DNA repair drives cisplatin-induced neuronal death Authors: William J. Nathan, Chuanyuan Chen, Rosy Sakr, Bruno Siqueira Mietto, Vincent van Batenburg, Jeroen van den Berg, Josette J. Wlaschin, Ferenc Livak, Elsa Callen, Nancy Wong, Eliza Y.H. Lloyd, Hanna Silberberg, Sushma Sharma, Raj Chari, Tzipporah Freeman, Baek Kim, Alexander van Oudenaarden, Alexander T. Chesler, Michael E. Ward, Lisa D. Boxer, Peter J. McHugh, Andrei Chabes, Claire E. Le Pichon, André Nussenzweig Published: 2026 ABSTRACT SUMMARYPlatinum agents are cornerstone therapies for many cancers but often cause neurotoxicity in post-mitotic tissues, for which effective interventions are lacking. This limitation reflects an incomplete understanding of neuronal responses to DNA damage. We show that nucleotide excision repair (NER) mediates cisplatin lesion removal in neurons; however, unlike its protective role in dividing cells, NER promotes neuronal death in response to cisplatin. This vulnerability arises because neurons possess low deoxynucleoside triphosphate (dNTP) pools. dNTPs are initially consumed during transcription-coupled NER to resolve transcription-blocking lesions. As dNTP levels become depleted, repair fails to complete, leading to accumulation of double-strand breaks, particularly during global-genome NER. Supplementation with deoxynucleosides or genetic upregulation of dNTP synthesis restores nucleotide pools, protects neurons from cell death, and reduces cisplatin-induced neuropathic pain. These findings identify limited dNTP availability as a key vulnerability in post-mitotic cells and suggest nucleoside supplementation as a potential strategy to mitigate chemotherapy-induced neurotoxicity. INTRODUCTION Platinum agents, such as cisplatin, are among the most frequently used cytotoxic cancer treatments. DNA damage from these agents inflicted across the genome causes cell death in rapidly dividing cells by inhibiting the essential process of DNA replication.1 While these agents were developed for their strong cytotoxicity in dividing cells, they also can promote cell death in post-mitotic cells, leading to significant, dose-limiting side effects in primarily post-mitotic tissues such as the nervous system (peripheral neuropathy) and the inner ear (ototoxicity).2–4 There is a lack of basic knowledge on how DNA damage from platinum or alkylating agents kills post-mitotic cells, and thus, there is no clear treatment to prevent or lessen the toxic effects of these chemotherapeutic agents. Cisplatin binds to nucleotide bases covalently, creating mainly intrastrand crosslinks between adjacent purines (~90%–95% of lesions) and more rarely bulky monoadducts (approximately 5% of lesions) or interstrand crosslinks (2% of lesions).5–8 These DNA adducts distort the DNA helix and block the essential cellular activities of replication and transcription; therefore, their removal is believed to be crucial for continued cellular viability.7,8 During replication, platinum interstrand crosslinks are repaired by the Fanconi Anemia pathway,9 whereas monoadducts and intrastrand crosslinks are removed by nucleotide excision repair (NER).10–12 NER consists of two subpathways, global-genome (GG) and transcription-coupled (TC) repair. GG-NER involves lesion recognition throughout the genome by a complex featuring the xeroderma pigmentosum group C (XPC) protein.13 TC-NER specifically removes DNA damage that blocks transcription and is initiated by binding of the Cockayne syndrome B (CSB) protein to stalled RNA polymerase.14,15 Both damage recognition mechanisms converge by recruiting the NER core complex, featuring XPA, the transcription initiation factor IIH (TFIIH) complex, RPA, and the endonucleases XPG and ERCC1-XPF, which mediate dual incision around the lesion to excise an approximately 26-nucleotide patch of DNA containing the damage.16,17 Finally, DNA repair synthesis and ligation fill and seal the excised gap.18 NER has been primarily studied using DNA damage induced by ultraviolet (UV) irradiation. In these studies, the relative contributions of TC-NER and GG-NER differ depending on the cell type and proliferation or differentiation status.19,20 Rapidly dividing embryonic stem cells or early embryos have been shown to be more reliant on GG-NER than on TC-NER for survival upon UV irradiation, whereas TC-NER is believed to be dominant in post-mitotic somatic cells.21 Because neurons focus their repair activity on actively transcribed genes rather than on the intergenic genome, it has been suggested that DNA damage accumulating outside of genes is largely ignored and therefore should not have deleterious consequences for proper cell functionality.22,23 To what extent GG-NER is active in post-mitotic neurons and what causes the switch from GG-NER to TC-NER during differentiation remains unclear. In this study, we clarify how cisplatin lesions are repaired in post-mitotic neurons and determine that NER paradoxically causes post-mitotic cell death due to limiting endogenous pools of deoxynucleoside triphosphates (dNTPs) needed for repair. dNTP pools are consumed initially by TC-NER to resolve transcription-blocking lesions from the transcribed strand of active genes. As dNTPs become exhausted, toxic double-strand breaks (DSBs) then accumulate across the genome, mainly because of incomplete GG-NER. Supplementation of dNTPs through pharmacological or genetic means improves neuronal resistance to cisplatin and reduces neuropathy in cisplatin-treated mice. We therefore suggest nucleoside supplementation as a strategy to protect patients from chemotherapy-induced side effects in post-mitotic tissues. RESULTS Measuring DNA repair of cisplatin lesions in human neurons In post-mitotic cells, the thymidine analog 5-Ethynyl-2’-deoxyuridine (EdU) is incorporated in the nuclear genome at sites of active DNA repair.24–26 To determine how cisplatin-DNA adducts are repaired in post-mitotic cells, we developed a sensitive EdU Click-iT-based assay to directly visualize DNA repair events in individual cells (Figure S1A).27 We treated induced pluripotent stem cell (iPSC)-derived glutamatergic neurons (i3Neurons) with 20 μM cisplatin and EdU for 18 h. Following cisplatin treatment, wild-type (WT) neurons showed pan-nuclear EdU incorporation (Figure 1A, bottom). Non-treated (NT) and cisplatin-treated i3Neurons also had cytoplasmic EdU signal likely derived from mitochondrial replication (Figure 1A, middle and inset). To measure nuclear-specific DNA repair events, we isolated nuclei from NT and cisplatin-treated neurons and measured DNA repair synthesis by flow cytometry. Nuclei from neurons treated with cisplatin showed 10-fold greater EdU incorporation than NT nuclei (Figure 1B). The much lower level of EdU incorporation in NT cells (Figure 1B) is likely due to long-patch base excision repair within neuronal enhancers24 or other endogenous DNA repair events (see Figure S1E). While we observed an increase in overall DNA repair levels between 5 and 20 μM of cisplatin, DNA repair became saturated at doses greater than 20 μM (Figure S1B). We next sought to measure where in the genome DNA repair was occurring in cisplatin-treated neurons. To this end, we developed a single-cell synthesis-associated with repair sequencing (scSAR-seq) assay, an expansion of scEdU-seq.24,28 SAR-seq detects and genomically maps sites of EdU incorporation in non-dividing cells. Single-cell sequencing allows quantification of reads across the genome in a single cell, whereas bulk sequencing (in this case, standard SAR-seq) provides only information about the relative enrichment of reads in genomic regions across a population of cells, thereby obscuring non-recurrent damage due to cell-to-cell variation. In the absence of exogenous damage, most i3Neurons had fewer than 2,000 scSAR-seq reads, whereas following cisplatin treatment, most i3Neurons had between 2,000 and 20,000 reads, while some had as many as 50,000 reads (Figure 1C). We were also able to detect scSAR-seq reads originating from mtDNA replication, with cisplatin treatment leading to a decrease in mtDNA reads (Figure S1C). By defining cells as being ‘‘EdU-positive’’ if they had greater than 2,000 scSAR-seq reads (EdU levels above background), we concluded that approximately 75% of i3Neurons were undergoing DNA repair synthesis in response to 20 μM cisplatin treatment (Figure S1D). In NT i3Neurons, scSAR-seq reads were generally located stochastically throughout the genome (Figure 1D, top), though upon aggregation, we were able to detect enrichment at neuronal enhancers, as previously reported in bulk SAR-seq experiments (Figure S1E).24,25 Upon cisplatin treatment, scSAR-seq reads were enriched in genes and correlated with RNA expression levels (Figures 1E and 1F). Within genes, scSAR-seq signal was depleted near transcription start sites (TSSs), mirroring previous reports that NER is inhibited at TSSs of active genes (Figure 1E),29 and otherwise was generally flat across the gene body, with elevated signal also beyond the transcription end sites (TESs) due to closely proximate neighboring genes (Figure S1F). Additionally, we were able to detect elevated DNA repair events outside active genes in the intergenic space following cisplatin treatment (Figures 1D and S1G). Of the total scSAR-seq reads in cisplatin-treated cells, approximately half were located on the transcribed strand of active genes (mean 42%), while the remaining were located in either intergenic space (32%) or on the non-transcribed strand (26%) (Figure S1H). Bulk SAR-seq in i3Neurons treated with cisplatin concentrations ranging from 5 to 50 μM confirmed enrichment of EdU signal in gene bodies as in scSAR-seq, irrespective of cisplatin concentration (Figures S2A and S2B). CDK7 or CDK9 inhibition greatly reduced SAR-seq gene body enrichment, indicating that DNA repair