Signal-Transduction Pathways: How Cells Compute With Molecular Switches

One epinephrine molecule can mobilize millions of glucose molecules. That is the quantitative signature of how cells handle information, and it is the reason signal-transduction pathways exist at all. A signal that arrives as a few molecules at the cell surface has to become a response large enough to change metabolism, and it does that by passing through a relay of enzymes that each multiply the message.

Your phone holds maybe ten billion transistors. A single cell runs on thousands of signaling proteins wired together in countless combinations, and with them it computes: it senses the outside, weighs competing inputs, and commits to an action. The vocabulary is large, but the grammar is small. Almost every pathway in this chapter is the same four-beat sentence. A switch turns ON. The signal spreads and amplifies. The switch turns OFF. And when an OFF mechanism fails, the switch sticks, and that is disease. Hold onto that fourth beat. Cholera, Cushing’s syndrome, and most cancers are, mechanistically, broken OFF switches.

The five parts every pathway shares

Strip any pathway down and the same five components appear in order: a primary (first) messenger arrives from outside, binds a receptor, which generates or activates a second messenger inside the cell, which drives effector molecules, which produce the physiological response. A sixth feature is non-negotiable and easy to forget: active termination. The pathway has to be switched off on purpose.

First and second messengers sit on opposite sides of the membrane. A first messenger is the extracellular signal: epinephrine, insulin, a growth factor. A second messenger is made after the receptor fires and carries the order inward: cAMP, IP3, DAG, calcium, PIP3. The split exists because of a physical constraint. Small, lipophilic signals (steroids, thyroid hormone) slip straight through the membrane and find receptors inside the cell. Large or polar signals (peptide hormones, growth factors) cannot cross, so they must be read by a receptor that spans the membrane and relays the news to the inside.

The cAMP paradigm: epinephrine to glucose

The canonical cascade runs: epinephrine to the beta-adrenergic receptor, to Gs, to adenylate cyclase, to cAMP, to protein kinase A (PKA), to phosphorylase kinase, to glycogen phosphorylase, which liberates glucose. Each arrow is a relay handoff, and most arrows multiply.

Earl Sutherland found cAMP by walking into a contradiction. Adrenaline needed the membrane fraction of broken liver cells to work, yet the glucose-releasing effect appeared in the soluble fraction where no receptors live. The hormone could not act in two places at once. The only resolution was that adrenaline never enters the cell. The membrane converts its arrival into a small, water-soluble messenger that carries the command inward. That messenger is cyclic AMP, the first second messenger ever identified. Adenylate cyclase makes it from ATP by clipping off two phosphates and cyclizing the rest. Think of cAMP as used ATP with a circular handle. The cyclic phosphate is the message; cAMP is not the same molecule as AMP.

The receptor itself is a GPCR, a seven-transmembrane protein. GPCRs are the largest receptor family in humans (roughly 800 of them) and the target of about 35% of approved drugs. The ligand binds on the outside and never passes through. Binding outside causes a conformational change that is detected inside, and that change is what activates the G protein.

The heterotrimeric G protein has three subunits: alpha, beta, gamma. At rest, alpha holds GDP and the trimer is intact. An activated receptor acts as a catalyst, prying GDP off alpha and letting GTP take its place. Alpha-GTP then dissociates from beta-gamma and carries the signal to its effector. There are three logics depending on which alpha you have. Gs stimulates adenylate cyclase and raises cAMP. Gi inhibits it and lowers cAMP. Gq activates phospholipase C and launches the calcium branch. Same architecture, opposite outputs. The same epinephrine can speed your heart, mobilize glucose, and constrict a vessel, because the response is written into the receptor and its G protein rather than the ligand.

cAMP’s job is to free PKA. PKA sits as an R2C2 tetramer, two regulatory subunits clamping two catalytic subunits silent. cAMP binds the regulatory subunits, they let go, and the catalytic subunits come free and active. This is the universal toggle in action: reversible phosphorylation. Kinases add a phosphate using ATP; phosphatases take it off. The phosphate brings two negative charges and some bulk, which forces a conformational change that can activate or inhibit depending on where it lands. The human genome encodes more kinases than phosphatases, which tells you something about how much of biology is run by this one switch.

