Abstract
Pseudouridylation is a common RNA modification that is catalyzed by the family of pseudouridine synthases (PUS). Pseudouridylation can increase RNA stability and rigidity, thereby impacting RNA splicing, processing, and translation. Given that RNA metabolism is frequently altered in cancer, pseudouridylation may be a functionally important process in tumor biology. Here, we show that the MYC family of oncoproteins transcriptionally upregulates PUS7 expression during cancer development. PUS7 is essential for the growth and survival of MYC-driven cancer cells and xenografts by promoting adaptive stress responses and amino acid biosynthesis and import. ATF4, a master regulator of stress responses and cellular metabolism, was identified as a key downstream mediator of PUS7 functional activity. Induction of ATF4 by MYC oncoproteins and cellular stress required PUS7, and ATF4 overexpression overcame the growth inhibition caused by PUS7 deficiency. Mechanistically, PUS7 induced pseudouridylation of MCTS1 mRNA, which enhanced its translation. MCTS1, a noncanonical translation initiation factor, drove stress-induced ATF4 protein expression. A PUS7 consensus pseudouridylation site in the 3′ untranslated region of ATF4 mRNA was crucial for the induction of ATF4 by cellular stress. These findings unveil an MYC-activated mRNA pseudouridylation program that mitigates cellular stress induced by MYC stimulation of proliferation and biomass production, suggesting that targeting PUS7 could be a therapeutic strategy selectively against MYC-driven cancers.
Significance: Oncogene activation of mRNA pseudouridylation is a mechanism that facilitates metabolic reprogramming and adaptive responses to overcome cellular stress during cancer development.
Introduction
Analogous to DNA and histone modifications, RNA undergoes extensive modifications (1, 2), including pseudouridylation (3–6). Pseudouridine (Ψ) was initially discovered (7, 8) and extensively studied in noncoding RNAs (tRNA, rRNA, and small nuclear RNA) due to their abundance (3). The presence of Ψ in low abundant, native mRNAs was unknown until about 10 years ago after advances in high-throughput sequencing technology (9–12). Transcriptome-wide mapping studies have identified thousands of Ψ sites in human mRNAs (refs. 9–11, 13–16; bioRxiv 2024.01.31.578250). Functional studies have shown that Ψ can increase the stability and rigidity of RNAs and impact various aspects of RNA metabolism, including splicing, processing, translation, and telomere maintenance (3–6).
Early studies have indicated that patients with cancer have elevated levels of serum and urinary pseudouridine (17, 18), suggesting a link between RNA pseudouridylation and cancer. Two recent studies have uncovered distinct roles for pseudouridine synthase 7 (PUS7)–dependent tRNA pseudouridylation in cancer (19, 20). In glioblastoma, PUS7 exhibits protumor activity: high PUS7 expression correlates with poor prognosis, and PUS7 promotes glioblastoma stem cell growth by repressing TYK2 mRNA translation (thereby blocking the interferon pathway) via tRNA pseudouridylation-regulated codon usage (19). In myelodysplastic syndrome, which can progress to acute myeloid leukemia, PUS7 has antitumor activity: low PUS7 expression is associated with acute myeloid leukemia progression, and PUS7-mediated pseudouridylation of tRNA fragments represses translation and enhances the differentiation of malignant myelodysplastic syndrome hematopoietic stem and progenitor cells (20). However, the functional significance and molecular regulation of mRNA pseudouridylation in cancer remain largely unexplored. We hypothesized that if mRNA pseudouridylation plays a role in cancer, the PUS family of enzymes responsible for this modification might be regulated at the level of expression or activity during cancer development. Our investigation revealed that the MYC family of oncoproteins, MYC and MYCN (21–23), transcriptionally activates a PUS7-dependent mRNA pseudouridylation program. This program sustains cancer cell proliferation and tumorigenesis by enhancing ATF4-mediated metabolic reprogramming and adaptive responses to mitigate cellular stresses associated with increased cell proliferation and biomass production.
Materials and Methods
Cell lines
All neuroblastoma cell lines and their culture conditions have been described in detail previously (24). The 293FT cell line (Thermo Fisher Scientific, R70007, RRID: CVCL_6911), 293T-ATF4 KO (25), 293T Lenti-X (TaKaRa 632180), HeLa (ATCC, CCL2, RRID: CVCL_0030), and SHEP MYCN-ER (26) were cultured in DMEM (HyClone, SH30022), and P493-6 (RRID: CVCL_6783; ref. 27) in RPMI 1640 (HyClone, SH30027). All media were supplemented with 10% FBS (Fisher Scientific, FB12999102). Cell lines from commercial sources and cell line repositories were authenticated using short tandem repeat profiling. Upon receipt, large frozen stocks were made to prevent cross-contamination with other cell lines. Neuroblastoma cell lines were regularly checked by immunoblotting and immunofluorescence for high-level nuclear expression of MYCN or the specific neuroblastoma marker PHOX2B through immunoblotting and immunofluorescence (28). All cell lines were used within 10 passages after reviving from frozen stocks and were free of Mycoplasma contamination as determined by DAPI staining regularly. Phase-contrast images were captured using an EVOS M5000 Imaging System (Invitrogen). Cell growth and survival were determined by trypan blue exclusion assay or crystal violet staining.
Xenograft models
NOD/SCID male and female 6-week-old mice (NOD.Cg-Prkdcscid/J, 001303, RRID: IMSR_JAX:001303) from The Jackson Laboratory were used in xenograft studies. BE(2)-C and SMS-KCNR cells expressing shGFP (control), shPUS7-33, shPUS7-34, pRetroX-Tight-Pur (tetoff, vector control), or pRetroX-Tight-Pur-PUS7 (tetoff-PUS7) were suspended in 100 μL Hanks Balanced Salt Solution (Thermo Fisher Scientific, 14170112) and injected subcutaneously into both sides of the flank (two sites per mouse) at ∼5 × 106 cells per injection site. Tumor volume was measured every other day using a digital caliper and estimated using the equation V = (L × W2)/2. Animals were euthanized when their tumors reached ∼1.0 cm in any diameter. The animal experiments were conducted at the Augusta University and approved by Institutional Animal Care and Use Committees of Medical College of Georgia, Augusta University.
Patient data
All survival and gene expression correlation analyses were conducted online using R2: Genomics Analysis and Visualization Platform (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi), and the resulting figures and P values were downloaded.
Overexpression and RNA interference
The Retro-X Tet-Off Advanced Inducible System (Clontech, 632106) was used for inducible PUS7 expression in the absence of doxycycline (Doxy) and the lentiviral vectors pCDH-CMV-MCS-EF1-puro (SBI System Biosciences, CD510B1) and pLV-EF1a-IRES-Hygro (RRID: Addgene, 85134) for constitutive gene overexpression. Human PUS7 coding sequence (CDS) in pCMV-SPORT6-PUS7 (Harvard Plasmids, HsCD00326998) was amplified by PCR, subcloned into pRetroX-Tight-puro, pCDH-CMV-MCS-EF1-pur, or pLV-EF1a-IRES-Hygro, and verified by sequencing. Site-direct mutagenesis was performed using the QuikChange II Site-Directed Mutagenesis Kit (Agilent, 200523) as instructed, with pCW57.1-ATF4 (variant 2, NM_182810; ref. 25), pGenLenti-ATF4 (variant 1, NM_001675, GenScript), or pCDH-myc-PUS7 as the template. Oligonucleotides for the mutagenesis are listed in Supplementary Table S1. Each mutation was confirmed by DNA sequencing. For cotransfection assays, 293T-ATF4 KO cells were cotransfected with the vector plasmid pCDH-CMV-MCS-EF1-pur or pCDH-MYCN mixed 1:1 with wild-type (WT) or mutant pGenLenti-ATF4 V1 or pCW57.1-ATF4 V2 for 24 hours before being collected for immunoblotting.
Lentiviral pLKO.1 shRNA constructs shGFP (RRID: Addgene, 12273) was obtained from Addgene and shMCTS1-25 (TRCN0000293825), shMCTS1-69 (TRCN0000293869), shMCTS1-83 (TRCN0000072283), shMYCN-94 (TRCN0000020694), shMYCN-95 (TRCN0000020695), shPUS7-33 (TRCN0000061933), and shPUS7-34 (TRCN0000061934) were from Sigma-Aldrich. For inducible expression of shRNA against PUS7, the shPUS7-33 sequence was cloned into the lentiviral Tet-pLKO-puro plasmid (RRID: Addgene, 21915).
For retrovirus and lentivirus production, gene expression or knockdown plasmids were mixed with the retroviral packaging plasmids pHDM-G and pMD.MLVogp (from R. Mulligan, Massachusetts Institute of Technology, Cambridge, MA) or with the lentiviral packaging plasmids pLP1 (RRID: Addgene, 22614), pLP2, and pLP/VSVG (Thermo Fisher Scientific, K497500) for transient transfection of 293FT cells (retrovirus) or 293T Lenti-X cells (lentivirus). Retroviral and lentiviral infection of cells was conducted according to standard procedures. For PUS7 knockdown rescue experiments, BE(2)-C cells were coinfected with lentiviruses expressing shGFP or shPUS7 along with vector (pLV-hygro or pCDH), pLV-PUS7, or pCDH-ATF4 lentiviruses. Three days after infection, viable cells were counted by trypan blue exclusion assay and collected for qRT-PCR and immunoblotting.
qRT-PCR
TRIzol (Thermo Fisher Scientific, 15596026) was used for isolation of total RNA from cells, an iScript Advanced cDNA Synthesis Kit (Bio-Rad 172-5038) for reverse transcription, and 2× SYBR Green qPCR master mix (Bimake, B21203) for qRT-PCR, which were performed on an iQ5 real-time PCR system (Bio-Rad) with primers against various genes (Supplementary Table S1). Data were normalized to β2 microglobulin (B2M) mRNA levels.
