Glioblastoma Stem Cells Resist Copper-Induced Cell Death Linked to Daily Changes in Copper Levels

The failure of cuproptosis-inducing agents in GBM clinical trials led us to hypothesize that cancer stem cells contribute to cuproptosis resistance, given their well-documented resistance to other therapies, such as temozolomide, platinum compounds, and radiotherapy (,,). GSC dependency on iron metabolism () suggested that GSCs may also display differential sensitivity to copper-dependent cell death. Therefore, we investigated the comparative cellular toxicity of a cuproptosis inducer (), ES, within the tumor hierarchy. To confirm the copper-dependent specificity of ES, we treated tumor cells with ES in the presence of iron, copper, or zinc and measured cell proliferation. Copper supplementation (1 μM) sensitized cells to ES-induced cytotoxicity, whereas iron or zinc had no effect (; supplemental material available online with this article;), supporting specificity of ES in copper homeostasis. Copper and ES cotreatment altered cuproptosis markers, decreasing FDX1 (ferredoxin 1) and LIAS (lipoic acid synthetase) and inducing DLAT (dihydrolipoamide-acetyltransferase) aggregation (). ES acts mainly as a pro-cuproptosis agent; thus, all experiments were performed in media supplemented with 1 μM CuClunless noted otherwise. ^{2 20 21 4 8} Supplemental Figure 1A Supplemental Figure 1B https://doi.org/10.1172/JCI192599DS1↗ S 2

To test whether GSCs differ in their response to cuproptosis induction, we treated patient-derived GSCs and their matched differentiated GBM cells (DGCs) with ES in media containing 1 μM CuCl. The ICvalues for ES in DGCs from 3 patients were low — 4.59 nM (DGC387), 11.14 nM (DGC3565), and 6.51 nM (DGC738) () — suggesting potential therapeutic efficacy against bulk tumor. GSCs were less sensitive to ES-induced cuproptosis, with higher ICvalues (47.54, 35.41, and 42.62 nM) than DGCs (), suggesting intrinsic resistance that may underlie the failure of copper-based GBM therapies (). This prompted us to investigate the mechanisms of GSC resistance to copper-induced cell death and identify strategies to overcome it. 2 50 50 Figure 1A Figure 1A ¹²

Cancer stem cells are defined by their capacity for self-renewal, typically assessed by sphere formation. Given the relative resistance of GSCs to cuproptosis, we evaluated the effects of ES on GSC self-renewal. Extreme limiting dilution and sphere size showed that 20 nM ES modestly inhibited GSC self-renewal and proliferation, whereas lower concentrations were more effective against DGCs (, and). ES preferentially induced cuproptosis markers — DLAT oligomerization and reduced LIAS and FDX1 — in DGCs compared with GSCs, indicating differentiated cells were sensitive, whereas GSCs were resistant (and). Figure 1, B and C Supplemental Figure 1C Figure 1D Supplemental Figure 1D

GSCs preferentially accumulate iron (), so we hypothesized that differential GSC sensitivity to cuproptosis may relate to altered copper regulation. We therefore quantified copper, iron, and zinc levels in GSCs, matched DGCs, and NSCs using inductively coupled plasma mass spectrometry (ICP-MS). Concordant with our prior observations (), iron content was elevated in GSCs compared with DGCs and NSCs, whereas copper and zinc levels showed no significant differences among all cell types (). Building on our prior findings of circadian regulation in GSC metabolism (), we investigated whether copper levels fluctuated temporally. We hypothesized that GSCs dynamically regulate copper with diurnal variation. Matched GSCs and DGCs were synchronized by a 2-hour dexamethasone pulse (100 nmol/L), followed by media replacement, and then metal levels were quantified by ICP-MS. Copper content displayed circadian oscillation in GSCs but not in DGCs, whereas iron and zinc showed no rhythmic variation in either cell type (). Copper imaging corroborated these findings (). For copper imaging analysis, GSC3565 exhibited rhythmicity whereas DGC3565 did not (). In addition, although both GSC387 and DGC387 exhibited rhythmic patterns, GSC387 showed a higher amplitude and more pronounced rhythmicity (). ^{4 4} Figure 2A ¹⁴ Figure 2B Figure 2C Figure 2C Figure 2C

To examine copper–circadian concordance, we analyzed gene expression data from asynchronously cultured 44 GSCs and 10 NSCs (GSE119834) (). Using the Gene Ontology (GO) circadian rhythm signature (GOBP_CIRCADIAN_RHYTHM, GO:0007623), we found that circadian activity was elevated in GSCs relative to NSCs (). Similarly, copper homeostasis (GO:0006878) activity was elevated in GSCs versus NSCs (). Next, we constructed a cuproptosis gene signature from 10 genes involved in cuproptosis (,,,,,,,,, and) (). In GSCs, circadian rhythm and cuproptosis signatures were negatively correlated (). These results suggested that GSCs regulate copper levels in a circadian manner, likely reflecting intrinsic molecular distinctions rather than synchronized time-of-day–dependent expression. ²² Supplemental Figure 2A Supplemental Figure 2B ⁸ Supplemental Figure 2C FDX1 LIAS LIPT1 DLD DLAT PDHA1 PDHB MTF1 GLS CDKN2A

