SeParate: Multiway fluorescence-activated droplet sorting based on integration of serial and parallel triaging concepts

Fluorescence-activated droplet sorting (FADS) has emerged as a versatile high-throughput sorting tool that is, unlike most fluorescence-activated cell sorting (FACS) platforms, capable of sorting droplet-compartmentalized cells, cell secretions, entire enzymatic reactions and more. Recently, multiplex FADS platforms have been developed for the sorting of multi-fluorophore populations towards different outlets in addition to the standard, more commonly used, 2-way FADS platform. These multiplex FADS platforms consist of either multiple 2-way junctions one after the other (i.e. serial sorters) or of one junction sorting droplets in more than 2 outlets (i.e. parallel sorters). In this work, we present SeParate, a novel platform based on integrating s̲e̲rial and p̲a̲r̲allel sorting principles for accura̲t̲e̲ multiplex droplet sorting that is able to mitigate limitations of current multiplex sorters. We show the SeParate platform and its capability in highly accurate 4-way sorting of a multi-fluorophore population into four subpopulations with the potential to expand to more. More specifically, the SeParate platform was thoroughly validated using mixed populations of fluorescent beads and picoinjected droplets, yielding sorting accuracies up to 100% and 99.9%, respectively. Finally, transfected HEK-293T cells were sorted employing two different optical setups, resulting in an accuracy up to 99.5%. SeParate's high accuracy for a diverse set of samples, including highly variable biological specimens, together with its scalability beyond the demonstrated 4-way sorting, warrants a broad applicability for multi-fluorophore studies in life sciences, environmental sciences and others.

The fluorescence intensities of every imaged droplet retrieved from their respective outlet.Dot colors represent the channel from which a droplet was retrieved at sorting threshold of 4 V (a) and 2.5 V (b).The dotted lines correspond to the fluorescence intensity value below which 99% of the droplets from the green and waste channel (for the horizontal line) or the red and waste channel (for the vertical line) are located.They indicate the effect the changing sorting threshold has on the sorting accuracy.(c) Sorting accuracy for all 4 channels for the 3 tested thresholds (the data for 1 V was already presented in Figure 5).Different letters above the bar indicate significant differences between the tested thresholds (α < 0.05).For every channel and condition, at least 2185 droplets were analyzed (Table S2-3).Error bars represent one standard error of the mean (n = 3).

Section S3. Sample variation in fluorescence intensity
Within one population (i.e. the whole library of droplets, existing of both fluorescent and nonfluorescent droplets), in the presented experiments, four subpopulations can be distinguished: (1) a green fluorescent subpopulation, (2) a red fluorescent subpopulation, (3) a green + red fluorescent subpopulation and an empty (non-fluorescent) subpopulation.In Figure S7a, a sample of the fluorescent bead population passing the interrogation zone over time is shown, illustrating that every bead type had its own, different, fluorescence fingerprint.However, within one subpopulation, beads showed a similar fingerprint, which simplified the determination of an optimal threshold, minimizing wrong sorting events.Figures S7b and c illustrate this for respectively picoinjected droplets and encapsulated cells.Compared to the low variation in fluorescence intensity between beads of the same type, picoinjected droplets and fluorescent cells showed higher intra-subpopulation variation in fluorescence intensity, potentially complicating accurate thresholding and sorting.These observations Figure S6.Green and red fluorescent widefield images of sorted picoinjected droplets for the three tested sorting thresholds (1, 2.5 and 4 V), retrieved from the green channel.Lowering the threshold to 1 V resulted in less green + red droplets in the green channel, increasing the sorting accuracy.Additionally, these images illustrate that an increase in threshold results in the lack of detection (and sorting) of low intense fluorescent droplets as for 4 V the droplets generally show a higher fluorescence intensity while for 1 V a mix between low and high fluorescence intensity droplets is observed.Scale bar = 100 µm.
are confirmed by Figure S8 as the average peak height and variation on this peak height are plotted, showing that the intra-subpopulation variation increases from beads to picoinjected droplets to cells.
Next to the cell sample showing the highest intra-subpopulation variation in fluorescence intensity, they pose an additional challenge as their fluorescence intensity is much lower compared to the other tested samples.This is reflected in the fact that a 40X short working distance was needed (instead of a more standard 20X) to capture fluorescent signals.While increasing the sensitivity, the background of the PMT measurements increased as well, which is clear from Figure S7c., letting all light above its value through to PMT 2. An eGFP tail is still observed above 590 nm, which leads to bleed-through when using the 590 nm LP, which is picked up when using a sorting threshold close to the background.Using the 638 nm LP configuration, together with the eGFP tail, the mCherry intensity peak is filtered out, leading to potential loss in sensitivity (i.e.low intense cells will not be detected).Table S1.The counts and sorting accuracy of sorted beads for three sorting repetitions on independent chips and the weighted average of these three repetitions.Table S2.The counts and sorting accuracy of sorted picoinjected droplets for a threshold of 4 V and 2.5 V for three independent chips.Next to the counts, the weighted average of these three repetitions is shown.Table S3.The counts and sorting accuracy of sorted picoinjected droplets for a threshold of 1 V for three independent chips.Next to the counts, the weighted average of these three repetitions is shown.The starting population was the same as depicted in Table S2.Table S4.The counts and sorting accuracy of sorted cells for three sorting repetitions on independent chips and the weighted average of these three repetitions for the optical setups using 638 nm LP.Table S5.The counts and sorting accuracy of sorted cells for three sorting repetitions on independent chips and the weighted average of these three repetitions for the optical setups using 590 nm LP.The starting population was the same as depicted in Table S4.

