Ras to MAPK: The Kinase Cascade

Give a PC12 cell EGF and it divides. Give the same cell NGF and it stops dividing and grows long, nerve-like fibres. Both growth factors send their signal through the same four proteins: Ras, Raf, MEK and ERK. PC12 is a cell line grown from a rat adrenal tumour. How can one chain of enzymes carry two instructions? You need the chain first: what each protein physically does to the next. After that, the answer turns out to be a matter of time.

The idea: Ras-GTP catches Raf and holds it at the membrane

The first steps are review. An active receptor tyrosine kinase carries phosphotyrosines on its tail. The adaptor Grb2 binds one with its SH2 domain and holds SOS with its SH3 domains, so SOS arrives at the membrane, where Ras sits on its lipid anchors (see adaptors and scaffolds and lipid anchors and addresses). Prof. Deppmann’s line for the next step is “Sos is a guanine exchange factor.” SOS makes Ras let go of GDP, and GTP, which is more plentiful in the cell, takes its place. Ras stays on until it cuts that GTP to GDP, the timer covered in GTPase switches.

Now think of “Ras as Pacman.” Picture Ras stuck to the inner face of the membrane with its mouth open. The mouth is a pair of surface loops, switch I and switch II. With GTP in the pocket they hold a shape that other proteins can grip. With GDP they go slack. What Ras-GTP catches in this lesson is Raf.

Raf is a kinase that phosphorylates serines and threonines. In a resting cell it floats in the cytosol folded shut, with its front half, which carries a Ras-binding domain, held against its kinase domain. Ras is not a kinase. It works by contact, and two effects are well established. Binding brings Raf from the cytosol to the membrane, and it helps pull the closed shape open. At the membrane two Raf molecules then pair, kinase domain beside kinase domain, and pairing is needed for normal activation. Phosphates are also added to and removed from Raf along the way, and the exact order of those events is still being worked out.

Three kinases in a row, with a two-phosphate lock at the end

Active Raf phosphorylates the next kinase, MEK, on two serines in its activation loop, the flexible stretch beside the active site that you met in the phosphorylation switch. MEK then does something unusual. Most kinases phosphorylate either serine and threonine or tyrosine. MEK phosphorylates a threonine and a tyrosine, one residue apart, in the activation loop of ERK. ERK needs both phosphates. Take either one off with a phosphatase and ERK goes quiet.

Melanie Cobb’s laboratory followed an insulin-stimulated kinase activity through eight chromatography columns, cloned its gene and named the enzyme ERK, for extracellular signal-regulated kinase. Its sequence resembled two yeast kinases that respond to mating pheromone. Natalie Ahn, in Edwin Krebs’s laboratory, then asked what switches ERK on. In 1991 she found two activities in extracts of growth-factor-treated cells that switched on inactive MAP kinase in a test tube. An activator that comes apart from ERK on a column is a different protein, so ERK was not switching itself on. Purifying it with Rony Seger showed that it phosphorylates ERK on both threonine and tyrosine. Several laboratories reached this enzyme, now called MEK, at about the same time.

What do three kinases in a row give the cell? Signaling circuits covered two answers: each tier can amplify, and the cascade can turn a graded input into an all-or-none output. The two-phosphate lock on ERK is one ingredient of that sharpness. A third answer is control. Each tier is a separate place where a phosphatase, a feedback loop or a scaffold can act.

MEK has almost no known substrates other than ERK, so the middle of the cascade is a narrow pipe. ERK is the opposite. It phosphorylates serines and threonines that are followed by a proline, and many proteins carry such a site. Some are in the cytoplasm. Part of the active ERK also moves into the nucleus and phosphorylates transcription factors. One of them, Elk-1, helps switch on c-fos, the best-known immediate early gene. An immediate early gene turns on within minutes of a signal with no new protein made first, because the proteins it needs are already present. Its product is often a transcription factor that turns on later genes.

The professor’s advice here is “Downstream Pathways—don’t memorize.” Learn the logic: a narrow relay that ends in a wide fan. “Putting it all together”:

growth factor → receptor → Grb2 and SOS → Ras-GTP → Raf → MEK → ERK → transcription factors → immediate early genes

One design, several cascades

Raf, MEK and ERK are one example of a general plan. A MAP kinase kinase kinase turns on a MAP kinase kinase, which turns on a MAP kinase by phosphorylating a threonine and a tyrosine in its activation loop. MAP kinase, or MAPK, is the family name, and ERK is the member you have met. Animal cells run at least two more modules built this way, ending in the MAP kinases JNK and p38. They use different kinases at every tier, and they answer mainly to stress, such as ultraviolet light or inflammatory signals.

