|Contents||EBJ Home||Single molecule measurements and biological motors|
Click to return to the top of this page; to view the graphical overview.
Typically, an optical tweezers apparatus is constructed around a commercial microscope body. It is quite possible to simply construct the apparatus from components mounted on an optical breadboard, but it is difficult to set up a good imaging system from scratch unless you are a real microscopy expert. On the other hand, it can be also quite difficult to make the necessary adaptations to a commercial microscope body, as they are not generally designed with this type of application in mind. A certain amount of ingenuity, and the courage to drill holes in expensive microscope components are required!
The design of the microscope body is itself important: as with many other components, mechanical stability is crucial. Not only must the microscope not be susceptible to vibration, but components such as the focussing mechanism need (ideally) to be free from drift.
Key to the function of the optical tweezers are a number of components of the microscope, most centrally the objective lens.
This, with the laser itself, is the fundamental trapping component. It must be selected for its suitability for trapping, but also be compatible with your other imaging requirements (e.g. TIRF, DIC). The key requirement for trapping is that the objective should have a high NA (typically 1.2-1.4). This ensures stable 3D trapping at the focal point. To achieve this high NA, an immersion medium is required, usually in the form of immersion oil (n = 1.51). However, spherical aberration limits oil immersion lenses to trapping quite close to the cover slip. Alternatively, water immersion lenses offer the possibility of trapping much deeper into the flow cell.
In most set-ups, the choice of condenser will be determined by specific imaging requirements such as DIC, and is usually purchased as part of the microscope package. The condenser may play a more critical role if it is being used as part of the position sensor, in which case issues such as mechanical stability should be considered.
The chief requirements of the mechanical stage are good stability and ease of sample manipulation. Typically, a piezoelectric substage is mounted on the mechanical stage and used for fine motions and forcing functions; therefore the main function of the microscope stage is to provide a solid platform, movable in the x, y plane, with minimal drift and vibration.
The choice of illumination also depends on specific imaging requirements. The example shown in the schematic has a mercury lamp for epifluorescence and a halogen lamp for bright field illumination. Bright field illumination is used for imaging trapped beads and also as the light source for the position sensors. The stability and intensity of this light source are limiting factors in the resolution and bandwidth of this detection system. Alternatively, laser light may be used (see position sensors).
If the apparatus incorporates fluorescence imaging, the normal complement of excitation and emission filters, and dichroic mirrors will be required. In addition, barrier filters are needed to block or attenuate trapping laser light from parts of the microscope where it is not needed, such as the cameras and eyepieces. Additionally, some form of dichroic mirror is needed to introduce the laser beam into the imaging path. This may be as simple as a "hot mirror" if the laser is being brought in separately, or a more complex dual wavelength dichroic mirror may be required if it is brought in in combination with the fluorescence excitation light (see schematic).
In general the eyepieces are little used on a typical trapping microscope, since using video and computer displays is more convenient (and fluorescence may be too faint for direct viewing). Also there are always safety concerns with using eyepieces in a laser-based microscope system. Barrier filters should be used to attenuate the trapping laser, and if possible the eyepieces should not be used while the laser is on. However, the wider field of view that the eyepieces give makes them useful for some tasks, such as adjusting condenser alignment.
A typical set-up will incorporate one or two video cameras. In the apparatus shown, an inexpensive CCD video camera is used for the bright field imaging. Even here, however, care in selecting components can pay dividends. Many inexpensive CCD cameras have very small sensors (e.g. 1/3"), which restricts the field of view in the microscope; we find a 2/3" sensor to be ideal. In addition to this, another camera may be required for fluorescence imaging. In the apparatus shown, we use an intensified CCD camera to allow real-time imaging of single fluorescently labelled actin filaments or DNA molecules.
A typical apparatus would include one or two monitors and a video recorder. The monitors can display the output from the video cameras and can also display system information generated by the computer. Additionally, a frame grabber may be used to permit digitisation of video.
Alternatively, the availability of fully digital cameras permits one to dispense with all of the above and display and store video data by computer. This obviously has many advantages, but storing large amounts of digital video is sometimes problematic.
The acquisition and storage of data, and control of the apparatus, are performed by computer. For typical data collection rates, the specification required is not too high; we typically use a 486 or Pentium computer running DOS. However, if the computer is to be used for data analysis (particularly in real time) more computing power is likely to be required. A colour monitor is useful for displaying information about the apparatus configuration and experimental conditions, and displaying data as it is acquired. A good sized hard disk and backup facility (e.g. CD writer or network storage) is advisable, as hopefully large volumes of data will be collected!
Data acquisition will generally require additional hardware. Many commercial data acquisition cards are available. It is worth evaluating carefully whether these will meet your requirements. It is also possible to build your own data acquisition card with some electronics and computing knowledge. A typical card will have a number of analogue inputs with analogue to digital converters; these are used, for example, to digitise data from the position sensors. Digital inputs are also useful for recording information such as the position of sliders and other microscope components (the inputs can be connected to micro-switches or opto-switches). Other useful inputs can include counters e.g. for photon counting in conjunction with a PMT. An accurate, programmable, timing function is also essential, to permit data collection at high rates. On-board memory may also be useful for high-speed acquisition.
