Nomad

The CoMPLEX ITPL programme is a crash course in the many different ways we can measure what's going on in biological systems. In a series of dense lectures, researchers co-opted from a variety of fields describe techniques either developed or advanced to solve problems in those fields. There's a lot of underlying theory (quantum mechanics, photochemistry, physiology, etc), a lot of complicated engineering (just to make the dratted processes happen), and then a lot more difficulty in taking the outputs and turning them into something that can perhaps go towards answering the real question, whatever that may be. The whole mess can get a bit bewildering at times; which is just how I like it :)

Anyway, following up on the lectures, last week was taken up by the ITPL practicals, which is to say: spending each day at a different lab getting variously exposed to a selection of those instrumental techniques. And what a fun-filled experience it was.

Monday was spent in the basement of the Physics department, in a room full of lasers¹. The particular techniques of concern were those of fluorescence lifetime imaging, with a side order of anisotropy and energy transfer.

A molecule fluoresces when one of its electrons is temporarily excited to a higher energy level by an incident photon, only to drop back to its ground state after an intermediate energetic relaxation. The redistribution of energy in the interval results in the emitted photon having a longer wavelength than the absorbed one -- it is red-shifted -- and that difference is (statistically) characteristic of the molecule². The bit of molecule that does this is called a fluorophore, and the time between it absorbing and emitting is the fluorescence lifetime.

An ordinary microscope allows you to look at a reasonably small point on your target, subject to various limitations³. If the target is illuminated with light at the wavelength your fluorophore absorbs, the microscope will pick up some of the corresponding emission photons, which will have a different wavelength. Add in a filter to block all the photons that aren't that wavelength and what comes out is your basic fluorescent image.

Now, suppose that the illumination occurs in very tiny pulses. Fluorophores will be excited by a pulse and then -- sometime later -- emit. If you know exactly when you fired the pulse and you know exactly when you picked up the emission, you can gauge the fluorescence lifetime. Do that a lot of times and you'll get a reasonable picture of the lifetimes at the point you're focussing on⁴.

If you repeat this process at different points across the sample, you can build up an image of the fluorescence lifetimes in different locations, which is a new layer of information on top of the fluorescence intensity image you already have. Fluorescence lifetimes are sensitive to the molecular environment, so this information may be valuable.

Hey! Where are you going? Come back, there's more!

Fluorescence is orientation dependent. Molecules exist in space and can point this way and that. The fluorophores have dipole moments⁵ along which their absorption characteristics manifest. The probability of an incoming photon causing fluorescence is tied to its polarization. Angled the wrong way, it's never gonna happen.

Since we control the polarization⁶ of the incident light and can also filter the light we detect, it is possible to use this to estimate orientational changes among the molecules over the fluorescence lifetime. This is particularly interesting if we have (courtesy of genetic engineering) a molecule with more than one fluorophore on it -- one end glows green, the other red -- because it can give information about the relative orientations of the two fluorophores, and thus about the conformation of the molecule. Which would take us on to FRET, but this post is already dead weight and I've only done Monday...

Tuesday found me in Medical Physics as both experimenter and guinea-pig.

Working out what's going on inside people's bodies -- and heads -- is a problem of pretty much unlimited scope. Bodies are complicated things and their owners have attitudes. They don't like being prodded and probed⁷. They get litigious when things go wrong or, you know, whenever they just feel like it. They are, basically, a pain in the arse.

They're also pretty fragile. You can't just go tearing in there to see what's going on, there'll be blood and guts and all sorts of mess. It'll end in tears.

As far as possible, you want to look inside people non-invasively. If only they were transparent,

Well, that may be too much to ask, but it turns out that many human body tissues are sort of translucent. We occlude, but not completely. There are wavelengths of light that pass through us relatively easily. Some, like X-rays, are horribly energetic and damaging, only to be used as a last resort. But the near infra-red region also turns out to be handily penetrative without all those nasty ionizing side effects.

It's not transparency -- IR beams don't pass neatly through the body to be focussed into a crisp image on the far side -- but flesh is permeable enough to IR to allow all kinds of measurements to be made. The rays are scattered, but only partially absorbed -- and the degree to which they're absorbed can reveal interesting things, such as the oxygenation level of the blood⁸.

