To detect bulk changes in nascent transcripts after irradiation in situ, we added the uridine analog ethynyl uridine (EU) to human U2OS cells for 30 min. The incorporated EU was detected by click chemistry to a fluorescent azide followed by immunofluorescence microscopy (Jao and Salic, 2008). A reduction in bulk nascent transcripts was apparent 30 min after exposure to 10 Gray ionizing radiation (IR), and the transcriptional recovery was already occurring 2 hr after irradiation (Figure 1A). Co-immunofluorescence with γH2AX showed that the DNA damage signal was greatly reduced at the same time point after IR where transcriptional recovery occurred (Figure 1—figure supplement 1), consistent with transcriptional recovery occurring after DSB repair. Given that the histone variant H3.3 histone chaperone HIRA promotes transcriptional recovery after UV repair (Bouvier et al., 2021; Adam et al., 2013), we tested whether that was also the case for transcriptional recovery after IR. We found that shRNA depletion of HIRA (Figure 1—figure supplement 2) had no effect on transcriptional inhibition nor recovery after IR (Figure 1A). In agreement, depletion of transcripts encoded from both H3.3 genes (Figure 1—figure supplement 2) had no effect on transcriptional inhibition or transcriptional recovery after IR (Figure 1B). As such, the requirement for HIRA for transcriptional recovery differs following IR and UV exposure, suggesting differences in the mechanism of these processes.

Given that the read-out of nascent EU labeled transcripts after irradiation is fluorescence, we established a flow cytometry-based assay to allow us to screen for factors regulating transcriptional inhibition and recovery after IR. We established this assay in a murine Abelson virus-transformed pre-B cell line (termed Abl pre-B cells) (Bredemeyer et al., 2006) which are non-adherent and are stably transformed with doxycycline-inducible Cas9 (Chen et al., 2021). By flow cytometry analysis, effective incorporation of EU into nascent transcripts was apparent in a manner dependent on ongoing transcription because it was inhibited by the global RNA polymerase inhibitor Actinomycin D (Figure 2A). The EU signal had 2 peaks (Figure 2A) and we asked whether this reflects cell cycle differences in the cells with higher and lower EU incorporation into the nascent transcripts. We conducted cell cycle analysis by labeling newly synthesized DNA with BrdU and staining DNA contents with FxCycle Violet at the same time as using EU to label nascent transcripts. It was apparent that the peak with less EU was from the G₁ phase cells, while the peak with more EU was derived from S and G₂ phase cells (Figure 2B). To determine whether there was a detectable reduction in EU incorporation by flow cytometry after irradiation, we irradiated cells, waited different lengths of times before EU labeling of nascent RNA (Figure 2C). The transcriptional inhibition after irradiation was clearly detectable by flow cytometry as early as 15 min after IR, while transcriptional recovery was complete by 4 hr after IR (Figure 2D). Also, the extent of transcriptional repression was similar regardless of whether the IR dose was 2, 5, or 10 Gray (Figure 2D, Figure 2—figure supplement 1). This is consistent with the possibility that the reduction of nascent transcripts after IR is a programmed/signaling response rather than due to proximity of the genes to the DSB, which would have led to a dose-response.

