Stem Cell

RBBP6 maintains glioblastoma stem cells through CPSF3-dependent alternative polyadenylation – Cell Discovery

Time Stamp: March 18, 2024 8:00 PM
Source Node: 558909

GSC derivation

The GSC468, GSC738, and GSC3565 cell lines were generated in our laboratory as reported previously³⁹. Patient-derived xenografts were generated and maintained as a reproducible source of GSC cells. The glioblastoma primary cells GBM12388157, and GBM12479390 in this study were obtained from the Department of Neurosurgery, the Second Affiliated Hospital of Zhejiang University School of Medicine. All glioblastoma tissues were obtained from excess surgical resection samples from patients at the Second Affiliated Hospital Zhejiang University School of Medicine with appropriate consent and in accordance with an IRB-approved protocol (IR2022453). All GSC and primary GBM cells were cultured as neurospheres in neurobasal medium (Gibco, Cat# 12349-015) supplemented with 2% B27 supplement (Gibco, Cat# 12587-010), 1% GlutaMax Supplement (Gibco, Cat# 35050-061), 1% sodium pyruvate (Gibco, Cat# 11360-070), 1% penicillin/streptomycin (Invitrogen, Cat# SV30010), 20 ng/mL basic human fibroblast growth factor (R&D systems, Cat# 4114-TC), and 20 ng/mL human epidermal growth factor (R&D systems, Cat# 236-EG). Short Tandem Repeat (STR) analyses were performed to authenticate the identity of each cell line used in this article. Mycoplasma testing was performed by qPCR with cellular supernatants on a yearly basis. Cells were grown for fewer than 20 in vitro passages from xenografts.

Other cell models

The HEK293T cells were purchased from the American Type Culture Collection (ATCC, Cat# CRL-3216). HEK293T cells were cultured in DMEM (Gibco, Cat# C11995500CP) supplemented with 1% GlutaMax Supplement, 1% penicillin/streptomycin and 10% fetal bovine serum (CellMax, Cat# SA211.02). The iPSC cells were purchased from Cellapy (Cellapy, Cat# CA4002106). iPSC-NPCs were differentiated from the iPSC cells using serum-free STEMdiff™ SMADi Neural Induction Kit (StemCell Technologies, Cat# 08581) according to the manufacturer’s instructions. Astrocytic precursors were generated from iPSC-NPCs using STEMdiff™ Astrocyte Differentiation Kit (StemCell Technologies, Cat# 100-0013) according to the manufacturer’s instructions. These astrocytic precursors are then matured further into astrocytes using STEMdiff™ Astrocyte Maturation Kit (StemCell Technologies, Catalog# 100-0016) according to the manufacturer’s instructions. iPSC-astrocytes were cultured in DMEM supplemented with 1% GlutaMax Supplement, 1% penicillin/streptomycin, and 10% fetal bovine serum. iPSC-NPC were cultured in a neurobasal medium supplemented with 2% B27 supplement, 1% GlutaMax Supplement, 1% sodium pyruvate, 1% penicillin/streptomycin, 20 ng/mL basic human fibroblast growth factor, and 20 ng/mL human epidermal growth factor. STR analyses were performed to authenticate the identity of each cell line used in this article. Mycoplasma testing was performed by qPCR with cellular supernatants yearly.

Animal experiments

All mouse experiments were performed under an animal protocol approved by the Institutional Animal Care and Use Committee of Westlake University and in accordance with the relevant guidelines. Intracranial transplantation of GSCs was performed as previously described³⁹. In brief, GSC spheres were dissociated into single cells, and 10,000 cells were intracranially injected into the right cerebral cortex of individual NSG immunocompromised mice. When any mouse was observed to exhibit neurological signs or signs of morbidity, including lethargy, gait changes, hunched posture, and weight loss, we sacrificed all mice in that cohort. To compare tumor growth in vivo, we isolated brains from mice transplanted with GSCs on the day that neurological signs or signs of morbidity were observed. Then, the brains were used for histologic analysis by H&E staining. In parallel survival experiments, mice were observed until the development of neurological signs or signs of morbidity.

Evaluation of JTE-607 in GBM model

GSC spheres were dissociated into single cells, and 10,000 cells were intracranially injected into the right cerebral cortex of individual NSG immunocompromised mice. Osmotic minipumps (ALZET, Cat# 21-3032) were implanted for the direct delivery of the vehicle or JTE-607 (dissolved in water, 7.143 μg per day per mouse) to the tumor site through a brain infusion kit (ALZET, Cat# 0008663) on one week after intracranial GSC injection.

