Copyright ©2007 NPR. For personal, noncommercial use only. See Terms of Use. For other uses, prior permission required.

IRA FLATOW, host:

You're listening to TALK OF THE NATION: SCIENCE FRIDAY. I'm Ira Flatow.

And now for something else, another scientific bit of research that sounds like a plot from a science fiction movie - tiny computers, so small that they can live inside a single cell in your body and monitor the health of the cell. We're talking about computers that can detect what - when the cell has gone bad.

Let's say when the genes inside have switched on to make the cell diseased or cancerous, and the molecular machine then sends out the signal from and to the cell to warn us of these events and determine what - and maybe even determine what the problem is and fix it. It really does sound fantastic, but researchers announced this week that they are one step closer to engineering a biological computer.

They report their advance in the current issue of Nature Biotechnology. And today, the lead scientist joins me to talk about this research. Kobi Benenson is a Bauer Fellow in the Center for Systems Biology at Harvard University. And he joins us today from his office. Welcome to the program.

Mr. KOBI BENENSON (Bauer Fellow, Center for Systems Biology, Harvard University): Well, hello. Thank you so much for having me on the show.

FLATOW: You're welcome. Now you say that each human cell has the ability to build its own biocomputers?

Mr. BENENSON: Well, we have to program the cells to actually build these computers. That is, we sort of hitchhike on the cellular mechanisms and program them to do what we need.

FLATOW: So they have the tools already in the cell to build the computer, and you give them the blueprint to how to do it.

Mr. BENENSON: Exactly. We engineer genes, basically lots of genes. We insert these genes into the cells, and the cells then engage into their - what they normally do because they cannot tell the difference between the naturally occurring genes and the synthetic genes that we have built.

So they just build all the active components that we have designed and these components start working on their own.

FLATOW: So the cell is building, actually, its own active computer?

Mr. BENENSON: Well, you could say that. Yes.

FLATOW: And has logic like a computer does?

Mr. BENENSON: Exactly, so the trick here was to actually design molecules that would interact in this logical way much like transistors operate in a silicon-based computer, and so we just - I think we did just that. We have come up with the design and was - implemented it in live cells.

FLATOW: 1-800-989-8255 is our number. Let's talk about the implications of this. You're able to take a cell, and you can instruct it by putting genetic information, something like its own genes in there, and to tell it how to make its computer inside the cell. And what can then - what can this computer then help the cell do? What can the cell learn from the computer?

Mr. BENENSON: Well, the cell probably would not learn much, but we can learn -we as humans (unintelligible). And still, I have to emphasize that we are not there yet in terms of being able to actually tap into cellular signals and, for example, diagnose cells with diseases. But once this technology matures, we really want to be able to interact with a variety of problems, which are, you know, manifested on a molecular level inside a cell and convert the signals into some action, for example, fixing a cell or targeting it for destruction or just labeling it for some future treatment.

FLATOW: So you've tried this in test tubes so far?

Mr. BENENSON: So in fact, three years ago, me and my colleagues at the Weitzman Institute, we have developed a test tube prototype that actually did all these tasks. Currently, what we build in live cells is only this logic component, the molecular component that have the capability of making decisions inside cells. But for now, we still feed this molecular computer with its own molecular zeros and ones, while in the future implementations, the data will come from molecular sensors. So these molecular sensors that actually look for mutations or look for, you know, a gene, look for genes that are incorrectly expressed is something which - that we are still working on.

FLATOW: So it would look for a gene that suddenly turns on, maybe, a cancer part of the cell, makes the cell cancerous, and your little detector would, in theory, be able to detect the change in that gene.

Mr. BENENSON: Right.

FLATOW: And does it signal to you that something bad is happening?

Mr. BENENSON: Exactly. So, in fact, we need computers to do a more complicated stuff. Normally, a single molecule or a single indicator is not enough to make precise decisions. What we instead - we need multiple indicators. And we may have to run a certain decision procedure much like human doctors do when they analyze symptoms of a given - of a patient. So the computational - micro-computation, really, is needed to make the right decision based on all these multiple cellular signals.

FLATOW: So it looks for a variety of signals - A or B, A and B - depending on what you're looking for.

Mr. BENENSON: Right. It could be quite complex, actually.

FLATOW: And then how does it make itself known that it's found something that you should look into?

Mr. BENENSON: Well, ideally, because the cell constructs this computer, it is in a way a - no. It is an automata(ph) system cell(ph) that it can generate a response, that is, it can synthesize a specific molecule within the cell, and this molecule can do things, not necessarily requiring an outside intervention. However, of course, if we do not, we may not be that sure about how the system operates. We may use this molecular output to just label a cell so that a physician may afterwards analyze, look for this marker, and then do something else.

