The Naked Scientists explore how scientists could work out what you look like from your genes; the algorithm that can decode your brain activity; how researchers have made a breakthrough in discovering what actually causes migraines; and why men shouldn't leave fatherhood too late.
Plus in 'Stuff and Non-Science', will you really explode in space?
Chris Smith: Coming up this week, more than just a DNA fingerprint, how scientists could work out what you look like from your genes.
Kat Arney: A group of scientists in the Netherlands have developed a way to tell what colour eyes a person will have just from a sample of their DNA. This is the first time that scientists have managed to show that a complex trait can be predicted with any kind of accuracy from DNA alone. So if we can manage to do it for this trait, perhaps we can do it for others like hair or skin colour. So you could draw up a basic profile of a criminal from DNA at a crime scene.
Chris Smith: And Kat Arney will be investigating that very soon. Also on the way, how scientists have come up with a computer programme, or algorithm, that can decode your brain activity to work out where you are.
Demis Hassabis: Once you’ve trained these algorithms up with these examples, you can then present these algorithms with new brain scans that they haven’t seen before, and you can then ask the algorithm, right, which room do you think the person was standing in while this brain scan was taken and, furthermore, which position in the room were they standing in? And it turns out that basically we can very accurately determine that.
Chris Smith: And Demis Hassabis will be telling us how that feat was achieved in just a moment. Plus, in this week’s ‘Stuff and Non-Science’, we’re tackling a mystery that’s out of this world.
Dr Stephen Juan: In the 1981 film, Outland, that starred Sean Connery, there was a construction worker who had a torn spacesuit that leaked, and he swelled up and exploded, and then Keir Dullea in the classic 2001: a Space Odyssey, he blows himself into the airlock from the pod without a helmet but he doesn’t blow up.
Chris Smith: But which film got it right? We’ll find out later in the programme. Hello, I’m Chris Smith and this is Breaking Science which is produced in association with the Open University.
Migraines are quite literally a major headache for people who suffer from them, and, in a moment, we’ll hear how researchers have made a breakthrough in discovering what actually causes them to happen. But first with news of why men had better not leave fatherhood too late, here’s Kat Arney.
Kat Arney: Yes, Hugh Hefner famously became a dad in his sixties and generally speaking men are fathering children at older ages, although only rarely while they’re in retirement, but are there health risks for the children of older dads? Now recent research has suggested that kids fathered by older men might have an increased risk of birth deformities and other health conditions, and now a new analysis by Australian researchers suggest that there might be other more subtle problems too.
Chris Smith: Such as what?
Kat Arney: Well this is a study led by John McGrath, and this is researchers at the University of Queensland. They reanalysed data from a study of more than 33,000 children born in the US between 1959 and 1965, and this was data from an impressive study called The Collaborative Perinatal Project which tested every child at eight months, four years and seven years for a variety of skills, coordination and intelligence, and writing in the journal PloS Medicine this week, the team reanalysed this data to take socioeconomic factors into account. So as well as looking at the age of the parents, they also looked at the mother and father’s education levels and their income as well as other factors.
Chris Smith: And this reanalysis, what did it show?
Kat Arney: Well the researchers discovered that after adjusting for these social factors, the children of older dads were more likely to have lower scores on the tests, with the exception of one test of physical coordination. But, interestingly, they found that the children of older mothers were more likely to score highly in the tests.
Chris Smith: That seems a bit surprising, doesn’t it, because you’ve got the children of the older dads being at a disadvantage, the same is not true with the mums, so what’s going on?
Kat Arney: Well it’s true, and some researchers have suggested that maybe kids born to older mothers have a more nurturing home environment, and studies suggest that this doesn’t hold true for older dads. But, looking for a biological explanation, we also know that men’s sperm can accumulate DNA damage, especially as they get older, as it can in female eggs. But the take-home message seems to be that actually we need to try and find new explanations for these observations that take into account both the social factors that are at work, and the underlying biology because there is this trend towards older parenthood.
Chris Smith: So you basically need a young dad and an old mum, is that what you’re saying?
Kat Arney: Erm, possibly, yes.
