YTHDF2 promotes ATP synthesis and immune evasion in B cell malignancies

Abstract

Long-term durable remission in patients with B cell malignancies following chimeric antigen receptor (CAR)-T cell immunotherapy remains unsatisfactory, often due to antigen escape. Malignant B cell transformation and oncogenic growth relies on efficient ATP synthesis, although the underlying mechanisms remain unclear. Here, we report that YTHDF2 facilitates energy supply and antigen escape in B cell malignancies, and its overexpression alone is sufficient to cause B cell transformation and tumorigenesis. Mechanistically, YTHDF2 functions as a dual reader where it stabilizes mRNAs as a 5-methylcytosine (m⁵C) reader via recruiting PABPC1, thereby enhancing their expression and ATP synthesis. Concomitantly, YTHDF2 also promotes immune evasion by destabilizing other mRNAs as an N⁶-methyladenosine (m⁶A) reader. Small-molecule-mediated targeting of YTHDF2 suppresses aggressive B cell malignancies and sensitizes them to CAR-T cell therapy.

Keywords: ATP production; B cell malignancies; CAR-T; CD19; MHC-II; YTHDF2; m(5)C; m(6)A; metabolism.

Figure 1.. YTHDF2 is overexpressed and plays an oncogenic role in B-cell malignancies.

(A) Heatmap of the m⁶A machinery genes’ expression patterns in patients with B-ALL. (B) YTHDF2 abundance as shown in (A). (C) Score of YTHDF2 in the dependency map (DepMap) portal. (D) YTHDF2 mRNA abundance in patients with DLBCL and healthy controls. (E) qRT-PCR of YTHDF2 mRNA in normal and malignant B cells. (F) OS, EFS or PFS of two grouped patients (divided by YTHDF2 abundance) with B-ALL, DLBCL (phs001444.v2.p1), or MCL (cBioPortal). (G) YTHDF2 protein abundance in leukemia and healthy cells. Statistical analysis was shown. (H) YTHDF2 protein abundance in normal and malignant B cells. (I-K) Growth competition assays of mCherry⁺-shRNA or RFP⁺-sgRNA malignant B cells. (L-O) Cell growth/proliferation changes upon YTHDF2 KD in B-ALL cells (L), lymphoma cells (M), and B-ALL-PDX cells (N), or upon Ythdf2 cKO in BCR-ABL1-transformed B-ALL cells (O), where Cre-ER^T2-IRES-GFP vectors (Cre) or ER^T2-IRES-GFP (EV) was transduced followed by 4-Hydroxytamoxifen (4-OHT) treatment. (P-S) Growth competition assays in BCR-ABL1 (P) or NRAS^G12D (Q) B-ALL cells (carrying EV/Cre), or B-ALL cells carrying either puro-ER^T2 (R) or puro-Cre-ER^T2 (S) which were further transduced with empty-vector-GFP (EV-GFP) or Ythdf2-GFP. Data are shown as mean ± SD and assessed by two-tailed Student’s t-test (B/D/ E/G) or two-way ANOVA (I-S). Log-rank tests are used for survival analyses (F). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. See also Figures S1

Figure 2.. YTHDF2 is promoted by TFs in expression and plays a pivotal oncogenic role in the development/progression of B-cell malignancies.

(A) Kaplan-Meier curves of B-ALL-PDX-xenotransplanted NSG mice. (B) Comparison of spleen weights of the mice in (A). (C) Kaplan-Meier curves of NSG mice transplanted with BCR-ABL1- or NRAS^G12D-B-ALL cells (carrying EV/Cre). (D) Survival curves of C57BL/6 mice transplanted with BCR-ABL1-Ythdf2^fl/fl-B-ALL cells. (E-G) Wright-Giemsa staining of PB smears (E), organ weight (F) and BM engraftment analyses (G) for the mice in (D). The scale bar represents 20 μm. (H) Survival curves of recipient mice with (cKO) or without (WT) Mx1-Cre transplanted with either BCR-ABL1-Ythdf2^fl/fl Mx1-Cre (cKO) or BCR-ABL1-Ythdf2^fl/fl (WT) pre-B cells. (I) UMAP showing clusters in scRNA-seq. (J) Cell proportion changes of T cell clusters (C14 and C15) in (I). (K-L) Photomicrographs (K) and colony number (L) of serial CFA for control- or Ythdf2-transduced pre-B cells. Scale bar, 100 μm. (M) Wright-Giemsa staining of PB smears of the NSG mice transplanted with EV-, YTHDF2-WT, or m⁶A-MUT-transduced pre-B cells. The scale bar represents 20 μm. (N) Engraftment analysis of mice BM in (M). (O) Kaplan-Meier curves of recipient mice for the primary and secondary BMT. (P) LDA showing leukemia stem/initiating cell frequency. (Q-R) Binding signals of TFs (ChIP-Atlas) (Q) and ChIP-qPCR (R) at the promoter of YTHDF2. (S) YTHDF2 protein abundance changes in TFs-KD malignant B cells. (T) Correlation analysis of gene expression in patient samples (GSE31312). r, Pearson’s rank correlation coefficient. Data are shown as mean ± SD and assessed by two-tailed Student’s t-test (B/F/J/L/R). Log-rank tests are used for survival analyses (A/C/D/H/O). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. See also Figures S2

Figure 3.. YTHDF2 promotes ATP synthesis in malignant B cells independent of m⁶A modification.

