Single Cell RNA Sequencing of Human Platelets Reveals a Stochastic Nature of the Platelet Transcriptome

Background: Platelets are anucleated cells derived from megakaryocytes that regulate blood coagulation and regulate immune function. While less than 10% of circulating platelets contain RNA, they have a demonstrated ability to transcribe their available RNA into protein. While it is known that megakaryocyte transcriptomes are changed in diseased states, it is not known how these changes affect the circulating platelet transcriptome. Furthermore, it has been observed that lung megakaryocytes possess distinct transcriptional profiles compared to those found in bone marrow. Notably, while platelets derived from bone marrow are devoid of MHC class II, lung megakaryocytes exhibit this characteristic. The small percentage of platelets that have RNA has hindered our ability to complete single cell sequencing on these cells. The lack of understanding regarding how disease changes their transcriptome or the heterogeneity of platelets originating from lungs and bone marrow hinders our comprehension of the roles platelets play in hemostasis and immunity. Aims The objective of this study is to single cell sequence human platelets, differentiate the signature of lung and bone marrow megakaryocytes, and profile the transcriptome of healthy human platelets. Methods Blood was taken via venous puncture. Platelets were isolated by slow centrifugation (100g), followed by a second centrifugation at 1200g, and subsequently stained with the RNA dye acridine orange. Platelets were identified by forward and side scatter and sorted based on positive staining for the RNA dye by flow cytometry. cDNA was prepared using the BD Rhapsody system. Files with molecules/cell were produced utilizing BD Rhapsody WTA Analysis Pipeline, employing genecode v43.primary_assembly annotation and GRCh38.primary_ assembly genome. The resulting Molecule Per Cell dense matrix was examined, resulting in 56,724 cells and 43,353 genes to analyze. Data were filtered, removing outliers for total counts, gene counts, and genes counts\cell to unit variance and zero mean. We employed Seurat to detect highly variable genes and regress out (mostly) unwanted sources of variation. Total counts were normalized with target sum of 10,000 counts per cell and log transformed. We computed a neighborhood graph of observations for UMAP. Upon completion, we clustered cells using the Leiden algorithm with low resolution, resulting in 61 detected clusters, and produced UMAP Dimension Reduction to investigate the structure of the data. For every cluster we utilized Wilcoxon approach to rank the differentiating genes in them. Results RNA positive platelets were collected from each of three individuals. Platelets had an average of 282±150 molecules. Moreover, approximately 75%±25 of the reads successfully passed the quality filter. Among the reads that were filtered out, individual comparisons showed percentages of 55%, 18%, and 11% and correspond with the effectiveness of the acridine orange staining and sorting. There was a rapid drop of transcript counts after the Principal Components in their variance, which can indicate simplicity and/or uniformity of the data, presence of outliers, or batch effects. Conclusion(s) Notable heterogeneity was observed among platelets, with a significant presence of mitochondrial genes. Utilizing a ratio of chromosomal and mitochondrial gene expression, we successfully assessed the age of the examined platelets. We identified 61 potential clusters among the sequenced platelets; however, after filtering out noise, we observed that 76% of the cells fell within two clusters, and 90% of the cells were represented in the first 14 clusters. While we were able to identify MHC class II transcripts and platelets expressing HLA-DR (class II), which signifies a lung origin, very few cells contained these transcripts (2550 cells) and those transcripts were not sufficient to separate out a lung specific platelet signature. Our data demonstrate that acridine orange can be used identify RNA containing cells for single cell sequencing and suggest that in healthy individuals, the separation of transcripts into cells is random in nature. We believe that this is the first report of single cell transcriptome from platelets. From these data, we set a baseline of the platelet transcriptome, which allows us to measure changes in transcription in disease states.