FLATOW: So you'd see the marker, let's say. Let's say it lights up or something.

Mr. BENENSON: Well, you could engineer a system to do that, right.

FLATOW: Yeah. So if you go for your checkup and the doctor looks for these cells that might be lighting up if something is wrong, he knows how to look for that.

Mr. BENENSON: Right. I mean, I could imagine, for example, if these damaged cells circulate in the blood stream and our biological computer were to tag these few cells that represent a problem, then just by looking at the blood sample under a microscope, a physician could identify those, you know, labeled cells and decide or tell that there is a problem.

FLATOW: So you could detect something really at its early, early stages here?

Mr. BENENSON: Well, we hope so. I mean, ideally, this technology is supposed to work on the level of individual cells. So even a single cell which has molecular problems, (unintelligible) could be identified using these tools. Of course, it is really - it is a big challenge to get it down to individual cells, to single cells in organs.

FLATOW: Would this be sort of genetic engineering?

Mr. BENENSON: Well, I would say of course. We're using tools of recombinant DNA and genetic engineering to build our system. They are composed of synthetic DNA, so we are not - we're really built on all the current technologies, synthetic biology and molecular computing, to engineer these systems.

FLATOW: Would it be possible to even go a step further and have the cell repair itself?

Mr. BENENSON: Well, it is a very interesting direction, but I'm afraid that just the level of biological knowledge may not be sufficient as of today to really fix the cells, because way too often, these molecular indicators are not the cause of the problem but the results. So we may detect these problematic signals without really understanding the cause of the problem. Now, to get down to the cause of the problem is a completely different issue, and this is really the subject of basic biological research.

FLATOW: Right.

Mr. BENENSON: But sure enough, once we really understand the mechanism of a disease and we know what we have to fix, our molecular outputs, our - that the computer generates can in fact go there and try and fix the problem.

FLATOW: Pierre(ph) in Rockville, Maryland, welcome to SCIENCE FRIDAY.

PIERRE (Caller): Hi.

Mr. BENENSON: Hello.

PIERRE: I was very intrigued by this. Obviously, the use of these cells as molecular reporters for therapeutic or diagnostic purposes is very interesting. But there was the other use of these that I would be interested in seeing -hearing a little bit about which you alluded to, and that's can these cells -because they are being used as computational items where there may be simple rules or identifiable rules - can they be used to test some of the predictions that you would have in models of cellular automata or self-organizing systems like the kind of stuff that Wolfram or Kaufman have been talking about?

Mr. BENENSON: Well, this is definitely an interesting idea. I would say…

PIERRE: How would you do that? How would you use them to test that?

Mr. BENENSON: Well, I would say you can - if once we can engineer cells that have a certain kind of non-natural and well-defined behavior, a sort of input-output behavior or computational behavior, then we may start thinking about integrating them into networks, into sort of artificial tissues. And you may start looking at how the simple rules of interactions lead to the emergence of the more and more complex phenomena on the level of the tissue or of the cell -assembly of cells. You know, in kind of our dreams, we may envision something like an artificial brain where our cells may function instead of neurons, but having all these, you know new properties. So yeah, I definitely agree that this is an interesting direction to pursue.

FLATOW: In your paper, you talked about biocomputers that can work in human kidney cells in culture. Can you explain what you did there?

Mr. BENENSON: Right. So a kidney - kidney cells are just a common lab system, a model system where people normally test hypotheses or systems, which are supposed to work in humans. There was no particular reason other than that for picking these cells, and - so practically, what we do, we grow these cells in Petri dishes, we provide them with nutrients, and these cells are also immortalized so they can divide indefinitely. We can just also follow their growth.

And when we do our experiments, we take these cells on plates, we introduce our sets of genes that encode the computer, then we actually - we sort of follow how the computer works by looking at different fluorescent proteins that are generated, that are the result of the molecular computation. And then - so we can tell whether, you know, the computation went right or wrong.

In a way, our experiments are just about debugging your system. It's very much like computer industry, what happens in the computer industry, where you run your programs under different conditions, and you want to make sure that the program gives you a correct answer every time. So we did just that, only in a very different setting.

FLATOW: So your program's then sort of, like, bits of DNA?

Mr. BENENSON: It's bits of DNA, bits of RNA, exactly.

FLATOW: And so you tinker with the structure of the DNA and RNA until you get the result that you want in the cells?