(A) Venn diagram of YTHDF2-regulated (RNA-seq) and m⁶A-modified (MeRIP-seq) genes in IAH8R cells. (B) Donut chart of YTHDF2-regulated genes with or without m⁶A modification. (C) Sankey diagram of module KEGG pathways enriched by the 816 genes in (B). (D) Bubble diagrams showing the top enriched pathways in malignant B cells. All the metabolism-related terms were labeled in red. (E) ATP amount in metabolites profiling assays. (F) Enrichment of YTHDF2-regulated metabolites. ATP-involved metabolite sets were labeled in red. (G) The integrative pathway analysis (non-m⁶A-modified & downregulated transcripts). (H) Seahorse assays showing the oxygen consumption rate (OCR) in B-ALL-PDX cells (left) and YTHDF2-tranduced pre-B cells (right). (I-J) ATP (I) and related protein (J) abundance changes upon YTHDF2 KD in cancer cells. (K) Correlation analysis by qRT-PCR across 45 healthy and malignant B cells. r, Pearson’s rank correlation coefficient (L) Target protein abundance in malignant B cells and healthy controls. (M-P) Changes of cell survival/growth (M), OCR (N), ATP abundance (O) and PDX mouse survival curves (P) in malignant B cells upon target genes KD. (Q-S) The rescue effects of overexpression of the three target genes simultaneously on ATP abundance (Q), cell survival/growth (R) and OCR (S) in malignant B cells with YTHDF2 KD. Data are shown as mean ± SD and assessed by two-tailed Student’s t-test (I/O/Q) or two-way ANOVA (M/R). Log-rank tests are used for survival analyses (P). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. See also Figures S3

Figure 4.. YTHDF2 recognizes/stabilizes m⁵C-modified mRNAs by recruiting PABPC1.

(A) Schematic of WT and mutated YTHDF2 proteins. (B) RIP followed by UHPLC-QQQ-MS. (C) UHPLC-QQQ-MS showing the abundance/intensity of m⁵C peaks in YTHDF2-bound RNAs. (D) RNA pull-down assays of YTHDF2 with m⁶A- or m⁵C-modified ssRNA compared to unmodified ssRNAs (A or C groups). (E) YTHDF2-bound mRNA transcripts by RIP-seq. (F) Venn diagram of genes overlapped in m⁵C MeRIP-seq and bisulfite-seq, and YTHDF2-RIP-seq. (G) Distribution of YTHDF2-bound sites relative to m⁵C peaks across the whole transcriptome. (H-J) Overlap of m⁵C and YTHDF2-bound regions (H), frequency analysis (I), and m⁵C-MeRIP-qRT-PCR (J) in IAH8R cells. (K) RNA pull-down assays of YTHDF2 with m⁵C-ssRNA corresponding to the top 2 sites in (I). (L) RIP-qRT-PCR. Statistical analysis was calculated and compared to D-MUT. (M) The mRNA stability detected by qRT-PCR in RS4;11-Cas9 cells. (N) Gene expression correlation analysis in patients with B-ALL and DLBCL. (O) PABPC1 protein abundance measurement. (P) qRT-PCR of PABPC1 mRNA in normal and malignant B cells. (Q-R) Analyses of cell proliferation/growth (Q) and mice survival curves (R). (S) Reciprocal co-IP assays in malignant B cells. (T) Overlapping genes in PABPC1-bound (RIP-seq), PABPC1-regulated (RNA-seq), m⁵C-modified (MeRIP-seq) and 816 genes in (Figure 3B) in IAH8R cells. (U-V) OCR (U) and ATP (V) changes upon PABPC1 KD. (W) RIP-qRT-PCR. (X) Target protein abundance changes upon PABPC1 KD. (Y) Diagram illustrating the molecular mechanisms. Data are shown as mean ± SD and assessed by two-tailed Student’s t-test (B/J/ L/P/V/W) or two-way ANOVA (Q). Log-rank tests are used for survival analyses (R). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. See also Figures S4

Figure 5.. YTHDF2 destabilizes m⁶A-modified CD19 and HLA-DMA/B, and its depletion sensitizes malignant B cells to T-cell-based immunotherapy.

