Microfluidic devices provide powerful, miniaturised platforms for the isolation and analysis of cells. Combined with other capabilities like optics, electronics and hydrodynamics, these systems can automate cellular isolation as well as multi-variable cellular analysis for applications ranging from drug discovery to molecular diagnostics. Not only can subpopulations of cells be isolated from mixed cell samples, but individual cells and their surrounding media can also be controlled independently for analysis in parallel. The two main focus areas covered in this article are isolation of cellular sub-populations analogous to conventional cell sorting, and the isolation of single (or a few) cells into isolated microenvironments, which performs a similar function to laser microdissection. In both the cases, we highlight strategies and techniques using microfluidic platforms, and examine their strengths and limitations.
by Dr G. J. Shah
Smaller, smarter tools for cell biology
Over the last couple of decades, microfluidics has emerged as an exciting technology, aiming to transform biological research, in much the same way as microelectronics transformed the field of information technology. Just as room-sized mainframe computers were replaced by microelectronic chips, so also are large pathological and biological labs envisioned to be compressed onto microfluidic “labs-on-chip” to achieve greater flexibility and automation, enabling complex studies of cells, proteins and nucleic acids. Potential areas of interest include drug discovery, molecular
diagnostics, cell metabolism, stem cell differentiation, proliferation studies etc.
The defining characteristic of a microfluidic system is the ability to manipulate tiny volumes of fluids (several nanolitres or less) in microchannels having sub-millimetre dimensions. Besides many other advantages of going micro (e.g. portability, less reagent consumption, lower power), the small feature sizes of microfluidic devices facilitate much more precise control over the spatial distribution of not just the fluids but also the species (e.g. cells) within them. Both these capabilities, or better still, their combination, can be utilised to isolate the species of interest from a mixed sample. The scale of microfluidic device features closely matches typical cell dimensions, making these devices particularly well-suited for cell-based assays.
Some of the functions that microfluidic devices perform are simply miniaturisations of large-scale cellular analyses such as flow cytometry and cell sorting. Others enable studies that would be very challenging with traditional techniques, such as single-cell studies in well-defined and well-controlled microenvironments. Moving from flasks and dishes to these “smarter vessels” (viz. microfluidic chips) enables the automation of fluidic and particle manipulation for such isolation and subsequent analysis.
In the next two sections this article considers, in turn, the strategies used by automated microfluidic systems for isolation of cellular subpopulations from mixed samples, and for isolation of cells into discrete microenvironments.
Cell isolation strategies in microfluidics
Conventional laboratory isolation procedures routinely separate sub-populations of cells of a particular type from mixed cell samples, perhaps the most popular technique being fluorescence-activated cell sorting (FACS). Several microfluidics attempts have focused on miniaturising FACS for point-of-care applications, and they achieve sorting rates and a range of parameters comparable to FACS. In this section, a few recent developments are discussed, with comments on some of their important features.
A direct adaptation of FACS into microfluidics would use sheath flow to hydrodynamically focus the sample, and laser-induced fluorescence to interrogate the cells. Perroud et al developed a µFACS system and sorted fluorescently-labelled, infected macrophage cells from non-infected cells [Figure 1(a)], [1]. Optical tweezers, which utilise the radiation pressure force from a high-intensity coherent laser beam to manipulate particles, were used to laterally deflect cells into the collection channel.
Analogous to FACS, cells could be separated using magnetophoretic forces (MACS) or electric forces.
The latter most commonly employs a dielectrophoretic (DEP) force (hence DACS), which varies according to the difference between the polarisability of target cells from other cells and the medium in an electric field gradient. Subpopulations of cells, such as circulating tumour cells, fractionated by a combination of dielectrophoretic, sedimentation and hydrodynamic forces, can be separated and collected [Figure 1(b)], [2].
Along the lines of multi-fluorophore cell sorting, more than one manipulation technique can be used on a single microfluidic platform in order to isolate multiple target subpopulations. For instance, in an integrated magnetic and dielectrophoretic cell sorting (“iDMACS”) system, genetically modified E. coli cells having different surface peptides are labelled with either DEP tags or magnetic tags and serially isolated in the respective separation modules [Figure 1(c)], [3]. The purely electromagnetic input required by DACS and MACS eliminates the optical setup that often tends to be bulky and expensive, and helps to make the entire system much smaller and cheaper. However, the purity of separation using DEP fractionation is usually lower. The use of dielectric or magnetic tags would enhance the isolation purity, but requires the added step of labelling (and perhaps de-labelling).
