IRA FLATOW, host
This is Talk of the Nation: Science Friday. I'm Ira Flatow. A team of biologists and mathematicians recently figured out how to make simple computers from living bacteria. These computing E. coli work away at the classic mathematical puzzle called the burnt pancake problem. That's something like maybe what goes on in your house in your house every morning, that experiment. But this is a little bit different. The E. coli really don't flip pancakes. They just jumble fragments of DNA instead.
And the research is out this week in the Journal of Biological Engineering. So, how on Earth do you get bacteria to become computers? And could they be someday be flipping one and zeros on your iPod or on your PC or Mac? Joining me now to talk about these bacterial computers is Karmella Haynes. She's a biologist and postdoctoral researcher at Davidson College in Davidson, North Carolina. She joins us from station WBUR in Boston. Welcome to the program, Dr. Haynes
Dr. KARMELLA HAYNES (Biology, Howard Hughes Medical Institute, Davidson College): Hi. Thank you, Ira.
FLATOW: Let's talk a bit about this. Is this the first time that microbes have been able to compute something?
Dr. HAYNES: Well, it is the first time that computational operations have been demonstrated inside living cells. Other researchers have used DNA outside of cells to attempt to address mathematical problems. And this is the first time that the operations that are used to actually carry out the computation have been executed inside living cells. So, it's pretty exciting.
FLATOW: Yeah, let's talk about the problem itself, the burnt pancake problem. Give us a little thumbnail on that, if you would
Dr. HAYNES: OK, so the burnt pancake problem, basically, what that is, is that it's a problem that's solved by sorting objects in an array of objects by flipping objects either singly or multiple objects at a time. Just imagine you have a stack of burnt pancakes. Each pancake is burnt on one side, golden on the other. And they're all different sizes and what you want to do is that you want to make a nice pretty stack where the biggest one is on the bottom, the smallest one is on the top, and every pancake is facing golden side up.
But one of the rules to that game is that you have a spatula where, in order to sort the pancakes, you actually have to insert the spatula in to the stack and then flip the pancakes on top of the spatula. You can't pull out a pancake and stick it in to the bottom, so you can't shuffle them. The only way that you can order that stack is by flipping them.
That's pretty much what the burnt pancake problem is. So, one of the goals of that problem is to try to figure out the minimal number of flips it takes to convert a scrambled ugly stack into a nice neat one where the biggest one is on the bottom, smallest ones on the top, and all of the pancakes are golden side up.
FLATOW: Now, how do you tell the bacteria to do that? That I'm not actually going to flip a pancake
Dr. HAYNES: Exactly. So - and I don't think a lot of people would want bacteria on their pancakes. That wouldn't be very appealing.
FLATOW: Certainly not E. coli
Dr. HAYNES: Yeah, exactly. So, what we have to do is first figure out how to represent a pancake inside of E. coli. So, what's nice about DNA, it looks a lot like a burnt pancake in that one piece of DNA has a specific directionality, so it can be either be pointing forwards or pointing backwards, OK. DNA pieces, in order for them to make any sense to the cell, they have to be arranged in a certain order. So, if we take something like a gene that has a promoter, which actually turns the gene on in a coding region. It has to follow that promoter.
If those two pieces are swapped into the wrong order, then the gene will not be able to be turned on. So, you can think of those fragments as burnt pancakes and that each piece of the gene needs to be in a certain order and in a certain orientation like golden side up in the pancake stack, in order for that gene to be switched on. So, in bacteria, we know that the pancakes' stack had been solved when the gene switches on, and then that gives us some visible phenotype that we can easily monitor.
FLATOW: So, you have basically then the bacteria working, working, working, just flipping things around until they get it right?
Dr. HAYNES: Exactly, exactly.
FLATOW: Do they know how to stop once they get it right
Dr. HAYNES: Aha, so that is a very good question. So what we have to do is that we have to keep careful watch on the bacteria as they are randomly trying to solve the problem. So, as the bacteria are solving the problem, they're swimming around in the cell culture, we sample at time intervals and check to see if any of those bacteria had reached the desired phenotype.
And that would help us answer the question of what's the minimal time or number of flips it takes to actually reach a solution. A bacteria could reach the solution and then flip away from it. But what we are interested in is just catching the bacteria when they first solve the problem at random.
FLATOW: 1-800-989-8255, talking with Karmella Haynes. But isn't this sort of like, you know, the old thing about monkeys on a typewriter and then they'll do Shakespeare after a while? If you give them enough time to flip, they'll come up with some guess will be right about getting the pancakes all lined up
Dr. HAYNES: Exactly. It's pretty much like that. One of the advantages of doing this sort of computation in bacteria is that you've got millions of little microcomputers all chugging away at this problem at the same time. So, that increases the likelihood that one of them will pick the exact right path in the minimal number of flips to solve the problem.