of cisplatin lesions within gene bodies is dependent on active transcription (Figures S2C and S2D). To examine the kinetics of DNA repair in individual cells, we compared sites of EdU incorporation within the first 7 h after cisplatin treatment versus 17–24 h following the start of cisplatin treatment using scSAR-seq. After 24 h of cisplatin treatment, scSAR-seq signal showed a similar genomic distribution as described above (Figures 1D and S2E). In the first 7 h of DNA repair, we observed enrichment of EdU in transcribed strands that correlated with gene expression (Figures 1G and S2F). After 17 h, DNA repair of cisplatin decreased in gene bodies, was less correlated with gene expression levels, and instead was greatest in non-transcribed intergenic regions (Figures 1G and S2F). We conclude that in neurons treated with cisplatin, repair is first concentrated within highly transcribed genes, followed by repair of lowly transcribed and non-transcribed genomic regions. NER acts on cisplatin-DNA lesions in human neurons We sought to understand whether NER repairs cisplatin-DNA damage in post-mitotic human neurons just as it does in other cell types.10–12 Using CRISPR-interference,30 we knocked down the NER core complex factors XPG and XPA in i3Neurons (Figures S3A–S3D) and found significantly lower levels of DNA repair compared with WT as measured by flow-cytometry-based detection of EdU (Figure 2A). Thus, NER does indeed repair cisplatin lesions in neurons. We next determined the relative contributions of TC-NER and GG-NER to cisplatin lesion repair by examining EdU incorporation in neurons deficient in CSB or XPC, respectively. We treated CSB−/− or XPC knockdown (KD) i3Neurons (Figures S3E–S3G) and primary cortical neurons derived from mice deficient in CSB31 or XPC (Figure S3H) with cisplatin and EdU. Relative to WT, XPC deficiency only reduced the EdU incorporation by 15%–20% 24 h post-cisplatin treatment, much less than the 50%–80% observed in CSB knockout (KO) (Figures 2B and 2C). However, combined loss of CSB and XPC in mouse primary cortical neurons and human i3Neurons reduced EdU incorporation upon cisplatin treatment to levels observed in NT cells (Figures 2B and 2C). Next, we used scSAR-seq to determine where CSB-mediated repair was occurring. As in flow-cytometry-based detection of EdU, loss of CSB reduced cisplatin-induced scSAR-seq signal per cell by roughly half compared with WT i3Neurons, whereas loss of XPG (and thus total NER) largely eliminated DNA repair synthesis (Figure 2D). CSB KO specifically reduced reads on the transcribed strand within gene bodies, proportionally shifting repair predominantly to intergenic regions (Figures 2E and 2F). We conclude that repair of cisplatin in neurons is largely CSB-dependent and transcription-coupled, though GG-NER also remains active to remove lesions genome-wide. Therapeutic concentrations of cisplatin induce cell death in neurons via NER Therapeutic doses of cisplatin, which can reach up to 30 μM in serum,32 are known to induce long-term side effects in post-mitotic tissues, likely due to cell death.33–35 To clarify the mechanisms by which neurons die upon cisplatin treatment, we treated WT i3Neurons with a range of doses of cisplatin and determined that 20 μM or greater initiates cell death within a 2–3 day period (Figure 3A). WT primary mouse cortical neurons were more sensitive and required only 5 μM of cisplatin to initiate cell death within that same time period (Figure S3I). The fact that cisplatin treatment led to prominent cell death was unexpected, as cisplatin-DNA damage was also being robustly repaired in neurons (Figures 1 and 2). Because EdU imaging and scSAR-seq demonstrated that NER was responsible for removing cisplatin-DNA lesions, we expected that NER deficiency in neurons would lead to increased sensitivity to lethal doses of cisplatin, as is the case in dividing cells (Figure 3B).10,36 Unexpectedly, KD of core NER factors XPA, XPG, or ERCC1 in i3Neurons (Figures S3J and S3K) or KO of XPA in primary mouse cortical neurons (Figures S3L and S3M) made cells resistant to doses of cisplatin lethal to WT cells (Figures 3C and 3D). While combined deficiency in CSB and XPC in i3Neurons and primary mouse cortical neurons led to cisplatin resistance similar to deficiency in core NER factors, loss of CSB and thus TC-NER alone did not lead to cisplatin resistance (Figures 3E and 3F). Interestingly, however, KD or KO of the XPC GG-NER factor did moderately improve neuronal survival after cisplatin, particularly at high doses (Figures 3E and S3N). We conclude that, in contrast to dividing cells, DNA repair activity, led by GG-NER, drives cell death in neurons following cisplatin treatment. Neurons treated with cisplatin fail to complete DNA repair While NER-dependent DNA repair synthesis was taking place in neurons treated with cisplatin, the resistance of NER-deficient cells to cisplatin clearly showed that this repair activity was also toxic. We hypothesized that NER events were being initiated but were not ultimately going to completion. Incomplete NER, in which the NER nucleases ERCC1-XPF and XPG coordinate and initiate removal of the damaged section of DNA,37 but DNA repair synthesis does not fill in the resultant gap, would lead to persistent DNA single-strand breaks or gaps that might be toxic. To test this, we measured chromatin-bound RPA,38 which coats single-stranded DNA at sites of DNA gaps during NER.39,40 Cisplatin induced a dose-dependent increase in RPA binding in WT i3Neurons, and KD of core NER factors XPG or XPA significantly reduced chromatin-bound RPA (Figures S4A–S4C). Moreover, cisplatin treatment induced an increase in PARylation within the nucleus, a marker of single-strand breaks (SSB) formation,41 which was dependent on XPG (Figure S4D). This suggested that NER was directly responsible for processing cisplatin lesions into prolonged single-stranded DNA. If NER failed to complete repair at two closely separated DNA gaps on opposite strands, this might create a DNA DSB. Indeed, cisplatin treatment of neurons induced co-localization of γ-H2AX and 53BP1 foci, markers of DNA DSBs (Figures 4A and 4B). γ-H2AX foci did not form immediately after cisplatin treatment (Figure S4E). This suggested that cisplatin itself was not breaking the sugar-phosphate backbone, but instead that cisplatin-DNA lesions were being processed into DNA breaks. Similar to its role in generating SSBs, we hypothesized that NER activity was responsible for DSB formation. Indeed, cisplatin-induced γ-H2AX, such as chromatin-bound RPA and PARylation, was dependent on the core NER factors XPA, XPG, and ERCC1 in i3Neurons (Figures 4C and 4D) and on XPA in mouse primary cortical neurons (Figure S4F). We sought to determine the individual contributions of TC-NER versus GG-NER to cisplatin-induced DNA breaks. Loss of TC-NER via CSB KO did not significantly reduce γ-H2AX levels in i3Neurons or primary mouse cortical neurons, while deficiency in GG-NER via XPC KD or KO did (Figures 4E and S4G). Loss of both CSB and XPC led to an even greater decrease in cisplatin-induced γ-H2AX levels (Figures 4E and S4G). This mirrored the effects that loss of these genes had on cisplatin-induced cell death, with XPC loss resulting in greater resistance (and decreased γ-H2AX formation) than CSB loss, while loss of both pathways led to phenotypes comparable to deficiency of core NER factors (Figures 3C, 3E, and 4D). Thus, although TC-NER is the primary repair mechanism of cisplatin-induced DNA damage in neurons, GG-NER is the primary driver of DNA break formation and cell death. To map cisplatin-induced DSBs genome-wide, we used END-seq.42 In WT neurons treated with cisplatin, the END-seq signal was elevated relative to NT neurons and present throughout the genome (Figure 4F). In contrast, the END-seq signal was not elevated in cisplatin-treated neurons deficient for both CSB and XPC (Figure 4F). To directly examine the relative contributions of TC-NER and GG-NER, we compared END-seq signals in neurons deficient in CSB or XPC, respectively. CSB−/− neurons had less END-seq signal in expressed genes and instead more END-seq signal in intergenic space compared with WT (Figure 4G). In contrast, END-seq in XPC KD neurons treated with cisplatin revealed a global decrease in signal both inside and outside of genes, confirming GG-NER’s dominant role in DSB formation (Figure 4G). To detect the presence of cisplatin-induced SSBs in addition to DSBs, we performed S1-END-seq in parallel with END-seq. S1 treatment cleaves single-stranded DNA into DSBs, which are subsequently detected by END-seq.24 S1-END-seq detected significantly more signal than END-seq in both expressed genes and in intergenic space, indicating the presence of both SSBs and DSBs in these regions (Figure S4H). In sum, these data demonstrate that cisplatin neuronal toxicity is associated with DNA breaks driven primarily by toxic NER activity. Nucleotide pool exhaustion underlies the sensitivity of human neurons to cisplatin and other bulky DNA damage To reconcile how NER could promote both DNA repair and cell death in response to cisplatin, we hypothesized that DNA synthesis during NER might not go to completion in certain regions of the genome, thereby leading to sustained break formation and cell death. To model whether incomplete NER indeed contributes to toxicity, we treated i3Neurons with the DNA polymerase inhibitor aphidicolin to block NER-associated DNA repair synthesis. Aphidicolin treatment eliminated cisplatin-induced EdU incorporation (Figure 5A). Cells treated with aphidicolin were also acutely more sensitive to a normally sub-lethal dose of cisplatin (5 μM) and had significantly greater levels of γ-H2AX (Figures 5B and 5C). Thus, blocking the completion of NER-initiated DNA synthesis can drive neuronal