Clever mice and drunken flies

How do mice learn tasks like playing the piano? And why do flies drink after a hangover?
14 January 2015
Presented by Kat Arney


How do we learn complex tasks like playing the piano? Why can we remember things better after a good night's sleep? And why do people - and fruit flies - drink again after the hangover from hell? The answers are all in your genes. Plus, why large-scale searches for so-called "genes for schizophrenia" and other psychiatric diseases are turning out to be trickier than we thought, and a gene of the month with a touch of Scottish - or maybe Hollywood - spirit.

In this episode

Beethoven playing the piano

01:12 - Bill Richardson - Piano-playing mice

Bill Richardson is investigating how myelin, which coats our nerves, is involved in learning and memory formation for complex tasks.

Bill Richardson - Piano-playing mice
with Bill Richardson, UCL

Kat - This month I'm reporting back again from the Genetics Society autumn meeting, held at the Royal Society at the end of November, which focused on the genetics and neurobiology of learning and memory. One of the most interesting talks, if only for the cute videos of mice running round on wheels, was from Bill Richardson from University College London. He's investigating the role of myelin - the fatty white insulating material that coats our nerves - and how it's involved in learning.

Bill - We and others have discovered relatively recently that new myelin continues to be formed in the brain of a mouse long after embryonic development and early postnatal development into young adulthood. That was quite unexpected because you would expect for the brain to work normally that everything should be set up early in order to be able to be programmed later.

Kat - You kind of think you've made enough, you don't need to make anymore.

Bill - Exactly, yeah. By finding that it continues to be made from these immature cells called oligodendrocyte precursor cells that raised the question of what is the function of the newly adult born myelin. Also, there's an increasing body of evidence in humans that people who learn complex motor tasks - that means complex sequences of movements like juggling or using the abacus or playing the piano - they develop changes, structural changes in their brain which are not necessarily to do with neurons.

They have structural changes which can be detected by MRI in the white matter tracts. The white matter is those parts of the brain which are called the information superhighways. They don't contain the neurons themselves. They just contain the connections between the neurons. So, it's where all the traffic occurs. Those superhighways contain myelin. And so, this was another link suggesting that new myelin might be formed in response to training or learning.

Kat - So, how do you go about discovering if making new roads in this superhighway for myelin is the key to learning?

Bill - Well, it's a very simple concept. We try and prevent the formation of new myelin and see if that impacts the ability of a mouse to learn complex motor task.

Kat - Now, a mouse can't learn to play the piano. So, how do you challenge a mouse to learn something complicated?

Bill - Well, we'd like the mice to play the piano, but we devised a task which is a little akin to playing the piano for a mouse, which to take a running wheel with rungs and remove some of the rungs so that it's irregularly spaced rungs. We call that the complex wheel.

Kat - So, it has to kind of figure out, "Ooh! Miss one, miss two, miss a few."

Bill - That's correct! It's like running up a staircase and some of the steps are taller than others and you have to remember where the trip steps are, so to speak. For some reason that I don't understand, they seem to enjoy it. So, you don't have to give them a reward. They will persevere and become very proficient at running on those wheels.

Kat - Just for the joy of running.

Bill - Just for the joy of running, yes.

Kat - So then how do you test whether it's this myelin growth that's involved in their learning?

Bill - We have to somehow prevent the production of new myelin and see if that affects the ability of the mice to learn to run on this complex wheel. So, how do we affect the production of myelin? Well, that's where the genetics comes in. We used genetic tricks to delete a specific transcription factor, a crucial protein or the gene encoding that protein that is required to elaborate the structure which is known as myelin. That genetic switch that we introduce into the mice is activated by a drug, tamoxifen, so that if we inject tamoxifen into the mice, it doesn't affect their pre-existing myelin, but it prevents any new myelin from forming after that point.

Kat - So, what happens then if you get a mouse and you treat it in this way and you try to make it learn something?

