Ultra-High-Throughput Absorbance-Activated Droplet Sorting for Enzyme Screening at Kilohertz Frequencies

Abstract

Droplet microfluidics is a valuable method to "beat the odds" in high throughput screening campaigns such as directed evolution, where valuable hits are infrequent and large library sizes are required. Absorbance-based sorting expands the range of enzyme families that can be subjected to droplet screening by expanding possible assays beyond fluorescence detection. However, absorbance-activated droplet sorting (AADS) is currently ∼10-fold slower than typical fluorescence-activated droplet sorting (FADS), meaning that, in comparison, a larger portion of sequence space is inaccessible due to throughput constraints. Here we improve AADS to reach kHz sorting speeds in an order of magnitude increase over previous designs, with close-to-ideal sorting accuracy. This is achieved by a combination of (i) the use of refractive index matching oil that improves signal quality by removal of side scattering (increasing the sensitivity of absorbance measurements); (ii) a sorting algorithm capable of sorting at this increased frequency with an Arduino Due; and (iii) a chip design that transmits product detection better into sorting decisions without false positives, namely a single-layered inlet to space droplets further apart and injections of "bias oil" providing a fluidic barrier preventing droplets from entering the incorrect sorting channel. The updated ultra-high-throughput absorbance-activated droplet sorter increases the effective sensitivity of absorbance measurements through better signal quality at a speed that matches the more established fluorescence-activated sorting devices.

Figure 1

(A) Schematic of the UHT-AADS showing the single 50 μm layer channel (light blue) and a close-up design of the sorting junction showing the gapped divider at the bifurcation (light blue). The gapped divider is on the second 50 μm layer: (1) input channel for droplets, (2) input channel for spacing oil, (3) input channel for bias oil, (4) positive channel outlet, (5) negative channel outlet, (6). The ground electrode is colored black and the positive electrode is colored red. Adding a single-layer droplet injection chamber allows for even spacing of the 75 pL droplets. The bias oil channel acts as a barrier to droplets from entering the negative chamber unless acted on by the dielectrophoretic force. (B) A diagram showing a typical droplet trace as it passes the optical fibers. At position A, light transmission is at the maximum value indicated by the baseline value set. At B, the droplet edge causes refraction at the water–oil interface producing a “shoulder” corresponding to the higher amount of light collected by the fiber. At C, the droplet is moving toward the center of the optical fibers. At D, the droplet is in the center of the optic fiber, and the true peak value of absorbance (minimum amount of light) is given in the droplet trace. At E, voltage increases as the droplet moves away from the optical fiber center. At F, there is the effect of the other droplet edge causing another peak (or a “shoulder”) of collected light.

Figure 2

(A) Effect of absorbance (455 nm) against different percentages of RI-matching oil for different concentrations of the absorbing compound. Blue points are 0 mM tartrazine, orange are 0.5 mM tartrazine, and green are 5 mM tartrazine. Different tartrazine concentrations were used to assess the dynamic range. As the RI-matching oil percentage increases, the absorbance also increases. The RI-matching oil percentage is the percentage volume of the additive added to the HFE-7500 oil. Each point corresponds to data from 20 s of droplet voltage recordings at 1000 Hz. Error bars are the standard deviation of the 20000 droplet peaks extracted from the data. (B–E) 6 ms droplet detection signal [V] for 0 mM tartrazine (water) droplets using (B) 25% RI-matching oil, (C) 27.5% RI-matching oil, (D) 30% RI-matching oil, and (E) 35% RI-matching oil. Droplet peaks are clearly distinguished from the baseline, and droplet edges have been removed, with a clear true peak at the center of each droplet trace. See section S4 (SI) for analogous droplet traces converted into absorbance values.

Figure 3

Effect of RI-matching oil on the droplet traces (V) at different droplet sizes with two droplet populations, 0.5 mM tartrazine (peaks at higher voltage) and 5 mM tartrazine (peaks at lower voltage) at approximately 1 kHz frequency of droplet measurement. The negative control is without 1,3-bis(trifluoromethyl)-5-bromobenzene and shows significant droplet edges for all droplet volumes. No droplet edges are seen when using 35% RI-matching oil, and peaks are clearly distinguishable. Droplets with a volume of 75 pL are distinguishable at 0.5 mM. The yellow asterisks show droplet peaks identified using a custom peak detection algorithm (SI). The algorithm cannot distinguish the droplet values for the 150 pL negative control trace due to broad peaks leading to an unidentifiable maximum.

Figure 4

(A) Histogram showing the fraction of correctly sorted droplets, partial false negatives, false negatives, partial false positives, and false positives at 1, 1.5, and 2 kHz. Partial false negatives are droplets that are false negatives that split at the junction. Partial false positives are false positives that split at the junction. A high-speed camera was triggered at every sorting event, and visual inspection in ImageJ was used to determine if the droplet moved into the correct outlet. Droplets were counted as correctly sorted only if the droplet that triggered the event was sorted alone. The events examined were ∼100. (B) Snapshots of droplets at 1000 Hz when the sorting electrode is triggered based upon the correct absorbance value. The droplet that is correctly sorted is shown with the white arrows. A video of this experiment is available as Supporting Information. (C) An example of a droplet fragmenting at 2000 Hz due to collision with the central barrier. Arrows show the droplet splitting into two sub-volumes. Scale bars are 50 μM.

Figure 5

(A) Illustration of a screening workflow, showing E. coli expression of a library, droplet generation for single cell encapsulation of library members, incubation off-chip and sorting for active phenylalanine dehydrogenase (PheDH) activity using the novel UHTS-AADS device. (B) The coupled reaction produces reduced WST-1 with absorbance at 455 nm. (C) Swarm plot of the mean absorbance at 340 nm for 38 colonies (blue, n = 3 for each) picked after transformation of DNA collected from the positive outlet of the AADS device sorting at 1 kHz. The red dot shows the mean (n = 9) of the positive control (pASK expressing wild-type PheDH), and the green dot shows the mean (n = 9) of the negative control (pASK expressing glycosidase). The confidently positive clones are classified as two standard deviations from the mean of the positive control and respectively for the negative control (dotted lines).