Immunofluorescence

The tissues were fixed overnight at 4 °C using 4% paraformaldehyde, followed by a 48-h incubation in 30% sucrose. Then, the tissues were embedded in an OCT embedding medium. Cryosections with a thickness of 10 µm were prepared for the immunofluorescence experiment. For immunofluorescence staining, the sections were fixed with 4% paraformaldehyde at room temperature for 10 min. Subsequently, the sections were washed three times with 1× PBS. To permeabilize the sections, 0.25% Triton-X100 was applied at room temperature, followed by another three washes with 1× PBS. After the washing steps, the sections were blocked with 5% serum derived from the donkey and then washed with 1× PBS. Staining for MYC (Cell signaling Technology, Cat# D3N8F, 1:200) or Ki-67 (Proteintech, Cat# 27309-1-AP, 1:200) was carried out at room temperature for 1 h. Following the staining, the sections were washed three times and cover-slipped using a fluorescent mounting medium containing DAPI (ZSGB-BIO, Cat# ZLI-9557). To detect apoptosis in the tumors, the sections were processed using a TUNEL kit (Beyotime, Cat# C1086) according to the manufacturer’s instructions.

Chemicals and oligos

JTE-607 was purchased from MedChemExpress (Cat# HY-110133). The antisense morpholino was purchased from Gene Tools. The antisense oligonucleotide sequences employed in this study are provided below: NC-morpholino: 5’-CCTCTTACCTCAGTTACAATTTATA-3’; ZNF281- morpholino: 5’-AATTTTGGATCAGCCCAGATGGAGA-3’; DNMT3B- morpholino: 5’-GGCTCCAGTTACAAAAAAAATTTTA-3’. Hsa-miR-590-3p antagomir was purchased from Ribobio (Cat# miR30004801-4-5).

CRISPR/Cas9 library

The CRISPR/Cas9 library in this study included 100 nontargeting controls and 3704 unique sgRNAs individually targeting 463 ubiquitin E3 ligases and deubiquitinases. A customized single-strand sgRNA oligonucleotide pool was synthesized by Twist Bioscience, amplified by PCR and then inserted into the lentiCRISPR-v2 vector (lentiCRISPR-v2, RRID:Addgene_52961, Addegene) using the NEBuilder HiFi DNA Assembly Cloning Kit (NEB, Cat# E5520). Next-generation sequencing was performed to ensure sgRNA abundance.

Lentivirus production

HEK293T cells were cotransfected with a lentiviral expression vector, the packaging plasmid psPAX2 (psPAX2, RRID:Addgene_12260, Addegene) and the envelope plasmid pMD2.G (pMD2.G, RRID:Addgene_12259, Addegene) using polyethylenimine (PEI) transfection reagent (Polyscience, Cat # 23966-1) following the manufacturer’s instructions. Lentiviral particles were collected 48 h after the medium change and concentrated using the lentivirus concentration kit (Genomeditech, Cat# GM-040801-100).

CRISPR screen of GSCs

GSC spheres were dissociated into single cells with TrypLE (ThermoFisher Scientific, Cat# 12604021) and were transduced with the lentiviral CRISPR library at a multiplicity of infection of 0.3–0.5 before puromycin selection. For the in vitro screens, GSCs were cultured for 1–2 weeks to ensure at least 10-fold cell doubling. For the in vivo screens, GSCs were implanted into the brains of NSG mice (1 million GSCs per mouse), and the brains were harvested when the mice showed neurological signs. Genomic DNA was isolated from GSCs at the experimental start points as well as from GSCs harvested at the experimental endpoints to prepare the library for next-generation sequencing.

Analysis of CRISPR screen data

FASTQ files were trimmed with Trim Galore (Trim Galore, RRID:SCR_011847). The trimmed FASTQ files were then analyzed using MAGeCK software to obtain the read count of each sgRNA. MAGeCK robust rank aggregation (RRA) algorithm was used to identify significantly enriched and depleted sgRNAs and genes by comparing the in vitro cultured samples and in vivo tumor samples with the control samples.

