Sorting the Glow from the Flow

Published: 03-02-2020
The Jimenez Lab has built a fast flow cytometry system which quickly sorts fluorescing cells from non-fluorescing ones. Image Credit: JILA/Steven Burrows and the Jimenez Lab

How do you find a single cell in sea of thousands? You make it glow.

Adding fluorescence helps track movement and changes in small things like cells, DNA, and bacteria. In a library of millions of cells or bacteria, flow cytometry sorts the glowing material you want to study from the non-glowing material.

In short, “it’s a fluorescence filter,” said Srijit Mukherjee, a graduate student in the Jimenez Lab at JILA.

With the help of JILA’s electronics shop and clean room, the Jimenez Lab has found a way to take droplet sorting time from days to hours. Their new setup not only improves the time it takes, you can better sort your material by how long or how brightly it glows.

“You gain an enrichment off a population in a matter of a few hours,” Mukherjee explained. “Then you can repeat the process again and again to enrich this population of a very rare event.”

Drop by drop

Here’s how flow cytometry works: You have a large library of material—cells, for example—which have been genetically modified so the cells with the traits you want to study glow. You encapsulate a group of those cells (and the medium they’re floating in) into individual droplets of water in oil. The droplets flow through a tube past a focused laser beam. When a glowing group is detected, it is separated out with an electric field which “pushes” it into the “keep” pile.

There are two obstacles scientists run into with flow cytometry systems. First, there’s a lot of junk floating around with the glowing material you want to study. The odds of getting any fluorescent cells at all in your droplet are low, Mukherjee pointed out—at single cell loading, fewer than 10% of the droplets have a cell, glowing or not. The other 90% are just oil and water.

“Even in that 10% the probability that you have a fluorescent droplet is even lower, so your throughput is really, really low,” Mukherjee explained.

Second, flow cytometry can be really tedious. That low throughput means it can take a long time to sort through with a large library of material, even with good flow cytometry systems.

The Jimenez Lab wanted to sort fluorescing E. coli bacteria. For their experiment, they needed to not only sort out the glowing bacteria, but sort by the lifetime of that fluorescence. The flow cytometry system they were using could only sort 50 cells a second; sorting through millions of bacteria would have taken days.

Plus, the system was complicated to use.

“To most of us, it was just a black box…it was just a cobweb of Labview codes,” Mukherjee said. “Getting it to sort was a challenge.”

Pumping up the drops

The group took a mathematical approach, Mukherjee said: if you increase the number of cells in each droplet, the probability that a droplet contains a fluorescing cell increases too.

“It's basically dumping out most of the non-fluorescing junk and selecting out the fluorescing population,” Mukherjee said.

Then, they repeat the flow cytometry process with the traditional single-cell per droplet approach—but this time, they sort out the material by more specific characteristics, such as a fluorescence lifetime or brightness.

The power of collaboration

To do that, they needed faster electronics and clean, precise tools. Those were all available in house at JILA, and the Jimenez Lab built their fast flow cytometry system completely at JILA.

JILA’s electronics shop was able to craft field-programmable gate array (FPGA) electronics which operate on a nanosecond scale—much faster than what they could order elsewhere. The clean room at JILA was used to fabricate all the microfluidic chips, so they were super clean. Being able to make everything in house also made this new system extremely cost-effective, Mukherjee added.

As a result, the Jimenez Lab enriched the proportion of fluorescing cells in their samples from 10% to 94%. They went from sorting 50 cells per second to about 2500 droplets per second—greater than a hundredfold improvement, Mukherjee said.

This type of system could make a difference not only to labs, but to anyone who needs to sort through a large library for a particular event, such as biomedical researchers who need to find the few abnormal cells in a pool of millions.

“We are trying to use it to approach fluorescent protein libraries but this is a very general approach to enrich any fluorescent event in a library of events.”

This study was published in Royal Society of Chemistry’s Lab on a Chip on February 21, 2020, and was supported by the NSF Physics Frontier Center Grant, and the NIH/CU Molecular Biophysics Training Program.


Ralph Jimenez

JILA’s NSF Physics Frontier Center follows University of Colorado polices for ensuring harassment-free environments.  For more detailed information please read the whole Discrimination and Harassment Policy and Procedures.