Ultra-low-input rG4-seq reveals the RNA G-quadruplex regulome in gene expression and genome integrity

Abstract

RNA G-quadruplexes (rG4s), formed through guanine self-recognition into stacked tetrads, serve as critical regulators of gene expression, yet their comprehensive mapping and dynamic regulation in physiological contexts remain technically challenging. Here, we develop Ultra-low-input rG4-seq (ULI-rG4-seq), enabling precise rG4 detection enabling precise rG4 detection with ∼140 bp resolution in samples as small as 100 oocytes, and reveal notable enrichment of rG4s near crucial regulatory regions, particularly transcription start sites and end sites. This technological advance, combined with Trim-away or oocyte-specific knockout of DHX36 (also known as G4R1 or RHAU), an rG4-specific helicase, reveals acute and chronic loss of DHX36 leads to opposing effects on rG4 levels. This observation extends beyond the traditional view of helicases as unwinding enzymes and suggests sophisticated cellular mechanisms maintaining RNA structural homeostasis. Through integrated analysis of rG4 landscapes and DHX36-binding profiles, we demonstrate coordination between cytoplasmic rG4 regulation and nuclear gene expression, revealing how RNA structure dynamics orchestrate RNA stability and translation, thereby influencing transcriptional elongation, genome stability, and alternative splicing. Finally, we show that deletion of DHX36 resulted in decreased oocyte quality, premature ovarian failure and complete female infertility due to transcriptional defects and genome instability related to R-loop accumulation. These technological and conceptual advances not only deepen our understanding of RNA-based regulation but also open new therapeutic possibilities for diseases involving RNA structure.

Graphical Abstract

Figure 1.

Development and validation of ultra-low-input rG4 sequencing (ULI–rG4-seq). (A) Representative images showing rG4 distribution in mouse oocytes visualized by QUMA-1 staining (red) with DMSO control or cPDS treatment for 12 h. Scale bar: 20 μm. (B) Quantification of QUMA-1 signal intensity in control and cPDS-treated oocytes. Data are presented as mean ± SEM; n = 30 oocytes per group; P < 0.0001 (Student’s t-test). (C) Schematic illustration of the ULI-rG4-seq protocol. The workflow consists of two main steps: (i) oocyte treatment and initial crosslinking, and (ii) library preparation including RNA immunoprecipitation, fragmentation, and sequencing; ZP, zona pellucida. (D) Correlation heatmap showing reproducibility between biological replicates of ULI–rG4-seq with DMSO control or cPDS treatment. Numbers indicate Pearson correlation coefficients. (E) Sequence motif analysis of rG4-forming regions identified by ULI–rG4-seq, showing enrichment of canonical G-quadruplex motifs. E-values and number of sites are indicated for each motif.

Figure 2.

ULI–rG4-seq maps rG4 sites in dynamic transcriptome. (A) Bar graph showing the number of genes detected by ULI–rG4-seq with DMSO control or cPDS treatment. Overlapping and non-overlapping genes are indicated in different colors. (B) Bar graph showing the number of rG4 peaks detected by ULI–rG4-seq with DMSO control or cPDS treatment. (C) Venn diagram showing the overlap between RNA-seq and ULI–rG4-seq datasets with DMSO control or cPDS treatment. The numbers indicate gene counts in each category: RNA-seq unique (4963), genes detected in both RNA-seq and ULI-rG4-seq with cPDS (1481), genes detected in all three datasets (3065), and other overlapping categories (555, 500, 634, and 312). (D) Pie chart showing the genomic distribution of rG4 peaks identified by ULI–rG4-seq. (E) Metagene analysis showing the distribution of rG4 signals around TSS and TES with DMSO control or cPDS treatment. (F) Representative genome browser tracks showing rG4 signals at the Bmp15 locus in biological replicates with DMSO control or cPDS treatment. (G) Network analysis of DHX36-specific target genes grouped by biological processes. Different functional categories are indicated by distinct colors.

Figure 3.