Immunoblotting
Cell lysates were prepared using standard SDS sample buffer, and protein concentrations were determined using a Bio-Rad Protein Assay Kit II. Proteins (20−50 μg) were separated on SDS-PAGE and transferred to nitrocellulose membranes, which were then probed with the following primary antibodies: rabbit anti-ASNS (1:1,000, Proteintech, 14681-1-AP, RRID: AB_2060119), mouse anti-ATF4 (1:400, Santa Cruz Biotechnology, sc390063, RRID: AB_2810998), rabbit anti-DENR (1:2,000, Proteintech, 10656-1-AP, RRID: AB_2092124), rabbit anti-DDIT3/CHOP (L63F6, 1:500, Cell Signaling Technology, 2895, RRID: AB_2089254), rabbit anti-GAPDH (FL335, 1:1,000, Santa Cruz Biotechnology, sc25778, RRID: AB_10167668), rabbit anti-GCN2 (1:2,000, Thermo Fisher Scientific, 720463, RRID: AB_2762413), rabbit anti-GCN2 phospho-T889 (1:1,000, Abcam, ab75836, RRID: AB_1310260), rabbit anti-MCTS1 (1:1,000, Thermo Fisher Scientific, PA5-31020, RRID: AB_2548494), rabbit anti-MYC (c-MYC, N262, 1:1,000, Santa Cruz Biotechnology, sc764, RRID: AB_631276), mouse anti-Myc tag (clone 4A6, 1:1,000, Millipore, 05724, RRID: AB_309938), mouse anti-MYCN (B8.4.B, 1:400, Santa Cruz Biotechnology, sc53993, RRID: AB_831602), rabbit anti-PHGDH (1:300, Sigma-Aldrich, HPA021241, RRID: AB_1855299), rabbit anti-PSAT1 (1:3,000, Novus Biologicals, 21020002, RRID: AB_2172599), rabbit anti-PUS7 (1:1,000, Thermo Fisher Scientific, PA5-54983, RRID: AB_2646152), rabbit anti-SLC7A5 (LAT1, 1:1,000, Cell Signaling Technology, 5347, RRID: AB_10695104), rabbit anti-SLC7A11/xCT (D2M7A, 1:1,000, Cell Signaling Technology, 12691, RRID: AB_2687474), rabbit anti-β-actin (1:2,000, Thermo Fisher Scientific, MA5-15739, RRID: AB_10979409), and mouse anti-α-tubulin (B-5-1-2, 1:5,000, Sigma-Aldrich, T5168, RRID: AB_477579). Horseradish peroxidase–conjugated goat anti-mouse (Jackson ImmunoResearch, 115-035-146, RRID: AB_2307392), and goat anti-rabbit IgG (Jackson ImmunoResearch, 111-035-046, RRID: AB_2337939) were used as secondary antibodies. Immunoblots were visualized using a Clarity Western ECL Substrate Kit (Bio-Rad 1705061) and quantified with Amersham ImageQuant 800 (Cytiva) or ImageJ (version 1.53k, RRID: SCR_003070).
ChIP-qPCR
Chromatin immunoprecipitation (ChIP) was performed as described (29) using ∼4 × 107 cells for each antibody. Cross-linked chromatin was sheared through sonication (Fisher Scientific, Model 150E) and immunoprecipitated using ChIP grade mouse anti-MYCN (B8.4.B, sc53993, RRID: AB_831602), mouse anti-ATF4 (sc390063, RRID: AB_2810998), or control mouse IgG (sc2025, RRID: AB_737182) from Santa Cruz Biotechnology. For qPCR, two independent ChIP samples were analyzed, and each sample was assayed in triplicate using the ChIP-qPCR primers listed in Supplementary Table S1. The number associated with each primer set indicates the position of the forward primer relative to its target gene transcription start site (+1).
Microarray
Total RNA was isolated using TRIzol from three biological replicates of BE(2)-C cells infected with lentiviruses expressing shGFP or shPUS7-33 for 6 days. Affymetrix microarray was performed using the Human Gene 2.0 ST microarray chip (Affymetrix). Data were normalized, significance determined by ANOVA, and fold change calculated with the Partek Genomics Suite (RRID: SCR_011860). Gene Ontology (GO) analysis by DAVID (RRID: SCR_001881) and gene set enrichment analysis (RRID: SCR_003199) were performed as described (30, 31). The NCBI Gene Expression Omnibus (RRID: SCR_005012) accession number for the microarray data reported in this article is GSE232058.
Stable isotope flux analysis
BE(2)-C–expressing shGFP or shPUS7-33 was cultured in glucose-deficient DMEM (Thermo Fisher Scientific, 11966025) supplemented with 10% dialyzed FBS and 25 mmol/L 13C6-glucose (Sigma-Aldrich, 389374-CONF) or in glutamine-deficient DMEM (Thermo Fisher Scientific, 11960044) supplemented with 10% dialyzed FBS and 2 mmol/L 13C5-15N2-glutamine (Sigma-Aldrich, 607983) for 24 hours. Cells were collected by scraping and centrifugation, and cell pellets were washed once with ice-cold PBS, snap-frozen in liquid nitrogen, and stored at −80°C until metabolite extraction. Six biological replicate samples (∼5 × 106 cells per sample) were analyzed for each cell line (shGFP or shPUS7). Metabolite extraction and deprivation, instrumentation, and data processing were performed as described previously (24).
Serine uptake assay
Serine uptake assay was performed as described previously (24). Briefly, (3H)-serine (11 Ci/mmol) was purchased from Moravek. Cells in 24-well plates were washed twice with transport buffer (25 mmol/L HEPES/Tris, pH 7.5; 140 mmol/L NaCl; 5.4 mmol/L KCl; 1.8 mmol/L CaCl2; 0.8 mmol/L MgSO4; and 5 mmol/L glucose). Transport of 5 µmol/L serine [0.05 µmol/L (3H)-serine plus 4.95 µmol/L unlabeled serine] was measured at 37°C for 5 and 10 minutes. Transport was terminated by aspiration of the transport buffer, followed by three washes with 2 mL of ice-cold transport buffer. The cells were lysed with 0.5 mL of 1% SDS in 0.2 N NaOH, and radioactivity was determined by scintillation counting.
PUS7 inhibitor
NSC107512-T/3 [PUS7 inhibitor (PUS7i)] was obtained from the NCI Developmental Therapeutics Program, dissolved in DMSO, and stored at −80°C. Cells were treated with PUS7i at indicated concentrations for various times and collected for cell survival (trypan blue exclusion assay), qRT-PCR, or immunoblot analyses.
Amino acid deprivation and ER stress responses
To assess the serine deprivation response, cells were cultured in MEM (Thermo Fisher Scientific, 11095080) supplemented with 10% dialyzed FBS (Thermo Fisher Scientific, 26400044), 4.5 g/L D-(+)-glucose (Sigma-Aldrich, G8769), 1× MEM vitamins (Thermo Fisher Scientific, 11120052), 1× sodium pyruvate (Thermo Fisher Scientific, 11360070), and nonessential amino acids alanine (A7479), asparagine (A4159), glutamic acid (G8415), glycine (50046), and proline (P5607) with or without 0.4 mmol/L serine (S4311) from Sigma-Aldrich. To assess the amino acid deprivation response (AAR) or ER stress, cells were cultured in DMEM treated with vehicle [H2O for histidinol (HisOH) control and DMSO for tunicamycin control], 5 mmol/L HisOH (Sigma-Aldrich, H6647), or 2 μg/mL tunicamycin (Sigma-Aldrich, T7765) for various times. Cells were then collected for qRT-PCR and immunoblot analyses.
Polysome profiling
Fractionation and collection of polysome-associated mRNAs were conducted essentially as described (32). Briefly, ∼1 × 107 control (shGFP) or PUS7 knockdown (shPUS7-33) BE(2)-C cells were treated with cycloheximide (Sigma-Aldrich, C7698) at 200 μg/mL for 10 minutes. The cells were collected, suspended in 500 μL polysome extraction buffer (20 mmol/L Tris-HCL, pH 7.5; 100 mmol/L KCl; 5 mmol/L MgCl2; and 0.5% NP40) supplemented with cycloheximide at 100 µg/mL, and lysed with Dounce homogenizer. The cytosolic extract was collected by centrifugation, overlaid on a 10% to 50% linear sucrose gradient, and centrifuged in a SW41Ti rotor (Beckman Coulter) at 35,000 rpm for 2 hours at 4°C. Then, 12 fractions were collected from the gradient using a density gradient fractionation system (Brandel, model BR186), and RNA was extracted from each fraction using TRIzol for qRT-PCR analysis. The qRT-PCR data were normalized to total RNA in the starting cytosolic extracts.
Proteomics
Proteomics analyses were carried out as previously described (33). Briefly, proteins were extracted using M-PER Mammalian Protein Extraction Reagent (Thermo Fisher Scientific, cat. #78501) and quantified using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, cat. #PI23225). Proteins of 40 µg per sample were reduced with DTT and denatured at 70°C for 35 minutes prior to loading onto 10% Bis-Tris protein gels for separation. The gels were stained overnight with colloidal Coomassie for protein visualization, and the entire gel lane representing each sample was cut into 6-MW fractions, equilibrated in 100 mmol/L ammonium bicarbonate, and digested overnight with Trypsin Gold, mass spectrometry grade (Promega, cat. #V5280) following the manufacturer’s instruction. Peptide extracts were reconstituted in 0.1% formic acid/ddH2O at 0.1 μg/μL. Mass spectrometry were carried out, and the data were processed, searched, filtered, grouped, and quantified by normalized spectral counting.