To investigate contributions of key copper modulators bridging the circadian clock and cuproptosis, we conducted CRISPR/Cas9 knockout screens in GSCs () using a custom library targeting 58 copper-related genes with additional control genes in 636 guide RNAs (gRNAs) (). We performed chemogenomic screens with ES + CuClor the circadian inhibitor SHP656 () to identify genes mediating resistance to cuproptosis or clock inhibition (). We employed pharmacologic CRY (cryptochrome) stabilizer SHP656, an orally bioavailable KL001 derivative that selectively stabilizes CRY2 and enhances its negative feedback on brain and muscle ARNT-like 1–CLOCK–mediated (BMAL1–CLOCK–mediated) transcription (). SHP656 inhibits the growth of GSCs in vitro and prolongs survival in GSC-engrafted mouse models without affecting DGCs or NSCs (). The ICof ES derived fromand SHP656 () were used as working concentrations. We identified 8 fitness, 8 circadian resistance, and 11 cuproptosis resistance genes (and). The only gene that intersected among these hits was(). ATP7A, a P-type copper-transporting ATPase, regulates intracellular copper homeostasis, and its mutations cause lethal infantile Menkes disease (). In response to elevated intracellular copper levels, ATP7A translocates to the plasma membrane to facilitate copper efflux, functioning as a copper exporter (). Figure 3A Supplemental Table 1 ²³ Figure 3A ^{23 23} Figure 1A ¹⁴ Figure 3B Supplemental Tables 2 and 3 Figure 3, B and C ^{24 24} 2 50 ATP7A

To validate the role of ATP7A in resistance to cuproptosis and clock inhibition, we modulated its expression and assessed ICchanges for ES and SHP656. ATP7A knockdown enhanced GSC sensitivity to both agents (). Overexpression of ATP7A rescued the increased sensitivity to cuproptosis or disruption of clock function caused by ATP7A knockdown (, D and E). ATP7A knockdown increased cuproptosis markers and enhanced ES or SHP656 effects (, and, F and G), also reducing GSC self-renewal in limiting-dilution and sphere assays, further sensitizing cells to ES (, A and B). 50 Figure 3, D and E Supplemental Figure 2 Figure 3, F and G Supplemental Figure 2 Supplemental Figure 3

Based on the impact of ATP7A modulation on GSC responses to cuproptosis, we next investigated its relative expression in GSCs. Chromatin immunoprecipitation sequencing (ChIP-Seq) revealed elevated acetylated H3K27 (H3K27ac) enrichment at the ATP7A promoter in GSCs compared with DGCs and NSCs (), indicating epigenetic activation of ATP7A in the tumor hierarchy. In patient samples, ATP7A was elevated in isocitrate dehydrogenase-WT (IDH-WT) and 1p/19q–non-codeleted gliomas, increased with tumor grade, and correlated with poor prognosis (, D–J). qPCR confirmed higher ATP7A expression in GSCs than in DGCs or NSCs (). We analyzed a single-cell glioma dataset for circadian rhythm and copper homeostasis gene expression, revealing preferential enrichment of these signatures in stem-like tumor cells (). Immunofluorescence staining showed cytoplasmic localization of ATP7A in both GSCs and DGCs, with no major differences in subcellular distribution; however, ATP7A expression levels were higher in GSCs than in their matched DGCs (). ATP7A expression was higher in GSCs than in matched DGCs (). Supplemental Figure 3C Supplemental Figure 3 Supplemental Figure 3K Supplemental Figure 3L Supplemental Figure 3M Supplemental Figure 3M

To assess the function of ATP7A in malignant and nonmalignant neural cells, we silenced its expression in GSCs, DGCs, and NSCs using 2 independent shRNAs. ATP7A knockdown markedly reduced GSC proliferation (). DGCs also showed reduced growth, though the effect was less pronounced, while NSCs exhibited only modest inhibition — one shRNA caused a mild decrease, whereas the other had no effect. Overall, the growth-suppressive impact of ATP7A loss was most evident in GSCs (). ATP7A depletion reduced GSC proliferation measured by 5-ethynyl-2′-deoxyuridine (EdU) staining (), markedly inhibited GSC self-renewal (, C and D), and suppressed the expression of the stemness markers, SOX2 and OLIG2 ().correlated withexpression in patients with IDH-WT GBM (). To confirm ATP7A's role in cuproptosis, copper depletion with tetrathiomolybdate (TM) partially reversed the effects of ATP7A knockdown, indicating tumor-suppressive action is at least partly copper dependent (). Supplemental Figure 4A Supplemental Figure 4A Supplemental Figure 4B Supplemental Figure 4 Supplemental Figure 4E Supplemental Figure 5A Supplemental Figure 5B ATP7A SOX2

