Journal: Nucleic Acids Research

Article Title: Alternative polyadenylation links RNA processing to iron metabolism in human erythropoiesis

doi: 10.1093/nar/gkag218

Figure Lengend Snippet: Single-cell atlas of PAS dynamics during human myelopoiesis and erythropoiesis. (A-B) Unsupervised transcriptional trajectory of myelopoiesis, including erythropoiesis, from Monocle 2, colored by cell states (A) and eight cell clusters (B) , which were clustered and defined based on lineage-specific markers expression . HSC, hematopoietic stem cells; G./M., granulocyte or monocyte lineage; Ery., erythrocyte lineage. (C) The top-scoring polyadenylation (PA) signals were identified through de novo motif analysis of all 14 510 PAS peaks across 15 606 cells in myelopoiesis. The sequences analyzed spanned from 30 nt upstream to 120 nt downstream of the peak’s 3′-end. Statistical significance was assessed using Fisher ’ s exact test. (D) Schematic representation of the proximal PAS usage index ( pPUI ). In 3′tag scRNA-seq data, reads from different RNA molecules of the same transcript isoform accumulate at distinct peaks, which can be used to infer PAS locations and assess their usage. The arrows indicate PASs classified as proximal (Prox.), middle (Mid.), or distal (Dist.) based on their distance from the 5′end of the 3′UTR. The pPUI , defined by scAPA , quantifies the relative usage of the most proximal PAS within a 3′UTR containing two or more peaks. C1, the read count of the most proximal peak; , the geometric mean (Geo Mean) of the counts of all the peaks associated with the 3′UTR. (E) The mean pPUI across all 2849 APA genes at the single-cell level. (F) pPUI calculated at the cell-cluster level for each APA gene. The plot displays pPUI across different cell types, with points representing the mean values of all genes and error bars indicating the standard deviation. A t -test was performed to assess pPUI differences between HSCs and other cell clusters along the differentiation trajectory. * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001; ns , not significant. (G) UMAP projection of 3132 single cells across healthy human erythropoiesis and basophilopoiesis. (H) Trajectory analysis of erythropoiesis and basophilopoiesis was performed similarly to (A) (top: cell states; bottom: cell clusters defined in ). The right panel displays the cell clusters and differentiation trajectory, with cell numbers indicated. (I) The upper right corner displays the top-scoring signal identified by de novo motif analysis of 12 456 PAS peaks from 9572 3′UTRs. The sequences analyzed spanned from 30 nt upstream to 120 nt downstream of the peak’s 3′-end. The density plot illustrates the distribution of PA signal locations relative to the peak’s 3′-ends. Note that the peak’s 3′-end does not precisely correspond to the actual PAS position. Given that the PA signal is typically located ∼20 nucleotides upstream of the PAS, the actual PAS position can be inferred to be ∼ 25 nt downstream of the peak’s 3′-end. (J) Cumulative percentage of predicted PASs supported by annotated PASs from PolyA_DB v3.2 . Distance cutoffs ranging from 10 nt to 100 nt (in 10-nt increments) were applied to determine the fraction of predicted PASs located within each distance threshold of known PASs. (K) The Integrative Genomics Viewer (IGV) displays a gene example with unannotated PASs. mRNA 3′-ends detected by 3′-seq are shown as peaks. Red arrows indicate newly identified PASs in erythropoiesis, which have no corresponding PASs in PolyA_DB v3.2 .

Article Snippet: RNA sequencing (RNA-seq) and 3′mRNA sequencing (3′-seq) were performed by Novogene (Beijing, China).

Techniques: Single Cell, Expressing, Standard Deviation

Journal: Nucleic Acids Research

Article Title: Alternative polyadenylation links RNA processing to iron metabolism in human erythropoiesis

doi: 10.1093/nar/gkag218

Figure Lengend Snippet: CPSF6 knockdown alters 3′UTR-APA. (A) Schematic diagram of experimental design. Human UCB-derived CD34 + HSPCs were induced to erythroid cells in vitro , and infected with shRNA lentivirus to knock down CPSF6. Cells on day 11 (D11) and 13 (D13) were subjected to bulk RNA-seq and 3′-seq. Schematic read distributions from 3′-seq and RNA-seq illustrating 3′UTR APA changes are shown on the right. (B) Cumulative distribution of the pPUI in differentiated cells with control and CPSF6 knockdown. Every shCPSF6 sample was compared to its control by the Kruskal–Wallis test, and the P values were all less than 2.2e-16. The pie chart shows the number and proportion of differentially shortening and lengthening APA-genes. (C) Volcano plot of differential APA-genes of CPSF6 knockdown samples compared to control samples identified by bulk RNA-seq at D11. (D) Venn diagram showing genes with differential APA identified in normal erythropoiesis (scRNA-seq) and in CPSF6 knockdown cells (bulk RNA-seq and 3′-seq at days 11 and 13). See . ( E and F ) Genome browser tracks of 3′-seq and RT-qPCR validation of representative iron metabolism-related differential APA genes after CPSF6 knockdown. (G) GSEA plot of APA-gene pPUI in 3′-seq datasets on day 11.

Article Snippet: RNA sequencing (RNA-seq) and 3′mRNA sequencing (3′-seq) were performed by Novogene (Beijing, China).

Techniques: Knockdown, Derivative Assay, In Vitro, Infection, shRNA, RNA Sequencing, Control, Quantitative RT-PCR, Biomarker Discovery