Nanopore direct RNA sequencing
Nanopore direct RNA sequencing (RNA-seq) of poly(A) RNA and ATF4 V1 RNA oligo with PUS7-mediated pseudouridylation was conducted as described (bioRxiv 2024.01.31.578250). MCTS1 pseudouridylation data were obtained from the reported nanopore sequencing datasets (bioRxiv 2024.01.31.578250). For ATF4 mRNA pseudouridylation study, poly(A) RNAs were isolated from control (shGFP) or PUS7 knockdown (shPUS7-33) BE(2)-C cells after treatment with 2 μg/mL tunicamycin for 5 hours, followed by nanopore sequencing and data analysis. ATF4 V1 DNA oligo design and in vitro transcription (to produce RNA oligos for in vitro pseudouridylation) were performed as described (34). The DNA oligo sequence is listed in Supplementary Table S1. For analysis of polysome-associated and input poly(A) RNA pseudouridylation, we downloaded the nanopore sequencing data from a recent study (35) and used high-confidence U-to-C basecalling errors to identify Ψ sites in polysome-associated and input (cytoplasmic) poly(A) RNA from HEK293T cells, as described previously (bioRxiv 2024.01.31.578250; ref. 14).
Statistics
Quantitative data are presented as mean ± SD and were analyzed for statistical significance using unpaired, two-tailed Student t test, Fisher exact test, and one-way or two-way ANOVA. For animal studies, the log-rank test was used to account for mouse survival by the end of animal experiments. Unless otherwise stated, all statistical analyses were conducted using GraphPad Prism 9.4.0 for Mac (RRID: SCR_002798).
Data availability
Microarray data reported in this study have been deposited in the Gene Expression Omnibus under the accession number GSE232058. The BioProject accession number for the nanopore poly-A RNA-seq data reported in this article is PRJNA961708. The patient data analyzed in this study were obtained from R2 Genomics Analysis and Visualization Platform (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi). All other raw data are available upon request from the corresponding author.
Results
PUS7 is a direct transcriptional target of the MYC family of oncoproteins
Genomic amplification of MYCN is a major cause of high-risk neuroblastoma (36). Microarray gene expression profiling of non–MYCN-amplified neuroblastoma SK-N-AS cells with MYCN overexpression (37) showed a significant increase (>1.5 fold) in mRNA expression of the PUS family members DKC1, PUS1, PUS7, and RPUSD4, with PUS7 being the most upregulated (Fig. 1A; Supplementary Fig. S1A). We performed qRT-PCR and immunoblotting to confirm this finding in non–MYCN-amplified neuroblastoma cell lines with either inducible (AS) or constitutive (SHEP1) MYCN overexpression. In both cell lines, MYCN overexpression markedly increased PUS7 mRNA and protein expression but had no consistent effect on PUS10 mRNA expression (Fig. 1B and C). Conversely, knockdown of MYCN expression in MYCN-amplified neuroblastoma cell lines by two independent short hairpin RNA (shRNA) constructs reduced PUS7 expression at both mRNA and protein levels (Fig. 1D and E). In agreement with these findings, MYCN-amplified neuroblastoma cell lines expressed significantly higher levels of PUS7 protein relative to non–MYCN-amplified cell lines (Supplementary Fig. S1B and S1C).
PUS7 is a direct transcriptional target of MYC oncoproteins. A, Microarray data show mRNA expression of PUS family genes in non–MYCN-amplified neuroblastoma SK-N-AS cells with MYCN overexpression compared with vector control cells (dashed line). B and C, qRT-PCR (B) and immunoblotting (C) show MYCN-mediated upregulation of PUS7 mRNA and protein in non–MYCN-amplified neuroblastoma SK-N-AS cells with inducible MYCN overexpression in the absence of Doxy and SHEP1 cells with constitutive MYCN overexpression. PUS7 protein levels were quantified against α-tubulin (α-tub). D and E, qRT-PCR (D) and immunoblotting (E) show downregulation of PUS7 mRNA and protein expression by shRNA-mediated MYCN knockdown in MYCN-amplified neuroblastoma cell lines. PUS7 protein levels were quantified against GAPDH. F, Immunoblotting show MYC upregulation of PUS7 in P493-6 cells with inducible MYC expression in the absence of Doxy (tetoff-MYC). PUS7 protein levels were quantified against β-actin. G, ChIP-qPCR show endogenous MYCN binding to the promoter and first intron of PUS7 in MYCN-amplified neuroblastoma BE(2)-C cells. The MDM2 and DDIT3 promoters were used as positive control for MYCN and ATF4 binding, respectively. The number associated with each primer set indicates the position of the forward primer relative to its target gene transcription start site (+1). H, Violin plot shows higher PUS7 expression in MYCN-amplified neuroblastoma tumors [the Sequencing Quality Control (SEQC) cohort, n = 498]. P values were calculated using one-way ANOVA. I, RNA-seq data show increased Pus7 mRNA expression in B cells during lymphoma development in Eµ-myc mice. PCR data (B, D, and G) are presented as mean ± SD (B and D, n = 4; G, n = 3) and analyzed using two-way ANOVA. **, P < 0.01; ***, P < 0.001.
PUS7 is a direct transcriptional target of MYC oncoproteins. A, Microarray data show mRNA expression of PUS family genes in non–MYCN-amplified neuroblastoma SK-N-AS cells with MYCN overexpression compared with vector control cells (dashed line). B and C, qRT-PCR (B) and immunoblotting (C) show MYCN-mediated upregulation of PUS7 mRNA and protein in non–MYCN-amplified neuroblastoma SK-N-AS cells with inducible MYCN overexpression in the absence of Doxy and SHEP1 cells with constitutive MYCN overexpression. PUS7 protein levels were quantified against α-tubulin (α-tub). D and E, qRT-PCR (D) and immunoblotting (E) show downregulation of PUS7 mRNA and protein expression by shRNA-mediated MYCN knockdown in MYCN-amplified neuroblastoma cell lines. PUS7 protein levels were quantified against GAPDH. F, Immunoblotting show MYC upregulation of PUS7 in P493-6 cells with inducible MYC expression in the absence of Doxy (tetoff-MYC). PUS7 protein levels were quantified against β-actin. G, ChIP-qPCR show endogenous MYCN binding to the promoter and first intron of PUS7 in MYCN-amplified neuroblastoma BE(2)-C cells. The MDM2 and DDIT3 promoters were used as positive control for MYCN and ATF4 binding, respectively. The number associated with each primer set indicates the position of the forward primer relative to its target gene transcription start site (+1). H, Violin plot shows higher PUS7 expression in MYCN-amplified neuroblastoma tumors [the Sequencing Quality Control (SEQC) cohort, n = 498]. P values were calculated using one-way ANOVA. I, RNA-seq data show increased Pus7 mRNA expression in B cells during lymphoma development in Eµ-myc mice. PCR data (B, D, and G) are presented as mean ± SD (B and D, n = 4; G, n = 3) and analyzed using two-way ANOVA. **, P < 0.01; ***, P < 0.001.
We also investigated whether PUS7 is a transcriptional target of MYC, another member of the MYC family. We used the Burkitt lymphoma model cell line P493-6 with inducible MYC expression in the absence of Doxy (tetoff-MYC; ref. 27). Silencing MYC expression by the addition of Doxy markedly reduced PUS7 expression, which was restored upon re-induction of MYC in the absence of Doxy (Fig. 1F).
We then asked whether MYC and MYCN directly targeted PUS7 for transcriptional upregulation. Inspection of PUS7 genomic sequence (NM_001318163) revealed the canonical MYC-binding E-box sequence CACGTG in the proximal promoter region (+16, +1 being the transcription start site) and the first intron (+663 and +2,357). Analysis of ENCODE data on ChIP-seq of transcription factors (38) showed specific binding of MYC and its partner MAX to the PUS7 promoter (Supplementary Fig. S1D). We also performed ChIP-qPCR assays using MYCN-amplified neuroblastoma cell lines BE(2)-C and SMS-KCNR, which showed the binding of endogenous MYCN to the E-box sequences within the PUS7 promoter and first intron, as well as to the promoter of MDM2 (Fig. 1G; Supplementary Fig. S1E), a known MYCN target gene (39). MYCN knockdown significantly reduced the endogenous MYCN levels at the promoter and first intron of PUS7 (Supplementary Fig. S1F). We found no significant levels of endogenous MYCN at the promoter of the stress response gene DDIT3 (also known as CHOP). By contrast, endogenous ATF4, a transcriptional activator of DDIT3 (40), is associated with the promoter of DDIT3 but not with that of PUS7 (Fig. 1G). Together, these data indicate that PUS7 is a direct transcriptional target gene of the MYC family of oncoproteins.
High PUS7 expression is associated with MYC-driven cancers and poor prognosis
The above findings prompted us to assess the clinical relevance of PUS7 upregulation by the MYC family of oncoproteins. We examined the gene expression profiling data from independent cohorts of patients with neuroblastoma (41–43) and found a strong positive correlation in mRNA expression between PUS7 and MYCN in neuroblastoma specimens (Supplementary Fig. S2A). Further analysis revealed that PUS7 mRNA expression is specifically upregulated in MYCN-amplified high-risk neuroblastoma tumors (Fig. 1H). Interestingly, there is no significant difference in PUS7 mRNA levels between low-risk tumors and high-risk tumors without MYCN amplification (Fig. 1H), suggesting a specific function of PUS7 in MYCN-amplified neuroblastoma. Moreover, higher PUS7 mRNA expression is significantly associated with advanced stages of neuroblastoma and poor prognosis in patients with neuroblastoma (Supplementary Fig. S2B and S2C). By contrast, PUS10 mRNA expression is negatively correlated with MYCN expression and is not associated with patient prognosis (Supplementary Fig. S2D and S2E).
For MYC, our analysis of published data (44) revealed a stagewise increase in Pus7 mRNA expression during B-cell lymphomagenesis and progression in Eµ-myc mice (Fig. 1I). Furthermore, gene expression profiling data from the diffuse large B-cell lymphoma dataset (45) showed a positive correlation in mRNA expression between MYC and PUS7 (Supplementary Fig. S2F), and higher PUS7 mRNA expression is associated with worse prognosis in these patients (Supplementary Fig. S2G).
Collectively, these data provide genetic evidence suggesting a functional role of PUS7 in the development and progression of MYC-driven cancers.