To gain a better understanding of transcriptional inhibition after IR, we sought to identify the genes whose transcription was being inhibited after IR. We isolated EU labeled nascent total RNA transcripts 30 min after IR and prior to IR from two independent experiments (Figure 3—figure supplement 1) and sequenced the EU-RNA (Figure 3A). Prior to isolation of the ER labeled nascent RNA, we added equal amounts of the commercial ERCC spike-in to RNA from the same number of cells. This enabled the subsequent normalization of the total read number from the human genome to total reads from the ERCC control, to detect global changes between samples (Chen et al., 2015). We observed that the total read count of nascent transcripts declined after IR (Figure 3B). Most of the read counts were due to rDNA transcripts, and the decline in bulk transcripts after IR was mostly due to a significant decline in rDNA transcripts (Figure 3B, Figure 3—figure supplement 2A). By contrast, the total read count from the protein-coding transcripts significantly increased after IR (Figure 3B). Analysis of the protein-coding transcripts showed that the transcripts of 3026 and 1388 protein-coding genes increased and decreased after IR, respectively (Figure 3C and D, Supp. File 1). To validate our EU-RNA sequencing results, we performed quantitative RT-PCR to measure the nascent transcript levels of genes that were up- and down-regulated after IR. Consistently, we found rDNA transcripts of 28 S and 18 S were significantly downregulated after IR; while the p53-regulated gene Cyclin-dependent kinase inhibitor 1 (Cdkn1a/p21) was highly induced (Figure 3—figure supplement 3). The gene ontology terms describing the genes that were activated after IR included known DNA damage response pathways and related genes (Figure 3E, Supp. File 2). For example, the intrinsic apoptotic signaling pathway in response to DNA damage (GO:0008630), type 2 response (GO:0042092), and cytokine-mediated signaling pathway (GO:0019221) were up regulated significantly. Pro-inflammatory cytokines are the major components of immediate early gene programs, being rapidly activated after irradiation in various cell types (Schaue et al., 2012) ultimately leading to radiation-induced fibrosis in cancer patients following radiation therapy (Kim et al., 2014; Yu et al., 2023). Meanwhile, many of the genes that were downregulated after IR included gene products that are involved in chromatin organization and nucleosome assembly (Figure 3E, Supplementary file 3). For example, chromatin silencing (GO:0006342), DNA packaging (GO:0006323), and nucleosome assembly (GO:0006334) genes were down regulated significantly.

To gain more insights on how the transcription of protein-coding genes was regulated after IR, we defined differentially expressed genes (DEGs) between samples before and after IR. We found read number from these DEGs became significantly greater after IR (Figure 4A). We sorted the protein-coding DEGs by average expression level of each gene in all four samples: two replicate samples before IR and two replicates after IR. The number of reads derived from the mostly highly expressed protein-coding DEGs became significantly smaller after IR (Figure 4A and B). If the gene repression after IR is due to their being in cis to the DNA lesion, it would be expected that genes that were repressed after IR would tend to be longer, because they would be more likely to be damaged. However, this was not the case because the length of the nascent transcripts was equivalent regardless of whether their transcription was repressed, activated, or not changed after IR (Figure 4C). Intriguingly, we found that the repressed genes of the top 100 high-expression DEGs tended to be shorter (Figure 4C). Next, we inspected the expression level of individual protein-coding genes and confirmed that most changes in gene expression after IR tended to occur for the genes that were activated after IR, while many of the genes that had a high-expression level were repressed after IR, for example, the histone encoding genes (Figure 4D). Strikingly, we found that the vast majority of the histone genes showed reduced transcription after IR (Figure 4E, Figure 3—figure supplement 2B), and was validated by RT-PCR analysis (Figure 3—figure supplement 3). Finally, to determine whether transcriptional repression was occurring at the initiation, elongation, or both stages of transcription, we examined the read counts throughout the open reading frames of the repressed protein-coding genes, before and after IR. We found that the decrease of transcripts mainly occurred in the gene body of these genes with similar intensity at both the 3’ and 5’ ends of the gene body, which indicates transcriptional repression after IR occurred at the initiation stage of transcription (Figure 4F). Therefore, these data indicate that the bulk reduction in nascent transcripts after IR is mainly due to reduced transcriptional initiation of the rDNA and histone genes.

