|
|
Contents | EBJ Home | Single molecule measurements and biological motors |
The choice of laser will largely be determined by the fluorophores of interest. Increasingly, solid-state lasers are becoming available in a wide range of wavelengths. For example, for fluorophores such as Cy3 and rhodamine, we use a frequency doubled Nd:YAG laser (532 nm). The laser is thermoelectrically cooled and so there are no fans to generate vibration, and the laser head itself is very compact. The laser produces 50 mW which should be ample for most purposes; if necessary the beam can be attenuated using e.g. neutral density filters. For EGFP we use an 50 mW argon ion laser with emission at 488 nm. These lasers are rather more unwieldy, but solid state alternatives are now becoming available. An excitation filter is necessary to eliminate other spectral lines emitted by this type of laser.
Although 50 mW sounds
very feeble, it is quite enough to do some serious damage. Many of the same
points apply as when using lasers for trapping.
Eyewear is an important precaution, but remember that the beam will usually
be invisible through it. Minimising or controlling stray reflections is important,
as is directing the outgoing beam to a suitable beam dump. Providing safety
interlocks is also advisable.
Background light is one of the main problems with TIRF microscopy - especially for the detection of single molecules. Background light can arise from many sources.
Light at the excitation wavelength can be a major source of background, particularly as it is so much brighter than the fluorescence that one is attempting to observe. The most significant measure that can be taken to reduce this effect is to choose an appropriate good quality interference filter. The filter chosen should have a very high absorption at the excitation wavelength, while transmitting the desired emission light very efficiently. Often a band-pass filter centred around the wavelengths of interest is most suitable, as this helps to exclude unwanted light at other wavelengths.
Minimising the extent to which the excitation light is scattered at various places in the set-up is also an important step, since some of this light will always penetrate the filter. With objective-style TIRF it is particularly important to avoid scattering within the objective lens. Other surfaces - such as prisms, mirrors, lenses and slides - should be kept clean and dust free to minimise scattering. Scattering from objects in the flow cell can also cause unwanted fluorescence in the solution away from the surface. The use of beam dumps is also important, since the outgoing beam from a TIRF set-up is very intense and potentially a huge source of background.
Another important source of background is fluorescence from sources other than the target molecules. There are many sources of such background fluorescence. One of the most significant, particularly for prism style TIRF, is microscope slides. Ordinary glass microscope slides are remarkably fluorescent, due to impurities in the glass. One alternative is to use quartz (fused silica) microscope slides, made from a material such as Spectrosil 2000, which is 99.9999% pure silica. Unfortunately, microscope slides in this material are not commercially available and must be custom made - which means that they are relatively expensive. We are currently investigating the availability of low fluorescence glass microscope slides. Similar issues occur with the cover slips to be used for objective style TIRF, although here it appears to be less critical, perhaps because of the shorter path length. However, for this method the refractive index is critical, so that quartz cover slips may not be suitable.
Other sources of unwanted fluorescence may include substances such as immersion oil, glycerol and various reagents which may be used in the flow cell. For many materials research grade reagents are sufficiently pure to avoid a problem. Nitrocellulose is often used in motor-protein studies to immobilise molecules on surfaces, and unfortunately here, even the highest grades of material have significant fluorescence. However it may be possible to pre-bleach this fluorescence before use (Conibear and Bagshaw, 2000). Dust and debris can also be a significant problem. The whiteners used in typical laboratory wipes are extremely fluorescent, and fibres from them are often abundant in laboratories. Furthermore, many dust particles appear to take up significant amounts of fluorescent dyes, or to scatter the excitation light. Therefore it is important to minimise dust, although extreme measures should not be necessary.
One source of fluorescence that is harder to eliminate is autofluorescence in cells. Using TIRF helps to reduce the level of autofluorescence because a much narrower depth of the cell is excited. However, autofluorescence can make it difficult to visualise single molecules in the region of the cell body.