Bill - So, we injected tamoxifen, we showed that we can effectively prevent 95 per cent of the new myelin forming and then we give them the complex will and find that they're pretty poor at learning the complex will compared to their littermates that still have the gene.

Kat - So, that seems to be pretty strong confirmation that it is the growth of this new myelin that's involved in their learning.

Bill - It's a very good clue, but of course, that on its own, that one experiment doesn't necessarily tell us it's learning per se. It could be that knocking at that gene has unexpected effects on fitness. It may have reduced cardiovascular tone or reduced muscle tone. So, we have to do a control, and the control was to allow the mice to learn to run on the wheel before we knocked out the gene. When we did that, of course, during the learning phase, while the gene was still present, they could learn normally like their wild type or normal littermates. Then we knocked out the gene and put them back on the wheel and they were now able to remember what they had previously learned. And that shows that they don't need new myelin to remember something that they previously learned, only to lay down the new skill.

Kat - I mean, this makes sense if you use the analogy of a highway - once you've laid down the tarmac, it's there.

Bill - Well, that's a pretty good analogy. We've also started to look at other learning and memory tasks. Still sticking with motor learning, we're training mice to pick up food pellets with one hand which they can do. Pretty clumsy at first, but they get better at that. After training with one hand, we can detect cellular changes in the white matter and the grey matter on the contralateral side - the opposite side of the brain to the hand that they're using because as we know, the pathways in the brain cross. That's rather useful because it means that we can compare the trained side of the brain with the untrained side and that would comprise a very nice internal control. We can compare one side of the same mouse with its opposite side if you see what I mean.

Kat - So now, we have these really nice clues that the growth of new myelin is really important for learning. This is in mice. How do we know that similar pathways may be at work in humans when we're learning things like playing the piano or running on a complicated wheel?

Bill - There is MRI evidence that there are structural changes in the white matter - the information superhighways. And that's what originally got us thinking about this whole question of adaptive myelination. But at the same time, it has recently been shown or reported at least that myelin production in humans does not extend throughout adult life as it does in mice. Most myelin production in humans is completed by age 10 or so. Although there is a small trickle that carries on into young adulthood into the early 20s. Maybe learning new tasks requires a relatively small proportional change in the myelin which cannot be detected so far.

Kat - So, could this explain how old dogs can learn new tricks that you do still need a little bit of myelin when you're older?

Bill - Dogs do learn new tricks but much more slowly than they would have done if they were younger and everyday experience tells us the same thing about humans. Another thing about myelin is, since it's a sheath that wraps around the axon, it has a kind of protective role and it preserves the circuit. As we know, if you learn to ride a bicycle, you can put the bicycle in the shed for maybe 20, 30 years and get it out at the end of that time and you can still ride the bicycle. So, motor skills are a very long lived thing. That also smacks of some kind of structural support by myelin in my opinion.

Kat - I always find it amazing. I'm a musician and sometimes if I haven't played a piece for ages, I sit down and I play and it's there.

Bill - Well, the very first MRI studies in humans were to do with pianists in fact, showing that the amount of time a professional musician practised before age 11 in fact was reflected in the amount of myelin in his or her white matter.

Kat - That was Bill Richardson, from UCL.

09:53 - Matt Jones - Sleeping and learning

Why does 'sleeping on it' help us learn and solve problems? Matt Jones, from Bristol University, is finding out.

Matt Jones - Sleeping and learning
with Matt Jones, Bristol University

Kat - You've probably heard the phrase "sleep on it", to help you solve a problem. And it's often said by students that a good night's sleep after revising is better than an all-nighter of cramming. But why? One person who's trying to figure out how sleep helps us learn, and how sleep problems might be involved in brain diseases such as schizophrenia, is Matt Jones from Bristol University, who also helped to organise the Genetics Society Autumn meeting.

Matt - The man on the street will acknowledge that sleep is important. We all feel a bit rough if our sleep is disrupted which obviously has, a long time ago, raised the question exactly, what is sleep for. It's clear, based on what's going on in the brain during sleep, that it's not just a time of rest. We're not just recuperating. We're also sorting through, in our sleeping minds, the wheat from the chaff and deciding which information to hold on to, to consolidate and which information perhaps is not so important and we can choose not to remember in the long term.