death. Following incision and removal of the damaged DNA segment by the ERCC1-XPF and XPG nucleases, repair synthesis fills in the resultant ~26-nucleotide gap. Among DNA repair pathways, NER is the most consuming of dNTPs. Post-mitotic cells have very low levels of dNTP pools compared with cycling cells.43–45 Indeed, we determined that i3Neurons had 10-fold lower dNTP levels compared with iPSCs (Figure 5D).43 We then compared dNTP levels in untreated and cisplatin-treated neurons and found that cisplatin treatment further reduced dNTP levels (Figure 5E), indicating the consumption of dNTP pools during DNA repair. Interestingly, dATP and dGTP were most depleted, reminiscent of the depletion pattern of ribonucleotide reductase (RNR) inhibitor hydroxyurea (HU) and deficiency in the RNR subunit RRM2B,46 which is responsible for de novo dNTP synthesis in non-dividing cells. We therefore hypothesized that DNA repair in neurons treated with lethal doses of cisplatin was not going to completion due to a shortage of dNTPs necessary to perform the required amount of DNA repair synthesis. Indeed, over time, a steady decrease in the levels of dATP and dGTP in neurons treated with cisplatin correlated with an increase in γ-H2AX (Figure 5F). To test whether low levels of dNTPs limit the completion of DNA repair in neurons, we sought to modulate dNTP pools. First, we questioned whether reducing dNTP production in neurons would increase their sensitivity to cisplatin. Indeed, treating neurons with HU increased γ-H2AX levels after cisplatin treatment and sensitized them to normally non-lethal doses of cisplatin (Figures 5G and 5H). We then queried whether increasing dNTP levels in neurons would do the opposite and lead to cisplatin resistance. As dNTPs cannot passively cross the cell membrane,47 we supplemented cellular media with deoxyribonucleosides (dNs), which are phosphorylated to dNTPs within the cell.48 Supplementation of i3Neurons or primary mouse cortical neurons with dN almost completely rescued their sensitivity to cisplatin (Figures 6A, 6B, and S5A) and decreased cisplatin-induced γ-H2AX and chromatin-bound RPA (Figures 6C, S5B, and S5C). Moreover, transcriptome analysis revealed that neurons treated with cisplatin and dN retained their neuronal identity (Figure S5D). We also confirmed neuronal health following dN supplementation by measuring nascent transcription and mitochondrial function. Neurons only completely recovered nascent transcription and maintained mitochondrial health after cisplatin treatment in the presence of dN (Figures S5E and S5F). Addition of individual dNs did not greatly improve neuronal viability following cisplatin treatment, and, to the contrary, addition of individual purine dNs increased neuronal cisplatin sensitivity (Figure 6D). Interestingly, supplementation of cycling iPSCs with dN did not rescue their sensitivity to cisplatin, which demonstrates that dNTPs are only limiting in non-dividing cells, such as neurons (Figure 6E). SAMHD1 is a phosphohydrolase that plays a critical role in regulating intracellular dNTP pools, particularly under conditions of limited dNTP synthesis, such as in post-mitotic cells.46,49 Genetically increasing dNTP pools by KD of SAMHD150 in i3Neurons or KO of SAMHD1 in primary mouse cortical neurons51 also increased neuronal cisplatin resistance (Figures 6F, S6A, and S6B) and reduced DNA break formation as determined by γ-H2AX staining (Figures S6C and S6D). Interestingly, degradation of the SAMHD1 protein in i3Neurons by the human immunodeficiency virus type 2 virion-associated protein Vpx52,53 improved neuronal cisplatin resistance (Figure S6E). Finally, we modulated dNTP pools by altering the expression of RRM2B.46 Overexpression of RRM2B (Figure S6F) increased i3Neuron cisplatin resistance (Figure 6G), whereas KD of RRM2B (Figure S6G) increased i3Neuron cisplatin sensitivity (Figure 6G). Altogether, these data demonstrate that low endogenous dNTP levels limit neurons’ ability to complete DNA repair following cisplatin treatment, and that increasing dNTP levels is a strategy to increase cisplatin resistance in neurons (see Figure 7G). To test whether low levels of endogenous dNTP pools are a vulnerability of post-mitotic cells to other types of DNA damage, we examined the impact of dN supplementation on i3Neurons following treatment with UV irradiation or methyl methanesulfonate (MMS). As with cisplatin, UV-DNA lesions are removed by NER, whereas MMS damage is primarily repaired by base excision repair.54,55 UV-irradiation-induced cell death in i3Neurons and supplementing neurons with dN after irradiation (and thus after induction of DNA lesions) greatly improved cellular resistance (Figures S7A and S7B). Additionally, neurons supplemented with dN had significantly fewer DNA breaks following UV treatment compared with neurons grown in standard medium, as measured by γ-H2AX immunofluorescence (Figure S7C). Similarly, dN supplementation reduced MMS-induced DNA breaks (Figure S7D). These results indicate that dNTP pools are insufficient for repair of various types of DNA lesions in post-mitotic cells. Genetic upregulation of dNTP pools reduces cisplatin-induced peripheral neuropathy Cisplatin therapy is associated with severe, dose-limiting side effects, including peripheral neuropathy, which arises from the effects of platinum on post-mitotic cells in the peripheral nervous system. It is estimated that up to 85% of patients experience chemotherapy-induced peripheral neuropathy after receiving platinum-based chemotherapy.56,57 Cisplatin-induced neurotoxicity is primarily caused by its impact on the dorsal root ganglion (DRG), clusters of sensory neurons that project from the spinal cord and innervate the extremities.34 To extend our in vitro models, we therefore sought to examine the repair of cisplatin in peripheral DRG neurons in vivo. To determine whether cisplatin-DNA damage is repaired in peripheral neurons in vivo, we applied our EdU flow cytometry assay in nuclei isolated from DRG of mice treated with cisplatin and EdU. Imaging flow cytometry demonstrated our ability to visualize an increase in nuclear DNA repair synthesis in WT DRG neurons after cisplatin treatment (Figure 7A). DRG neurons occasionally also had non-nuclear EdU signal likely derived from mitochondria still associated with the nuclear membrane (Figure 7A). Standard EdU flow cytometry showed that both neuronal and non-neuronal nuclei from the DRG of cisplatin-treated WT mice had an increase in DNA repair synthesis (Figures 7B and S7E). However, neuronal and non-neuronal nuclei from DRG of cisplatin-treated XPA-deficient mice had no such increase in DNA synthesis (Figures 7C and S7F). Thus, repair of cisplatin-DNA lesions does occur in peripheral neurons in vivo via NER, as in neurons in vitro. Moreover, non-neuronal cell types in the peripheral nervous system, likely satellite glial cells that surround the cell bodies of neurons, also repair cisplatin-induced DNA damage via NER. Daily injections of a clinically relevant dose of cisplatin (1.5–2.8 mg/kg) into adult mice is a well-established model for inducing peripheral neuropathy58 (Figure 7D). Neuropathy is commonly quantified by hypersensitivity to touch using von Frey filaments on the paw (Figure 7D). If insufficient dNTP pools necessary to complete the DNA repair of neuronal damage play a role in cisplatin-induced peripheral neuropathy, loss of SAMHD1 should reduce these symptoms. WT mice showed the expected mechanical sensitization after 3 and 5 days of treatment with 1.5 mg/kg cisplatin, as revealed by increased responses to normally innocuous filaments (Figure 7E; Table S1). Strikingly, SAMHD1-deficient animals developed significantly less hypersensitivity relative to controls (Figure 7E; Table S1). At day 3 and day 5 of treatment, SAMHD1−/− mice maintained approximately 2-fold and 3-fold greater mechanical withdrawal threshold compared with WT animals, respectively, indicating attenuation of cisplatin-induced hypersensitivity (Figure 7F). SAMHD1-deficient mice also showed better recovery than WT mice 5 days following the cessation of treatment (Figures 7E and 7F). This demonstrates dNTP depletion as one of the drivers of cisplatin-induced peripheral neuropathy and supports augmenting dNTP pools in vivo as a therapeutic strategy toward protection against cisplatin-induced peripheral neuropathy. DISCUSSION Here, we identify NER activity and a shortage of dNTPs as drivers of cell death in post-mitotic cells treated with cisplatin. Under normal physiological conditions, post-mitotic cells do not require large pools of dNTPs to maintain their genomes, but under conditions of high levels of exogenous DNA damage, such as when a patient is undergoing chemotherapy, the low levels of dNTPs in post-mitotic cells become an acute vulnerability. We observed that genetically increasing dNTP levels in mice reduced the severity of cisplatin-induced neuropathy and improved recovery following the cessation of treatment. Increasing dNTP levels further by pharmacological means may provide even greater resistance to neuropathy in vivo, just as supplementation of neurons with dN improved neuronal cisplatin resistance in vitro. While supplementation of neurons with increased levels of dNTPs provides a durable resistance to cisplatin (Figure 6A), total loss of NER does not. Indeed, we see that after approximately 4 days of cisplatin exposure, eventually NER-deficient neurons do begin to die (Figure S7G). This is the same time-frame for cell death to be induced by transcription inhibition (Figure S7H), suggesting that even when NER-dependent DNA breaks are not formed, transcription inhibition by unrepaired cisplatin lesions within active genes is toxic. Thus, unlike with increasing dNTP levels, which improve neuronal health long-term during