High PUS7 expression promotes the proliferation and tumorigenicity of MYC-driven cancer cells
We next investigated the functional significance of high PUS7 expression. We generated MYCN-amplified neuroblastoma cell lines with inducible PUS7 expression in the absence of Doxy (Fig. 2A, Doxy−). Compared with parental (tetoff) and tetoff-PUS7 cell lines cultured in the presence of Doxy (no PUS7 induction), PUS7 upregulation increased cell proliferation (Fig. 2B), as evidenced by a significant reduction in the cell population doubling time (Fig. 2C). In addition, PUS7 overexpression markedly accelerated the growth of neuroblastoma xenografts and the death of tumor-bearing mice (Fig. 2D and E). We also silenced PUS7 expression using two distinct shRNA constructs (Fig. 2F). Knockdown of PUS7 expression induced G1 arrest (Supplementary Fig. S2H) and inhibited the growth of MYCN-amplified neuroblastoma cell lines in culture (Fig. 2G) and in vivo (Fig. 2H; Supplementary Fig. S2I) and prolonged the survival of tumor-bearing mice (Fig. 2I; Supplementary Fig. S2J).
High PUS7 expression promotes neuroblastoma cell proliferation and tumorigenicity. A, Immunoblotting of PUS7 in neuroblastoma SK-N-DZ cells, with inducible PUS7 expression in the absence of Doxy (Doxy−). B, PUS7 overexpression (Doxy−) promoted the proliferation of BE(2)-C and SK-N-DZ neuroblastoma cells. The same cell lines cultured in the presence of Doxy (Doxy+, no PUS7 induction) and their parental tetoff cell lines were used as control. C, PUS7 overexpression reduced the cell population doubling time based on the data from B. D and E, PUS7 overexpression accelerated BE(2)-C xenograft growth (D) and decreased event-free survival (E) of xenograft-bearing NOD/SCID mice. Log-rank test P value is indicated. F, Immunoblotting of PUS7 in BE(2)-C expressing shGFP or shPUS7. G, PUS7 knockdown inhibited the proliferation of neuroblastoma cell lines. H and I, PUS7 knockdown impeded BE(2)-C xenograft growth (H) and prolonged event-free survival (I) of xenograft-bearing NOD/SCID mice. Log-rank test P value is indicated. J, Diagram illustrates shPUS7-33 and shPUS7-34 targeting the PUS7 3′-UTR and CDS, respectively. The myc-PUS7 construct lacks the PUS7 3′-UTR, making it resistant to downregulation by shPUS7-33. K, Immunoblotting of PUS7 in BE(2)-C cells expressing shGFP, shPUS7-33, or shPUS7-34 that were then infected with vector or myc-PUS7 lentiviruses. L, Cell growth assay shows PUS7 expression abrogated the growth inhibitory effect of shPUS7-33 but not that of shPUS7-34. All cell growth data (B, G, and L) are presented as mean ± SD (n = 4) and quantitative data (B, D, G, H, and L) were analyzed using two-way ANOVA. ***, P < 0.001.
High PUS7 expression promotes neuroblastoma cell proliferation and tumorigenicity. A, Immunoblotting of PUS7 in neuroblastoma SK-N-DZ cells, with inducible PUS7 expression in the absence of Doxy (Doxy−). B, PUS7 overexpression (Doxy−) promoted the proliferation of BE(2)-C and SK-N-DZ neuroblastoma cells. The same cell lines cultured in the presence of Doxy (Doxy+, no PUS7 induction) and their parental tetoff cell lines were used as control. C, PUS7 overexpression reduced the cell population doubling time based on the data from B. D and E, PUS7 overexpression accelerated BE(2)-C xenograft growth (D) and decreased event-free survival (E) of xenograft-bearing NOD/SCID mice. Log-rank test P value is indicated. F, Immunoblotting of PUS7 in BE(2)-C expressing shGFP or shPUS7. G, PUS7 knockdown inhibited the proliferation of neuroblastoma cell lines. H and I, PUS7 knockdown impeded BE(2)-C xenograft growth (H) and prolonged event-free survival (I) of xenograft-bearing NOD/SCID mice. Log-rank test P value is indicated. J, Diagram illustrates shPUS7-33 and shPUS7-34 targeting the PUS7 3′-UTR and CDS, respectively. The myc-PUS7 construct lacks the PUS7 3′-UTR, making it resistant to downregulation by shPUS7-33. K, Immunoblotting of PUS7 in BE(2)-C cells expressing shGFP, shPUS7-33, or shPUS7-34 that were then infected with vector or myc-PUS7 lentiviruses. L, Cell growth assay shows PUS7 expression abrogated the growth inhibitory effect of shPUS7-33 but not that of shPUS7-34. All cell growth data (B, G, and L) are presented as mean ± SD (n = 4) and quantitative data (B, D, G, H, and L) were analyzed using two-way ANOVA. ***, P < 0.001.
We conducted rescue experiments to confirm the target specificity of the PUS7 shRNAs using a lentiviral construct expressing only the PUS7 CDS with an N-terminal myc tag. The shRNA construct shPUS7-33 targets the 3′ untranslated region (UTR) of PUS7, while shPUS7-34 targets its CDS (Fig. 2J). Because the myc-PUS7 construct lacks the PUS7 3′-UTR (Fig. 2J), it should be resistant to downregulation by shPUS7-33. As expected, shPUS7-33 did not reduce the expression of myc-PUS7, whereas shPUS7-34 effectively silenced myc-PUS7 expression (Fig. 2K). Overexpressing myc-PUS7 completely reversed the growth inhibition induced by shPUS7-33 but did not significantly affect the growth inhibition caused by shPUS7-34 (Fig. 2L). Thus, at least for shPUS7-33, its growth inhibitory effect is mediated by the suppression of PUS7 expression.
Together, these findings demonstrate that high PUS7 expression is required for the proliferation and tumorigenicity of MYCN-amplified neuroblastoma cell lines.
PUS7 is required for the expression of genes involved in adaptive stress responses and amino acid metabolism
To gain a molecular understanding for the growth-promoting activity of PUS7, we performed microarray gene expression profiling of the MYCN-amplified neuroblastoma BE(2)-C cells with PUS7 knockdown by shPUS7-33. A total of 1081 PUS7 knockdown-responsive genes (≥ ±1.50-fold, P < 0.05) were identified, with 682 genes being upregulated and 399 genes downregulated (Supplementary Table S2). GO analysis revealed that PUS7 silencing led to upregulation of genes that are highly enriched for GO terms associated with DNA replication and repair, G1 to S phase transition, and cholesterol biosynthesis (Supplementary Fig. S3A; Supplementary Table S3). These findings suggest that PUS7 knockdown may trigger DNA damage and cholesterol depletion responses. Alternatively, PUS7 may have an inhibitory effect on mRNA expression of genes involved in DNA replication and repair and cholesterol synthesis. These possibilities are currently under investigation.
The genes downregulated by PUS7 silencing are primarily involved in the response to endoplasmic reticulum (ER) stress, tRNA aminoacylation, serine biosynthesis, one-carbon metabolism, and amino acid synthesis and transport (Fig. 3A and B; Supplementary Table S3). Gene set enrichment analysis confirmed the finding (Supplementary Fig. S3B) and, additionally, showed that PUS7 knockdown reduced mRNA expression of genes involved in the AAR (Fig. 3C), which promotes amino acid biosynthesis and imports in response to reduced amino acid levels in the cells (46, 47). The microarray data are consistent with the gene expression profiling data from neuroblastoma tumors, showing co-upregulation of PUS7, ATF4, and amino acid synthesis enzymes and transporters in MYCN-amplified neuroblastoma (Fig. 3D). Similarly, neuroblastoma cell lines with MYCN amplification exhibited higher levels of ATF4 protein expression than non–MYCN-amplified neuroblastoma cell lines (Supplementary Fig. S3C). It should be noted that the human ATF4 gene produces two main mRNA variants: V1 (NM_001675.4) and V2 (NM_182810.3). Although both variants encode the same protein, they differ in their 5′-UTR, with the V2 transcript missing an internal segment that is present in the V1 transcript (Supplementary Fig. S3D). RT-PCR analysis using primers specific to V1 and V2 revealed that MYCN-amplified neuroblastoma cell lines expressed either similar levels of V1 and V2 [BE(2)-C and SMS-KCNR], higher levels of V1 (LA1-55n), or higher levels of V2 (IMR5), whereas non–MYCN-amplified cell lines SHEP1 and SK-N-AS expressed significantly higher levels of the V2 variant (Supplementary Fig. S3E).
PUS7 sustains the expression of genes involved in stress responses and amino acid metabolism. A, GO analysis of microarray data shows the top biological processes for genes downregulated by PUS7 knockdown. B, Volcano plot shows the representative genes involved in stress responses and amino acid metabolism. C, Gene set enrichment analysis of microarray data shows the downregulation of genes involved in AAR by PUS7 knockdown. D, Volcano plot shows the co-upregulation of PUS7 and representative genes involved in amino acid biosynthesis and transport. E and F, qRT-PCR (E) and immunoblotting (F) confirm the downregulation of mRNA and protein expression for representative genes involved in stress responses and amino acid metabolism. qRT-PCR data (E) are presented as mean ± SD (n = 4) and analyzed using two-way ANOVA. ***, P < 0.001.
PUS7 sustains the expression of genes involved in stress responses and amino acid metabolism. A, GO analysis of microarray data shows the top biological processes for genes downregulated by PUS7 knockdown. B, Volcano plot shows the representative genes involved in stress responses and amino acid metabolism. C, Gene set enrichment analysis of microarray data shows the downregulation of genes involved in AAR by PUS7 knockdown. D, Volcano plot shows the co-upregulation of PUS7 and representative genes involved in amino acid biosynthesis and transport. E and F, qRT-PCR (E) and immunoblotting (F) confirm the downregulation of mRNA and protein expression for representative genes involved in stress responses and amino acid metabolism. qRT-PCR data (E) are presented as mean ± SD (n = 4) and analyzed using two-way ANOVA. ***, P < 0.001.
