QKI shuttles internal m7G-modified transcripts into stress granules and modulates mRNA metabolism

Abstract

N⁷-methylguanosine (m⁷G) modification, routinely occurring at mRNA 5' cap or within tRNAs/rRNAs, also exists internally in messenger RNAs (mRNAs). Although m⁷G-cap is essential for pre-mRNA processing and protein synthesis, the exact role of mRNA internal m⁷G modification remains elusive. Here, we report that mRNA internal m⁷G is selectively recognized by Quaking proteins (QKIs). By transcriptome-wide profiling/mapping of internal m⁷G methylome and QKI-binding sites, we identified more than 1,000 high-confidence m⁷G-modified and QKI-bound mRNA targets with a conserved "GANGAN (N = A/C/U/G)" motif. Strikingly, QKI7 interacts (via C terminus) with the stress granule (SG) core protein G3BP1 and shuttles internal m⁷G-modified transcripts into SGs to regulate mRNA stability and translation under stress conditions. Specifically, QKI7 attenuates the translation efficiency of essential genes in Hippo signaling pathways to sensitize cancer cells to chemotherapy. Collectively, we characterized QKIs as mRNA internal m⁷G-binding proteins that modulate target mRNA metabolism and cellular drug resistance.

Keywords: G3BP1; METTL1; N(7)-methylguanosine; QKI; drug resistance; m(7)G; mRNA metabolism; mRNA stability; stress granule; translational regulation.

Figure 1.. Distribution and quantitation of internal m⁷G modification in mRNA

(A) The chemical structure of m⁷G methylation. (B-C) The internal m⁷G/G levels in various RNAs (B) and the nuclear and cytoplasmic cap-depleted polyA⁺ RNAs (C), as detected by LC-MS/MS. (D-E) METTL1 (D) and WDR4 (E) KO efficacy in HeLa and HepG2 Cas9 single clones. (F-G) LC-MS/MS quantification of internal m⁷G abundance changes in small RNA (F) and cap-depleted polyA⁺ mRNA (G) isolated from HeLa and HepG2 cells upon METTL1 or WDR4 KO. (H) Validation of wild-type METTL1 (METTL1-WT), catalytically inactive METTL1 (METTL1-Mut), or WDR4 overexpression in HeLa and HepG2 cells. (I-J) LC-MS/MS quantification of internal m⁷G abundance changes in small RNA (I) and cap-depleted polyA⁺ mRNA (J) isolated from HeLa and HepG2 cells upon METTL1 and WDR4 overexpression. Two-tailed student’s t-test (C); One-way ANOVA (F, G, I, J). *p < 0.05; **p < 0.01; ***p < 0.001. Data in B, C, and J are presented as mean ± SEM (n = 4); Data in F, G, and I are presented as mean ± SEM (n = 2). See also Figure S1.

Figure 2.. Selective binding of QKIs to internal m⁷G residues in mRNA

(A) Schematic of RNA affinity chromatography and MS analysis. (B) Dot blot showing the m⁷G levels (left) and Biotin levels (right; loading controls) in the 60-mer m⁷G/G ssRNA probes. (C) Scatter plot of proteins bound to m⁷G-Oligos versus G-Oligos as detected by the mass spectrometry. (D) Western blotting showing the endogenous Pan-QKI, IGF2BP1-3, LRPPRC and IDH1 proteins pulled down by biotin-labeled G-Oligos and m⁷G-Oligos from HepG2 whole cell lysates. (E) RNA pulldown assays showing the dose-dependent interaction between endogenous Pan-QKI and biotin-labeled G-Oligos (left) or m⁷G-Oligos (right). (F) The top five binding motifs of METTL1 and QKI. (G) Western blotting showing the cell-free binding between recombinant QKI proteins and biotin-labeled G-Oligos or m⁷G-Oligos. (H) Western blotting showing the flag-tagged QKI5, QKI6 and QKI7 proteins pulled down by biotin-labeled G-Oligos or m⁷G-Oligos using whole cell lysates from QKI5-, QKI6-, and QKI7-overexpressing HeLa cells, respectively. (I) Quantification of m⁷G/G ratio by LC-MS/MS in decapped polyA⁺ RNA bound by recombinant QKIs as determined by cell-free RIP assay (mean ± SEM; n = 2). (J) Quantification of m⁷G/G ratio by LC-MS/MS in decapped polyA⁺ mRNA bound by QKIs in HEK293T cells as determined by the ex vivo RIP assay (mean ± SEM; n = 2). (K) Metagene profile showing the distribution of m⁷G peaks in HepG2 cells in two biological replicates along a normalized transcript composed of 5’UTR, CDS and 3’UTR. (L) Top internal m⁷G motif in HepG2 mRNAs, identified by HOMER. (M) Overlap between m⁷G-modified transcripts and QKI-bound transcripts in HepG2 cells. (N) Top two motifs of the binding sites of QKI5 (left), QKI6 (middle) and QKI7 (right) in HepG2 cells, identified by HOMER. (O) Circos plot showing the QKI7-binding peaks (blue), internal m⁷G peaks (green), and overlapping peaks (red) in HepG2 cells. (P) MA plots displaying the hyper (increased) and hypo (decreased) QKI7 binding peaks in HepG2 mRNAs upon METTL1 KO. (Q) The frequency distribution of QKI7 binding peaks in sgNS and sgMETTL1 HepG2. See also Figure S2 and Table S2–3.

