Wnt: β-Catenin and Planar Polarity

Most of your cells make a protein called β-catenin all day long and, unless a signal tells them to stop, destroy it soon after it is made. That looks like waste. It is a design: a protein that is always being made and always being destroyed can be raised quickly by one move, which is to stop the destruction. A secreted signal called Wnt works this way. This lesson follows Wnt from its discovery to the nucleus and into colon cancer. It ends with a second use of the same receptor, in which the cells of a sheet work out which way to point.

The name: a mouse tumour gene and a fly gene were one gene

Around 1980 Roel Nusse, a postdoctoral fellow in Harold Varmus’s laboratory in San Francisco, began to study mammary tumours in mice infected with mouse mammary tumour virus. This virus copies its DNA into a mouse chromosome wherever it happens to land, and the inserted copy is called a provirus. If a provirus sits in the same stretch of mouse DNA in tumour after tumour, that stretch probably holds a gene that helps a tumour grow when the virus disturbs it.

In 18 of the 26 tumours they examined, a provirus sat inside one region of about 19 kilobases (19,000 base pairs). In the few tumours tested, an RNA was made from that region, and lactating mammary gland made none. They named the place int-1, for integration site 1. Their 1982 paper proposed only that insertion there strongly favours a tumour, perhaps by raising the expression of a cellular gene.

What the gene does in a healthy animal stayed unknown for five years. Then Frank Rijsewijk, in Nusse’s group in Amsterdam, cloned the fly version. Its protein was 54 percent identical to the mouse protein, with all 23 cysteines in the same positions, and the gene proved to be one that fly geneticists already knew as wingless. A weak wingless mutation, described by R. P. Sharma and V. L. Chopra in 1976, gives adult flies that often lack wings. Stronger ones kill the embryo and scramble the pattern of its segments.

So one gene, kept in both lineages for more than 500 million years, causes tumours in a mouse when it is switched on in the wrong tissue and lays out pattern in a fly embryo. In 1991 eight researchers signed a one-page note asking the field to use one name, spliced from wingless and int: Wnt. The fly match was also practical. Other fly mutants with the same segment defect became candidates for the rest of the pathway, and one of them, armadillo, is the fly’s β-catenin.

The idea: β-catenin is made all the time and destroyed all the time

A Wnt is a protein that one cell secretes and a nearby cell reads. To see what it changes, start with a cell that receives none.

β-Catenin has a second job, linking the adhesion proteins called cadherins to the cytoskeleton at cell junctions, and the copies doing that job are stable. Every other copy meets the destruction complex. Its scaffold is Axin, which holds an enzyme next to its substrate as the scaffolds in the lesson adaptors-and-scaffolds do. A second large protein, APC, also binds β-catenin and is needed for the complex to work. The enzymes are two kinases, CK1 and GSK3.

The order matters. CK1 puts a phosphate on serine 45 of β-catenin. That phosphate primes GSK3, which adds three more, on threonine 41, serine 37 and serine 33. Then comes the step to remember: SCF targets β-catenin. The phosphates on serines 33 and 37 are read by β-TrCP, the part of an SCF ubiquitin ligase that chooses the substrate. SCF adds a chain of ubiquitin, and the proteasome destroys the protein. The lesson ubiquitin-and-proteolysis gave the general rule, and these are the particulars.

Picture a bathtub with the tap running and the drain open. The water level depends on both.

steady level of β-catenin = rate of synthesis ÷ rate constant of destruction

Slow the destruction ten-fold and the level rises ten-fold, with no change in how fast the protein is made.

You have met GSK3 before, as the kinase that Akt switches off after insulin (lesson pi3k-akt-insulin). In the cultured cells where this has been tested, insulin does not raise β-catenin. The explanation offered is that the GSK3 held on Axin is a separate pool that Akt does not reach.

What Wnt changes: it stops the destruction

Wnt is an unusual secreted protein because it carries a lipid. In the cell that makes it, an enzyme called Porcupine attaches a fatty acid to one of its serines. The lipid makes Wnt cling to membranes, so it is hard to dissolve, and in many tissues it acts mostly on cells near its source.

The receptor is Frizzled, a seven-pass membrane protein with the overall shape of the receptors in Unit 3. In the crystal structure of a Wnt bound to the outer domain of Frizzled, the fatty acid lies in a groove on the receptor, so the lipid is part of the grip. Frizzled has a partner, a single-pass protein called LRP5 or LRP6, and Wnt brings the two together.