We validated the microarray data by qRT-PCR (Fig. 3E) and immunoblotting (Fig. 3F; Supplementary Fig. S3F and S3G), showing that PUS7 knockdown reduced the expression of ATF4 and genes involved in stress responses, amino acid synthesis and transport, and tRNA aminoacylation but had no effect on the expression of enzymes responsible for nucleotide synthesis, such as CAD and PPAT (Fig. 3E). Conversely, PUS7 overexpression increased the expression of ATF4 and enzymes for amino acid biosynthesis (Supplementary Fig. S3H and S3I). Furthermore, we examined the catalytically inactive mutant PUS7 D294A in which the aspartate residue at the position 294 is replaced by alanine (19, 48). Relative to WT PUS7, PUS7 D294A showed a reduced capacity to upregulate ATF4 and amino acid synthesis enzymes (Supplementary Fig. S3H and S3I), indicating that the ability of PUS7 to upregulate its target genes is dependent on its enzyme activity.
Collectively, these data suggest a functional role for PUS7 in maintaining the expression of genes involved in adaptive stress responses and amino acid metabolism.
PUS7 sustains amino acid biosynthesis and transport
We next assessed the functional significance of gene expression data. The top amino acid synthesis pathway regulated by PUS7 is the serine–glycine biosynthesis pathway (Fig. 3A). PUS7 knockdown decreased the expression of the pathway enzymes PHGDH, PSAT1, PSPH, and SHMT2 (Fig. 3; Supplementary Tables S2 and S3), whereas PUS7 overexpression increased PHGDH and PSAT1 expression (Supplementary Fig. S3I). These enzymes use the glycolytic intermediate 3-phosphoglycerate to produce serine and glycine (Fig. 4A). We performed stable isotope tracing experiments using uniformly labeled 13C6-glucose to assess the impact of PUS7 knockdown on serine and glycine synthesis from glucose (Fig. 4A). Knockdown of PUS7 expression markedly reduced the production of the m+3 isotopolog of serine (13C3-serine) and the m+2 isotopolog of glycine (13C2-glycine) from the glucose tracer (Fig. 4A and B; Supplementary Table S4). By contrast, we found no significant effect of PUS7 knockdown on the synthesis of m+3 alanine from m+3 pyruvate (Fig. 4A and B; Supplementary Table S4), which is catalyzed by the cytosol enzyme GPT (Fig. 4A), and on the production of various 13C isotopologs of aspartate from 13C-oxaloacetic acid, which is catalyzed by the mitochondrial enzyme GOT2 (Supplementary Fig. S4A and S4B). It is well established that cellular aspartate is primarily derived from oxaloacetic acid amination (49). The 13C6-glucose tracing data are consistent with the gene expression data in that PUS7 knockdown reduced serine synthesis enzymes but had no significant impact on GPT and GOT2 mRNA expression (Supplementary Table S2). Thus, PUS7 has a key role in sustaining the biosynthetic activity of the serine–glycine synthesis pathway by maintaining the pathway enzyme expression.
PUS7 is required for sustaining amino acid biosynthesis and transport and stress responses. A, Diagram for incorporation of 13C from glucose into serine, glycine, and alanine. Fold changes (numbers in parentheses) in mRNA expression of relevant enzymes were from microarray profiling. B, PUS7 knockdown reduced 13C6-glucose flux into serine and glycine production but had no effect on alanine production from 13C6-glucose. C, Labeling patterns of metabolites derived from 13C5-15N2-glutamine via anaplerosis. Fold changes (numbers in parentheses) in mRNA expression of relevant enzymes were from microarray profiling. D, PUS7 knockdown reduced incorporation of 13C and 15N from glutamine into asparagine. Data (B and D) are presented as mean ± SD (n = 6) and analyzed using two-tailed Student t test. ***, P < 0.001. E,3H-serine uptake assays show PUS7 regulation of serine transport. Data are presented as mean ± SD of three biological replicates and analyzed using two-way ANOVA. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. F and G, Immunoblotting shows that PUS7 knockdown diminished stress-induced upregulation of ATP4 protein and its downstream targets involved in stress responses and amino acid synthesis and transport. HisOH, AAR inducer; tunicamycin, ER stress inducer. H, qRT-PCR shows PUS7 knockdown diminished ER stress–induced upregulation of ATP4 and its target genes. Data are presented as mean ± SD (n = 4) and analyzed using two-way ANOVA. ***, P < 0.001.
PUS7 is required for sustaining amino acid biosynthesis and transport and stress responses. A, Diagram for incorporation of 13C from glucose into serine, glycine, and alanine. Fold changes (numbers in parentheses) in mRNA expression of relevant enzymes were from microarray profiling. B, PUS7 knockdown reduced 13C6-glucose flux into serine and glycine production but had no effect on alanine production from 13C6-glucose. C, Labeling patterns of metabolites derived from 13C5-15N2-glutamine via anaplerosis. Fold changes (numbers in parentheses) in mRNA expression of relevant enzymes were from microarray profiling. D, PUS7 knockdown reduced incorporation of 13C and 15N from glutamine into asparagine. Data (B and D) are presented as mean ± SD (n = 6) and analyzed using two-tailed Student t test. ***, P < 0.001. E,3H-serine uptake assays show PUS7 regulation of serine transport. Data are presented as mean ± SD of three biological replicates and analyzed using two-way ANOVA. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. F and G, Immunoblotting shows that PUS7 knockdown diminished stress-induced upregulation of ATP4 protein and its downstream targets involved in stress responses and amino acid synthesis and transport. HisOH, AAR inducer; tunicamycin, ER stress inducer. H, qRT-PCR shows PUS7 knockdown diminished ER stress–induced upregulation of ATP4 and its target genes. Data are presented as mean ± SD (n = 4) and analyzed using two-way ANOVA. ***, P < 0.001.
To further assess the function of PUS7 in amino acid biosynthesis, we conducted stable isotope tracing experiments using 13C5-15N2-glutamine because glutamine provides carbon and nitrogen for the biosynthesis of several nonessential amino acids (50, 51). In agreement with the 13C6-glucose tracing study discussed above, PUS7 knockdown had no effect on the production of various isotopologs of aspartate from 13C5-15N2-glutamine via the tricarboxylic acid cycle (Supplementary Fig. S4C; Supplementary Table S5). However, PUS7 knockdown markedly reduced the production of asparagine isotopologs from 13C-15N-labeled aspartate (Fig. 4C and D; Supplementary Table S5). Again, this finding is consistent with our gene expression data. PUS7 knockdown markedly decreased the expression ASNS (Fig. 3; Supplementary Table S2), which catalyzes the conversion of aspartate to asparagine (Fig. 4C).
Our gene expression profiling analysis also showed that PUS7 knockdown reduced mRNA expression of genes coding for amino acid transporters (Fig. 3; Supplementary Table S2), including the alanine–serine–cysteine transporters ASCT1 and ASCT2 (also known as SLC1A4 and SLC1A5, respectively; ref. 52). Therefore, we conducted [3H]- serine uptake assay. Consistent with the gene expression data, PUS7 overexpression or knockdown significantly increased or reduced serine uptake, respectively (Fig. 4E).
Collectively, these data provide evidence for a key role of PUS7 in maintaining nonessential amino acid biosynthesis and transport.
PUS7 sustains adaptive stress responses
The PUS7 knockdown gene expression data also led us to explore the functional role of PUS7 in cellular stress responses (Fig. 3A), using the model systems of ER stress and the AAR. We treated control (shGFP) and PUS7 knockdown BE(2)-C and HeLa cells with tunicamycin (ER stress inducer) or HisOH (AAR inducer; refs. 53, 54). In control cells, treatment with HisOH or tunicamycin induced ATF4 protein expression, and tunicamycin also induced the stress protein CHOP (Fig. 4F). The induction was markedly diminished in PUS7 knockdown cells (Fig. 4F). We further examined other known AAR genes downstream of ATF4, including ASNS, SLC7A5, and SLC7A11, and their protein induction was likewise reduced by PUS7 knockdown (Fig. 4G). Moreover, we found that PUS7 knockdown inhibited ATF4 protein induction by serine deprivation (Supplementary Fig. S4D). We obtained similar results with qRT-PCR analyses of the stress-induced, ATF4-dependent mRNA expression program (Fig. 4H; Supplementary Fig. S4E and S4F).
The diminished stress response in PUS7 knockdown cells was not a result of defects in upstream stress signaling, because HisOH or serine starvation induced robust GCN2 phosphorylation in these cells (Fig. 4F; Supplementary Fig. S4D), which occurs upstream of ATF4 induction during the AAR (55). Interestingly, in all the cell lines examined, PUS7 knockdown cells showed higher basal levels of GCN2 phosphorylation than control shGFP cells under standard culture conditions (Fig. 4F, untreated; Supplementary Fig. S4D, time 0; shGFP vs. shPUS7). The observed constitutive GCN2 phosphorylation suggests that cells with PUS7 knockdown are constantly under the stress of amino acid starvation or have increased levels of uncharged tRNAs, which triggers GCN2 phosphorylation (55). This notion is supported by our gene expression data, showing an essential role of PUS7 in maintaining the expression of enzymes for tRNA aminoacylation and amino acid biosynthesis (Fig. 3).
Together, these findings reveal the role of PUS7 in driving the cellular responses to AAR and ER stress.
PUS7 protects MYC-driven cancer cells from MYC-induced cellular stress by enabling ATF4-mediated adaptive responses
It is increasingly recognized that the high rate of cell proliferation induced by MYC oncoproteins renders cells dependent on adaptive responses to mitigate cellular stress caused by increased biomass production (24, 56–58), including the unfolded protein response as a result of ER protein overload (59) and the AAR caused by depletion of amino acids (60, 61). These stress signals activate ATF4 to initiate a transcription program that drives cellular adaptive responses (60, 62, 63). Indeed, overexpression of MYCN or MYC increased the protein levels of ATF4 and its targets, including amino acid synthesis enzymes (e.g., PHGDH and PSAT1) and transporters (e.g., SLC7A5; Fig. 5A–C). The upregulation was largely abrogated by PUS7 knockdown (Fig. 5A, shPUS7-33) or by PUS7 inhibition with the compound NSC107512-T/3 (C17; Fig. 5B and C, PUS7i; ref. 19). Similarly, PUS7 knockdown or inhibition significantly reduced the ability of MYCN to upregulate mRNA expression of PHGDH, PSAT1, and SLC7A5 (Supplementary Fig. S5A and S5B). Thus, PUS7 is required for MYC-induced adaptive responses.