cisplatin treatment (Figure 6A), total inhibition of NER does not. If and why neurons and other terminally differentiated cells focus or even exclusively repair the transcribed regions of the genome through TC-NER has been a mystery.22,23 Here, we show that DNA repair outside of genes does take place in neurons via GG-NER, though repair occurs at a faster rate within genes. Interestingly, loss of GG-NER alone (via XPC depletion), but not TC-NER, was able to improve neuronal survival in response to high doses of cisplatin (Figure 3E). This is despite TC-NER contributing a greater proportion of NER activity compared with GG-NER, as measured by EdU incorporation (Figures 2B and 2C). Thus, there is a paradox: GG-NER is less active in neurons, but it nonetheless contributes more to cell death. This leads to the following model: cisplatin adducts are first recognized and repaired in active genes by TC-NER. This activity protects important highly transcribed areas of the genome but consumes the finite pool of dNTPs in neurons. Unlike unicellular organisms, in which DNA damage causes a large increase in dNTP pools,59,60 mammalian cells do not increase dNTP production in response to DNA damage.61 As time progresses, GG-NER identifies and removes cisplatin adducts across the genome. However, in an environment of depleted dNTP pools, there are too few available dNTPs to perform gap fill-in DNA synthesis at these sites. Consequently, DNA gaps and breaks form and persist genome-wide, which inevitably trigger cell death (Figure 7G). In the absence of XPC and GG-NER, fewer DNA breaks form, and more dNTPs can be committed to DNA repair in active genes by TC-NER, which promotes longer-term cell survival. This model might also explain the observation that loss of C. elegans xpc-1 provides greater protection against UV-induced apoptosis in pachytene-stage oocytes compared with loss of csb-1, whereas combined deficiency of xpc-1/csb-1, such as ablation of core NER, provides full protection to these post-mitotic cells.20 In the absence of CSB and TC-NER, it follows that more dNTPs are available for GG-NER, and thus, there should be fewer DNA breaks detected by γ-H2AX immunofluorescence. We nonetheless detect DNA breaks at the same levels in CSB−/− neurons as in WT neurons (Figures 4E and S4G), and CSB−/− neurons die at a similar rate to WT neurons in response to cisplatin (Figure 3E). When we examined the kinetics of EdU incorporation in cisplatin-treated CSB−/− neurons (proficient only in GG-NER), we found that EdU incorporation gradually increases to levels observed in WT neurons after 48 h (Figure S7I). Thus, GG-NER processes cisplatin adducts slower than TC-NER, consuming the dNTP pool more slowly, but thoroughly enough to lead to DSB formation and cell death by 48 h post-treatment. Indeed, we found that addition of dN in CSB−/− neurons reduced γ-H2AX levels, confirming that available dNTP pools are still insufficient to complete GG-NER (Figure S7J). An additional explanation for why GG-NER is more toxic than TC-NER is that it acts on both DNA strands and thus is more likely to lead to toxic DSBs when repair patches are not filled in (Figure 7G). In contrast, TC-NER acts only on one DNA strand (the transcribed strand), making DSB generation less frequent. Currently, there are no clear preventative treatments for platinum-induced neuropathy or ototoxicity, though some have argued that antioxidants such as N-acetylcysteine (NAC) and the recently clinically introduced sodium thiosulfate (STS) should be further explored as treatments to reduce cisplatin-induced side effects.62 There are questions over the mechanism of action of these antioxidants on cisplatin, with reports that the thiol groups on NAC or STS interact with and deactivate cisplatin itself.63 Thus, these antioxidant therapies can pose an increased risk to the patient’s overall survival through negating the positive anti-tumor effects of cisplatin.64,65 Notably, our findings that dN supplementation promotes survival not only in response to cisplatin but also after infliction of UV damage indicate that dN does not act by simply quenching DNA-damaging agents. Moreover, dN supplementation did not reduce cisplatin’s cytotoxic effects in dividing cells (Figure 6E), suggesting it could spare healthy post-mitotic cells while not interfering with its anti-cancer effects. We therefore believe that targeting the biological mechanism behind cisplatin sensitivity in post-mitotic cells may provide a rational treatment paradigm. Low levels of dNTPs in neurons may also have implications in normal aging and neurodegeneration in the brain. Recent sequencing studies of postmortem brains have revealed a steady increase in DNA damage and mutagenesis associated with aging.66–68 If dNTPs became exhausted, or if dNTP production decreased with age, there could be an increased frequency of neuronal death or other downstream consequences following endogenous or exogenous DNA damage. When dNTP pools become low, one consequence could be an increased rate of ribonucleotide incorporation within the genome. Indeed, a recent study demonstrated that in neurodegenerative disorders, there is evidence for loss of neuronal RRM2B expression (and thus lower dNTP levels) and increased mutations associated with erroneous ribonucleotide incorporation within nuclear genomic DNA.69 Additionally, recent work identified that a decrease in dNTP levels is associated with increased ribonucleotide incorporation into mitochondrial DNA and a greater inflammatory phenotype in senescent cells associated with aging.70 In summary, our work identifies a unique feature of the DNA damage response in post-mitotic cells and suggests that exogenous dN supplementation or other means to increase dNTP levels in patients undergoing chemotherapy may be a strategy to protect post-mitotic cells in the nervous system from cell death. Limitations of the study Cytotoxic chemotherapy agents differ from one another in the broad range of DNA lesions they inflict on cells. While our work with cisplatin forms a foundation for understanding how DNA-binding agents are toxic to neurons, different chemotherapy agents may be toxic to neurons through additional mechanisms. Furthermore, we identify increasing nucleotide pools as a strategy to reduce platinum-induced neuropathy in mice. While this may improve neuronal health, the effects on the immune system by cisplatin may also contribute independently to neuropathy. Finally, the effects of nucleotide pool augmentation on other neurotoxic phenotypes, such as ototoxicity, should be investigated in future studies. RESOURCE AVAILABILITY Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, André Nussenzweig (nussenza@exchange.nih.gov). Materials availability All unique and stable reagents, plasmids, and animals generated in this study will be shared by the lead contact with a materials transfer agreement. Data and code availability Microscopy data reported in this paper will be shared by the lead contact upon request. Microscopy data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. This paper does not report original code. All scSAR-seq, SAR-seq, END-seq, S1-END-seq, and RNA-seq datasets have been deposited at GEO (accession numbers GEO: GSE329935 and GEO: GSE329361) and are publicly available as of the date of publication. All scSAR-seq, SAR-seq, END-seq, S1-END-seq, and RNA-seq datasets have been deposited at GEO (accession numbers GEO: GSE329935 and GEO: GSE329361) and are publicly available as of the date of publication. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. STAR★METHODS EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS Mice All animal procedures were conducted according to the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals, under NCI Animal Study Proposal no. 23–042 and NICHD Animal Study Proposals 23–003 and 26–003 approved by the NIH Animal Care and Use Committee. C57BL/6J mice were obtained from the Jackson Laboratory. XPA and XPC-mutant mice were generated as follows. Guide RNAs targeting early and late into the protein coding sequence of murine Xpa and Xpc were designed using sgRNA Scorer 2.0.73 Upon testing multiple candidates in vitro for cutting activity, guide RNAs for Xpa (5’-AAGCGGGCUCUUGAAGUGUG-3’; 5’-GGCUGCGCGGCCCUACCCCG-3’) and Xpc (5’-GCAGAGGCAAAGCAACCGCC-3’; 5’-GCACCGCAGACGGAAGGCGG-3’) were chosen for subsequent mouse knockout experiments. Synthetically modified versions of the guide RNAs (Synthego) were then complexed with purified recombinant Cas9 protein and injected into C57/BL6 NCr mouse embryos using methodology previously described.74 CSB-mutant mice were obtained from the Laboratory of Vilhelm Bohr.31 SAMHD1 mice were obtained from the Laboratory of Jan Rehwinkel via the Laboratory of Andrei Chabes.51 CSB−/−/XPC−/− mice were generated by intercrossing. Mice were housed under a 12-h light–dark cycle at room temperatures maintained between 20 and 26 °C and a relative humidity of 30–70%. For the isolation of embryonic cortical neurons, timed pregnancies were set up using heterozygote. Vaginal plug detection was performed the following morning to confirm the presence of a plug. iPSC cell culture The iPSC WTC11 cell line derived from a healthy male was obtained from the Coriell cell repository and all policies of the NIH Intramural Research Program for the registration and use of iPSC cell lines were followed. iPSC cell culture was performed as previously described.24,26 In brief, tissue culture plates were coated with matrigel and then, following matrigel removal, cells were plated in Essential 8 Medium (E8) supplemented with 10 μM ROCK inhibitor (Ri) and grown in a 37°C incubator at 5% CO2. The medium was replaced the following day with E8 medium without Ri and cells were allowed to grow until passaging. Cells were passaged using accutase