Dr. Xin Jin was a key player in the development of in vivo Perturb-seq (published here in 2020) during her research fellowship at Harvard. She recently became a junior PI and just started her own laboratory at the Scripps Research Institute in San Diego, California.
Perturb-seq, first described in 2016, is an innovative and high throughput technique that employs CRISPR to introduce gene perturbations in cells and measure resulting changes in transcriptomes at the single-cell level. In 2020, Jin et al. further developed this technique for use in vivo. As mentioned in their paper published in Science, “Due to the labor-intensive and time-consuming nature of generating and analyzing individual knockout animal models for functional investigation, it is crucial to develop phenotyping methods that are scalable, general-purpose, high-resolution, and high-content, to identify tissue- and cell-type specific effects of genetic perturbations in vivo.”
In the study, Jin et al. used in vivo Perturb-seq to evaluate the effects of perturbations in a panel of Autism Spectrum Disorders (ASD) and Neurodevelopmental Delay (ND) risk genes on transcriptome changes at the single-cell level in the brain during development.
Dr. Xin Jin is recruiting for various positions in her lab! Contact her at firstname.lastname@example.org for more information.
How the idea for in vivo Perturb-seq was born
“I did my PhD in worms and genetics is very powerful in this research model. I started thinking about whether it was possible to apply this systematic type of genetic screening in the mammalian system. I’m also a neuroscientist by training, and I was fascinated by how neurons are born and form mature networks in a living tissue.”
Other factors were at play around the same time, which contributed to the technical aspects of the technique. “We were really excited about the CRISPR-Cas9 gene editing technique once it came out and I was wondering whether it could be done in vivo, but introducing multiple mutations can be very complicated without an appropriate readout,” Jin continues.” Then, first versions of droplet-based sequencing (InDrop and DropSeq) became available when I was finishing my graduate training. I was interested in seeing whether I could combine all these pieces together, conceptually that was my goal when I started as a Junior Fellow at the Harvard Society of Fellows.”
Collaborations were crucial in helping orchestrate the link between developing and optimizing the technique that will become Perturb-seq, determining which genes to perturb, and choosing an appropriate readout method.
“This work was done in collaboration with teams of scientists who each brought a crucial element to the table: Aviv Regev (genomics), Paola Arlotta (developmental neurobiology), and Feng Zhang (CRISPR and neurobiology). They had extensive experiences with the different techniques needed and this facilitated the work. Genome-wide association studies and whole-exome sequencing studies in ASD and ND were picking up through multi-institute collaborations including the Broad around this time too. We had access to a list of de novo variant genes – thanks to generous researchers – that were affected in these disorders.”
Once the pieces were in place, the team hit the ground running and a few years later, the paper describing in vivo Perturb-seq was published, in which 35 of these risk genes were perturbed in the brain during development (in utero in mice) and subsequent effects on single-cell transcriptomes and cell populations were evaluated and subsequently validated.
“Single-cell technology is revolutionary because it is allowing us to appreciate the fundamental unit of biology, a cell, and understand how it operates and behaves. There’s even more excitement when it becomes scalable; we can talk about assaying half a million cells, which is an engineering miracle!
Surprises along the way
“One of my humbling moments along the way was when I realized how naïve I was to believe that we would see a neat and obvious effect of the gene perturbation on one cell type, and that would be it! If you see the data, the first thing you realize is so many things are profoundly impacted when risk genes are acting.”
“Technological advancements can keep strengthening our ability to uncover biological questions, but we still need the intellectual space and ability to decode complicated data and translate it into meaningful information.”
“Once we started the experimental work on in vivo Perturb-seq, it just made me realize how complex the system is and that there is so much more to learn. It’s also a justification for why we’re trying to build a better, cleaner, high-throughput technology. The goal is to try to do this in a systematic way and be as unbiased as possible.”
The technical stages of in vivo Perturb-seq
“We started this project around 2017 with a pilot study in which we did all the experimental steps, but with only two or three genes. We saw that the different technical parameters and steps were working (CRISPR, cell dissociation, cell sorting, cell viability, multiplexing). Ironically, in the paper, this entire optimization and troubleshooting period is summarized in one little diagram of cells -> analysis.”