Some microfluidic systems provide added functionality compared to existing technologies, such as the ability to isolate cells based on temporal observation of their behaviour. For example, unlike conventional cell sorting where cells are sorted based on a single time-point, (viz. when they cross the interrogation area), the image-based optofluidic platform developed by Kovac et al. is capable of observing cells in an array of wells for extended periods of time. Upon locating cells of interest, they can be lifted out of the well using an infrared laser to be dragged downstream with the flow [Figure 1(d)], [4]. An additional capability of this platform is its ability to distinguish between whole-cell fluorescence and localised fluorescence within the cell.
Isolated cell analysis in microfluidics
The amazing complexity of the “cellular machine” makes it far more intricate than arguably any human-made machine. The heterogeneity of cells, even within a population of similar cells, entails that the bulk response of cells is often very different from the individual cell response. Experiments that use bulk measurements obtain averaged-out responses, which could yield misleading information and overlook important phenomena or phenotypical differences that single-cell analysis might reveal. Moreover, not only is cell behaviour influenced by a wide gamut of environmental stimuli, but also by the cell-cell interactions from neighbouring cells. Isolating cells helps separate the effects of environmental factors from cell-cell interactions, leading to greater insight into cell behaviour. A platform that can perform an array of isolated cell analyses, where each cell is subjected to independent microenvironments and/or treatments, would enhance multi-parameter single-cell studies as well as insight into inter-cellular interactions.
Non-microfluidic techniques for individual cell isolation, such as laser microdissection and micromanipulation, often require skilful and extensive manual involvement, as well as specialised equipment, proprietary films etc. On the other hand, automated single-cell isolation in microfluidic devices has been achieved through several approaches, each with their unique capabilities and limitations. The rest of this section highlights representative strategies from among these.
Cells can be trapped individually using, for instance, an array of physical structures, and can then be subjected to uniform environmental conditions to perform parallel cell-studies, but applying different conditions to different cells is challenging. More explicit cellular isolation can be achieved by enclosing the cells within discrete microenvironments, such that the surrounding media of neighbouring cells are separated from each other. Using a device made from multiple layers of an elastomer that helps provide on-chip pneumatic valves, small microfluidic chambers with valves on either end are used to isolate cells along with the surrounding liquid. Kamei et al. isolated human embryonic stem cell colonies as shown in Figure 2(a) [5]. Large-scale integration of several fluidic components and flexibility of layout for chambers and valves makes this an increasingly popular and attractive platform for microfluidics, especially in cell studies, where harsh chemicals or high temperatures are not encountered. Integrating a particle manipulation technique that can selectively pick out specific cells could make this platform even more powerful for isolated cell analysis.
An alternative microfluidic platform to flow in channels of a stationary device is to use the centrifugal force on a rotating disc much like a compact disc (CD), to move the fluids. This force could also be used to push cells into traps in order to isolate them. Lee et al. trapped single cells in “pitfalls” along the edge of a channel using centrifugal force. The concept was demonstrated by performing an apoptosis assay on HEK 293 cells [Figure 2(b)], [6]. CD-like microfluidic systems are advantageous in that they do not need pumps or pressure sources for fluidic actuation, but are limited in their controllability of individual cells, since almost all fluidic actuation and cell manipulation on the entire device are controlled by the same rotation mechanism.
Cells could be individually encapsulated by flowing two immiscible (oil and aqueous) fluids in cross-channel flow-geometry to form droplets of one (usually aqueous) phase while injecting cells into it. Edd et al. achieved automated, highly monodispersive single-cell encapsulation into 10-100 pL drops, by appropriately adjusting the rate of cell-sample flow and oil flow (for droplet formation) in a high-aspect ratio channel [Figure 2(c)], [7]. Very high throughput of encapsulated single cells (up to 103 – 104 cells/s) can be produced with this technique. Selective isolation of specific cells, for instance by optical manipulation, can also be integrated with such encapsulation, but it is usually at the cost of throughput. Optical tweezers have been used to trap and manipulate specific lymphocytes to the edge of the aqueous-oil interface before the aqueous droplet is split [9]. Encapsulating cells in different microenvironments is, however, an extreme challenge for this platform.
