Fourth Annual DNA Grantees' Workshop

Tuesday, June 24, 2003

AFTERNOON SESSION

A Chip-Based Genetic Detector for Rapid Identification of Individuals
Ronald Sosnowski
Biography

(Note: The PowerPoint presentation that supplements Dr. Sosnowski's transcript is not available.)

MR. COFFMAN: Our next speaker is Dr. Ronald Sosnowski. He received his training as a molecular cellular geneticist at Johns Hopkins University and at the University of California at San Diego. He is currently the senior director of the molecular biology division at Nanogen, Incorporated, in San Diego. He is going to speak to us today about a chip-based genetic detector for rapid identification of individuals.

DR. SOSNOWSKI: Good afternoon. I'm also going to update you on our progress with the chip that we're using and developing for human identification. In doing that I'm also going to give you a short review of what we do, because I always run into people who have heard of Nanogen but don't always connect with it.

This may refresh your memory. The chip on the left-hand side is one that we currently have. The DNA is being addressed to alternatively bias negative and positive electrodes. That's occurring in real time. The DNA doesn't have any attachment chemistry, so it floats in solution.

Down here is a comparison between an electronic hybridization, where there is actually DNA attached to the chip, and a passive hybridization. The time on the right-hand side was a 16-second hybridization, and I think it was several hours on the left-hand side.

Again just by way of review, the chip surface looks like this [indicating]. These are platinum electrodes. Platinum is the wire that's used in electrophoresis material and electrophoretic gels that you have on our benches. There are traces that go out to the edge. A dielectric or insulating material falls in between the platinum electrodes, thus separating them, and makes the traces distinct. A permeation layer of embedded straptavidin encompasses the whole surface. It is part of the chemistry that attaches to the DNA, and it also keeps the DNA molecules from contacting the surface of the electrode. Electrons that reside on the electrode will destroy any analyte that comes in contact with it.

Conceptually, the way to think about our technology is that each of these electrodes is equivalent to a microliter well or to an individual tube, and we can do separate reactions. That is, we can put different DNA down in these places and react that with the DNA that's in solution to get a different reaction at each of the points.

This is just another closeup of the chip. Here you can see from the electrodes that there is a trace going out to the edge, and then this edge connects with wires that go to power supplies and switches that control the electric field.

I'm going to go through this fairly quickly again because I think a lot of you have already seen it. The diameter of one of the bare electrodes is about 80 microns. Keeping in mind that a eukaryotic cell is about 40 microns in diameter, it's a microscopic system. Here you can see the permeation layer and its relative thickness of about 8 to 10 microns.

It's a silicon-based wafer constructed out of the microelectronics industry, which gives us a lot of overlap with well-established technology. And again, this is a platinum electrode surface.

What happens is that large molecules are put into solution over the array, and different points in the array can be activated. Whatever has the opposite charge of these molecules is going to be attracted to those points. As you can see here, this is being activated by the instrumentation, and the molecules are going to concentrate to that point.

Not all the molecules react. Some are left over to react at other sites. Pads can be turned on in parallel, and a large number of molecules are interrogated. One can then do a fluid wash or an electronic wash to get rid of the molecules that are not perfectly hybridized. Here, as represented by the different colors, you can see different types of reactions. This would be one capture sequence, and this would be another capture sequence.

This is currently being used for SNPs (single nucleotide polymorphisms), mostly because SNPs have evolved as the choice among professionals in the clinical diagnostic industry who are looking at DNA associations with diseases. We've also done, through NIJ's support, STRs (short tandem repeats) on this system, which I've talked about in the past.

There are a couple of different formats that can be done: amplicon down, capture down, and sandwich assays. But what I'm going to focus on today is an amplicon-down format.

Focusing on the SNP assay is the direction that we're headed in with this technology for forensics applications. What we do is amplify. There are two ways of doing it. You can amplify the locus offsite or offline by using a biotintillated primer, and then you can address that biotintillated primer to a particular pad. Through a combination of WAVNAS action, concentration effect, and electrochemistry, the biotin interacts with the straptavidin only at the activated pad. It creates pH conditions and other conditions that favor the reaction but not at an unactivated pad.

So the DNA is addressed and it becomes almost permanently attached. There are two ways to then interrogate the molecule that's present. One can either put a labeled reporter probe directly on the molecule or use a stabilizer oligonucleotide, which is what we've gained a lot of performance with. The presence of the stabilizer oligo in the presence of the reporter oligo gives an extra bit of stability in the dipole-induced dipoles and the vertical-base stacking. There's actually a lot of energy when molecules stack, and it gives a significant amount of discrimination level.

