eLife Episode 54: Dodgy cells and big neurons
Why one in five published papers that use cultured cells may be wrong, the frog that sings underwater without air, genes that make you live longer, seeing evolution through bats' eyes, and do brainier people have bigger brain cells? Join Chris Smith as he talks to the authors of five hard-hitting new papers published in eLife...
In this episode
00:32 - Mistaken identity: contaminated cell lines
Mistaken identity: contaminated cell lines
Anita Bandrowski, University of California San Diego
Up to 20% of the papers that have used cultured cells to study basic biology and disease processes like cancer may have drawn incorrect conclusions, new research is showing. This is because the cell lines used by the authors of these publications may not have been what they thought they were. Speaking to Chris Smith, UCSD scientist Anita Bandrowski has been using a machine learning approach to sift through millions of publications. In the process she’s found potentially hundreds of thousands of papers that may now need looking at…
Anita - We want to make sure that what we're doing is what we think we're doing. But, sometimes, we can be fooled by bad reagents; and this is one of those things we can't really control. If the real cell line that we're using is one type of cancer and we think that it's another type of cancer, because that's what we've been told by the stock centere, it's a big deal, because we're actually testing the wrong cancer!
Chris - Does that seriously happen though, Anita?
Anita - Yes absolutely! I mean the hep G2 cell line is one of the most commonly used liver cell lines. And if you just use it as, you know, some kind of a liver cell great, you can do that. But if you don't know that it's a hepatocarcinoma versus a hepatocytoma, you're looking at actually the wrong kind of cancer!
Chris - And how common are these sorts of mistakes? If you look at the field, how often do you think this is happening?
Anita - So, before our paper, there have been multiple estimates as high as 60 to 70 per cent at certain stock centres of cell lines that were actually contaminated or misidentified. Our estimate, I believe, is the most accurate estimate - it's certainly the largest sample size. And we looked at it and we said OK, sixteen point one per cent of the papers out there that use these kind of cell lines are using one of these cell lines that happens to be either misidentified - so people might think that it's the wrong cell line; it had some contamination - either it's partially contaminated or fully contaminated.
Chris - So that means, potentially, if you're right that between one in six one in five papers to have used cell have potentially drawn invalid or incorrect conclusions because they're not the cells I think they've used?
Anita - That is correct. And so Christopher Korch, one of the leaders in this field, has called for - you know - tens of thousands of papers to be looked at again. He was one of the reviewers on our paper and was very harsh, trying to ask us "what are you gonna do with this data? How are we going to address this field moving forward?"
Chris - So how did you do this. What was your method?
Anita - We actually did a method called "text mining". So we took the papers, we looked only at the method sections, which is where authors say this is the recipe of what I did including the reagents that are being used including the cell line. So we looked at all the sentences that said "I used a hep G2 cell line", or some other cell line HeLa cell line.
Chris - Was this a machine that was doing this for you. And how many papers did you feed into it?
Anita - So we fed in about 2 million papers. So this is everything in the entire open access corpus including all of the eLife papers, including all of the PLoS papers; the algorithm was actually trained to look for sentences that had these different cell lines. And we simply matched them. We said, OK here is a nice list of the things that could be potentially bad - potentially contaminated. And we matched them up, and turns out that it's about sixteen point one per cent of papers that are potentially affected by this problem.
Chris - Now did you - when you were testing your algorithm - did you go and scrutinise to see how sensitive and specific the filter was, to make sure that when you subjected these 2 million journal papers to this filter it wasn't mis-calling them?
Anita - Yes. So we we actually tested an accuracy rate of about 95 per cent.
Chris - This puts you in an interesting position and doesn't it, because you're sitting on a huge amount of data which you can point the finger at certain papers and say "well, these might not be reliable". What are you going to do with this data?
Anita - That is exactly what Christopher Korch asked us. And that's a really tricky question, because I don't have enough manpower - enough muscle - to kind of go through tens of thousands of papers that are generally problematic. Certainly I would not want to drag the reputations of any scientists through the mud unnecessarily. However, it should be one of those things that, at least moving forward, people should be able to look at this.