Given that intracellular copper levels exhibited circadian oscillation in GSCs and ATP7A regulates copper homeostasis, we next examined whether ATP7A is subject to circadian control. BMAL1, a core component of the mammalian circadian clock, forms a heterodimer with CLOCK to activate transcription of key clock genes, such as() and(), via E-box promoter elements. The rhythmic expression of these genes establishes and maintains circadian regulation of diverse physiological processes, including metabolism and immune function (). Our previous study has shown that BMAL1 plays an important role in glioma stem cells (). We performed a circadian expression analysis of,, andin matched GSCs and DGCs using JTK_CYCLE algorithm. Both cell types showed circadian variation, but DGC oscillations were less synchronized and consistent than those in GSCs (). Notably, ATP7A showed stronger circadian amplitude in GSCs, indicating enhanced rhythmic regulation in the stem-like state (). Consistently, ATP7A protein levels exhibited circadian oscillation relative to BMAL1 in GSCs but were less variable in DGCs (). Period Per Cryptochrome Cry CLOCK BMAL1 ATP7A ^{25 14} Figure 4, A and B Figure 4, A and B Supplemental Figure 5C

Given the link between ATP7A and the transcriptional regulator BMAL1, we examined whether BMAL1 directly regulates ATP7A. ChIP-Seq revealed stronger BMAL1 binding at the ATP7A promoter in GSCs (). This promoter region harbored canonical E-box motifs (CACGTG), known binding sites for the BMAL1–CLOCK heterodimer, potentially accounting for BMAL1 recruitment in GSCs. Consistently, ChIP-qPCR analysis confirmed BMAL1 occupancy at thepromoter in GSCs but not in DGCs (). When cells are treated with a cuproptosis inducer to trigger copper-induced cell death, BMAL1 remains bound to the promoter region ofin GSCs but not in DGCs (). BMAL1 showed rhythmic expression in both GSCs and DGCs. ATP7A rhythmicity occurred only in GSCs, likely due to stronger BMAL1 promoter binding. BMAL1 knockdown reduced ATP7A expression in GSCs but not in DGCs (, and).expression positively correlated withexpression in patients with IDH-WT GBM in the Chinese Glioma Genome Atlas (CGGA;). Although direct BMAL1 inhibitors are unavailable, compounds that suppress clock function via feedback stabilization were tested. Circadian inhibition reduced ATP7A expression in GSCs (and). Genetic BMAL1 knockdown disrupted circadian ATP7A expression and reduced overall ATP7A expression (and). Supplemental Figure 5D Supplemental Figure 5E Supplemental Figure 5F Figure 4, C and D Supplemental Figure 6A Supplemental Figure 6B Figure 4E Supplemental Figure 6C Figure 4F Supplemental Figure 6D ATP7A ATP7A ATP7A BMAL1

As BMAL1 regulates the copper exporter ATP7A, we examined its effect on copper levels. ICP-MS showed that BMAL1 knockdown in GSCs increased copper without affecting iron or zinc (). Live-cell imaging showed circadian oscillations of intracellular copper in control and BMAL1-deficient GSCs, though the rhythmic amplitude was markedly reduced in shBMAL1 cells (and). This dampened oscillation was accompanied by elevated baseline copper levels (and). ATP7A overexpression reversed copper accumulation from BMAL1 knockdown, indicating BMAL1 controls copper via ATP7A (). Figure 4G Figure 4H Supplemental Figure 6E Figure 4H Supplemental Figure 6E Supplemental Figure 6F

To investigate how ATP7A promotes GSC growth, we compared RNA-Seq profiles of shATP7A- and shCONT-transduced GSCs. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed changes in metabolic, mineral absorption, and fatty acid metabolism pathways (). Reactome analysis showed altered expression of acyl-CoA desaturase and SREBP1A/1C–stearoyl-CoA desaturase (SCD) binding pathways (). ATP7A knockdown positively correlated with gene sets associated with fatty acid metabolism on gene set enrichment analysis (GSEA) (). Figure 5A Figure 5B Figure 5C