“The experiments themselves were intense and long. We made sure we had snacks and coffee before starting because once we did, we couldn’t pause until absolutely everything was done since a half-hour break would mean neurons dying. It was also a huge team effort and I was grateful to have the support of the Regev and Zhang labs. The different steps of the experiments were all done in the same building, which was a great help – it can get very complicated if the logistics aren’t in place. These may seem like details, but the devil is always in the details.”
“After the experiments were done, we spent probably a solid year and a half just trying to make sense of the data, working closely with a brilliant computational biologist, Sean Simmons. Designing and conducting experiments are very important, but as the data starts to get more and more sophisticated, analysis is really a crucial part.”
The future of in vivo Perturb-seq is bright
Dr. Jin is now heading her own lab and their current focus is to further develop in vivo Perturb-seq, with exciting prospects both on the technology side and on the biology side.
“In the original paper, we only looked at one developmental timepoint, but now we’re also starting to look at later timepoints that are equivalent to the adolescent period to be able to distinguish the transient phenotypes from those that persist. We’re also looking at more genes [the original study involved a panel of 35 risk genes].”
“Our readout was cells in the cortex, not only because they’re implicated in ASD, but also because it’s a region we know really well and for a pilot setup we wanted to have high confidence in the cell types. Since there are many single-cell transcriptome atlases of the cortex, it was easier to determine effects of perturbations on those cell populations. Now we’re starting to look at other relevant brain regions.”
“We are also working on improving and developing the technology on the perturbation side and on the readout side. In the initial study, the perturbation was an insertion/deletion but now we’re working on more subtle changes in non-coding regions. Our goal is to modify gene expression without necessarily interfering with the genome. With the readout, we’re working on introducing multi-omics approaches to in vivo Perturb-seq.”
Using other single-cell technologies as a read-out for genetic perturbations
“For now, we’re focusing on spatial transcriptomics which can allow us to map out not only cellular and transcriptomic changes but also where these perturbed cells end up in the brain. Spatial technology will also allow us to evaluate secondary effects on nearby cells too, which provides us with a fuller understanding of the context.”
Jin further explains why this is important and how it may act as a limitation when relying exclusively on single-cell RNA-seq to determine the effects of a gene perturbation. “We always hear the question, ’Gene expression changes but what does this actually mean?’ Accordingly, single-cell spatial technology can help us detect whether, for example, the cell ends up migrating to another layer or exhibits altered interactions with other cells. We can’t answer this question by dissociating cells and looking at them independently.”
Tips for other researchers
Dr. Jin also has a wealth of advice for other researchers who are interested in diving into single-cell research.
“It’s very important to have a solid, well-controlled experimental plan; this includes the analysis part. As biologists, sometimes we have this approach of doing a quick and dirty experiment just to see if it works. I learned a lot from computational people, because they don’t start a project if they haven’t thought about it from start to finish and they’re much more stringent in their criteria.”
The importance of this mix between experiments and analysis can be seen in Dr. Jin’s new lab, where a core value is cross-talk between biologists and bioinformaticians. “I don’t want anything being 100% wet lab or 100% dry lab; everyone has to be at least at 80%/20%. Understanding the rationale, questions, and ways of approaching problems on both sides makes everyone better at what they do.”
Everyone can do single-cell data analysis to some extent! “Many giants in the field have developed single-cell data analysis packages that are easy to install and then apply to a demo dataset using a very well-documented tutorial. I think there’s really no excuse for not being able to have some familiarity with how single-cell analysis works, and I feel very strongly about this interdisciplinary approach!”
With respect to using in vivo Perturb-seq, Dr. Jin emphasizes important points about the overall rationale that researchers should use when applying it to their research projects. “It’s important to plan things out in a way that makes sense for the specific context that someone is investigating. The two essential questions in this technology are: What kind of perturbation do you need and what is the most relevant read-out? For example, depending on the disorder or disease being studied, it might be relevant to either have only one gene perturbation per cell or many co-occurring perturbations. For the readout, epigenetic assays – like single-cell chromatin accessibility (e.g. scATAC) – may be more relevant for some than transcriptomic changes.”
Take home message about single-cell research in biology
“Single-cell technology is revolutionary because it is allowing us to appreciate the fundamental unit of biology, a cell, and understand how it operates and behaves. There’s even more excitement when it becomes scalable; we can talk about assaying half a million cells, which is an engineering miracle! The technique has proved its value already by helping us build single-cell atlases of different tissues, but hopefully this is just the beginning, as it can help us answer so many biological questions.”