We cell biologists get around.

After being awarded our PhD, we tend to move from lab to lab, taking on whatever new project is on offer. Like guns for hire, we follow the stimulating scientific puzzles from city to city, and often from country to country.

The variety of cell biology out there is staggering. Cells can be part of a human being, or a turkey, or a frog, or a worm. Cells can fly solo, as do single-celled baker’s yeast. Cells can go awry in diseases ranging from Alzheimers to zygomycosis. You can study cells in the context of an organism, an organ, a tissue, or as single tiny machines made up of thousands of working parts – or you can just study one of those parts, such as a single gene. So moving to a new lab can be like entering a brand-new galaxy of opportunity and adventure.

As I’ve negotiated my own path through various labs over the years, explaining my project to my family and friends used to be easy:

I’m studying a cat virus that invades cat cells and makes them into cancer cells.

I’m studying a molecular switch that causes out-of-control cells to commit suicide.

I’m studying a chicken virus protein that slaughters tumor cells but leaves their normal healthy counterparts unscathed.

But my current stint has been giving me problems in the making-conversation-at-cocktail-parties department. In fact, I’ve begun to dread the usual question, So what is it you do, exactly?

The thing is, I don’t study one discrete disease, or organism, or particular gene any more. I’m studying a condition of cells: their shape. Animal cells are incredibly plastic and dynamic – they can spread, contract, round up, split in two, crawl around, send out ruffles and fingers and spikes and ripples. Their shape is controlled by a complex network of scaffolding, motors, moving parts and regulators in a way that’s been studied intensely for years but that we don’t yet entirely understand at a whole-systems level. The way cells move and take up various shapes is absolutely crucial for carrying out their function, and if this goes wrong, disease can result. For example, the UFO shape of our blood cells is important for allowing them to be whisked around our arteries and veins in an efficient, streamlined fashion, and diseases that mess around with the UFO shape (like sickle-cell anemia) lead to dangerous obstructions in our circulatory system.

So far so good: that’s all understandable. But it’s the way I’m investigating them that makes it difficult to explain to my mother. I’m doing a so-called “reverse genetic screen” to work out what makes an animal cell take up the particular shape that it does – a sort of brute-force approach where you can study hundreds or even thousands of an animal’s 20,000+ genes in one single experiment. “Genetic”, because I’m looking at the effect of various genes; a “screen”, because I’m surveying a large number of genes at one time, in the hope that a few interesting ones will pass through my intellectual filter. And “reverse”, because I’m going about it backwards. Instead of the classic method of looking at a cell effect and then trying to work out what genes are responsible, I’m doing it the other way around: messing around with the genes and then seeing what effect that has on the cell.

The way I think about screens has been permanently colored by a lovely analogy that was doing the rounds when I was in graduate school in Seattle in the 1990s, about alien auto aficionados. (Other great analogies have since been used too, such as Yuri Lazebnik’s paper about shooting radios with BB guns, but I like the car version because cars, like cells, are dynamic and move around.) Imagine you’re an alien from across the galaxy, observing Earth and trying to figure out what a car is and how it works. You’ve noticed that cars can do a wide variety of things: open up to allow humans to enter and leave, start up, move, brake, turn, reverse. Some of a car’s behavior is easy to work out by detailed observation: the car door and its handle are simple mechanical devices and it’s clear how they allow their passengers in and out. The wheels clearly allow the car to roll. Other processes, however, are more complex and mysterious: how does the car engine start? What makes the wheels turn, or brake? Using simple observational techniques, the aliens soon reach the limits of what they can understand easily.

Clearly a shift in scientific tactics is needed. So, after a little brainstorming session, the aliens decide to gather up thousands of cars, as many as there are working parts inside a single car. They systematically damage a different component in each of the otherwise identical car, and watch what happens. Alien scientists soon realize that snapping off a wing mirror has no effect on ignition, but removing the distributor abolishes it completely. Assuming they’ve press-ganged some human guinea pig drivers to their scientific cause, they would soon see that removing the steering wheel leads to loss of directional control, and breaking the axle causes the wheels to fall off – they gradually learn to distinguish between mechanical effects and regulatory effects. Some damaged parts can be clustered into the same category: sabotaging the pedal, master cylinder and brake discs all lead to the same outcome: failure of the car to stop when it needs to. So these three parts are probably all in the same “pathway” – in other words, they control the same ultimate effect, so you can guess that they might work together in the same subsystem. Our aliens might initially place the wing mirror into a “useless” category – until, in its absence, some guinea pig drivers were seen to crash into other cars.