Fatty acid synthesis promotes GBM growth, including GSC maintenance (). SCD1 converts saturated to monounsaturated fatty acids, serving as a key regulator of lipid metabolism (). Monounsaturated fatty acids (MUFAs) are essential for cancer cell survival (). Lipidomics analysis showed that ATP7A depletion decreased the MUFA/saturated fatty acid (SFA) ratio while leaving the polyunsaturated fatty acid (PUFA)/SFA ratio unchanged (and). ATP7A knockdown reduced SCD1, FADS1, SREBF1, and FASN protein levels () and decreased fatty acid–related genes by qPCR ().expression correlated withandin patients with IDH-WT GBM from CGGA (). However, the heatmap of lipidomics measurements showed ATP7A depletion led to overall increased levels of all SFAs, MUFAs, and PUFAs (and). SFAs accumulated to cytotoxic levels, lowering MUFA/SFA ratios. SCD1 inhibition further reduces this ratio, while excess SFAs can convert to MUFAs or PUFAs, increasing their absolute levels.expression measured by qPCR was higher in GSCs compared with DGCs or NSCs ().expression in patients with IDH-WT GBM from CGGA correlated with SOX2 and OLIG2 expression (). To assess SCD1's role in GSC maintenance, we inhibited it with A939572 (), which reduced SOX2 and OLIG2 expression by Western blot and qPCR (, E and F). ^{26 27 28} Figure 5D Supplemental Table 4 Figure 5E Figure 5F Supplemental Figure 7A Supplemental Figure 7B Supplemental Table 4 Supplemental Figure 7C Supplemental Figure 7D ²⁹ Supplemental Figure 7 ATP7A SCD1 FADS1 SCD1 SCD1

The original description of cuproptosis noted a link to fatty acid biology (), and copper regulates hepatic fatty acid synthesis. Thus, our findings indicate that ATP7A links circadian and copper metabolism while regulating fatty acid synthesis, suggesting that inhibiting fatty acid synthesis may enhance cuproptosis. Accordingly, combined treatment with the SCD1 inhibitor A939572 and the cuproptosis inducer ES produced a greater reduction in GSC growth (). To connect this process to the circadian clock regulation, BMAL1 knockdown in GSCs decreased expression of fatty acid–related genes (). Similarly, circadian inhibitors SR9009 and SHP656 decreased,,,, and, with minimal effect on(, G and H). Collectively, these findings suggest that circadian regulation of copper influences tumor cell behavior at least in part through fatty acid synthesis, revealing an additional therapeutic axis for intervention. ⁸ Figure 5G Figure 5H Supplemental Figure 7 SCD1 FADS1 SREBF1 FASN ACC1 ACADS

As BMAL1 regulates ATP7A, we tested whether copper affects the circadian clock. Increasing copper concentrations reduced BMAL1 expression in GSCs (). Similarly, ES decreased BMAL1 expression in the presence of copper (). Copper suppressed total protein levels and oscillations of BMAL1 in GSCs () but notmRNA levels (), suggesting that copper regulates BMAL1 at the posttranscriptional level. We treated GSCs with the translational inhibitor cycloheximide (CHX) following copper or control pretreatment and monitored BMAL1 protein levels over time, verifying that copper accelerates BMAL1 degradation (). Chloroquine elevates lysosomal pH, disrupting its acidic environment and blocking autophagosome–lysosome fusion, thereby inhibiting autophagic degradation. Chloroquine treatment prevented copper-induced BMAL1 degradation, whereas the proteasome inhibitor MG132 had no effect (), indicating copper-induced BMAL1 degradation involves the autophagy–lysosomal pathway. Figure 6A Figure 6B Supplemental Figure 8A Supplemental Figure 8B Figure 6C Figure 6D BMAL1

We performed MS of BMAL1-binding proteins (), then prioritized among identified targets autophagy-related proteins (GOBP:SELECTIVE_AUTOPHAGY, GO0061912;), demonstrating SQSTM1 as a top target (). IP confirmed that BMAL1 bound SQSTM1 but not other autophagy receptors (OPTN, TAX1BP1, NBR1, NDP52) (). To map the SQSTM1 domain interacting with BMAL1, FLAG-tagged SQSTM1 fragments were coexpressed with HA-BMAL1 in HEK293T cells for co-IP. Constructs lacking the PB1 domain (~80 amino acids mediating dimerization) failed to bind BMAL1 (). Oligomerization of SQSTM1 is essential for aggrephagy cargo assembly and autophagic degradation. Copper supplementation enhanced SQSTM1 oligomerization and its interaction with BMAL1, effects that were augmented by chloroquine treatment (and). SQSTM1 oligomerization was markedly induced in response to cotreatment with copper and ES (). Consistent with their interaction, BMAL1 and SQSTM1 colocalized in punctate cytoplasmic structures by immunofluorescence and exhibited circadian rhythmicity in synchronized GSCs (). Copper increased the number of BMAL1-P62/SQSTM1 colocalized puncta and abolished rhythmicity (). SQSTM1 or ATG7 knockdown blocked copper-induced BMAL1 degradation (, E and F). Under excess copper, SQSTM1 bound BMAL1 via its PB1 domain, inducing oligomerization and autophagic–lysosomal degradation, forming a negative feedback loop linking copper to circadian regulation. In DGCs, copper or ES treatment similarly reduced BMAL1 levels (, G and H), suggesting that exogenous excess copper induces BMAL1 degradation in tumor cells. Supplemental Tables 5 and 6 Supplemental Table 7 Figure 6E Figure 6F Figure 6G Figure 6H Supplemental Figure 8C Supplemental Figure 8D Figure 6I Figure 6I Supplemental Figure 8 Supplemental Figure 8