Mr. BENENSON: Right, and then we also test it under different - we know what to expect from our systems, and we want to make sure that the experimental results really are as expected.

FLATOW: So you can genetically engineer the bits of DNA to make these crude little computers. How complex can you make the computer, theoretically?

Mr. BENENSON: Yeah, this is a very interesting question. And so one could look inside, you know, in nature, look into cells. And in fact, cells already have their own quote, unquote, "biological computers" that process, for example, hormone signals or make sure that a cell divides correctly every so and so hours.

So these natural systems are quite complex. They have up to, I would say, hundreds of components. Our systems currently have about on the order of (unintelligible) 10 different components. So it is a good question of whether we can get to a hundred components. However, it seems like many real-life problems, do not or may not require these very complex systems. Instead, you may be looking for just a few very characteristic disease indicators, and it will be good enough to make a decision.

FLATOW: Talking with Kobi Benenson this hour at TALK OF THE NATION: SCIENCE FRIDAY from NPR News.

Talking about these computers. Have - are the computers are - the size of the DNA? I'm trying to get an idea for everyone to get an idea how small and tiny these little computers are. They're molecules.

Mr. BENENSON: Well, that's why maybe to talk, really, about the size is not entirely appropriate, because they're not little black boxes that float around inside cells.

FLATOW: Right.

Mr. BENENSON: Other than that, we sort of - they float all over. Our components, which are proteins and RNAs, just float inside the whole volume of a cell, and they transfer information between each other just by basically bumping into each other. And each of these collisions is an act of information transfer.

So, one could tell, you know, the size of the computer is pretty much like the size of a cell. And of course, the size of DNA molecule is small, it is known. A few nanometers or (unintelligible) nanometers, but this is really different, I would say, from a nanotextile technology, which envisions these tiny, tiny robots that are totally separate from their environment.

FLATOW: Right. These are not little robots. These are little bits of instructions floating around?

Mr. BENENSON: Again, these bits of instructions are all encoded in our molecules, which are generated from the DNA templates.

FLATOW: Right.

Mr. BENENSON: And some - indeed, some of them are instructions, yes. So…

FLATOW: Just waiting for the right trigger to react.

Mr. BENENSON: Right, exactly.

FLATOW: They're prime, they're waiting and if they find something, they create a protein that tells you something has happened.

Mr. BENENSON: Right.

FLATOW: And then you can - you pick up the signals for that. How soon until you can scale this up - or is that someone else's job - to try it out in, you know, out of the dishes and into other things?

Mr. BENENSON: Well, we feel it that is our job, and we are working on that. And maybe we're trying to - so I would say the next big challenge is to be able to read out all these disease-related indicators in a reliable fashion or maybe not tailgating within the same side of the system, something like five different signals and answer whether we can really integrate these five different signals to produce a reliable response in individual cells. And from that point, of course, we want - we would like to take it to some model organisms and test them there.

FLATOW: Well, good luck to you.

Mr. BENENSON: Okay, thank you so much.

FLATOW: And thank you for taking time to be with us. We are going to follow this. This is quite interesting research.

Mr. BENENSON: Thanks.

FLATOW: Take care.

Mr. BENENSON: Goodbye.

FLATOW: Kobi Benenson is a Bauer fellow in the Center for Systems Biology at Harvard University in Cambridge, Massachusetts. We're going take a break, change gears - a whole lot of gears here, moving to forth - because it's hurricane season. June 1st is hurricane season, coming up.

And we had a prediction last year. Remember, there was a prediction for a very heavy hurricane season? It didn't quite work out that way. We'll talk about why it didn't happen, what we might expect this year, and why this year might be a whole different ballgame.

So stay with us. We will be right back after this short break.

(Soundbite of music)

I'm Ira Flatow. This is TALK OF THE NATION: SCIENCE FRIDAY from NPR News.

Copyright © 2007 NPR. All rights reserved. No quotes from the materials contained herein may be used in any media without attribution to NPR. This transcript is provided for personal, noncommercial use only, pursuant to our Terms of Use. Any other use requires NPR's prior permission. Visit our permissions page for further information.

NPR transcripts are created on a rush deadline by a contractor for NPR, and accuracy and availability may vary. This text may not be in its final form and may be updated or revised in the future. Please be aware that the authoritative record of NPR's programming is the audio.

Comments

 

Please keep your community civil. All comments must follow the NPR.org Community rules and terms of use, and will be moderated prior to posting. NPR reserves the right to use the comments we receive, in whole or in part, and to use the commenter's name and location, in any medium. See also the Terms of Use, Privacy Policy and Community FAQ.