20 19 12 3 1 (really caught the tail of 20 19! ), the research team of Wellcome Sanger Institute published a paper entitled "Scrna-Seq Evaluation of the Stability of Human Lung, Spleen and Esophagus Tissue after Cryopreservation", and discussed the influence of sample cryopreservation and cryopreservation time on the sequencing of human lung, spleen and esophagus single cell transcriptome.
In order to solve the rapid transcription change/stress reaction caused by tissue dissociation or preservation, a series of cell freezing or fixing methods have been developed. Guillaumet-Adkins and others proved that although the viability of cultured cells or chopped mouse biopsy tissues decreased after DMSO cryopreservation, the transcription profile did not change significantly. However, some types of cells are more easily frozen than others. For example, in human endometrial biopsy, stromal cells have a higher survival rate than epithelial cells under frozen conditions. In addition, some studies have evaluated the ability of traditional cross-linking fixatives, reversible cross-linking agents, non-cross-linking substitutes (such as methanol) and other new stabilizers to fix cells. Although fixation usually produces 3' bias, it can prevent the change of transcription and ensure that the cell type does not change. So far, these reagents have only been tested on dissociated cells or the most chopped tissues, but not on complete tissue samples. Unfortunately, it is usually impractical to separate human clinical samples before transportation, and it is often difficult to separate fixed tissue blocks by traditional mechanical or enzymatic methods.
In this study, the author aims to determine a tissue preservation method, which can ensure that the transcription status of complete tissue samples used for scRNA-seq remains stable, with minimal clinical treatment and certain transportation time. In order to contribute to the human cell mapping project, the author tested this method on three kinds of human primary tissues with different sensitivities to ischemia: spleen (the most stable), esophageal mucosa and lung (the most unstable). These tissues contain many different cell types, from immune cells to keratinocytes. Samples were collected from dead organ donors and perfused with cold organ preservation solution quickly after the death of the donors.
The 240,000 single-cell data sets generated by this study include the largest single-cell data set of human esophagus and spleen published so far. You can view and analyze data in www.tissuestabilitycelllas.org. It was found that the intact tissue blocks of these three organs were stored in 4℃ low temperature sol FRS for 24 hours or 72 hours (most samples were stored for 72 hours), and then the transcriptome of common and 10x 3' enriched single cells was sequenced, and the transcriptome had little effect. The results of population diversity obtained by scRNA-seq did not change with the increase of freezing time. This scheme should be easily adopted by many clinical institutions, and allow at least a 24-hour window to transport samples to collaborators, thus increasing the chances of obtaining fresh human tissues for research.
The authors obtained lung, esophagus and spleen samples from 12 organ donors, and evaluated them by genome-wide sequencing and pathological sections. All samples are healthy samples to avoid some glands or inflammation in the samples from affecting the evaluation of the results of scRNA-seq. Fresh samples were dissociated (T0) or re-dissociated immediately after cold ischemia 12h, 24h and 72h at 4℃, and then single cell sequencing was performed with 10X 3'v2 scRNA-seq. As shown in the figure below, after comparing and standardizing the scRNA-seq data, the quality control indicators of all samples are evaluated. For lung and esophagus, the quality indexes such as the number of readings per sample, the number of cells per sample and the median number of genes detected per cell did not change obviously with time, but the comparison rate of samples stored in spleen organs for 72 hours decreased obviously.
Conclusion: The author did not detect the change of quality control indexes related to the length of cold storage time within 24 hours of cold ischemia. (Cold ischemia time: the time from cold perfusion (cold preservation) to blood supply after transplantation. )
Figure 1
Further detailed analysis showed that the percentage of exon reading decreased significantly only in the spleen (fig. 2a, b). In addition, only the percentage of introns in spleen samples increased with the increase of storage time (fig. 2c, d). Changes in the proportion of high-quality readings in the spleen at 72h (Figure 2b, c) may lead to cell type-specific differences, which will be further discussed in the following results. The ratio of intron reading to exon reading changes more obviously in top 25% high-quality exon reading and top 75% high-quality intron reading, suggesting that non-splicing reading is more stable for degradation (this conclusion is a bit strange, please guide your familiar friends). Finally, the researchers also found that the percentage of mitochondrial gene expression in spleen only increased significantly with the passage of time (Figure 2e, F, G).
Figure 2( R Language Learning-boxplot (violin diagram, jitter diagram, regional scatter diagram))
Firstly, the double fraction of droplets in the three tissues does not change with time.
Next, the author evaluates the changes of non-cellular droplets. All droplet-based sequencing reactions will produce many droplets that contain no cells but capture mRNA in the environment, which is usually called environmental RNA or soup. A relatively arbitrary threshold is set for the normalized UMI number of each droplet, which is defined as the environmental RNA is 0-0.25, the debris is 0.25-5, and the normalized umi of each droplet is > 5 to reflect the distribution of readings (Figure 3a). The cold storage time of spleen, lung or esophagus samples has no obvious effect on the proportion of droplets containing UMI. The average normalized UMI of fragmented droplets in spleen increased at 72 hours, while the cellular droplets decreased (fig. 3b, c). This phenomenon was not observed in lung or esophageal samples. However, the average normalized UMI of debris droplets and cellular matter in the three organs varies greatly.