Plasmids

The following sgRNA sequences were chosen from our CRISPR library or designed with CHOP-CHOP (CHOPCHOP, RRID:SCR_015723). SgRNA sequences were inserted into the lentiCRISPR v2 plasmid (lentiCRISPR v2, RRID:Addgene_52961, Addgene). SgRNAs used in this study were sgNT: 5’-CTCTGCTGCGGAAGGATTCG-3’, sgRBBP6#1: 5’-AAGTCGAACTGAACCAGCGA-3’, sgRBBP6#7: 5’-AAGTCGAACTGAACCAGCGA-3’, sgCPSF3#1: 5’-ATGTTCATGATTGAGATCGC-3’, sgCPSF3#2: 5’-ATTCATAGACTAACCACATG-3’. The shRNA sequences were inserted into the pLKO.TRC.1 plasmid (pLKO.TRC.1, RRID: Addgene_10878, Addgene). ShRNAs used in this study were shNT: 5’-CAACAAGATGAAGAGCACCAA-3’, shRBBP6#2: 5’- GACTCTCCTTCTCGGAATAAA-3’, shRBBP6#7: 5’-GATGACTCTTCCGCGTCTATT-3’, shRBBP6#UTR: 5’-GGGTCTCTGGATTATTGTT-3’.

Quantitative RT-PCR

Total cellular RNA was extracted with TRIzol reagent (Invitrogen, Cat# 15596018) according to the manufacturer’s instructions. Then, RNA was reverse transcribed to cDNA with a cDNA reverse transcription kit (Novoprotein, Cat# E047-01B). Additionally, miRNA was reverse transcribed to cDNA with a miRNA 1st Strand cDNA Synthesis Kit (Vazyme, Cat# MR101-01) and specific probes (miR-590-3p: GTCGTATCCAGTGCA-GGGTCCGAGGTATTCGCACTGGATACGACACTAGC, GADPH: GGCTGTTGTCAT-ACTTCTCATGG; RNU6: AGGGGCCATGCTAATCTTCT). qPCR was performed with Novostart SYBR qPCR Supermix Plus (Novoprotein, Cat# E096) in a Bio-Rad CFX instrument. The qPCR primers used in this study were as follows: GAPDH forward primer: 5’-GGAGCGAGATCCCTCCAAAAT-3’, GAPDH reverse primer: 5’-GGCTGTTGTCATACTTCTCATGG-3’; 18 S forward primer: 5’-GGCCCTGTAATTGGAATGAGTC-3’; 18 S reverse primer: 5’-CCAAGATCCAACTACGAGCTT-3’; RBBP6 forward primer: 5’-CATCTCCCTCTGCGACTTAAAG-3’; RBBP6 reverse primer: 5’-TAGTTGCCATCGCTGGTTCAG-3’; MYC forward primer: 5’-GGCTCCTGGCAAAAGGTCA-3’; MYC reverse primer: 5’- CTGCGTAGTTGTGCTGATGT-3’; DNMT3B (coding sequence, CDS) forward primer: 5’-AGGGAAGACTCGATCCTCGTC-3’; DNMT3B (CDS) reverse primer: 5’-GTGTGTAGCTTAGCAGACTGG-3’; DNMT3B (UTR) forward primer-: 5’-TGGAGCCACGACGTAACAAA-3’; DNMT3B (UTR) reverse primer: 5’-GCATCCGTCATCTTTCAGCC-3’; ZNF281 (CDS) forward primer: 5’-AGGACCTCAGTATTCTCCACC-3’; ZNF281 (CDS) reverse primer: 5’- CCATCTCCAACCAAAGAAGGTTT-3’; ZNF281 (UTR) forward primer: 5’- TGCTTTACTCTCAGGAAAGTGT-3’; ZNF281 (UTR) reverse primer: 5’-TGTTACAGTTGAGATCAAGAGAGGG-3’; miR-590-3p forward primer: 5’-GCGCGCGCGCTAATTTTATGTATAA-3’; miR-590-3p reverse primer: 5’-CAGTGCAGGGTCCGAGGTAT-3’.

APA analysis of a single gene was performed as previously described¹⁶. In brief, the levels of transcripts amplified with common primers targeting the CDS were used for normalization to the total transcript level. The distal primers targeted sequences just upstream of the dPAS and were used to detect long transcripts that used the dPAS. Values were calculated as previously described¹⁶. ΔCT (common or distal) = CT_{common or distal}−CT_GAPDH. ΔΔCT = ΔCT_distal− ΔCT_common. Normalized ΔΔΔCT = ΔΔCT_{average of case} – ΔΔCT_{average of control.}

Apoptosis assay

The apoptosis assay was performed with Annexin V-Alexa Fluor 647 (Yeasen, Cat# 40304ES60). Samples were analyzed using a Beckman Coulter Cytoflex Cytometer.