We performed a genome-wide gRNA library CRISPR-Cas9 screen in Abl pre-B cells, allowing 7 days for the gRNAs to inactivate their target genes (Figure 5A). We then sorted the 10% of the cells with the most nascent RNA (high EU) 30 min after IR, as these would include cells with gRNAs corresponding to gene products that are required for transcriptional inhibition after IR. We sequenced the gRNAs within the high EU cells and within the total cell population. We calculated an enrichment score for each of the 5 gRNAs that were included in the library against each gene (Supp. File 4, Figure 5—figure supplement 1). Figure 5B shows the enrichment scores for the 5 gRNAs for some of the top hits (high EU) from the screen. We found that gRNAs against ATM were enriched in the high EU cells, consistent with the fact that transcriptional inhibition of a gene after induction of a DSB requires ATM signaling (Shanbhag et al., 2010). Other top hits included most of the machinery that mediates neddylation, the post-translational covalent addition of NEDD8, a small ubiquitin-like peptide, onto other proteins (Rabut and Peter, 2008). These hits included the Nedd8 gene encoding the NEDD8 ubiquitin-like modifier, Nae1 encoding NEDD8 activating enzyme E1 subunit 1 NAE1, Uba3 encoding the Ubiquitin-like modifier activating enzyme 3 UBA3, Ube2m/Ubc12 encoding the NEDD8-conjugating enzyme UBC12, Ube2f encoding a neddylation E2 enzyme UBE2F, and Rbx1 and Rbx2 which encode linkers that facilitate NEDD8 transfer from the E2 enzyme to Cullins (Figure 5B). We also uncovered gRNAs against the gene encoding the neddylation substrate CUL4B enriched in the high EC cells (Figure 5B). These data suggested that neddylation may have a novel role in transcriptional inhibition after IR.

We validated the role of ATM in bulk transcriptional inhibition after IR with the ATM inhibitor Ku55933 (Figure 5C). To validate a role for neddylation in transcriptional shut off after IR, we depleted NAE1 with a gRNA to Nae1 (Figure 5—figure supplement 2) and showed that depletion of NAE1 also greatly reduced transcriptional inhibition after IR (Figure 5D). Similarly, a 16 hr treatment with the neddylation inhibitor MLN4924 also greatly reduced transcriptional inhibition after IR (Figure 5E). To ensure that the role of neddylation in transcriptional inhibition after IR is not unique to the murine Abl pre-B cells, we found that transcriptional inhibition, detected by fluorescence microscopy, in U2OS cells was significantly reduced upon inhibition of neddylation (Figure 5F). As such, these data suggest that neddylation is required for efficient transcriptional inhibition after IR.

While most neddylation substrates have only been identified upon overexpression of the neddylation machinery, the Cullins have been shown to be bone fide neddylation substrates in vivo (Pan et al., 2004). Given that we found gRNAs to Cul4b encoding CULLIN 4B enriched in the EU high cells (Figure 5B) we asked whether its neddylation is induced after IR. We found that when compared to the level of unneddylated CUL4A there was not a significantly higher proportion of neddylated CUL4A or CUL4B in U2OS cells after IR (Figure 6A). We used the neddylation inhibitor MLN4924 as a positive control to confirm which bands were neddylated CUL4A and CUL4B (Figure 6A). To determine whether CUL4B or CUL4A were required for IR induced transcriptional inhibition, we made stable Abl pre-B cell lines lacking each protein by gRNA-mediated disruption of the Cul4a or Cul4b genes. Loss of CUL4A had no effect on transcriptional inhibition after IR, while loss of CUL4B partially reduced transcriptional inhibition after IR (Figure 6B, Figure 6—figure supplement 1A). Given that CUL4B and CUL4A show some functional redundancy (Hannah and Zhou, 2015; Brown et al., 2015), we depleted both proteins at the same time. Additional transient bulk gRNA transfection-mediated depletion of CUL4A from cells lacking CUL4B (depletion of both CUL4A and CUL4B is lethal) did not further increase the block of transcriptional inhibition after IR (Figure 6—figure supplement 1B). As such, these data indicate that CUL4B but not CUL4A contributes to transcriptional inhibition after IR. Given that neddylation of CUL4A/CUL4B regulates cell cycle progression (Hannah and Zhou, 2015), and we saw changes in the distribution of the cell cycle phases upon CUL4A/CUL4B depletion and upon blocking neddylation as indicated by the change in relative heights of the EU low (G₁) and EU high (S/G₂) peaks (Figures 5D, E, 6B and Figure 6—figure supplement 1B and C), we wondered if cell cycle arrest may be influencing transcriptional inhibition after IR. Accordingly, we inhibited neddylation for 1–3 hr in Abl pre-B cells, which effectively inhibited neddylation (Figure 6C), and tested the effect on transcriptional inhibition after IR. Strikingly, we observed transcriptional inhibition after IR even upon neddylation inhibition treatment for 1–3 hr (Figure 6D), indicating that neddylation is not required for transcriptional inhibition after IR. We also observed that the length of time required for treatment with the neddylation inhibition to block transcriptional inhibition after IR (Figure 5E) caused arrest of the cell cycle in G₂ phase (Figure 6—figure supplement 2). Therefore, these data indicate that neddylation per se is not required for transcriptional inhibition after IR. They also raise the possibility that the cell cycle arrest caused by persistent loss of neddylation may prevent transcriptional inhibition after IR.

