Microfluidic devices typically have at least one dimension in the sub-100 µm range. This allows close control over the system as the liquid flow is predictably laminar. In addition, the small size of microfluidic systems allows scale down of reagents used in experiments, greatly reducing costs. Devices can be made optically transparent with structures to mix, measure, and meter liquids as well as trap particles such as cells. Microfluidic devices generally have scope for parallelisation for multiplexing and high throughput analyses.
Liquid handling for microfluidics requires control over the flow rate through passive or active pumping, and direction of the fluid through chan- nel geometries and valving. Pumping can be carried out directly using sy- ringe pumps, by relying on centrifugal force or by relying on the liquid height in a reservoir to form a pressure head. Valving can present a chal- lenge, whether to use passive single use valves or active valves that can be actuated again and again, although some devices for single cell analysis use valving many use unvalved systems (Roman et al., 2006; Huh et al., 2005). Further considerations include the materials for manufacturing the device, the fluid itself and how it interacts with the cells and the device surfaces, how long the assay will be carried out, and how the cells will be studied.
For single cell analysis, the geometries used to capture the cells are of particular interest. Some geometries allow detailed probing of large numbers of cells while other systems are limited in cell number or in how they can be assayed. A variety of geometries are discussed here, with Lab in a Trench specifically expanded on in section 3.1.
Rowat et al. (2009) presented a device that allows for the trapping of single cells in a linear channel designed to monitor single cells as they divide and multiply. The device allows for medium term culture and monitoring of cells, and later staining of the cells for analysis.
An elegant design to capture two different cell types for fusion studies was devised by Skelley et al. (2009) based on flowing cell into cup shaped traps. By capturing the cells in a shallow trap, then reversing the flow, the
cells are transferred into a deeper trap, ensuring a single cell per deep trap. The secondary cells are then flowed into the system to fall into the deep traps directly and allowed to fuse with the cells for production of hybridomas.
Another cup based method for single cell capture is the centrifugal V-cup array presented by Burger et al. (2015, 2012). The advantages of centrifugal microfluidics is that all fluidic flow is controlled by the rotation of the disc, and therefore requires no intervention beyond changing spin speeds. The cups are oriented axially, so the settling force on the cells is due to the centrifugal force of the spinning disc. As the capture chamber is primed with fluid before cell capture, shear forces on the cells can be better controlled. The cups are scaled to fit single cells, to reduce the capture of multiple cells in the cups. A further addition of optical tweezers (Burger et al., 2015) increases the flexibility of this device by allowing selective manipulation of cells, and their transfer to a side chamber for potential downstream analysis. Lin et al. (2013) present a system based on sieve traps with side pillars to maximise trapping of cells in the sieves. Not only are the cells trapped in the sieve, but one side of the device is micropatterned with adhesive patches for the cells to adhere to and grow. As the spaces between the micropatterning are made of non-fouling materials, stray cells cannot adhere outside of the defined micropatterns. The microfluidic traps can be removed, leaving the cells growing on their designated patches. This system allows for tracking of cells that are grown on specifically defined adherent patches.
Microwell traps for single cells come in a variety of formats. Kobel et al. (2010) trap cells in microwell divots in the sidewall of long channels, where the cells are retained by the pressure differential between the fluid flow on either side of the divot. A limitation with Kobel’s device is that the area required to capture a single cell means that it is not possible to have many cells located under the microscope objective at the same time. Carlo et al. (2006) captured HeLa cells in an array of microwells, and observed their growth in relation to the shear stress on the system, noting that when the cells divided, the daughter cells also remained trapped in the wells. Cao et al. (Cao et al., 2015) use microarrays of wells for single cell trapping of cells, where the cells flowed into the device and allowed to settle into the microwells. The trapped cells can be further probed, for example for their extracellular glycans. Park et al. (2010) employ settling to fill their microarrays also, but their system is open, so the cells are allowed to settle
from a droplet on top. The open array allows probing of the cell supernatent with an antibody-conjugated membrane, which can then be used to identify cells to pick out and subculture.
Pillar arrays have been used to capture cells with antibodies (Nagrath et al., 2007; Vickers et al., 2011) and with lectins (Vickers et al., 2011). The cells are forces through convoluted paths and interact with the pillars, binding where the cell has the appropriate ligand for the capture molecule. As lectins have poor affinity, the lectin capture pillars were demonstrated to be not as effective as the antibody pillars.
Using an acoustic field with a wavelength on the same order as the cell size, Collins et al. (2015) were able to distribute single cells across a plane, with the ability to keep cells in position while media is exchanged or the cells are washed. The single cell array can thus be stained and analysed, and the device is reusable, which is unusual for many lab on a chip devices. Droplet microfluidics involves encapsulating cells or reagents in aqueous droplets and transporting them in an oil-based carrier liquid. Single cells can be trapped in the droplets and merged with droplets containing reagents to study them. Droplet microfluidics enables relatively easy sorting of the droplets for recovery afterwards. A summary of the state of droplet microflu- idics by Guo et al. (2012) covers a number of high throughput applications, including drug screening, antibody generation and virally infecting cells.
These various microfluidic devices for single cell analysis have advan- tages and disadvantages. Recovery of cells of interest is not practical from the acoustic set-up or linear traps, and can be challenging from cup based systems. Droplet microfluidics allows cell recovery, but as a population of droplets. Most microfluidic set ups are based on light and epi-fluorescent microscopy, which for some geometries limits the high throughput capabil- ities that are often portrayed as a benefit of microfluidics. Microwells and cup arrays generally allow multiple cells in the field of view of microscopes at low to moderate magnifications. Microwell arrays and some cup-based arrays are well suited to multiply probing cells with a variety of probes, whereas droplet fluidics only permit the summed addition of reagents.
Microfluidic devices are often designed as disposable, made from rela- tively cheap materials. PDMS is a common material for making microflu- idic devices in an academic setting as it is easy to use for rapid prototyping purposes and for casting devices with high aspect ratios. PDMS is how-
ever, impractical for scaled-up production, not ideal for cell culture, and hydrophobic (Berthier et al., 2012). In some cases, it may be possible to manufacture the devices by hot embossing or injection moulding thermo- plastics such as PMMA, techniques which allow for scale up but can be beyond the scope of the academic laboratory. Other devices that are made from materials such as SU-8 on silicon are generally reusable, as the cost of manufacture is quite high.
All devices also require consideration of sterility for cell work. Some materials are suitable for autoclaving, such as PDMS, other sterilisation options include running 70 % alcohol through the system for a time followed by rinsing with sterile buffer (Lu et al., 2004), gamma irradiation, or low- pressure plasma treatment (Meyvantsson & Beebe, 2008). Open systems require the use of antibiotics, and, where possible, carrying out the work in a sterile environment such as a biological safety cabinet.
Another consideration is the shear force on the captured cells. It is known that shear forces can influence the physiology of cells (Meyvantsson & Beebe, 2008; Christophis et al., 2010). For some microfluidic geometries, such as droplet microfluidics, the cells can be maintained in shear free en- vironments, whereas other geometries such as shallow cups and microwells, there is a shear force over the top of the cells at all times there is flow in the system. Depending on the nature of the cell and the magnitude of the shear force, shear should be taken into account when observing cells in microfluidic devices.