Now the amplification. It is multiplicative, not additive. One active receptor can switch on perhaps ten G proteins; each adenylate cyclase makes hundreds of cAMP; each PKA phosphorylates many phosphorylase kinases; and so on down the line. The final output is the product of the per-step gains, which is why one hormone molecule can release millions of glucose molecules. Knock out a single amplifying step and the whole output collapses, because you are dividing the product, not subtracting a term. The widget above lets you feel this: set the gains, watch the running tally climb a log-scale bar, then remove one step and watch millions become thousands.

Switching OFF, and why it is the hard part

A signal that cannot be turned off is not a signal; it is a stuck cell. Termination is active and has several layers. Alpha’s intrinsic GTPase hydrolyzes its own GTP back to GDP, so alpha reassociates with beta-gamma and goes quiet. Phosphodiesterase chews cAMP into ordinary AMP, ending the message. Caffeine inhibits phosphodiesterase (one of several effects), so cAMP lingers. Ligand dissociates as its concentration falls. And the receptor desensitizes and internalizes after sustained stimulation.

Spatial control comes from AKAP scaffold proteins, which tether PKA next to the specific targets it should phosphorylate. The clever part is that AKAPs also recruit phosphatases, building the OFF switch right beside the ON switch. The scaffold is a surgical tray with every instrument pre-arranged and within reach.

Three diseases make the point that the OFF switch is where things break. In cholera, choleragen chemically modifies Gs-alpha by ADP-ribosylation so it can no longer hydrolyze its GTP. Gs is locked ON, cAMP never stops, and chloride and water pour into the intestine. Severe dehydration can kill within hours. Pertussis (whooping cough) toxin uses the same chemistry, ADP-ribosylation, on a different target with the opposite result: it locks Gi in its inactive form, so Gi can no longer inhibit, and signaling runs unchecked. Two toxins, one mechanism, opposite effects. In some cases of Cushing’s syndrome, a mutation leaves PKA’s catalytic subunit unable to bind its regulatory subunit, so PKA is never inhibited and cortisol production runs without a hormone telling it to. The clinical picture (muscle weakness, thin and easily bruised skin, osteoporosis) is the downstream cost of a switch that will not turn off.

A reassuring counter-example from the same GPCR family: GLP-1 receptor agonists (Ozempic, Wegovy, Mounjaro). GLP-1 is released from gut L-cells after a meal and acts through a Gs-coupled GPCR with the same architecture as the beta-adrenergic receptor. Its safety feature is built into the wiring: GLP-1 stimulates insulin release only when blood glucose is already high. That glucose-dependence means these drugs rarely cause hypoglycemia, unlike the older sulfonylureas, while still acting on pancreas, brain, and stomach.

The Gq branch: one cut, two messengers

When Gq fires, it activates phospholipase C, which cleaves the membrane lipid PIP2 into two messengers at once. IP3 is soluble; it floats to the endoplasmic reticulum and opens an IP3-gated calcium channel, releasing a flood of Ca²⁺. DAG stays in the membrane, and together with that Ca²⁺ it activates protein kinase C. Conventional PKC needs both signals, which makes it a coincidence detector: it fires only when the pathway has confirmed itself two different ways. One PLC cut, a two-for-one deal, with the two products going to different places to do different jobs.

Dimerizing receptors: recruiting a kinase or being one

Not every receptor talks to a G protein. The second class is dimeric receptors that recruit a kinase. Growth hormone has two receptor-binding sites on one molecule, so it grabs two receptor chains and pulls their intracellular tails together. That clustering recruits JAK2, and the JAKs cross-phosphorylate each other, then phosphorylate STAT5, which dimerizes, enters the nucleus, and changes gene expression. JAK is named for Janus, the two-faced Roman god, because it carries two kinase-like domains though only one actually works. Cross-phosphorylation is the recurring trick here: the kinases phosphorylate each other, not themselves. Overdrive this pathway and you get acromegaly in adults or gigantism in children.

