Skip to main content
Premium Trial:

Request an Annual Quote

New Single-Cell RNA-seq Method Enables Transcriptomic Profiling of Live Cells


BALTIMORE – A team of collaborators from the Swiss Federal Institute of Technology (EPFL) in Lausanne and ETH Zurich has developed a new single-cell RNA sequencing (scRNA-seq) technology that can profile the transcriptome of a cell while preserving its viability.

Described in a proof-of-concept study in Nature last month, the method, aptly named Live-seq, combines a technology called fluidic force microscopy (FluidFM), which can extract cytoplasmic content from a live cell, with a highly sensitive RNA-seq method that can generate complementary DNA from a trace amount of mRNA. By avoiding destruction of the cell, Live-seq offers a way to track transcriptomic dynamics over time throughout the cell cycle, potentially helping researchers to decipher biological questions such as cell heterogeneity. 

"The beauty of [Live-seq] is that the cell stays alive, so you can really follow its fate," said Julia Vorholt, a faculty researcher at ETH and one of the corresponding authors of the paper. Live-seq is made possible by FluidFM, she explained, a technology that was invented by another group at ETH over a decade ago. Combining atomic force microscope with nanofluidics, FluidFM made it possible to inject or extract molecules from a cell while keeping it alive.

"When I heard about this [technology], I really thought this is cool for biology … it [evokes] a lot of imagination of what you can do with this," said Vorholt. "Since the beginning, we have been interacting with engineers to develop the technologies such that it becomes useful in biology."

In 2016, Vorholt's team published a landmark study in Cell that demonstrated FluidFM could draw out precise amounts of intracellular fluid from live cells, monitoring the process with sub-picoliter resolution. The group also found a detectable amount of mRNA in the extracted cellular content, raising the possibility that FluidFM could be harnessed for transcriptomic profiling of a live cell. That prospect, Vorholt said, sparked her collaboration with Bart Deplancke, a bioengineering professor at EPFL who is experienced in single-cell method development and served as the other lead investigator of the study.

Given that researchers had to deal with a minuscule amount of mRNA from picoliter-scale cytoplasmic samples, the development of Live-seq entailed a multitude of technical hurdles. "The challenge was how we could actually preserve this mRNA as well as possible and then actually put it into a library preparation workflow that is sufficiently sensitive," said Deplancke.

To achieve this, the researchers designed ways to minimize RNA degradation and sample loss during the FluidFM extraction process. These included reducing the extraction time, preloading the FluidFM probe with a sampling buffer that contains RNase inhibitors to protect the extracted cytoplasmic fluid, and releasing the extracted content into a droplet containing reaction buffer that is compatible with the downstream RNA-seq method.

Meanwhile, the team also had to come up with a highly sensitive scRNA-seq protocol in order to generate sufficient cDNA from the low mRNA input. For that, the group turned to a method called Smart-seq2, a well-received scRNA-seq method developed by a group of researchers at the Ludwig Institute for Cancer Research in Sweden.

Despite the high sensitivity of Smart-seq2, the method still could not effectively meet the demand for Live-seq, Deplancke said. Therefore, the researchers set out to systematically assess and optimize each step of the Smart-seq2 workflow.

"We functionalized our primers, for example, because what we actually noticed is that since we were dealing with so little mRNA, basically the primers themselves amplified," he said, adding that, along with many other improvements, the group also tested a whole range of different mRNA reverse transcriptases, in search for the most sensitive one for Live-seq.

After a series of optimizations, the group eventually came up with a scRNA-seq method that was deemed to be "sufficiently sensitive," Deplancke said, generating "reasonable amounts of cDNA" out of the scant starting material.

To benchmark Live-seq and demonstrate that the method can accurately profile different cell types while preserving their integrity, the researchers tested the method on live cells and investigated their growth, functional responses, and whole-cell transcriptome readouts. 

In brief, the results showed that Live-seq is capable of distinguishing cell types and states, while it appeared to preserve the normal functions of the cell. The authors said they arrived at this conclusion after observing normal cell cycles, concordant lipopolysaccharide (LPS)-induced responses, and preserved adipogenic differentiation capacity in cells that underwent Live-seq when compared to control cells. 

