Cancer GPS?

The thing that pretty well everyone knows about cancers is that most are furtive little blighters. They kill one in three of us but usually we don’t they’re there until they are big enough to make something go wrong in the body or to show up in our seriously inadequate screening methods. In that sense they resemble heart problems of one sort or another, where often the first indication of trouble is unexpectedly finding yourself lying on the floor.

Meanwhile, out on the highways and byways you are about 75 times less likely to be killed in an accident than you are to succumb to either cancers or circulation failure. Which is a way of saying that in the UK about 2000 of us perish on the roads each year. That it’s ‘only’ 2000 is presumably because here your assailant is anything but furtive. All you’ve got to do is side-step the juggernaut and you’ll probably live to be – well, old enough to get cancer.

Did you know, by the way, that ‘juggernaut’ is said to come from the chariots of the Jagannath Temple in Puri on the east coast of India. These are vast contraptions used to carry representations of Hindu gods on annual festival days that look as though walking pace would be too much for them. So, replace the monsters on our roads with real juggernauts! Problem largely solved!!

Flagging cancer

But to get back to cancer or, more precisely, the difficulty of seeing it. After centuries of failing to make any inroads, recent dramatic advances give hope that all is about to change. These rely on the fact that tissues shed cells – and with them DNA – into the circulation. Tumours do this too – so in effect they are scattering clues to their existence into blood. By using short stretches of artificial DNA as bait, it’s possible to fish out tumour cell DNA from a few drops of blood. That’s a pretty neat trick in itself, given we’re talking about fewer than 100 tumour cells in a sea of several billion other cells in every cubic millimeter of blood.

There are two big attractions in this ‘microfluidics’ approach. First it’s almost ‘non-invasive’ in needing only a small blood sample and, second, it is possible that indicators may be picked up long before a tumour would otherwise show up. In effect it’s taking a biochemical magnifying glass to our body to ask if there’s anything there that wouldn’t normally be present. Detect a marker and you know there’s a tumour somewhere in the body, and if the marker changes in concentration in response to a treatment, you have a monitor for how well that treatment is doing. So far, so good.

And the problem?

These ‘liquid biopsy’ methods that use just a teaspoonful of blood have been under development for several years but there has been one big cloud hanging over them. They appear to be exquisitely sensitive in detecting the presence of a cancer – by sequencing the DNA picked up – but they have not been able to pinpoint the tissue of origin. Until now.

Step forward epigenetics

Shuli Kang and colleagues at the University of California at Los Angeles and the University of Southern California have broken this impasse by turning to epigenetics. We noted in Twenty More Winks that an epigenetic modification is any change in DNA, other than in the sequence of bases (i.e. mutation), that affects how an organism develops or functions. They’re brought about by tacking small chemical groups (commonly methyl (CH3) groups) either on to some of the bases in DNA itself or on to the proteins (histones) that act like cotton reels around which DNA wraps itself. The upshot is small changes in the structure of DNA that affect gene expression. You can think of DNA methylation as a series of flags dotted along the DNA strand, decorating it in a seemingly random pattern. It isn’t random, of course, and the target for methylation is a cytosine nucleotide (C) followed by a guanine (G) in the linear DNA sequence – called a CpG site because G and C are separated by one phosphate (p). Phosphate links nucleosides together in the backbone of DNA.

Cancer cells often display abnormal DNA methylation patterns – excess methylation (hypermethylation) in some regions, reduced methylation in others – that contributes to their peculiar behavior. It’s possible to determine the methylation profile of a DNA sample (by a method called bisulfite sequencing).

Kang & Co. developed a computer program to analyse methylation profiles from solid tumours and healthy samples in public databases and compare them to patient DNA of unknown tissue origin.

The peaks represent CpG clusters that characterize normal cells (top) and a variety of cancers. The key point is that the different patterns identify the tissue of origin (from Kang, S. et al., 2017).

The program’s called CancerLocator and in this initial study it was used to test samples from patients with lung, liver or breast cancer. In the modest words of the authors, CancerLocator ‘vastly outperforms’ previous methods – mind you, they struggle to even to distinguish most cancer samples from non-cancer samples. Nevertheless, CancerLocator’s a big step forward, not least because it can detect early stage cancers with 80% accuracy.

It’s also reasonable to expect major improvements as methylation sequencing becomes more extensive and higher resolution reveals more subtle signatures. What’s more, in principle, it should be able to detect all types of cancers – meaning that, after all so many centuries we may at last have a way of side-stepping the juggernaut.


Kang, S. et al. (2017). CancerLocator: non-invasive cancer diagnosis and tissue-of-origin prediction using methylation profiles of cell-free DNA. Genome Biology DOI 10.1186/s13059-017-1191-5.

Transparently Obvious


Scientists have a well-earned reputation for doing odd things – by which I mean coming up with a ‘finding’ that leaves me, at least, wondering how, in the name of all things wonderful, they ever got money to do their study. To be fair, it’s the ‘social scientists’ – rather than the ‘real’ lot – that excel in this field. An example? Take your pick. They crop up pretty well weekly in the press. I liked the one on how something called ‘personal congruence’ affects marriage survival. The more congruence you and your partner have the better your chances: if, over time, your congruence goes down the tubes, your relationship will surely follow. But what on earth is congruence? Seemingly it’s a ‘state of agreeing.’ Lots of it equals harmony, loss of it = discord. So, it is what you remember from school geometry: it means more or less equal. Wow! Now I’ve grasped the upshot of this ‘study’: agreeably happy couples tend to make it: pairings based on whacking each other with frying pans tend to end in tears. Why didn’t they tell us earlier!!