To directly determine the relationship between cell cycle arrest caused by prolonged neddylation inhibition and transcriptional inhibition after IR, we labeled the Abl pre-B cells with both EU (nascent transcripts) and 7-Aminoactinomycin D (7-AAD) (DNA stain). Prolonged neddylation inhibition led to accumulation of cells with a 4 N DNA content and greatly reduced transcriptional inhibition after IR (Figure 7A, Figure 6—figure supplement 2). To investigate if this correlation was specific to neddylation inhibition, we used an unrelated inhibitor that causes cell cycle arrest with a 4 N DNA content, RO-3306, a CDK1 inhibitor (Vassilev et al., 2006). We also saw accumulation of cells with a 4 N DNA content and no transcriptional inhibition after IR upon CDK1 inhibition (Figure 7B, Figure 6—figure supplement 2). To determine whether this effect was unique to cells arrested with a 4 N DNA content or was shared with conditions that cause arrest in G₁ phase, we treated cells with 10% or 0.1% FBS, where the later lead to arrest in G₁ phase (due to serum depletion) and prevented transcriptional inhibition after IR (Figure 7C). It is relevant to point out that the level of bulk nascent transcripts in cells arrested in G₁ by serum depletion is still far higher than cells lacking EU (Figure 7C), suggesting that the limited transcriptional inhibition after IR in G₁ arrested cells is not just because there is only minimal transcription occurring. In agreement, two CDK4/6 inhibitors, Ribociclib and Palbociclib, that lead to G₁ arrest also prevented transcriptional inhibition after IR (Figure 7D and E). Intriguingly, these data indicate that multiple different treatments that lead to cell cycle arrest in G₁ or G₂ per se prevent transcriptional inhibition after IR.

To gain molecular insight into why ATM-dependent bulk transcriptional inhibition occurs after IR in cycling cells but not in arrested cells, we investigated the possibility that ATM-dependent pathways that inhibit rDNA transcription after IR in cycling cells (Kruhlak et al., 2007) fail in arrested cells. In cycling cells, DSBs within the rDNA trigger the formation of nucleolar caps due to damaged rDNA and associated proteins relocalizing to the nucleolar periphery (van Sluis and McStay, 2017). In cycling cells, DSBs within the rDNA trigger ATM-dependent phosphorylation of Treacle which promotes recruitment of NBS1 and TOPBP1 to the nucleolar caps to inhibit rDNA transcription (Larsen et al., 2014; Mooser et al., 2020). Accordingly, in cycling cells, TOPBP1 accumulates in the nucleolar caps after induction of DSBs, concomitant with repression of rDNA transcription (Sokka et al., 2015; Mooser et al., 2020). We asked whether TOPBP1-eGFP recruitment to nucleolar caps was disrupted after DNA damage in cells arrested by treatment with MLN4924 or RO-3306. The overall induction of TOPBP1 expression was not affected by inhibitor treatment (Figure 7—figure supplement 1). However, the percentage of cells with TOPBP1-eGFP localizing to nucleolar caps after IR was markedly reduced upon MLN4924 and RO-3306 treatment (Figure 7F and G). These results are consistent with arrested cells failing to repress rDNA expression after global DNA damage, as a consequence of compromised ATM-dependent localization of TOPB1 to nucleolar caps.

