by Jeffrey M. Perkel
If you want to view some of the best of microscopy, look no further than the galleries of the Nikon Small World and Olympus BioScapes competitions. Art, science and technology come together in equal measure to produce stunning images that would be equally at home in art galleries and scientific journals.
One of the key technologies driving the winners of these two competitions in recent years is confocal microscopy. Confocal produced two of the top three finalists in Nikon's 2010 competition and the first-place finalist in Olympus' 2010 competition, a dazzling kaleidoscopic image of a pair of daddy longlegs eyes.
Confocal microscopy is a fluorescence-imaging technique that produces exquisitely sharp optical sections through biological specimens such as tissue slices. The technique generally relies upon rasterizing an excitation laser over the sample and collecting emission data, point by point, to reconstruct the final image. The resolution is more or less the same as in traditional widefield microscopy, but the images are dramatically sharper thanks to the use of a pinhole aperture that blocks fluorescence-emission light not originating in the correct optical plane from reaching the detector. (In contrast, in widefield-fluorescence microscopy all light from the sample reaches the detector, producing blurry images.)
"Confocal microscopy offers several advantages over conventional optical microscopy, including controllable depth of field, the elimination of image degrading out-of-focus information, and the ability to collect serial optical sections from thick specimens," explains the Molecular Expressions website (an excellent source of information and tutorials on all things microscopic).
"Any time people want to have high resolution in fluorescence microscopy, confocal is typically considered the standard," says Brendan Brinkman, product manager for laser scanning confocal microscopes at Olympus America. The technique, he says, is "ideal for those working with tissues or specimens where they need to precisely know the location of proteins that have been marked with fluorescent dyes."
All four of the major optical microscopy vendors—Olympus, Nikon Instruments, Leica Microsystems and Carl Zeiss MicroImaging—offer confocal microscopes. These come in two basic flavors: single-point (laser-scanning) instruments, which generally produce the sharpest fluorescence images; and spinning-disk (or, more generally, field- or array-scanning) instruments, which traditionally yield the fastest images and thus are preferred for live-cell applications. Whether you plan to image fixed tissues or living cells, there's a confocal microscope to fit your needs. Just remember to carefully consider your current and future plans before making a purchase. "No one microscope handles all applications," says Michael Davidson, who runs the Molecular Expressions site from his office at the National High Magnetic Field Laboratory at Florida State University. "If you don't have the right scope for your application, then you are out of luck."
Laser-scanning vs. multipoint scanning
There are two fundamental ways to build a confocal microscope. Laser-scanning microscopes rasterize the excitation laser across the sample, building an image point by point. The emitted light from any given point (or pixel) typically is captured by one or more point detectors, such as photomultiplier tubes (PMTs), then fed to a computer, which then reassembles the image.
Multipoint (or array) scanners, in contrast, image larger swaths of a sample at a time. One common approach, exemplified in the Zeiss Cell Observer® SD, uses a spinning (or Nipkow) disk. A Nipkow disk is a spinning wheel studded with pinholes that enables the system to build images relatively rapidly (and with less phototoxicity than single-point, laser-scanning systems) by scanning samples at multiple pixels simultaneously.
The Cell Observer SD uses the Yokogawa CSU-X1 spinning-disk assembly, scans 1,000 points at one time and features two disks: a pinhole array (for confocality) and a microlens array (to focus the excitation light). Between the two disks is a dichroic mirror, which bounces the returning emission light to a detector—in this case, a charge-coupled device (CCD). (Unlike point-scanning systems, multipoint systems typically use array detectors, such as CCDs or EM-CCDs.) The system can capture images of 512 pixels x 512 pixels at up to 30 frames per second, says Duncan McMillan, group product marketing manager, BioSciences, Carl Zeiss MicroImaging, LLC. Another Zeiss multipoint scanner, the LSM 7 LIVE, uses line scanning to collect 512- x 512-pixel images at up to 120 frames per second.
According to Davidson, whose lab uses both point and array scanners, the former provide sharper images and better axial resolution, making them better for optical sectioning. For live cells, array scanners typically are preferred because of their speed. (The trade-off, though, is that array scanners, especially spinning-disk systems, typically exhibit poorer axial resolution and some loss of confocality due to point-to-point crosstalk and limitations in terms of objective, magnification and pinhole size, Davidson notes.)