Lastly, a microfluidic platform that synergistically combines two technologies to create a highly powerful tool for individual cell studies is discussed. The first technology, namely optoelectronic tweezers (OET), produces an optically-modulated DEP force for particle manipulation. By illuminating a photoconductive layer with an image feedback-based dynamic optical pattern to create virtual electrodes for DEP manipulation, the technology can achieve automated, individual cell manipulation over a large field of view while using an optical power orders of magnitude lower than optical tweezers. The second technology is an electrically-controlled droplet manipulation technique, namely Electrowetting-on-dielectric (EWOD). Droplet manipulations like creating, splitting, merging and transporting can be performed on-the-fly by purely electrical input without the need for external pumps, pressure sources, valves etc. Optically-controlled manipulation of HeLa cells using lateral-field OET (LOET) followed by electrically-controlled droplet movement by EWOD, were demonstrated on the integrated EWOD-LOET device [Figure 2(d)], [8]. Cells could be observed over time, and based on differences in optical (or dielectric) properties, specific cells could be picked out by OET and isolated into separate droplets by EWOD. Both EWOD and LOET can be automated to simultaneously manipulate multiple droplets and cells respectively. One limitation for this platform is the relatively
large droplet size (typically 0.1-1 µL).
Conclusions and future directions
Several exciting microfluidic platforms for automated cellular isolation and analysis have been developed. Some research has focussed on replicating large-scale functions such as cell sorting, but with less space, time and cost and without sacrificing the capabilities. For instance, the use of non-optical sorting methods such as electrical or magnetic techniques [2,3] could help eliminate the large optical setup. On the other hand, some microfluidic cell sorters [4] offer additional features such as longer-term interrogation before sorting. Other efforts have been made to encapsulate individual (or few) cells into isolated microenvironments. Such isolation is hard to automate at the macro-scale, but microfluidic devices, with their smaller feature sizes, are better suited for this purpose. A variety of microfluidic platforms have been used for isolated cell studies, each with their strengths and weaknesses. The trade-off between precise positional control and high throughput does remain a challenge. The emergence of droplet microfluidics, and the synergistic combination of techniques for multi-variable, individual cell-studies in parallel show great promise. An automated platform capable of investigating several independently isolated cells, with the ability to address specific cells (e.g. to move to a different microenvironment or apply a stimulus) seems to be one of the “holy grails” of current research.
Given the abundance of exciting microfluidic technologies for automated cellular studies, though, it is somewhat surprising that very few if any of them have been translated into commercial products so far. One possible cause is the fact that many of these devices are still fabricated and operated in the prototype mode. Adapting them to large-scale manufacturing and easy-to-use commercial products is another key focus area of research and development.
Acknowledgements
The author would like to thank Prof. Yong Chen, the Nano Research Group and the California NanoSystems Institute (CNSI) at University of California, Los Angeles (UCLA), and the Bill and Melinda Gates Foundation for their support, as well as Prof. Chang-Jin “CJ” Kim and the Micro/Nano-manufacturing Lab at UCLA. Many thanks also to Prof Dino Di Carlo for his valuable suggestions and comments during the
manuscript preparation.
References
1. Perroud TD et al. Analytical Chemistry 2008; 80: 6365-6372.
2. Gascoyne PRC et al. Electrophoresis 2009; 30: 1388-1398.
3. Kim U, Soh HT Lab on a Chip 2009; 9: 2313-2318.
4. Kovac JR, Voldman J. Analytical Chemistry 2007; 79: 9321-9330.
5. Kamei K-i et al. Lab on a Chip 2009; 9: 555-563.
6. Lee S-W et al. Sensors and Actuators A 2008; 143: 64-69.
7. Edd JF et al. Lab on a Chip 2008; 8:1262-1264.
8. Shah GJ et al. Lab on a Chip 2009; 9: 1732 - 1739.
9. He M et al. Analytical Chemistry 2005; 77: 1539-1544.
The author
Dr G. J. Shah
University of California Los Angeles
USA
e-mail gjshah@ucla.edu
