Name: Derek Stein
Title: Assistant professor, Department of Physics, Brown University, since 2006
Experience and Education:
Postdoc (with Cees Dekker), Delft University, the Netherlands, 2003-2006
PhD in applied physics (with Jene Golovchenko), Harvard University, 2003
MSc in applied physics, 1998
BSc in physics, McGill University, 1997
Researchers have analyzed DNA by mass spectrometry for many years, but Derek Stein, an assistant professor of physics at Brown University, wants to take the technology to a new level. He proposes to combine solid-state nanopores and electrospray mass spectrometry for single-molecule sequencing.
Last month, Stein presented an outline of his concept at the National Human Genome Research Institute's Advanced Sequencing Technology Development meeting in La Jolla, Calif.
In Sequence recently spoke with Stein about his idea, and the challenges he needs to overcome to make it work.
Tell me about your idea of nanopore DNA sequencing by mass spectrometry. How do you envision this working?
I envision this to be similar to electrospray ionization in conventional mass spectrometry. In other words, on one side of a small hole, you have biomolecules in a fluid, and on the other side, you have a mass spectrometer sitting in a vacuum environment. The way that electrospray works is that typically, you push the fluid through the nozzle of this little aperture, and at the same time, you are applying a strong electric force by applying a voltage between the capillary and an electrode that is sitting in the vacuum chamber. What typically happens then is that little charged droplets are pulled off and pulled into the mass spectrometer by electrostatic forces. Eventually, these droplets evaporate and sort of blow apart when the charge becomes too high to contain them, and you are left with a bunch of singly charged molecules.
My idea is to essentially shrink down the size of those electrospray nozzles to the single-molecule, nanometer range. The idea for doing that is twofold: On the one hand, when large, charged droplets are injected into mass spectrometers, and when they blow apart, a lot of the molecules are missed by the detector. So it's not really a single-molecule technique that can [only] maybe catch a percent or two of the molecules that are transferred.
Empirically, as the aperture gets smaller, the efficiency goes way up, and these droplets that get pulled out are smaller, so they don't blow apart as much, and you don't get as much spreading of the charged molecules. But the smallest [aperture] people really ever use is conventionally in the micron range, and maybe a few mavericks have tried things down into the hundreds-of-nanometer range. The question is, what happens if this thing gets pushed down into the nanometer range?
The second reason is that it may well be possible to get the order of the components of the molecules as they are coming through. Right now, protein sequencing uses mass spectrometry, and the way it works is, you weigh a piece of some protein you are interested in with a first stage of mass spectrometry, and then you fragment it into many pieces by having it either collide with a gas of some molecules, or irradiate it with laser light of sufficient energy and intensity. All of these are known to break up molecules, and they are known to break up proteins in very convenient places, namely the peptide bond.
Similarly, if you try to do this with DNA molecules, researchers have found that the DNA molecule tends to break apart at convenient locations as well. The weakest link is the single bond that holds the base of DNA onto the backbone. Also, the backbone itself is very fragile. In either case, if you pluck those things off, you still have a way of determining what base it was at that particular location.
[ pagebreak ]
In protein sequencing, when these pieces are fragmented, it's not possible to know in what order they were sitting before they were broken apart. What you have to do is make all possible fragments. You need many, many molecules, and you need to do all possible combinations, and then you can reconstruct by deduction what their sequence must have been.
On the other hand, if you can fragment this molecule in a way that preserves the order, namely, fragment it before it even totally leaves the aperture, then the order in which the fragments leave that spot will tell you the sequence information. If your detector is sensitive to these individual ions, then it should be possible to reconstruct the sequence from the sequence of masses that hit a detector array.
What do you consider the most challenging part of this process?
To my mind, the fragmentation process is one that has to be extremely well controlled. While it would be a very welcome thing to be able to cut each molecule at the same place along the backbone — and there is good reason to think that we will be able to do that — at the same time, we want to avoid breaking apart the bases themselves and destroying the information that's contained within them. I see that challenge as the most important one, because it's the one for which we have the least amount of experience and information available. This has not been a question that people have really addressed, so we are not sure what's exactly waiting for us.
What have you done already?