PUS7 protects cells from MYC-induced cellular stress by enabling ATF4-mediated adaptive responses. A–C, Immunoblot analysis shows PUS7 knockdown (A) or inhibition (B and C) abrogated MYCN- or MYC-induced ATF4 and proteins for amino acid synthesis and transport. Protein levels were quantified against α-tubulin (A and B) or β-actin (C). D and E, MYCN activation by 4-hydroxytamoxifen (4OHT) promoted cell proliferation but sensitized cells to PUS7 knockdown (D) or inhibition (E). Data are presented as mean ± SD (n = 4) and analyzed using two-way ANOVA. F, Immunoblot analysis of autophagy (LC3B-II) in SHEP MYCN-ER cells without or with MYCN activation (4OHT) and PUS7 knockdown. G, ATF4 overexpression alleviated the growth inhibitory effect of PUS7 knockdown on neuroblastoma cell lines. H, qRT-PCR shows ATF4 overexpression abrogated PUS7 knockdown–mediated repression of genes for amino acid synthesis and transport. Data are presented as mean ± SD (n = 4) and analyzed using two-way ANOVA. ***, P < 0.001; ****, P < 0.0001.
PUS7 protects cells from MYC-induced cellular stress by enabling ATF4-mediated adaptive responses. A–C, Immunoblot analysis shows PUS7 knockdown (A) or inhibition (B and C) abrogated MYCN- or MYC-induced ATF4 and proteins for amino acid synthesis and transport. Protein levels were quantified against α-tubulin (A and B) or β-actin (C). D and E, MYCN activation by 4-hydroxytamoxifen (4OHT) promoted cell proliferation but sensitized cells to PUS7 knockdown (D) or inhibition (E). Data are presented as mean ± SD (n = 4) and analyzed using two-way ANOVA. F, Immunoblot analysis of autophagy (LC3B-II) in SHEP MYCN-ER cells without or with MYCN activation (4OHT) and PUS7 knockdown. G, ATF4 overexpression alleviated the growth inhibitory effect of PUS7 knockdown on neuroblastoma cell lines. H, qRT-PCR shows ATF4 overexpression abrogated PUS7 knockdown–mediated repression of genes for amino acid synthesis and transport. Data are presented as mean ± SD (n = 4) and analyzed using two-way ANOVA. ***, P < 0.001; ****, P < 0.0001.
Next, we assessed the impact of PUS7 silencing or inhibition on cell proliferation and survival after activation of MYC proteins. SHEP MYCN-ER neuroblastoma cells express an MYCN protein fused to the α-domain of the estrogen receptor (ER), which allows conditional activation of MYCN by 4-hydroxytamoxifen (26). MYCN activation increased the proliferation of shGFP control cells but markedly reduced the survival of PUS7 knockdown cells (Fig. 5D). Similarly, MYCN activation significantly sensitized cells to PUS7i (Supplementary Fig. S5C), leading to a 13-fold reduction in IC50 from 155 to 12 nmol/L (Fig. 5E). Consistent with these observations, MYCN activation induced higher levels of autophagy in PUS7 knockdown cells, as evidenced by a marked increase in the levels of LC3B-II (Fig. 5F), a quantitative marker for autophagic activity (64). In addition, MYCN-amplified neuroblastoma cell lines exhibited significantly higher sensitivity to PUS7 inhibition than non–MYCN-amplified cell lines, with an average 15-fold decrease in IC50 (Supplementary Fig. S5D and S5E). These findings suggest an essential role of PUS7 in supporting cell survival and proliferation after MYCN activation.
We further identified ATF4 as an essential mediator of this protective function of PUS7. ATF4 overexpression rescued the growth inhibition caused by PUS7 knockdown in MYCN-amplified neuroblastoma cells (Fig. 5G; Supplementary Fig. S5F) and abrogated the inhibitory effect of PUS7 knockdown on the expression of genes involved in amino acid synthesis and transport, such as ASNS and SLC7A11 (Fig. 5H).
The data presented above collectively suggest that a key function of PUS7 in MYC-driven cancer is to enhance ATF4 expression, thereby mitigating MYC-induced cellular stress. This implies that inhibiting PUS7 could be a strategy to selectively kill MYC-driven cancer cells.
PUS7 enhances ATF4 protein expression via distinct mechanisms
We next investigated the molecular basis for PUS7 to enhance ATF4 expression. Given that ATF4 expression is regulated primarily at the translation level (46, 65), we performed polysome analysis to assess the translational activity of ATF4 mRNA, as actively translating mRNAs are associated with higher numbers of ribosomes (polysomes; ref. 66). Cell lysates from control or PUS7 knockdown BE(2)-C cells were centrifugated through a sucrose gradient, followed by a collection of fractions that contain ribosomal subunits, monosomes, or polysomes (from the top to the bottom of the gradient). RNAs from each fraction were extracted and analyzed by qRT-PCR. PUS7 knockdown markedly reduced the levels of polysome-associated ATF4 mRNA but had no significant effect on the levels of polysome-associated GAPDH mRNA (Supplementary Fig. S6A and S6B). This observation suggests a key role of PUS7 in promoting ATF4 mRNA translation. Further investigation revealed distinct mechanisms for this function of PUS7.
MCTS1 is a downstream target of PUS7 required for efficient ATF4 mRNA translation
Two independent lines of evidence prompted us to investigate malignant T-cell amplified sequence 1 (MCTS1), a noncanonical translation initiation factor that forms a complex with density-regulated protein (DENR) to promote ATF4 mRNA translation in response to ER stress and AAR (67, 68). First, we performed quantitative profiling of PUS7-dependent Ψ sites in poly(A) RNAs (mRNAs and long noncoding RNAs) from BE(2)-C cells by nanopore direct RNA-seq (bioRxiv 2024.01.31.578250) that detects Ψ modifications by identifying high-confidence U-to-C basecalling errors (14, 16, 69). Using the stringent 10% U-to-C basecalling error threshold (P < 0.05), we identified five Ψ sites at the base positions 107, 339, 371, 674, and 713 in the MCTS1 transcript ENST00000371317.10 (MCTS1-202, corresponding to MCTS1 mRNA NM_014060.3) with Ψ modification levels at ∼14% to 26% (Fig. 6A and B), which agree with the reported ∼10% to 20% median Ψ modification levels in the human transcriptome (bioRxiv 2024.01.31.578250; ref. 15). PUS7 knockdown significantly reduced the Ψ levels at these sites (Fig. 6B), indicating that their pseudouridylation is primarily mediated by PUS7. Moreover, single-read analysis revealed that PUS7 knockdown eliminated the subset of MCTS1 mRNAs with multiple Ψ sites and increased the population of MCTS1 mRNAs that lack Ψ modification (Fig. 6C; Supplementary Table S6), leading to a 3.7-fold reduction in MCTS1 mRNA Ψ-strength (sum of Ψ fractions at all the Y sites within one mRNA molecule; Fig. 6D; ref. 13).
PUS7 targets MCTS1 to sustain ATF4 mRNA translation. A, Diagram of MCTS1 mRNA with indicated PUS7-dependent PSI (Ψ) sites. B, Quantification of Ψ fractions at MCTS1 mRNA Ψ sites shows significant reduction in Ψ levels after PUS7 knockdown. Data represent two biological replicates with 100 or more mRNA reads. C, PUS7 knockdown eliminated the subset of MCTS1 mRNA with two or more Ψ modifications. Nanopore data (B and C) were analyzed using Fisher exact test. D, PUS7 knockdown reduced the Ψ-strength of MCTS1 mRNA, defined as the sum of Ψ fractions at all Ψ sites within each MCTS1 mRNA molecule. E, Volcano plot shows proteomic profiling of PUS7 knockdown vs. control BE(2)-C cells. MCTS1 and the ATF4 target ASNS are highlighted along with several translation initiation factors. F and G, Immunoblotting (F) and quantification (G) of MCTS1 protein expression in HeLa and BE(2)-C cells without (shGFP) or with PUS7 knockdown (shPUS7). Quantitative data (G) are from two or more independent experiments, with samples from each experiment analyzed by immunoblotting in two or more technical replicates. Data were analyzed by one-way ANOVA. H, Polysome profiling show PUS7 knockdown reduced polysome-associated MCTS1 mRNA. Data are representative of two independent experiments and are presented as mean ± SD of three technical replicates. I, Immunoblotting shows MCTS1 knockdown abrogated serine deprivation (AAR)–induced and tunicamycin (Tm)–induced ATF4 protein expression in BE(2)-C cells. ATF4 protein levels were quantified against α-tubulin. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.
PUS7 targets MCTS1 to sustain ATF4 mRNA translation. A, Diagram of MCTS1 mRNA with indicated PUS7-dependent PSI (Ψ) sites. B, Quantification of Ψ fractions at MCTS1 mRNA Ψ sites shows significant reduction in Ψ levels after PUS7 knockdown. Data represent two biological replicates with 100 or more mRNA reads. C, PUS7 knockdown eliminated the subset of MCTS1 mRNA with two or more Ψ modifications. Nanopore data (B and C) were analyzed using Fisher exact test. D, PUS7 knockdown reduced the Ψ-strength of MCTS1 mRNA, defined as the sum of Ψ fractions at all Ψ sites within each MCTS1 mRNA molecule. E, Volcano plot shows proteomic profiling of PUS7 knockdown vs. control BE(2)-C cells. MCTS1 and the ATF4 target ASNS are highlighted along with several translation initiation factors. F and G, Immunoblotting (F) and quantification (G) of MCTS1 protein expression in HeLa and BE(2)-C cells without (shGFP) or with PUS7 knockdown (shPUS7). Quantitative data (G) are from two or more independent experiments, with samples from each experiment analyzed by immunoblotting in two or more technical replicates. Data were analyzed by one-way ANOVA. H, Polysome profiling show PUS7 knockdown reduced polysome-associated MCTS1 mRNA. Data are representative of two independent experiments and are presented as mean ± SD of three technical replicates. I, Immunoblotting shows MCTS1 knockdown abrogated serine deprivation (AAR)–induced and tunicamycin (Tm)–induced ATF4 protein expression in BE(2)-C cells. ATF4 protein levels were quantified against α-tubulin. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.