Figure 3
In the lung, 57,020 cells passed the quality control and were labeled as 25 cell types. Cilia, alveoli 1 and 2 cells, fibroblasts, muscles and endothelial cells were detected in blood vessels and lymphatic vessels. The cell types identified from the immune compartment include NK, T and B cells, two kinds of macrophages, monocytes and dendritic cells (DC). Several DC cell subsets were detected, such as cDC 1, plasma cell-like DC(pcDC) and activated DC, accounting for 0.3% (163 cells), 0.08% (46 cells) and 0.2% (122 cells) of all cells, respectively. Alveolar club cell marker genes were detected in a few cells, but the clustering algorithm did not cluster these cells separately (the resolution of this classification is also very high, and all 46 cell groups can be separated, either because of the high initial resolution or because of cell group subdivision. Although clustering is intuitive, it needs to be tried again and again. Divide as many meaningful clusters as possible to keep the stability of clustering results. )。
After quality control, 87,947 cells remained in esophageal samples, of which more than 90% belonged to four main epithelial cell types: upper basal layer, graded cells, cells on basal layer and splinter cell on basal layer. Other cells from epithelial basal layer gather more closely in glandular ducts and mucus secretion cells. Immune cells in esophagus include T cells, B cells, monocytes, macrophages, DC and mast cells. Interestingly, almost 80% of mast cells (87 cells) came from a single donor. The proportion of other immune cells (B cells, DC, monocytes/macrophages) in the donor increased. The donor was diagnosed with ventilator-associated pneumonia and was excluded from later comparison.
All 94,257 cells from the spleen were labeled as immune cells. Follicular B cells and cuff B cells are considered as the largest cell groups, accounting for 17%(: 16000 cells) and 20%(>: 18000 cells) respectively. More than 6000 plasma cells were detected, which were labeled as plasma mother cells and expressed IgG or IgM. More than 28,000 T cells are labeled as CD4+ classic cells, CD8+ activated cells, CD4+ immature cells, CD4+ follicular helper cells (fh), CD8+MAIT-like cells, CD8 +γ-δ cells, CD8+cytotoxic lymphocytes (CTL), CD4+ regulatory cells or mitotic T cells. Two subsets of natural killer (NK) cells, mitotic NK subsets, monocytes, macrophages and DC were also identified. The proportion of multicellular groups is very low, such as DC subgroups, including activated DC(0.04%%), conventional DC 1 (0.3%) and PCDC-S (0.3%), as well as congenital lymphocytes (0.6%), CD34+ progenitor cells (0.2%) and platelets (0.08%). A subset of cells containing more than 2207 T cells and B cells simultaneously can represent the duplex of interacting cells, which is called T_B duplex.
Figure 4
After labeling the cell types, we can study the changes of cell type composition ratio with refrigeration time. The proportion of cell types between samples and donors varies greatly. When examining the changes of cell types in donors with time, we noticed that the proportion of B cells in spleen (Figure 4f) increased with the increase of storage time (note the orange column area in the figure), while the proportion of T cells in lung (Figure 4d) and spleen (Note the blue column area in the figure 4f) decreased with the increase of storage time (Figures 4d-f, D lung, E esophagus, and F spleen).
Next, the author studied the effect of storage time on whether the transcriptome has cell type specificity. Notably, the UMAP map calculated from highly variable genes did not change significantly with time (fig. 4g, h). Integrate the gene expression matrices of all tissues and calculate the percentage of variability explained by different variables. As shown in fig. 4j, different donors, tissues, cell types and total number of UMI cells explain the highest proportion of variance sources, while the storage time has the least influence (purple line). (If you want to do this analysis, see the article "How to identify confounding factors in single cell transcriptome". )
The author puts forward a cold storage method of human original tissue samples, which does not need any treatment except clinical field collection, and gives at least 24 hours of operation time for transportation, tissue dissociation and scRNA-seq sequencing. The quality indexes of lung and esophagus samples remained stable within 72 hours after cold storage. In the spleen, we observed the change of intron and exon reading ratio and the increase of mitochondrial reading percentage at 72 hours.
The results show that putting tissue samples into cold tissue preservation solution immediately after collection can ensure that different types of cells are least affected after ischemia. Refrigeration time has no effect on the diversity of cell population or the change of overall RNA-seq in scRNA-seq data within 24 hours.
In addition, the author also provides data resources of lung, spleen and esophagus, covering single cells, common transcriptome, genome-wide sequencing and clinical data. For further analysis and use by scientists in this field.
With such a large amount of data, only one genome biology has been published. Is it a bit overqualified, so can this data resource be further explored?
| Proofreading: Life Letters Collection