Fluorescence image processing

We performed all DNA imaging with an upright microscope (AZ100, Nikon, Tokyo, Japan) equipped with a 5× objective (NA 0.5, AZ Plan Fluor, Nikon, Tokyo, Japan) set to 3× optical zoom, an LED light source (Sola, Lumencor, Beaverton, OR) and a CMOS camera (Zyla, Andor, Belfast, UK). We used a Cy3 filter (TRITC-B-NTE, Semrock, Rochester, NY) for observing labelled DNA (focusing experiments), a Cy5 filter (49006, Chroma, Bellows Falls, VT) for detecting molecular beacons (hybridization experiments).

In the LVF chips, we chose a detection point 1.4 cm downstream from the end of the converging region. For the DNA focusing experiments, we calculated the final (focused) concentration from a calibration curve constructed by filling the channel with known concentrations of DNA solution. The total amount of focused sample could then be found by spatially integrating over the resulting concentration map and multiplying by the depth of the channel. In the integration, we used a threshold of 10% of the peak value. We confirmed that no cross-contamination occurred between runs by comparing the fluorescence signature after ITP focusing in a new device in the absence of DNA to that in a device previously used to focus a high concentration of DNA. For the hybridization experiments we calculated the signal by integrating over a fixed 100 × 200 μm² region in the channel.

We imaged the bacteria on an inverted microscope (Eclipse Ti-U, Nikon, Tokyo, Japan) equipped with a 10× objective (NA 0.45, WD = 4 mm, AZ Plan Apo, Nikon), an Intensilight C-HGFI light source (Nikon, Tokyo, Japan) and a Neo sCMOS camera (Andor, Belfast, UK). We used a FITC filter (49011, Chroma, Bellows Falls, VT) for SYTO9 (bacteria focusing). In the LVF chips, we chose a detection point 1.4 cm downstream from the end of the converging region. We manually counted the bacteria over 5 frames to account for any bacteria moving vertically into and out of the focal plane.