When BioArray News first interviewed Andy Berlin, director of biotechnology research at Intel, in February (see BAN, 2/28/2003), his group, the Intel Precision Biology program, was just preparing to step into the public spotlight with details of the work it was conducting within the corporate walls of the semiconductor manufacturing giant.
Since then, the researchers under Berlin’s leadership have begun to publish results from their work and have entered into a research collaboration announced in October with the Fred Hutchinson Cancer Research Center of Seattle. In that agreement, Intel was to install at the Hutch an instrument it uses in chip fabrication, the Intel Raman Bioanalyzer System, which uses silver particles in a process called surface-enhanced Raman spectroscopy. The company would provide technical personnel to run the system and collaborate with the cancer researchers there with the hopes of using the presently room-size system to detect subtle traces of early-stage diseases.
BioArray News visited Berlin, who is in his third year at the company, last week at Intel’s Santa Clara research facilities to speak about the progress the group has made, and its plans for the future.
Let’s talk about this instrument, the Raman Bioanalyzer, which you are installing at the Hutch. What guided you to use this tool?
I’d like to tell you that it was this really carefully planned out thing. I wish I could tell you that. I think part of the credit goes to David Tennenhouse, director of research at Intel, who had the vision to fund a project that didn’t actually know what it was going to do. Our charter was essentially to take a look at Intel’s technology and see how it could play out for medical diagnostics. He basically hired me to do this. So, what do you do? You are Intel’s biotechnology effort, you have no staff, and you have no labs. So, you look around the company at anything that might possibly see a biomolecule. We very quickly focused in on the chemistry labs, where Intel actually has a tremendous analytical chemistry capability that we use for everything. People like to look at Intel as kind of the PC company, but it is actually not. It is actually a chemical-based manufacturing company. The vast majority of Intel is assembling atomically thin layers on chips using chemical procedures, and validating those using different types of chemical procedures. In looking around, Raman spectroscopy was one of the techniques that Intel had a lot of expertise in. And, it looked like it had some real nice potential for detecting biomolecules; it also has a nice property that we believed could be moved onto chips, eventually. So, we started there because it was convenient. There was a lab down the hall from the office. They were friendly people, they wanted us to succeed and they wanted to help, so we started that way with simple test samples. We started to get good results, comparable to some of the best ones that had been published and we started to say: What if we made this better? We worked our way up to the point where we could go back to David [Tennenhouse] with a proposal to start a fairly significant project to build the next generation of the Raman system along with the assays and the chemistries, microfluidics, and everything else it takes. It’s kind of a bootstrapping, and it’s really emerging from Intel’s expertise in chemical quality control, semiconductor process development and characterization.
How does the Raman system operate?
Raman is a fairly well-known technique in the chemistry community. And, there is a small handful of places that are pushing the sensitivity for biological applications. Basically, you have a set of lasers that goes off into the back of the microscope and right down into an array. Sample size is limited by the wavelength of the optics. For the systems that we are delivering to the Hutch, you may want a spot that is about a micron or so. This thing is real sensitive. The thing that we have done is push on the enhancement and the applications. There are various chemical enhancements, there are various silver tricks to get silver nanoparticles to right where you want them.
So, you have used this technique to get a graphic readout of a molecule of adenine suspended in a solution. Have you been able to reproduce these results?
We are getting to the point where we can see it every day, repeatedly. There are about four or five places you could go and see a picture where you can see the difference between one molecule and none. [For most labs], at best, it works once every two months, if it ever works again. We can do it every day and that took about a year. It required some very non-obvious optimization steps, like hiring an expert in sewage treatment. To get signal out of these things in solution, you are using these little colloidal chemistries that bring a little piece of silver to the biomolecules. To get the biomolecules really near silver, they interact and you get more signal out than you would otherwise. It’s called surface-enhanced Raman spectrometry, where the surface of the silver enhances the signature. Traditionally, that has been a technique that has not been very repeatable, and hard to make work because the silver colloids have lots of different shapes and different properties. It turns out the sewage treatment industry is very good at silver colloids in general — that’s how they get junk out of water, they cause it to self-assemble and get this little cluster that eventually gets big enough that you can filter it out.
How accurate is the readout?
I can do this with one adenine molecule. But if I had 100 types of molecules, all these peaks [in the graphic readout] would blur together.
One thing people want is to make sense of complicated mixtures, and separation technologies and detection technologies that can work in the presence of complicated combinations. The second thing that people really want is sensitivity. And the third thing is selectivity. Can you tell me for sure that this is that type of molecule? This helps to some extent with that, but to the extent that you have multiple types of mixes, no one says this is useful. It could be that when you look at medical samples, they are really so complicated that being able to see more details of that fuzzy noise just gets you a brighter picture of that fuzzy noise. But we are fairly excited about some of the ideas that the people at the Hutch have had about how they would use the sensitivity. They want to try it out. There are enough people up there with enough different ideas that the risk was low enough that the at least one of those would be good for something.
In this field there is such a matrix of intellectual property stakes, how does your group navigate those?
We are a small research group, just moving from seven to eight full-time people and we have no intention of shipping a product. [The group] is a corporate learning vehicle, asking from a technology perspective, does Intel have something to bring to the table, and what, and in which area? In our view, it is so pre-competitive, we are not seriously thinking about those sorts of issues. At this stage, we are just trying to do some things that nobody has ever done.
We do give [IP] a thought at an industry level. The fact is that it is very clear that the semiconductor industry has matured and people cross-license with each other, and have reasonable royalty expectations such that the industry can operate. To put it in context, it takes 100,000 patents to create a PC — that’s the level of complexity that it takes to put lots of pieces together to do something that is significant for people. As the convergence of nanotech and biology continues, you are going to start to see people creating much more complicated systems than people have ever built before and the [biotech] industry’s IP models will have to evolve.
Could you describe your views on the pace of science and innovation in this field?
There is lots of very good science moving forward very rapidly and it’s primarily people looking at crossing energy domains. The way that I think about it is: Can you take something that is happening in chemical energy where two biomolecules are binding together and convert it to light? Or converting [that energy] to motion of mechanical devices on chips, where if we take a little diving board on [a] cantilever, and change its motion if a biomolecule does something on it? There is a tremendous flurry of activity in the academic community looking at these couplings, but there is much less activity, unfortunately, in the medical community, looking at how to actually make use of it for patients.
Is the application of this knowledge for the benefit of patients the next challenge?
I think Intel’s relationship with the Fred Hutchinson Cancer Center is an attempt to catalyze that transition. We actually see the biology and the medical communities as vastly different — the people are different, the cultures are different, there are different time scales to results. At least in the US, the academic and research communities have done a wonderful job of bridging nano and bio, but if you walk into almost any major medical research institution, none of that is there. At best, they have some Affymetrix chips. I think part of it is that the medical community has a higher standard for where the technology needs to go to be useful to them, which means that either you have to wait for the technology to progress a bit further and become a bit more productized so that you are shipping medical diagnostic instruments to them, or you need people with the technology background to make the technology useful to them early-on. We are actually putting Intel’s personnel up there to work side-by-side with the medical researchers and to take what would otherwise be an instrument that would probably available to them in five years, and make it useful today.
We are asking: How can we help? We aren’t asking what products we can build for them. It’s a different kind of interaction than Intel routinely does. When we first got set up, I think they felt that we wanted to sell them PCs. We do, in fact, try to sell them a lot of PCs. But if there is something we have to help them do this research, we will do it. But it is way too soon to know how this is all going to turn out. Is this going to turn into a chip-based healthcare business that has high profit margins of the form that Intel would want to enter? Or will it not? No one is going to know that for five years.