Kat - And there was kind of a bit of a thing at university. You'd say, if you're studying for an exam, do some revision then 'sleep on it' and you'll know it better the next day.

Matt - That's right. there's well-controlled evidence that good night sleep will improve your performance subsequently. So, we should all strive to sleep more and better I think.

Kat - How do you try and study the impact of sleep on learning? What are you doing?

Matt - We're using a translational approach. So, we're studying sleep in rodent models - in rats and mice in which we can study brain activity during sleep in great detail, trying to understand how information is processed in the brain during sleep in these animals. And then at a more clinical extreme, we're studying patient groups who have disrupted sleep and trying to understand how their disrupted sleep impacts on their symptoms and in particular, in memory impairment. So to date, we focused on Schizophrenia which is a disease that if you ask a psychiatrist, they'll be quite dismissive about the role of sleep disruption in that disease. But if you ask a patient, they often cite bad sleep is having a major impact on their quality of life. So, we've seen in one cohort, patients suffering Schizophrenia that their brain activity during sleep is abnormal. So, it's not as well coordinated as it is in healthy volunteers and that the extensive disruption in brain activity during sleep correlates with the extent of memory impairment. So, this obviously offers an opportunity to intervene, to try and normalise brain activity during sleep in Schizophrenia and hopefully, show that has a beneficial effect for patients.

Kat - So, how are you trying to unpick maybe what's going on at a deeper level in the kind of the nerve cells and the molecules that are involved as we learn when we sleep?

Matt - In those experiments in animals where we can record from individual identified nerve cells and track their activity both during learning so, during waking behaviour and during sleep. There's the possibility for us to capture those cells, control their activity. So for example, you could bias the kind of information that animals store in long term memory by activating particular groups of nerve cells during sleep. And so, it's a sort of an interesting experiment. Our goal is not to control the minds of rats but it offers us insights as to what is going on in the human mind during sleep.

Kat - So, by stimulating different bits of a rat's brain while it sleeps, you can kind of make it remember something more or make it forget something that it really needs to remember.

Matt - Yeah, exactly. And so, you can immediately think all sorts of nefarious military motivated experiments that the CIA might be interested in. but I can assure you, we're not funded by them!

Kat - What do you still need to find out about what's going on when we sleep? What are your your known unknowns?

Matt - One of the big issues in the field remains what's going on during different stages of sleep. So, as most people know, we have different stages of sleep that we cycle through during the course of the nights Most famously, we have REM - Rapid Eye Movement sleep - and non-REM sleep. It's still not clear what those two different stages are contributing, particularly in the context of learning and memory. So, if someone chose to give me a few million quid and said, "Go and do what you want", I would try and focus on what's REM doing, what's non-REM doing and how do the two stages of sleep work together to finesse learning and memory.

Kat - And you've presented your results here at the Genetics Society Autumn meeting. What do we know about how genetics or genetic variations affect this sort of sleep and memory issue?

Matt - Well, I certainly don't know enough, obviously. In relation to the Schizophrenia work we're recruiting healthy volunteers at the moment on the basis of their genotype, that's at loci that are related to risk psychiatric disease. So, we're trying to use that approach to understand whether sleep disruption earlier on in the disease can exacerbate symptoms. But one appeal of sleep is that it's quite easy to measure at a population levels. So, lots of people wear wristbands that might monitor their movement for example. If hundreds of thousands of people are doing that, if we can capture those data and relate them to genotype, perhaps we can begin to tie these things together.

Kat - Matt Jones from Bristol University.

Alcohol bottles

15:06 - Karla Kaun - Hungover flies

Why do fruit flies forget how terrible their hangover was and head straight back to the pub? And why do we? Karla Kaun is after answers.

