Figures and data
Construction of an O-GlcNAc responsive GFP biosensor.
a. Schematic of the GFP reporter (left, drawn to scale) and predicted changes in reporter splicing and expression upon varying cellular O-GlcNAc conditions (right). ISS – Intronic splicing silencer; LHA – Left homology arm; RHA – Right homology arm; HBB – Hemoglobin subunit β; PGK – Phosphoglycerokinase; CMV – Cytomegalovirus; bGH – Bovine growth hormone; SV40 – Simian virus 40.
b. Semi-quantitative RT-PCR of RNA isolated from the reporter line under different treatment conditions using DNA primers (NC3378 and NC2094) that hybridize within the GFP ORF and just upstream of the polyadenylation signal sequence as shown below. The PCR conditions make it unlikely to detect the full-length detained intron isoform, so only the mRNA is observed.
c. Northern blot analysis of total RNA isolated from the reporter line after treatment with either DMSO, 1 μM TG or 10 μM OSMI-1 for 6 hours. The blot was probed for GFP. The retained intron band is heterogeneous and difficult to discern clearly due to its co-migration with the large ribosomal RNA. Methylene blue stain of the blot (right) is shown as a loading control.
d. GFP fluorescence levels of the reporter line as measured by flow cytometry after treatment with DMSO, TG or OSMI-1 for 24 hours.
e. Validation of GFP reporter protein levels by western blot analysis after treatment of the reporter line with various modulators of cellular O-GlcNAc levels (left). Steps in the hexosamine biosynthesis pathway targeted by the modulators are shown on the right. Treatment with modulators indicated in red are expected to lead to reduced cellular O-GlcNAc levels, while treatment with those indicated in green are expected to lead to increased cellular O-GlcNAc levels. A broad specificity O-GlcNAc antibody (RL2) and β-actin are used as controls.
f. Northern blot analysis of RNA isolated from either 30 nM non-target (siNT) or OGT specific (siOGT) siRNA-treated reporter line. Cells were treated for 6 hr with DMSO, TG or OSMI-1 3 days after siRNA treatment. The blot was probed for GFP as above.
SFSWAP is a negative regulator of OGT intron 4 splicing.
a. Top, timeline of CRISPR screen. Bottom, MAGeCK analysis of CRIPSR screen results from TG-treated gain of GFP screen in three biological replicates. Top hits are color coded based on predicted function of the protein. Target genes are arranged alphabetically on the x-axis.
b. GFP fluorescence of TG-treated reporter cells 4 days post treatment with siRNA corresponding to non-target (siNT), OGT (siOGT) or SFSWAP (siSFSWAP).
c. Northern blot analysis of RNA isolated from either the TG-treated reporter line (left, probed for GFP) or TG-treated 293A-TOA cells (right, probed for OGT) 4 days after treatment with siRNA corresponding to either non-target (siNT) or SFSWAP (siSFSWAP). Cells were treated with TG for 6 hours just before RNA isolation.
d. RT-qPCR analysis of the splice junctions of interest after treatment of cells with either DMSO, TG or OSMI-1 in the presence or absence of SFSWAP knockdown (n=3). Primers used correspond to either the OGT intron 4 spliced junction (e4-e5) or retained intron junction (RI-e5).
SFSWAP is a global regulator of retained intron splicing and exon skipping.
a. Alternate splicing analysis in untreated SFSWAP knockdown (siSFSWAP) cells compared to non-target (siNT)-treated cells using rMATS (n=3). The number of events of each type are plotted as proportion of total events detected. A3SS – alternate 3’ splice site; A5SS – alternate 5’ splice site; MXE – mutually exclusive exon; RI – retained intron; SE – skipped exon. Assignment of sample labels for events was done based on the value of IncLevelDifference (events with positive IncLevelDifference were designated as siNT and negative IncLevelDifference were designated as siSFSWAP).
b. Scatter plots of retained intron (RI) and skipped exon (SE) events plotted using the JC model of rMATS (top). The difference in inclusion values between siNT and siSFSWAP-treated cells is plotted on the x-axis. Only statistically significant events (FDR <= 0.05) are shown. Violin plots of inclusion levels corresponding to the individual samples are shown below. Median inclusion value for the sample is indicated by the black dot. Only significant events with 20% or greater change in inclusion levels are plotted for the violin plots.
c. IGV screenshot of read coverages of a few significant retained intron events. The intron of interest is marked by the red rectangle.
d. GC content of relevant regions of significant RI and SE events (FDR <= 0.05, >= 10% change in inclusion levels). Red bars indicate the mode of GC content in each region.
SFSWAP regulates OGT decoy exon inclusion.
a. RT-qPCR analysis of OGT intron 4 splicing in UPF1 knockdown background. SFSWAP knockdown was performed for 5 days and cells were then treated with TG for 6 hours before RNA isolation and reverse transcription using a mixture of dT20 and a DNA oligomer complementary to exon 5 of OGT. qPCR was performed for either the spliced junction (e4-e5), retained intron junction (RI-e5) or decoy-e5 junction as shown. DE – decoy exon.
b. IGV screenshot of aligned reads after nanopore sequencing of semi-quantitative RT-PCR amplicons generated from the above samples using DNA oligomers complementary to exon 3 and exon 8 of OGT. A zoomed version is shown below to better show changes in the decoy exon region.
c. Quantification of OGT decoy exon inclusion from RNA-seq data in the presence or absence of UPF1 knockdown and/or TG treatment. Inclusion levels and p-values are calculated from the JCEC model of rMATS performed after alignment against a custom reference annotation of the human genome containing decoy exon annotations.