DHX36 regulates global RNA G-quadruplex structures. (A) Representative immunofluorescence images showing the co-localization of DHX36–GFP with QUMA-1in mouse oocytes. Scale bar: 10 μm. (B) Line-scan analysis showing the fluorescence intensity profiles of DHX36–GFP and QUMA-1 signals along the blue arrow indicated in (A). (C) Metagene analysis showing the distribution of rG4 signals (with DMSO control or cPDS treatment) and DHX36-binding sites around TSS and TES. (D) Venn diagram showing the overlap between rG4-containing transcripts and DHX36-bound targets identified by ULI-rG4-seq and LACE-seq, respectively. (E) Gene Ontology (GO) enrichment analysis of rG4-containing genes. The x-axis shows gene ratio, dot size represents gene count, and color intensity indicates P-value. Selected biological processes are shown on the y-axis. (F) Representative images showing rG4 structures (QUMA-1, red) in control and Dhx36^fl/fl;SKO oocytes. Scale bars: 10 μm. (G) Quantification of QUMA-1 signal intensity in control and Dhx36^fl/fl;SKO oocytes. Data are presented as mean ± SEM; n = 30 oocytes per group; P < 0.0001 (Student’s t-test). (H) Metagene analysis showing the distribution of rG4 signals around TSS and TES in control and Dhx36^fl/fl;SKO oocytes. (I) Western blot analysis showing DHX36 protein levels in control and Dhx36^fl/fl;SKO oocytes. β-Actin serves as loading control.

Figure 4.

Acute depletion of DHX36 reveals temporal dynamics of rG4 regulation and mRNA stability. (A) Schematic illustration of the Trim-away strategy for acute depletion of endogenous DHX36–GFP protein in oocytes. (B) Representative images showing DHX36–GFP and TRIM21-mCherry in control and Trim-away oocytes after 3 h treatment. Scale bars: 10 μm. (C) Quantification of DHX36–GFP signal intensity in control and Trim-away oocytes. Data are presented as mean ± SEM; P < 0.0001 (Student’s t-test). (D) Representative images showing DHX36–GFP and QUMA-1 in control and Trim-away oocytes; scale bars: 10 μm. (E) Quantification of QUMA-1 signal intensity in control and Trim-away oocytes. Data are presented as mean ± SEM; P < 0.0001 (Student’s t-test). (F) Metagene analysis shows the distribution of rG4 signals (with control or Trim-away). (G) Cumulative distribution analysis of expression changes for rG4-containing and non-rG4 transcripts in Dhx36^fl/fl;SKO oocytes. KS test P-value is indicated. (H) Bar graph shows the proportion of up- and down-regulated genes with rG4 structures in control and Dhx36^fl/fl;SKO oocytes. (I and J) RNA-seq analysis shows the expression levels of Zar1 mRNA in control and Dhx36^fl/fl;SKO oocytes. Data are presented as mean ± SEM; **P < 0.01, ***P < 0.001 (Student’s t-test). (K and L) Genome browser tracks showing rG4 signals at the Zar1 and Zar1L loci in biological replicates. (M) Schematic diagram of transcription inhibition experiment using DRB treatment in control and Dhx36^fl/fl;SKO NSN (non-surrounded nucleolus) oocytes. (N) Density plot showing the distribution of mRNA half-lives in control and Dhx36^fl/fl;SKO oocytes. Dashed lines indicate median half-lives. (O) Cumulative frequency distribution of mRNA half-lives after DRB treatment. rG4-containing transcripts (blue) show shorter half-lives than non-rG4 transcripts (red) in control oocytes (solid lines). DHX36 loss (dashed lines) selectively destabilizes rG4-containing transcripts. (P) Half-life changes of 85 RNA stabilization genes with rG4 in Dhx36^fl/fl;SKO versus control. Numbers indicate gene counts. Increased (orange); Dncreased (blue); n.s.: unchanged (gray).

Figure 5.

