RAD23B acquires a copper metalloadaptor function in amphibian-to-reptile evolution to increase metabolism and regulate genomic integrity

SUMMARY

Increasing brain complexity is a major step in the evolution of species. Here, we show that in the transition from amphibians to reptiles that the DNA repair protein, RAD23B, acquires a metalloadaptor function that allows it serve as a central hub for both metabolism and protection of genomic integrity. More specifically, RAD23B gains an allosteric H274/H323 copper-binding site to enable transfer of copper from the universal copper transporter 1 (CTR1) uptake protein to all known copper metallochaperone pathways, while simultaneously making its canonical functions in DNA repair copper-dependent. This layer of nutrient regulation allows organisms to withstand elevated levels of potentially toxic copper while augmenting metabolism in cells with high energetic needs across both physiology and disease, including neurons in the locus coeruleus, a key brain structure that regulates sleep, and cancer cells.

Graphical Abstract

eTOC

Xiao et al identified RAD23B evolves as metalloadaptor in Amniota to regulate copper homeostasis, thus improving cellular metabolism fitness while protecting cell integrity.

INTRODUCTION

Brain development and the evolution of species are intimately connected, as the brain is the body’s most metabolically active organ and its energy demands increase with higher organism complexity¹. Nowhere is this relationship better illustrated than in the shift of life from water to land, where oxygen is made more accessible to cells. The transition from amphibians to reptiles, occurring 350 million years ago, required large expansions of energy supply^2,3, with mammals having much higher cellular metabolic rates compared to fish^4,5. However, the molecular basis of how this change occurred is unknown.

In this context, metals are required nutrients for all kingdoms of life, with cell-specific metal quotas tailored to their particular function. For example, because they can participate in oxidation-reduction (redox) chemistry, transition metals like copper accumulate at unusually high concentrations in the brain⁶ as they are needed for driving mitochondrial function to meet increased energy demands in higher organisms, and for the biosynthesis of neurotransmitters in specialized brain cell types. In particular, neurons in the locus coeruleus (LC), a central brain structure whose development during the transition from invertebrates to vertebrates coincides with the evolution of norepinephrine as a vertebrate-specific neurotransmitter, accumulate 10-fold higher copper levels compared to other brain regions^7,8. This high demand for copper in the LC is due in part to its role as a redox cofactor in the terminal enzyme for norepinephrine (NE) biosynthesis, dopamine beta-hydroxylase (DBH). However, the redox activity of copper also makes it potentially toxic, which raises the fundamental question of how such large pools of copper can be tolerated given its ability to cause oxidative stress and damage.

To meet its physiological requirements yet minimize toxicity, copper homeostasis is tightly regulated by proteins with dedicated functions for metal import, metal export, and metal insertion into individual protein targets^9-14. Our current understanding of cellular copper regulation encompasses four major steps: (1) import by the high-affinity copper transporter 1 (CTR1/SLC31A1)^15,16; (2) distribution by copper metallochaperones including antioxidant protein 1 (ATOX1)^17-19, copper chaperone for superoxide dismutase, (CCS)^20-22, and cytochrome c oxidase chaperone 17 (COX17)^23-25; (3) utilization by cuproenzymes or structural cuproproteins such as cytochrome c oxidase (COX) and copper-zinc superoxide dismutase 1 (SOD1); and (4) export by P-type copper-transporting ATPases (ATP7A/B)^26-28. However, this situation gives rise to a disconnect between how cells acquire metals through a single metal uptake source, such as CTR1, and how they get distributed to the multiple metallochaperones needed to deliver metals to diverse downstream targets in the cell.

Here, we report that RAD23 homolog B (RAD23B) provides the missing link that directly connects CTR1 copper uptake to distribution via all known metallochaperone pathways, representing the first example of a metalloadaptor protein. We found this new layer in metal nutrient homeostasis by developing a mouse model of brain copper deficiency localized to the LC. Specifically, we showed that genetic loss of the CTR1 copper ion channel alters sleep-wake cycles by decreasing wakefulness and correlates with a decrease in RAD23B expression. During the amniote transition from animals living in water (amphibians) to on land (reptiles), this ancient DNA repair and ubiquitin adaptor protein^29-33 acquires an allosteric H274/H323 copper-binding motif to not only introduce a new metalloadaptor gain-of-function, but also to confer copper dependency to its canonical functions in DNA nucleotide excision repair (NER). Biochemical studies on RAD23B identify direct protein-protein interactions with both CTR1 and multiple copper metallochaperones (e.g., ATOX1, CCS, COX17) in a copper-dependent manner, and solution NMR studies identify structural changes in RAD23B induced by allosteric copper coordination to regulate its binding to XPC and endow copper-dependent NER to combat UV-triggered DNA damage. Thus, RAD23B metalloadaptor function couples copper acquisition to a broader array of functions to meet evolutionary demands, enabling this metal nutrient to augment respiration and metabolic fitness, tune circadian function, and protect cellular integrity.