Chris - Presumably, you've got the raw data from your analysis and that will be available. So, if I were to come along with a paper I'm thinking of publishing and and basing my results and interpretations on other people's findings, I could go and have a look on your list presumably to see if the papers I'm thinking of citing, or people I'm thinking of collaborating with, are on your watch-list?
Anita - Yeah. And actually one of the other things that we will look at in the near future is actually putting that out into an easily-digestible public place. So as people are thinking about looking up these cell lines, they should be able to find these references.
Chris - But how do you think the scientific community who might be on your list of "naughty cells" react to this? Because there could be some really important papers in there, and some people with very big reputations and even bigger egos, who actually could have quite a lot of beef with what you're about to say about their work?
Anita - Well that's that's what makes this very very tricky right. Because we can't tell very easily just from knowing that somebody is using a cell whether they're doing it appropriately or inappropriately and therefore are their conclusions up to snuff. So this is a very tricky thing. We're going to proceed with a lot of caution. One of the other things that can be done, using text mining, is something called "sentiment analysis". So sentiment analysis is a method in text mining that allows you to start to understand things like "does the paper know that they are using a potentially contaminated cell line?" Christopher Korch's paper will actually be flagged with - you know - with "hey, there's a bunch of contaminated cell lines that appear here!" But, of course, his paper says "hey these are contaminated!" So, if we can tell that a particular valence of the language is actually correct, then we should not bother those people, right. We should only bother the ones that probably either don't know that their cell lines are contaminated, or maybe trim down this this larger list of of tens of thousands of papers to looking at sort of the most egregious potential problems...
07:41 - Frogs singing submerged
Frogs singing submerged
Darcy Kelley and Ursula Kwong-Brown, Columbia University
Now here’s a question - if you breathe air but you live and socialise underwater, and you rely on sounds to attract a mate, how can you sing and not drown? As they explain to Chris Smith, it took the musical ear and tenacity of Ursula Kwong-Brown to convince her boss, Darcy Kelley, that there was something special about the sounds these Xenopus frogs were making, and that led the duo to the answer. Darcy first…
Darcy - For many years, I've studied a frog that started out - as all frogs did - came from the sea and went on to land, and then at some point it went back under water. And it's been a mystery for over 120 years of how they can actually create sounds underwater and how those sounds manage to contain the information that they need in order to communicate with each other socially.
Chris - Why is it a problem for a frog to make a sound underwater?
Darcy - The way we make sounds is we have air flowing over our vocal cords to create these sounds. But when you're underwater you can't afford to breathe or else you'll drown. So how do you manage to make the sounds that are so important in your social life without actually having air move through the vocal organ?
Chris - And what had scientists speculated might be the mechanism? Well there's a long history of it - it started about 120 years ago; there were all these theories: they have rods in there maybe they knocked against the side of the voice box. The most recent one was that the sounds are actually made by the implosion of little teeny bubbles of air like the noises that are made by snapping shrimp or propellers until this paper came out. That was the accepted version of how the sounds were made, but nobody had actually seen these little tiny bubbles. So we set out to see if they actually exist.
Chris - And, Ursula, did you come to Darcy with the idea, or did she find you?
Ursula -That is a long story. My background is in music and and biology. I was sitting in the lab listening to other members analyse recordings of these male advertising calls that I heard a musical interval and I said, "why are you singing a perfect fourth?" - Dun dun - Here comes the bride! That's a perfect fourth. Even more astoundingly, it was a harmonic perfect fourth: they were sounding at the same time, and I had never heard this before in any species. And so that's when I came to Darcy and said "I want to know what this is. I want to know how it happens and also how it matters to the frogs!"
Chris - And Darcy, when Ursula came to you with this did it mean anything to you?
Darcy - Well the first thing we said was, "Ursula, you're on crack!" Seriously, we found it very hard to believe because we couldn't actually hear it ourselves. But Ursula went and taught herself how to code. She showed us. We believed her.
Chris - You got she got a demo, you can play us, Ursula, to demonstrate this for those of us who are not musically minded like you...
Ursula - Yes. So this was the very first frog call in which I heard the musical interval. [Sounds]
Chris - What have we just heard there?
Ursula - So first we heard the frog's advertising call. And then we heard the exact same pitches played on the piano. Dun dun.