ELDA

The neurosphere formation capacity was assayed by ELDA. In brief, different numbers of cells per well (100, 50, 20, 10, and 5) were plated into 96-well plates. Seven days later, the number of each well containing neurospheres was recorded. ELDA software was used to analyze the number of wells containing neurospheres as previously described⁴⁰.

Cell proliferation assay

Twenty-five hundred cells per well were plated in 96-well plates. Cell viability was measured with a CellTiter-Glo Luminescent Cell Viability Assay Kit (Promega, Cat# G7572) at the indicated times.

Western blotting analysis

Cells were collected, washed, lysed with RIPA buffer (Beyotime, Cat# P0013C) and incubated on ice for 30 min. Lysates were centrifuged at 4 °C for 15 min at 12,000 rpm, and the supernatants were collected. A Bradford (Beyotime, Cat# P0006C) kit was used to determine protein concentrations. Samples containing equal amounts of protein were mixed with LDS Sample Buffer (Invitrogen, Cat# B0007), boiled for 10 min, separated using SDS-PAGE, and then transferred onto PVDF membranes. The membranes were blocked with 5% nonfat milk for 1 h. Next, the membranes were incubated with a primary antibody at 4 °C overnight. The membranes were washed with TBST buffer and were then incubated with secondary antibodies in 5% nonfat milk for 1 h. For all western blot analyses, membranes were imaged using Bio-Rad Image Lab software (Image Lab, RRID:SCR_003073). The antibodies used in this study wereRBBP6 (Bethyl Laboratories, Cat# A304-975A), CPSF3 (Proteintech, Cat# 11609-1-AP), CPSF2 (Proteintech, Cat# 17739-1-AP), NUDT21 (Proteintech, Cat# 10322-1-AP), CSTF2 (Proteintech, Cat# 26825-1-AP), HA-tag (Cell signaling Technology, Cat# 3724 S), Flag-tag (Sigma, Cat # F1804), MYC-tag (Proteintech, Cat # 16286-1-AP), MYC (Cell signaling Technology, Cat# D3N8F), GAPDH (Proteintech, Cat# 60004-1-Ig).

IP experiments

Cells were collected, washed, lysed with NP-40 lysis buffer (Beyotime, Cat# P0013F) and incubated on ice for 30 min. Lysates were centrifuged at 4 °C for 15 min at 12,000 rpm, and the supernatants were collected. A Bradford (Beyotime, Cat# P0006C) kit was used to determine protein concentrations. For IP experiments, samples containing one milligram of protein were incubated with the indicated anti-Flag-M2 beads (Sigma, Cat# M8823) or anti-HA beads (ThermoFisher Scientific, Cat# 88838) for 1 h at room temperature. After incubation, the beads were washed with lysis buffer five times. After washing, the beads were mixed with LDS Sample Buffer (Invitrogen, Cat# B0007) and boiled for 10 min for western blot analysis.

Luciferase reporter assay

The indicated plasmids were transfected into HEK293T cells (ATCC, Cat# CRL-3216) for 48 h. Then, we used a Dual-Luciferase Reporter Gene Assay Kit (Yeasen, Cat# 11402ES60) to measure firefly and Renilla luciferase activity.

3’ RACE

The full length 3’UTR mRNA were generated with GoScript™ Reverse Transcriptase (Promega, Cat# A5001) kit, following the manufacturer’s protocol with Oligo dT18-N(22) primer. GSP-1 primers and Oligo N(22) primer were used to perform the first-round PCR. GSP-2 primers and Oligo N(22) primer were used to perform the second-round PCR. The 3’RACE primers used in this study were as follows: Oligo dT18-N(22) primer: 5’-CTGATCTAGAGGTACCGGATCCTTTTTTTTTTTTTTTTTT-3’; Oligo N(22) primer: 5’-CTGATCTAGAGGTACCGGATCC-3’; ZNF281 GSP-1 primer: 5’-GGAGTGTGGTTTCGGCCAA-3’; ZNF281 GSP-2 primer: 5’- AGTGTGGTTTCGGCCAATCT-3’; DNMT3B GSP-1 primer: 5’- TCTTTGGCTTTCCTGTGCAC-3’; DNMT3B GSP-2 primer: 5’- TGGCTTTCCTGTGCACTACA-3’. The PCR products were further sequenced.

