Skip to main content
. 2017 Dec 14;6:e29224. doi: 10.7554/eLife.29224

Figure 2. APEX-RIP mapping of the nuclear-cytoplasmic RNA distribution.

(A) Fluorescence imaging of nuclear and cytosol-targeted APEX2 fusion constructs. HEK 293T cells expressing the indicated constructs (‘NLS,’ nuclear localization signal; ‘NES,’ nuclear export signal) were labeled with biotin-phenol, crosslinked and stained as indicated. Scale bars, 10 µm. DAPI is nuclear stain. (B) APEX-RIP recovers known nuclear and cytosolic standard RNAs, defined here as long noncoding RNAs (nuclear markers, blue) and RNAs proximal to the Endoplasmic Reticulum (cytoplasmic markers,red—defined by (Jan et al., 2014), with measured p-values≤0.05—see Materials and methods). Top: APEX2–NLS-RIP enriches nuclear standards. Middle: APEX2–NES-RIP enriches cytoplasmic standards. Bottom: Combined analysis of the APEX2–NLS and APEX2–NES RIP experiments distinguish the two classes. Fold changes are defined as (FPKMpost-enrichment/FPKMpre-enrichment); combined fold change as [FPKMNLS–post-enrichment /FPKMNES–post-enrichment]. Dotted line indicates the significance threshold for nuclear localization. (C) Global analysis of nuclear and cytoplasmic RNA localization by combined APEX2-NLS and APEX2-NES RIP. Vertical dashed lines indicate the cutoffs for nuclear and cytosolic RNAs. Horizontal dash line indicates p-value=0.05. Top histogram illustrates the distribution of RNAs with p-value=5×10−5, which are boxed in gray in the scatter plot. The average of data from three biological replicates are shown. See Supplementary file 2A-B. (D) APEX-RIP reveals classes of RNAs with canonical and noncanonical nuclear-cytoplasmic distributions. Left: the same data as in (C), separately parsed into mRNAs (top) and lncRNAs (bottom). Right: read density plots of example RNAs from each class that exhibit stereotypical and atypical localization. EEF1A1 and C1orf63 are mRNAs; XIST and SNHG5 are lncRNAs. For each gene, a common y-scale is used for all read tracks. SnoRNAs encoded in the SNHG5 gene body are indicated as gray rectangles. (E) Venn diagrams comparing APEX-RIP and fractionation-based RNA datasets (Sultan et al., 2014). (F) Nuclear APEX-RIP is more sensitive than is biochemical fractionation. Left: Specificity of the APEX-RIP and nuclear RNA datasets (Sultan et al., 2014). Off-target RNAs were defined as actively translated ER-proximal mRNAs (Jan et al., 2014). Right: Recall of nuclear standard RNAs, defined as a set of 827 lncRNAs annotated by GENCODE hg19 with average pre-enrichment FPKM ≥ 1.0. See also: Figure 2—figure supplements 1–2.

Figure 2.

Figure 2—figure supplement 1. Characterization of APEX2 fusion constructs.

Figure 2—figure supplement 1.

HEK 293 T cells stably expressing the indicated constructs (right) were labeled and crosslinked via Protocol II (Figure 1—figure supplement 1A). Cell lysates were analyzed by SDS-PAGE, blotting with streptavidin-HRP, anti-V5 and anti-FLAG. L: ladder; U: untransfected HEK 293T cells. Anti-V5 and anti-FLAG blots (bottom left) measure fusion construct expression.
Figure 2—figure supplement 2. Reproducibility of nuclear–cytoplasmic APEX-RIP experiments.

Figure 2—figure supplement 2.

Pairwise correlations between global RNA abundances (FPKM) in biological replicates of the APEX2–NLS-RIP experiment, (A) prior to, and (B), following enrichment, and similarly, of the APEX2–NES-RIP experiment, (C) prior to, and (D) following enrichment.
Figure 2—figure supplement 3. Precision and specificity of nuclear–cytoplasmic APEX-RIP, and its comparison to subcellular fractionation.

Figure 2—figure supplement 3.

(A–B) Determination of cutoffs for APEX-RIP data by ROC analysis (see Materials and methods). (C) APEX protein localization isn’t perturbed during the APEX-RIP labeling and crosslinking protocol. APEX2-NLS HEK293T cells expressing a whole-cell GFP marker were processed using the optimized APEX-RIP protocol (bar diagram). Cells were fixed prior to, and following the mild formaldehyde crosslinking step (arrowheads), and imaged by immunofluorescence. Left: Representative images for cells fixed at each time point. Right: average nuclear-vs-cytosolic biotinylation signal ratio measured at both time points; each dot represents an individual cell (n = 30). p-value: student’s paired t-test. (D-—E) APEX-RIP recapitulates the results of nuclear-cytosolic fractionation-sequencing. (D) Global analysis of nuclear and cytoplasmic RNA localization [FPKMnuclear/FPKMcytosolic] by biochemical fractionation-sequencing of HEK293 cells (Sultan et al., 2014). Horizontal dash line indicates p-value = 0.05. Top histogram illustrates the distribution of RNAs with p-value = 5×10−5, which are boxed in gray in the scatter plot. Average data from two biological replicates are shown. (E) Distribution of RNAs based on enrichment of nuclear over cytosolic fractionation, similar to the bottom panel of Figure 2B (see Materials and methods). (F–G) Determination of cutoffs for fractionation data (Sultan et al., 2014) by ROC analysis (see Materials and methods).