Chips were manufactured in PDMS cast from SU-8 masters as described in subsection 2.1.3. Multiple revisions of the chip manufacturing process were required to address issues during sample loading. The assembly of the PDMS was optimised to reduce bursting of the chips during loading (see subsection 2.1.3). Hydrophilisation of PDMS surfaces is also required to permit fluid flow and reduce the occurrence of bubbles within the system.
Despite the improved manufacturing process, there is some variability in the chip assembly process, which can lead to differing distances from the inlet to the trench itself. This reduces the predicted control over the system somewhat, but can be countered by experience using LiaT with a range of distances from the inlet. The velocity of the cells appears to vary with the distance of the inlet to the trench, and this may be due to the cells not having time to reach an equilibrium state in the fluid flow.
3.3.1 The move to partial curing
In initial experiments, the PDMS had been fully cured and plasma treated before bonding the lid and channel segments together (as described in sub- section 2.1.4). The bond between the surfaces was of variable quality and the upper and lower parts of the chips regularly separated on loading.
Partial curing of the PDMS before assembly eliminated bursting of chips. The PDMS was more difficult to manipulate, and more prone to collecting debris during assembly, but these issues were offset by the improved loading. Partial curing gives a high bond strength between the layers without resort- ing to adhesive (Eddings et al., 2008). Punching inlet holes in the partially cured PDMS also gave cleaner inlet walls and made them less inclined to
leak around the reservoirs. The fully cured PDMS tended to crumble on punching, and could leave fractures around the inlet that leaked on loading. In addition, the chips were assembled in groups of 3 to 6 trenches, rather than attempting to align the entire disc as was initially performed. This improves alignment, and is also much easier to manipulate, especially when working with the partially cured polymer.
3.3.2 Hydrophilisation of channel surfaces
The hydrophobic nature of the PDMS surface made loading with aqueous solutions challenging. When loading the plasma bonded chips, loading was easiest when the chips had been bonded on the same day as the assay was to be performed. However, small air bubbles often accumulated and expanded as the surface wetting was not always complete.
The chips assembled from partially cured PDMS were particularly dif- ficult to fill if they were not treated for hydrophilisation. The use of pro- teinaceous buffers, such as 5 % BSA in TBS, improved loading but was not always desirable as it can interfere with upstream applications that require low protein background.
Plasma treatment of the channels in the completed chip was impossi- ble using a corona plasma probe (Electro-Technic Products, Chicago), as it could not reach internal surfaces. The probe tip could be inserted into the inlet hole and plasma treat the channels in a localised fashion, however the plasma treatment did not extend for enough of the channel to provide ad- equate hydrophillicity, and in addition, prolonged plasma treatment caused damage to the surfaces of the chip.
It was decided to trial plasma treatment in a plasma cleaning cham- ber (Harrick Plasma, Ithaca) under reduced air pressure. The resulting hydrophilisation is successful but short lived, requiring that the chips are loaded immediately after removal from the plasma chamber (see subsec- tion 2.2.1).
3.3.3 Lab in a Trench in use
The manufacturability of the platform was improved by assembling the chips using partially cured PDMS instead of fully cured. Furthermore, loading was improved through the use of plasma hydrophilisation of the channel
surfaces. Loss of liquid from the reservoir and through evaporation through the PDMS was reduced by humidifying the microscope chamber.
Loading the chips with cell media containing 10 % serum also improved loading. In addition, it was noted that the cells were less inclined to stick to the channel surfaces, possibly due to blocking of the surface by the proteins. For glycosylation studies however, it is preferred to minimise the amount of protein loaded into the system as it may contain glycoproteins that could give a background signal during imaging.
Managing the flow rate is challenging using the current system. As flow rate is regulated by the height in the channel, it is necessary to adjust the rate by adding or removing microlitre volumes of liquid. This can be tricky when working with the small volumes that are required for the slow flow rates during cell capture. The liquid meniscus is often below the height of the PDMS, making it hard to see. Moreover, as the reservoir tips are not always at the exact depth in the PDMS, finding the right liquid height can often be a case of trial and error.
Sterility of the platform was not considered during testing, as the plat- form is used as an open system, with the potential for bacteria and yeasts to fall from the air and the operator into the reservoir. It is possible to autoclave PDMS chips, but it is uncertain if it would be necessary, as the present system requires plasma treatment of the device surfaces. Precau- tions to reduce contamination that were taken include, wearing gloves and clean labcoats, and using antibiotics in the media to ensure bacteriostatic conditions.
Optimisation of the system manufacture and loading conditions has im- proved the robustness of the device for use in the laboratories, although some further refinements are still required.
Chapter 4
Sequential glycoprofiling of
single cells using Lab in a
Trench
The Lab in a Trench (LiaT) platform was used to investigate cell surface gly- cosylation on Ramos B-lymphocytes. It was demonstrated that it is possible to label cells with lectins and to elute the lectin with free sugar, allowing, for the first time, sequential glycoprofiling of the cell surface (O’Connell et al., 2014). The method for sequentially glycoprofiling live cells was further ex- panded for other uses including investigation of fixation on lectin binding.