Innovation in Single-Cell Analysis

Sarah Boswell, PhD, director of the recently established Single-Cell Sequencing Core at Harvard Medical School, talks to contributing editor Tanuja Koppal, PhD, about her efforts to bring inDrops single-cell RNA sequencing (scRNA-seq) to the greater Boston scientific community. She talks about the different types of scRNA-seq technologies that are currently available, her experiences working with and helping investigators understand the best practices for sequencing, and how sequencing can be put to use when addressing certain biological questions.

Q: Can you provide some background on why and how the single-cell RNA sequencing core lab came to be established?

A: Our core lab started in July 2016. Allon Klein’s lab in the Department of Systems Biology at Harvard Medical School is part of the team that developed the inDrops technology for scRNA-seq, and he had too many people interested in collaborating with him to use this technology. So, I was brought on as someone with a lot of sequencing expertise and experience to get the core lab started. Currently in the core we use only inDrops technology, which is similar to other microfluidics-based techniques for scRNA-seq like Drop-Seq and the one by 10x Genomics. inDrops is a microfluidics-based technology where you make a suspension of single cells that are captured and barcoded in nanoliter droplets on a chip. Each droplet has a hydrogel carrying photocleavable combinatorially barcoded primers. Reverse transcription takes place within each droplet to generate single-cell mRNA transcripts labeled with random cell barcodes as well as unique molecular indices to identify each transcript from a cell. The transcripts from thousands of individual cells are then pooled to prepare the RNA sequencing library, which is analyzed using nextgeneration sequencing. After we do the library prep, we hand the samples and all library information back to the user so they can get the sequencing and bioinformatics done at any center they choose. 1CellBio is currently commercializing the inDrops technology.

Q: How do the different technologies for scRNA-seq compare against each other?

A: There are lot of comparative studies going on right now with these different methodologies. We are looking at some of that data as well, along with our collaborators, to see how these different microfluidicsbased platforms compare and perform. The microfluidics-based scRNA-seq instruments from commercial vendors like 10x Genomics are certainly easier to use, but they are also more expensive than inDrops. When inDrops is used optimally, the prices at our core lab can drop to around 7 cents per cell, while other methods can be as high as 25 cents, and that is even before you get to the sequencing. On average, our users are looking at about 3,000 cells per sample. Even though the inDrops technology is not that expensive, when you look at thousands of cells it can still get costly.

Besides the microfluidics droplet-based technology, people use plate-based methods where single cells are usually flow-sorted into 96 or 384 well plates containing lysis buffer, and then RNA sequencing libraries are prepared using protocols like Smart-seq2. These methods can give you the full-length transcript information, but they are also more expensive per cell. With plate-based technology, it’s difficult to collect and store thousands of cells; that is much easier to do with droplet-based methods. Using a flow sorter can sometimes also lead to a higher dropout rate. However, microfluidics-based single-cell technologies require thousands of cells in order to work well and give good data. If you have a rare cell population with a few hundred cells, you probably should consider a plate-based method. Hence, we sometimes recommend that people bring us more cells for analysis by using some broader criteria for cell sorting, if possible.

Q: What are some of the applications that people are using scRNA-seq for?

A: When the sample is limiting, such as with single cells, you can’t get as much information out as you would in a traditional RNA sequencing experiment. The more RNA you can extract from the sample and input into your library preparation, the more information you can get about the whole transcriptome at the back end. With single cells you get information about only a small percentage of the transcriptome, and this can vary further by cell type, the sample prep method, and the scRNA-seq technology used. That said, we have used the inDrops technology on a lot of different cell types, from the fly and sea worm to induced pluripotent stem (iPS)- derived cells and patient samples.

Lots of people are using single-cell technologies for looking at heterogeneity in immune cells and immune populations in response to different conditions or treatments. In patient samples, this may have to do with looking at various disease states. We have people looking at different organisms to see how their cells function. We have many users working on iPS cells and organoids, hoping to understand how these models compare when looking at disease pathways and the effects of various treatments in healthy donor or patient cells. Hence, single cells are being used to answer a lot of different types of biological questions.

Q: What are some of the challenges that you have run into with scRNA-seq?

A: The real challenge is for the user to bring us the sample that is ready to go on the instrument. Dissociating the sample into a single-cell suspension, keeping the cells viable, and not altering the transcriptome during the sample processing can be very tricky. Many questions, such as what does a three-hour dissociation process do to the transcriptome, still need to be answered. Some cells are definitely more challenging than others because many of them die during processing, and that creates a bias among the population of cells that survive. We are very much aware of these challenges and make sure our users understand them, too. We cannot do the sample prep for every user, as it would be hard to optimize protocols for each sample and would add to the cost of analysis. We also don’t have many papers published with our technology yet, so we try to connect our users so they can exchange protocols for sample prep. Can different labs using different sample prep protocols get the same results? A lot of this has yet to be determined.

With inDrops, we can work on only one sample at a time. So, we must coordinate well and make sure the samples are not sitting on ice for too long before being analyzed. Each sample is loaded into the microfluidic chamber, where it undergoes gentle lysis in the hydrogel and reverse transcription takes place with the barcoded primer. Each single cell, still encapsulated within a droplet, is frozen as a DNA:RNA hybrid that can be used for library prep along with all other samples at a later date. This helps eliminate some of the batch effects by prepping all the user’s libraries on the same day, even if the samples were collected on different days. We run about six to eight samples a day, depending on the sample and how many cells we are collecting. Some instruments, such as the one from 10x Genomics, have parallel channels on the chip so you can run many samples at the same time, although it may not always be feasible for the user to prep and hand over all the samples at once.

Q: Do you see some of these challenges being overcome soon?

A: There is a lot that can be done to move the field forward. The long sample prep required to make single-cell suspensions on some samples can lead to nonuniform processing, and many cells die in the process. Many sample types cannot be frozen if you want to look at cellular heterogeneity, and there is work being done on cryopreservation for single-cell analysis. So far, that seems to work well only on certain robust cell types. Similarly, people are working on encapsulation and freezing for high-throughput, multistep analysis. Sample coordination, especially when it comes to patient samples, can be tricky. These cells must be drawn from the patient, stored, and processed properly and quickly. We have now managed to optimize this protocol so we can accommodate more patient samples, but if freezing cells were possible it would make things much easier.

Many labs are working on optimizing technology so you can capture more genes per cell. If we can develop inDrops to accommodate multiple samples, that will be a big step forward. A lot of innovation is also being done in the data analysis because the single-cell data is quite different from the standard RNA sequencing data. It’s far sparser, and we need the right replicates and well-controlled experiments to understand what we see in the cell population data. Before jumping on the single-cell bandwagon, we need to do good science and be sure to follow up on results with secondary methods. However, as far as looking at patient data and understanding the heterogeneity in cell response to various drugs and treatments, the single-cell technology can certainly prove to be very powerful.