Several look-alike cascades in one cytoplasm create a wiring problem, and scaffolds are part of the answer. A scaffold holds the kinases of one module side by side so that they pass the signal to each other and not to a neighbour, as Ste5 does in yeast and KSR does for Raf, MEK and ERK (see adaptors and scaffolds). A scaffold can also decide where the cascade runs: “Scaffolds can localize ras-mapk.” Two examples are paxillin, which gathers Raf, MEK and ERK at focal adhesions, the spots where a cell grips its surroundings, and MP1, which holds MEK and ERK on endosomes.

How long is the signal on?

Back to the PC12 cell. A signal has a size and also a duration. Negative feedback sets the duration, and ERK supplies its own. Active ERK phosphorylates proteins above it in the chain, among them SOS and Raf, in ways that weaken them. It also turns on genes for dual-specificity phosphatases, enzymes that take both phosphates back off ERK. The second kind is slower, because a protein has to be made.

Different stimuli leave a pathway on for different lengths of time before feedback closes the window. A 2007 review by Amit, Wides and Yarden (Molecular Systems Biology 3:151) sketched how feedback loops set such windows: about 30 minutes for EGF acting through MAP kinase, about 90 minutes of gene activity after NGF, set by a relay of two gene-activation steps, a short pulse of NF-κB activity after TNF, an inflammatory cytokine, and about 120 minutes of NF-κB activity after LPS, a molecule from the outer membrane of bacteria. These are schematic values for different readouts, not one measurement in one kind of cell, and real windows shift with cell type and dose.

In PC12 cells, ERK activity after EGF rises and falls back within about an hour. After NGF it climbs to a plateau that lasts for hours. In 1995 Christopher Marshall proposed that the difference in duration is itself the instruction: a brief ERK signal goes with division, and a sustained one with differentiation.

Design problem

In a PC12 cell the same Ras-to-ERK cascade answers EGF with division, and answers NGF with an end to division and the growth of nerve-like fibres. Design a way for one cascade to give two answers. Then predict what an always-on MEK would do in this cell with no growth factor at all, and compare your prediction with what Cowley and Marshall found.

When the cascade is stuck on

Cowley and Marshall put the same always-on MEK1 into mouse fibroblasts and got a different result. The cells piled up into foci, the disordered heaps that mark a cell transformed toward cancer. Unending ERK activity means “differentiate” to a PC12 cell and “keep dividing” to a fibroblast.

Human cancers reach the same state from higher up. In 1982 the laboratories of Robert Weinberg, Mariano Barbacid and Michael Wigler each compared a ras gene cloned from one bladder-tumour cell line with the normal gene. The two differed at a single base, which changed glycine 12 to valine: the G12V mutation, whose effect on the GTP timer you saw in GTPase switches. A 1989 survey found mutant RAS genes in roughly nine of ten pancreatic tumours and about half of colorectal cancers. Reviews have long quoted up to about 30% for all human tumours. A 2020 analysis that cross-checked the large public cancer-mutation databases put it nearer 19% of patients, and the share varies widely by tissue.

One tier down, roughly half of melanomas carry a mutation in B-Raf, one of three Raf kinases in human cells, most often V600E. The glutamate sits in the activation loop and makes B-Raf active by itself, with no Ras-GTP and no partner needed. Vemurafenib fills the ATP pocket of the mutant kinase, and trametinib binds MEK in a pocket beside its ATP site and holds it inactive. Pairing explains a surprise that came with the Raf drugs. In a cell with normal B-Raf and active Ras, a drug-bound Raf can still pair with a drug-free Raf and switch it on, so the same drug raises ERK activity there. A MEK inhibitor blocks the cascade below that point, one reason the two kinds of drug are commonly given together.

The RASopathies are syndromes caused by mutations, inherited or present from conception, that mildly over-activate this pathway in every cell. They include Noonan syndrome, Costello syndrome, cardio-facio-cutaneous syndrome and neurofibromatosis type 1. The genes are this lesson’s cast: SOS1 and RAF1 in some cases of Noonan syndrome, HRAS in Costello syndrome, BRAF and the MEK genes in cardio-facio-cutaneous syndrome, and loss of the Ras GAP neurofibromin in neurofibromatosis type 1. Affected children share features of the face, heart and growth.

Raf is not the only protein that Ras-GTP can catch. PI 3-kinase is another, and it is the next lesson.