Chris Smith: But you don’t need a migraine. You need one of those like a hole in the head. Researchers may have some help on that front.
Kat Arney: Exactly, migraines can be extremely debilitating as any sufferer will know. But the trouble is we still don’t really know what causes them. And now researchers writing in the latest edition of the journal Neuron have made a step forward in understanding what may make the brain vulnerable to migraine attacks.
Chris Smith: What have they found?
Kat Arney: This is work by Daniela Pietrobon and her colleagues in Italy and the Netherlands. Now we know from previous research that the so-called aura in migraine, this is a visual disturbance, is caused by something called cortical spreading depression. This is an electrical wave that passes across the brain. Now it was thought that this wave is basically what brings on the migraine. And in this work the researchers studied mice with a gene fault called FHM1. This is the same fault that causes a condition in humans called familial hemiplegic migraine, and these mice also showed cortical spreading depression, so they’re thought to be a pretty good model for human migraine.
Now the scientists studied the brains of these mice in depth, and they discovered that the brains with the FHM mutation showed high levels of release of a neurotransmitter called glutamate, and this is the main chemical in the brain that excites or activates nerve cells. But when the researchers dropped the levels of glutamate in these FHM mice, the mice didn’t show cortical spreading depression, so presumably weren’t experiencing these migraines.
Chris Smith: One thing to say, it’s true in mice though, what about in humans?
Kat Arney: Well the research does show that the overactive release of glutamate might explain why this cortical spreading depression is more likely in the mice with the mutation. It does suggest that perhaps migraines are down to imbalance between activation and suppression of nerve cell activity in the brain. Now this idea does need following up with studies in humans, but it may also help to explain why some people are susceptible to migraines and could even point towards new avenues for treatment in the future.
Chris Smith: Let’s hope so, because for people who have them they really are not very pleasant.
Now sticking with the brain, BSE made a lot of headlines in the 80s and early Nineties. Now scientists have got new insights into actually the protein that underlies BSE (the prion) and what it does.
Kat Arney: Yes, this isn’t one about mad cow disease but possibly mad fish disease. So BSE, this mad cow disease, and its human equivalent, Creutzfeldt Jakob Disease, or CJD, are very rare but very serious degenerative brain diseases. Now we’ve known for over 20 years that they’re caused by rogue protein, this prion protein, and it exists in two different forms or shapes in the brain.
Now the healthy, normal shaped protein plays a useful role in the brain, but the rogue form can spread rapidly causing healthy proteins to adopt this abnormal shape and causing disease. But it’s not really clear why we produce this prion protein in the first place, or what its exact function is in the body. But now new results from German researchers have shed light on the role of the normal prion protein which could explain more about the causes of prion diseases.
Chris Smith: It’s quite surprising though isn’t it because you’re saying we’ve known about this for a long time, 20 years, why has the real role of this protein only surfaced now?
Kat Arney: Well it’s been a really tough problem to crack, mainly because, for example, experiments with genetically-modified mice that lack the prion protein have just shown them to be perfectly healthy.
Chris Smith: I guess we infer from that there must be something else which is taking over the role of the missing protein when you get rid of it, but how have the scientists now gone on to solve the problem?
Kat Arney: Well they’ve used a simpler organism to try and answer this, and writing in this week’s edition of PloS Biology, Edward Málaga-Trillo and his colleagues studied the developing zebra fish. Now zebra fish are a handy model for these kind of studies because they’re very quick to grow. Now scientists injected zebra fish eggs with chemicals called morpholinos. These are a bit like DNA and they block the production of specific proteins.
So in this case, they used morpholinos that were specifically designed to target the prion protein, and they found that the zebra fish were unable to develop properly. For example, they didn’t really develop a proper nervous system, and they died.
Chris Smith: But why is that? Why would it have such a dramatic effect when you said earlier that when they removed it from mice the mice didn’t have any ill effects?
Kat Arney: Well there’s obviously something in mice that’s compensating for the lack of the protein, but that’s not there in the zebra fish, and the researchers showed that other proteins that are normally found at the sites of contact between brain cells disappeared when the prion protein was taken away so the cells couldn’t communicate properly. So it looks like this prion protein is playing some kind of important role in organising cell-to-cell contact, particularly in the brain.