Dr. HAYNES: So, what use can you make of this? Now we have - bacteria can flip the DNA around, but so what
Dr. HAYNES: Right, so this has applications in - well, basically any problem that requires massive amounts of parallel processing capacity - problems that can be addressed by sorting objects by reversals. So this would be very useful in that regard, for instance, in studying evolutionary genomics where we compare species by comparing how their genes are all arranged compared to each other. Evolutionary biologists are interested in figuring out the minimal number of steps it takes to convert one species genome into the other by rearrangement.
So, that's one practical use of being able to solve very complex versions of this problem. And larger amounts of computational power are required as the pancakes stack gets bigger. It would probably take a computer months and months just to solve a pancake stack that was on the order of 11 pancakes big. At that point, the problem gets massively complex and that's where having tons of parallel processing power becomes handy
FLATOW: So, in other words, you turn each of the E. coli into a little computer?
Dr. HAYNES: Yes.
FLATOW: And you all work on it together, or parallel. Yeah.
Dr. HAYNES: In parallel.
FLATOW: In parallel. That's interesting. Wouldn't it be great if you could have it stop itself and signal hey, I've finished, you know?
Dr. HAYNES: Yeah, you know, that wouldn't be too difficult, at least in concept, to accomplish. So for instance, we could have that unscrambled gene be a gene product that would then turn around and attach itself to the enzyme that's doing the flipping to stop the flipping. You see, so sort of a negative feedback loop
FLATOW: Maybe you could have..
Dr. HAYNES: Yeah, but that can be engineered
FLATOW: Yeah? Could you do that, yeah
Dr. HAYNES: Yeah, I think so
FLATOW: Could you get it to turn a little light on or something? You know, would they have these fluorescent bacteria that maybe, you know, hey it turns green, it's done
Dr. HAYNES: Right, so the gene that we are using in our work right now is tetracycline-resistant, so it's a gene that renders the bacteria resistance to an antibiotic. So, the phenotype that we look for is growth. So, any bacteria that lives solve the problem and everybody else in the culture is dead when we check. It's unfortunate for the bacteria.
So alternately, yes, you could easily use a fluorescence protein to get a visual of when the bacteria actually solve the problem. Further engineering could allow the bacteria to stop flipping the DNA around once they got to the solution. And that's where we would have to engineer in something like a negative feedback loop.
Dr. HAYNES: That would be pretty nifty.
FLATOW: Let's go to Becky in Gunnison. Hi, Becky, Gunnison, Colorado
BECKY (Caller): Hi. My question is I just wonder - how do you get the bacteria to do this or want to do this? Or is this something they just automatically do when given the opportunity
FLATOW: Good question. How do you - do they have to talk them into that
Dr. HAYNES: Well, we have to engineer them to be able to do so. So, what we've done is that we've taken - there's an enzyme that regulates the scrambling of the DNA in the bacteria. That enzyme, we cloned that of a bacteria called salmonella, this is the same bug that will grow in eggs that are sitting too long, raw eggs that are sitting too long on a counter overnight.
So, same bug. We get the enzyme from salmonella and that's the enzyme that can cut the DNA, flip it 180 degrees and stick the ends back together. That enzyme also needs to be told where to cut by specific DNA sequences. So those are also engineered into the E. coli. So, we basically bring in components from Salmonella bacteria in order to engineer E. coli to carry out these manipulations
BECKY: Great. Thank you
FLATOW: Thank you, Becky. What would be the next problem for it? A little chess came? Checkers
Dr. HAYNES: All right. That's a very a good question. So, right now we're focused on the burnt pancake problem because the system is so specific to sorting by inversions, what we would essentially have to do is take other math problems and convert them into sorting by reversals problems. And we haven't quite done that just yet, but the collaborative team at Davidson College and Missouri Western State University, the mathematicians are thinking away at ways to be able to do that. And once that..
FLATOW: Like at one point
Dr. HAYNES: OK, what's that?
Dr. HAYNES: Oh, that would be awesome
FLATOW: Wouldn't that be great? Yeah, bacteria solving Sudoku problems.
Dr. HAYNES: Yes! Brilliant! You want to come and join our team?
FLATOW: I don't know why it just seemed perfect. It just struck me as you were saying that, you know
Dr. HAYNES: Yeah. Yeah
FLATOW: Because its own numbers and you just have a random - put all those numbers in there and see how long it takes to do the Sudoku problem. Give them a hard one.
Dr. HAYNES: Yeah
FLATOW: Not the entry level
Dr. HAYNES: That makes it - that would be yes. OK
FLATOW: All right?
Dr. HAYNES: That's something to think about
FLATOW: Well, that's your assignment
(Soundbite of laughter
Dr. HANES: It's definitely something to think about
FLATOW: Well, let's see. Sometimes they do come up with something intelligent on this program. OK, let's go to the phones. Let's go to Ty in Reno, Nevada. Hi, Ty
TY (Caller): Hey, how are you doing
FLATOW: Hi, there. Do you like bacteria to do Sudoku problem for you?
TY: I'm not really into Sudoku.