In one analysis, the researchers compared the transcriptional profiles obtained by Live-seq with those from conventional scRNA-seq and performed unbiased clustering on the cells. "We saw, basically, that we could not distinguish the probed cells from the unprobed cells," Deplancke said. "This was a very surprising result because it suggests that the cells have not really developed the system to defend themselves against something being extracted from them."

However, he cautioned that the result came only from one cell type, which is derived from mice, and whether it is the case in other cell types remains to be further investigated.

In terms of applications, by tracking transcriptomic changes of a cell over time, Live-seq can hopefully help researchers uncover the molecular mechanisms behind a wide array of biological issues pertaining to cellular differentiation and heterogeneity, he said.

Additionally, Vorholt said that because Live-seq employs an optical microscope to spatially observe, pick out, and even trace the desired cell for analysis, it can be tapped as a single-cell spatial analysis tool.

"I think it's a very creative strategy to measure the temporal dynamics of gene expression in the cell," said Junyue Cao, head of the single-cell genomics and population dynamics laboratory at Rockefeller University.

According to Cao, broadly speaking, researchers have previously come up with two strategies to overcome the temporal constraints of conventional scRNA-seq. One approach, he said, is to computationally infer the cellular transcriptional trajectory based on the transcriptome similarity of the sequenced cells. However, he said, this method only allows researchers to extrapolate whole-transcriptome dynamics over pseudo time instead of real time points.

To address such limitation, researchers have also come up with ways to chemically label newly synthesized mRNA inside a cell with a molecular tag to help distinguish it from existing molecules. This strategy, which includes a scRNA-seq method that Cao helped to develop, enables researchers to obtain the transcriptome profile of the cell at two different time points.

However, Cao said that regardless of the strategy used, these scRNA-seq methods generally require killing the cells in order to extract the RNA for sequencing. By preserving the viability of the cells, Live-seq further pushed the boundary of scRNA-seq analysis over time, Cao noted, enabling scientists to sample the transcriptome of a cell across multiple time points.

Additionally, he thinks that the temporal transcriptomic data generated by Live-seq could potentially be helpful for developing statistical models, which can help augment the analysis of in vivo cell stage changes using conventional end-point-based scRNA-seq techniques.

Despite Live-seq's promises, Vorholt said that parts of the workflow, especially RNA extraction using FluidFM, can still be labor-intensive at this point. "So far, we have done this manually for our study; it was certainly a lot of work," she said, adding that it normally takes an hour to extract the mRNA from four cells. As such, there is potential in automating the extraction. Cytosurge, a Swiss biotech company that has commercialized FluidFM, is working on automating the technology, she added, which hopefully can make it more widely accessible.

Besides operability, Deplancke said there is also a need to continue improving Live-seq's sensitivity. "We have to be honest; we're still at a 40 percent, 50 percent success rate," he pointed out. "We still have to improve further the overall sensitivity of the method in order to make the efficiency a bit higher."

While the study did not specify the per-sample cost for Live-seq, Deplancke said that after the upfront investments for the equipment, the cost per cell for Live-seq should be "pretty marginal." However, he said that labor costs can be a significant factor for the technology.

The researchers have not applied for a patent for Live-seq, but perhaps with future improvements of the method, "there's definitely room for IP there," Deplancke said.

Moving forward, Vorholt said she is interested in applying Live-seq to examine the spatial aspects of cellular infections and learn how cells react to and sense infections. Additionally, her team is working to extend the method toward cells with a cell wall, such as those from fungi, which are much harder to manipulate.

The lab already has some "promising results" showing that the probes can be injected into fungal cells, she said, and the next step is to see whether the researchers can also extract cytoplasm and eventually perform Live-seq on these cells.

Meanwhile, for Deplancke, whose lab has been studying fat cell differentiation, future goals include going back to the adipogenesis model and trying to identify potential molecular factors that can explain the differentiation among fat cells.

"Wouldn't it be cool if you could actually probe a cell and know which genes are expressed in that cell and then let that cell go and see whether it differentiates or not?" he said.