RNA shortening upon cellular stress.

a) Schematic of experimental design b) Volcano plot for the differential expression of arsenite-treated and unstressed cells. Green color indicates genes with more than 2-fold difference and red indicates statistical significance higher than 10−5. c) Cumulative distribution of read length for arsenite-treated and unstressed cells. d) Scatter plots of average transcript length for arsenite-treated and unstressed cells stratified by their differential expression change. Down-regulated: (-Inf, -0.5), Unchanged: (−0.5, 0.5), Up-regulated (0.5, Inf) fold-change. Only transcripts with at least 5 aligned reads are shown. Red dotted indicates the y=x line. Color indicates transcripts below (blue) and above (orange) the diagonal. e) Cumulative distribution of transcript length for arsenite-treated and unstressed cells using only reads with adaptor ligated at the 5’ end. f) Schematic of read meta-length calculation. Each annotated transcript is divided into 20 equally sized bins. Each read is then assigned meta-coordinates depending on the bin in which its 5’ and 3’ templated ends align. The read meta-length is calculated as the difference of the meta-coordinates and presented as a percentage of full length. g) Scatter plot of average transcript meta-length for arsenite-treated and unstressed cells. Coloring is the same as (d) h) Scatter plot of average transcript length for heat shock and unstressed cells.

Characterization of stress-induced shortened RNAs identified by NanopLen:

a) Left: Line plot of average length difference estimate for simulated data of varying read depths. True value is depicted by dashed grey line. Right: Percentage of simulated genes that are detected as significantly different in length. The null simulation of no shortening corresponds to simulated shortening proportion 1.0. b) Box plot of average transcript length for statistically significant and non-significantly shortened transcripts identified through differential length analysis in arsenite-treated and unstressed cells using NanopLen. c) Gene Ontology analysis for biological processes of significantly shortened transcripts. d) IGV screenshot of ASCC3 aligned reads for arsenite-treated (red) and unstressed cells (blue). Libraries were randomly downsampled to maximum 50 reads per window and the libraries were overlayed. All reads were used, irrespective of adaptor ligation status. e) Short-read RNA-Seq density at the 5’ half of transcripts for unstressed, AsO2, H2O2 and heat shock -treated cells. Shade indicates standard error of the mean for replicates. f) Scatter plot of Transcription Start Site position for arsenite-treated and unstressed cells.

The effect of XRN1 knockdown on RNA shortening.

a) Immunoblot for XRN1, eIF2α-P and eIF2α for cells transfected with a mock (siCTRL) or XRN1-targeting (siXRN1) siRNA. ACTB is used as control. b) Cumulative distribution plot of transcript length for XRN1 knockdown (siXRN1) and control (siCTRL) cells with and without arsenite treatment (unstressed). c) Same as (b) but only reads with ligated 5’ end adaptor are used. d) Box plots of differential transcript length in arsenite-treated versus unstressed cells for significantly and non-significantly shortened transcripts upon XRN1 knockdown and control. e-f) Cumulative distribution plot of transcript length for NOT8* D40A E42A and GFP-expressing cells (e) or DCP2* E148Q and GFP-expressing cells (f) with or without (unstressed) arsenite treatment.

Association of poly(A) tail length and cis-regulatory elements with RNA shortening.

a) Nucleotide composition around the 5′ end of reads in arsenite-treated and unstressed cells. All reads were used, irrespective of adaptor ligation status. b-d) Scatter plot of annotated transcript length (b), annotated coding sequence (CDS) length (c) and GC content (d) against transcript differential length in arsenite-treated and unstressed cells. The box plots on the right side summarize the y-axis variable for significant and non-significantly shortened transcripts. The Pearson’s correlation coefficient and the Mann-Whitney-U test p-value are shown.

Translation and RNA shortening.

a, b) Immunoblot for eIF2α and eIF2α-P upon increasing concentration of arsenite (a) and cell treatment with 200 nM of ISRIB at different arsenite concentrations (b). c) SGs visualized by immunofluorescence of HeLa cells treated with indicated concentration of arsenite in the presence or absence of 200 nM ISRIB. A secondary goat anti-rabbit IgG H&L (Alexa Fluor 594) against G3BP1 (SG marker) and DAPI were used for visualization. d) Ribosome sedimentation curve following cell treatment with ISRIB (200 nM) for arsenite-treated and unstressed cells. e) Cumulative density plot of transcript length for arsenite-treated cells in the presence or absence of ISRIB. f) Same as (e) for significantly shortened transcripts only. g) Box plots of differential transcript length for comparisons indicated on the x-axis. h-i) Same as (e) and (f) for cycloheximide CHX instead of ISRIB.

Inhibition of stress-granule formation in cells devoid of G3BP1/2 rescues RNA decay

a) Scatter plot of average gene length for arsenite-treated and unstressed cells stratified by gene SG localization. **** indicates p-value < 10−172. b-c) Cumulative distribution and box plots of average gene length difference in arsenite-treated and unstressed cells stratified by gene SG localization. d) Box plot of GC content percentage for SG enriched and depleted gene transcripts. e) Scatter plot of average transcript length for arsenite-treated and unstressed U-2 OS and ΔΔG3BP1/2 U-2 OS cells. Only transcripts with at least 5 aligned reads are used. Red dotted indicates the y=x line. f) Cumulative distribution plot of average transcript length in arsenite-treated and unstressed U-2 OS and ΔΔG3BP1/2 U-2 OS cells. g) Scatter plot and box plot of differential transcript expression and shortening in arsenite-treated and unstressed ΔΔG3BP1/2 and WT U-2 OS cells.