U2OS cells (ATCC, HTB-96) were cultured in McCoy’s 5 A (Corning, 10050CV) medium supplemented with 10% fetal bovine serum (FBS) and 1% Penicillin-Streptomycin. Abelson virus-transformed pre-B cells were maintained in DMEM (Thermo Fisher, 11960–077) supplemented with 10% FBS, 1% Penicillin-Streptomycin, 1 x nonessential amino acids, 1 mM sodium pyruvate, 2 mM L-glutamine, and 0.4% beta-mercaptoethanol. HEK-293T cells were maintained in DMEM (Corning, 10–013 CM) supplemented with 10% FBS and 1% Penicillin-Streptomycin. All the cells were grown at 37  °C under a humidified atmosphere with 5% CO2.

SiRNA oligos against human H3F3A and H3F3B (SMARTPool) for RNAi in U2OS cells were purchased from Horizon Discovery (Dharmacon). 100 nM of H3F3A and H3F3B were mixed with Lipofectamine RNAiMAX transfection reagent (Thermo Scientific, 13778150) according to the manufacturer’s protocol to knockdown H3.3 for 48 hr. The siRNA control (ON-TARGETplus non-targeting) was also purchased from Horizon Discovery (Dharmacon) and used as a negative control. shRNA lentiviral plasmids against HIRA (5’-TAGAGCATACCAAGATGCC-3’) and the control were described in a previous study (Huang et al., 2018). HEK-293T cells were transfected with a mixture of shRNA plasmids and the viral packaging and envelope vectors, pCMV-dR8.2 and pCMV-VSVG. The media containing shRNA virus particles were collected 48–72 hr after transfection and filtered through a 0.45 μm filter. Cells were incubated with the lentiviral supernatant containing 5 μg/ml polybrene (Sigma-Aldrich, S2667) for 24 hr, followed by 1 μg/mL Puromycin selection for another 48 hr. To inactivate Nae1, Cul4a, and Cul4b in bulk cell populations, guide RNAs (gRNAs) against each gene were cloned into pKLV-U6gRNA-EF(BbsI)-PGKpuro2ABFP (Addgene, #62348) modified to express human CD2 or Thy1.1 as cell surface markers. The pKLV-gRNAs lentiviruses were prepared in 293T cells as described above. The Abl pre-B cells containing pCW-Cas9 (addgene, #50661), which can express Cas9 with doxycycline induction, were mixed with viral supernatant supplemented with 5 μg/ml polybrene and centrifuged at 1800 rpm for 1.5 hr at room temperature. After the spin-infection, the transduced cells were maintained in DMEM with 3 μg/ml doxycycline (Sigma-Aldrich, D9891) for 3 days before flow cytometric cell sorting based on hCD2 or Thy1.1 expression. To make stable cell lines depleted of Cul4a and Cul4b, serial dilution of the sorted cells into a 96-well plate was used to isolate single cells. Western blot analysis was used to determine the knockdown efficiency of each target gene.

In EU flow cytometric and immunofluorescence analysis, final concentrations of 15 μM ATM inhibitor Ku55933 (Selleck Chemicals, S1092) and 5 μM of Actinomycin D (Sigma-Aldrich, A9415) were added to cell culture 1 hr prior to irradiation. 1 μM and 10 μM of neddylation inhibitor MLN4924/Pevonedistat (Active Biochem, A-1139) were used for long (16 hr) and short (1–3 hr) treatments, respectively. To arrest cells in the G₁ cell cycle phase, cells were incubated in media supplemented with 5 μM Palbociclib (Selleck Chemicals, S1116) or 5 μM Ribociclib (Selleck Chemicals, S7440) for 24 hr. For serum starvation, Abl pre-B cells were grown in a complete medium containing 10% of FBS to the desired density and collected and washed in a medium with a reduced concentration (0.1%) of FBS. The cells were maintained in the medium with reduced FBS for 72 hr to arrest in the G₁ phase. Representative data are shown for experiments repeated three of more times with consistent results.