While this is medically useful already -- and you've quite likely had the measurement taken yourself, you can do it in Boots these days -- for our purposes it's just a stepping stone towards more detailed information. The body needs oxygen in order to do work and is generally quite good at getting it to where that work is happening. So measuring local oxygenation in different places can give a good functional view of where the action is.

Which is why I wound up with optodes all over my head doing anagrams in a darkened room.

Actually constructing a useful image from such measurements -- and even just making the measurements themselves -- is a painstaking business because of the low level of transmission, the high levels of noise, various confounding physiological processes and the awkwardly complex mixture of stuff we contain. The resolution is never going to be high, but it's safe and practical and potentially important.

Wednesday was spent at the Cell Biophysics lab at Cancer Research UK, down by Lincoln's Inn, doing some sample preparation and mass spectrometry. The samples in question were membrane lipids from sea urchin gametes, and the preparation involved many stages of dissolving and filtering and centrifuging and drying and such: proper biochemistry with test tubes and everything. The mass spec was much duller by comparison, very impressive machinery but mostly computer-controlled and rather slow.

Phospholipids are fatty molecules that make up the bulk of cell membranes, key structural components of all lifeforms more complex than a virus. Membranes were traditionally viewed as rather simple things, thin waterproof bilayers held together by hydrophobia⁹, almost trivial beside the mind-boggling complexity of DNA and proteins. However, given the enormous range of interactions between the inside of a cell and the big bad world, and the astonishingly precise control most cells can exercise over what goes in and out, it is clear that membranes are actually seething hotbeds of vital activity. While much of that is mediated by embedded proteins, the lipid composition is also intricately involved. Determining that composition in detail is no simple task.

Mass spectrometry, which uses an electric field to separate out particles according to their mass-charge ratio, can help, provided the particles are actually charged. To get it to work, the lipids are broken into fragments and the resulting spectrum analysed to see which fragments came from where. Of course, there are a lot of different possible lipids with varying arrangements of fatty acid chains, fragment masses and charges. With only a single spectrum there would be too much noise to get much sense of where the spikes were, so the feed into the mass spec is first run through a high-pressure liquid chromatography column, which spreads the fragments out according to their hydrophobicity.

The resulting data set is still pretty intimidating. I'm very glad not to be responsible for analysis beyond a casual conversation about a single experiment. Our mentor for the day was dealing formally with hundreds or maybe thousands of them...¹⁰

Thursday was back to UCL for some very hands-on patch clamping in the Pharmacology.

Among their many exciting features, lipid membranes can contain little protein assemblies called ion channels that control the movement in and out of particular ions. These nanomechanical marvels can discriminate¹¹ the types of individual atoms and if they're not on the guest list they just don't get through. That's atoms we're talking about. Do you have any idea how small they are?

The passage of specific ions through the ion channels constitutes an electric current, and handling such currents is one of the things cells do that makes them alive. The currents are not large -- of the order of pico-amps -- but they're important to the cell and also -- rather amazingly -- measurable in the lab.

In order to do that, you need to isolate some channels on the cell surface with a very tiny glass tube: both the glass and the cell membrane are good insulators, and when they're firmly clamped together the only route left through which current can flow is the ion channels.

First catch your tiny glass tube.

Creating a microscopic pipette for patch-clamping is a surprisingly simple, surprisingly manual and also somewhat hit and miss business. Basically, you take a small glass tube, heat it in the middle until it gets soft, and then pull the ends apart until it breaks. There are a bunch of refinements along the way, but given favourable circumstances that process will give you the pipette you need; on top of which it's also rather fun.

You then fill it with electrolyte and attach it to an electrode on a rig that can gear down simple hand cranks to produce very sensitive adjustments in each direction. The whole caboodle is attached to a powerful microscope: you carefully maneouvre it down onto a cell et voilà. There's a giga-ohm resistance seal and the only currents that can flow are through the ion channels.

It's not quite as haphazard as I paint it, but you do have to take a certain amount on trust. Which may be why it's so gratifying when it works.

Ion channels are far too small to see with your own eyes, no matter how powerful the microscope, thanks to the resolution limit mentioned in footnote 3. So there's no telling what's included in the patch of membrane to which your pipette is attached. But as you increase the membrane potential -- which is one of the things that can tell the ion channels to open -- you start seeing a tiny current flow, increasing or decreasing in very obvious steps, all the same size.