SFSWAP is a global regulator of decoy exon splicing.
a. Scatter plot of global decoy exon inclusion level changes upon SFSWAP knockdown in a TG-treated UPF1 knockdown background. Significant events with greater than 20% change in inclusion levels are colored. Not all analyzed cassettes may function as splicing decoys.
b. Exon types of the events shown in (a) classified based on the predicted translation outcome. Events shown in blue introduce an in-frame stop codon in the CDS, thus functioning as poison cassettes.
c. Inclusion level changes in decoy-containing retained introns upon SFSWAP knockdown in a TG-treated UPF1 knockdown background. Significant events with 10% or greater change in inclusion levels are colored.
d. Length distribution of the decoy-containing retained introns compared to non-decoy containing retained introns.
Model for the mechanism of action of SFSWAP on intron retention and exon skipping.
See text for details. In the case of retained introns without decoy exons or cassette exons (top and middle), SFSWAP (green oval) restricts splicing subsequent to definition of the exons by U1 and U2. For retained introns with decoys, this inhibition of decoy exon inclusion is linked to the decoy exon’s function to promote intron retention. For simplicity, we showed SFSWAP functioning in combination with U2 snRNP, but it could also be functioning at the 5’ splice sites (see Discussion).
CRISPR screen for the identification of factors regulating OGT intron detention in the absence of TG treatment (n=1). Results of MAGeCK analysis are plotted with target genes arranged alphabetically on the x-axis.
Validation of targets identified in the TG-treated CRISPR screen by RT-qPCR analysis of endogenous OGT intron 4 splicing after knockdown of the target of interest (n=2). Percent spliced OGT RNA was calculated as the ratio of spliced OGT mRNA (exon 4-exon 5 junction) to the sum of spliced and retained (RI-exon 5 junction) forms of the transcript. RNA knockdowns were not verified by western blot or RT-qPCR, so negative results are interpreted to be inconclusive.
Alternate splicing analysis in SFSWAP knockdown background using Whippet (n=3). Scatter plots of RI and SE events are shown with significant events colored (probability >=0.9 and greater than or equal to 20% change in ΔPSI). Violin plots of PSI distribution of significant events from individual samples is shown below. The black dot corresponds to the median PSI value of each sample.
Alternate splicing analysis in SFSWAP knockdown background using MAJIQ (n=3). Scatter plots of RI and SE events are shown with significant events colored (probability >=0.9 and greater than or equal to 10% change in ΔPSI). Violin plots of PSI distribution of significant events from individual samples is shown below. The black dot corresponds to the median PSI value of each sample.
IGV screenshot of read coverage of the OGT transcript from RNA-seq analysis in the presence or absence of SFSWAP knockdown (n=3) without any inhibitor treatment.
Sashimi plot of splice events around the OGT retained intron showing enhanced removal of the retained intron upon SFSWAP knockdown. Splicing to generate the mature mRNA (e4-e5) is shown in green.
Scatter plot of significant RI and SE events from rMATS analysis of RNA-seq data from TG-treated cells in the presence or absence of SFSWAP knockdown.
a. Scatter plot of global decoy exon inclusion level changes upon SFSWAP knockdown in a TG-treated background. Significant events with greater than 20% change in inclusion levels are colored.
b. Exon types of the events shown in (a) classified based on the predicted translation outcome. Events shown in blue introduce an in-frame stop codon in the CDS, thus functioning as poison cassettes.
c. Inclusion level changes in decoy-containing retained introns upon SFSWAP knockdown in TG-treated cells. Significant events with 10% or greater change in inclusion levels are colored.
d. Length distribution of the decoy-containing retained introns compared to non-decoy containing retained introns.
Enrichment of different SRSF1 binding sites on significant alternate splicing events as analyzed by rMAPS. Enrichment on RI and SE events is shown as a motif score across the regions of interest. Colored solid lines (red or blue) correspond to motif scores for upregulated and downregulated events respectively, and dotted lines indicate corresponding -log(p-values). The solid black line indicates the background motif score for the region.
Co-immunoprecipitation (co-IP) analysis of SF1 interaction with SFSWAP. SF1 was immunoprecipitated with mouse anti-SF1 antibodies from 293A-TOA cells treated with either DMSO, TG or OSMI-1 and the blot was probed with rabbit anti-SFSWAP antibodies. The ∼150 kDa band corresponding to SFSWAP is marked. Notably, IP efficiency does not change upon treatment with TG or OSMI-1.
Validation of SFSWAP knockdown for untreated RNA-seq samples (data shown in Fig. 3) by western blot analysis (top) and corresponding quantification (normalized to actin, bottom). siRNA #1 is not an effective siRNA.
IGV screenshot showing knockdown of SFSWAP in the untreated RNA-seq samples (data shown in Fig. 3).
Validation of SFSWAP and UPF1 knockdown in TG-treated RNA-seq samples in the presence or absence of UPF1 knockdown (data shown in Figs. 4 and 5) by RT-qPCR.
IGV screenshot showing knockdown of SFSWAP in the TG-treated RNA-seq samples in the presence (bottom) or absence (top) of UPF1 knockdown (data shown in Figs. 4 and 5).
Key resources