Figure 3.. QKI proteins interact with SG core protein G3BP1 under stress conditions

(A) Subcellular localization of the potential m⁷G binding proteins in HeLa and HepG2 cells (C = cytosol; N = nuclear). (B) IF staining of ectopically expressed flag-tagged QKIs in HeLa cells. (C) Profile intensity showing Flag/Phalloidin/DAPI fluorescence signals of the white lines in (B). (D) The predicted interaction network of QKI binding proteins. (E-F) Co-IP (E) and reciprocal Co-IP (F) showing the interaction between QKI6 (Flag) or QKI7 (Flag) and G3BP1 in HEK293T cells under normal (Ctrl) and stress conditions. (G) In situ detection of QKI6-G3BP1 and QKI7-G3BP1 interaction via PLA in untreated (left) and NaAsO₂ treated (right) U2OS cells. (H) Quantification of in situ PLA puncta in NaAsO_2-treated U2OS cells. (I-J) IF staining of Flag-tagged QKIs and G3BP1 in U2OS cells under normal (I) or NaAsO₂-mediated stress (J) conditions. (K-L) Pearson correlation analysis showing the colocalization of Flag-tagged QKIs with SGs (G3BP1) (K) or the RFP-tagged QKIs with G3BP1-GFP (L) in NaAsO₂-treated U2OS cells (mean ± SEM; n = 4). (M) Co-IP showing the potential interaction between G3BP1 and several truncated QKIs in HEK293T cells under stress conditions. See also Figure S3.

Figure 4.. QKI transfers internal m⁷G-decorated mRNAs into SGs

(A) Workflow for IF staining of internal m⁷G-modified polyA⁺ mRNA in adherent cells. (B) IF staining of internal m⁷G-modified mRNA in HeLa cells under normal and NaAsO₂-induced stress conditions. (C) Relative enrichment of m⁷G signal and polyA signal in SGs (G3BP1-positive dots) of HeLa cells with or without NaAsO₂ treatment. (D) m⁷G/G levels in cap-depleted total, nuclear, cytoplasmic, and insoluble RNA-granule (RG)-enriched polyA⁺ mRNAs isolated from U2OS under normal and stress conditions (mean ± SEM; n = 4). (E) Western blotting showing the level of wild-type or catalytic inactive mutant METTL1 in U2OS cells with METTL1 KO. (F-G) Violin plots (F) and representative IF pictures (G) of internal m⁷G modification in NaAsO₂-treated U2OS cells with or without METTL1 KO and rescued METTL1 expression. (H) Western blotting showing the level changes of indicated proteins in U2OS cells upon QKI KD. (I-J) Western blotting showing the levels of indicated proteins in U2OS cells upon QKI overexpression (I) or KO (J). (K) Relative fluorescence intensity of m⁷G signals in SGs upon QKI KD in U2OS cells. (L) Western blotting showing the rescued levels of wild-type and truncated (Δ KH domain) QKI7 in U2OS cells with QKI KO. (M-N) Violin plots (M) and representative immunofluorescence pictures (N) of internal m⁷G modification in NaAsO₂-treated U2OS cells with or without QKI KO and forced expression of wild-type or truncated QKI7. (O) LC-MS/MS quantification of m⁷G/G ratio in decapped insoluble RG-enriched polyA⁺ mRNA fractions from NaAsO₂-treated U2OS cells upon QKI7 KO (left) and overexpression (right) (mean ± SEM; n = 4). Two-tailed student’s t-test (C, K, and O); Two-way ANOVA (D, F, and M). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001. See also Figure S4.