So the prion protein’s probably got a similar role in mice or humans, but because we’re a bit more complicated than fish, it hasn’t been possible to tease it out before.
Chris Smith: And going back to where we started this story, which was a mention of BSE and CJD, does it give us any help with insights into those conditions and how we might be able to deal with them?
Kat Arney: Well this isn’t going to lead to a cure for CJD right now, but it’s certainly an important piece of the jigsaw. And if we can understand what the prion protein normally does, we’ll have more of an idea how it goes wrong in CJD, and, as we discussed a couple of weeks ago, the prion protein may actually play a role in Alzheimer’s Disease so this is pretty important research for that area too.
Chris Smith: Indeed, and let’s just finish off by taking a look, not around the eyes but into the eyes as one comedian famously said, what can you tell by looking into my eyes about my DNA?
Kat Arney: Yes, looking around at families, it’s obvious that certain characteristics, your hair colour, your eye colour can be inherited, but it’s actually surprisingly difficult to tell just from a DNA sequence exactly what characteristics a person might have. But according to a report in this week’s edition of Current Biology a group of scientists in the Netherlands have developed a way to tell what colour eyes a person will have just from a sample of their DNA.
Chris Smith: But that sounds pretty basic. Can we not do that already?
Kat Arney: Well, you may think that eye colour is actually quite simple, but it’s what biologists call a complex trait. It means it’s determined by several different genes. And over the years a number of these genes have been found, and it’s a combination of certain variations of these that give us our unique eye colour.
Chris Smith: So how have these researchers made a breakthrough here? What have they done that’s different?
Kat Arney: Well, Manfred Keyser and his team have analysed the DNA of more than 6,000 Dutch people. They’ve linked variations in eight eye colour genes to their actual eye colour. They started by looking at 37 gene variations known as SNPs in the eight genes, and they narrowed it down to the six best variations or SNPs in six genes that were the most strongly linked to eye colour. And they found that testing these six SNPs could predict whether a person’s eyes would be brown or blue with around 90 per cent accuracy. Now for people with other colour eyes, I mean mine are green, they were about 75 per cent accurate.
Chris Smith: So what does this mean if you commit a crime? Will we be able to, by examining the crime scene and picking up some of your DNA, predict that you’ve got green eyes?
Kat Arney: Well, for a start, this is more of a proof-of-principle experiment. It’s the first time that scientists have managed to show that a complex trait can be predicted with any kind of accuracy from DNA alone. So if we can manage to do it for this trait, perhaps we can do it for others like hair or skin colour, so you could draw up a basic profile of a criminal from DNA at a crime scene. But, of course, there’s a lot to be done.
This study was only done in Dutch Europeans, so we still need to find out if the results hold true across other populations. And, of course, you can always pop in some coloured contact lenses to fool the police. But it could probably be useful for drawing up a profile of murder victims or other deceased people who are too decomposed to keep their identifying features.
Chris Smith: DNA photo-fits on the way then potentially. Thank you, Kat. That was Dr Kat Arney from The Naked Scientists with a look at some of this week’s top science news stories. And, of course, if you’d like to follow up on any of those items, the details and the references are all on the Open University’s website, and that’s at open2.net/breakingscience.
In just a moment, how surface scratches on your favourite phone could heal themselves with nothing more than just a bit of sunlight. But first to decoding what the brain’s doing. We tend to think of our thoughts as very much personal to us and something that other people don’t know anything about, but now researchers in London have found a way to read someone’s brain activity and, in this instance, to work out where people think they are in a virtual room.
The team placed volunteers in a brain scanner and then asked them to use a joystick to wander round in a virtual reality environment that was shown to them on a screen. All the time that this was going on, the activity in a part of their brains called the hippocampus was being logged by the scanner. And what the team found were large groups of interconnected nerve cells that do the job of telling you where you are. Demis Hassabis.