The third class is the receptor tyrosine kinases (RTKs), which do not recruit a kinase because they are one. EGFR dimerizes on ligand binding, the two halves cross-phosphorylate, and the resulting phospho-tyrosines become docking sites for the adaptor Grb2, which brings in SOS, which activates Ras. Ras is a monomeric small GTPase, a tiny version of G-alpha: GDP-off, GTP-on, with its own intrinsic GTPase. Active Ras drives the Raf-MEK-ERK cascade toward proliferation. About 30% of human cancers carry a Ras mutation that blocks GTP hydrolysis, locking Ras (and the proliferation signal) ON. It is the broken-OFF-switch story again, one tier down.

Insulin: the lipid-kinase deep dive

The insulin receptor is an RTK with one twist: it is a pre-formed dimer, already assembled before insulin arrives. Insulin binding triggers a quaternary rearrangement, the halves cross-phosphorylate, and the receptor phosphorylates adaptor proteins called IRS. Here the pathway forks away from the EGFR route. IRS recruits PI3K, and PI3K is a lipid kinase: it phosphorylates PIP2 to make PIP3, a membrane second messenger. PIP3 recruits PDK and Akt, and Akt drives the two outcomes that define the fed state: GLUT4 translocation to the membrane for glucose uptake, and glycogen synthesis.

Hold the two PIP2 stories side by side, because it is the cleanest contrast here. In the Gq branch, PLC cleaves PIP2 into IP3 and DAG. In insulin signaling, PI3K phosphorylates PIP2 into PIP3. Same substrate, opposite chemistry, opposite outcome.

Termination matters as much here as anywhere. Protein phosphatases strip the activating phosphates, and a lipid phosphatase called PTEN converts PIP3 back to PIP2, shutting down the Akt arm. PTEN is a tumor suppressor: lose it, and Akt signals constitutively, which drives cancer. Every clinical failure of this pathway is a failure at a specific node. Any block along insulin, receptor, IRS, PI3K, PIP3, Akt, GLUT4 reduces glucose uptake, and that is where type 2 diabetes lives.

Calcium and the cancer-as-signaling view

Calcium is a ubiquitous second messenger, and the common sensor that reads it is calmodulin, which carries four Ca²⁺-binding EF-hand motifs, each a helix-loop-helix fold (the motif is named for helices E and F of parvalbumin, where it was first seen). The finger pose, where index and thumb form the two helices and the curled middle fingers form the loop, is the mnemonic for the fold. Ca²⁺-loaded calmodulin then activates targets including the plasma-membrane Ca²⁺ ATPase (which pumps calcium back out, terminating the signal) and CaM kinase.

Step back and cancer is fundamentally a disease of signaling. Proto-oncogenes promote growth and become oncogenes when stuck ON; tumor-suppressor genes brake growth and cause cancer when lost. Ras stuck ON, PKA stuck ON, PTEN lost: each is a switch that failed.

Two drug strategies attack the wiring from opposite sides. Gleevec (imatinib) treats chronic myelogenous leukemia, where the Philadelphia chromosome fuses two genes into a constitutively active Bcr-Abl tyrosine kinase. Gleevec wedges into the kinase’s ATP-binding pocket and shuts it off, turning a rapidly fatal leukemia into a manageable chronic condition. It acts from the inside. Cetuximab is an antibody that blocks EGF from binding EGFR, preventing dimerization from the outside. If Gleevec is like putting gum in a lock, cetuximab is like covering the doorbell so no one can ring it. But cetuximab only helps if the cancer is actually EGFR-driven. A downstream KRAS mutation leaves the pathway stuck ON below the receptor, so blocking the doorbell does nothing. That is why precision medicine requires knowing the tumor’s genetics, and why single-pathway drugs so often fail: pathways cross-talk, and the cell reroutes around a block.

Cells compute with molecular switches. Broken switches cause disease. Knowing the wiring is what makes a therapy targetable. That is signal transduction.