Mass-Activated Droplet Sorting for the Selection of Lysine-Producing Escherichia coli

Abstract

Synthetic biology relies on engineering cells to have desirable properties, such as the production of select chemicals. A bottleneck in engineering methods is often the need to screen and sort variant libraries for potential activity. Droplet microfluidics is a method for high-throughput sample preparation and analysis which has the potential to improve the engineering of cells, but a limitation has been the reliance on fluorescent analysis. Here, we show the ability to select cell variants grown in 20 nL droplets at 0.5 samples/s using mass-activated droplet sorting (MADS), a method for selecting droplets based on the signal intensity measured by electrospray ionization mass spectrometry (ESI-MS). Escherichia coli variants producing lysine were used to evaluate the applicability of MADS for synthetic biology. E. coli were shown to be effectively grown in droplets, and the lysine produced by these cells was detectable using ESI-MS. Sorting of lysine-producing cells based on the MS signal was shown, yielding 96-98% purity for high-producing variants in the selected pool. Using this technique, cells were recovered after screening, enabling downstream validation via phenotyping. The presented method is translatable to whole-cell engineering for biocatalyst production.

Figure 1.

Workflow for E. coli cell screening with MADS. Droplets are generated with single E. coli cells then cultured to create isogenic colonies. Droplets are split with one portion analyzed by ESI-MS, and the other is sorted using DEP based on the MS signal intensity. Selected droplets are added to fresh cell media for bulk culture, enabling validation by LC-MS/MS and phenotyping.

Figure 2.

(A) Droplet ESI-MS traces alternating cell media with media spiked with 250 μM lysine. The top trace shows the 10 μL/min sheath flow condition, and the bottom trace shows the 30 μL/min sheath flow condition. (B) Signal intensities of 250 μM lysine in media and blank cell media as a function of sheath flow rate. Mean signal for 70 droplets with ± 1 SD for the error bar. (C) Calibration curve using droplets. Droplets were generated with increasing lysine in cell media. Each point represents mean signal from 30 droplets with ± 1 SD for error bar. A line of best fit (R² >0.99) is included. The dashed line represents the background signal when no droplet is introduced.

Figure 3.

(A) Initial and (B) 48 hours of in droplet cell growth. Panels show images using a 10x and 20x objective. Scale bars represent 200 μm, and 100 μm for 10x and 20x objectives respectively. (C) An image of droplet reinjection and respacing for ESI-MS analysis after cell growth. (D) An example trace of droplet ESI-MS following cell growth. Lysine trace (black, 147 m/z) and neostigmine (blue, 223 m/z, offset) are shown.

Figure 4.

(A) Reformatted trace of droplet heights for 120 minutes of screening. Blue points represent mixture droplets and red points represent marker droplets. The sorting threshold is shown with the dashed line (B) Histogram of all analyzed droplet heights. (C) Example images of droplets in the sorted winner pool (top) and the waste pool (bottom). Fluorescent droplets were generated from the DapA E84T variant culture. Scale bars are 500 μm. (D) Concentrations of lysine determined by LC-MS/MS.

Figure 5.

(A) Droplet ESI-MS trace monitoring 147 m/z over ten minutes of screening. The sorting threshold is shown by the horizontal dashed line. (B) Droplet signal heights for 100 min of screening, with the sorting threshold shown. Droplets were collected for validation during three time blocks, shown by the vertical black lines. The final collection was taken at the end of the experiment. (C) Percentage of colonies formed from liquid cultures. Pre-sort pools represent liquid cultures generated by adding merged droplets to fresh media. Selected pools represent droplets collected from sorting at three time windows.