Single-cell RNA sequencing (scRNA-seq) allows the study of individual cells in complex tissues in an unbiased manner. This, among others, has proven to be successful for the discovery of previously undescribed cell types or the dissection of the role of individual cell types in complex diseases. However, to acquire representative, high-quality scRNA-seq data with limited batch effects, several challenges that are encountered during the preparations of cells need to be tackled. First, scRNA-seq requires the use of single-cell suspensions. This can be easily acquired for cells that are already in suspension, like blood cells. However, for tissues that are solid this can be a challenge. Second, good experimental design dictates to combine multiple samples and experimental conditions together in one experiment to prevent that potential batch effects are associated with the experimental questions that are tackled. In our research we use patient-derived gut biopsies to study complex intestinal diseases. To obtain such precious samples through invasive endoscopy, extensive coordination with doctors is required. Thus, a single-cell dissociation method for gut tissue that is compatible with cryopreservation is critical for conducting a solid scRNA-seq study using gut tissue.
The gut mucosa is a complex organ that can be divided into the epithelial layer, located in the luminal part; and the lamina propria layer, lying underneath the epithelium. Each compartment is composed of a mixture of cell types, which may have specialized and local functions, and that are differently distributed across the gastrointestinal tract. The ideal dissociation protocol for gut tissue should recover each of the cell types present in gut biopsies without affecting the original cell proportions and transcriptome profiles. However, dissociation protocols are known to impact the cell recovery and may induce transcriptional changes.
In our publication, we compared three different dissociation methods using two different dissociation enzymes to recover viable cells from cryopreserved biopsies: one-step collagenase, two-step collagenase and three-step protease dissociation. The first one was developed in our lab to speed-up the dissociation process and to overcome the different dissociation procedures that were previously required for the different compartments within the tissue. This protocol combines enzymatic digestion with collagenase IV and mechanical disruption of the tissue. The second method, which is the most commonly used one in the field, relies on an initial separation of the epithelial layer from the lamina propria using an EDTA incubation, after which the lamina propria is further digested using enzymatic digestion with collagenase. The last method, relatively new, is very similar to the two-step collagenase but with the difference that it adds an extra step which takes advantage of using a protease that is active at 4°C, thereby intending to minimize dissociation-induced transcriptional changes. We evaluated the impact of these dissociation protocols on cell viability, cell type proportions and the transcriptome profile of each recovered cell type by using flow cytometry and scRNA-seq read-outs.
We found that each dissociation protocol performs differently at recovering specific cell types from fresh and cryopreserved samples. For example, the two-step collagenase is most suitable when aiming to recover myeloid cells and fibroblasts in fresh samples, but the viability of cells is compromised when using cryopreserved samples. In contrast, both one-step collagenase and three-step protease protocols show comparable viability of the recovered cells in cryopreserved and fresh samples. The one-step collagenase protocol provides the same dissociation treatment for all compartments, and thereby provides no dissociation-induced differences across the compartments. Therefore, being the method of choice when interested in studying all compartments together. On the other hand, the multi-step protocols have the benefit of enabling enrichment of cells coming from a specific compartment, which may be the preferable choice to cost-efficiently study specific cell types within these compartments.
Next, we assessed the impact of cell dissociation on the transcriptome of individual cell types using scRNA-seq. This revealed dysregulation of specific gene sets by the different dissociation protocols and the impact varied across cell types. For example, within the immune cells, the myeloid and B cells were most sensitive to collagenase treatment showing the largest fraction of known collagenase-induced genes, whereas the plasma cells were most sensitive to protease treatment showing the lowest number of remaining cells after dissociation. Similarly, in fresh versus cryopreserved samples cells showed a different transcriptomic profile. These results may help to carefully interpret future work, especially when observing dysregulated genes that are likely to be influenced by dissociation.
Finally, we showed that the one-step collagenase protocol is compatible not only with scRNA-seq analysis, but also with flow cytometry and the expansion of cells (e.g. epithelial stem cells for organoid studies and IELs) for follow-up functional studies. Thereby we believe that the one-step collagenase protocol using cryopreserved samples is the preferred approach for explorative, broad characterization of the gut. Nevertheless, each protocol has its unique benefits and disadvantages. Unfortunately, there is not a single protocol that can suit everyone’s needs. As the balance between benefits and disadvantages can be different depending on the research question, knowing the impact of dissociation protocols is crucial when planning single (gut) cell experiments. Based on our findings, we designed a decision three summarizing our main findings, to help other researchers to pick the dissociation method best suiting their needs. We hope that our work also inspires other researchers to think about which factors, other than biological, are affecting their results.