Bench philosophy: Mechanotyping of cells
by Steven Buckingham, Labtimes 05/2016
There has been a growth in methods of quickly sorting cells on the basis of their physical properties. This approach is attractive because it is label-free and amenable to rapid sorting. New methods, developed in parallel with microfluidics, are pushing the boundaries of this important set of new techniques even further.
Sorting cells is a key step for a lot of biological and medical applications. It means we can perform experiments on sub-populations of cells to determine their physiology en masse. And there are those situations where you want to separate out cells from a tissue for medical applications, such as getting hold of a population of stem cells from bone marrow.
A micropipette force probe is used to measure the tiny forces a T-lymphocyte (l.) is exerting on its target. Photo: Julien Hussson
Automated cell sorting often works by finding some characteristic feature of the cells that can be used as a basis to separate the cells. This separation can be brought about using a physical setup: we do that every time we use a paper filter. Or you can do the separation in a more sophisticated way, using a machine to measure a cell's defining characteristic and use that measure to sift the cells. FACS (Fluorescence-Activated Cell Sorting) is a well-known example, in which cells are identified by the presence of a fluorescent label and automatically directed into different fluid streams.
But this kind of approach often means you have to do something to the cells to make them recognisable in some way. In the case of FACS, for example, you have to introduce a fluorescent marker so that a machine can tell one cell type from another. FACS and related methods are famously effective but they can be a bit laborious (requiring cells to be labelled in some way or even genetically transformed), and stuffing a cell with a protein that's not meant to be there may introduce some unwanted side-effects.
So, wouldn't it be great if there was some way of sorting cells based on some “innate” property?
Fortunately, cells do indeed differ naturally in many ways. Increasingly, scientists are looking at ways of sorting cells by their biomechanical properties. The great thing about exploiting mechanical properties is there are so many of them to get working on and they are conceptually easy to understand. Like cell density, for example – cells of course increase in weight when they make new proteins.
But who has got a balance capable of measuring such tiny changes? Actually, several labs have come up with some ingenious devices for doing just this, taking advantage of advances in piezoelectric technology. For instance, you can create an array of piezoelectric pins and culture your cells on that, a bit like a fakir lying on a bed of nails. As the cells do their usual business of crawling around their spiky bed, the forces they generate are read off from the piezoelectric pins.
This, however, is a very unrealistic environment for cells. For one thing, they are stuck in two dimensions. A method that overcomes this is to embed the cells in a gel along with lots of microdots. As the cells pull themselves through the gel matrix, the deformations of the gel are revealed by the movements of the microdots. You can record and analyse these movements using video cameras and motion tracking software. These approaches have been very fruitful in distinguishing subtypes of cells in a population and adding insights into phenotypic changes in health and disease.
By applying an electrical charge, you can also use piezoelectric devices to move cells around. That means, once you have cells growing on a piezoelectric tip, you can poke and prod them to measure their biomechanical properties. A bit like an Atomic Force Microscope – but on a chip.
Very impressive, of course, but this is all very low-throughput. Is there a way of applying these principles to get something working on the same scale as a FACS machine? Fortunately, cells differ in many ways other than just how heavy they are. In fact, they differ quite markedly in several different mechanical properties. Think of the contrast between tough, inflexible epithelial cells compared to soft, pliable macrophages. Cells have been characterised using technologies like Atomic Force Microscopy, Traction Force Microscopy and Optical Tweezers, to uncover details of their tensile strength, viscosity and elasticity. To a first approximation, you can think of these properties in the same way as you would for a solid. Do you remember physics lessons at school, when you measured how much a rubber band stretches per unit of force applied? Don't tell me you have forgotten all about Young's modulus! Go to the back of the class!
Actually, cells have physical properties somewhat different to rubber bands. After all, they are watery bags of gel, packed with organelles of different physical properties. The physical properties of cells have been described by experts as “tensegrity”, a combination of tension and compression, which is thought to result from a balance between a tensile force coming from the actin and an opposing force from the microtubules. So, we have a complex set of biomechanics, resulting in cells that do all sorts of different things when you pull or squash them.
Can this be used to phenotype cells? Yes. An established and conceptually very simple technique is to pull the cells into a micropipette and look at how they deform. This is called “micropipette aspiration”. The pressure in the pipette is carefully controlled and measured, and the way the shape of the cell changes as it gets sucked into the pipette is observed and measured using microscopy.
Think back to the rigid epithelial cell and the squashy macrophage again. Ignoring any size differences (which you can adjust for by using different sized pipettes), it is easy to picture the epithelial cell keeping its shape but the macrophage bending quite easily, all at the same pressure. All you have to do now is to record the cell shape and you have found a way to mechanotype cells. But we are still nowhere near high throughput yet. That is going to take a bit more biomechanical subtlety. And, once again, the inventiveness of the bench scientist has been at work. Last year, a paper by Per Augustsson's group at MIT and Lund University (Sweden) described how you can use the acousto-mechanical properties of cells to sort them automatically in a microfluidics device (Augustsson et al., Nat Comm DOI: 10.1038).
