PI 3-Kinase, Akt and Insulin

In January 1922 a teenage boy in Toronto was injected with a purified extract of beef pancreas, and his blood sugar fell by about three quarters. Nobody could have said how. Insulin never crosses the cell membrane, yet within minutes of its arrival a muscle cell is pulling glucose out of the blood and storing it as glycogen. This lesson follows the signal from the receptor to one kinase, Akt, and then asks what happens when the path is too weak or cannot be shut off.

The hormone before the pathway: Toronto, 1921 to 1922

The pancreas makes digestive enzymes and, in small clusters of cells called islets, a hormone released into the blood. By 1920 several researchers had made pancreatic extracts that lowered blood sugar, and none of those extracts could be given safely to a patient.

Frederick Banting, a surgeon with no research training, took an idea to John Macleod, professor of physiology at Toronto, in November 1920: tie off the pancreatic duct in dogs so that the enzyme-making tissue wastes away, then extract what is left. Macleod gave him laboratory space, dogs and two student assistants, one of them Charles Best. Through the summer of 1921 Banting and Best prepared extracts that lowered blood sugar in diabetic dogs. On 11 January 1922 their extract was injected into Leonard Thompson, then in his early teens. His blood sugar fell a little, and an abscess formed at the injection site.

The extract was active and too impure to use. James Collip, a biochemist visiting from Alberta who had joined the group in December, raised the alcohol concentration of the extract in steps so that impurities dropped out of solution while the active substance stayed in, and he tested each batch on rabbits. His preparation went into the same boy on 23 January, the injection this lesson opened with. The 1923 Nobel Prize named Banting and Macleod. Banting shared his prize money with Best and Macleod shared his with Collip, so the two-name story you may know leaves out half the people.

They had a drug and no mechanism. That the receptor is a tyrosine kinase was not known until 1982, and most of the steps after it were found in the 1990s.

Recruitment of AKT: the receptor writes an address on the membrane

“Ligand activates receptor…now what?” The insulin receptor phosphorylates its own tyrosines and then those of a docking protein, IRS-1, which you met with the adaptors and scaffolds. IRS-1 has a PH domain and a PTB domain at one end, which hold it at the membrane beside the activated receptor, and a long tail with many tyrosines. Once the receptor has phosphorylated that tail, proteins with SH2 domains dock on it: PI 3-kinase, the adaptor Grb2 that leads to Ras, and the phosphatase Shp2. Adaptors mediate pathway branching, and this lesson takes the PI 3-kinase branch.

PI 3-kinase has two subunits. The regulatory subunit carries two SH2 domains, and the catalytic subunit does the chemistry. When the SH2 domains bind phosphotyrosines on IRS-1, two things follow. The enzyme is now at the membrane, where its substrate is, and the regulatory subunit loosens the hold it keeps on the catalytic one. The catalytic subunit adds a phosphate to the 3 position of the lipid PIP2 and makes PIP3, the membrane address from the lesson on lipid anchors.

Akt is a serine/threonine kinase, also called protein kinase B, with a PH domain at its amino end. Before insulin arrives it drifts in the cytosol, inactive. PIP3 binds the PH domain and holds Akt against the membrane. A second kinase with its own PH domain, PDK1, is held there too, and it adds a phosphate to a threonine in Akt’s activation loop (Thr308), the loop a kinase must set before it works. A separate kinase complex, mTORC2, adds a second phosphate to a serine near Akt’s tail (Ser473). The Thr308 phosphate does most of the work, the Ser473 phosphate adds to it, and the two together give full activity. mTORC2 belongs to the next lesson; here you need only its name.

insulin → receptor → IRS-1 (tyrosine-P) → PI 3-kinase → PIP3 → Akt at the membrane → Thr308-P (PDK1) + Ser473-P (mTORC2) → active Akt

Why a lipid and two phosphates? Each is a check: Akt is on only where PIP3 has been made, under an active receptor, and only while the lipid lasts. Two experiments show what the lipid is for. One is a way of visualizing PIP3: fuse Akt’s PH domain to green fluorescent protein, add insulin, and the glow moves from the cytosol to the plasma membrane within minutes. The other starts from a plain fact: lipids can be attached. Give Akt a permanent fatty-acid anchor and it sits at the membrane, gains both phosphates and is active with no insulin present. The main job of PIP3 is to put Akt in the right place.

PI3K–AKT regulation of metabolism: one kinase, three substrates

Different tissues need different things from the same hormone, and Akt provides them by phosphorylating a different protein for each job.

