Scientists at the Vanderbilt University School of Medicine and the Oak Ridge National Laboratory have developed a technique for imaging whole cells in liquid using a scanning transmission electron microscope. Electron microscopy requires a high vacuum, which has previously prevented the imaging of biological samples, such as cells, in liquid.
The investigators claim that their liquid STEM approach images single molecules in whole cells with significantly improved resolution and imaging speed compared to existing ultrahigh-resolution optical imaging methods such as stochastical optical reconstruction microscopy, photo-activation localization microscopy, and stimulated emission depletion microscopy.
In a study described in this week's Proceedings of the National Academy of Sciences, the team imaged in liquid single gold-tagged epidermal growth-factor molecules bound to the cellular EGF receptors of fixed COS7 fibroblast cells. The cells were placed in a buffer solution in a silicon microfluidic device with electron-transparent windows inside the vacuum of the STE microscope.
The investigators obtained a spatial resolution of 4 nm and a pixel dwell time of 20 µs. The liquid layer was sufficient to contain a layer of cells with a thickness of approximately 7 µm.
Niels de Jonge, an assistant professor of molecular biology and biophysics at Vanderbilt University and a staff scientist at Oak Ridge National Laboratory, spoke with CBA News this week about liquid STEM and its application in drug discovery. De Jonge is the lead and corresponding author on the PNAS paper.
The following is an edited transcript of the interview.
Give me a little background on STEM microscopy and the liquid STEM microscopy technique.
I think there have been some good results where you have a single gold particle or thin bacteria, but the resolution was never high enough for use on eukaryotic cells. That is really what is happening in our paper.
The main thing is that, traditionally, people have used a transmission electron microscope to image biological samples, which is a little different than a scanning transmission electron microscope. The way that you make images and the contrast mechanism is different.
It just really needs very thin samples, and there have been some really nice results, especially in materials science. But it is always limited to 500 nm to 1 µm of thickness.
I work in the biomedical area. There, you really want to investigate eukaryotic cells, because they have relevance for human health. That is so far not possible at a high resolution.
That is the key here. The trick is to use the STEM in combination with the silicon chip technology.
How does what you discuss in this paper differ from what was done previously using STEM microscopy?
People have traditionally used STEM microscopy for material science. We have tried this in combination with a liquid cell. People have tried to do some things using transmission microscopy, but were always limited in terms of sample thickness, the same with STEM.
So what you are talking about here was using STEM to image biological material, and using a microfluidic chip in combination with the microscope to image that material?
Yes.
To your knowledge, is this the first time a paper on this topic has been published?
I think we are brand new on this. To give you some perspective, there have been some results published on microfluidic chips, but they focused more on materials science and used transmission microscopy, and they never got a high resolution on such a thick layer of liquid.
Actually, people believed that this would be impossible to do. I have spoken in the last year to several authorities in the field and they told me, "What you are doing is not possible."
But then I showed them my calculations, and my calculations said that it was possible. I also have my experimental proof.
How exactly does this technique work?
I have actually been working for four years on this. I am originally from the Netherlands, and I was working for Phillips Research and I was doing research for a microscopy company. But I wanted to establish this new technology.
[ pagebreak ]
So I came to Oak Ridge National Lab to work on it, and I developed the silicon microchips in collaboration with Protochips, which is a startup company. Over the four years' time, I developed four prototype holders, because it is very difficult to place these microchips in the microscope, actually.
In the end, I also worked with another startup company, Hummingbird Scientific, and in collaboration with them, I built the fluid holder. So that means you have the silicon technology and the holder technology, but it is all precision mechanics.
So you have a whole piece of hardware, in this case, that you have to develop.
I have also developed all this theory, to really understand what is going on. We also had to develop the biological procedures, so we had to find a good biological system where it works, and we also had to make sure the labels worked. So it was a year's work for the postdoc to get all of these biological experiments running.
It then takes a lot of time running experiments on the microscope to sort out these four components — the biological experiments and the labels, the chip technology, the holder, and the theory behind it. In the last year, it was about 50 days on the microscope to sort out the details, and you run into little problems and optimize the imaging conditions.
For example, you really need to provide proof that you are imaging in liquid.
How would you provide proof that you are imaging in liquid?
The best way was to measure the current in the detector, and to compare that with a theoretical model. So you can see when we have a sample in vacuum, and only 1 percent of the current is basically scattered into the detector. Most of the current goes in the current direction. It is not scattered.
Then if you have liquid, and often 50 percent or more of the current is scattered away, you see that your image gets really intense and there is a lot of noise, and you can measure that directly.
There were a few more indications written in the paper, where we proved that there was liquid. Also we have the flow going, so we know that we have liquid going in the microscope.
There are two tubes — input and output. We initiate flow, and we see that we put water in one tube and there is water coming out of the output tube.
How would this technique be applicable to drug discovery?
That is really on a fundamental level. My aim for the future is to investigate cellular processes. What happens when a certain substance comes into the cell and finds a receptor? What exactly happens there?
We have some indications with the EGF receptor, although there are still many questions about what happens if receptors combine to start some process, and how that happens.
We hope that on that fundamental level, we can do experiments.
Could this technology be used be used to screen drug candidates or small molecules?
If you can image what the receptors are doing, basically, you can then inject some drugs into our flow system, and see if those reactions change.
I have to be honest. I am really a physicist or a biophysicist. So far, I have put most of my effort into getting this technology to work. I do not have much knowledge of the pharmaceutical area.
That is, for example, one reason why I am at Vanderbilt's medical school — to find collaborators among the biomedical scientists.
What do you see as the immediate next step in developing this technology?
We want to try different types of labels, for example, different sizes. We have to prove that we can distinguish different types of labels. That is one project we are working on.
I am also involved in a lot of discussions with the biomedical people to find different applications for this technology.
We also want to try and use this technology to image live cells. That is really a major challenge because we do not know what the electron beam does, the radiation.
So this is not currently being used to image live cells?
No. At the moment they are fixed.
Could these be used in conjunction with ultrahigh-resolution optical microcopy methods such as stochastical optical reconstruction microscopy, photo-activation localization microscopy, and stimulated emission depletion microscopy? If so, how?
Basically, liquid STEM is kind of a competing technique. What I am hoping for is that you would have the standard high-resolution confocal microscopy, and we would like to combine that to see, for example, the entire cell, or the cellular processes.
If we see some interesting events, we would then zoom in using STEM to see what is happening on that precise spot with the receptors. In our paper, we describe it as a snapshot of cellular function. I think if that works, then it could be used as an alternative to these techniques, but maybe it could be used together.
That is one thing that we need to sort out in the future.