That said, newer laser-scanning systems with "resonant scanning heads," such as Nikon's A1R and Leica's SP5 systems, also can capture images at video frame rates (30 frames per second). "The whole argument that spinning disks are faster is no longer true," says Leica marketing manager, Chris Vega.
Combining traditional galvanometer-based scanning with ultrafast resonant scanning, the A1R can even stimulate and acquire data at different wavelengths, says Jeff Larson, Nikon's confocal product manager. This enables researchers to perform tricks such as photostimulation of activatable fluorophores with one head while simultaneously imaging with the other. "You can stimulate and record data simultaneously," Larson says. "So you can watch an event unfold, as opposed to seeing it after the fact."
Detectors
The pinhole apertures that help confocal instruments produce such sharp images also are the source of one of their limitations: sensitivity. Because these pinholes block out-of-focus light, there are fewer photons hitting the detectors, meaning users need to use either more intense laser excitation or excite for longer. Both alternatives can lead to photobleaching. Alternatively, users can just use more sensitive detectors.
EM-CCD array detectors, such as those used on spinning-disk systems, are highly sensitive—they have quantum efficiencies in excess of 90% (meaning nine of every 10 photons to hit the detector will register a signal), says McMillan. But scanning confocal systems don't use CCDs; they use photomultiplier tubes, and most PMTs have quantum efficiencies of 20% to 25%, McMillan says.
Some confocals, though, including Zeiss' LSM 780 laser-scanning system, use what's called a GaAsP (gallium- arsenide-phosphide) detector. With peak quantum-efficiency values of around 45%, says McMillan, these detectors are about twice as sensitive as regular PMTs.
For Davidson, that's a significant improvement. There's "a big difference" between 45% and 25%, he says. "You sacrifice a lot of photons with the pinhole, so every photon is precious."
Another variable is the number of detectors. You don't necessarily need one for every color you want to detect in an experiment, says McMillan, unless you need to image those colors simultaneously. If you can get away with imaging sequentially (especially in fixed-cell applications), you can save some money, he says. For instance, the Zeiss LSM 700 has two PMTs that can be reassigned to detect different wavelengths on the fly. Using that configuration, researchers can detect four colors with just two detectors.
Spectral unmixing
Most confocal systems use dichroic mirrors and optical filters to send specific wavelengths to their detectors. Sometimes, though, you need to use fluorophores with overlapping spectral characteristics. That's where spectral detectors come in.
Spectral dispersion devices are tunable elements that use prisms or diffraction gratings to separate light into its component wavelengths. "Think of Pink Floyd's Dark Side of the Moon," says Vega. That separated light is then directed onto a series of detectors without the use of dedicated (i.e., fixed-wavelength) optical elements, where it can be resolved into component elements to, for instance, subtract autofluorescence.
Leica's SP5, for example, uses a prism and a series of sliding mirrors to capture any five user-selected wavelengths of emitted light in the system's PMT array. Zeiss' LSM 710 and LSM 780 and Nikon's A1R feature a dispersion device based on a diffraction grating that directs light onto a 32-element PMT array. McMillan estimates an upgrade to Zeiss' Quasar 32-element detector array from a three-element PMT detector would cost about $100,000.
Such detectors (which also are available as an optional add-on to the Nikon C2 confocal) "allow you to separate closely overlapping fluorescent probes quantitatively, such as CFP and YFP," Larson says.
According to Larson, these devices, which can add some $80,000 to the price of a Nikon confocal, are particularly useful for applications such as FRET (fluorescence resonance energy transfer) "or if you need to extract quantitative intensity information from your data and you have problems with spillover between channels or interference from autofluorescence." However, for those interested mostly in taking (non-quantitative) pictures to capture structural detail, unmixing algorithms combined with data from standard detectors should suffice, Larson says; in that instance, if you buy a spectral detector, "you're wasting your money."
Excitation sources
Key variables to consider when purchasing a confocal microscope are the number and type of excitation sources. Confocal systems use laser excitation, and most systems can accommodate a half dozen or more separate laser "lines" (or wavelengths), the choice of which will determine the fluorophores you can excite.