DHX36 regulates translation of rG4-containing transcripts and protein synthesis. (A) Representative images showing protein synthesis (HPG-594, red) in control and Dhx36^fl/fl;SKO oocytes. Yellow arrows indicate nuclei; scale bars: 10 μm. (B) Quantification of HPG-594 signal intensity in cytoplasmic and nuclear compartments of control and Dhx36^fl/fl;SKO oocytes. Data are presented as mean ± SEM; P < 0.0001 (Student’s t-test). (C) Cumulative distribution analysis of protein expression changes for rG4-containing and non-rG4 targets in Dhx36^fl/fl;SKO oocytes. Wilcoxon test P-value is indicated. (D) Schematic diagram of the dual-reporter system for monitoring rG4 structure-specific translation in live cells. The construct contains an rG4-regulated mCherry and a control EGFP reporter. (E) Time-lapse imaging showing the translation of rG4-mCherry and control-EGFP reporters in control and Dhx36^fl/fl;SKO oocytes over 150 min. (F) Quantification of fluorescence intensity for rG4-mCherry and Ccnb1–EGFP reporters in control and Dhx36^fl/fl;SKO oocytes over time. Data are presented as mean ± SEM; ^****P < 0.0001 (Student’s t-test). (G) Genome browser tracks showing rG4 signals at the Ybx2 locus in biological replicates. (H) Representative images showing YBX2 protein (red) in control and Dhx36^fl/fl;SKO oocytes. Yellow arrows indicate nuclei; scale bars: 10 μm. (I) RNA-seq analysis showing Ybx2 mRNA expression levels in control and Dhx36^fl/fl;SKO oocytes. Data are presented as mean ± SEM; ***P < 0.001 (Student’s t-test). (J) Quantification of YBX2 protein levels in control and Dhx36^fl/fl;SKO oocytes. Data are presented as mean ± SEM; ***P < 0.001 (Student’s t-test).

Figure 6.

DHX36 regulates transcriptional elongation through modulation of RNA G-quadruplex structures. (A and B) Representative images (A) and quantification (B) of nascent transcription visualized by EU incorporation (green) in control and Dhx36^fl/fl;SKO oocytes. Scale bars: 10 μm. (C and D) Representative images (C) and quantification (D) of RNA Polymerase II (Pol II, green) in control and Dhx36^fl/fl;SKO oocytes; scale bars: 10 μm. (E and F) Metagene analysis showing Pol II distribution around TSS and TES in control versus Dhx36^fl/fl;SKO oocytes (e) and control versus cPDS-treated oocytes (F). (G) Heatmap showing genome-wide Pol II distribution patterns in control, Dhx36^fl/fl;SKO, and cPDS-treated oocytes. (H) Cumulative distribution of Pol II pausing index in control and Dhx36^fl/fl;SKO oocytes. Inset shows boxplot of log2 pausing index. (I) Bar graph showing the number of up- and down-regulated transcription elongation-related genes with rG4 structures. (J and K) Genome browser tracks showing rG4 signals and DHX36 binding at the Ccnt1 (J) and Tcea1 (K) loci. (L and M) Representative images (L) and quantification (M) of CCNT1 protein in NSN and SN stage oocytes from control and Dhx36^fl/fl;SKO mice. (N and O) Representative images (N) and quantification (O) of TCEA1 protein in control and Dhx36^fl/fl;SKO oocytes. (P–R) Representative images (P) and quantification (Q and R) of phosphorylated Ser2 RNA Pol II (pS2) and CDK9 in control and Dhx36^fl/fl;SKO oocytes. (S) Metagene analysis showing CDK9 distribution around TSS and TES in control and Dhx36^fl/fl;SKO oocytes. (T) Schematic model illustrating DHX36-mediated regulation of transcriptional elongation through rG4 unwinding. Data are presented as mean ± SEM; P < 0.0001 (Student’s t-test); scale bars: 10 μm.

Figure 7.