Glucose uptake in muscle and fat. GLUT4 is the glucose transporter of these two tissues. In the resting cell most of it sits in small vesicles inside the cell. In 1980 two laboratories working with rat fat cells (Cushman and Wardzala; Suzuki and Kono) showed that insulin moves transporters from an internal membrane pool to the cell surface. Akt’s part, found later, is to phosphorylate AS160, a GAP for the Rab GTPases that direct those vesicles. A GAP switches its GTPase off, as you saw with Ras, and Akt’s phosphate switches the GAP off. With the Rabs left in their GTP form, the vesicles fuse with the plasma membrane and glucose enters. This step was worked out mostly in fat cells.

Glycogen synthesis. An enzyme called glycogen synthase kinase 3 (GSK3) phosphorylates glycogen synthase and keeps it slow. Akt phosphorylates GSK3 and inhibits it, a link reported in 1995 by the laboratories of Philip Cohen and Brian Hemmings. Phosphatases then strip the inhibitory phosphates from glycogen synthase, and it speeds up. This is one of several ways insulin raises glycogen synthesis.

Gene expression in the liver. FOXO1 is a transcription factor, a protein that binds DNA and switches genes on. Between meals it sits in the nucleus of a liver cell and helps drive the genes for enzymes of gluconeogenesis, including glucose 6-phosphatase and PEP carboxykinase. Akt phosphorylates FOXO1, the phosphates become binding sites for 14-3-3 proteins, and FOXO1 is moved out of the nucleus and held in the cytosol. Those genes are then transcribed less. FOXO1 leaves the nucleus quickly; the enzymes already made take longer to disappear.

Akt also switches on mTORC1, a controller of cell growth, which is the subject of the next lesson.

Design problem

Twenty minutes after a meal a muscle cell must take up glucose and store it as glycogen, and a liver cell must stop making new glucose. Using one kinase, Akt, choose a substrate for each job and say whether Akt switches that substrate on or off. Then predict what happens to each job in a person with one faulty copy of the Akt gene that muscle and liver use most.

More AKT: the same pathway says survive and grow

Insulin is one of many signals that use this route. Growth factors such as PDGF switch on PI 3-kinase through their own receptor tyrosine kinases, and the PDGF receptor binds the regulatory subunit directly on its own phosphotyrosines, with no IRS-1 in between. In those cells the output of Akt is read as an instruction to stay alive and grow. Its substrates include proteins that would otherwise start the cell-death programme, and FOXO again, which in many cell types switches on genes that stop division.

A tumour gains from holding such a pathway on, and two breaks are common. PIK3CA, the gene for a catalytic subunit of PI 3-kinase, is among the most often mutated oncogenes in human cancer, including cancers of the breast, colon and endometrium. The mutations cluster at a few positions and raise the enzyme’s activity without the proper signal. PTEN, the phosphatase that removes the 3-phosphate, is a tumour suppressor that many tumours have lost. Either way the membrane carries more PIP3 and holds more Akt.

Ending the signal, and what insulin resistance means

Every step above has an eraser. Tyrosine phosphatases reset the receptor and IRS-1. PTEN removes the 3-phosphate from PIP3 and returns it to PIP2. With the address gone, Akt leaves the membrane and protein phosphatases take off its two phosphates.

The pathway also turns itself down. Kinases that it switches on, among them mTORC1 and its target S6 kinase from the next lesson, phosphorylate IRS-1 on serines. Phosphoserine makes IRS-1 a poorer substrate for the receptor and can mark it for destruction by the proteasome. This is negative feedback of the kind you met in the signaling circuits lesson: the output reduces its own input, so a long signal fades.

Insulin resistance means that a given concentration of insulin produces less response than it should. In muscle and fat from insulin-resistant people and animals, the receptor is usually present and binds insulin, but the steps after it are weak, and in severe obesity the receptor itself also signals less. Less phosphotyrosine appears on IRS-1, less PI 3-kinase is recruited, and fewer GLUT4 vesicles reach the surface. One well-supported explanation is the feedback loop driven too hard. With a chronic excess of nutrients, lipids and inflammatory signals keep serine kinases active, IRS-1 collects phosphoserine, and the docking protein docks less. The dose-response curve for insulin shifts to the right, so the same effect needs more hormone. The pancreas supplies it, often for years, and type 2 diabetes begins when it no longer can.

How much of human insulin resistance this one mechanism explains is still argued, and the break is not always at IRS-1. There is also a puzzle: the insulin-resistant liver fails to shut off glucose production, yet still makes fat in response to insulin. One proposed explanation for that split involves the kinase in the next lesson, mTORC1.