Chris - Now what does that mean the frog must be doing. Why did that jump out at you - or croak out at you I suppose I should say - as significant and important?
Ursula - One of the first questions we had to address was whether or not this was real. It is so rare to see two simultaneously-sounding frequencies like this that everyone we brought it to thought, "this is an artifact of the glass tanks you're recording in. There's no way this could be real." It wasn't until we recorded with laser vibrometry, which is measuring vibrations at the surface of the animal, that we could really prove that it was real.
Chris - And is that what you did. You actually went and looked at the apparatus the animals are using to produce these sounds in order to work out how they're doing it and why it's so unusual?
Darcy - Yeah. Because they produce these sounds underwater you can actually isolate the vocal organ and put it in a dish and stimulate the nerves that would normally make the muscles contract and have sounds in the dish. We called this "vox in vitro". And then to study the vibrations we actually went and recruited a scientist who worked on the courtship song of spiders; you know male spiders vibrate their webs and you can turn that into sound by shining light at it and having the light reflected back at you. So that's called laser vibrometry. So we called him up and we said we want to figure this out. He hopped on a plane with his laser and came out. We started doing experiments on the vox in vitro.
Chris - What does that reveal about the mechanism though?
Darcy - It means that it has to be intrinsic to the vocal organ itself. It means that you don't need airflow and it means that the acoustic qualities - these two harmonic intervals - have to be shaped entirely within the larynx.
Ursula - This larynx in a dish is creating these two frequency peaks - this musical interval - and at first we were like, "this is amazing. This is proof that this was not an artifact of the glass tank. This is it. The larynx is making it!" But when we tried to ask what about the larynx is shaping these, it was really difficult to get rid of those frequency peaks. We put large glass beads on top of it, inside of it, drilled holes in the top. Almost nothing made a difference to these frequency peaks. They were so strong until we took a pin and actually poked the elastic cartilage inside, disrupting the central lumen which is normally divided into three chambers and making it into one. And that got rid of the two frequency peaks.
Chris - The intriguing thing you said was you don't need airflow to get this sound because, when I'm speaking to you my vocal folds are opening and closing producing pressure changes which then resonate through my mouth, and those are the vibrations that we call speech. So these animals are clearly not doing that. So describe this apparatus then, with the three chambers, that you had to physically perturb in order to get rid of the sound. What's that actually look like?
Darcy - At the front of the larynx there are these discs - there are pieces of cartilage that are flat, held very tightly together. And when they move apart at a certain key speed, they cause the larynx to vibrate and you can record that vibration either on the surface or as a sound wave. And the key feature that enables the larynx to vibrate in two modes right, with creating simultaneously two different sound pitches, is the cartilage that runs through the larynx that separates the central chamber from the side chambers that surrounds the discs that create the sound. So it's intrinsically vibrating at these two harmonic intervals.
Chris - So how does the sound get out of the frog underwater what's actually vibrating and how does it transmit the sound into the environment? Is it mechanically coupled to its whole body then?
Ursula - It is indeed mechanically coupled to its whole body. I could shine that laser on the frog's littlest toe and I would get the exact same frequency peaks, because it's using the entire body of that frog to radiate that signal.
Chris - And so this blows out of the water the idea that bubbles are collapsing, because you could demonstrate with this series of experiments it's not collapsing bubbles. It is literally the apparatus itself making these vibrations and then turning the whole frog basically into a resonant chamber to get the sound out into the environment?
Darcy - Boy you're great! Yes, exactly! And furthermore we never saw any bubbles and we should have been able to see it in the equipment that we had when we did these experiments. Yeah. So it's it's quite fantastic really; it's a new way of creating sound underwater that's extraordinarily efficiently coupled to the medium.
16:03 - Genes and lifespan
Genes and lifespan
Peter Joshi, University of Edinburgh
Inside each of us are 3 billion letters of DNA that control how our bodies put themselves together in the first place, and how our systems operate day to day. And, as he explains to Chris Smith, what Edinburgh University’s Peter Joshi wanted to understand is how those genetic letters also affect how long we’ll live...
Peter - Each of us has got three billion letters of DNA. And we differ from each other, typically, at about 3 million different places on a genome, which is our collection of DNA. And we were interested in how those differences might affect how long you live relative to how long I might live.