To assess how circadian disruption influences cuproptosis sensitivity, we silenced BMAL1 and evaluated GSC and DGC responses. ATP7A overexpression rescued the enhanced cuproptosis sensitivity caused by BMAL1 knockdown in GSCs (). BMAL1 knockdown enhanced cuproptosis markers measured by Western blot in GSCs but not in DGCs (and). BMAL1 depletion enhanced ES-induced inhibition of GSC self-renewal in limiting dilution and sphere assays (, and). Thus, genetically targeting the circadian clock sensitizes GSCs to cuproptosis. Figure 7A Figure 7B Supplemental Figure 9A Figure 7, C and D Supplemental Figure 9B

Given the feedback loop between copper and the circadian clock, concurrent pharmacologic targeting of both pathways may enhance GSC suppression. Circadian inhibitors, including the CRY stabilizer SHP656 and the REV-ERB agonists SR9009 and SR9011, synergized with ES to augment anti-GSC efficacy (, and) without increasing toxicity in NSCs (). Treatment with SHP656 or SR9009 enhanced cuproptosis markers on immunoblot, including increased DLAT oligomerization and reduced LIAS and FDX1 expression following ES treatment (). Figure 7, E and F Supplemental Figure 9C Supplemental Figure 9D Figure 7, G and H

The gold standard for assessing cancer stem cell function is in vivo tumor initiation. We transduced luciferase-expressing GSCs with shCONT or shATP7A and implanted them intracranially into immunocompromised mice. ATP7A knockdown prolonged survival and reduced tumor burden compared with controls (, and). To evaluate ATP7A's role in glioma stemness in vivo, intracranial limiting dilution assays (10,000–100 cells) showed that ATP7A knockdown extended survival, indicating reduced tumor initiation (). Figure 8, A and B Supplemental Figure 10A Supplemental Figure 10B

To test therapeutic potential, we evaluated the in vivo effects of the CRY stabilizer SHP656 alone or combined with copper gluconate, ES, or both. We previously reported the intracranial antitumor activity of SHP656 (), and ES has shown monotherapy effects against GBM in vivo (), supporting their potential for intracranial antitumor activity; however, further studies are necessary to definitively establish blood–brain barrier penetrability. Orthotopic GSC tumor growth was reduced by copper gluconate plus ES or SHP656 alone, and SHP656 further enhanced the efficacy of copper gluconate, ES, or their combination (and, C and D). Combined SHP656, ES, and copper gluconate treatment prolonged survival compared with single therapies (and). Previous reports link GSC CLOCK/BMAL1 signaling to glioma immunosuppression (–). Using the CT2A syngeneic glioma model in C57BL/6 mice, combination therapy extended survival compared with single agents (). Neither single-agent treatment nor combination therapy altered the proportion of CD206microglia (). These results indicate that SHP656, copper, and ES act independently of microglia-mediated immunosuppression and that cotargeting the circadian clock and cuproptosis enhances therapeutic efficacy. Together, circadian control of intracellular copper supports GSC resistance to cuproptosis by coordinating BMAL1-dependent copper homeostasis, ATP7A-driven fatty acid desaturation, and copper-mediated BMAL1 degradation, revealing a targetable circadian-copper vulnerability in GBM (). ^{23 30} Figure 8B Supplemental Figure 10 Figure 8C Supplemental Figure 10E ^{31 33} Supplemental Figure 10F Supplemental Figure 10G Figure 8D +

Our study used both female and male immunodeficient NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) mice (IMSR catalog JAX:005557, RRID: IMSR_JAX:005557, The Jackson Laboratory), and similar findings are reported for both sexes.

GBM tissues from consented patients at Duke University or University Hospitals Cleveland Medical Center (IRB 090401) were used to isolate and validate GSCs. ENStem-A are human embryonic stem cell–derived neural progenitors (MilliporeSigma). NSC11 are NSCs derived from human induced pluripotent stem cells (ALSTEM). hNP1s are NSCs derived from human induced pluripotent stem cells originating from the hESC WA09 line. Cell identity was confirmed by short tandem repeat profiling and mycoplasma PCR. GSCs/NSCs were cultured in Neurobasal medium (Gibco) with B27 minus vitamin A, pyruvate, GlutaMAX (Gibco), and 20 ng/mL EGF/basic FGF. DGCs were derived from GSCs in DMEM + 10% FBS for at least 7 days.