Demis Hassabis: So we’ve known for a long while now from studies undertaken on rats that if you record from single neurons in a rat’s brain, you can actually read out where a rat is in a location just from the activity of single neurons in a part of the rat’s brain called the hippocampus. And all mammals have this part of the brain, the hippocampus, and it seems that it’s essential for all sorts of memory activities such as navigation, spatial memories and also remembering what you’ve done at the events in your daily life, record autobiographical memories. So whereas in rat studies they record from, you know, maybe a few dozen neurons at the same time, with neuro imaging techniques we’re going to scan the whole brain at once.
Chris Smith: So looking at how this might work for a second then, what you’re saying is that to-date what people have done is to stick electrodes into a brain and then record animals as they move around an environment and when they go to certain places in the environment, certain cells or collections of cells become more active?
Demis Hassabis: That’s right, and it’s likely to be a large number of cells that all collectively, together, code for a position, and if you look at a whole population of cells, it may turn out to be that there’s some very interesting property of that population that you can’t see if you just look at an individual cell.
Chris Smith: So how are you doing that because that doesn’t sound like a trivial question to answer?
Demis Hassabis: No, it’s a difficult question to answer in humans with scanners because they’re non-invasive techniques. They obviously don’t involve electrodes or anything like that. And these new imaging techniques, they are relatively quite gross resolution, so, you know, you’re seeing maybe at best 10,000 neurons as little pixels. We have to come up with new techniques to extract the information that might exist in those brain scans that we’ve borrowed from computer science and engineering; machine-learning techniques or pattern recognition techniques, you could call them.
Chris Smith: So what did you actually do?
Demis Hassabis: So what we did was we set up a virtual reality environment which looked very much like a very simple block environment to reduce the amount of visual distractions there were in the environment, and we got people to play a very simple game. What they had to do in this virtual reality environment was to navigate as quickly as possible between four positions which were labelled A, B, C and D, and they were randomly told to go to a new position and then they would have to navigate using a keypad as quickly as possible to that new position.
Chris Smith: And what were you recording when they were doing that? Obviously, you’ve got them in the brain scanner. What are you actually looking for in their brains when they’re doing this?
Demis Hassabis: So what we do is when they get to the destination position, during that point in time obviously brain scans were being recorded. After the scanning was finished, we’d take those brain scans, put them into a computer and show the computer algorithms several examples of somebody’s brain scans when they were standing in a particular position, let’s say Position A, and what these algorithms do is they learn to find patterns of activity in those brain scans that are common across all examples they’re given of somebody standing in Position A.
Chris Smith: So, in other words, by doing it enough times and looking at the nerve activity enough times you begin to see the pattern rising above the noise?
Demis Hassabis: Exactly, a pattern emerges. So even though there’s quite a lot of noise because, you know, you can’t control everything about their thought even though they’re standing in the same position, but despite that what we found quite surprisingly is that these machine-learning algorithms are sensitive enough, given enough examples, and it took about 12 examples of each position, that was enough to train up these sophisticated algorithms so that they could recognise common patterns in the hippocampus of activity.
Chris Smith: So what you’re basically saying is that you could read someone’s brainwaves and know which room they were standing in just by looking at the pattern of brain activity.
Demis Hassabis: That’s right, so once you’ve trained these algorithms up with these examples, then what you can do is you can then present these algorithms with new brain scans that they haven’t seen before, and you can then ask the algorithm, right, which room do you think the person was standing in while this brain scan was taken and, furthermore, which position in the room were they standing in. And it turns out that, basically, we can very accurately determine that.
Chris Smith: And what about in terms of what insights this gives us into how the brain is storing the information in the first place? So you’ve taken this from small clusters of cells at most to whole populations of cells. Does it give us any insights into how the brain stores this kind of information?
Demis Hassabis: Yes, I think it does. I mean in order for these algorithms to work they need quite a lot of data to work on, which means quite a lot of the brain scan must be usable for it, and we found that the actual important areas of the brain scan are actually hundreds of voxels in size. If you actually work out what that translates to in number of neurons, then we’re talking about two to five million neurons are actually coding for any one location. So it seems like from that that therefore the neural code or the population size representing a particular spatial position must be quite large.