Chris - But that sort of study takes a lifetime to do which you don't have. None of us do so have you approached it?
Peter - That's a really good question. So the problem as you highlight is that if we were to recruit a whole lot of people in their say 40s, take their blood, read their DNA, we would perhaps have to wait 50 or more years until they died to understand how their DNA had affected how long they lived. What we realised was, if we take your DNA, say, Chris, we could ask you how long your parents or your grandparents had lived and because your DNA is of course their DNA we're able to understand how their DNA actually affected how long they lived.
Chris - So the obvious approach here is that because I'm the product of the mixing of the genes of both parents, I therefore am a proxy for what's going on in my parents, so if we find out what happens to them we can make a reasonable prediction about what genetically is driving that?
Peter - That's right. Genetically, you're half your mum and half your dad and we can use that - and a little bit of clever statistics - to make fairly precise probabilistic inferences about what your mum and dad's DNA are.
Chris - So talk us through then how you actually did this. What was the resource you used in order to get access to the sort of data you need to make those deductions?
Peter - In essence we took half a million people who had their DNA read, asked them "is your mum alive? If so, what age is she? If not what age did she die at?" And the same for your dad. We didn't actually gather all of that data. We've been really lucky. There's a fantastic resource UK Bio Bank where the UK Medical Research Council gathered together half a million people and asked these sort of questions; and then we top it up that with 24 other cohorts from around the UK Europe and the United States to give us an even larger resource.
Chris - And basically if you look at enough people enough times you're going to start to see hotspots in the genome regions which appear to be associated with the answers to those questions about how old you your parents were when you died. And so you can begin to say well that region must have some kind of role or association in how long we live.
Peter - That's right. By having enough people we can pinpoint specific DNA letters, amongst your three billion, that are associated with how long your mum lived or your dad lived.
Chris - Now we know that some regions of the genome that have already been quite powerfully linked to longevity. So did your study pull out those ones we already know about?
Peter - That's right. I mean, in fact, remarkably few regions were known to associate with longevity directly. As recently as 2014 there was only one robust association at a genomic region called up APO-E, which associates with Alzheimer's disease. Over the last two or three years, myself and other research groups around the world had shown three or four regions robustly associated with lifespan; and then our study increased that to 12 robust associations. As you ask, we did verify the previously-known loci in many cases. In some cases we actually refuted some tentative findings that people had made before as well.
Chris - And what are the 12 regions that you've now been known to highlight through this. And what do they do?
Peter - The 12 regions - essentially, we see mainly influenced cardio-metabolic health. So basically heart disease. Aongside those, we're seeing regions that interestingly are affecting propensity to smoke. You might not have thought of whether you're a smoker or not is a genetic trait but in fact it turns out the way your brain responds to nicotine is influenced by your genetics. That in turn influences how much you smoke and that in turn affects how long you live.
Chris - So you've got these 12 regions. If I take a person who has all the risk factors and a person who has none of them how many added years do I get through having none of them compared to someone who's got all of them?
Peter - So I don't have that precise number to hand; it must be about half a dozen. Typically the effects that we see - if you carry the good variant and I carry the less beneficial variant - it would be around half a year. So I would estimate it's about six years difference in the way that you describe, but the chances of you having all the good ones and me having all the bad ones are pretty small!
Chris - Are you surprised that you found these associations for so few regions? Twelve doesn't seem like a very big number when you've got such an extensive genome that we have...
Peter - Well that's right. We've studied height for example and with about quarter of a million people we found hundreds of genomic regions that affect how tall you are. Essentially there's two reasons for that. Firstly we've had to ask questions about parents' lifespan so the statistical signal is diluted. But really the main reason is that actually signals of lifespan effects from the genome have these relatively small effect sizes and those require a much bigger sample sizes. So what we would expect is if we were to increase the sample from a million to say three million then we would begin to see many more tens of signals. It's just that we want to be really confident when we report an association.
Chris - Where we going with this then? Do you foresee a situation where we will take a healthy person - in inverted commas - and we throw a genetic analysis at them like this and this highlights a potential set of risk factors that may truncate or extend their life. Is that where you're sort of going with this?