The system, which was done in a sequencing lab as part of the Human Genome Project, turned out to be surprisingly accurate. We really didn't think it was going to show up as well as it did. We had about 93-percent accuracy or a 7-percent discordancy rate. We went back and, after re-reading the electrophoretogram or resequencing the material, all of the discordance resolved in our favor. So we were actually 100-percent accurate, overturning a lot of the sequencing issues.

There was one sample that never really completely resolved. But this one was always sort of in between. We went back and found that the sample was actually some placental DNA bought from a commercial supplier. It was a mixture of homozygote and heterozygote genotypes, which is exactly what you would expect.

This is one of the reasons I think that we were able to overturn some of the sequencing calls. You can see that these are thiamines in different backgrounds, and there are nearest-neighbor effects that change the relative height of the peaks. Because of that, the calls are very difficult to make, and it's not always as accurate as you would like. By developing for specific points and specific loci, we can obtain a much higher accuracy in the assays.

Our experience to date has been that nothing is 100 percent, so this is a misnomer. We just haven't seen the first real error. But we are still almost absolutely correct against the gold standard, and in a test with Jim Schaum at Boditech, we were 99.9-percent accurate with STRs. They used some different rules than we used in our software that would be reversed in today's technology to be 100 percent.

Just real quickly, here are some data that you've seen in the past. This is a look at both STRs and SNPs on the same chip, which has advantages in being backward compatible with the databases. This is just after discrimination, whereas the previous slide was before any discrimination. You can see a heterozygote CSF result and a homozygote Tpox result. Here are the duplicates going in different directions, and they resolve correctly with homozygote, heterozygote, and homozygote loci.

This is just quantitation of the results that I just showed you.

One thing that we've incorporated into the chip is anchored amplification. I think it's going to have a significant advantage over most other systems in terms of being able to provide automation. The anchored amplification allows you to simply take the prepared DNA, address it to the chip, have the amplification take place on the chip, and then the analysis can take place immediately on the amplified material.

We chose strand displacement amplification (SDA) for several reasons. It is isothermal, which reduces some of the requirements on the instrumentation. Our format allows it to be easily multiplexed, which overcomes one of problems with SDA. It's not easily multiplexed in solution, but by separating it into the different virtual tubes, we are able to fairly easily multiplex it. SDA also reduces some of the sample preparation in terms of purifying the PCR fragment, and it gives the amplification genetic determination on a single platform.

Here you can see two different scenarios. The first scenario includes collecting the sample, isolating the genomic DNA, PCR amplification, purifying the amplicon, and putting it on the chip. The second scenario includes isolating the sample, preparing the genomic DNA, putting it on the chip, and walking away.

It's actually a very rapid method because it's asynchronous. You can have any stage of amplification going on at any particular time point depending on the molecule. So you don't get the separation of the molecules, reannealing, extension, and so on. Any individual molecule can be at any stage at any time, resulting in a very quick amplification.

In solution, it's been reported to have 10- to 15-fold amplification in as little as 20 minutes. It's also isothermal, as I mentioned, and that gives you advantages in being able to build instruments that are more simple, and it can be done in an anchored format.

It's a little more complex molecularly, and it requires two amplification primers, just as PCR does. They're opposed primers, one for each strand, so you get the exponential amplification. It can take between zero and two bumpers, and we're working on methods to eliminate the bumpers because we want to reduce the complexity of DNA in the reaction.

It also requires two enzymes. One is a polymerase enzyme, and the other is a restriction enzyme. It's a fair amount of work to optimize these two enzymes and make sure they have the same magnesium, temperature, and salt requirements, but we're doing that for you, so you don't have to worry about it.

Reaction time is slow, and it takes modified dNTP (deoxynucleotide triphosphate) but we're working on nicking enzymes, which will reduce that requirement.

Very simply—and this is, I will admit, a gross overview—what is done is the biotintillated amplification primers are addressed to the pad and firmly attached to the pad. Then, the purified genomic DNA—this is the DNA that really doesn't have any extraordinary preparation done to it—is put on the chip and hybridized specifically to the complementary primer. If you have a different sequence, you'll get a different piece of genomic DNA hybridized.

Then the amplification step takes place, and when you're done you get extended strands that contain the locus that you want to interrogate, be it STRs or SNPs. This has been demonstrated for use with bacterial warfare agents, antibiotic resistance detection, and also for forensics with SNPs and STRs.