Chris Smith: And those linked groups of nerve cells, rather than being spread out over a wide area of the brain, seem to be clustered together in groups, which is giving researchers entirely new insights into how spatial memories and perhaps other memories might actually be stored. That was Demis Hassabis from the Wellcome Trust Centre for Neuro Imaging, and he’s published that work in the journal, Current Biology, this week.
Now from mental images to maintaining an immaculate image; self-healing materials are still waiting to hit the shelves but the latest discovery needs nothing more than a bit of sunlight to get it going. This is the work of US-based scientist, Marek Urban.
He’s taken a chemical called chitosan which is related to the substance chitin which forms the shells of animals like crabs and lobsters, and he’s linked that to a square-shaped molecule called an oxetane ring. This mixture is blended into polyurethane, the same chemical that’s used to make paints and surface coatings at the moment. But when the surface is scratched, the damage causes that oxetane ring to break open which activates the chemicals and makes them much more reactive, and if UV light is then shone on to the injured area, it triggers the activated molecules to link together, pulling the wound closed and repairing the damage.
Marek Urban: What we’ve done, we’ve essentially picked up a few components of known chemistries and put this together in a thoughtful way that allows us to essentially scratch the surface, and upon exposure of the surface to sunlight, or to ultraviolet light, that scratch that was mechanically created will be able to repair itself.
Chris Smith: What’s in the mixture? What’s the chemistry that’s going on here?
Marek Urban: It’s a very simple approach where we’ve taken known polyurethanes, which are utilised in a variety of applications ranging from automotive paints all the way to biomedical devices to polyurethane forms, and we’ve incorporated it into those networks certain chemical entities which are able to break apart upon mechanical damage, and one of those chemicals is chitosan. It’s a derivative of chitin, and that chitin can be found in crab shells and shrimp shells.
That chitosan incorporated into polyurethane network, which was prior modified with highly reactive oxetane, which has like four bonds forming a square with one oxygen in one corner and the other corner attached to the chitosan; that modification allows opening those highly strained bonds of that square to break apart and consequently create reactive groups that, upon exposure to the sunlight or ultraviolet light, are able to react back and consequently the wound that heals.
Chris Smith: So what you’re saying is that when the surface, this material, is damaged, the damage physically breaks open this ring of this square structure which is under strain or tension, and because it breaks open, it then becomes chemically active and can then begin to react with the other constituents of the polyurethanes which will effectively cause the wound that’s been created, the piece of damage, to heal again?
Marek Urban: That is correct. Of course, we still need to learn a lot about specific chemistries and mechanisms responsible for that healing process, but it’s a really fascinating process because without those components, chitosan and oxetane, polyurethanes won’t be able to self-repair.
Chris Smith: So if you watch the damage repairing itself, say down a microscope, how seamless is the repair of the surface and how long does it take to happen?
Marek Urban: First of all, let me say this, that the concentration levels of those covalently attached additives is relatively small, that’s one requirement, but they are very sufficient to be able to repair the network within let’s say 30, 40 minutes upon exposure to ultraviolet source.
Chris Smith: And what does the ultraviolet do to make the repair happen?
Marek Urban: Ultraviolet light essentially generates another reactive site, which in turn reacts with those that opened up as a result of damage, and consequently reacts with other constituents of the network, therefore self-repairing the entire network.
Chris Smith: And is the amount of ultraviolet that you need to trigger this repair reaction, is there sufficient of it in sunlight for this to happen naturally?
Marek Urban: It will just take a little longer. So typical experiments we’ve done are really going from zero to about 30 minutes, exposing to ultraviolet light, and that specific portion of the ultraviolet light that we utilise, which was about 300 nanometres, the energy density of that source corresponds to the energy density of the sunlight. So we are confident this will take a little longer, but it will obviously self-repair.
Chris Smith: And how would you see this actually being used? Could you just work this into standard paints and surface coatings that are used on, say, cars now so that you could have a car that would, if it got scratched, would repair itself?