Loss of DHX36 leads to R-loop accumulation and DNA damage. (A and B) Representative images (A) and quantification (B) of R-loops detected by S9.6 antibody in control and Dhx36^fl/fl;SKO oocytes. DNA was stained with DAPI (blue). Data are presented as mean ± SEM; P < 0.0001 (Student’s t-test); scale bars: 10 μm. (C) Western blot analysis showing levels of DHX36, phosphorylated CHK1 (p-CHK1), and γH2AX in control and Dhx36^fl/fl;SKO oocytes. β-Actin serves as loading control. (D) Representative images showing co-localization of R-loops (S9.6, green) with DNA damage markers RAD51 and 53BP1 (red) in control and Dhx36^fl/fl;SKO oocytes; scale bars: 10 μm. (E) Representative images showing R-loops (S9.6) and DNA double-strand breaks (γ-H2AX) in control and Dhx36^fl/fl;SKO oocytes with or without RNaseH1 mRNA or α-amanitin treatment; scale bars: 10 μm.

Figure 8.

DHX36 regulates alternative splicing through modulation of splicing factors. (A) Schematic representation and distribution of different types of alternative splicing events. SE: skipped exon; A5SS: alternative 5′ splice site; A3SS: alternative 3′ splice site; MXE: mutually exclusive exon; RI: retained intron. (B) Distribution of differential splicing events in Dhx36^fl/fl;SKO oocytes compared to controls. PSI: percent spliced in. (C and D) Sashimi plots showing alternative splicing patterns of Zar1 (C) and Nlrp14 (D) in control and Dhx36^fl/fl;SKO oocytes. Numbers indicate read counts supporting each splicing junction. (E) Bar graph showing the number of up- and down-regulated RNA splicing-related genes containing rG4 structures in control and Dhx36^fl/fl;SKO oocytes. (F) Genome browser tracks showing rG4 signals and DHX36 binding at the Srsf9 locus in biological replicates. (G) RNA-seq analysis showing expression levels of Srsf9 mRNA in control and Dhx36fl/fl;SKO oocytes. Data are presented as mean ± SEM; *P < 0.05 (Student’s t-test). (H and I) Representative images (H) and quantification (I) of SRSF9 protein in control and Dhx36^fl/fl;SKO oocytes. Data are presented as mean ± SEM; P < 0.0001 (Student’s t-test); scale bars: 10 μm.

Figure 9.

DHX36 deficiency leads to reduced oocyte quality, premature ovarian failure and female infertility. (A) Cumulative number of pups per female in control and Dhx36^fl/fl;SKO mice over a 6-month breeding period. Data are presented as mean ± SEM. (B) Number of ovulated oocytes per mouse in control and Dhx36^fl/fl;SKO females at different ages. Data are presented as mean ± SEM. (C) Representative bright-field images showing ovulated oocytes from control and Dhx36^fl/fl;SKO females at 2 months (2M) and 3 months (3M) of age after Day 1 and Day 2 culture. Yellow arrows indicate germinal vesicles; scale bar: 100 μm. (D) Abnormal meiotic arrest in Dhx36^fl/fl;SKO oocytes after superovulation. While control females yielded normal zygotes, Dhx36^fl/fl;SKO females produced immature oocytes arrested at the germinal vesicle (GV) stage or abnormal embryos derived from GV-stage oocytes (yellow arrows indicate GV structures), suggesting essential roles of DHX36 in meiotic progression. Nuclei were visualized by DAPI staining; scale bars: 10 μm. (E) Representative H&E-stained ovarian sections from control and Dhx36^fl/fl;SKO females at 2, 3, 5, and 7 months of age; scale bars: 200 μm. (F) Quantification of total follicle numbers per ovary in control and Dhx36^fl/fl;SKO females at different ages. Data are presented as mean ± SEM. (G) Working model illustrating how DHX36-mediated rG4 regulation maintains genome stability and oocyte quality through rG4 regulation. Left: In normal conditions, DHX36 unwinds rG4 structures to facilitate proper transcription elongation. Right: DHX36 depletion leads to transcription elongation defects, R-loop accumulation, and subsequent DNA damage.