Every time one of your cells divides, it has to copy its entire genome, roughly six billion letters, and hand an exact duplicate to each daughter cell. It does this while its DNA is under constant attack, from reactive molecules made by its own metabolism and from radiation and chemicals outside.1 Strands break. Letters get damaged. Replication forks stall when they run into knots and tangles in the DNA ahead of them.
Most of the time, none of this matters, because cells carry an extraordinary set of repair systems. They detect damage, pause the cell cycle, and fix the problem before it can be copied into the next generation of cells. When those systems fail, errors pile up. If the errors land in genes that control growth, a cell can start to divide when it shouldn't.
That accumulation of damage, called genomic instability, is a feature of most cancers.23 It is also what I spent my PhD studying.
Where the instability comes from
The obvious explanation is that cancers start with broken repair genes. For some, that's exactly right. People who inherit a faulty copy of BRCA1, BRCA2 or one of the mismatch repair genes have a much higher lifetime risk of cancer, and their tumours show the scars of the missing repair pathway.
But most cancers are not inherited, and here the story is less tidy. When researchers sequenced large numbers of sporadic tumours, they found that mutations in DNA repair genes were actually uncommon before treatment. The instability seemed to come from somewhere else: oncogenes driving cells to divide so hard and fast that their replication machinery stalls and breaks, a process called replication stress.3
Instability is a double-edged trait for a tumour. It makes cancer cells less fit, because many of them accumulate damage they can't survive. But it also lets the population evolve faster under pressure, including the pressure of treatment.4
A weakness you can aim at
That double edge has given us one of the cleverest drug strategies in oncology. In 2005, two groups in the UK showed that cells missing working BRCA1 or BRCA2 are exquisitely sensitive to drugs that block a second repair enzyme, PARP.56 A normal cell can lose PARP and fall back on BRCA-dependent repair. A BRCA-deficient tumour cell has no fallback, and dies.
Geneticists call this synthetic lethality: two defects that are each survivable but fatal together. PARP inhibitors became the first approved drugs designed around the idea, and they are now part of standard treatment for several cancers in patients with BRCA mutations.7 The tumour's repair defect, the thing that helped it arise, became the thing that kills it.
The RecQ helicases
Repair depends on more than enzymes that cut and paste DNA. It also depends on helicases, motor proteins that travel along DNA and pull its two strands apart. Helicases untangle structures that would otherwise block replication, and they take apart repair intermediates that could cause trouble if left in place.
One family, the RecQ helicases, has a striking link to human disease. Humans have five of them. Inherited defects in three, called BLM, WRN and RECQL4, cause rare syndromes marked by chromosomal instability, a high risk of cancer, and in some cases features of premature ageing.8 The other two, RECQL1 and RECQL5, have not been tied to any human disease.49
RECQL5 interested me because the evidence that it matters was already there. In 2007, a team in Cleveland showed that mice lacking it are prone to cancer, and that their cells have more spontaneous DNA breaks and more recombination than normal. In the test tube, RECQL5 pulls the recombination protein RAD51 off single-stranded DNA, which suggested it acts as a brake on recombination happening where it shouldn't.9 What nobody knew was where in the genome that brake was needed.
Watching recombination in single cells
One of the clearest signs of recombination is the sister chromatid exchange. When a cell copies a chromosome, it makes two identical sister copies. If a replication fork breaks, the cell can repair it by borrowing the matching sequence from the sister, and sometimes the two sisters end up swapping segments. Under a microscope, after special staining, this shows up as a chromosome that changes shade partway along its length.
For decades that microscope view was the best available, and it could only place an exchange to within millions of bases. Strand-seq, developed in Peter Lansdorp's lab in Vancouver, changed that. It labels newly copied DNA in a single dividing cell so that only the original template strands get sequenced. A sister chromatid exchange then appears as a switch in strand identity partway along a chromosome, and in the original study those switches could be located to within as little as 23 bases.10
My PhD, in the same lab, used Strand-seq to map exchanges in human cells lacking different RecQ helicases. The idea was simple: if you know exactly where exchanges pile up when a particular helicase is missing, you learn what that helicase normally protects.11
Without RECQL5, cells had more sister chromatid exchanges and more DNA double-strand breaks. The extra exchanges weren't scattered at random. They clustered in genes that were being actively transcribed and that were rich in G-quadruplexes, knotted four-stranded structures that guanine-rich DNA can fold into.12 These are places where the machinery that reads genes and the machinery that copies them are likely to collide, and where RECQL5 was already thought to help clear the way.4 The picture that emerged is of RECQL5 as a caretaker for a particular kind of trouble spot: it stops cells from reaching for risky recombination where transcription and replication get in each other's way.
Why the map matters
A map of where the genome breaks is more than a curiosity. It shows which regions depend on which caretakers, and so which tumours might be vulnerable when a particular caretaker is lost. That is exactly the logic that turned BRCA deficiency into a target for PARP inhibitors.
We don't yet know whether RECQL5 loss creates a similar weakness that a drug could exploit. But the cells that taught us about BRCA and PARP were once just a puzzling repair defect in a dish. Every caretaker we understand is one more place a tumour might be made to pay for the shortcuts it took.