When it comes to counting, examining and sorting cells in a moving fluid stream, researchers are not content to go with the flow. They are constantly looking for ways to improve the flow cytometers that are used for these tasks. In a flow cytometer, a fluid stream that carries particles or cells is hydrodynamically focused into a small volume and then interrogated, often optically. From these readings, researchers deduce cell characteristics. Three recent innovations promise improvements for different parts of the process. One delivers better fluid focusing through changes in chamber design. Another offers better multispectral analysis through the use of quantum dots — nanometer-size semiconductor crystals whose fluorescence is narrow, stable and size-dependent. The third combines fluorescence imaging with cell population information in a live animal. Tiny devices, focused results Commercial flow cytometers tend to be sophisticated and expensive, and they require large sample volumes. At the University of California, San Diego, postdoctoral researcher Claire Simonnet and assistant physics professor Alex Groisman have demonstrated a different approach — a microfluidic cytometer. In this high-throughput microfluidic flow cytometry device, arrows show the direction of fluid flow. The top image shows a micrograph of the device from above. The middle shows a schematic of the microchannel network, with the dashed box indicating the cytometry channel. A suspension of particles is injected into port B; flow focusing is provided by the liquids injected into ports A and C (focusing from top and bottom), along with liquid injected into port E. Port F is the outlet. The channels are a few hundred microns deep and of similar width. The schematic diagram on the bottom shows the structure of the flow in the device, from the 3-D focusing element to the cytometry channel. The liquid injected in inlet port B is dark in color. Image courtesy of Alex Groisman. Microfluidic devices are small and inexpensive, suitable for lab-on-a-chip and one-time-use applications, but, to date, microfluidic cytometry chambers have lacked something, according to Groisman; namely, 3-D flow focusing. Without tight flow focusing, cell velocities tend to vary a lot. The stream of particles spreads out, making optical detection difficult and correction necessary. The researchers, therefore, set out to modify the flow chamber and to improve the fluid focusing. They worked with poly(dimethylsiloxane) (PDMS), a transparent rubber. Using soft lithography fabrication, they fashioned the rubber into two devices capped by cover glass. One device was intended for high throughput, while the other was for high-resolution applications. Each had inlet and outlet ports, with connecting channels in the PDMS. One inlet was for injecting a particle-bearing fluid, while the others were for injecting focusing fluids. Both devices had 110-μm-deep cytometry channels, some of which were 200 μm wide, 200 μm deep and 6 mm long, while others were 120 μm wide, 400 μm long and 8 μm deep. Others had other dimensions. Viewed from above, the devices appeared to be a series of intersecting channels. Simonnet noted a bit of trial and error in determining the layout and size of the channels. “We have made a number of iterations to optimize the 3-D flow focusing and to enhance the performance,” she said. An important part of getting the fluid flow right, she continued, was the use of hydrostatic pressure. That approach ensured a stable forcing pressure, guaranteeing stable flows. For high-resolution imaging, the researchers used a Nikon inverted fluorescence microscope equipped with a standard mechanical stage and a 60x objective. They attached a Diagnostic Instruments Inc. cooled camera with a 1360 x 1024-pixel CCD and a maximum rate of 10 fps to the microscope. For high-throughput fluorescence detection, they used a Hamamatsu photomultiplier tube with a 488-nm-wavelength emission from an argon-ion laser for a light source. They focused the beam down to a 12-μm spot using microscope optics. In a series of experiments, they verified device performance. For the high-throughput device, they injected three types of 2.5-μm polystyrene beads, with nominal relative fluorescence intensities ranging from 100 down to 2.6 percent. They varied the inlet pressure and demonstrated detection rates of up to 17,000 particles per second. The work is detailed in the Aug. 15 issue of Analytical Chemistry. At the time these experiments were done, noise arising from the electronics would have kept the device from working with fluorescently marked live cells. Groisman reported improvement of the signal-to-noise ratio but no testing yet with live cells. For the high-resolution device, the researchers moved the microscope stage at 0.60 mm per second over 300 μm so that moving particles and cells in the device appeared stationary. Using exposure times of up to 100 ms over a field of view spanning 146 x 110 μm, they took images of polystyrene beads and yeast cells. These pictures did not show any streaks or blurring and were comparable with still micrographs. Fluorescently labeled yeast cells were focused to a thin flow