All mouse experiments followed protocols approved by the University of Pittsburgh IACUC. For intracranial xenografts, 4- to 6-week-old NSG or C57BL/6J mice were implanted with 10patient-derived GSCs into the right cortex (3.5 mm depth). Animals were housed under veterinary supervision and euthanized upon neurological decline. Brains were fixed in 4% paraformaldehyde, paraffin-embedded, and analyzed by H&E staining. Survival was assessed using GraphPad Prism (log-rank test). For luciferase-labeled GSCs, mice received-luciferin (50 mg/kg, i.p.; Promega) and were imaged under isoflurane anesthesia with an IVIS system (PerkinElmer). For in vivo studies, mice were similarly implanted intracranially. After 7 days, the mice were subjected to treatment with vehicle, ES (25 mg/kg, intragastric administration [i.g.], once every other day) (), copper gluconate (0.15 mg/kg, i.g., once every day) (), or SHP656 (10 mg/kg, i.g., twice a day) () until reaching specified endpoints. 5 d ^{30 53 14}

Tumor-bearing brains were microdissected, and tumors were dissociated into single cells using Collagenase D (2 mg/mL; Roche) for 30 minutes at 37°C. RBCs were removed with lysis buffer (Thermo Fisher Scientific) for 3 minutes and quenched with medium. Cells were stained with fluorochrome-conjugated antibodies against CX3CR1, CD11b, CD45, and CD206 (BioLegend) for 30 minutes at 4°C. Data were collected on an Attune flow cytometer and analyzed with FlowJo (BD Biosciences). Microglia were defined as CD45CD11bCX3CR1cells. lo + +

A total of 636 sgRNA oligonucleotides targeting 58 genes involved in copper regulation () were synthesized by VectorBuilder (Lib230221-1474xzd). GSC3565 cells stably expressing Cas9 (lentiCas9-Blast, Addgene #52962) were transduced with the pooled lentiviral library at MOI 0.3 (>1,000× coverage, 2 replicates). After 2 days, puromycin selection was applied. On day 5, cells were collected as the day 1 baseline. Remaining cells were divided into 3 groups, SHP656 (8.5 μM), ES (35 μM) + CuCl(1 μM), or untreated control, and cultured 14 days. Comparison of day 14 versus day 1 controls identified fitness genes; SHP656-treated versus untreated identified circadian drug resistance genes; and ES-treated versus untreated identified cuproptosis resistance genes. Genomic DNA was extracted (QIAGEN), PCR-amplified (primers in), and sequenced (Illumina PE150, CD Genomics). Data were processed and analyzed using MAGeCK on Galaxy (RRID:SCR_006281), with< 0.05 defining significant hits (). Supplemental Table 1 Supplemental Table 8 Supplemental Tables 2 and 3 2 P

shRNAs used were shBMAL1 (TRCN0000019097, TRCN0000019096), shATP7A (TRCN0000043173, TRCN0000043177, TRCN0000418612), shSQSTM1 (TRCN0000007237), and shATG7 (TRCN0000007584). A nontargeting shRNA (Sigma-Aldrich, SCH002) served as control. ATP7A- and BMAL1-overexpression plasmids were from VectorBuilder (VB220919-1171bzh, VB221011-1324hfh). FLAG-SQSTM1-WT and FLAG-SQSTM1 fragment plasmids that lacked specific domain were constructed according to. Figure 5H

For stable gene modulation, lentiviruses were produced by cotransfecting HEK293T cells (ATCC) with transfer plasmid (3 μg), psPAX2 (6.75 μg), pMD2.G (2.25 μg), and PEI (24 μL). Viral supernatants were collected at 48 hours and 72 hours, concentrated (Lenti-X, Takara), and used to transduce GSCs with 10 μg/mL polybrene for 24 hours. Infected cells were selected with 2 μg/mL puromycin for 2 days, and infection efficiency was verified by qPCR or immunoblotting.

Cell viability was measured using the cell counting kit-8 (CCK-8) assay (APEXBIO, K1018) in 96-well plates seeded with 3,000 cells/well on days 0, 2, 4, and 6 per the manufacturer's instructions.