Putting together the results of all of the thousands of breakages, together with their older observational studies, the aliens would probably be able to get a fairly good idea how many car parts work, and a rough idea about what makes a car do what it does. Dr Lazebnik, in the BB gun/radio paper I cited above, gives a very thoughtful explanation about the limitations of this approach, and how biologists are sometimes misled by these sorts of screens, and the mentality behind them. In my view, I agree that screens aren’t perfect, but they can be a productive first step. For example, having established that the wing mirror wasn’t just a piece of shiny, decorative bling but was somehow involved in preventing collisions – unlike the hood ornament, whose damage had no effect in all tests – the aliens could abandon the crude screen method, and set up a team to study how the mirror worked in more detail.

Despite all the frailties and limitations of their approach, I’ve been emulating the alien auto aficionados now for the past few years. Instead of studying thousands of cars, I’m growing cells in thousands of individual partitions. In the image below, each of those black plates contains 384 wells, and in turn, each tiny well can grow about 3000 cells comfortably – I can use as many plates as I need.

Then, using a technique called RNA interference, I’ve been systematically switching off the function of one individual gene in the group of cells growing in each individual well. So in each well of cells, I’m studying the effect of removing one working part of the cell, and seeing what happens to cell shape and movement as a result.

I started off doing this with fixed images – imagine our aliens breaking a car part, putting a driver inside and then taking a static photo of the resulting effect a few hours afterwards. Here is an example of what normal cells (top) look like compared to cells where I’ve silenced a gene called Citron (bottom).

As you can see, the silenced cells have multiple nuclei instead of just the one they are supposed to have – which makes sense because we’ve known for years that Citron helps the cell body divide in two: if the cell body doesn’t divide properly, then you get two copies of the genome stuck inside one big cell.

But looking at still images is pretty crude, especially for subtle or rare behaviors. An alien would have to be pretty lucky to catch a wing mirror-less car in the act of colliding, for example. So after a first pass with fixed images, I repeated my entire screening experiment using a microscope set-up for timelapse video. If you’re studying movement (like how a car steers, accelerates and breaks), this approach is going to be a lot more useful. Similarly, if you want to know how a cell wiggles its membranes or divides into two, a video is going to be a lot more informative than a still photo taken at an arbitrary time. In the first video below is an example of how my normal cells look on film – and in the second video, you can see what happens when you silence Rac1, a gene that helps make dynamic protrusions or ruffles at the edge of the cell. In the absence of Rac1, the cell edge is a lot less mobile.

So how successful has this approach been? The underlying principles of RNA interference – using a small piece of ribonucleic acid to “silence” any gene you want – is based on a cellular defense mechanism first discovered in the 1990 by Napoli and colleagues, and later, researchers began exploiting this phenomenon for genetic screens. Since then, the technique has become routine and highly automated, with the use of robotics to perform the experiment and computers to analyze the data. Unfortunately we started out our project a few years before all the high-tech solutions became routine, so I’ve been going about things in a more old-school way – with smaller robots, and more reliance on what my eyes can see. But that’s okay. Although staring at many gigabytes worth of image data can sometimes be tedious – and was probably single-handedly responsible for my recent drastic shift of spectacles prescription – I’ve nevertheless enjoyed seeing first-hand what my cells are getting up to. This was never more true than with results that are spectacularly beautiful as well as informative:

This past summer I published the results of my screen for cell shape, and currently I’m studying a wing mirror of my own in more detail – a gene whose silencing doesn’t seem to have much effect, until you catch it in the act during cell division. If viewed at just the right time, shows one of the most bizarre defects in cell division that I’ve ever seen:

What is going on here? I wish I knew. The DNA, which was replicated before division commenced, is supposed to partition evenly into two daughter cells. Instead, both sets of DNA go into one daughter cell, and the other re-spreads without any DNA at all. And herein lies the limits of a genetic screen: the screen has told me that this gene is important for faithful cell division, but it cannot tell me how. For this, I’ve got to go back to more traditional techniques to study how the gene might be doing what it’s doing: biochemistry, forward genetics and molecular biology. Screens, for all their hard work and cinematic scope, are only the first step.

Which is why, sometimes, we cell biologists take a break and relax at cocktail parties. So forget my work: let’s talk about you.