In all fluorescence microscopy, photobleaching can be a problem. In single molecule microscopy, where the molecule under observation effectively disappears abruptly, this can be a major headache. This is particularly the case where one is studying association and dissociation processes, when dissociation is hard to distinguish from photobleaching. One approach to minimising photobleaching is to reduce the excitation intensity; however this may not be practical if the sensitivity of the detector is not adequate. In general, the photobleaching process is not well understood but is known to require oxygen. Therefore photobleaching can be minimised by use of an oxygen scavenger system. We use an enzymatic system of glucose oxidase and catalase, together with DTT which, together with degassing of buffers, we find to be quite effective.
Other poorly understood photochemical and photophysical processes can be observed in single molecule fluorescence (Moerner and Orrit, 1999). Spectral shifts, intensity fluctuations, and the effects of fluorophore orientation and environment can all play a role. The effects observed will depend on the fluorophore and the molecule to which it is attached. They may cause significant problems, or be easily ignored or avoided.
The field of low-light level imaging is developing rapidly, and new technologies are emerging all the time. Probably the most likely detector is some kind of intensified CCD camera, but this covers a wide range of different technologies. The key issues to focus on are sensitivity and noise at different intensifier gains; spectral characteristics; and spatial resolution (which is often limited by the intensifier, rather than the CCD).
Another factor is to consider whether a camera with analogue or digital output is required. Analogue cameras are usually cheaper and typically use the CCIR or NTSC video standards. These standards limit resolution and frame rate (to 25 or 30 Hz respectively), but inexpensive video recorders and other equipment are readily available for storage, processing and digitisation of data. However, these cameras are rather inflexible and the analogue video data must be digitised for analysis.
Cameras with digital output get around most of these problems, and often data acquisition and camera control can be done purely by computer. Often there is great flexibility of frame rate, and other parameters. However, storing large volumes of digital video data can be awkward - although the increasing size of hard drives is helping. Cutting out the analogue stage eliminates several forms of noise and improves the quantitative aspect of the data.
Often it is desirable to follow fluorescence in a quantitative way over time. This can be done with imaging sensors, but this can be problematic. Some imaging detectors are not particularly easy to use in a quantitative way. Furthermore, the vast amount of data that is collected by an imaging detector can be very useful (if it is desired to collect large amounts of data in parallel) but it can also make data analysis and storage unwieldy. Another disadvantage is that the time resolution of imaging detectors will be quite poor (e.g. 25-30 Hz). Image artefacts such as interlace, intensifier noise and so on complicate data analysis. Quite often, it will be enough to collect data from a single point over time, perhaps in parallel with other measurements.
Essentially, there are two types of detector with sufficient sensitivity for single molecule fluorescence studies. Photomultiplier tubes (PMTs) are specialised vacuum tubes in which electrons released from a photocathode are amplified with very high gain (~106) and excellent signal to noise ratio. Because the detector gain is so high, the amount of amplification needed downstream is reduced and the contribution of amplifier noise is minimised. For single molecule studies a PMT is used in “photon counting” mode where each pulse of electrons is individually recorded. Perhaps the main disadvantage of PMTs is that their quantum yield (the fraction of photons hitting the photocathode which are observed as counts) is typically low (20% or less) particularly at the red end of the spectrum where many popular fluorophores' emission maxima lie. Avalanche photodiodes have much better quantum yields (up to 90%) but do less well in terms of gain (102-103), so that amplifier noise becomes significant. They also suffer from a relatively high dark current. Careful consideration needs to be given when choosing these detectors.
One problem that may be experienced in TIRF applications, particularly at the single molecule level, are various kinds of variation in the level of illumination. Foremost among these is the variation in intensity of the evanescent wave across the field of view. This is primarily due to the gaussian intensity distribution of the input laser beam, although it becomes elongated as it reflects from the interface. However, if the area of illumination is wide enough then the variation across the centre of the field of view is relatively small.
Another problem that may occur is the formation of interference fringes and other artefacts due to the highly coherent nature of the illumination. We have not found this to be a problem, but some workers have used a rapidly-rotating ground glass disc to "smear out" the fringes. Another issue which may be significant is the polarisation of the evanescent wave, which is not isotropic and may have an effect if the fluorophore is not free to rotate. The polarisation depends on the polarisation of the input beam. Some workers have addressed this issue by using a quarter wave plate to convert a linearly polarised input beam into a circularly polarised one.
|
Previous |
Contents | Next |