Karla Kaun - Hungover flies
with Karla Kaun, Brown University

Kat - I'm sure this has never happened to you, but I bet you know someone who's been for a big night out on the booze, woken up feeling awful and swearing they'll never drink again, only to be back on the sauce a few days later. It turns out that forgetting the horrible effects of drinking is something we share with tiny fruit flies. Karla Kaun, from Brown University in the USA, is training flies to associate a particular smell with an alcoholic tipple and figuring out how they switch between so-called aversive memory - when they recall how bad they feel - to appetitive memory, when they can't wait to get back to the fly pub.

Karla - So, what I study is memories for the intoxication experience. What I found is that even in flies, the initial effects of alcohol are always aversive, something like a hangover effect. But what's intriguing is, like going to the bar on a Friday and Saturday morning not feeling well, but wanting to go out again Saturday night, the long lasting effects of alcohol are appetitive. And what I'm interested in figuring out is, how are neural circuits mediating this effect and what the molecular mechanisms acting in these circuits to affect them.

Kat - Obviously, fruit flies don't go out to the pub, what's fruit flies' relationship with alcohol normally like? Are they big drinkers?

Karla - So, flies, like humans have a long natural history with alcohol. So, humans have been consuming alcohol for centuries and flies actually spend a good portion of their life and little concentrations of alcohol. They lay their eggs in fermenting fruit and in fermenting fruit of course are patches of moderate concentrations of alcohol. So, as larvae, they eat this alcohol and it's evolutionary advantageous to them. For example, a larvae that has a 6 per cent alcohol concentration in its body is less likely to be parasitised by wasps. So, what's very interesting also is that flies' effects to higher concentrations of alcohol are remarkably behaviourally similar to those in humans. They go through the same stages. At first, you get disinhibition and then you get a loss of locomotor coordination and then the flies will just pass out in the bottom of the vial. The time it takes for him to recover is almost the same as the time it would take us to recover.

Kat - How long does it take a fly to kind of go from, "I'm never doing that again" to "Yeah, let's have another drink.?

Karla - Well, with the parameters I tested, it's somewhere between 12 and 15 hours.

Kat - That's pretty good going. So, the next time that flies encounter alcohol, they think, "Yeah, this is great! Let's party!"

Karla - Pretty much, yeah. They find the reward long lasting, so it lasts up to seven days which is a long time for a memory for a fly. They'll also walk over a 120 volt electric shock to get to the odour that was previously paired with alcohol. So, this suggests that it's an extremely appetitive and intense memory.

Kat - So, I'm sure some of us would deeply sympathise with that. But tell me then, you're trying to understand what's going on at a molecular level. What have you found out so far about how they form these different types of memories, the kind of, "Oh my God, no" and then the, "Yeah, let's do it again"?

Karla - So, what I think is happening is you're getting parallel circuits that encode both the aversive memory and the appetitive memory, and then you get feedback between these circuits. The appetitive memory can turn off the aversive memory circuit for example or potentially vice versa if something goes wrong. I'm interested in figuring out what inside of these neurons is affecting it and one of the molecules that I looked at is a regulator in the Notch signalling pathway. Notch is really important for maintaining long term memory. So, we're trying to figure out how alcohol affects this to potentially result in aberrant memory formation.

Kat - So, Notch is one of these kind of signals that tells cells what to do, what kind of cell to be, what to get up to.

Karla - Pretty much. It's a cell-cell signalling molecule. so, it's like an early signal that affects a lot of downstream effectors.

Kat - So now, you're kind of starting to understand some of the molecules that are involved in these different types of memory and how the, "Yes, let's do it" overwrites the, "Oh God! Never again" kind of thing. How are you going forward with this and do you think that there may be similar mechanisms at work in human brains?

Karla - So, I absolutely do think there are similar mechanisms affecting human brains. Notch signalling is actually one of the most well-conserved signalling pathways. It's been studied for over 100 years in flies now. And biochemically, things work extremely similarly. I'm interested in figuring out how to target different types of Notch signalling changes to different kinds of brain regions. I think this will be extremely informative for developing pharmacological treatments. There are currently treatments for cancer being developed for drugs in the Notch signalling pathway and I'm hoping that I'll be able to perhaps use some of these drugs to treat addiction-related disorders.