Figure S2 .
Figure S2.Picoinjection design with the most important dimensions in µm.

Figure S3 .
Figure S3.Microfluidic sorting design with the most important dimensions indicated in µm.

Figure S5 .
Figure S5.Sorting performance of the picoinjected droplet population for different thresholds.(a-b) The fluorescence intensities of every imaged droplet retrieved from their respective outlet.Dot colors represent the channel from which a droplet was retrieved at sorting threshold of 4 V (a) and 2.5 V (b).The dotted lines correspond to the fluorescence intensity value below which 99% of the droplets from the green and waste channel (for the horizontal line) or the red and waste channel (for the vertical line) are located.They indicate the effect the changing sorting threshold has on the sorting accuracy.(c)Sorting accuracy for all 4 channels for the 3 tested thresholds (the data for 1 V was already presented in Figure5).Different letters above the bar indicate significant differences between the tested thresholds (α < 0.05).For every channel and condition, at least 2185 droplets were analyzed (TableS2-3).Error bars represent one standard error of the mean (n = 3).

Figure S7 .
Figure S7.Recorded output signal from PMT 1 and PMT 2, depicted in green and red respectively, during the sorting process.Three peak types are visible: peaks in PMT 1 (i.e.green fluorescence), peaks in PMT 2 (i.e.red fluorescence) and in both PMTs (i.e.green + red fluorescence), indicated with respectively green, red and yellow arrows.Dotted lines represent the sorting threshold used for sorting.(a) The encapsulated mixed bead population, recorded using the 20X objective and 638 nm LP.(b) Picoinjected droplets, recorded with the 20X objective and 638 nm LP.Multiple thresholds are depicted as this population was sorted using 3 different thresholds.(c) The encapsulated cell population, recorded using the 40X objective and 590 nm LP.

Figure S10 .
FigureS10.The emission spectra of eGFP and mCherry, the two fluorescent labels used in the cell sorting experiments.The dotted lines indicate the emission filter (for both optical setups used, 590 nm LP and 638 nm LP) positioned in front of PMT 2, letting all light above its value through to PMT 2. An eGFP tail is still observed above 590 nm, which leads to bleed-through when using the 590 nm LP, which is picked up when using a sorting threshold close to the background.Using the 638 nm LP configuration, together with the eGFP tail, the mCherry intensity peak is filtered out, leading to potential loss in sensitivity (i.e.low intense cells will not be detected).

Figure S9 .
Figure S9.The relative frequency of green, green + red and red fluorescent cells, imaged prior to encapsulation.

Figure S8 .
Figure S8.Average peak height as detected by the PMTs.(a)The peak heights for green and green + red droplets, as detected in PMT 1 (i.e.green fluorescent signal).For the green droplets, CV values were 0.3%, 22% and 21% for respectively beads, picoinjected droplets and cells, while the green + red droplets showed CV values of 12%, 25% and 50% for respectively beads, picoinjected droplets and cells.For green droplets, the signal in PMT 1 saturated.(b) Peak heights for red and green + red droplets detected by PMT 2 (i.e.red fluorescent signal).For the red droplets, CV values were 12%, 30% and 42% for respectively beads, picoinjected droplets and cells, while the green + red droplets showed CV values of 12%, 22% and 36% for respectively beads, picoinjected droplets and cells.Error bars indicate one standard deviation, n > 8.

Figure S11 .
Figure S11.Recorded output of PMT 1 and 2 for green fluorescent beads in the two tested optical configurations.Every peak represents a green fluorescent bead passing by the interrogation zone.(a) Due to the more strict filtering by the 638 nm LP filter, green fluorescent beads only show in PMT 1 and not in PMT 2. (b) A zoom on one of the peaks, showing no bleedthrough (i.e.no peak for PMT 2).(c) Using the 590 nm LP filter, green fluorescent beads also show small peaks in PMT 2, which could lead to wrongful identification during sorting.(d) A zoom on one of the peaks, clearly showing a peak in PMT 2 due to bleed-through.