layer in a high-resolution microfluidic flow cytometer. The cells were moving at a velocity of 0.57 mm/s. Researchers at the University of California, San Diego, moved the microscope stage with the same velocity in the opposite direction and took the images under bright-field illumination with a 1-ms exposure (top left) and under fluorescence illumination with a 100-ms exposure (top right). The cells are not blurred, indicating smooth flow in the cytometer and no rotation. Images of similarly labeled cells immobilized on a substrate were taken under bright-field illumination with a 1-ms exposure (bottom left) and under fluorescence illumination with a 100-ms exposure (bottom right). The still images and those from the cytometer appear equally sharp. Images courtesy of Alex Groisman, University of California, San Diego. To use the device for mammalian cells will require scaling it up roughly 2.5 times to accommodate up to 15-μm-diameter cells. Groisman said it would not be necessary to expand the already relatively long stable flow region by a similar amount; instead, a more powerful light source, different optics, another camera, a slower flow rate or some combination of these could be used. “All we need is enough photons of the fluorescent light from cells,” he noted. A different path to better flow cytometry was taken by a team that included researchers from the National Institutes of Health in Bethesda, Md., Solus Biosystems of Palo Alto, Calif., the University of Pennsylvania in Philadelphia, John Radcliffe Hospital in Oxford, UK, the University of Alabama at Birmingham, the University of Washington in Seattle and Carnegie Mellon University in Pittsburgh. The group improved the cytometer’s ability to characterize cells by labeling them with quantum dots. By adding these to more traditional labeling techniques, they resolved 17 fluorescence colors. Certain aspects of quantum dots make multicolor detection easier. The emission spectra (colored lines) of various quantum dots range from blue to red (a). Emission is narrow and symmetrical, allowing long-pass dichroic filters (blue diamonds) and bandpass filters (gray bars) to separate emission for different detectors. This octagonal photomultiplier tube detection system is configured to detect eight quantum dot fluorescences (b). Dichroic long-pass filters transmit light at a certain wavelength to bandpass filters. Light below a particular wavelength is reflected to the next dichroic filter in the sequence. Reprinted from Nature Medicine with permission of the researchers. Limit is colors Michael R. Betts of the University of Pennsylvania said that the practical limit today is 12 colors, a capability not often used. “The vast majority, 95 percent, of researchers typically use eight colors or less,” he said. Part of the problem lies in traditional labeling reagents. They have broad emission outputs and require different excitation wavelengths. To compensate, some systems have as many as four lasers and multiple detectors. The setup has to be extensively optimized, making it expensive to buy and to operate. Quantum dots, on the other hand, can be excited by any wavelength below their emission. They also have narrow emission peaks, making detection simpler because there is less overlap. The researchers used quantum dots with a cadmium-selenide semiconductor core. For biological applications, the quantum dot cores are encased in a zinc-sulfide shell coated with organic polymers. In building their 17-color cytometry system, the researchers used quantum dots with emission peaks of 525, 545, 565, 585, 605, 655, 705 and 800 nm. They combined these with nine standard fluorochromes, using diode lasers at 408, 488 and 532 nm, along with a helium-neon laser at 635 nm for excitation. As described in the August issue of Nature Medicine, they steered the lasers individually into the cytometry chamber with a series of mirrors. The lasers emitted in a timed sequence of 488, 408 or 635 nm and then 532 nm. For detection, they used a series of photomultiplier tubes and optical filters. The light entered an octagonal assembly containing the photomultiplier tubes. They directed the light inside using dichroic mirrors to the correct photomultiplier tube/filter pair and detected the emission. Betts noted that the quantum dots provided additional colors to those of the other fluorophores without complicating things. “The equipment to view these reagents is expensive but relatively straightforward — a violet laser and an octagon detector setup,” he said. Using this equipment, the researchers analyzed populations of T cells, which play a central role in the immune system. They characterized the cells from a single HIV-positive individual in terms of response to HIV, to cytomegalovirus and to Epstein-Barr virus. This in vivo image shows a single T cell in a mouse blood vessel (flowing from right to left), labeled with the fluorescent probe DiD and imaged with two consecutive pulses separated by 5 ms. The cell velocity is ~1.6 mm/s. In the T-cell population, they found almost every possible combination of negative, dim and bright expression for the various fluorescent markers, validating the need for