TY: I thought it was kind of funny and ironic that this was basically a genetic algorithm but you're using bacteria's genetic material to solve it. Anyway, I was wondering, in that respect, is there any sort of like pressure toward the fitness, like a genetic algorithm would actually run to sort of pressure the bacteria toward the actual solution? Or is it just completely random like the monkeys in the typewriters
Dr. HAYNES: Right. What we were going for was complete randomness so we don't introduce the selection pressure. Its selection pressure would be antibiotic resistance, OK. The bacteria are rewarded..
FLATOW: Explain what that means, this selection pressure
Dr. HAYNES: The selection pressure. OK. So it's whatever environmental condition rewards a certain arrangement of fragments and in the E. coli right, so..
FLATOW: See, you punish it - you'd punish it for going in the wrong direction or something like that
Dr. HAYNES: Yes, yes. But we don't do that, so the caller is asking whether during that flipping process, is there any reward or punishment for having a certain arrangement of fragments? And the answer to that is no. What we do is in the - after we allow them to flip at random, we then take out a sample of bacteria, and then subject those to the antibiotic to screen for antibiotic resistance, just to check to see if anybody had solved the problem.
FLATOW: Yeah, we're talking with Karmella Haynes this hour of Talk of the Nation: Science Friday from NPR News about the bacteria that solve problems, and our listeners have their thinking caps on. Homer in Raleigh. Hi, welcome to Science Friday.
HOMER (Caller): Hey, how is it going?
FLATOW: Hi, go ahead
HOMER: I'm wondering if we can get some bacteria that's living in our bodies, maybe in our digestive system or something to do this, to maybe trigger some sort of internal response and I'll take my answer off the air
Dr. HAYNES: OK. Well, the system that we've engineered, this using an enzyme from Salmonella and the DNA recognition sites that can manipulate DNA, that can be probably applied to manipulating genes and a system. So more like less in terms of solving a mathematical problem and more in terms of just perhaps driving a toggle switch to regulate the expression of some useful genes, something that would be useful to the bacteria's function.
FLATOW: Now - yeah..
Dr. HAYNES: That would allow us control over the expression of genes
FLATOW: I imagine there are people who are worried now that you're playing around with E. coli, flipping their genes around, it's going to escape out of the laboratory. We're going to have some brainy E. coli taking over our food or whatever. Any possibility that might happen
Dr. HAYNES: You know, at this point I, to be honest, I seriously doubt that these bacteria would pose any public threat. The - so, all of this flipping is taking place on circular pieces of DNA called plasmids. If those plasmids are not selected for, so in order to maintain these little micro - DNA microprocessors in the bacteria, we constantly have to grow them up into in conditions that encourage the bacteria to maintain them
FLATOW: OK, let me interrupt it because we have someone on the phone from - Rebecca from Chicago who has sort of the same question, right, Rebecca
REBECCA (Caller): Yeah, hi
FLATOW: Go ahead.
REBECCA: Yeah, I'm actually just curious as to what this process actually looks like when you're in the lab
FLATOW: What does it look like? Yeah! What are you looking at and how does it work? Good question
Dr. HAYNES: Well, you know, it's surprisingly not very exciting. It's a big glass flask full of brownish-looking liquid swirling around at 37 degrees. So, we're basically just growing up bacteria in culture. Each one - those bacteria carry the plasmids that - where the flipping is taking place. So, it's all - it's pretty much just normal standard molecular genetics techniques that we use..
FLATOW: So, do you take samples out of this big flask every once in awhile to see if it has done something
Dr. HAYNES: Yes, yes, at time intervals and grow them up on an agar plate that has antibiotic on it and then see if we have any growth
FLATOW: So, this is a long process, this is not something that happens overnight
Dr. HAYNES: No, it's not. Yes, so, it does require some human intervention. However, it's a very efficient process and that the actual computations generating all the permutations of the problem happens inside the bacteria, and all we have to do is sample every once in awhile to see if we have any growth.
Now, we have engineered the system to occur very slowly so that we could study the kinetics of flipping the DNA. There are ways to actually speed up the flipping. The natural process of flipping a piece of DNA and salmonella takes somewhere in the order of under a minute. So it happens very rapidly as well
Dr. HAYNES: So there are ways to speed it up if we wanted to
FLATOW: Well, Dr. Haynes, I want to thank you for taking time to be with us.
Dr. HAYNE: All right
FLATOW: And we'll be watching for that paper you publish about the Sudoku problem solving.
(Soundbite of laughter
Dr. HAYNES: All right
FLATOW: And have you back.
Dr. HAYNES: All right, we'll be working on that
FLATOW: OK, Karmella Haynes. She's a biologist and post-doctoral researcher at the Davidson College in Davidson, North Carolina, and she was talking to us about using E. coli as computers. We're going to switch gears, take a break and come back and talk lots more about the new "Indiana Jones" movie and those crystal skulls that are the subject of the movie. Is there real sciences? Is there real geology? Real archeology? Real mythology? What is real about those skulls? So, stay with us. We'll be right back after the short break
(Soundbite of music
FLATOW: I'm Ira Flatow. This is Talk of the Nation: Science Friday from NPR News