Peter - I think that may well be where society is going. Basically I think two sorts of screening going to come out. One is for rare diseases you might be a carrier for and pass on a susceptibility. And the other is common complex diseases with these small effects that we've been talking about today really. People will be told essentially whether they're in the top or bottom part of the risk spectrum for the disease. I have seen some very nice work in Estonia, for example, that highlighted risk scores for heart disease and how if someone had a BMI of say 35 - was obese - and were to lose weight, the dramatic difference they can make to their risk of heart disease if they're in that part of the population that is susceptible to heart disease in the first place due to their genetics.
23:02 - Evolution through bat's eyes
Evolution through bat's eyes
Alexa Sadier, UCLA
Darwin did the hefty lifting when it came to explaining the broad basis of evolution, but now we need to understand at the level of the genetic machinery how this actually works across the species spectrum. Speaking with Chris Smith, UCLA's Alexa Sadier has been doing this by looking at the evolution of the visual systems of bats...
Alexa - We choose bats because bats are really really really diverse. You have a thousand four hundred species approximately, and they all look different in terms of shape, in terms of size, and in terms of colour vision as well. We have a huge difference in terms of that, also in bats. So that's why we choose bats, because of these diversity that can help us to then go broader about other species.
Chris - And what specifically were you asking off the bats? So how did you approach this?
Alexa - The methods, yeah. So first we need to go to the field we need to go where they are, to the tropics. For example Dominican Republic, Puerto Rico or Trinidad, because that's where they live. To catch bats, we put some mist nets - like a big volleyball net - between the trees in the forest and they fly in, and we put some traps in the caves as well. So now we have the bats. We were studying the eyes, so we took some eye samples to see the final protein that is used to see colours.
Chris - Now there's this saying - I know that you're from France but I'm sure the same thing exists there -there's this saying "as blind as a bat". But actually that's not true is it, because bats actually have quite good vision. We should probably really say things like "as blind as a rat", because bats can see quite well can't they?
Alexa - Yeah exactly. They see very well at night, and during the day they can see colours - only two - we can see three. So, yeah, that's something that people don't expect for sure.
Chris - But when you actually begin to probe into these visual pigments that the bats are using to endow them with their vision, what did you learn and how did that inform the overall sort of thesis of this work, which is trying to understand how they've evolved?
Alexa - So we had this idea before that maybe all bats can see two colours: so green and UV; and when we studied that we realised that a lot of bats can see both green and UVB, so they can distinguish colours during the daylight or dim light. But a lot of them can only see green light. So they are a kind of monochromatic, because they can't distinguish colours. What it showed us is one colour is lost in many bats; it just show that losing things it seems to be an important rule during the evolution of species, because you can think that you will gain new things but what we show here is, at the beginning it's likely that we had green and UV vision in bats and that UV vision was lost independently, repeatedly in different species of bats.
Chris - Do they all lose that visual pigment via the same mechanism? Is it the loss of the gene? Because there are lots of different ways that you can lose something: the gene can go wrong; the process that turns the gene into the physical protein in the cell can go wrong. So there are a range of different ways of losing something. Are they all losing it the same way, or different ways?
Alexa - So different ways. And that's one of the other main findings of the paper actually. Some of the bats have lost everything from the gene so they don't have a functional gene anymore. Some species have the gene, they have the RNA, but they have lost the protein; and some species are in-between: they have lost the final product, the protein, they have lost the RNA, but they still have a functional gene. So we believe that we are in different steps at the process of losing colour vision in these different species, and it's never done exactly in the same way.
Chris - So what are the implications of this? Obviously you've widened the repertoire of knowledge about this important groups of bats, but what does this tell us in terms of the bigger picture?
Alexa - So first we had this idea about these evolution by what we called parallel losses, which is losing things. But here it inform us how it works. And here we can see it can works at different levels: gene expression, going from the gene to the protein. That's very important because we when we had this idea about having rules to have evolution we really want to understand how it works. Now that we have that, we can apply this idea to many different protein, many different genes and many different other species.