multiparametric analysis. For example, they found Epstein-Barr-virus-specific T cells to express a particular sequence of dim and bright fluorescent marker emission, with cytomegalovirus-specific cells expressing a similar sequence for many of the markers. The difference came largely in two specific labels, which were mostly negative and bright for Epstein-Barr but often dim for cytomegalovirus. Even within HIV-specific T cells, the variation was great. Applications for the quantum dot cytometry include human cancers, stem cell research, vaccine studies, mouse and primate models, and the study of antigen-specific T cells and their differentiation. The latter, the researchers reported, might benefit the most from the technique. Because quantum dots contain heavy metals such as cadmium, their use in cytometers presents a potential environmental risk. The fluorophore disposal problem may have a simple solution, and it may not be an issue at all. Research team member Marcel P. Bruchez is an associate research professor and program manager for the technology center for networks and pathways at Carnegie Mellon. Previously, he was a founder of a company commercializing quantum dots. He noted that more research must be done. “We have not got a complete picture of what the safety of these quantum dots materials is,” he said. Going live The third cytometry innovation comes from researchers at the Wellman Center for Photomedicine, part of Boston-based Harvard Medical School. The group added fluorescence imaging to a cell-counting flow cytometer to create an in vivo imaging flow cytometer. The instrument can detect and capture information on fluorescently labeled cells circulating in a live animal, noted team leader Charles P. Lin. “Immediate applications are in the studies of circulating tumor cells and immune cells in mouse models of disease,” he said. Human applications are more long-term. Methods to identify particular cell populations in the bloodstream will have to be developed. One approach might be to use intrinsic optical cell signatures for identification, while another might be to use fluorescent probes approved for humans. In building the instrument, the researchers started with an in vivo flow cytometer that used a Melles Griot helium-neon laser operating at 632 nm for a counting beam and a Hamamatsu photomultiplier tube for cell detection. They focused the beam onto a selected blood vessel, exciting fluorescence in labeled cells. Optics captured that emission, which traveled through a confocal slit before reaching the photomultiplier tube. To this setup the researchers added a second HeNe laser, and an electron-multiplying CCD from Andor Technology. They offset the second beam from the first by about 25 μm at the target surface. Via a beamsplitter and other optics, they directed the image onto the CCD. They used the counting signal from the photomultiplier tube as a trigger for the second, imaging beam. When a cell passed the counting window, the imaging laser flashed, producing a fluorescent image captured by the CCD. To control the delay between trigger and strobe, they used an Andersen Laboratory acousto-optic modulator to gate the imaging beam. Lin noted that the technical requirements weren’t that challenging. “The temporal resolution, 10 to 100 μs, was not all that demanding — much faster imaging systems have been used by many groups, including our own,” he said. The modulator was versatile enough that the delay could be adjusted, allowing capture of multiple images of a cell. That ability meant that such information as cell velocity and even direction could potentially be extracted. The researchers demonstrated the instrument using a mouse model. To locate the blood vessel of interest, they placed a green LED behind an anesthetized animal’s ear and imaged that onto a CCD camera with a large field of view. They labeled isolated T cells with a Molecular Probes dye that binds to the cell membrane, with an excitation peak at 647 nm and an emission peak at 669 nm. After injecting the cells into test animals, they observed the arteries of five mice, with an average diameter of 20 μm. An analysis of 119 images showed that 98 percent of signal peaks from the counting photomultiplier tube were due to single cells, with the rest doublets where two cells were imaged together. The actual fluorescence intensity varied some thirtyfold between cells, a difference that led to variability in peak photomultiplier tube intensity. Another contributing factor to intensity variation, the researchers found, was the axial position of the cell when it passed the counting window. The work is described in the Aug. 21 issue of Optics Express. As for uses of the instrument, Lin noted that his group is doing studies on the kinetics of T cells and cancer cells in circulation. It also is working to enhance the in vivo imaging flow cytometer. “We are improving multichannel detection capabilities and detection sensitivity, or cell-counting efficiency,” he said. The goal, he added, is to detect a single circulating cell, which may not be possible with the current technique. However, it may detect as few as 100 cells, an important threshold for research.