RNA-seq analysis

Total RNA was extracted with an RNeasy Small RNA Isolation Kit with Spin Columns (Beyotime, Cat# R0028) according to the manufacturer’s instructions. RNA samples, namely, 4 samples of GSC468 cell lines under two different conditions (Control: GSC468_shNT; Case: GSC468_RBBP6 knockdown) with 2 biological replicates per condition, were sequenced on the Illumina platform, and data were obtained in FASTQ format. For each FASTQ file, quality checks were conducted using FastQC (FastQC, RRID:SCR_014583)⁴¹. Contaminating data, such as low-quality reads, adaptor sequences, and poor-quality bases, were removed with Trimmomatic software (Trimmomatic, RRID:SCR_011848)⁴², and 252,765,061 clean reads were generated. The trimmed reads were mapped to the human reference genome (GRCh38) using STAR (STAR, RRID:SCR_004463)⁴³ and were then sorted by SAMtools (SAMtools, RRID:SCR_002105)⁴⁴. Subsequently, the uniquely mapped reads were quantified by featureCounts (featureCounts, RRID:SCR_002105)⁴⁵. The transcript per million (TPM) values were calculated using RSEM. The genes with TPM < 1 in more than 80% of samples were removed. The remaining genes were then used to perform downstream analysis. DEG analysis was performed with the DESeq2 package (DESeq, RRID:SCR_000154)⁴⁶.

APA event analysis

We used DaPars2 to identify the most significant APA events between the shNT and RBBP6 knockdown conditions. To quantify APA events, we first downloaded the human gene annotation file (GRCh38) from UCSC and extracted a 3’UTR annotation for each transcript. Subsequently, .wig files were generated using STAR (STAR, RRID:SCR_004463). The PDUI values were extracted by the DaPars2 algorithm (version 2.0)⁴⁷ with the .wig files. A PDUI score close to 0 indicates that the gene tends to use a proximal PAS, whereas a PDUI score close to 1 indicates that the gene tends to use a dPAS. Statistical significance was assumed when the adjusted P-value of the PDUI difference was less than 0.05 and the absolute mean PDUI difference was greater than 0.2.

Trans-effect analysis of 3’US

MAT3UTR software²⁰ was used to quantify the trans-effect of 3’US. First, we extracted the 3’UTR information of transcripts from UCSC. Second, we collected the miRNA binding data from TarBase⁴⁸, miRecords⁴⁹, miRTarBase⁵⁰, and TargetScanHuman version 6.2⁵¹; these data contained the transcript ID, gene name, miRNA family name, chromosome, start and end coordinate of the binding region, and strand information. Third, we assumed coexpression when a pair of transcripts shared at least five miRNA binding sites in their 3’ UTRs. We identified the potential ceRNA partners of the given genes with 3’US. Finally, we used the script ‘MAT3UTR.py’ to estimate the trans-effect of 3’US for each mRNA and miRNA pair. Enrichment analysis was conducted to examine the ceRNA partner genes’ association with OGs. The OGs utilized in this study were determined by the TUSON algorithm, which identified residue-specific activating mutations in over 8200 tumor/normal pairs through genome sequencing. The genes were ranked based on their TUSON prediction P-values, with the top 1000 genes (P < 0.01) considered as the reference oncogenes for the enrichment analysis. To enhance statistical power, we selected 551 highly expressed OGs (TPM > 1) for further analysis. Among these 551 OGs, 37 were identified as 3’UTR ceRNAs in our study, while 514 were not in the ceRNA networks (ceRNET). Welch’s t-test, which accounts for different variances in the two groups being compared, was employed to compare means. To verify the normality assumption for the t-test, a Shapiro–Wilk normality test for small sample sizes (n < 50) was conducted. All statistical analyses were performed using R (version 3.6).

APA regulatory analysis

The 3’ end processing factors datasets were retrieved from NCBI GEO (GSE151919 and GSE149204) and ENCODE (ENCSR594DNW vs ENCSR067GHD, ENCSR815JDY vs ENCSR856ZRV, ENCSR895BTE vs ENCSR913CAE). We then used DaPars2 to identify the most significant APA events. We applied a hypergeometric test to test for significant overlap between the pre-mRNA 3’ end processing factors shortening genes and the RBBP6 shortening genes.