for dissociation (10 minutes at 37°C), washed with PBS, and replated onto fresh matrigel-coated plates in E8+Ri medium. For viability experiments, 20,000 cells were plated per-well onto 24-well plates in E8+Ri medium and then treated the following day during a media change to E8 medium. i3Neuron differentiation and cell culture i3Neuron differentiation and cell culture was performed as recently described.24,26 The WTC11 human iPSC cell line was previously engineered to express mouse neurogenin-2 (NGN2) via a doxycycline-induced promoter that was integrated into the AAVS1 safe harbor site. Additionally, these cells were engineered to express CAG-dCas9-BFP-KRAB at the Clybl promoter for use in targeted gene knockdowns. To differentiate iPSCs into neurons, iPSCs were grown on matrigel pre-coated plates in N2 medium (KnockOut DMEM/F12 medium; Life Technologies Corporation, cat. No. 12660012) with N2 supplement (Life Technologies, cat. No. 17502048), 1x MEM nonessential amino acids (Thermo Scientific, cat. No. 11140050), 1x GlutaMAX (Thermofisher Scientific, cat. No. 35050061), 10 μM ROCK inhibitor (Y-27632; Selleckchem, cat. No. S1049), and 2 μg/mL doxycycline (Clontech, cat. No. 631311). N2 medium was removed and changed daily for two additional days. 3 days post-plating, cells (‘‘day3 neurons’’) were dissociated in accutase for 10–15 minutes at 37°C to obtain single-cell suspension, washed in PBS, and replated onto new plates freshly coated with poly-L-ornithine (PLO) (0.1 mg/mL; Sigma, cat. No. P3655) in BrainPhys medium (STEMCELL Technologies, cat. no. A3582801) to complete differentiation. BrainPhys medium was supplemented with B27 Plus Supplement (ThermoFisher Scientific, cat. no. 450–03), 1 mg/mL mouse laminin (Sigma, cat. no. L2020–1MG), 10 ng/mL BDNF (PeproTech, cat. no. 450–02), 10 ng/mL NT-3 (PeproTech, cat. no. 450–03), and 2 μg/ml doxycycline (Clontech, cat. no. 631311). For SAR-seq experiments, 10–15 million day3 neurons were plated onto 10-cm plates. For scSAR-seq and all FACS experiments, 2–3 million day3 neurons were plated per well onto 6-well plates. For immunofluorescence imaging experiments, 50,000 d3 neurons were plated per well onto μ-Slide 18 well glass bottom slides (Ibidi, 81817). For viability experiments, 200,000 d3 neurons were plated per well onto 24-well plates. For nascent RNA imaging experiments, 500,000 d3 neurons were plated per well onto 12-well plates. Usually, cells were treated on day 6 (3 days after plating on PLO-coated plates) unless otherwise noted. In cases where neurons were grown longer-term, half media changes were done every 3–4 days. The following drugs were frequently used to treat cells. Cisplatin (Sigma Aldrich, 479306) was dissolved in PBS. 2’-deoxycytidine (dC) (Sigma Aldrich, D3897) and thymidine (dT) (Sigma Aldrich, T9250) were dissolved in PBS, and 2’-deoxyguanosine monohydrate (dG) (Sigma Aldrich, D7145) and 2’-deoxyadenosine monohydrate (dA) (Sigma Aldrich, D7400) were dissolved in DMSO. Equal amounts of these nucleosides were combined to obtain ‘‘dN’’ solution. EdU (Vector Laboratories, CCT-1149) was dissolved in PBS. Briefly, for treatment of i3Neurons with cisplatin, 5 μM was used as a low dose (sub-lethal in WT), 20 μM as a medium dose (lethal in WT), and 75 μM as a high dose (faster cell death). For functional assays (eg. EdU imaging/FACS or γ-H2AX immunofluorescence staining), 20 μM was used as a standard dose (unless otherwise indicated) in i3Neurons because this dose begins to overwhelm nucleotide pools and induces cell death in WT cells 24–48 hours after cisplatin treatment. 5 μM cisplatin was used as a sub-lethal dose that becomes toxic in the presence of reduced dNTPs (eg. HU treatment). When dNTPs are elevated (eg. SAMHD1 KD, Vpx treatment, RRM2B overexpression), improved survival can be seen even at high doses of cisplatin (75 μM). Primary cortical neuron cell culture Embryonic day 16.5 mouse embryo cortices were dissected as previously described.75 The amniotic sac was opened, embryos were decapitated, and meninges were removed prior to isolation of cortices in sterile dissection medium (DM; Hanks’ balanced salt solution (HBSS, Thermo Fisher, 14170161), 10 mM HEPES-NaOH, pH 7.0, 10 mM MgCl2 and 1 mM kynurenic acid). Cortices were digested using papain (Roche, 10108014001) and DNase I (Sigma-Aldrich, DN25) for 10 min at 37 °C. Digested cortices were washed three times with DM containing 10 mg ml–1 ovomucoid trypsin inhibitor (Worthington Biochemical, LS003086) and BSA (Sigma-Aldrich, 9647), once with DM containing 1 mg ml–1 ovomucoid trypsin inhibitor and BSA, and once with cold Neurobasal Plus Medium (Thermo Fisher, A3582901). They were then triturated 20 times in Neurobasal ‘‘complete’’ medium (Neurobasal Plus Medium supplemented with B-27 Plus (Thermo Fisher, A3582801), CultureOne (Thermo Fisher, A3320201), GlutaMax (Thermo Fisher, 35050061) and penicillin–streptomycin (Thermo Fisher, 15140122)) and cells were strained through a 40-μm nylon filter (Falcon, 352340). Neurons were seeded onto plates coated with poly-d-lysine hydrobromide (Sigma-Aldrich, P7405) and laminin (Thermo Fisher, 23017015) in Neurobasal ‘‘complete’’ medium at the following densities: 200,000 cells/well in 24-well plates; 1 million cells per well in 6-well plates; 50,000 cells/well in 18-well Ibidi slides. The medium was replaced the day after seeding with medium supplemented with aphidicolin (4 μM, Sigma-Aldrich, 178273) to eliminate dividing cells in culture. For longer term culture, culture was refreshed with 1/3 media changes every seven days. For treatment of primary mouse neurons with cisplatin, 5 μM was used as a medium dose (lethal in WT) and 20 μM as a high dose. METHOD DETAILS PCR confirmation of gene knockout in mice Genomic DNA was isolated from tails and CRISPR target sites were amplified by polymerase chain reaction. For PCR of the Xpa gene, the following primers were used: Forward, 5’-ACTGCCTCGCCCTTCATAGG-3’, Reverse, 5’-CTACCCCACCAGACACCACT-3’, Internal, 5’-AGCTCTGCACCATAGACGTG-3’. For PCR of the Xpc gene, the following primers were used: Forward, 5’-GATCTACGTCGTCCGCCATG-3’, Reverse, 5’-GTTGGGTCTTGGAAGCACTC-3’, Internal, 5’-TCCCTCCGTCTTGGTTTTGG-3’. For PCR of the Samhd1 gene, the following primers were used: KO-Forward, 5’-ACGAAGTTATATTGTACCTGACTGAT-3’, WT-Forward, 5’-TTTGGTTCTAGGCACTCTTCTG-3’, Reverse, 5’-TCTCCTTATGGATGGTTGTAAGCC-3’. PCR products were separated by electrophoresis on 2% agarose gels. Mechanical sensitivity assessed by von Frey testing For von Frey mechanical sensitivity assays, adult mice were administered 1.5 mg/kg cisplatin dissolved in normal saline daily for five days. All injections were i.p. All behavioral testing was performed with the experimenter blinded to genotype. Mice were placed individually on an elevated wire mesh platform within inverted glass chambers. Opaque barriers were positioned between chambers to prevent visual interaction, and animals were allowed to habituate for at least 10–15 minutes prior to testing. Only mice of the same sex were evaluated within a given session, and testing order was randomized on each test day. Littermates of the same sex were randomly assigned to experimental groups. Mechanical sensitivity was measured using calibrated von Frey filaments (0.008, 0.02, 0.04, 0.07, 0.16, 0.4, 0.6, and 1 g), following established procedures76 with minor adaptations. Testing began with the 0.4 g filament. Each filament was applied to the plantar surface of the hind paw 10 times, with a minimum inter-trial interval of 10 seconds. Responses were considered positive when stimulation elicited a rapid paw withdrawal, paw shaking, or directed licking behavior. All filaments were tested sequentially, and responses were recorded for scoring. Lentivirus production and CRISPRi knockdown For lentivirus production, Lenti-X-HEK293T cells were plated onto 6-well plates at 2 million cells per well in DMEM containing 10% FBS (Sigma, TMS-013-B) and cultured overnight to reach 90% confluency. The following day, cells were transfected using Lipofectamine 3000 (Life Technologies, L3000150). 150 μL room temperature Opti-MEM (Fisher Scientific, 31985062) and 3.75 μL Lipofectamine 3000 were combined and allowed to incubate at room temperature for 5–10 minutes. Separately, 150 μL room temperature Opti-MEM, 5 μL of P3000, 1.2 μg of the lentiviral vector of interest, 0.8 μg psPAX2, 0.3 μg of pMD2G, and 0.1 μg pAdVantage were combined and then added to the Opti-MEM + Lipofectamine 3000 mixture. After incubating the complete mixture at room temperature for approximately 30 minutes, it was added dropwise to a well of Lenti-X cells. The next day, Lenti-X cell medium was replaced with fresh DMEM/FCS with 6 μL ViralBoost (Alstem, VB100). 