A single ion channel lets through a fixed amount of current when it's open; all channels of the same type having the same conductance. So each of those steps represents a single open channel. What you're "seeing" as the current jumps up and down is the opening and closing of those channels -- something so small it is barely possible to imagine.

How cool is that?

Finally, on Friday I was in Physiology, somewhere in the depths of the medical school. The experiment combined both the electrophysiological patch clamping of the day before with fluorescence techniques related to Monday's, only this time it was two-photon fluorescence. Used, in this case, to observe a living neuron in something approaching 3d.

In ordinary one-photon fluorescence, as described earlier, you shine energetic (usually UV) light through your sample and everywhere it hits a fluorophore some less energetic light is emitted. This is a great process, but it has disadvantages. All that UV light pouring in can damage the target tissues. Fluorophores are often unstable and can stop being responsive after even relatively short exposures. Since any fluorophore in the light's path will fluoresce, not only can you photobleach what you're looking at but everything nearby too. Worse, you have to jump through some vexing optical hoops to eliminate all the emissions that aren't coming from your focal point, and in the process you sacrifice a lot of photons that you can't afford to lose.

In the two-photon technique, the fluorescence is triggered by two lower-energy photons hitting the fluorophore at the same time. That may seem obvious, but there's a problem: the ground state and the excited state are strict quantum energy levels. There is nothing in between. Really, it should be impossible to get bumped from one to the other in two stages.

Enter Heisenberg's Uncertainty Principle, which states that there is an innate vagueness in the energy of a system in a certain amount of time, and vice versa. This uncertainty is very, very small: even for the really tiny energy state differences we're talking about in our fluorophore, the margin for error is minuscule. But it does exist. As long as the two photons hit the fluorophore virtually simultaneously, they can be absorbed and everyone's happy.

In order for there to be any chance of it happening at all, there have to be a vast number of photons in the immediate vicinity of the fluorophore, which sounds like a nuisance but is actually the greatest benefit of two-photon fluorescence: it only happens where the photon flux is highest, which is exactly at the focal point.

The rest of your sample doesn't fluoresce!

You don't need to worry about filtering emissions from everywhere else, because there aren't any. The lower energy light you're shining into your sample just blunders harmlessly through. Every higher-energy emission photon you collect has to be from the point you're looking at. You can scan a live cell in rastered layers and build up a complete 3d picture of the fluorophores within it.

Of course, it's a delicate procedure and you'll need hundreds of thousands of pounds worth of equipment to do it, but at least it's easier than three-photon fluorescence...

---

1 Cover your eyes if you bend down to pick something up.
2 This is the process that makes dayglo items show up under the ultraviolet lights in nightclubs -- they absorb the high energy UV photons and emit lower energy visible ones. The loss goes, pretty much, to heat, though this effect is vanishingly small compared to that of you shaking your funky groove thang.
3 Probably the most important limitation is the wavelength of light you're viewing with. Imagine trying to measure a grain of sand with a yardstick marked in inches -- there just isn't enough resolution to make a meaningful measurement. Light is a yardstick with much finer markings than that, but there's still space between the markings and that space limits what you can resolve.
4 Naturally, things are not quite so simple. You're dealing with molecules and photons en masse. A single excitation event will (perhaps) lead to a single emission event after a particular time, and the emitted photon may or may not head towards your detector. Other events will have other lifetimes and go off in other directions. What you detect is a fraction of the available events with a range of different lifetimes. So the result is not a single lifetime, but a distribution of them.
5 Don't even ask, okay?
6 We control the vertical. We control the horizontal.
7 Oh come on. Just insert your own smutty innuendo here.
8 Another simplification, of course. You need to take measurements at more than one wavelength and do a bit of linear algebra, but the details are relatively straightforward and the basic technique has been around since WW2.
9 Not, in this case, a synonym for rabies; rather an expression of the tendency of some molecules or parts thereof to interact unfavourably with water, a substance overwhelmingly present throughout all biological systems.
10 That's a PhD for you. In a year's time I may be facing something similar...
11 Let's just ignore the implicit anthropomorphism of this sentence.
Posted by matt at November 27, 2006 06:32 PM