Fortunately, in my world, even the weirdies usually turn out to be quite sensible, once you know what’s going on. Many moons ago a girl-friend asked me if I’d like to see her collection of axolotls. Not having a clue what she was on about I gave it an excited ‘yes please’. Whilst it mayn’t have been what I was hoping for (I was very young back then), I immediately fell in love with these wonderful amphibians that I’d never heard of as she explained what I should have known: these ‘Mexican walking fish’ have very large embryos which makes them particularly useful for studying development. These sensational salamanders really are amazing, not least because they can regenerate entire limbs after they’ve been chopped off.

More recently there’s been another unlikely recruit to the scientific armoury: the zebrafish – a tropical freshwater fish from the Himalayas. This mighty minnow was the first vertebrate to be cloned which led to its being genetically modified to give a transparent variety. That’s all good fun but what on earth is the point of a see-through fish? Well, in Betrayed by Nature we pointed out that you can actually watch tumours growing in transparent zebrafish and we got so excited by that we even included a photo – kindly provided by Richard White of the Dana Farber Cancer Institute in Boston. The cancer was a melanoma which had grown into a black mass about 1 cm in diameter in the fish’s body after a small number of tumour cells had been injected a couple of weeks earlier.

And the driver is …

Nearly 15 years ago, just as the first complete sequence of human DNA was being unveiled, Mike Stratton and his colleagues at the Sanger Centre in Cambridge discovered a mutation that arises in about two-thirds of all malignant melanomas. It’s in a gene called BRAF. The protein made by the gene is an enzyme that’s part of a signalling pathway that pushes cells to divide. The mutation changes the shape of BRAF protein so it works 24/7 as an enzyme: the pathway is no longer controlled by a message from the world beyond the cell. It’s a ‘molecular switch’ that’s been flipped by mutation to act as a cancer ‘driver.’

Richard White and his colleagues showed that the same mutation drove melanoma development in zebrafish and that when it did so something remarkable happened. As the tumours got going they turned on a gene that is normally only required during the first 72 hours after fertilization. The gene’s called crestin – because it’s switched on in a tissue called the neural crest where crestin protein helps to form the bony support for the gills. After that it’s switched off and crestin protein never appears again. Except in the pigment-containing cells called melanocytes when they are turning into a tumour.

Seeing the problem

In a great example of how science can work, Charles Kaufman, Leonard Zon and colleagues in Boston and other centres took this finding and made another transgenic variant of the transparent zebrafish. They cut out the stretch of DNA that controls whether the crestin gene is ‘on’ or ‘off’ and hooked it up to a gene that makes a green fluorescent protein (GFP). Result: when the machinery of a cell turns crestin on, GFP is also made – and the cell glows green under the appropriate light. Hence you would expect to see a glowing neural crest early in development but thereafter a non-glowing fish. Unless it has a melanoma. And Zon & Co saw exactly that. Because green fluorescent protein glows so brightly, a single cell shows up and it turned out that whenever one green cell was detected it always went on to expand and grow into a large melanoma tumour.

1 cell to mel

Tracking a single cell turning into a tumour over 6, 9, 11.5 and 17 weeks. The green fluorescence marks an early developmental gene (crestin) being re-activated in a melanoma tumour (from Kaufman et al., 2016).

But why might it be useful to ‘see’ single cells?

Since the original finding by Stratton & Co more detailed studies have confirmed that mutated BRAF is indeed an important ‘driver’ in about two-thirds of malignant melanoma. But here’s the odd thing: lots of melanocytes (the cells that can turn into melanomas) have mutated BRAF – but they don’t become cancerous. Why not? And there’s something else: it’s well-known that ultraviolet radiation in sunlight causes many melanomas and they do indeed often arise on exposed skin – but they can also crop up in places where, as they say, the sun doesn’t shine. So clearly, important though mutated BRAF and sunlight are, there’s something else that’s critical for malignant melanoma.

The Kaufman experiment was remarkable, not least because it offers a way of getting at this key question of what happens in a cell to kick it off as a tumour, by comparison with a near neighbour that remains ‘normal.’

The tumour cells used in this model carry mutated BRAF and another gene, P53, was knocked out. This gives two major genetic drivers and it may be that further genetic changes aren’t needed. If that’s the case, then the decisive push must come either from epigenetic changes (that affect gene expression without change in DNA sequence) or from adaptations of the tumour microenvironment to provide an optimal niche for expansion. At the moment we don’t know very much about these critical areas of cancer biology. Being able to follow single cells may lead us to the answers.

Keep your eye on the transparent minnows!


Kaufman, C.K., Zon, L.I. et al. (2016). A zebrafish melanoma model reveals emergence of neural crest identity during melanoma initiation. Science 351, Issue 6272, pp. DOI: 10.1126/science.aad2197