Inside the cell two things follow. Frizzled recruits a cytoplasmic protein called Dishevelled, and the tail of LRP6 is phosphorylated and then binds Axin. The destruction complex is pulled to the receptor, and there it stops delivering β-catenin to the ligase. How it is stopped is still argued. The outcome is clear: newly made β-catenin survives, builds up and enters the nucleus.

In the nucleus it finds TCF, a protein that sits on the DNA of Wnt target genes whether or not a signal has come. Without β-catenin, TCF holds a co-repressor called Groucho and keeps those genes off. β-Catenin displaces Groucho and brings in the proteins that start transcription, so TCF turns from a repressor into an activator. Which genes answer depends in part on where TCF sits in that kind of cell. One common target is Axin2, a relative of Axin, which gives negative feedback of the kind in the lesson signaling-circuits.

Wnt in disease: a pathway stuck on

Most people who inherit one faulty copy of the APC gene grow hundreds to thousands of small growths called polyps in the colon, a condition called familial adenomatous polyposis. Nearly all of them get colon cancer unless the colon is removed. APC is also mutated in most colon cancers in people with no family history, about four in five.

Why APC? In 1993 the laboratory of Paul Polakis and the laboratory of Bert Vogelstein and Kenneth Kinzler each reported that the APC protein binds β-catenin. In 1997 Hans Clevers’s laboratory and the Vogelstein and Kinzler laboratory published side by side. Colon cancer cells without working APC had β-catenin and TCF switching genes on all the time, and putting APC back turned that transcription off. Some colon tumours with normal APC carried mutations in β-catenin instead, at the serines and the threonine that CK1 and GSK3 phosphorylate.

Both routes end in the same place. Without APC the complex cannot mark β-catenin. Without the phosphorylation sites β-catenin cannot be marked. Either way the drain is plugged, and TCF is on with no Wnt. The gut is where this matters most, because the dividing cells of the gut lining sit at the base of pits called crypts and depend on Wnt: newborn mice made without TCF4 lack the dividing compartment of the small intestine. A cell without APC behaves as if it had never left the base of the crypt.

The same pathway sets bone mass. People with two inactive copies of LRP5 have fragile, low-density bone, and a single change in LRP5, glycine 171 to valine, gives unusually dense bone.

The other Wnt pathway: planar cell polarity tells a cell which side is which

Look at a fly wing. Each cell grows one hair, and every hair points toward the wing tip. A direction shared by the cells within the plane of a sheet is called planar cell polarity, or PCP.

Design problem

Each cell in a sheet grows one hair, and every hair must point the same way. No cell can see the whole sheet. Design a way for each cell to tell its two sides apart using only proteins in its own membrane and in the membranes of the cells it touches. Then predict what a patch of cells that lacks one of your proteins does to the normal cells beside it.

UVa moment: Paul Adler. The fly solves the problem this way, and the patch experiment was done at the University of Virginia. In 1987 Charles Vinson and Paul Adler made patches of cells lacking the gene frizzled in otherwise normal wings. Normal cells on the tip side of a patch grew misdirected hairs. Normal cells on the other side did not. Two years later the laboratory cloned frizzled and found that it encodes a seven-pass membrane protein. It is the same Frizzled: the Wnt receptor family is named for untidy hairs on a fly.

Later work placed the parts. Frizzled and Dishevelled gather on the tip-side edge of each wing cell. A membrane protein called Van Gogh gathers on the opposite edge with its partner Prickle. Flamingo, a cadherin, sits on both edges and links each cell to the next. The hair grows from the Frizzled side. None of this uses β-catenin or TCF. In the wing the output is the place where the cell builds the actin bundle that becomes its hair.

Vertebrates use the same genes to move cells. The neural tube, the future brain and spinal cord, starts as a flat plate that must narrow and lengthen before its edges can fold up and meet. Cells narrow the plate by squeezing between their neighbours toward the midline, and to do that each cell must know which of its sides faces the midline. Disrupting PCP causes neural tube closure defects, the group that includes spina bifida. In mice with two mutant copies of Vangl2, a vertebrate Van Gogh, the tube stays open along most of its length, and rare variants of the related human gene VANGL1 have been found in some patients with neural tube defects. Vertebrates need particular Wnts for these movements, Wnt5a and Wnt11 among them, which is one reason the system is counted as a Wnt pathway.