Figure 5.. Transcriptome-wide analysis of QKI7-bound and internal m⁷G-modified transcripts in SGs

(A) Motifs enriched in QKI7-binding peaks and internal m⁷G peaks in U2OS cells. (B) Cumulative distribution function (CDF) plot of mRNA enrichment in U2OS SGs. (C) Overlap among internal m⁷G-modified transcripts, QKI7-bound transcripts, and SG-enriched transcripts (left) or SG-depleted transcripts (right) in U2OS cells. (D) Volcano diagram displaying the enrichment or depletion of SG-purified mRNAs compared to total mRNA. (E) The mRNA composition of SG-enriched genes (left) or SG-depleted genes (right) in U2OS cells. (F) KEGG pathway analysis of the internal m⁷G-modified & QKI7-bound genes in U2OS entire transcriptome (top) or U2OS SG transcriptome (bottom). (G) List of core-enriched genes in the “Hippo signaling pathway” and “pathway in cancer”. (H) SG isolation workflow. (I) Validation of efficient SG isolation from U2OS cells. G3BP1, GAPDH, and PRL7 were used as the marker of SG, cytoplasm, and ribosome, respectively. (J) Heatmaps showing the relative abundance changes of the candidate target genes of QKI7 in SGs of U2OS cells upon QKI7 overexpression (left, normalized to EV group) or KO (right, normalized to sgNS group) as determined by qRT-PCR. GAPDH, negative control. Statistical analysis: Two-tailed student’s t-test (B, J); Person’s chi-square test (C). See also Figure S5.

Figure 6.. QKI7 regulates the metabolism of a set of its target transcripts in an internal m⁷G-dependent manner

(A) CDF plot showing the effects of QKI level changes on global mRNA stabilities in NaAsO₂-treated U2OS cells. (B-C) Overlap between transcripts with significantly decreased half-lives upon QKI KO (‘KO-down’) and QKI7-bound (B) or SG-enriched (C) transcripts in U2OS cells. (D) Heatmap showing the relative half-life changes of the representative target genes in 0.2 mM NaAsO₂-treated U2OS cells with manipulation of QKI expression. (E) Western blotting showing the distribution of QKI7 and RPL7 in 10%−50% sucrose gradient fractions of U2OS or HeLa cells under normal (ctrl) and stress conditions. (F) Volcano diagram displaying the up-regulated and down-regulated genes in NaAsO₂-treated U2OS cells upon QKI7 overexpression. (G) Scatter dot plot displaying the effect of QKI7 overexpression on translation efficiency of QKI7-bound (left) or SG-enriched (right) genes in U2OS cells under stress conditions. (H-I) Distribution of representative targets in 40S fractions and polysome portions of U2OS cells upon QKI7 overexpression under normal (H) or NaAsO₂-treated (I) conditions (mean ± SEM; n = 3). (J) Screening strategy for functionally essential targets of the QKI7-m⁷G axis. (K) QKI7-RIP qPCR showing the effects of METTL1 KO on the enrichment of QKI7 on representative target mRNAs in U2OS cells (mean ± SEM; n = 3). (L) Western blotting showing the levels of representative targets in control and QKI7-overexpressing U2OS cells upon NaAsO₂ treatment. (M) Schematic illustration showing the effect of QKI on shuttling the internal m⁷G-modified mRNA into SGs and regulating their translation efficiency and/or stability. Two-tailed student’s t-test (A, G, H, I, K); Hypergeometric test (B, C). *p < 0.05; **p < 0.01; ***p < 0.001. See also Figure S6.

Figure 7.. QKI confers sensitivity to chemotherapy drugs in cancer cells

(A) Correlation between QKI and SG core genes in various cancer types. (B-C) Effect of QKI overexpression on doxorubicin IC₅₀ in parental (B) or METTL1 KO (C) cancer cells. (D) Effect of QKI overexpression on cancer cell proliferation. (E-F) Effect of QKI7 overexpression or QKI KO on the apoptosis of HeLa (E) or U2OS (F) cells treated with DMSO or doxorubicin for 24 hours. (G) Clonogenic assay showing the effect of QKI overexpression (left) or KO (right) on survival of HeLa cells treated with DMSO or doxorubicin for 14 days. (H) Images of xenografts formed by control or QKI7-overexpressing HeLa cells treated with doxorubicin (1 mg/kg) or vehicle. (I-J) Tumor growth curve (I) and weight (J) from the indicated groups. (K-L) Dot plots showing the relative (K) and exact (L) IC₅₀ values of 90 drugs in QKI_Low and QKI_high HCC cell lines. (M) Heatmap presenting the relative IC₅₀ value of a variety of chemotherapy drugs in QKI_Low and QKI_High HCC cell lines. Extra-sum-of-squares F test (B, C); One-way ANOVA (D, E, F); Two-way ANOVA (G, I, J); Two-tailed student’s t-test (L). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001. Data in B-G are presented as mean ± SEM (n = 3); data in I and J are presented as mean ± SEM (n = 7). See also Figure S7.