Marek Urban: There is no reason why it shouldn’t work. If you look at a typical example of, let’s say, a four-layer automotive paint, we tested damages in the range of about 50 microns which is about the thickness of two layers on the top. But there are other applications. I could envisage even electronic devices. Let’s say you take an iPod, you scratch it, or laptop computer, and that scratch is kind of permanent, you expose to sunlight and that will self-repair. Of course, it may take a little longer but, as a user, I would be perfectly happy if that repair takes even a week, as long as it disappears.
Chris Smith: So in the future used cars might not look quite so used after all. That was Marek Urban from the University of Southern Mississippi in the US. He’s published that work in this week’s edition of Science.
You’re listening to Breaking Science with me, Chris Smith, and time now for this week’s ‘Stuff and Non-Science’, where we massacre myths and bash bad science, and managing to contain herself, thankfully, here’s Diana O’Carroll.
Diana O’Carroll: ‘Stuff and Non-Science’ this week is about the exploding spaceman. If you go out without your space suit can you expect to explode, snap-freeze or boil? With the answer, here’s Dr Stephen Juan.
Dr Stephen Juan: Yes, well, in the 1981 Outland that starred Sean Connery there was a construction worker who had a torn space suit that leaked, and he swelled up and exploded, and then Keir Dullea in the classic 2001: A Space Odyssey, he blows himself into the airlock from a pod without a helmet but he doesn’t blow up, and the reason is because the air pressure would drop and the human diaphragm could not pull hard enough for normal respiration.
You wouldn’t explode; your difficulty would be with anoxia (no oxygen), a difficulty in breathing, it would be harder and harder for you to inhale. You could probably last for a few seconds, but you may not even get any trouble until about a minute. And probably in outer space your real exposure would be to the skin, you’d get a sunburn, and if anything is going to boil up, rather than your blood, it would be your saliva.
Diana O’Carroll: Still, could be a new way to fry an egg. That was Oddbody author, Dr Stephen Juan, from Sydney University. If you know of any more science myths, then email me with them and that’s firstname.lastname@example.org.
Chris Smith: So not all that dramatic if you are naked in space. The worst you can hope for might be a rather profound suntan. Thank you, Diana. That was Diana O’Carroll with this week’s ‘Stuff and Non-Science’. That’s it for this week.
We’ll be back next week with the final edition of Breaking Science. That’s right, the series is almost over, but in recognition of the fact that it’s the last show, we’ll bringing you a special programme from the Open University’s Darwin Lectures that are taking place in London this week.
In the meantime, don’t forget that Breaking Science is produced in association with the Open University and that means that you can follow up on any of the items that are covered in the programme via the OU’s website. That’s at open2.net/breakingscience. The other way to get there is to follow the links from the BBC Radio Five Live Up All Night web pages.
The production this week was by Diana O’Carroll from thenakedscientists.com, and I’m Chris Smith. Until next time, goodbye.
Does it sound good? Why not get the Breaking Science podcast, and receive new episodes as they're released?
These are the sources used by the Naked Scientists in making the show:
In the news
'Regulation of embryonic cell adhesion by the prion protein'
by Malaga-Trillo E, (2009)
in PLoS Biology volume 7(3): e1000055.
'Advanced paternal age is associated with impaired neurocognitive outcomes during infancy and childhood'
by Saha S et al. (2009)
in PLoS Medicine volume 6: e1000040
'Eye color and the prediction of complex phenotypes from genotypes'
by Liu F et al (2009)
in Current Biology volume 19, p5
'Enhanced Excitatory Transmission at Cortical Synapses as the Basis for Facilitated Spreading Depression in CaV2.1 Knockin Migraine Mice'
by Tottene et al (2009)
in Neuron 61, 762–773
Demis Hassabis on 'Decoding Neuronal Ensembles in the Human Hippocampus', by Maguire, Hassabis et al. in Current Biology 19, 7, April 14, 2009
Marek Urban on 'Self-Repairing Oxetane-Substituted Chitosan Polyurethane Networks', by Biswajit Ghosh and Marek W. Urban in Science
Stephen Juan for 'Stuff and Non-Science'