U2OS cell lines were authenticated by STR profiling, and MCF10A and murine cell lines tested negative for mycoplasma contamination.

The following antibodies were used for western blot: CUL4A (Cell Signaling Technology, 2699 S, 1:1000), CUL4B (Proteintech, 12916–1-AP, 1:1000), H3.3 (Millipore Sigma, 09–838, 1:1000), HIRA (Abcam, ab20655, 1:1000), NAE1 (Thermo Fisher, PA5-59836, 1:500), GAPDH (Sigma-Aldrich, G8795, 1:5000). Representative data are shown for experiments repeated three of more times with consistent results.

For immunofluorescence, the Click-iT RNA Alexa Fluor Imaging Kit (Thermo Fisher, C10330) was used to label newly synthesized RNAs in the cells. Briefly, 50,000 U2OS cells grown on cover slips in a 24-well plate were irradiated with 10 Gray and allowed to recover for indicated times at 37 °C with 5% CO2. 0.5 mM EU was added to the medium and incubated for 45 min for EU incorporation. Cells were then washed with PBS, fixed in 4% paraformaldehyde PBS for 15 min at room temperature, and permeabilized in cold 0.5% Triton X-100 PBS for 10 min. Cells were blocked in 3% BSA-PBS for 1 hr at room temperature and subsequently incubated overnight at 4 °C in primary antibody (anti-γH2AX (S139), EMD Millipore, 05–636). Coverslips were then washed 3 x with PBST (0.05% Tween 20), and incubated with a secondary antibody diluted in 3% BSA PBS (Alexa Fluor 488 Goat anti-mouse IgG, BioLegend, 405319) in the dark for 1 hr at room temperature and washed 3 x with PBST. Click-iT reaction cocktail was prepared according to the manufacturer’s protocol and immediately added to the cells to perform click reaction in the dark for 30 min at room temperature. After washes with Click-iT reaction rinse buffer (Component F) and PBS, cells were stained with Hoechst (1:2000) or DAPI (Sigma-Aldrich, D9542) in PBS for 10 min and mounted in Prolong Gold Antifade Mountant (Life Technologies, P-36930). Images were taken on Biotek Lionheart Automatic Microscope and EU intensity quantification was conducted using Biotek Gen5 software. For eGFP-TOPBP1 fluorescence microscopy, 50,000 cells grown on coverslips in 24-well plates were treated with doxycycline (1 μg/mL) for 12 hr to express eGFP-TOPBP1. MLN4924 (1 μM) and RO3306 (10 μM) were added to the cell culture, and the cells were incubated for another 16 hr followed by IR (10 Gy) and recovery for 2 hr. Cells were fixed in 4% Paraformaldehyde for 20 min followed by Hoechst (1:2000) staining for DNA and mounting in Prolong Gold Antifade Mountant. Images were taken and quantified on Biotek Lionheart Automatic Microscope. Representative data are shown for experiments repeated three of more times with consistent results.

Click-iT RNA Alexa Fluor Imaging Kit was adapted to label newly synthesized RNAs in the cells for flow cytometry. Abl pre-B cells grown in a 24-well plate were irradiated with 10 Gray and allowed to recover at different times. 2 mM EU was added to the medium and incubated for 30 min for EU incorporation. Cells were then washed with PBS, fixed in 4% paraformaldehyde PBS for 15 min at room temperature, and permeabilized in cold 0.5% TritonX-100 PBS for 5 min. Click-iT reaction cocktail was prepared according to the manufacturer’s protocol and immediately added to the cells to perform click reaction in the dark for 30 min at room temperature. Cells were then washed with Click-iT reaction rinse buffer (Component F) and 3% BSA-PBS, respectively. For cell cycle analysis, BrdU (10 ug/mL) was added to the cells and incubated for 30 min to label new DNAs. Cells were washed with PBS, fixed in 4% paraformaldehyde PBS for 15 min at room temperature, and permeabilized in cold 0.5% Triton X-100 PBS for 5 min. Cells were then digested with DNase (BD Biosciences, 51-2358KC) for 1 hr at 37 °C. Subsequently, cells were incubated with Alexa Fluor 488 Mouse anti-BrdU (BD Biosciences, 51–9004981, 1:500) in 3% BSA-FBS for 1 hr at room temperature and washed 2 x with PBS, followed by FxCycle Violet (Thermo Scientific, R37166) or 7-AAD (BD Pharmingen, 559925) staining for 10 min. Cells were resuspended in PBS and analyzed on BD LSRII Flow Cytometer or BD LSRFortessa Flow Cytometer. Flow cytometry results were further analyzed using FlowJo software. Representative data are shown for experiments repeated three of more times with consistent results.

