Skip to main content

Purdue Team Eschews Microscopy in Favor of Holography to Study Cellular Drug Response

Premium
Most cell-based assays rely on high-powered microscopy to observe how cells respond to certain stimuli, but a research group at Purdue University is taking a different approach.
 
Using a digital holographic system based on a laser and a charge-coupled device, the research team led by physics professor David Nolte has demonstrated that it can observe how organelles move within tumor cells at the nanometer scale.
 
More importantly, the system can quantifiably measure how the motion of these organelles changes when introduced to small-molecule compounds.
 
Earlier this month at the annual meeting of the American Physical Society in Denver, Nolte presented preliminary findings showing how the platform can be used to screen anticancer molecules.
 
Cell-Based Assay News caught up with Nolte by phone after the meeting to find out how the approach might be adapted for high-throughput screening.
 
How does this technology differ from other imaging methods?
 
It differs a lot. In microscopy, you go in with a high-powered microscope and see cells and maybe see what’s inside of cells. And the area that we’re working in is known as cellular motility, and virtually all cellular motility is done under high-magnification microscopes. You might be looking at a single cell or maybe inside a cell and seeing how things move and then how that changes when you apply drugs or other stimuli to it. So that’s sort of the current state of motility assays, high-magnification microscopy. But what it also means is that they’re only really looking at a single cell, and typically it’s a cell sitting on a culture dish, so it’s some sort of two-dimensional environment for that cell.
 
What we’re doing is very different from that. We’re actually almost working in the far other extreme. We use very low magnification. You can’t even call it microscopy, in fact. It isn’t. We basically broadly illuminate tissue with a laser, so we can’t even see cells. Our resolution isn’t anywhere near where we can even see cells, so it’s really just broad illumination of tissue.
 
This is where the interferometry comes in, and interferometry is the most sensitive way of using light to detect small thickness changes or distance changes. Essentially, in interferometry, you’re using the wavelength of light, which is about a micron, as a yardstick. So essentially, one wavelength, or one micron, becomes your yardstick, and then you’re able to measure very small displacements — very sub-micron, down even to about the level of about 10 nanometers of displacement. You can actually detect that because you’ve got this yardstick that’s only a micron in size. So 10 nanometers is 1 percent of that, and you can typically pick up a 1-percent spatial shift of something compared to a wavelength of light.
 
So the bottom line is that while we are broadly illuminating tissue, and therefore have no microscopic resolution at all, we can pick up 10-nanometer motion using the interferometric aspect of our technique. That’s really novel, and a lot of people have a hard time zeroing in on that, actually. The idea that you’re sort of doing this really wide-field view, and yet getting nanometers’ worth of displacement information out is amazing to many people.
 
I guess it’s counterintuitive.
 
Well, the idea that you’re getting nanometer-scale information and yet you’re interrogating millimeters’ worth of tissue — the ratio of those two numbers, one millimeter to one nanometer, is a million. So we basically have a million-to-one dynamic range in terms of our ability to sense these really small displacements, but over these really wide-field views.
 
And that’s what makes this more biomedical and less biology. The motility people in biology really want to see the cells and see the organelles moving around in the cells. But in our case, we’re more interested in what the response of a tissue is to a stimulant. In our case, we work with anti-mitotic drugs, AMDs, and our AMDs so far have been microtubule based. So that’s actually what we’ve been doing, is looking at things like [the ovarian, breast, and non-small cell lung cancer drug] taxol, [the gout treatment] colchicine, and [the antineoplastic drug] nocodazole. Colchecine in small doses is an anti-inflammatory, but in large doses it actually inhibits microtubule polymerization. And nocodozole also inhibits microtubule polymerization, but taxol actually stabilizes the microtubules against depolymerization.
 
So we’re now able to look at this broad tissue illumination — and a millimeter isn’t exactly a big number, so by calling it broad tissue I’m using words that a microscopist would use. A doctor would think that a millimeter was small, but somebody who does microscopy regularly would think that a millimeter was really big, so it’s all kind of how you look at it.
 
But we look at a millimeter-size tumor, and we apply the anti-cancer drugs, these AMDs specifically, and then we can watch what happens to the sub-cellular motion. So we’re actually doing motility assays with nanometer-scale resolution, and we can watch organelles.
 
For example, with nocodozole and colchecine, you’re inhibiting polymerization, but microtubules are very dynamic things, so they have dynamic instability and they’re constantly polymerizing and depolymerizing. But if you inhibit polymerization, then basically they dissolve, they depolymerize, and those are the main highways that the organelles move along. So if the microtubules dissolve, then the organelles don’t have their highways anymore and so they just kind of go into random Brownian motion.
 
And we can pick that up. We can really pick up the very high motility when everything is healthy, but then when we inhibit the polymerization of the microtubles after several minutes, you can watch the organelle motion settle down, really diminish.
 
When you say you can watch, what exactly is the readout for this system?
 
The image of the tumor shimmers under laser light. It doesn’t shimmer if you just use ordinary white light. You don’t see this effect without a laser, so this is very much a laser optics effect, which is the interferometry part of it. And we’re actually doing what’s known as digital holography. So we’re coming in with a reference beam that interferes destructively and constructively with the signal beam.
 
