Mammalian cells have dominated the scRNA-seq field for years, but single-cell studies on plants are slowly but surely catching up. A research lab at the Vlamms Institute for Biotechnology (VIB, Ghent, Belgium) has recently published a review on the recent efforts to adapt scRNA-seq to plant cells. Dr. Carolin Seyfferth, first author of the review, discussed with us the state of single-cell plant transcriptomics today.
Breaking the cell wall
Scipio: “From your review, I can see that many of the steps for scRNA-seq for plants cells are similar to how you would treat mammalian cells. According to you, what is the main difference to be aware of when using plant cells and is there a bottleneck to achieve similar outcomes to mammalian models?
Carolin Seyfferth: “Unsurprisingly, the largest issue to overcome is the cell wall, which is there to protect the cell and makes it very difficult to dissociate them. And unfortunately, the majority of existing single-cell RNA sequencing technologies require to get rid of this wall.”
“You can try manually enriching the cells, but it is extremely labour-intensive and only make sense if you want to study a certain rare cell type population. If you want to look at a whole population, it is just too much work.”
“So the trick that we are using is to generate protoplasts, plant cells without cell walls. It can be done mechanically, but it also involves chemical steps or chemical degradation. This is the main bottleneck in single-cell research in plants for sure, because not all cells like to be protoplasts on their own. It is quite specific, depending on the cell type, on the tissue, or on the function that the cell has.”
“For us in general in plant science, I think the main first question with a new technology is ‘does it work with plant cells?’. The answer is almost 98% ‘no, we have never tested it in plant cells.’ “
S: “So you could end up with relative abundances of cell types biased according to their ability to be split from the cell wall, which is also an issue encountered in some mammalian tissues with extracellular matrices.”
CS: “Exactly. You will have to make sure that the data is representing the cell types that you are expecting, and there is a long-term process for actually validating the dataset. You can use histological slides or generate marker lines specific to your clusters of interest, but it is quite tedious.”
S: “Another issue with mechanical, chemical, and enzymatic dissociation is that it induces stress on mammalian cells, and stress-related gene clusters have been identified to skew gene expression profiles. What about plant cells?”
CS: “From our modern species, we know quite well what genes are responsive to the stress of dissociation. But even if you set up a new experiment, maybe not done on a typical model species where you have research information, for the majority of cell types you would be able to dissect what is stress-induced or what is coming from the protoplast, and what is actually coming from your data set of interest.”
Interested about scRNA-seq for prokaryotic organisms? Have a look at what Prof. Manuel Martinez-Garcia has to say about single-virus genomics!
Single-nuclei sequencing would be best in parallel with scRNA-seq
S: “A compromise that you can make with mammalian cells that are difficult to dissociate is to perform single-nuclei experiments, even if you only capture, of course, nuclear mRNA, which may not represent the actual relative abundances of mRNA transcript in the cytoplasm. Could we do the same for plant cells and bypass the cell wall altogether?”
CS: “We could, but I think the best way is to combine single-nuclear and single-cell RNA sequencing datasets. For single-cell RNA-seq, generating the protoplasts would most likely mean not capturing all of the cell types that we actually want. With single-nucleus RNA seq, you are capturing cell types that you are not capturing with scRNA-seq, but you also have much, much less information because of the lower content of RNA. So I think that the combination of both is where you get very powerful results.”
“In plant cells, you also have a lot of other organelles such as chloroplasts mediating the photosynthesis rate. So I think that the main issue in single-nuclei RNAseq is the purity of the nuclei. We have very good methods that are working really well but nevertheless, there are a lot of components inside the plant cell that are very similar in size to the nuclei. So you really need to be very careful in planning a very good strategy.”
Plant-specific features, such as varying ploidy levels, need to be taken into account
S: “Regarding the poly-A tail sequences, could we expect major differences from other eukaryotes that might affect the capture rate of mRNAs?”
CS: “When we compare our data sets to mammalian counterparts, the capture rate is quite similar. In general, I would say that there shouldn’t be a problem and we can, for example, use kits that have been designed for capturing polyadenylated mRNA, and they work nicely. But I am not sure if this is true for all plant species or for all plant tissues.”