Second, our proteomic analysis showed that PUS7 knockdown led to decreased protein levels of MCTS1 and several canonical translation initiation factors (Fig. 6E; Supplementary Table S7). We confirmed the proteomic data by immunoblotting, showing that PUS7 knockdown significantly reduced MCTS1 protein expression in BE(2)-C and HeLa cells (Fig. 6F and G). PUS7 knockdown had no significant effect on MCTS1 mRNA expression (Supplementary Fig. S6C) but markedly reduced the level of polysome-associated MCTS1 mRNA (Fig. 6H). These data suggest that PUS7-dependent pseudouridylation promotes MCTS1 mRNA translation and protein expression.
To further investigate the notion that pseudouridylation may promote translation, we analyzed the nanopore direct RNA-seq data from a recently published study (35) and identified 1,487 uridine sites in polysome-associated transcripts and cytoplasmic poly(A) RNA with significant levels of Ψ modification (P < 0.01; Supplementary Table S8). The vast majority (94.4%) of these Ψ sites in polysome-associated transcripts showed higher Ψ levels than the same sites in the input (cytoplasmic) mRNAs, leading to a 5.9-fold increase in the median Ψ stoichiometry (Supplementary Fig. S6D and S6E). Thus, mRNAs with higher Ψ levels are preferentially associated with polysomes. This model is supported by the observation that polysome-associated mRNAs with higher Ψ levels have increased translation efficiency (35).
Considering above data, we examined the function of MCTS1 in regulation of ATF4 expression in our experimental systems. We used shRNA to silence MCTS1 expression, which markedly diminished ATF4 protein induction by AAR or ER stress in BE(2)-C (Fig. 6I) and HeLa cells (Supplementary Fig. S6F), phenocopying the effect of PUS7 knockdown. In addition, we conducted site-directed mutagenesis to disrupt the interaction of MCTS1 with the ATF4 mRNA 5′-UTR, which contains several upstream open reading frames (uORF) that regulate stress-induced ATF4 mRNA translation (Fig. 7A; refs. 25, 70–72). Recent studies suggest that the 1-amino acid uORF1, consisting of a start and a stop codon (ATGTAG or AUGUAG in mRNA; Fig. 7A), is targeted by the MCTS1-DENR complex (67, 73). Mutations of the uORF1 start and stop codons (G27C,T28G) or only its stop codon (T28C) markedly inhibited stress-induced ATF4 protein expression (Fig. 7B–E; Supplementary Fig. S6G and S6H), phenocopying the effects of PUS7 or MCTS1 knockdown. Random mutations of other nucleotides within the ATF4 mRNA 5′-UTR had no effect (Fig. 7B–E; Supplementary Fig. S6G and S6H, T205C and T223C).
The one amino acid–stop codon uORF1 in the 5′-UTR of ATF4 mRNA is critical for stress-induced ATF4 protein expression. A, Diagram of the ATF4 V2 mRNA with indicated uORFs and the main ORF. The uORF1 nucleotide sequence is shown. B–E, Immunoblotting (B and D) and quantification (C and E) of ATF4 protein expression after the AAR (HisOH; B and C) or ER stress (tunicamycin). D and E, 293T-ATF4 KO cells expressing WT or mutant ATF4 V2 in the presence of Doxy (teton). ATF4 protein levels were quantified against β-actin. Data are representative of three independent experiments. F and G, Immunoblotting (F) and quantification (G) of ATF4 protein expression 24 hours after cotransfection of 293T-ATF4 KO cells with vector (pCDH) or pCDH-MYCN in combination with either pGenLenti-ATF4 V1 or pCW57.1-ATF4 V2, including WT or uORF1 mutants. Data represent results from two independent experiments.
The one amino acid–stop codon uORF1 in the 5′-UTR of ATF4 mRNA is critical for stress-induced ATF4 protein expression. A, Diagram of the ATF4 V2 mRNA with indicated uORFs and the main ORF. The uORF1 nucleotide sequence is shown. B–E, Immunoblotting (B and D) and quantification (C and E) of ATF4 protein expression after the AAR (HisOH; B and C) or ER stress (tunicamycin). D and E, 293T-ATF4 KO cells expressing WT or mutant ATF4 V2 in the presence of Doxy (teton). ATF4 protein levels were quantified against β-actin. Data are representative of three independent experiments. F and G, Immunoblotting (F) and quantification (G) of ATF4 protein expression 24 hours after cotransfection of 293T-ATF4 KO cells with vector (pCDH) or pCDH-MYCN in combination with either pGenLenti-ATF4 V1 or pCW57.1-ATF4 V2, including WT or uORF1 mutants. Data represent results from two independent experiments.
Next, we examined the impact of ATF4 mRNA uORF1 mutations on MYCN regulation of ATF4 protein expression. Cotransfection experiments showed that MYCN overexpression upregulated ATF4 protein expression from either V1- or V2-expressing plasmid (Fig. 7F and G). Mutations of the uORF1 start and stop codons (V1, G30C,T31G; V2, G27C,T28G) abrogated the upregulation, whereas the T205C mutation within the ATF4 mRNA 5′-UTR had no effect (Fig. 7F and G).
Collectively, the data presented above suggest that MCTS1 is a key downstream target of PUS7 for efficient ATF4 protein induction in response to cellular stress. Moreover, the data provide evidence supporting the model that MYCN targets the PUS7-MCTS1 axis to promote ATF4 expression, thereby mitigating oncogenic stress.
PUS7 consensus pseudouridylation site in the 3′-UTR of ATF4 mRNA regulates the timing of stress-induced ATF4 protein expression
To investigate whether PUS7 directly targets ATF4 mRNA for pseudouridylation, we analyzed the nanopore RNA-seq data from control and PUS7 knockdown BE(2)-C cells (bioRxiv 2024.01.31.578250). The numbers of native ATF4 mRNA reads were too low to obtain robust results, most likely because of the low abundance of ATF4 mRNA under normal culture conditions. We then treated control and PUS7 knockdown BE(2)-C cells with tunicamycin to increase ATF4 mRNA expression and performed nanopore sequencing, which revealed several PUS7-dependent Ψ sites in the ATF4 transcript ENST00000337304.2 ATF4-201, corresponding to ATF4 mRNA variant 1 (V1) nucleotides 6 to 2,041 (NM_001675.4; Fig. 8A). PUS7 knockdown decreased Ψ levels at these sites by more than 50% and largely eliminated the pseudouridylation of ATF4 V1 mRNA under ER stress, leading to a fivefold reduction in Ψ-strength (Fig. 8B–D; Supplementary Table S9). These data suggest that ATF4 V1 mRNA is a direct target of PUS7-mediated pseudouridylation.
PUS7 consensus pseudouridylation site in the ATF4 mRNA 3′-UTR regulates the timing of stress-induced ATF4 protein expression. A, Diagram of ATF4-201 transcript (ATF4 V1 mRNA) with indicated PUS7-dependent Ψ sites following ER stress. B, Quantification of Ψ fractions at ATF4-201 transcript Ψ sites shows >50% reduction in Ψ levels following PUS7 knockdown. Data represent two biological replicates with 32–35 mRNA reads. C, PUS7 knockdown largely eliminated Ψ modifications in ATF4-201 transcripts under ER stress. Data were analyzed using Fisher exact test. D, PUS7 knockdown reduced the Ψ strength of ATF4-201 transcripts. E,In vitro pseudouridylation of ATF4-201 transcript U1984 by immunopurified PUS7. F–I, Immunoblotting (F and H) and quantification (G and I) of ATF4 protein expression at various time points following AAR (HisOH) or ER stress (tunicamycin, Tm)). Quantitative data (G and I) are presented as mean ± SD (n = 3). J, Model for MYC/MYCN activation of the PUS7-MCTS1-ATF4 axis to mitigate cellular stress induced by oncogenic activation. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (J, Created with BioRender.com.)
PUS7 consensus pseudouridylation site in the ATF4 mRNA 3′-UTR regulates the timing of stress-induced ATF4 protein expression. A, Diagram of ATF4-201 transcript (ATF4 V1 mRNA) with indicated PUS7-dependent Ψ sites following ER stress. B, Quantification of Ψ fractions at ATF4-201 transcript Ψ sites shows >50% reduction in Ψ levels following PUS7 knockdown. Data represent two biological replicates with 32–35 mRNA reads. C, PUS7 knockdown largely eliminated Ψ modifications in ATF4-201 transcripts under ER stress. Data were analyzed using Fisher exact test. D, PUS7 knockdown reduced the Ψ strength of ATF4-201 transcripts. E,In vitro pseudouridylation of ATF4-201 transcript U1984 by immunopurified PUS7. F–I, Immunoblotting (F and H) and quantification (G and I) of ATF4 protein expression at various time points following AAR (HisOH) or ER stress (tunicamycin, Tm)). Quantitative data (G and I) are presented as mean ± SD (n = 3). J, Model for MYC/MYCN activation of the PUS7-MCTS1-ATF4 axis to mitigate cellular stress induced by oncogenic activation. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (J, Created with BioRender.com.)
