta, theta	C1; C2-like domain that does not bind calcium	DAG only
Atypical	zeta, iota	C1-like domain that does not bind DAG; PB1 domain for protein partners	neither

“Phorbol esters mimic DAG to activate PKC.” Phorbol esters come from croton oil, and they promote tumors in animal skin without mutating DNA. In 1982 Monique Castagna, working with Nishizuka’s group, showed why: the phorbol ester takes DAG’s place and activates the kinase directly. The difference is time. A cell clears DAG quickly, but it has no efficient way to break down a phorbol ester, so the kinase is held on for hours.

This is the hinge of a case study on environmental chemicals and cancer. Soot has been tied to cancer since Percivall Pott’s 1775 report on chimney sweeps, and the hydrocarbons later isolated from coal tar act by damaging DNA. A phorbol ester contributes to the same disease by impersonating a second messenger.

The direction of the effect took decades to read correctly. PKC that is held on for hours is degraded, so long exposure to a phorbol ester removes the enzyme. Alexandra Newton’s laboratory tested 46 PKC mutations from human tumors and found that most disabled the kinase and none activated it.

Coincidence detection: an answer only when two signals meet

“Let’s name some signals”: calcium, phosphorylation, a membrane phospholipid, cAMP, an active small G protein, a protein partner. Each one rises for many reasons. Calcium also rises when a voltage-gated channel opens, and other lipid enzymes also make DAG. A kinase that answered to any single one would fire all day.

A coincidence detector is a protein that responds only when two signals arrive in the same place at the same time. Conventional PKC is one. Each of its membrane contacts is weak alone. Binding energies add, which means preferences multiply:

ΔG(both) ≈ ΔG(C1) + ΔG(C2), so K(both) ≈ K(C1) × K(C2)

If each contact alone favors the membrane thirtyfold, say, the pair favors it about nine hundredfold. Only the pair holds the enzyme there firmly enough to free the active site. And calcium and DAG appear together at the inner face of the plasma membrane for one main reason: PLC has just cut PIP2 there. The kinase recognizes the cut by sensing both halves of it.

The signals

calcium is high (C2 can grip) DAG is in the membrane (C1 can grip)

Each contact alone favors the membrane by (say): 30-fold

Keep the rule: neither signal alone is enough, because only the pair holds a conventional PKC at the membrane firmly enough to free its active site. Thirtyfold is the lesson's example, not a measured value, and this is true of conventional PKC, not of every PKC. This is a sketch of the logic. The slider's range, 2 to 100, is this sketch's own, and the lesson states no threshold.

The signal ends on both halves too. Phosphatases strip IP3 and a DAG kinase converts DAG to phosphatidic acid, and both products are recycled into new PIP2. Meanwhile Gq hydrolyzes its GTP and calcium is pumped away, so PKC folds shut and returns to the cytosol.

Design problem

Design a detector. It should respond only when two of these signals are present together: calcium, a membrane lipid such as DAG, phosphorylation, cAMP, an active small G protein. Say which domains you would build it from, and why one signal alone must not be enough.

One way to do it

Choose DAG and an active small G protein, then copy the PKC plan and swap one part.

1. Output. A kinase domain kept off by a pseudosubstrate on a flexible tether.
2. First sensor. A C1 domain, for DAG.
3. Second sensor. In place of C2, a domain that binds a small G protein only in its GTP form. The kinase Raf has one for Ras. Ras is anchored in the membrane by lipid, so this is a second membrane contact.
4. Tuning. Make each contact weak, so that either one alone holds the protein at the membrane too briefly to pull the pseudosubstrate out.

One signal alone must not be enough because each signal alone is ambiguous. DAG says that some enzyme cut a lipid. Ras-GTP says that some receptor switched Ras on. Together they say that both happened at this patch of membrane, now. If either one could switch the kinase on, you would have built an OR gate, and its output could not tell the pair of events from a single one. You can test the design the way PKC was tested: disable one domain and the response should vanish, and DAG alone should do nothing while Ras holds GDP.