Just in case you are one of the tiny minority of biologists (a mere 99% of them, I would guess), who didn't know that cells even had acousto-mechanical properties, or even what acousto-mechanical properties are, let me explain.
When sound passes through a medium, it gets reflected or refracted just like light does. Now, we have already mentioned, in passing, the complicated viscoelastic properties of cells. Well, as if that wasn't complicated enough, it turns out that their acoustic properties are just as byzantine. So let's simplify things a little. When a sound wave hits a cell, it will get scattered and that will cost the sound wave a little bit of energy. Where does that energy go? Why, into the cell of course. That means you can push cells around using sound.
If you are feeling a high-throughput, cell sorting idea coming along, let me tell you that you are on the right track. But sorry – others have had the same thoughts too, and there are several methods for sorting cells by pushing them around with sound. Here's the basic idea: cells have different acoustic properties, so if you apply a sound wave of just the right amplitude and frequency, some cells get pushed further than others. It's like electrophoresis, so they called it acoustophoresis.
But there's a snag – and it is to do with cell size. Cells with the same acoustic properties will also get sorted according to their size, simply because bigger cells absorb more energy but have more inertia. Now, sadly, the similarity with electrophoresis is working against us.
That is where the really clever part of Augustsson's technique comes in. They got around the problem by creating a standing sound wave in a medium of varying acousto-mechanical properties. This is how it works. First, you get a cell medium flowing along a tube. You find a substance that alters the acoustic impedance and inject it into the middle of the stream. As it spreads out, you create a lateral acoustic impedance gradient. Add your cells and apply the sound wave. As the sound wave hits the cells, it pushes them along the gradient. But eventually, the cells arrive at the part where the gradient's acoustic impedance is the same as that of the cells. Now, instead of scattering, the sound passes straight through the cells, imparting little or no energy to them on the way. No more force, no more movement, cells sorted. It's like HG Wells' Invisible Man and that old SciFi trick of making something invisible, by giving it the same optic properties as the air around it. Iso-acoustic focussing is the same idea – but with sound instead of light. Okay, so does it work? Well, Augustsson showed how they could very effectively separate out neutrophils from monocytes and lymphocytes using this approach.
Oliver Otto's lab at the Technische Universität at Dresden took another approach to getting over the cell size problem (Xavier et al., 2016 Integrative Biology DOI: 10.1039). The name of their technique, “real-time deformability cytometry”, says it all. Their idea is to force cells in suspension through a parallel array of 20 μm x 20 μm channels. The squeezed cells get deformed, and you measure this and the cell size, using video analysis. Cell size and deformation are plotted in two dimensions for each cell, and Otto's group found that they could easily distinguish skeletal stem cells from other mesenchymal stromal cells cultured from bone marrow.
Sure, we have long been able to enrich these cells using traditional techniques like gradient centrifugation; it is just that you get a better yield this new way.
Another neat trick, used by several labs, exploits the inertial and deformational differences between cells, to separate them in a flowing stream. In this case, you get the cells to flow through a series of strategically-placed pillars. Once again, what happens to a cell when it hits a pillar depends on several mechanical features but what is important, is that they will differ in how much they are displaced.
So imagine a cell that has just bumped into a pillar and has been deflected to the side. We now place a pillar in just the right place, so that the cell gets bumped aside by about the same amount again. You can do this if you place the pillars in rows, offset by just the right amount, so that the cells hit the pillars at the same place in each row. It's a bit like dropping balls through a peg-board – get things just right and the balls will tend to go to one side.
Many mechanotyping techniques are striking for their sheer simplicity. Amy Rowat at the University of California, Los Angeles, used a simple (simple in principle, at least) micro-filtration device made of a series of polycarbonate membranes with different pore sizes (Qi et al., Scientific Reports DOI: 10.1038/srep17595). The cells are pushed through by applying pressure to the medium – simple as that. By getting the pressure and the cell density just right, they were able to get quite large differences in retention, even between transformed versus untransformed cells of the same cell type.
The viscoelastic properties of cells affect the way they respond to compression, and Todd Sulchek and his team at the Georgia Institute of Technology exploited this fact in a particularly elegant way (Wang et al., Lab on a Chip 15 doi 10.1039). They set up a microfluidic device, in which cells encounter a set of parallel ridges set at an angle to the direction of flow. The height of the ridges allows just sufficient clearance for the cells to squeeze under. This means that when the cells encounter a ridge's leading edge, they experience a compressive force. Because the ridges are set at an angle, part of this force acts sideways to the direction of flow, pushing the cell sideways. Eventually, the cell passes under the ridge, where the cell now has no lateral force and so continues along the direction of flow. The point is that the cell moves laterally, for as long as it takes for the compression to be complete and that, of course, differs from cell to cell. Sulchek's device was able to separate K562 cells (a leukaemia cell line) from HL60 cells (another leukaemia cell line).
Many of these new mechanotyping methods need several parameters to be carefully matched to the cell you are working on, such as the diameter of microfluidic channels and the separation of pegs. But these devices obviously increase the set of tools available for automated cell sorting and plug a gap where other sorting techniques fail. And of course, the big advantage is that you don't need to alter the cells' phenotype.
Last Changed: 05.10.2016