But basically, we’re just shining lasers on tissue, and then they shimmer — the image of them when we look at them on a digital camera, it shimmers. And then because it is digital, we download that data and we can analyze the statistics of the shimmering and come up with these numbers that are basically motility metrics. So then you look at the healthy tumor before you apply anything and you set a good baseline, so you know how motile it was. And then you apply the drug and you actually watch the motility decrease with time as you affect the microtubules.
 
So then we can get dose-response curves. We can get the amount of motility change as a function of dose and time. And again, it’s rather remarkable that we’re doing this as a broad number coming off of this broad area of tissue, and yet it’s this sub-microscopic motion. So we’re able to do this really high sensitivity detection of motion without having to have the high microscopic resolution.
 
And that’s what gives it potential for high-throughput [screening], because to do motility assays, if you’re working under high resolution, you’re pretty much just seeing one cell at a time and frankly, they aren’t dividing that much. It’s like looking for a needle in a haystack to find the cells that are actually in mitosis because most cells are spending their time in interphase. Mitosis is actually a fairly small portion of the cell cycle in terms of total time.
 
It’s actually kind of hard to go hunting and poking around with a microscope looking for mitotic cells to then maybe investigate what these AMDs are doing. And while people do that, you’re not going to automate that. You’re not going to be looking at hundreds of drug candidates all at the same time.
 
So what we have is actually perfect for looking at hundreds of drug candidates at a time because, for instance, you just take a 384-well titer plate, which is now pretty much the industry standard. So you could actually have several millimeter-scale tumors in each well, and then you could go with even lower resolution so that you’re now imaging that whole microtiter plate onto the digital CCD. And so in some sense, you might be able to look at 384 wells all at the same time. And since we’re just looking for this shimmering effect — we don’t need to be working in high resolution — we might be able to look at all 384 tumors at the same time. And you might have applied different drugs to different wells, so in a matter of 10 minutes, we might be able to get something on the order of 100 dose response curves, for either 100 different doses or 100 different candidates.
 
So that’s the high-throughput part of it and that could be interesting.  
 
Do you have any kind of proof of concept for that application?
 
Nope. We’ve been playing around a little bit with scale affects — seeing what happens if we go out to lower resolution and things like that, but that’s all in the development phase right now. This is something for the future.
 
What would you estimate is a reasonable timeline for getting to that point?
 
Well, my graduate student [Kwan Jeong] is about to graduate, so that’s going to delay things. But if it weren’t for that I’d say within a year we’d probably be at that point.
 
Is this something where you’d be willing to partner with a commercial entity for further development?
 
I’d be willing to partner with a commercial entity, sure. 
 
What other applications do you envision this having in terms of drug screening and response?
 
Well, the more general way of looking at this is that we measure motion, so you can almost think of motion as a new type of contrast agent. And that’s pretty novel, because people have not been able to see motion inside of tissue before. So this is three-dimensional biomedical imaging, and yet our contrast agent is the motion itself. Therefore, anything that affects the motion, we pick up. In fact, you could probably apply all types of drugs — not just these anti-mitotic drugs — that could influence motility. Even things as simple as if you shut down oxidative phosphorylation, you could shut down ATP generation, and that’s certainly going to effect the kinesin molecules and the myocins that are dragging the organelles around and things like that.
 
So in a way, it’s almost like taking the internal temperature of tissue. It’s not real temperature, but just like you use a thermometer to tell whether someone’s sick or not, this is almost along the same lines. It’s a way of using all the motion going on inside of cells, and there’s a lot of it. Cells are very motile and the things inside of them are moving around a lot. But it’s really using motion as a gauge of how healthy the tissue is.
 
Therefore, [you could measure] anything that affects the health of the tissue, where either you want to kill it off, like some kind of an anticancer drug, or you don’t want to kill it off, meaning, let’s say, you’re screening for toxicity of some drug, … and that toxicity might show up in terms of motion. If you’re killing the cells, or if you’re making the cells unhappy, then the motion is going to decrease.
 
What are your next steps in terms of developing this?
 
I’ve become very interested in a broader class of AMDs. All the AMDs that we’ve looked at so far have gone after tubulin specifically, but that affects all cells. If you mess up the tubulin, or the polymerization, even good, healthy, non-cancerous proliferating cells, you’re damaging them. So there are other, more specific, classes of AMDs, where they might go after a kinesin that is only specific to mitosis, so you’re not messing with the microtubules, so that non-dividing cells remain healthy and happy, and you’re only affecting those cells that are in mitosis. And then, probablistically, the cancer cells tend to be in mitosis more often than health cells. So then you can just target the dividing cells specifically rather than targeting all cells.
 
You presented this work at the American Physical Society meeting. Do you plan to publish this in a peer-reviewed journal?
 
We are just about to submit this to a journal.

The Scan

Rise of B.1.617.2 in the UK

According to the Guardian, UK officials expect the B.1.617.2 variant to soon be the dominant version of SARS-CoV-2 there.

Anne Schuchat to Retire

Anne Schuchat is retiring after more than 30 years at the US Centers for Disease Control and Prevention, Politico reports.

US to Share More Vaccines

CNN reports that the US will share 20 million doses of the Moderna, Pfizer, and Johnson & Johnson SARS-CoV-2 vaccines with other countries.

PNAS Papers on Gene Therapy Platform, Aspergillus Metabolome, Undernutrition Model Microbiome

In PNAS this week: approach to deliver protein-based treatments to cells, pan-secondary metabolome of Aspergillus, and more.