2–3 days later, the cell medium was carefully collected, centrifuged to remove any cells or debris, and then concentrated using Lenti-X Concentrator as per the manufacturer’s instructions (Takara Bio, 631231). Concentrated virus was resuspended in 300 μL PBS and aliquoted and stored at −80°C. For CRISPRi knockdown in iPSC, 500,000 iPSCs containing the CAG-dCas9-BFP-KRAB construct (see i3Neuron differentiation and cell culture above) were transduced with an aliquot of lentivirus containing a pU6-sgRNA EF1Alpha-puro-T2A-BFP vector (gift from J. Weissman; Addgene, 60955) targeting the gene of interest. Cells were treated with accutase to remove them from the plate, pelleted, and then resuspended in lentivirus for two minutes before replating onto fresh matrigel-coated plates in E8 + Ri medium. The following day, media was removed and replaced with E8 medium. The next day (two days post-transduction), medium was removed and replaced with E8 medium with 10 μg/mL puromycin. Cells underwent puromycin-selection for 1–2 days before removing the replacing the medium with fresh E8 medium and grown normally or differentiated into i3Neurons. Guide RNAs were expressed under the mouse U6 promoter. The following gRNA sequences were used: sgXPA, 5′-GGGAGCTAGGTCCTCGGAGT-3′; sgERCC5, 5′-GAGTGGGGGACGGCACTAAT-3′; sgERCC1, 5′-GGACTGCAGAGGGATCGAGG-3′; sgXPC, 5′-GCAGGGCGGGAGTCCAGACG-3′; sgSAMHD1, 5′-GAGGCGGAGGAAGAAGACAT-3′; sgRRM2B, 5′-GGTGCTGTCGTAGTTGGAGG-3′. CRISPR knock-out For CRISPR knockout in iPSCs with Ngn2 and the CAG-dCas9-BFP-KRAB cassettes, iPSCs were transiently transfected by electroporation with LentiCRISPRv2 vectors (Addgene, 52961) containing sgRNAs targeting the gene of interest. Two days after transfection, cells were selected for with 4 μg/mL puromycin. Clones were screened for successful knock-out by PCR and Western blot. sgRNAs cloned into LentiCRISPRv2 used to target CSB were: 5′-ATGTCGGTCGATGTGCAGCA-3′; 5′-GAAGGAGTATCGG TCGGTCC-3′; and 5′-CGTCTATGACCAGGACGTGC-3′. EdU Click-IT imaging i3Neurons or primary mouse cortical neurons were treated with or without 20 μM EdU and/or 20 μM cisplatin for 24 hours as indicated. Neurons were then washed once with PBS and fixed with 4% paraformaldehyde in PBS for 10 minutes at room temperature. Fixed neurons were permeabilized in 0.2% Triton X-100 in PBS for 8 minutes and then washed with PBS before the EdU Click-it reaction was added (Click-iT EdU kit (Thermo Fisher Scientific, C10337) with 1 μM AZDye 488 Picolyl Azide (Vector Laboratories, CCT-1276) instead of the standard kit-provided azide) for 30 minutes at room temperature. Click-it reaction mix was removed, neurons were washed twice with PBS, once with PBS and DAPI, and then resuspended in PBS for imaging. Images were captured on a Stellaris confocal microscope (Leica). EdU FACS Neurons were treated with 20 μM EdU and the indicated concentration of cisplatin for ≈24 hours. Neuronal nuclei were harvested by addition of lysis buffer (10 mM Tris pH 7.5, 10 mM NaCl, 3 mM MgCl2, 0.1% Igepal CA-630), scraping, and resuspending by repeated pipetting. Nuclei were washed once with lysis buffer, once with PBS and fixed with BD Cytofix/Cytoperm buffer (BD Biosciences, 554714) for 20 minutes on ice. Nuclei were washed with PBS and resuspended in 100 μL of click-it reaction mix (Click-iT EdU kit (Thermo Fisher Scientific, C10337) with 1 μM AZDye 488 Picolyl Azide (Vector Laboratories, CCT-1276) instead of the standard kit-provided azide) for 30 minutes at room temperature. Nuclei were washed with PBS and then resuspended in 1% BSA in PBS with DAPI and analyzed using a CytoFLEX flow cytometer (Beckman Coulter). Flow cytometry data was analyzed using FlowJo (v.10). Single-cell SAR-seq i3Neurons were treated with 20 μM cisplatin for 18–24 hours with 20 μM EdU added for the indicated times (either for the full cisplatin treatment (two independent experiments) or for hours 1–7 or 17–24 of cisplatin treatment). Neurons were harvested as single cells by incubating with accutase for 20 minutes and resuspending by repeated pipetting. Cells were washed with PBS, resuspended in 1 mL PBS, and then fixed by dropwise addition of 4 mL ice cold methanol under mild vortexing. Fixed cells were stored overnight at −20°C. The next day, the methanol solution was removed, and cells were resuspended in storage buffer (10% DMSO, 20 mM HEPES pH 7.5, 150 mM NaCl, 2 mM EDTA, EDTA-free protease inhibitors) and kept at −20°C until processing. Processing of neurons for scSAR-seq was performed as previously described in detail for scEdU-seq.28 Bulk SAR-seq i3Neurons were treated with 20 μM EdU and the indicated concentration of cisplatin for 24 hours. For inhibiting transcription elongation, neurons were also treated with 1 μM THZ1 or 100 μM DRB. Cells were harvested as single cells by incubating with accutase for 15–20 minutes and resuspending by repeated pipetting. Cells were washed with PBS, resuspended in 1 mL PBS, and then fixed by dropwise addition of 4 mL ice cold methanol under mild vortexing. Fixed cells were stored at −20°C. Bulk SAR-seq was performed as previously described.24 Fixed cells were washed with PBS, permeabilized with 0.2% Triton X-100/PBS for 10 minutes on ice, washed with PBS, and then resuspended in the Click-iT reaction mix for 2 hours shaking at room temperature to label EdU with biotin. The Click-iT reaction mix was made by adding the following to PBS (in order): 3 mM copper sulfate (Sigma), 50 μM biotin azide (Thermo Fisher, B10184), 1x Click-iT additive (Thermo Fisher, C10424). Cells were washed in PBS and then incubated in 50 mM Tris pH 8.0 with 1% SDS and proteinase K overnight at 37 °C. The following day, DNA was isolated using UltraPure phenol:chloroform:isoamyl alcohol (25:24:1, v/v) (Invitrogen) according to the manufacturer’s instructions, followed by 2.5:1 volume ethanol and 1:10 volume sodium acetate precipitation. Precipitated DNA was resuspended in TE buffer and sheared to approximately 200-bp fragments using a Covaris S220 sonicator at 10% duty cycle, 175 peak incident power, 200 cycles per burst, for 240 seconds. DNA was then precipitated in 2.5:1 volume ethanol and 1:10 volume sodium acetate and resuspended in 70 μL TE buffer. Biotin-conjugated DNA fragments were isolated using MyOne Streptavidin C1 Beads (Thermo Fisher, 650–01). 35 μL of beads were washed twice with 1 mL 1x wash and binding buffer (1x W&B; 10 mM Tris-HCl pH 7.5, 1 mM EDTA, 1 M NaCl, 0.1% Tween 20) on a DynaMag-2 magnetic separator (Invitrogen, 12321D) and resuspended in 70 μL of 2x W&B (10 mM Tris-HCl pH 7.5, 2 mM EDTA, 2 M NaCl, 0.2% Tween 20). This was added to the fragmented sample DNA and incubated at 24°C while in a ThermoMixer while shaking for 30 minutes. Dynabeads (now bound to DNA) were washed three times in 1 mL of 1x W&B, twice in 1 mL EB, and once in 1 mL 1x T4 DNA ligase buffer (NEB). Dynabeads were resuspended in 50 μL end-repair reaction mix (1x T4 DNA ligase buffer, 2.7 U T4 DNA polymerase (NEB), 9 U T4 Polynucleotide Kinase (NEB), 1 U Klenow fragment (NEB), and 0.4 mM dNTPs) and incubated with shaking at 24°C for 30 minutes. Beads were washed once in 1 mL 1x W&B, twice in 1 mL EB, once in 1 mL NEBuffer 2 (NEB), and resuspended in 50 μL A-tailing reaction mix (1x NEB dA-Tailing Buffer with 20 U Klenow fragment exo- (NEB)). Samples were incubated at 37°C with shaking for 30 minutes. Dynabeads were then washed in 1 mL NEBuffer 2 and resuspended in 115 μL ligation reaction mix (1x Quick Ligase Buffer (NEB), 6,000 U Quick Ligase (NEB), 5 nM annealed TruSeq truncated adapter) and incubated with shaking at 25°C for 20 minutes. 50 mM EDTA was added to terminate the reaction and Dynabeads were then washed three times with 1 mL 1x W&B and three times with 1 mL EB. Beads were resuspended in 8 μL EB + 10 μL 2x Kapa HiFI HotStart Ready Mix (Kapa Biosciences). Primers (10 mM) 5′-CAAGCAGAAGACGGCATACGAGATxrefGTGACTGG AGTTCAGACGTGTGCTCTTCCGATC*T-3′ and 5′-AATGATACGGCGA CCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T-3′ (asterisk indicates a phosphothiorate bond and a NNNNNN TruSeq index sequence) were added to 37 μL PCR reaction mix (20 μL 2x Kapa HiFI HotStart Ready Mix added to 17 μL H2O) for a final volume of 60 μL. DNA was amplified using the following PCR parameters: 98°C (45 seconds); 16 cycles of 98°C (15 seconds), 63°C (30 seconds), 72°C (30 seconds); 72°C (5 minutes). The finished PCR reaction was cleaned using 1.8x volume AMPure Beads XP (Beckman Coulter), following the manufacturer’s instructions. Samples were then separated on a 2% agarose gel, with 150–200 bp DNA being excised and purified using a QIA-quick Gel Extraction Kit (Qiagen). Library concentrations were measured using a KAPA Library Quantification Kit for Illumina Platforms (Kapa Biosystems) and then sequenced using Illumina NextSeq 550 or NextSeq 2000 instruments. CellTiter-Glo luminescent cell-viability assay Mature i3Neurons or primary mouse cortical neurons were treated with cisplatin and other drugs during a half media change. iPSCs were treated with cisplatin when changing media from E8+Ri to E8 media. For UV cell viability experiments, media was removed and replaced with PBS, cells were irradiated, and then PBS was replaced with media supplemented with or without dN. To measure viability, half the media was removed and replaced with CellTiter-Glo reagent (Promega, G8461) as per the manufacturer’s instructions. After incubation at room temperature while shaking, samples were analyzed in triplicate on a Tecan Infinite 200 Pro plate reader. Viability was normalized to non-treated neurons. γ-H2AX and 53BP1 immunofluorescence microscopy Unless specified, i3Neurons or primary cortical neurons were treated overnight (approximately 18 hours), with the indicated drugs added during a half media change. Cells were fixed with 4% paraformaldehyde in PBS for 10 mins at room temperature, washed with PBS, and permeabilized with 0.2% Triton-X 100 in PBS for 8 mins. They were then washed with PBS and blocked with 3% BSA in PBS for 30 mins at 37°C before staining with primary antibodies diluted in 3% BSA for 1 hr at room temperature. Neurons were then washed with PBS and stained with secondary antibodies diluted in 3% BSA in PBS 1 hour. Finally, cells were washed with PBS with DAPI and then imaged. Quantification of γ-H2AX nuclear staining was done at 40x magnification on a LionHeart LX microscope (BioTek Instruments). Confocal images were captured on a Nikon SoRa spinning disk microscope equipped with a 60x apo TIRF oil immersion objective lens (1.49 NA). The following primary antibodies and dilutions were used: anti-phospho-histone H2AX (Ser139) (1:5,000, Millipore, clone JBW301, 05–636), anti-53BP1 (1:1,000, Novus Biologicals, NB100–305), anti-TUBB3 (1:1000, BioLegend, clone TUJ1, #98130, conjugated in-house to CF488A using the Mix-n-Stain CF Dye Antibody Labeling Kit (Biotium, 92273)). The following secondary antibodies and dilutions were used: Alexa Fluor 488 