So far, we have, in a conventional mass spectrometer, convinced ourselves that it is in fact very easy to inject single nucleotides and identify them by their mass. We have also tried to establish that the nanopores themselves would be compatible with mass spectrometry, in the sense that if we have fluid on one side and a vacuum chamber on the other, we need to know that these pores are going to survive. We have been pressure-testing, for example, their structures, and we have been able to show that these things can easily survive about 9 atmospheres worth of pressure, which is as high as our regulators could go currently. And we have been designing and building [devices] that will allow us to interface a nanopore with a vacuum chamber.
What kinds of nanopores are you using?
These are solid-state nanopores made in free-standing silicon nitrite membranes. As far as the nanopore field goes, this is rather conventional. There is nothing extraordinarily special about this. The other thing that we realized we might need is control over the electric field in the vicinity of the nanopore. For that, we have been working on incorporating electrodes into the nanopore membrane itself, either at the surface or embedded inside the membrane, so we can have control of the electric fields right in the vicinity of the pore, where we expect to be creating ions, so we can efficiently inject them into the mass spectrometer.
How will you deal with inefficiencies of the ionization and the detection?
What we have to know first is how inefficient a process might be, and that's something that nobody can answer yet. There are some very basic scientific experiments that need to be done. Nobody has pushed this to this level. You can suppose things: Let's suppose that we can only detect 1 out of 10 bases. At that point, we might be willing to just discard this method altogether. However, if, as we suspect, we can push these efficiencies well above 90 percent, then an error in reading a base by this method is in principle no different than errors that you encounter by any other sequencing method, in the sense that even DNA polymerase is known to induce errors in incorporation and give you mistakes. And then, if you want a certain accuracy, and you know the error rate, then all you have to do is a little bit of calculation, and you know how much coverage you need of the molecule.
Can the detection keep up with the speed with which the DNA goes through the nanopore?
Right now, typical experiments in nanopore research will show that a molecule of double-stranded DNA goes through a relatively wide pore at a rate of something like 25 bases per microsecond, for a typical applied voltage of about a tenth of a volt. That is an extremely high rate for an electronic detector to keep up with. It means that you need bandwidth in the tens or even hundreds of megahertz range. That is not very easy to come by, especially if you are detecting very small electronic currents. For that reason, nanopore researchers have been focusing a lot of attention recently on means of controlling the translocation process. Many of these strategies are showing very interesting results and promise.
[ pagebreak ]
What I find appealing about this mass spectrometry idea is that the detection of a single ion in a conventional, let's say, Channeltron mass spec detector, produces a pulse that's only about 20 nanoseconds in duration. So in principle, even if the DNA molecule comes through at 25 bases per microsecond, and every single base hits the exact same detector one after another, that detector has the bandwidth to resolve those things and count them individually. And that is without even pushing on the technical capabilities. There are other kinds of single-ion detectors that are considerably faster. They have pulse widths that are only about a nanosecond. So bandwidth does not appear to be a limiting factor if we can count every ion.
What is going to be the next step in testing the feasibility of this concept?
I think the fragmentation is probably the most difficult [question to solve]. The other question is, 'How do you get the efficiencies up?' There are good theoretical reasons to believe that a high efficiency can be obtained, but we have to still do that, we have to demonstrate it experimentally, and then we also want to know what's happening to these molecules as they are injected into a mass spectrometer. How are they damaged? Are they cleaved? Are they ionized? And so on.
To answer both of these questions, the ideal tool is some kind of a prototype single-molecule mass spectrometer. We want to do mass spectrometry that people haven't tried yet, so we are going to need some kind of a new tool to do that. And we want it to be as flexible as possible, so we can understand what we are doing, and that understanding will guide technical development. So the next step is to try to build a prototype single-molecule mass spectrometer. It would combine a very traditional form of mass separation, for example a quadrupole element, with a nanopore ion source, essentially replacing the capillaries that are used in traditional mass spectrometers with a nanopore-type source of ions.
And then furthermore, we would like to have the ability to extract ions and to fragment them with electric fields, with laser irradiation, and even with bombardments of particles — the easiest ones I can think of would be electrons. So we would want to build some kind of research system that would test both the feasibility and study the efficiency of these different probes, let’s say high electric fields, laser beams, and electrons, at doing what we want, which is generating ions and fragmenting molecules.
How do you want to parallelize the instrument eventually?