More than 140 million wild-type Abl pre-B cells carrying inducible Cas9 transgene were transduced with a lentiviral gRNA library containing 90,230 gRNAs targeting over 18,000 mouse genes (Addgene, 67988) by spin-infection as described above. 3 days post-infection, cells transduced with gRNAs were sorted on a BD FACSAria II Cell Sorter based on BFP expression. BFP-positive cells were treated with 3 µg/ml doxycycline for 7 days to induce gRNA expression and gene inactivation. Cells were irradiated with 10 Gray, allowed to recover for 30 min, processed as described above for EU labeling of newly synthesized RNAs, and analyzed on BD FACSAria II Cell Sorter. Cells with high (top 10%) EU staining and unsorted cells were collected, and the genomic DNA of the cells were isolated for library preparation using nested-PCR. The library was sequenced on an Illumina HiSeq 2500 platform. Raw fastq files were demultiplexed by the Genomics and Epigenomics Core Facility of the Weill Cornell Medicine Core Laboratories Center. The gRNA sequence region was then retrieved from the sequencing data using Seqkit Shen et al., 2016 and mapped to the gRNA sequence library (Koike-Yusa et al., 2014; Tzelepis et al., 2016). The number of reads of each library sequence was counted and then normalized as follows Shalem et al., 2014. Normalized reads per gRNA = reads pers gRNA total reads for all sgRNAs in sample ×10⁶ + 1. Hereby, the generated normalized reads from each guide RNA were used and compared between the EU high cell and unsorted cell. P values were measured by the Poisson test to compare guide RNAs between EU high cells and unsorted cells. FDR was used for adjusting the p-value. CRISPR score = log2 (final sgRNA abundance/initial sgRNA abundance). The EU high genes were defined as these genes that have at least one guide RNA with p adjust value ≤ 0.01 & FC ≥ 1.5. Gene ontology (GO) analysis was performed by the R package cluster Profiler v3.18.1.

Click-iT Nascent RNA Capture Kit (Thermo Fisher, C10365) was used to label and capture the nascent transcripts. In brief, the same number of Abl pre-B cells was plated in two T-25 Polystyrene flasks, one for irradiation (10 Gray) and the other for no IR control. After irradiation, both flasks of cells were recovered for 30 min and incubated in a medium with 2 mM EU for 30 min to allow for the incorporation of EU into the nascent transcripts. Total RNAs were harvested using TRIzol reagent (Thermo Fisher, 15596018) following the manufacturer’s protocol. The click-iT reaction was performed as per the manufacturer’s protocol in 50 µL total volume for 30 min in the dark. Subsequently, 1 µL ultrapure glycogen, 50 µL 7.5 M ammonium acetate, and 700 µL of chilled 100% EtOH were added to the reaction. The mixture was incubated at −80 °C for 16 hr. RNA pellet was spun down at 13000×g for 20 min at 4 °C, washed 2 x with chilled 75% EtOH, and resuspended in nuclease-free water. EU-RNAs were pulled down with Dynabeads MyOne Streptavidin T1 magnetic beads and extracted with TRIzol reagents. The same amount of ERCC spike-ins (Thermo Fisher, 4456740) were added to the purified EU-RNAs, followed by cDNA library generation using NEBNext Ultr II Directional RNA Library Prep Kit for Illumina (NEB, E7760) according to the manufacturer’s protocol and deep sequencing on Illumina HiSeq 2500 platform.