Both gas and solid-state diode lasers are available. Gas lasers can emit multiple laser lines, depending on tuning, but "they typically don't last as long [as diode lasers] and may not be as stable," says Brinkman. Therefore, diode lasers are preferred, he says, though many system configurations still use both classes.
For instance, Brinkman says one common configuration of the Olympus FluoView® FV1000 laser-scanning confocal includes six laser lines: 405 nm (diode), 458 nm/488 nm/515 nm (multiline argon gas), 559 nm (diode) and 635 nm (diode). The Zeiss LSM 710 and LSM 780 can have eight lines, says McMillan: diode lasers at 405 nm, 440 nm and 561 nm; an argon gas laser; and a pair of helium-neon gas lasers at 594 nm and 633 nm.
Some systems, such as the Leica SP5, use white-light lasers. Coupled with an Acousto-Optical Beam Splitter (AOBS®), or tunable beam splitter, the SP5 can emit "up to eight independent and simultaneous lines of emission" between 470 nm and 670 nm, Vega says. (Systems using physical devices, such as fixed-wavelength filter wheels and dichroic mirrors, says Vega, can select only one wavelength at a time, precluding simultaneous excitation.) Best of all, white-light lasers can accommodate new fluorophores and fluorescent proteins that haven't yet been invented. "Conventional lasers have fixed wavelengths, which are defined to work well with fluorophores that exist. But new fluorophores are being developed all the time," Vega says. "We are trying to use these fixed laser lines to image these fluorescent molecules, and they don't always match well for maximum efficiency."
Fortunately, laser selection is highly configurable, and users can upgrade their systems with new laser lines as needed. According to McMillan, it can cost anywhere from a few thousand dollars to about $40,000 to add a new laser. That price reflects more than the price of the laser itself, he stresses. It includes installation, a way to control the laser’s intensity and integrate it into the software, the beam profile, noise specifications and so on.
Other considerations
When it comes to purchasing a confocal system, the application is key. Will you be doing live-cell imaging or fixed-tissue sections? How about FRET, multiphoton microscopy or super-resolution imaging? The system you buy, says Davidson, "all depends on what you want to do," so make sure you get the features you think you need—or at least, make sure the system can accommodate those features moving forward.
For one Olympus customer mapping neural connectivity in brain slices, Brinkman says, a motorized stage was necessary. "That's typically a big force multiplier," he says, "because rather than having a manual stage that gives only one position at a time, you can program these stages to acquire very large areas."
Other variables include:
Ease of use: Confocal systems are complicated, with lots of moving parts, so to speak. The software to control them is likewise complicated. And though it is generally becoming more user-friendly, says Davidson, it still isn't easy. "Even the simplest software has a very steep learning curve," he says. So, consider how easy it is to use the software, to train new users and to perform basic operations.
Some companies offer what you might call "lite" versions of confocals, specifically for those who lack either the dedicated darkroom for full-sized systems or the know-how to use such a system effectively. Olympus' FV10i, for instance, is like a confocal-in-a-box—"self-contained," Brinkman says. "It's robust, but tailored for making it easy to use and to get people out of the dark."
Upgradability: When spending six figures on a piece of hardware, it pays to know it will be pulling its weight in years to come. Ensure that the system has sufficient upgradability to support your future plans.
Maintenance contracts: With so much going on under the hood, confocals are high-maintenance instruments. Mirrors can come out of alignment, lasers can go down—the possibilities are endless. And the repair bills can be substantial. So, if at all possible, purchase a maintenance contract. They are relatively expensive, says Davidson, "but if you lose a major component, buying a new part would be more expensive than the cost of the maintenance contract." McMillan says a good rule of thumb is that the contract will cost 5% to 10% of the system purchase price per year. But the contracts don't necessarily include everything, McMillan adds, so "read the fine print."
Finally, whatever you do, be sure to take any prospective system out for a test drive. "We encourage people to ask for demos, because seeing is believing," says Brinkman.
The image at the top of this article is LSM7 LIVE from Carl Zeiss MicroImaging.