We further investigated the Ψ modification at U1984 in the 3′-UTR (Fig. 8A) because it is the Ψ site of the consensus PUS7 target sequence “UGUAG” (underlined U indicates the Ψ site; refs. 9–12, 15). PUS7 knockdown reduced the Ψ level by 2.3-fold at U1984 (Fig. 8B), although the difference was not statistically significant because of the low numbers of mRNA reads. We performed in vitro pseudouridylation assay (34, 48) using an in vitro transcribed ATF4 V1 RNA oligo that included nucleotides upstream and downstream of U1984 (Supplementary Table S1). After incubation with immunopurified human PUS7, U1984 showed a significant level of Ψ modification (Fig. 8E). Thus, both in vivo and in vitro results suggest that U1984 is a target of PUS7-mediated pseudouridylation.
We then assessed the functional significance of U1984 by site-directed mutagenesis. (U1984 in the ATF4-201 transcript corresponds to U1989 in the ATF4 V1 mRNA, NM_001675.4. We used U1984 here to be consistent with the nanopore sequencing data.) We replaced T1984 with C (T1984C) in the ATF4 V1 gene and established 293T-ATF4 KO cells (25) with a stable expression of exogenous WT or mutant ATF4 V1 (T1984C). Time course studies revealed that T1984C mutation delayed the induction of ATF4 protein by AAR or ER stress (Fig. 8F and G); whereas the induction of WT ATF4 peaked around 4 hours after stress, maximal induction of ATF4 V1 T1984C was not observed until 16 hours after stress. We obtained essentially the same results with the T-to-C mutation at the corresponding T1384 in the ATF4 V2 gene (NM_182810.3; Fig. 8H and I). These findings suggest that the consensus PUS7 target site U1984 is a positive regulatory element in the ATF4 mRNA 3′-UTR that is essential for timely induction of ATF4 protein expression by cellular stress.
Discussion
In this report, we present evidence of a MYC-activated, PUS7-dependent mRNA pseudouridylation program essential for the survival, proliferation, and tumorigenicity of MYC-driven cancer cells. We demonstrate that MYC and MYCN directly target the PUS7 gene for transcriptional upregulation, explaining the specific upregulation of PUS7 in MYCN-amplified neuroblastoma tumors in patients and the stage-wise increase in PUS7 expression during lymphoma progression in the Eµ-Myc mouse model. Moreover, our investigation suggests a molecular basis for the functional significance of PUS7 upregulation by the MYC family of oncoproteins: to enhance ATF4 protein expression, thereby mitigating cellular stress induced by MYC-driven biomass production, which supports cell growth and proliferation (24, 56–58). This model is supported by evidence showing that ATF4 overexpression can rescue the growth inhibitory phenotype caused by PUS7 knockdown. The ATF4-induced transcriptional program drives cellular adaptive responses, including increased expression of ER chaperones to facilitate protein folding, 4E-BP1 to reduce protein synthesis and proteotoxic stress, amino acid synthesis enzymes and transporters to raise cellular amino acid levels, and aminoacyl-tRNA synthetases to capture amino acids for protein synthesis (Fig. 8J; refs. 60, 62, 63). Thus, ATF4 upregulation offers a survival advantage for MYC-driven cancers, as shown by our study and recent reports that ATF4 is essential for cell survival after MYC activation and is critical for MYC-driven tumor progression (58, 60). A key finding of our study is the identification of PUS7-dependent mRNA pseudouridylation as a mechanism driving ATF4 expression after MYC activation.
In addition to mitigating oncogenic stress, our data suggest a broader role for PUS7 in amino acid metabolism and adaptive responses by upregulation of ATF4 expression. We demonstrate that PUS7 is essential for timely and efficient ATF4 protein induction in response to ER stress and amino acid deprivation, as well as for maintaining the expression of ATF4 target genes involved in amino acid biosynthesis and transport. PUS7 knockdown reduced the production of serine and glycine from glucose, asparagine from glutamine, and reduced serine transport. Collectively, our data indicate that PUS7-dependent mRNA pseudouridylation is an RNA epigenetic mechanism that promotes stress responses and stress-induced metabolic reprogramming.
Our findings suggest two distinct mechanisms by which PUS7 sustains ATF4 protein expression (Fig. 8J). First, PUS7 functions through MCTS1 to target the uORFs in the 5′-UTR of ATF4 mRNA, which are critical for stress-induced ATF4 mRNA translation (25, 70–72). Under stress conditions, eIF2α, a subunit of the translation initiation factor eIF2, is phosphorylated by stress-activated eIF2α kinases, including GCN2 (the AAR) and PERK (ER stress). Phosphorylation of eIF2α reduces the level of active eIF2, impairing the recruitment of the initiator Met-tRNAMeti to the 40S ribosomal subunit for translation initiation, leading to a reduction in global mRNA translation (55, 74, 75). MCTS1 is a noncanonical translation initiator (76, 77) that is essential for stress-induced ATF4 mRNA translation by targeting 5′-UTR uORFs, including the 1-amino acid uORF1 (67, 68). Mechanistically, the MCTS1-DENR complex promotes eIF2-independent recruitment of Met-tRNAMeti (76) and translation reinitiation at ATF4 mRNA uORFs (67, 68). We identified five PUS7-dependent Ψ sites in MCTS1 mRNA. PUS7 knockdown eliminated the subset of MCTS1 mRNAs containing multiple Ψ sites and reduced the overall Ψ-strength of MCTS1 mRNA, leading to a decrease in polysome-associated MCTS1 mRNA and MCTS1 protein expression. Additionally, we demonstrated that MCTS1 knockdown was sufficient to abrogate ATF4 induction in response to amino acid deprivation and ER stress. Thus, our data suggest a PUS7-MCTS1 axis in the regulation of stress-induced ATF4 protein expression.
Second, we present evidence that PUS7 also targets the 3′-UTR of ATF4 mRNA to regulate ATF4 protein expression. We identified a PUS7 consensus Ψ site within the 3′-UTR of ATF4 mRNA (U1984 in ATF4-201 transcript or U1989 in ATF4 V1 mRNA) and confirmed that U1984 is a target of PUS7-dependent pseudouridylation. Importantly, we found that U1984 is required for the timely induction of ATF4 protein expression in response to amino acid deprivation and ER stress. Thus, our study uncovers a previously unrecognized positive regulatory element in the 3′-UTR of ATF4 mRNA that controls the timing of ATF4 protein induction and stress responses.
Our findings suggest that PUS7-dependent mRNA pseudouridylation enhances translation. PUS7 knockdown reduced the Ψ levels and polysome association of MCTS1 and ATF4 mRNAs, leading to the downregulation of MCTS1 and ATF4 protein expression. Furthermore, we present evidence that mRNAs with higher Ψ levels are preferentially associated with polysomes, offering an explanation for their increased translation efficiency (35). Interestingly, our transcriptome-wide profiling of pseudouridylation by nanopore native RNA-seq revealed a median Ψ stoichiometry of 16.8% for PUS7-dependent Ψ sites in poly(A) RNAs (bioRxiv 2024.01.31.578250), similar to the MCTS1 and ATF4 mRNA Ψ modification levels reported here. Bisulfite-based sequencing has also shown an ∼10% median modification level at Ψ sites in the human transcriptome (15). These observations raise the question of how such low-stoichiometric pseudouridylation impacts mRNA translation or stability. The same question may also be raised for other low stoichiometric (<20%) mRNA modifications, such as m5C (78) and m6A (5, 79–81), which, nevertheless, have been shown to regulate mRNA stability and translation (82–85). We noted that PUS7 modifies multiple uridine sites within single mRNA molecules, which increases mRNA Ψ-strength. PUS7 knockdown essentially eliminated the subsets of MCTS1 and ATF4 mRNAs with multiple Ψ sites and markedly decreased their Ψ-strength. It has been reported recently that mRNA molecules with higher Ψ-strength show higher levels of expression (13). Thus, we speculate that it is the overall Ψ-strength of individual mRNAs that dictates their stability and translation efficiency. Similarly, it has been suggested that m6A modification is likely to be functionally significant for mRNAs that carry either a high stoichiometry site or multiple low stoichiometry sites (81).
Recent studies have identified a role for PUS7 in cancer through tRNA pseudouridylation and translational repression (19, 20). Our study suggests a new mechanism for PUS7 action in cancer via mRNA pseudouridylation to enhance ATF4 protein expression. This MYC-driven, PUS7-mediated mRNA pseudouridylation program enables an efficient and timely ATF4-dependent adaptive response to alleviate oncogenic and nutritional stress associated with MYC-induced cell growth and proliferation. These findings illuminate the molecular regulation and functional significance of mRNA pseudouridylation in cancer development.
Authors’ Disclosures
S. Sudarshan reports grants from NIH and the Department of Veterans Affairs during the conduct of the study. H.-F. Ding reports grants from NIH during the conduct of the study. No disclosures were reported by the other authors.
Authors’ Contributions
J. Ding: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft. M. Bansal: Conceptualization, data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft. Y. Cao: Formal analysis, investigation, methodology. B. Ye: Formal analysis, investigation. R. Mao: Formal analysis, investigation. A. Gupta: Formal analysis, investigation. S. Sudarshan: Conceptualization, resources, writing–review and editing. H.-F. Ding: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, writing–original draft.
Acknowledgments
The authors thank Kyoko Kojima and James A. Mobley at the UAB Mass Spectrometry and Proteomics Core for proteomic analysis, Brian J. Altman at the University of Rochester Medical Center for sharing SHEP MYCN-ER cells, and Carson C. Thoreen at Yale University for providing pCW57.1-ATF4 (V2) and 293T-ATF4 KO cells. The Mass Spectrometry and Proteomics Core of the O’Neal Comprehensive Cancer Center at the UAB was supported in part by the NIH/NCI Cancer Centers grant P30CA013148. This work was supported by the NIH grant R01CA190429 (to H.-F. Ding); the UAB Center for Clinical and Translational Science, the NIH National Center for Advancing Translational Sciences research award UL1TR003096; and NIH R01CA200653 and Department of Veterans Affairs I01BX002930 (to S. Sudarshan).
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).