Chip fabrication and MACS properties

MACS chips were produced via soft lithography using PDMS. The base (part A) and the curing agent (part B) of a two-part silicone elastomer kit (Slygard 184, Dow Corning) were mixed at particular ratios (part A:part B) in weight to produce PDMS. The master mould for the flow channel was produced by spin-coating positive photoresist (PR) AZ10xt (AZ Electronic Materials) to a height of 10 μm on a silicon wafer. After ultra violet patterning the PR using a transparency mask (Output city), the wafer was heated for rounding the features to achieve dome-shaped channels. After rounding, the channel height becomes 8 μm. The wafer was then baked on a hotplate overnight to stabilize the positive PR. The control layer master was made by spin-coating the negative PR SU-8 2025 (MicroChem) to yield a height of 25 μm, and ultra violet patterning it using a transparency mask defining the channels. To produce the soft MACS chip, 20:1 PDMS was spin-coated on the flow channel master at 1,250 r.p.m. for 45 s to yield an ∼65-μm-thick PDMS membrane. For this condition, the minimum pressure required for closing the valve is ∼5 p.s.i. For even gentler handling of cells, if required, a thinner membrane can be made to achieve valve closing at lower pressures. For control channels, PDMS with a 5:1 ratio was poured onto the control layer master. After both masters were partially cured at 65 °C for 33 min, they were aligned and cured for another 6 h at 65 °C to achieve thermal bonding. Finally, the two-layer PDMS chip was plasma-bonded permanently against glass coverslips. Since the freshly bonded chips did not work due to altered surface properties following plasma treatment, they were kept at room temperature for at least 1 day to regain the native surface properties. For single-molecule counting experiments, the chips were kept at the 65 °C for a total of 3 days after cover glass bonding since ‘cytoplasmic slowing-down' works better with stiffer PDMS.

Closing properties of the valve depend on multiple parameters⁴⁸. To achieve the half-open state for the 200-μm-wide control and flow channels, we typically used ∼20 p.s.i. both for P_valve and P_flow, though different combinations of P_valve and P_flow also work. Compared with the full footprint of the valve (200 μm × 200 μm), cell trapping happens within a subregion (approximately, 100 μm × 50 μm), which can be varied by slight modifications of P_valve and P_flow. The number of cells captured per field-of-view (FOV) also depends on the relative values of P_valve and P_flow, as well as durations of the valve states. Since Quake valves can be actuated millions of times without signs of fatigue²², the bottleneck for long-term stability of MACS is the accumulation of debris within the valving intersection. This is problematic only during actuation of the valve and the presence of sample flow: debris does not get stuck permanently unless pressed against the surface during valve actuation and eventually gets washed away otherwise. Therefore, the intersections that remain passive do not collect debris and a neighbouring intersection can be used on demand if the actively used intersection becomes clogged. To minimize debris, we filtered all buffers and media using 0.22-μm-pore-size filters (Corning). Cells were grown in plastic tubes (BD Falcon, round-bottom) instead of glass vials to prevent crumbled glass. Sonicating the PDMS chips in isopropanol for 30 min, followed by 4 h of drying at 65 °C before bonding them to the cover glass removes PDMS crumbs that form at the inlets during hole punching⁴⁹. In addition, prior to using the chips, flow channels were extensively rinsed with PBSA buffer (1 × PBS with 4 mg ml⁻¹ BSA) to wash away debris that was stuck on the walls of the chip, as well as for passivating the channel surfaces to minimize cell sticking. We were able to keep the same chip on the microscope and use it for multiple days, everyday using a fresh flow channel.