This type of card will generally also furnish a number of analogue (with digital-to analogue converters) and digital outputs. These can be used to control various components of the apparatus. For example, the positions of the optical traps and of the piezoelectric stage should be controlled via computer.
Other hardware may be interfaced in different ways; for example, many lasers can be controlled via the computer's serial port. Parallel and "game" ports and even the sound card provide alternative methods of interfacing hardware.
Typically, additional electronic components may be required between the computer and the apparatus. For example, the signal from a quadrant photodiode requires pre-processing to produce the x and y signals which are fed into the computer.
Analogue video data can be acquired using a frame grabber; alternatively, digital video may be acquired using a digital frame grabber or other interface card. A frame grabber may also be used to overlay information on the video screen.
Another aspect needing careful thought is the software to be used to control the apparatus and acquire the data. We have tended to use DOS as the environment in which our programs run, because of its relative simplicity and the extent of low-level control of the hardware which it makes possible. However DOS is a 16-bit environment and this imposes severe restrictions, most notably the limited amount of memory which can be accessed (although there are some tricks which enable this problem to be reduced). Programming in a GUI such as Windows is more complicated and may also limit low level access to hardware, but has many advantages in terms of the user interface and access to system resources. One can also choose whether to program in a low level language such as C, or a software environment such as LabView.
The software used to analyse the data "off-line" is also an important factor, although it is not restricted by the need for low level hardware access. Running the software under a 32-bit operating system has many advantages. Custom software will generally be required, which may be written in a language like C, or in a development environment like MATLAB which comes with many routines for signal processing and other analytical tasks.
By active components we mean those which can move in response to computer control. These can be used to set up the experimental geometry (e.g. to position an actin-tethered bead pair over a myosin coated pedestal) or to impose a forcing function on the experimental system.
In the typical trapping apparatus, the acousto-optic deflectors serve four main functions:
(See the detailed discussion of AOD's).
An x-y stage is essential for many trapping applications. As well as providing positional control, it allows forcing functions to be imposed, which are useful for calibration of trap stiffness and detector sensitivity. Piezoelectric transducers suffer intrinsically from problems such as creep and hysteresis, so to ensure maximum stability a feedback stabilised stage is ideal. The stages we use have capacitative sensors, and give excellent stability, positional resolution and repeatability. The control units can be interfaced to a computer by analogue or digital means. It is often necessary to fabricate a custom adapter to link the piezo stage to the microscope stage, and a slide or specimen chamber holder to fit the piezo stage.
The trapping laser is, of course, a key component of the set-up. The key parameters of the laser to be considered are:
There has been considerable discussion in the literature about what are the most suitable trapping wavelengths for biological trapping experiments; the best type will depend on the application. The main factors to consider are discussed below.
Because of the very high powers (typically 107 W·cm-2 for a 100 mW laser power; Neuman et al., 1999) at the focal point of a trapping laser beam, there is potential for a number of problematic side effects. The problems are particularly acute when trapping living cells. Very high intensities may lead to cell death (opticution); intermediate intensities may simply impair cell viability. The problems are less acute, however, when using isolated molecules which are manipulated with bead "handles" - as in molecular motors experiments. Even in this case, it is wise to be aware of the potential for photochemical effects, particularly in fluorescence experiments. Three mechanisms for optical damage have been considered:
The main cause of "opticution" under typical trapping conditions has been shown to be probably a single-photon process mediated by oxygen (Neuman et al., 1999); thermal effects are likely to be negligible. Two-photon processes may come into play under some conditions.
The key to minimising such effects is, not surprisingly, to minimise the absorption of laser energy. The general trend has been to use lasers with near infrared wavelengths. This spectral region lies between the absorption maxima of most biological molecules and those of water. Lasers in this region of the spectrum include the popular neodymium lasers Nd:YAG, Nd:YVO4 (both 1064 nm) and Nd:YLF (1047 nm); diode lasers and tunable Ti:sapphire lasers are also available in this region. Neuman et al. (1999) have determined the action spectrum of opticution in this region, which should aid in the selection of lasers if this issue is a major concern. Another observation is that removing free oxygen (if possible) may minimise any problems.
It is worth considering whether the wavelength of the laser will interfere with other aspects of the experimental design, e.g. fluorescence imaging or quantitation. Provided the wavelength is sufficiently different, the monochromatic nature of laser light generally makes it fairly easy to filter out.
A number of the laser types mentioned above fall into the favoured wavelength range for trapping. However, the different types of laser vary in much more than their output wavelength. Some of the issues to consider are the beam quality, power output and stability (discussed in more detail below) as well as cost, which may vary significantly. It is for this reason that the Nd lasers have been most popular.