goat anti-mouse (1:1,000, Invitrogen, A11029); Alexa Fluor 568 goat anti-rabbit (1:1,000, Invitrogen A11011). γ-H2AX, RPA32, or PAR FACS Neurons were treated with the indicated drugs for ≈24 hours. i3Neurons were harvested as single cells by incubating with accutase for ≈20 minutes and resuspending by repeated pipetting. Primary cortical neuron nuclei were harvested by addition of lysis buffer (10 mM Tris pH 7.5, 10 mM NaCl, 3 mM MgCl2, 0.1% Igepal CA-630), scraping, and resuspending by repeated pipetting. For RPA32 staining, single neuron suspension was washed once with PBS and pre-extracted with 0.05% Triton X-100 in PBS on ice for 10 minutes before fixation in 2% paraformaldehyde in PBS for 10 minutes at room temperature in the dark. For γ-H2AX staining, whole cells or nuclei were not pre-extracted and were instead fixed directly in 2% paraformaldehyde. Following fixation, cells or nuclei were washed in PBS, blocked with 1% BSA in PBS, and then incubated with primary antibody for 1 hour on ice in saponin buffer. Cells or nuclei were washed and then incubated with secondary antibody for 1 hour at room temperature. Following staining, cells or nuclei were resuspended in PBS with DAPI and analyzed using a CytoFLEX flow cytometer (Beckman Coulter). Flow cytometry data was analyzed using FlowJo (v.10). The following primary antibodies and dilutions were used: anti-phospho-histone H2AX (1:1,000, Millipore, clone JBW301, 05–636), anti-RPA32 (1:1,000, Cell Signaling Technology, clone 4E4, 2208), anti-poly-ADP-ribose binding reagent (1:1000, Millipore Sigma, MABE1031). The following secondary antibodies and dilutions were used: Alexa Fluor 647 chicken anti-mouse (1:1,000, Invitrogen, A21463); Alexa Fluor 647 goat anti-rat (1:1,000, Invitrogen, A21247); Alexa Fluor 647 goat anti-rabbit (1:1000, Invitrogen, A21245). Western blotting Cells were collected and lysed in RIPA buffer: 10 mM Tris-HCl (pH 7.5), 0.1% SDS, 0.1% sodium deoxycholate, 1% Triton X-100, EDTA-free protease inhibitor cocktail tablet (complete Mini, Roche Diagnostics). Equal amounts of protein per sample were resolved by SDS-PAGE using NuPAGE 4–12% Bis-Tris Gels (Invitrogen, NP0321). Protein was transferred from gels onto nitrocellulose membranes, blocked with Intercept Blocking Buffer in TBS (LI-COR, 927–60001), and incubated with primary antibodies diluted in Intercept Blocking Buffer overnight at 4°C. The next day, membranes were washed with TBS-tween and incubated with secondary antibodies diluted in Intercept Blocking Buffer for at least 1 hour at room temperature. Membranes were imaged using Image Studio (v.5.2) on a LI-COR Odyssey CLx instrument. The following primary antibodies were used at the following dilutions: anti-CSB (1:1000, Abcam, ab96089); anti-XPA (1:1000, Abcam, ab85914); anti-XPG (1:1000, Bethyl, A301–484), anti-ERCC1 (D-10) (1:1000, Santa Cruz, sc-17809), anti-XPC (D1M5Y) (1:1000, Cell Signaling, #14768), anti-RRM2B (EPR8816) (1:1000, Abcam, ab154194); anti-β-tubulin (1:10,000, Sigma-Aldrich, T5168). The following secondary antibodies were used at the following dilutions: IRDye 800CW goat anti-rabbit IgG (1:10,000, LI-COR, 926–32211); IRDye 680RD goat anti-mouse IgG (1:10,000, LI-COR, 926–68070). END-seq and S1-END-seq 10 million d7 i3Neurons were treated with or without 20 μM cisplatin for 24 hours and harvested by treatment with accutase for 20 minutes and gentle pipetting to reach approximate single-cell suspension. Neurons were counted and spike-in cells (G1-arrested Lig4−/− mouse pre-B cells expressing a site-specific zinc-finger endonuclease42) were added at 10% of cell count. Cells were embedded in 1% agarose plugs and processed as previously described in detail.42,77 Agarose-embedded cells were lysed and digested using Proteinase K (50°C for 1 hour, then 37°C for 7 hours). Plugs were washed in TE buffer and treated with RNase A at 37°C for 1 hour. For S1-END-seq, plugs were then washed in EB and equilibrated in S1 nuclease buffer (40 mM sodium acetate pH 4.5, 300 mM NaCl, 2 mM ZnSO4) for 30 minutes. Plugs were added to 100 μL S1 nuclease buffer containing 1.8 U S1 nuclease and incubated for 15 minutes on ice to allow the enzyme to diffuse into the plug. Plugs were then incubated with the reaction mix at 37°C for 20 minutes before addition of EDTA (10 mM final concentration) to terminate the reaction. After this, the standard END-seq protocol was followed. Plugs were then washed first in TE buffer, then in ExoVII Buffer (NEB), before incubating with ExoVII (NEB, M0379S) in ExoVII buffer for 1 hour at 37°C. Plugs were washed twice in NEBuffer 4 and then incubated with ExoT (NEB, M0265S) in NEBuffer 4 for 1 hour at 24°C. After blunting, plugs were washed twice in NEBNext dA-Tailing Reaction Buffer (NEB, B6059S) and then incubated with Klenow exo- (NEB, M0212S) in dA-Tailing Reaction Buffer at 37°C for 30 minutes. Plugs were washed once in NEBuffer 2 and incubated with END-seq hairpin adapter 1 and NEB Quick Ligase in NEB Quick Ligation buffer for 20 minutes at 25°C. Agarose plugs were then melted and digested with β-Agarase I (NEB) and DNA was sonicated to approximately 200 bp using a Covaris S220 sonicator for 4 min at 10% duty cycle, peak incident power 175, 200 cycles per burst, 4°C. DNA was ethanol-precipitated and dissolved in 70 μL TE buffer. Biotin-containing DNA fragments were isolated using MyOne Streptavidin C1 Beads (Thermo Fisher, 650–01). 35 μL of beads were washed twice with 1 mL 1x wash and binding buffer (1x W&B; 10 mM Tris-HCl pH 7.5, 1 mM EDTA, 1 M NaCl, 0.1% Tween 20) on a DynaMag-2 magnetic separator (Invitrogen, 12321D) and resuspended in 70 μL of 2x W&B (10 mM Tris-HCl pH 7.5, 2 mM EDTA, 2 M NaCl, 0.2% Tween 20). This was added to the fragmented sample DNA and incubated at 24°C while in a ThermoMixer while shaking for 30 minutes. Dynabeads (now bound to DNA) were washed three times in 1 mL of 1x W&B, twice in 1 mL EB, and once in 1 mL 1x T4 DNA ligase buffer (NEB). Dynabeads were resuspended in 50 μL end-repair reaction mix (1x T4 DNA ligase buffer, 2.7 U T4 DNA polymerase (NEB), 9 U T4 Polynucleotide Kinase (NEB), 1 U Klenow fragment (NEB), and 0.4 mM dNTPs) and incubated with shaking at 24°C for 30 minutes. Beads were washed once in 1 mL 1x W&B, twice in 1 mL EB, once in 1 mL NEBuffer 2 (NEB), and resuspended in 50 μL A-tailing reaction mix (1x NEB dA-Tailing Buffer with 20 U Klenow fragment exo- (NEB)). Samples were incubated at 37°C with shaking for 30 minutes. Dynabeads were then washed in 1 mL NEBuffer 2 and resuspended in 115 μL ligation reaction mix (1x Quick Ligase Buffer (NEB), 6,000 U Quick Ligase (NEB), END-seq hairpin adaptor 2) and incubated with shaking at 25°C for 30 minutes. 50 mM EDTA was added to terminate the reaction. Hairpin adaptors were digested using USER enzyme (NEB, M5505S) for 30 minutes at 37°C. Dynabeads were then washed three times with 1 mL 1x W&B and three times with 1 mL EB. Beads were resuspended in 8 μL EB + 10 μL 2x Kapa HiFI HotStart Ready Mix (Kapa Biosciences). Primers (10 mM) 5′-CAAGCAGAAGACGGCATACGAGATxrefGTGACTGG AGTTCAGACGTGTGCTCTTCCGATC*T-3′ and 5′-AATGATACGGCGA CCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T-3′ (asterisk indicates a phosphothiorate bond and a NNNNNN TruSeq index sequence) were added to 37 μL PCR reaction mix (20 μL 2x Kapa HiFI HotStart Ready Mix added to 17 μL H2O) for a final volume of 60 μL. DNA was amplified using the following PCR parameters: 98°C (45 seconds); 15 cycles of 98°C (15 seconds), 63°C (30 seconds), 72°C (30 seconds); 72°C (5 minutes). The finished PCR reaction was cleaned using 1.8x volume AMPure Beads XP (Beckman Coulter), following the manufacturer’s instructions. Samples were then separated on a 2% agarose gel, with ≈200 bp DNA being excised and purified using a QIA-quick Gel Extraction Kit (Qiagen). Library concentrations were measured using a KAPA Library Quantification Kit for Illumina Platforms (Kapa Biosystems) and then sequenced using Illumina NextSeq 550 or NextSeq 2000 instruments. RT–qPCR RNA was extracted from i3Neurons using TRIzol reagent (Thermo Fisher, 15596018). cDNA synthesis was performed with SuperScript VILO cDNA Synthesis kit (Thermo Fisher, 11754050) following the manufacturer’s instructions. qPCR was performed with iTaq Universal SYBR Green supermix (Bio-Rad, 1725121) following the manufacturer’s instructions. Results were collected using a Bio-Rad C1000 Touch thermal cycler. Relative fold change in gene expression was calculated using the ΔΔCt method with β-actin used as a reference gene. The following primers were used: ERCC5 forward, 5′-CTCTTGACGATGACGAAGATGT-3′, reverse, 5′-GTTCTCGGTGGAGCTGTTTAT-3′; XPA forward, 5′-TGTCTGTGATTGCCTTCTTACA-3′, reverse, 5′-CTAGGGTTTGCCTTGGTATCTT-3′; ERCC1 forward, 5′-GAGTGGCCAAGCCCTTATT-3′, reverse, 5′-GAGGCTGTGAGATGGCATATT-3′; XPC forward, 5′-GACCATACCAGAGCCCATTTAT-3′, reverse, 5′-AGGCTGGTCCATGTGTTTAG-3′; SAMHD1 forward, 5′-TCCACACTCGCAACTCTTTAC-3′, reverse, 5′-TTTCCTCCAGCACCTGTAATC-3′; RRM2B forward, 5′-CATCTCACCCAGCTAGTGTTC-3′, reverse, 5′-CCTACCATGTTCTCCCTTTCTC-3′; β-actin forward, 5′-CGTCACCAACTGGGACGACA-3′, reverse, 5′-CTTCTCGCGGTTGGCCTTGG-3′. Nucleotide extraction and Quantification 180 million wild-type i3Neurons were treated with 20 μM cisplatin or PBS for 24 hours. Following treatment, cells were harvested in 700 μL of ice-cold lysis buffer containing 24% trichloroacetic acid (TCA) and 30 mM MgCl₂. Lysates were transferred into pre-chilled Eppendorf tubes, snap-frozen on dry ice, and stored at − 80 °C until further processing. Nucleotide extraction was performed as previously described,78 with minor modifications. Briefly, cell lysates were vortex-mixed in ice-cold lysis buffer using an Intelli-Mixer at 4 °C for 10 minutes. Samples were centrifuged at 14,000 rpm for 3 minutes at 4 °C, and the supernatant was neutralized twice by mixing