(A) Gene ontology (GO) enrichment analysis of the 1,023 genes listed in Figure 3B. (B) The most enriched pathways of the above 1,023 genes. (C) Venn diagram of upregulated genes upon YTHDF2 KD and the enriched pathways. (D) GSEA plots using a pre-ranked m⁶A-modified gene list weighted by FC in IAH8R cells. (E) Gene expression correlation analysis of in-house samples. r, Pearson’s rank correlation coefficient. (F) The mRNA stability in RS4;11-Cas9 cells. (G) qRT-PCR of mRNA abundance. (H) MFI plots showing the cell surface protein abundance changes in the same patient between diagnosis (2016-23) and relapse (2016-54), or in malignant B cells upon YTHDF2 KD. (I) MeRIP-seq tracks of m⁶A distribution. m⁶A were shown in red. (J) BST-qRT-PCR assays. RT primers were shown in (I). (K) RNA pull-down assays of YTHDF2 proteins with m⁶A-modified ssRNAs corresponding to (I). (L) MeRIP-qRT-PCR for m⁶A presence/enrichment. (M) RIP-qRT-PCR. (N) T cell co-culture assays. (O) Representative florescent images. White arrows indicated immune synapse. Scale bar, 10 μm. (P) Quantitative comparison of synapse size between CD19-CAR-T cells (GFP⁺) and B-ALL cells (RFP⁺). Length of synapse was measured and normalized to T cell membrane diameters. (Q-R) CD19-CAR-T cell co-culture assays in mCherry⁺-SU-DHL-4 cells (Q) and doxycycline-induced Cas9-ZsGreen-PDX2 cells (PDX2-iCas9) carrying RFP⁺-sgRNAs (R). Data are shown as mean ± SD and assessed by two-tailed Student’s t-test (G/H/J/L/M/N/P/Q/R). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. See also Figures S5

Figure 6.. Identification of CCI-38 as an effective YTHDF2 inhibitor.

(A) Early B cell development in mouse BM detected by flow cytometry. (B) Mature B cell analysis in mice splenic CD19⁺B220⁺ pregated cells. (C) Splenic CD4⁺ or CD8⁺ T cell populations analysis in Ythdf2 KO mice. (D) MFI plots showing the cell surface CD19 and intracellular H2-M abundance in Ythdf2 cKO mice. (E) Gating strategy for splenic GC B cells and subpopulations post immunization. (F) Analysis of splenic GC B cell proportion in CD19⁺-pregated and that of DZ and LZ in GC B-pregated populations after immunization. (G) Virtual screening for compounds binding with YTHDF2 and other m⁶A-related proteins. (H) Docking model and structure of YTHDF2 protein (4rdn) with compounds (dark red and green). (I-J) DARTS assays (I) and the MST time trace (J) for CCI-38 and YTHDF1/2/3. (K) FTSA showing binding of CCI-38 (50 μM) with WT and mutated YTHDF2 proteins. (L) RNA pull-down assays for m⁶A-ssRNAs & YTHDF1/2/3 with or without CCI-38 presence. (M-Q) Changes of cell surface protein abundance (M), mRNA stability (N), protein abundance (O) and mRNA abundance (P and Q) upon treatment with CCI-38 for 24 hours. (R) DARTS showing CCI-38 binding with YTHDF proteins with or without Cysteine mutations in 293T cells. Data are shown as mean ± SD and assessed by two-tailed Student’s t-test (A/B/C/D/F/M/P/Q). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. See also Figures S6

Figure 7.. CCI-38 exhibits potent anti-tumor efficacy and synergizes with T-cell-based immunotherapy in treating B-cell malignancies.

(A) IC₅₀ of CCI-38 for 72 hours in malignant B cells. (B-C) Changes of ATP abundance (B) and OCR (C, 1 μM) after CCI-38 treatment (24 hours). (D) T cell co-culture with CCI-38 pretreated malignant B cells with blinatumomab. (E) CD19-CAR-T cell co-culture with CCI-38 pretreated malignant B cells. (F) YTHDF2 protein abundance in B-ALL patient samples. (G) Cell surface CD19 abundance changes in patient samples after CCI-38 treatment (24 hours). (H) CD19-CAR-T cell co-culture with relapsed patient cells after CCI-38 treatment (24 hours). (I) Bioluminescence imaging of luciferase-IAH8R-xenotransplanted mice after treatment with CCI-38, CD19-CAR-T cells or combination. (J) Kaplan-Meier analysis of recipient mice in (I). (K) Survival curve of 2016-54-xenotransplanted NSG mice. (L) Schematics showing YTHDF2’s role in immune response. Data are shown as mean ± SD and assessed by two-tailed Student’s t-test (B/D/E/G/H). Log-rank tests are used for survival analyses (J/K). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. See also Figures S7