There are many ways of doing mass spectrometry. If one wants to do it in parallel, then what you would need are sources of ions, let's say fabricated on the same chip, for example an array of nanopores. And then you could also imagine an array of detectors that would catch the ions coming from these different pores. The kind of mass spectrometer for which this would be the easiest to do is a time-of-flight mass spectrometer, because you just measure how long it takes for an ion to go from the nanopore to the detector. The detector may be somewhere else in the vacuum system. You could even imagine putting a detector on the same kind of chip on which the nanopore is located. Any one of these strategies would, in principle, give you a way of parallelizing it.
Why has nobody else tried this before? Scientists have analyzed DNA by mass spectrometry for a long time.
That's a question I have sometimes asked myself. The best that I can come up with is that nanopores have, until recently, been very much of a research curiosity, rather than an established commodity that's available to the research field in general. It's not that easy to produce nanopores. There are only a handful of groups in the world that routinely make these things, and none of them, as far as I am aware, are mass spectrometry groups. For that reason, the availability of nanopores has opened up new doors. I know that of course mass spectrometry-based sequencing techniques have been conceived for some time now, but to try to do things at the single-molecule level is the thing that was really not the orthodox way of thinking about mass spectrometry.
How did you come to this project? Are you coming from the mass spec side or from the nanopore side?
The DNA sequencing idea is one in which I have been very involved since my days as a PhD student. I was the first PhD student to work on solid-state nanopores in the group of Jene Golovchenko at Harvard. In working on that project, the first problem that we had to tackle was how to actually make nanopores. At this stage, there are a few methods that people have devised to do this, but in those days, we had to come up with one, and we had to be confident in our prospects for success.
[ pagebreak ]
The technique that was ultimately developed was called ion-beam sculpting. It uses the fact that in a vacuum chamber, intermediate-energy ions, namely 3-kilovolt argon ions, can induce nanoscale changes at the surface of a solid-state material like silicon nitrite. With ion-beam sculpting, the idea was, you hit the surface of silicon nitrite with a low-energy beam of ions, and then you count the ions that come through the nanopore using Channeltron detectors in order to know the size of the pore. So by counting every single ion, you have information on the size of this pore down to the single nanometer scale. And that's the trick that gave the ion-beam sculpting its real power, both as a fabrication technique and as a tool for studying material science.
From that perspective, this project, in terms of combining nanopores and mass spectrometry, was sort of a natural thing for me, because I know that we can count every single ion, I know how to build optics that can refocus every single one, and I know that if they come through this nanopore one at a time, there is no problem in refocusing them. The efficiencies are extremely high, and the detection speed of the Channeltron [detector] is extremely high, so it was just a question of, 'Can we get the molecule to fragment?' That's the part to me that I don't have experience doing, [so] that's why I consider that the biggest challenge. But looking through the literature, I know that it's certainly possible.
My teammate here at Brown, Peter Webber, is in fact an expert in the field of photo-fragmentation. So it was a very natural thing to come together with him. It turns out that more than 10 years ago, he himself had proposed an idea to do a kind of single-molecule mass spectrometry for DNA sequencing. There were some differences in that proposed approach, but in those early days, I guess, it was an idea that was too far ahead of its time. Then, it didn't get funded, but at this point, we are more optimistic that the environment is right to be able to pursue this.
Have you patented your method?
Brown has protected the intellectual property on this. We have filed patent applications on the concept of combining nanopores and mass spectrometry for single-molecule analysis. In fact, I actually think the promise of combining these two tools goes beyond just DNA sequencing, because in principle, you could imagine protein sequencing, and if you can sense single ions with high efficiency, then the notion of, for example, lysing a cell and doing an inventory of all of the molecules that are inside becomes attractive.
Since you are based at Brown University, do you have any connections to Nabsys, a Brown spin-out?
Yes, I do. Nabsys is just up the street. They have their own sequencing technology based on nanopores. About two years ago, I was approached by them to consult. And, in fact, one of the principals at Nabsys, John Oliver, was also involved with Peter Webber, my current collaborator, in this idea more than 10 years ago to try to do single-molecule mass spec sequencing, although it did not involve nanopores, because they did not exist at the time. But this is not their core technology; they don't have the IP on this — their core technology is based on detecting tagged DNA electronically.
This idea of mass spectrometry[-based sequencing] is one that I would be happy to explore with industrial partners. The most natural ones that would come to my mind are ones who already have, of course, experience with mass spectrometry.