RESULTS

High-affinity copper transporter 1 (CTR1) selectively enriches copper in mouse locus coeruleus (LC)

To decipher how certain cell types can tolerate the high amounts of copper needed for metabolic function, we studied LC neurons in the vertebrate brain, an important example of cells with a heightened requirement for copper. In zebrafish, both ctr1 copper uptake and atp7a copper export genes are expressed at a higher level in LC neurons compared to other types of brain cells³⁴. To examine whether this elevated ctr1 expression results in increased copper accumulation within LC neurons, we mapped copper distributions in adult Tg (dbh-mcherry, dopamine beta-hydroxylase) zebrafish brains with Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICP-MS). We did not detect colocalization of high copper concentrations with the soma of LC neurons marked by mCherry (Figure S1A), due to few LC neurons present in these models (6-8 cells per side), along with their small cell size (3-7 μm in diameter), copper levels fell below reliable spatial detection limits of our current LA-ICP-MS instrumentation. We thus switched to mammalian models to map copper distribution by LA-ICP-MS, owing to the presence of a larger number of LC neurons and larger cell size (> 10 μm in diameter). Both mice (Allen Brain Atlas mouse brain in situ hybridization) and human (Allen Brain Atlas, Microarray data) expression databases suggest that the expressions of Ctr1 gene are elevated in LC neurons. We first confirmed the coexpression of Ctr1 and a LC marker, Dbh mRNA, by fluorescence in situ hybridization (Figurer 1A), then mapped copper distributions along with other transition metals, iron and zinc, in mouse brain (Figure 1B). Only copper is selectively enriched in LC neurons, marked by dtTomato using Tg (Dbh^Cre; Ai14).

Figure 1. Conditional Ctr1 knockout mice show that copper is selectively enriched in mouse Locus Coeruleus (LC) neurons and promotes wake behavior.

(A) A 20 μm coronal section of the mouse brain. mRNA expression of Ctr1 and Dbh in mouse LC neurons were detected with fluorescence in situ hybridization. Scale bar, 200 μm.

(B) Transition metals Cu, Fe, and Zn mapped by LA-ICP-MS with Dbh^Cre; Ai14 reporter mice, in which the LC expresses tdTomato. Elevated Cu pools colocalize with tdTomato-positive LC neurons. White boxes highlight regions with LC neurons. Scale bar, 1 mm.

(C) Conditional Ctr1 knockout mice lines generated by crossing TH^Cre or Dbh^Cre driver lines to Ctr1^flox/flox mouse line (Slc31a1tm2Djt/J) were used for this study.

(D) Relative copper levels mapped with LA-ICP-MS are reduced in Th^Cre; Ctr1^flox/flox and Dbh^Cre; Ctr1^flox/flox mice compared to wildtype Ctr1^flox/flox control. Scale bar, 200 μm.

(E) Representative sleep recordings in the control (top) and TH^Cre; Ctr1^flox/flox (bottom) mice from time of day/Zeitgeber time (ZT), over a time-course from 0 to 24 hours. Shown are EEG spectrogram in frequency (Freq.), EMG amplitude, and color-coded brain states.

(F) Percentages of time in each brain state in control and TH^Cre; Ctr ^flox/flox mice (n = 5 mouse pairs).

Data are means ± SEMs.

To test whether high concentrations of copper observed in the mammalian LC are caused by import via CTR1, we generated a conditional knock-out mouse line by crossing TH^Cre (tyrosine hydroxylase, a marker for LC) or Dbh^Cre to Ctr1^flox/flox mice³⁵ and the reporter line Ai14 (Figure 1C). The resulting conditional knock-out of the Ctr1 gene lowers copper levels in the LC region (Figure 1D). DBH uses copper as cofactor, and it has been suggested that NE neurons release copper-bound DBH protein³⁶. To test whether DBH is required to maintain high copper concentrations in the LC, we compared a copper map of LC regions in Dbh^+/− and Dbh^−/− brains. Copper levels appear comparable between these sets of mice (Figure S1B), suggesting that loss of DBH does not appear to significantly alter copper concentrations in LC neurons.

One of the behaviors regulated by the LC is maintaining a wake state. To investigate the functional effects of low LC copper induced by CTR1 loss-of-function, we performed 24-hour sleep recordings in TH^Cre; Ctr1^flox/flox mice and their littermate control mice (Figure 1E). We found that during the active/wake period (i.e., dark phase), but not during the inactive/sleep period (i.e., light phase), low copper altered sleep-wake behaviors. During the dark phase, the percentage of time spent in wakefulness was significantly reduced in TH^Cre; Ctr1^flox/flox mice, resulting in higher non-rapid-eye movement (NREM) and rapid-eye movement (REM) sleep (Figure 1F).