These are some data similar to the data I showed you with PCR fragments, where the loci have been amplified on the chip. All the data that I'm going to show you is where the genomic DNA has been amplified in situ. This is the resolution of the different heterozygote loci. This is the quantitation of that data, and then on that same chip, here are the Y-chromosome SNPs, which have been sequenced in collaboration with Mike Hammer.

This was also done on the same chip, and there's the resolution of the homozygote Y–SNPs on the system.

So, where do we stand? We've amplified three different STR loci on the same chip and resolved them, and we also have six Y-chromosome SNP loci that are on the chip.

Moving forward, Nanogen is changing from a passive chip to an active complementary metal oxide semiconductor (CMOS) chip that will have some of the features for thermal control and electronic feedback built into the chip. It will also have the capability of memory and will keep an onboard map of what's been done to each position. (The passive chip is actually kind of a dumb chip. The traces are inert, they don't have anything to do, and they don't cooperate with the reaction or any of the information going to the pads.)

These are some of the advantages: One of the major advantages that we hope to see is not just an increase in the number of sites—which is something that's required for the NIJ project because we plan on having a fairly large panel that we're going to put on here—but also with a greater control of the individual sites. We anticipate, and actually we are seeing, much faster hybridization and better control over the molecular reaction in an electric field. There's a new cartridge that will be with the chip. This is the size of the chip and cartridge that's in the new system, and the chip has 400 pads on it, so you can do 400 individual reactions on that chip.

Again, some of the work in the development of the chip in the CMOS system for remote biowarfare detection agents is being funded by Defense Advanced Research Projects Agency (DARPA).

I just want to show you this to show you that we have an active chip that's working. It's similar to what you saw before, but what I want you to notice is that this concentration gradient is directly proportional to the amount of electric field that we applied to each of those places. It just demonstrates the control that we have there.

The new chip is going into a new next generation system. The platform is based on a 400-site chip. There's only going to be one instrument rather than a loader and a reader combined. It will have better reagent capacity and a lower cost, and you'll be able to walk away from it.

I can only show you an artist's rendition at this point. This is the robotics station that's built into the system, and the rest of the works are down here. If any of you are familiar with the old system, you can appreciate the difference.

I'm going to go through this pretty quickly. I think a lot of it is probably pretty evident. These are things that were targeted for a diagnostic lab, but, as I mentioned this morning, there are a lot of parallels between the needs of the forensics labs and the needs of the diagnostics labs. Both could result in decisions that could significantly affect the outcome of someone's life, so the need for stringent results is remarkably similar. There's a lot of leverage and overlap between the development of these things.

While we're waiting for the next generation system to be built, we're leveraging against some work for a remote biowarfare system that's being supported by DARPA. This is your standard laptop that's on top of a box that has functionality of pretty much everything that I just showed you, including thermal control, and the smart chip allows us to greatly reduce the size of this. This whole instrument can be put onto the platform of a robot, and you can access it fluidically through this port here.

So we're doing some of our development while we're waiting for the next generation to come fully online. This is just gene detection on the chip using that system.

New loci that we've developed include four more SNPs. We've gone through PCR amplification. We've developed the PCR primers. We do that to validate the reporters that we're using. And these loci have now gone through development with on-chip amplification and are ready to be added to the mix. Here are the results that we're getting for those SNPs.

So, where we are now? In terms of the STRs that we have on the chip, these three have been worked out with PCR, and we've also checked out the reporters. These are kind of on hold. Our priority is to develop the SNP assays. We now have a total of 10 SNP loci that we're investigating.

As far as instrumentation goes, we're still working on the molecular biology workstation and the 100-pad chip. We go back and forth between these two. That's good news. It means the assays are pretty portable between platforms, which makes development easier, and it lets us have a fallback. There is still some development being done with that device.

What are the next steps? We need to get to the point where we can put a total of 35 loci on the pad. What we want to do is select from the Y-chromosome mitochondrial sequence and autosomal chromosomes. That means we need to put 25 more on the chip and then integrate those loci.

"Instrumentation" means we need to go on to the next generation instrument.

I just want to acknowledge our support from NIJ and the help of Lisa Forman and Lois Tully in finding our way through the budgets and all of those things. Bode Technology has supported this from very early on. I also want to thank several people at Nanogen who have worked directly on this: Ashton Schroder, Edy Wong, Jack Shiragian, Tim Summers, Hira Halalli, and Zushi Yao. Rick Gelbart and Paul Swanson are some of the engineers that built the BWA help support for the NIJ applications.

Thanks for your attention.