Kat - Could you make an anti-hangover pill?

Karla - I'll tell you what, I'll work on it. More realistically, what we'll probably try to do is decrease the really strong appetitive memories by enhancing the aversive memories, so that people don't crave alcohol quite as strongly.

Kat - That was Karla Kaun, from Brown University in the States.

20:17 - Danielle Posthuma - Searching for genes

Danielle Posthuma is taking a new approach in the search for genes involved in psychiatric disorders such as schizophrenia.

Danielle Posthuma - Searching for genes
with Danielle Posthuma, University of Amsterdam

Kat - You're listening to the Naked Genetics podcast with me, Dr Kat Arney. Still to come, we'll be meeting our bold and brave gene of the month. But now it's time to return to the Genetics Society autumn meeting, where we heard an intriguing talk from Danielle Posthuma, from the University of Amsterdam in the Netherlands. For many years, scientists have been searching for genes involved in psychiatric diseases, such as schizophrenia and depression, but although many gene variations have been found, each of them seems to have a tiny effect on the risk of developing an illness. Yet we know that a significant chunk of the risk for these conditions must be in our DNA somewhere. Danielle's trying to look at this problem in new ways, by studying whole networks of genes rather than single suspects, and by using an intriguing new technique based on reprogrammed stem cells made from adult cells, known as induced pluripotent stem cells.

Danielle - This all started I think 30, 40 years ago when we said, if traits are heritable then we should be able to find genes for a trait. And then so, we are selecting one gene and then trying to find association with that gene and a trait.

Kat - So, kind of the usual suspects.

Danielle - Right.

Kat - This should be important, let's see if it links up to this behaviour.

Danielle - Yeah, exactly. That didn't really yield a lot of reliable associations and then a couple of years ago, we were finally able to do this at a genome wide scale - thanks to the development of micro-arrays. So, we started searching for genes for various disorders by scanning the whole genome. So, that's what many people have been doing for the last 8 years or so.

Kat - I see quite a lot of headlines, scientist find a hundred genes for autism, they find a hundred genes for Schizophrenia, these kind of studies.

Danielle - Yeah. So, that's what's really happening. I mean, that's all very exciting, but what we have to keep in mind is that the effects of those genes are very small and that together, they explain very little of those heritability of traits. So, that means that there are more genes that we still have to discover and those genes will have smaller effects than the genes that we currently have identified.

Kat - We've actually found the biggest ones already and they're not very big.

Danielle - Yeah, so it will be increasingly smaller, the effect sizes of the genes.

Kat - Are there missing genes or are we just thinking about how these genes work in the wrong way?

Danielle - Well, I think both. So, we are missing genetic variants that are very rare for example because these are very difficult to detect and they might have a large effect. But we need different strategies and different genotyping for that. But there also might be different statistical strategies that we have to employ in order to find common variants such as gene set analysis for example.

Kat - What do you mean by gene set analysis?

Danielle - Yeah, so what we've been doing mostly up until now is to determine the effects of every single SNP, every genetic variant at a time.

Kat - So, each single one.

Danielle - Yeah. What we would like to do is to say, "Well, we're not really interested in the single variant effects, but we are interested in all of the variants that are related to this and this particular pathway." For example, the dopamine pathway, we have a certain idea which genes are involved in that pathway. So, we can select all the variants that are important for that pathway and then test the effect of those variants as a group instead of looking at the single effects. And that will increase our effect size and it will also be more easy to interpret.

Kat - So, instead of just going, okay, this is single thing here, single thing here, you're saying, as a whole, all the genes involved in this kind of thing, are they important? What do you find when you take that kind of approach?

Danielle - So, we and others have taken this approach and we were able to find certain synaptic pathways that were associated with Schizophrenia for example, and we also found a specific gene set associated with IQ. We weren't able to detect any of those genes if we hadn't done a gene set analysis. So by themselves, these genes were not strongly enough associated with the trait. But when we looked at them in their context, in their functional genetic context, we were able to associate them with the trait.