The power required for trapping will depend on a number of experimental parameters, but 100 mW at the specimen plane would usually be ample. However, it is worth bearing in mind that much of the output of the laser will not reach the trap. Losses in objectives and acousto-optic deflectors may together amount to 75%, without considering other optical components. Typically a laser with a nominal power output of a few watts is used. Most lasers have the capacity for a variable power output; however it is worth noting that manufacturer's specifications for stability are usually given at the nominal power output. In our experience laser instability, particularly amplitude noise, may be much worse at lower (or higher) powers. If this instability is a problem, the laser may be run at nominal power and the laser intensity controlled by changing the AOD efficiency or by other means.
To produce an optical trap with an approximately parabolic potential well, the laser needs to produce a single transverse mode beam, typically TEM00, which gives a Gaussian beam profile. Alternatively, the "doughnut" shaped TEM01* mode may be used (Ashkin, 1992), or the Laguerre-Gaussian modes used for optical spanners.
For any experiment where the position of the bead in the trap is being measured, it is obviously desirable to minimise any positional noise introduced by pointing instability in the laser beam. This becomes an even more significant issue if trapping laser is also being used for position detection.
As well as the inherent pointing instability of the laser cavity, extraneous noise can also be introduced by air currents or vibrations. The first point can be addressed by completely enclosing the beam (we achieve this using a combination of cardboard tubing and foam rubber). The second by isolating the system from mechanical vibration which can propagate throughout the system. If the system is mounted on anti-vibration table, it is important to avoid sources of vibration on the table itself. For example, some designs of laser head incorporate a cooling fan.
Variations in the output intensity of the laser can be considered in two main categories at different time-scales. Drift in output power occurs over time periods longer than the typical duration of an experiment, and will cause variations in trap stiffness between experiments. This can be controlled for to some extent by regular stiffness calibrations. More pernicious is higher frequency noise which contaminates experimental data. This is particularly an issue where the laser is the light source for position detection. If this type of noise is a problem it will usually be apparent from the power spectrum of a trapped bead, giving rise to a non-Lorentzian distribution. For example, the spectrum may contain peaks at particular frequencies or be contaminated at low frequencies by so-called 1/f noise.
It is worth considering how the laser will be controlled and monitored. Many lasers can now be controlled remotely by computer, often using an RS232 (serial port) connection. Incorporation of laser interfacing into the software facilitates automation and permits laser parameters (e.g. operating power) to be stored in data files with other experimental conditions.
At the risk of stating the obvious, a powerful laser is a significant safety hazard, most critically to the eyesight of the operator and his colleagues. Laser safety is beyond the scope of this article, but laser operators should ensure that they are trained in laser safety and take all necessary precautions such as safety interlocks, eyewear, beam containment etc. In a well designed apparatus the beam should be largely contained within the apparatus; enclosing the beam in tubing also helps to exclude air currents which can affect pointing stability. Bear in mind that the lasers typically used emit invisible beams!
The position sensor(s) are a key component of any optical trapping system. A variety of different detection mechanisms are used, which may permit the motion of a trapped bead to be followed in one, two or even three dimensions. The use of multiple traps may complicate matters, requiring the use of multiple detectors, or a method for separating the signals from different traps. They also require supporting electronics to produce a suitable output for digitisation. See the separate discussion of position sensors.
Additional electronics will often be necessary to interface between the computer and external components. These circuits may be incorporated in some kind of rack system or in separate boxes. The latter system is more awkward but can give improved noise isolation between circuits. Example circuits are given for processing the output from a quadrant photodiode; for producing dual traps and for separating the signals from dual traps. Access to good facilities and expertise in making custom electronics is invaluable.
The construction of an optical tweezers apparatus requires many optical components such as lenses and mirrors, and possibly beam-splitters, polarisers and the like. It is useful to choose an optical table with a grid of mounting holes, although magnetic mounting is also useful. Additional optical breadboards may be required, as well as rails, mirror and lens mounts and the like. Unfortunately these components can be very expensive. It is certainly worth shopping around and trying to strike a compromise between cost, design and quality. Components from several manufacturers may need to be used and this can result in problems of compatibility. Here access to a good mechanical workshop is very useful; custom components or adapters between other components can then be produced as required.
A very simple flow cell can be constructed from a slide, cover slip and lines of vacuum grease dispensed from a syringe needle. This will suffice for a simple motility assay, but for optical trapping experiments more mechanical stability is needed. We generally construct flow cells using slivers of cover slip as spacers and a suitable UV curing glass adhesive. Many flow cells can be prepared weeks in advance.
Good isolation from vibration is essential for good quality data. There are a number of measures that can be taken:
The most important reason for temperature control is to avoid mechanical drift due to the thermal expansion or contraction of microscope components. Ideally an air conditioner should be running continuously to maintain a stable temperature. Obviously a room without windows will be preferable (and in any case probably necessary for laser safety). A stable temperature is also necessary to ensure the consistency of biochemical measurements. It its worth noting that many aspects of the experiment are only slightly affected by temperature - for example the variance of the thermal motion is proportional to the absolute temperature - in other words at room temperature, a 1C change will give only a 0.34% difference. If it is desired to perform experiments over a range of temperatures it will probably be necessary to control the temperature of the whole room - regardless of operator comfort! (This will avoid issues such as mechanical drift due to temperature fluctuations, condensation, etc.)