Single-cell RNA-Seq data from NCBI GEO (GSE174554) were reanalyzed using Seurat. Uniform manifold approximation and projection was generated after standard preprocessing, and tumor cells were classified as stem-like or differentiated glioma via AUCell using a validated GSC gene signature (). GOBP module scores were calculated and normalized (from 0 to 1) using Seurat functions in R. ⁵⁴

Cells (3,000/well) were seeded in Matrigel-coated, 96-well plates and treated with 2-fold serial drug dilutions. After 48 hours' incubation (37°C, 5% CO), CCK-8 reagent was added for 1 hour at room temperature, and luminescence was measured. Dose–response curves were generated in GraphPad Prism (RRID:SCR_002798), and drug synergy was analyzed using the R package SynergyFinder. 2

The Click-iT EdU assay (Thermo Fisher Scientific) was performed per manufacturer instructions. Cells were incubated with 10 μM EdU for 2 hours, and EdU-positive fractions were quantified relative to DAPI-stained nuclei using ImageJ (NIH).

Cells were lysed in RIPA buffer with protease inhibitors on ice for 15 minutes and centrifuged at 12,000for 10 minutes at 4°C. Protein concentrations were measured by BCA assay. Equal protein amounts were mixed with 4× lithium dodecyl sulfate buffer, boiled for 5 minutes, and used for NuPAGE or stored at –80°C. PVDF membranes were blocked with 5% milk in TBS-Tween for 1 hour and incubated overnight with primary antibodies at 4°C. Antibodies used in this study are detailed in. Membranes were incubated with HRP-conjugated secondary antibodies, developed using Immobilon ECL Ultra substrate (MilliporeSigma), and imaged with a Bio-Rad ChemiDoc MP system. g Supplemental Table 9

Cells were lysed in IP buffer (Thermo Fisher Scientific, 87788) with protease inhibitors, and cleared lysates were incubated overnight at 4°C with target or control IgG antibodies. Pierce Protein A/G Magnetic Beads (88802, Thermo Fisher Scientific) were added for 2 hours at 4°C, washed 4 times with IP buffer, and then boiled and analyzed by immunoblotting.

Total RNA was extracted using TRIzol (Sigma-Aldrich) per manufacturer instructions. cDNA was synthesized with the High-Capacity Reverse Transcription Kit (Thermo Fisher Scientific, 4368814), and qPCR was performed using SYBR Green Master Mix (Thermo Fisher Scientific, 4309155) on a CFX Connect Real-Time PCR System (Bio-Rad). β-Actin served as control; primer sequences are in. Supplemental Table 8

Cells were cross-linked with 1% formaldehyde for 10 minutes and quenched with 0.125 M glycine. Nuclei were lysed and sonicated to yield 200–500 bp chromatin fragments, which were incubated overnight at 4°C with anti-BMAL1 (CST 14020S) or control IgG. Immunocomplexes were captured with Protein A/G magnetic beads, washed, eluted, and reverse cross-linked. The purified DNA was analyzed by PCR.

Cells grown on Matrigel-coated coverslips were fixed with 4% paraformaldehyde for 15 minutes, permeabilized with 0.1% Triton X-100 for 10 minutes, and blocked in serum for 1 hour. Cells were then incubated with primary antibodies overnight at 4°C, followed by secondary antibodies. Nuclei were counterstained with DAPI, and slides were mounted with ProLong Diamond Antifade Mountant (Thermo Fisher Scientific). All antibodies used in this study are detailed in. Supplemental Table 9

Live-cell copper imaging was performed using BioTracker Green Copper Live Cell Dye (SCT041, MilliporeSigma) according to the manufacturer's instructions. Briefly, cells cultured on Matrigel-coated dishes were incubated with 5 μM dye for 2 hours, rinsed twice with observation buffer, and imaged using a Leica SP8 microscope.

Cells were treated with CHX (100 μg/mL; Sigma-Aldrich), chloroquine (10 μM; Selleck, S4157), or MG132 (10 μM; MedChemExpress, HY-13259) for the indicated times before Western blotting.

For RNA-Seq following ATP7A knockdown, total RNA was extracted from GSC387, GSC3565, and GSC738 cells transduced with shCONT or shATP7A using TRIzol (Invitrogen) and Direct-zol RNA Miniprep Kits (Zymo Research). Purified RNA was eluted in RNase-free water and subjected to paired-end 150 bp sequencing. Raw reads were trimmed with Trim Galore and aligned to the human genome (hg38) using HISAT2. Samtools was used for sorting, indexing, and file conversion, and gene quantification and differential expression analysis were performed with FeatureCounts and DESeq2 (paired design). Differentially expressed genes (fold-change > 1,< 0.01) were analyzed for pathway enrichment (KEGG, Reactome) and GSEA using Omicshare tools. RNA-Seq data are available in GEO (GSE278671). P