Kat - So, apart from doing maybe bigger and bigger ever studies or this kind of analysis of lots of genes all bundled together, how else can we try and understand perhaps what some of these variations actually do? Because sometimes it feels to me, we've collected a lot of stamps, but we don't know how they work.

Danielle - Yeah. I think that's an important next step to find out how does it actually work. And so, gene set analysis might be one step in the right direction, but it might point us towards the important pathways, but it doesn't really tell us how things work. So, what we need is functional genomic follow up studies and we need molecular biologists who will look at our findings and to design experiments where they can actually manipulate the gene or the set of genes and look at their effect on a cellular level.

Kat - I guess the problem with some of these variations that we found is they're not necessarily in genes. Also, they're in humans, a lot of molecular biologists work in animal models. Tell me about the way that you're starting to move towards trying to understand these in cells?

Danielle - Yeah, so that's true. I mean, not all the results that we get are directly available for use in functional genomics experiments. So, one of the recent developments in biology is IPSC - induced pluripotent stem cells.

Kat - These are the 'turn back the clock' cells.

Danielle - Yeah. So, those are cells that you can take for example from a hair or from the skin and you reprogramme into an embryonic state, and then you can differentiate them to any kind of cell you like.

Kat - They're like magic!

Danielle - Yeah. I also think it's magic because if I can't do it myself, my colleagues do this. So, you can reprogramme cells and differentiate them from patients and controls and then you can for example, select patients that have a whole bunch of risk factors for Schizophrenia or another disease. So, you don't need one gene and you don't even have to know the function of the genes, and your findings don't even have to be inside genes. You simply select people based on their genotypic array and you do this for patients and controls. And then you differentiate their cells into neurons and use cellular assays to look at different phenotypes of the cell. And then hopefully, you'll find some differences which will tell you something about how the cells function in patients and controls.

Kat - So, you're almost making a model organism from an individual patient.

Danielle - Yeah.

Kat - What are some of the things that these studies are starting to show? It seems so exciting to me.

Danielle - Yeah, I think it's a very exciting era that we live in. I really like being part of this although I can do part of this myself and the other part, I need other people to collaborate with. So, science at least in my field is no longer something that's very individual. You have to collaborate with people from your own field to increase your sample size, but also, from other fields to increase your knowledge of what you're investigating. So, I think it's very nice.

Kat - And with these reprogrammed cells, what sort of results have come out so far? I'm aware it's still at the very early stages.

Danielle - Yeah, so there have been some initial studies and these used one or two patients or one or two controls. So, this is a very small scale but they were published in a very good journal because these were the first ones to do this. For Schizophrenia for example, they found differences in synaptic pathways between cells from patients and controls. But these studies, they do need replication because it's N or one or N of two studies. So, we do need larger samples for these kinds of studies.

Kat - That was Danielle Posthuma from the University of Amsterdam.


28:11 - Gene of the month - Braveheart

Our gene of the month sounds like a Scottish warrior - it's Braveheart

Gene of the month - Braveheart
with Kat Arney

And finally it's time for our gene of the month, and this time it's Braveheart. Nothing to do with Scottish warriors, Mel Gibson or the allegedly most historically inaccurate film ever made, Braveheart is a mouse gene. But it's one with a difference. Many genes carry the instructions that tell cells to make a particular protein - these are called protein coding genes. The DNA of the gene is 'read' to make an intermediate message, called RNA, which then acts as a kind of molecular recipe that the cell's factories use to build the right protein. But now there's a growing number of genes we know about that don't make proteins. Instead, the RNA they produce, known as non-coding RNA, is useful to the cell in different ways, for example by helping to switch genes on and off. Braveheart is one of these non-coding RNAs, and is needed to help turn early embryonic cells into heart cells during the early stages of development in the womb. No kilt required.


Add a comment