with an equal volume of ice-cold dichloromethane–trioctylamine solution. After vortexing for 20 seconds, samples were centrifuged again at 14,000 rpm for 1 minute at 4 °C. From the resulting aqueous phase, 650 μL were adjusted to pH 8.9 using 1 M ammonium carbonate and applied to boronate affinity columns (Affi-Gel Boronate Gel, Bio-Rad) to separate dNTPs from NTPs. Elution was carried out using 2.5 mL of a buffer containing 50 mM ammonium carbonate (pH 8.9) and 15 mM MgCl₂. The eluate was adjusted to pH 4.6 and purified using Oasis WAX solid-phase extraction cartridges (Waters). Samples were dried using a SpeedVac concentrator and reconstituted in 1.25 mL of distilled water. For dNTP analysis, samples were adjusted to pH 3.4 and analyzed by high-performance liquid chromatography (HPLC) with UV detection using a Partisphere SAX column (125 mm × 4.6 mm, 5 μm; Hichrom, UK). NTPs were quantified directly from 18 μL aliquots of the aqueous phase, also adjusted to pH 3.4. Reverse transcriptase assay Cellular dNTPs were measured by HIV-1 RT-based dNTP assay as previously described.43 Briefly, cellular dNTPs were extracted from known numbers of cells with 70% methanol and dried. The dNTP extracts were resuspended in water. A 5′ 32P-end of the 18-nucleotide primer annealed to 19-nucleotide templates with different single nucleotide overhangs was mixed with the extracted dNTPs, and the primer was extended by HIV-1 RT at 37°C for 5 min. The reactions were terminated with stop buffer (40 mM EDTA and 99% formamide) and incubated at 95°C for 5 min. Reactions were analyzed by a 14% urea-PAGE denaturing gel, and the extended 18-mer primer were analyzed using Amersham Typhoon (Cytiva) within the linear range of the assay, to quantify single-nucleotide extensions of the samples. The percents of the extended primer were converted to dNTP levels as previously described.43 Dorsal root ganglion harvest, nucleus preparation, and Imaging Flow Cytometry and Flow Cytometry For in vivo EdU imaging assays, adult mice were administered 15 mg/kg cisplatin dissolved in PBS 28 hours prior to harvest along with 200 mg/kg EdU dissolved in PBS 28 hrs, 20 hrs, and 6 hrs before harvest. All injections were i.p. Mice were anesthetized with isoflurane and decapitated. DRGs were dissected and harvested. All DRGs per mouse were combined and placed in 700 μL homogenization buffer (20 mM Tris-HCl (pH 7.5), 25 mM KCl, 5 mM MgCl2, 0.25 M sucrose, 0.2% BSA, and protease inhibitor cocktail tablet (complete Mini, Roche Diagnostics) in a 1 mL glass Dounce tissue grinder. Tissues were homogenized with 10 strokes, before 11.2 μL of 5% Igepal in homogenization buffer was added and tissues were further homogenized with 5 strokes. Sample was strained using a 40 μm cell strainer, centrifuged for 5 mins at 500g, and nuclei were resuspended in PBS. Nuclei were spun down again and then fixed with BD Cytofix/Cytoperm buffer (BD Biosciences, 554714) for 20 mins on ice. EdU Click-iT was performed as described above with AZDye 488 Picolyl Azide (Vector Laboratories, CCT-1276 (see EdU FACS). Nuclei were then incubated with Alexa Fluor 647 anti-NeuN antibody (1:500, Abcam, ab190565) at 4°C for 1 hr, washed with 1% BSA in PBS, stained with DAPI (1 μg/mL, Thermo Fisher Scientific, 62248), and analyzed. Conventional flow cytometry analysis was performed on a 5-laser LSRFortessa flow analyzer (BD Biosciences). Imaging flow analysis was performed on an Amnis Imagestream MarkII dual camera Imaging Flow Cytometer (Cytek). Images were collected on 60x magnification and illuminated with 405nm, 488nm and 640nm lasers. Neurons and glia were distinguished with NeuN intensity. RNA-seq WT i3Neurons were treated (or not) with 20 μM cisplatin and 150 μM dN for 7 days. Total RNA was harvested using TRIzol reagent (Thermo Fisher, 15596018) following the manufacturer’s instructions. DNA was removed using DNAse and rRNA was removed using NEBNext rRNA Depletion kit v.2 (Human/Mouse/Rat) (NEB, E7400). Libraries were prepared using NEBNext Ultra II Directional RNA Library Prep kit for Illumina (NEB, E7760), using 1 μg of starting RNA material and following the manufacturer’s instructions. cDNA was amplified for 11 cycles. Libraries were sequenced on an Illumina NextSeq2000 with 100-bp single-end reads. Production of Vpx virus-like particles and treatment of i3Neurons HEK293T cells were plated in 6-well plates and transfected using 7.5 μL TransIT-LT1 (Mirus) with 0.4 μg pCMV-VSV-G envelope expressing plasmid and 2.6 μg pSIV3 Vpx expressing plasmid per well. Media were changed to OptiMEM 18 hrs after transfection (2 mL per well). Two days after transfection, media was collected, filtered, and stored at −80°C. For treating i3Neurons, 10 μL of Vpx solution per 10,000 cells was added to cell cultures at day 5 of differentiation, 24 hours before beginning treatment with cisplatin. Nascent RNA FACS 500,000 i3Neurons were plated per well on 12-well plates and treated with the indicated concentrations of cisplatin, dN, and THZ1 (CDK7i) for 24 hours. During the last hour of treatment, 1 mM 5-ethynl uridine (EU) (Vector Laboratories, CCT-1261) was added to media. Neurons were harvested as single cells by incubating with accutase for ≈20 minutes and resuspending by repeated pipetting. Cells were washed in PBS and fixed in 2% paraformaldehyde in PBS for 10 minutes. Cells were then washed with PBS and resuspended in 100 μL of click-it reaction mix (Click-iT EdU kit, Thermo Fisher Scientific, C10424) for 30 minutes at room temperature. Cells were washed with PBS and then resuspended in 1% BSA in PBS with DAPI and analyzed using a CytoFLEX flow cytometer (Beckman Coulter). Flow cytometry data was analyzed using FlowJo (v.10). Active mitochondrial staining i3Neurons were treated with 20 μM cisplatin and 150 μM dN for 24 hours, or 1 μM carbonyl cyanide m-chlorophenylhydrazone (CCCP) for 1 hour, where indicated. 100 nM Tetramethylrhodamine, Ethyl Ester, Perchlorate (TMRE) (Invitrogen, T669) was added to neurons for 15 minutes prior to harvest. i3Neurons were harvested as single cells by incubating with accutase for ≈20 minutes and resuspending by repeated pipetting. Cells were washed once with 1% BSA in PBS, resuspended in 1% BSA in PBS, and then analyzed using a CytoFLEX flow cytometer (Beckman Coulter). Flow cytometry data was analyzed using FlowJo (v.10). QUANTIFICATION AND STATISTICAL ANALYSIS Details of statistical tests used for each experiment can be found in the figure legends and STAR Methods. Single-cell SAR-seq analysis Unless otherwise mentioned all data analysis was done in bash and R. First the cell barcode was extracted from position 4–11 in read-1, after which only read pairs were kept where the barcode had an exact match with a cell barcode and where position 12 corresponded to T (complementary A was added during single-base overhang ligation). The cell barcode and the umi (position 1–3 in read-1) were put in the name of the read pair. The reads were then trimmed using trimgalore (wrapper for cutadapt) using the default settings and the –paired flag. Trimmed reads were mapped with STAR version 2.710b to un-masked GRCh38 with following settings; –outFilterMultimapNmax 1 –peOverlapNbasesMin 10 –alignIntronMax 1. Only properly mapped read pairs were kept and deduplicated by grouping on cell barcode, umi, chromosome, and mapping position of read-1 within 10 bases, after which the longest insert size is kept per group. https://github.com/FelixKrueger/TrimGalore https://github.com/alexdobin/STAR https://github.com/marcelm/cutadapt For single-cell data, cells with > 2,000 reads were defined as EdU positive cells, after excluding reads in blacklist. Read counts from the transcribed and non-transcribed strands were quantified using intersect command in the bedtools suite (version 2.31.1)79 with ‘‘different strandedness’’ and ‘‘strandedness’’ parameters, respectively. Remaining reads were assigned to intergenic regions. Transcribed-strand reads from EdU-positive cells were quantified within each gene body and normalized by TPM. Correlations with gene expression levels were calculated by Person correlation coefficient. For visualization, aggregate heatmaps of EdU signals were generated from bigwig files created by combining EdU-positive cells per condition, using the bamCoverage function from deepTools (version 3.5.5)72 with parameters ‘‘-bs=1 -normalizeUsing BPM.’’ END-seq data analysis For analysis of END-seq data, raw sequencing reads were trimmed by Trim Galore! (version 0.6.7) followed by adjusting the parameter ‘‘“–stringency 3’’. Trimmed END-seq reads were aligned to the human genome (hg19) using bowtie (version 1.3.1)80 with parameters ‘‘-n 3 -l 50 -k 1.’’ The END-seq raw counts within gene bodies were quantified using featureCounts with ‘‘reversely stranded’’ parameter in Bioconductor Rsubread package (version 2.18.0).81 Intergenic regions, defined as intervals between genes after excluding blacklist and short regions (<100 bp), were quantified with default parameters. Counts from both genic and intergenic regions were normalized by TPM (Transcripts Per Million). For spike-in normalization, reads from the samples were aligned to both human and mouse genomes. The scaling factor for the spike-in normalization was calculated using the number of reads mapped to the spike-in locus (TCRβ region) of the mouse genome and the total number of reads mapped to the human genome. To generate the spike-in normalized reads, the normalized END-seq signal was divided by the scaling factor. Public data analysis Public RNA-seq from i3Neurons was downloaded from Gene Expression Omnibus (GEO: GSE167256)24 and normalized by TPM. Expressed genes were defined as having TPM > 1. Supplementary Material SUPPLEMENTAL INFORMATION Supplemental information can be found online at https://doi.org/10.1016/j.cell.2026.05.025.