Genome-wide screening of circadian and non-circadian impact of Neat1 genetic deletion

Abstract

The functions of the long non-coding RNA, Nuclear enriched abundant transcript 1 (Neat1), are poorly understood. Neat1 is required for the formation of paraspeckles, but its respective paraspeckle-dependent or independent functions are unknown. Several studies including ours reported that Neat1 is involved in the regulation of circadian rhythms. We characterized the impact of Neat1 genetic deletion in a rat pituitary cell line. The mRNAs whose circadian expression pattern or expression level is regulated by Neat1 were identified after high-throughput RNA sequencing of the circadian transcriptome of wild-type cells compared to cells in which Neat1 was deleted by CRISPR/Cas9. The numerous RNAs affected by Neat1 deletion were found to be circadian or non-circadian, targets or non-targets of paraspeckles, and to be associated with many key biological processes showing that Neat1, in interaction with the circadian system or independently, could play crucial roles in key physiological functions through diverse mechanisms.

Keywords: CRISPR/Cas9 edition; Circadian gene expression; Long non-coding RNA Neat1; Paraspeckles; RNA-sequencing.

Graphical abstract

Fig. 1

A-C: Schematic representation of Neat1 genetic edition by CRISPR/Cas9 and characterization of Neat1 KO cell line. A. The positions of guide RNAs used to delete the Neat1 gene are indicated as well as the positions of qPCR primers used to verify deletion of Neat1 gene (primer sequences are given in Supplementary Table 1). B. RNA-Seq peaks for the Neat1 locus, that is not officially annotated in the rat genome, demonstrate preferential expression of Neat1_1 in Neat1WT cells and residual Neat1_1 and Neat1_2 expression in Neat1 KO cells. C. Significant down-regulation of total Neat1 and Neat1_2 in Neat1 KO GH4C1 lines as revealed by RT-qPCR analysis. qPCR primers (Supplemental Table 1) for total Neat1 are in shared sequences between Neat1_1 and Neat1_2 and detect both isoform RNAs. D. Impact of Neat1 genetic edition on circadian expressed genes. Neat genetic deletion leads to a loss of rhythmic pattern in 112 genes (Unique in WT), in a different rhythmic pattern in 362 genes (With different pattern in WT and KO) and in a de novo genesis of rhythmic pattern in 303 genes (Unique in KO). 4485 circadian genes were not affected (Both in WT and KO).

Fig. 2

Analysis of circadian genes affected by Neat1 deletion. A. Analysis of genes from the category “Unique in WT”. Left Panel Heat map representation of genes found to belong to the category “Unique in WT”. Heat maps have been built using the euclidean distance as a measurement of how similar gene expressions in WT group are to each other and the ward.D2 method to group them based on that distance. Corresponding genes are displayed in the same order in KO group. The dendrogram obtained from hierarchical clustering is shown on the left side of the heatmap. A min–max normalization was applied for each gene independently and is represented by colors on the heatmap: higher, blue; lower, black. The hour of cell collection is indicated below (T). Right Upper Panel Cosinor analysis of RNA-Seq counts of two genes, Alcam and Hexim1, shows that counts from WT cells could be adequately fitted (R² > 0.50) with a non-linear cosinor equation in which the period value was set to 24 h (cosinor fit values given in Supplemental Table 4), while counts from Neat1 KO cells can’t. At each time point, data are means ± SEM of samples from the three biological replicates of RNA-Seq. Right Lower Panel Functional characterization by Panther Analysis showing the Gene Ontology biological processes with the highest p value. The numbers inside the columns correspond to the fold enrichment. B. Analysis of genes from the category “With different pattern in WT and KO”. Left Panel Heat map representation of genes found to belong to the category “With different pattern in WT and KO”. Heat maps have been built using the euclidean distance as a measurement of how similar gene expressions in WT group are to each other and the ward.D2 method to group them based on that distance. Corresponding genes are displayed in the same order in KO group. The dendrogram obtained from hierarchical clustering is shown on the left side of the heatmap. A min–max normalization was applied for each gene independently and is represented by colors on the heatmap: higher, blue; lower, black. The hour of cell collection is indicated below (T). Right Upper Panel Cosinor analysis of RNA-Seq counts of two genes, Cdc20 and Nsf, shows that counts from WT and KO cells could be adequately fitted (R² > 0.50) with a non-linear cosinor equation in which the period value was set to 24 h (cosinor fit values given in Supplemental Table 4). At each time point, data are means ± SEM of samples from the three biological replicates of RNA-Seq. Right Lower Panel Functional characterization by Panther Analysis showing the Gene Ontology biological processes with the highest p value. The numbers inside the columns correspond to the fold enrichment. C. Analysis of genes from the category “Unique in KO”. Left Panel Heat map representation of genes found to belong to the category “Unique in KO”. Heat maps have been built using the euclidean distance as a measurement of how similar gene expressions in KO group are to each other and the ward.D2 method to group them based on that distance. Corresponding genes are displayed in the same order in WT group. The dendrogram obtained from hierarchical clustering is shown on the left side of the heatmap. A min–max normalization was applied for each gene independently and is represented by colors on the heatmap: higher, blue; lower, black. The hour of cell collection is indicated below (T). Right Upper Panel Cosinor analysis of RNA-Seq counts of two genes, Ets1 and Zfp39, shows that counts from KO cells could be adequately fitted (R² > 0.50) with a non-linear cosinor equation in which the period value was set to 24 h (cosinor fit values given in Supplemental Table 4), while counts from WT cells can’t. At each time point, data are means ± SEM of samples from the three biological replicates of RNA-Seq. Right Lower Panel Functional characterization by Panther Analysis showing the Gene Ontology biological processes with the highest p value. The numbers inside the columns correspond to the fold enrichment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Analysis of down-regulated and up-regulated genes after Neat1 deletion. A. Volcano plot for Neat1 KO cells versus WT cells is shown; the top 1000 down-regulated and the top 1000 up-regulated genes are highlighted in red. B. The proportion of down-regulated and up-regulated genes is also shown (gene numbers are given in pie charts). C. Functional characterization by Panther Analysis of the top 1000 down-regulated genes showing the Gene Ontology biological processes with the highest p value. The numbers inside the columns correspond to the fold enrichment. D. RNA-Seq counts of two genes, RT1-CE5 and Tlr3, from the top 1000 down-regulated genes are given as an example. At each time point, data are means ± SEM of samples from the three biological replicates of RNA-Seq. E. Functional characterization by Panther Analysis of the top 1000 up-regulated genes showing the Gene Ontology biological processes with the highest p value. The numbers inside the columns correspond to the fold enrichment. F. RNA-Seq counts of two genes, Pou2f2 and Rara, from the top 1000 up-regulated genes are given as an example. At each time point, data are means ± SEM of samples from the three biological replicates of RNA-Seq. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

A. Contribution of Paraspeckle RNA targets to the circadian transcriptome. Venn diagram representing overlaps between circadian transcripts and Paraspeckle (PS) RNA targets. B-D. Contribution of Paraspeckle RNA targets to Neat1 dependent transcriptome. Venn diagrams representing overlaps between Paraspeckle (PS) RNA targets and total genes affected by Neat1 deletion (B), circadian genes affected by Neat1 deletion (C) and genes whose expression is affected by Neat1 deletion (D).