S: “And regarding the number of mRNAs that you capture and their isoforms, can you relate them to the sometimes varying ploidy level in your plant model?”
CS: “Ploidy levels is actually quite an interesting point. Again, we only have knowledge for very few model species, but for those, we know how we can bioinformatically correct for the ploidy levels. However, ploidy levels are whole-genome events. There are several crop and plant species that undergo multiple rounds of genome duplication, and it is much more difficult to identify isoforms because at the moment the majority of single-cell studies are using 3’ based amplification mechanisms. These are not covering 100% of all of the variations that you would see, for example, in gene variants that come from whole-genome duplication.”
S: So is scRNA-seq limited to existing plant models then, if we have to rely on prior research information?”
CS: “Not at all, and this is what is exciting: as long as you have a genome or transcriptome information, you can apply the method to every kind of plant species. It is opening the gates for a lot of research questions that so far have only been done on model plant species, although we know that it would actually make sense to broaden the view and to address the biological research question much better within other plant species.”
“One of those questions is, for example, photosynthesis, or plants that are different in their photosynthesis apparatus. There are so many plant species that have really specialized methods on how to capture light and have good photosynthesis rates. This requires that you look down on the single-cell level, sequencing for example an entire leaf would never give you the resolution you need.”
Spatial transcriptomic for plants is an upcoming development
S: “Can you apply recent technical developments in single-cell sequencing to plants, such as spatial transcriptomic?”
CS: “I think that spatial transcriptomics is really something that is opening the market in research. But for us in general in plant science, I think the main first question with a new technology is ‘does it work with plant cells?’. The answer is almost 98% ‘no, we have never tested it in plant cells.’ ”
“Spatial transcriptomics is a really good example of this. If we see that the importance of the technology is so vital for future research, then we on the platform start heavily investing in it to give it a shot, as we need to start somewhere.”
“Unfortunately the methods that were available were not at the single-cell resolution yet. The distance between hybridization spots [featuring the spatial barcodes for mRNA capture] was still wide enough that you can actually put a whole Arabidopsis root in between spots. But newer technologies with better resolutions are coming up.”
Key points of advice before diving into single-cell plant transcriptomics
S: “To finish our discussion, what would be three 3 main points you would advise a fellow researcher to look into before trying to generate their single-cell datasets?”
CS: “The first thing is to consider the time to spend on it. For study and protocol optimization, protoplast isolation can take quite a long time. However, if you are more interested in manual isolation of the cells, it is, of course, a possibility but it will also require some tweaking. So do not underestimate the time that you need in order to achieve the isolation of the cells.”
“My second point of advice is to look at the question that you want to answer. For example, is it necessary to go for whole-transcript sequencing? Would it make sense to choose something like the smart-seq protocol? Or is your project more about identifying particular of cell types and their transcriptomes from a massive amount of cells?”
“Finally the third piece of advice that I would give is that it must be clear what should be done on the bioinformatics side. You get an amazing amount of data and you realize that you can easily spend years on just one dataset or two, trying to come up with creative new ideas and a creative way of analyzing things. So what you actually want to address must be clear in your mind from the start.”
Dr. Carolin Seyfferth is the coordinator of the plant single cell accelerator program at the VIB, hosted by the De Rybel lab. The program provides a platform for knowledge transfer between academic and industrial institutes to join efforts in promoting single cell technologies in various plant species.
EDIT: After verification, Dr. Seyfferth can affirm with certainty they do not foresee any issue with Poly-A tail enrichment in single-cell plant transcriptomics.
- Seyfferth C, Renema J, Wendrich JR, Eekhout T, Seurinck R, Vandamme N, Blob B, Saeys Y, Helariutta Y, Birnbaum KD, De Rybel B. Advances and Opportunities of Single-Cell Transcriptomics for Plant Research, Annu Rev Plant Biol. 2021 Mar 17. doi: 10.1146/annurev-arplant-081720-010120.