EU-labeled RNA was prepared from the same number of cells as described above. All the isolated EU-RNAs were used for cDNA synthesis using Superscript III (Thermo Fisher, 18080044) reverse transcription with random hexamer as primers following the manufacturer’s protocol. The same proportion of cDNA products in each sample was used as a template for the quantitative RT-PCR reaction with Light Cycler 480 SYBR Green I Master Mix (Roche, 04707516001). The Ct value was used to represent the absolute amount of EU-RNAs in each sample, in which a smaller Ct value indicates higher nascent transcript levels of an individual gene, given the same initial number of cells and the same proportion of EU-RNAs were used for the analyses. Primer sequences for the analyses are as follows:

ms28S-fwd, 5’-TGGGTTTTAAGCAGGAGGTG-; ms28S-rev, 5’-GTGAATTCTGCTTCACAATG-3’(Watada et al., 2020); ms18S-fwd, 5’-CTTAGAGGGACAAGTGGCG-3’; ms18S-rev, 5’-ACGCTGAGCCAGTCAGTGTA-3’ (Stephens et al., 2011); msHist1h2ab-fwd, 5’-GCCTGCAGTTCCCCGTA-3’; msHisth2ab-rev, 5’- ATCTCGGCCGTCAGGTACTC-3’; msHist1h2ac-fwd, 5’-GGCTGCTCCGCAAGGGT-3’; msHist1h2ac-rev, 5’-CTTGTTGAGCTCCTCGTCGTT-3’; msH2afz-fwd, 5’- ACTCCGGAAAGGCCAAGACA-3’; msH2afz-rev, 5’-GTTGTCCTAGATTTCAGGTG-3’ (Nishida et al., 2005); msCdkn1a-fwd, 5’-GTGGCCTTGTCGCTGTCT-3’; msCdkn1a-rev, 5’-TTTTCTCTTGCAGAAGACCAATC-3’ (Béguelin et al., 2017).

The mouse genome version GRCm38.p6 release M23 and the associated GENCODE version of the mouse reference gene set were downloaded from the GENCODE website (https://www.gencodegenes.org/mouse/release_M23.html). We trimmed adapter sequences and low-quality sequences in RNA-seq data using the Trim Galore v0.6.6 (Martin, 2011) with default parameters. To avoid rRNA homologous sequences (i.e., in the intron regions of Zc3h7a or Cdk8) prior to subsequent genomic and other RNA analysis, we first mapped the reads to mm10 rDNA sequences by TopHat v2.1.1 (Kim et al., 2013). The unmapped reads were then further mapped to the mouse genome version mm10 and ERCC spike-in version ERCC92 using TopHat v2.1.1 (Kim et al., 2013). Successfully mapping reads were sorted by SAMtools v1. 5. Afterward, read counts in several types of genomic feature, i.e., protein-coding genes, rDNA, and ERCCs (ERCC92.gtf), were quantified by Htseq-count v0.11.2 (Anders et al., 2015) using the union gene region option. The read number per gene was normalized based on the total ERCC read numbers in each sample.

To visualize read coverage across the genome, DeepTools v3.5.0 (Ramírez et al., 2014) was used to convert BAM files into bigwig files using scale factors calculated by the total ERCC read number in each sample. Next, DeepTools was used to plot the average read depth per sample across interested groups of genomic regions (i.e. repressed protein-coding genes from 3 kb upstream to 3 kb downstream of gene bodies). Screenshots of read density at individual regions were generated by IGV 2.8.13 (Thorvaldsdóttir et al., 2013).

We then used one tail Poisson test to evaluate the difference in gene expression level based on the read counts normalized by total ERCC read counts. We defined differentially expressed RNAs as those with a fold change greater than 1.5 and an FDR value smaller than 0.05. To detect highly expressed genes, we ranked genes by RPKM in the control cells, whereas RPKM was calculated using ERCC-normalized read counts further normalized by gene length. GO analysis was performed by the R package clusterProfiler v3.18.1 (Wu et al., 2021). Heatmap were generated by pheatmap.