Metal quantification was performed by Sarah Jantzi at the Plasma Chemistry Laboratory, University of Georgia. Samples were digested in polytetrafluoroethylene vessels with 0.5 mL trace metal–grade nitric acid and 0.5 mL hydrogen peroxide (1 hour each at 95°C), diluted to 2% HNO, and analyzed by ICP-MS (Thermo X-Series 2) using an indium internal standard in kinetic energy discrimination mode with 8% H/He to reduce interference. 3 2

Lipidomics analysis was performed by the West Coast Metabolomics Center, UC Davis Genome Center. Frozen cell pellets were processed using the Matyash extraction method (MTBE/MeOH/HO). The organic phase was dried, resuspended in 9:1 methanol/toluene containing 50 ng/mL CUDA, and centrifuged (14,000). Aliquots were prepared for positive- and negative-mode analyses on an Agilent 1290 Infinity LC system coupled to an Agilent 6546 QTOF mass spectrometer. Samples were separated on an Acquity Premier BEH C18 column (1.7 μm, 2.1 × 50 mm; Waters) using a 5.5-minute gradient (15%–99% B) at 0.8 mL/min. Positive-mode scans covered120–1,200 Da; negative-mode scans,60–1,200 Da. Mass resolutions were 10,000 (electrospray ionization) and 30,000 (electrospray ionization). Lipidomics data are provided in. 2 g m/z m/z + – Supplemental Table 4

Single-sample GSEA of GOBP pathways was performed using the GSVA package (ssgsea method) in R. Circadian rhythm (GO:0007623) and copper homeostasis (GO:0006878) signatures from the GO database were used to infer pathway activity. Ten cuproptosis-related genes () (,,,,,,,,, and) were used to construct a cuproptosis signature. RNA-Seq data of 44 GSCs and 9 NSCs were from GSE119834 () and were analyzed using the limma package. ChIP-Seq data of BMAL1 from GSCs and NSCs were from GSE134972 (). ChIP-Seq data of H3K27ac were from GSE54047 () and GSE119755 (). Integrative Genomics Viewer (RRID:SCR_011793) was used for peak visualization. ^{8 22 14 55 22} FDX1 LIAS LIPT1 DLD DLAT PDHA1 PDHB MTF1 GLS CDKN2A

MS was performed by the Proteomics Resource Center at The Rockefeller University. Nuclear BMAL1-interacting proteins were immunoprecipitated from GSC3565 cells and eluted by partial on-bead digestion with 300 ng trypsin (Promega) in 50 mM ammonium bicarbonate for 3 hours at room temperature. Supernatants were reduced with 10 mM DTT, alkylated with 100 mM iodoacetamide for 1 hour in the dark, and further digested overnight with 500 ng trypsin and 500 ng Lys-C (Wako) in 50 mM ammonium bicarbonate. Digestion was stopped with trifluoroacetic acid. Peptides underwent reversed-phase-based micro-solid-phase extraction (). Triplicates of 10 μL were injected and analyzed by nano–liquid chromatography-tandem mass spectrometry (Fusion LUMOS coupled to an Easy-nLC 1200, Thermo Fisher Scientific). Mass spectrometers were mass-calibrated weekly and operated with lock mass (). MS and MS/MS spectra were acquired at resolutions of 120,000 and 30,000 (@200 Thomson), respectively, with automatic gain control targets of 4 × 10(MS) and 5 × 10(MS/MS). Peptides were separated on a 12 cm × 100 μm packed emitter column (NIKKYO TECHNOS) using a 70-minute gradient from 2% to 35% solvent B (A: 0.1% formic acid; B: 80% acetonitrile/0.1% formic acid). Peptide counts are listed in. Proteins with IgG/IP-BMAL peptide ratios less than 0.3 were defined as BMAL1-binding proteins (). GOBP:SELECTIVE_AUTOPHAGY, GO0061912 (), were considered autophagy-related proteins. ^{56 57} Supplemental Table 4 Supplemental Table 5 Supplemental Table 6 5 4

Public glioma datasets from GlioVis were used to analyze mRNA expression, correlations (,,,,), and ATP7A survival in the CGGA cohort. ATP7A BMAL1 SOX2 OLIG2 SCD1

A significance threshold of< 0.05 was used. Rhythmicity was analyzed using the RTK (JTK_CYCLE–based) algorithm. Statistical tests, including 2-tailed unpaired t test, 1-way or 2-way ANOVA followed by Tukey's multiple-comparison correction, Pearson's correlation, likelihood-ratio test, and log-rank test, were performed using GraphPad Prism (RRID:SCR_002798). Data are presented as mean ± SEM unless otherwise noted. P

All mouse experiments were approved by the University of Pittsburgh IACUC.

The data that support the findings of this study are available within the article and its supplement. Values for all data points in graphs are reported in thefile. The RNA-Seq and ChIP-Seq data are deposited in the NCBI's GEO database (GSE278671). Supporting Data Values