27:29 - Higher IQ associated with bigger brain cells
Higher IQ associated with bigger brain cells
Huib Mansvelder, Vrije University, Netherlands
Famously, since Einstein died, researchers have studied his brain to try understand the basis of his genius. But they’ve largely drawn a blank, with no clear consensus. But, as he tells Chris Smith, Huib Mansvelder at Amsterdam’s Free University has been doing a more modern and better controlled version of this sort of experiment to try to understand why some people are more intelligent than others, and in the process answering a very important outstanding question...
Huib - Nobody has ever checked or looked at whether properties of brain cells are associated with mental abilities. So we really wanted to study whether we could find a link between properties of brain cells of the human brain and their mental abilities - IQ. So we consider the size of the brain cells - and that's not just the size of the cell body but also of the of the processes that have come out of the cell body where the cells receive information make contact with other cells - and we also consider their electrical behaviour.
Chris - And what are you relating those neuronal parameters to, in terms of an assessment of brain function? How are you measuring how good someone is at thinking?
Huib - Right. So we use IQ score. We've measured that with IQ tests. And, of course, that's a gross simplification - you cannot of course get somebody's intelligence into a single number, but at least it gives a good approximation of a general intelligence of a person.
Chris - So in a nutshell then what you're saying is let's relate how someone performs on an IQ test - notwithstanding there are limitations of an IQ test - with physically what the nerve cells are like in their brain in terms of the numbers of those nerve cells and how well those nerve cells work, and how fast they work?
Huib - Right. And the point was others have shown certain parts of the brain - when you look at the brain on an MRI scan - the thickness of the cortex associates or correlates with IQ scores over the very same person. So people with higher IQs tend to have thicker cortical regions in temporal cortex and frontal cortex. So that's why we thought that maybe people with thicker cortex have room for bigger brain cells.
Chris - So how did you then bring the two together and access or gain access to brain cells so that you could make those sorts of measurements and relate them to IQ?
Huib - Human brain tissue is very hard to come by. So there's very few people where actually brain tissue is taken out. One of them is epilepsy and other one is a brain tumour. So with the epilepsy patients, about 20 to 30 percent are resistant to a pharmacological treatment. And when the epileptic focus is localised well enough then the surgeon can remove the epileptic focus in the tissue that he cuts out on the way to get access to those deeper structures - that this is a cortical tissue that we take to the lab - and there we can study the shape of the cells but also the electrical activity of the cells, because we take every care that the brain cells in the tissue are alive.
Chris - So you're able to electrically interrogate the tissue while it's still viable but then also go in and physically measure what the cells look like in the same specimen. So you know that you're comparing apples with apples?
Huib - Exactly. During our electrical recordings we fill the cells with a chemical that gives the cells a colour and that we can then later on analyse.
Chris - And what relationship emerges, because presumably you then marry up all of those measurements with the IQ tests that the patients have filled in before they underwent surgery. So what's the relationship? What do you see?
Huib - Right. What we see is that patients that had higher scores on the IQ tests tend to have larger brain cells with larger processes that are more complex and more bifurcations. Also that these cells are larger, generate electrical signals faster and can maintain faster signals much better than patients with a lower IQ scores that have smaller cells
Chris - And do you think the cells are bigger, because they've always been bigger, or do you think they're - in the same way that the taxi drivers in London who did the knowledge of the streets got a bigger hippocampus - that these cells are responding to more use and more stimulation and they're becoming more metabolically active and that's why they're bigger? Is it cause and effect, or were they big to start with?
Huib - There are important genetic factors that associate with IQ scores, but then again training of the brain definitely affects the structure of the brain, so we think that will affect also the shape of neurons.
Chris - And you don't think the fact that you were forced to study tissue that it come from people with an underlying neuropathology might have skewed the results in some way?
Huib - So that's a good question. All the people used this study were epilepsy patients. So we compared similar patients with their similar disease backgrounds with each other. We never take tissue from the epileptic focus; so the tissue that we take to the lab was only removed for the surgeon to gain access and also for the properties that we studied, like the size of brain cells and the electrical properties, we compare these between different patient groups. We also have tissue from tumour patients: they have a different medical history than the epilepsy patients. We compare the properties of these cells between the patient groups and then check whether there is a bias in one of those groups or not or whether it generalises across the disease background. We can never get around the fact that we have to work with tissue from patients but at least we can quantify whether there are some influence of the background or not...