Transcriptome analysis in the LC show that expression of Rad23b, but not Rad23a, is downregulated by Ctr1 loss-of-function

To understand how LC neurons store and tolerate such high levels of copper, we performed Translating Ribosome Affinity Purification followed by RNA sequencing (TRAP-seq) analysis comparing gene expression profiles of LC neurons between Ctr1 KO and wildtype (WT) mice during light and dark phases to assess how copper deficiency affects the LC transcriptome (Figure 2A). Dbh^Cre; Ctr1^flox/flox mice were crossed with EGFP-L10a mice³⁷ to generate mice with EGFP-tagged ribosomal protein L10a in LC neurons. LC mRNA was purified by affinity purification³⁸ and sequenced. For this study we focused on male populations, as male and female LC neurons have different gene expression profiles, including genes known to be enriched in females (Figure S2A and Table S1). LC neurons appear to have 1/3 of their genes showing circadian-regulated differential expression (threshold > 1.5 fold, p < 0.01) (Figure S2B). Interestingly, this list includes the NE transporter Slc6a2 and key copper homeostasis genes showing differential expression bias in inactive and active periods (Figure S2C). Specifically, plasma membrane-associated copper importer Ctr1 and copper exporter Atp7a genes are expressed at higher levels during the inactive/sleep phase similar to Slc6a2, while intracellular copper-binding proteins such as metallochaperones are expressed at higher levels during active/wake phase (Figure S2C). Because expression of Ctr1 is higher during the inactive phase than active phase, we reasoned that Ctr1 knockout during the inactive/sleep phase would produce stronger transcriptome differences. Indeed, we confirmed this hypothesis by comparing gene expression profiles between WT and Ctr1 knockout LC mice (Figure 2B) and we identified one gene, Rad23b, that showed particularly strong downregulation by Ctr1 loss-of-function (Figure 2B and Table S2). Other genes downregulated by Ctr1 loss-of-function include Chchd10, Cartpt, Pakap, Selenow, and Ndufb8, which potentially are required for maintaining LC neuronal function through boosting mitochondria function, producing neuropeptides, or offering neural protection. Upregulated genes include Alad, Hdhd3, Gm20075 (BT3L4), Mroh7, and Hoxb5. Rad23b is an ancient and highly conserved gene, evolved from Rad23 in single-cell eukaryotes such as Saccharomyces cerevisiae, after a gene duplication event in chordata (Figure S2D). In contrast, the gene expression of its paralog, Rad23a, shows no correlation with Ctr1 (Figure 2B).

Figure 2. Transcriptome analysis in the mouse LC identifies Rad23b as a copper binding protein.

(A) Design of TRAPseq analysis for LC neurons in wildtype and LC conditional Ctr1 knockout mice using Dbh^cre; EGFP-L10a mice.

(B) Differential gene expression of Dbh^cre; EGFP-L10a and Dbh^cre; EGFP-L10a; Ctr1^flox/flox LC neurons during resting/sleep phase.

(C) Immunofluorescence analysis of TH and RAD23B protein expression in LC neurons in 20 μm coronal sections of the mouse brain. Scale bars, 200 μm.

(D) The binding of RAD23B to indicated metal was assessed by immunoblot analysis of eluted proteins from the indicated metal-charged resins.

(E) RAD23B sequence alignment from indicated species. Orange arrows show key histidine sites conserved in amniotes (H274/H323 in human sequence). Blue arrows and dashed line indicate mutation from NQ to HH in the transition from amphibians to reptiles.

(F) Evolutionary path for copper regulation in LC neurons, with acquisition of the bis-histidine motif in RAD23B within the vertebrate phylogenetic clade.

(G) AlphaFold2-predicted structure of XPC binding domain in human RAD23B showing the putative copper-binding residues H274 and H323. Distance is measured between H274 and H323 Nε₂ atoms. Confidence is based on the predicated local-distance difference test (pLDDT). Very high, pLDDT > 90, confident, 90 > pLDDT > 70, low, 70 > pLDDT > 50, very low, pLDDT < 50.

(H and I) The binding of RAD23B variants to copper was assessed by immunoblot analysis of eluted proteins from the Cu-charged resins. The variants were ectopically expressed in HEK293T cells. Quantification of V5 intensity relative to RAD23B-WT was shown (I, n = 3). ACTIN, loading control.

(J) SDS-PAGE analysis of recombinant RAD23B protein from uncharged or Cu-charged resin.

(K) Spectra of Cu(II)-DP2 upon addition of recombinant RAD23B proteins.

(L) X-band CW EPR spectra of the WT (blue) and NQ (yellow) RAD23B proteins with simulations for the WT (red) confirming Cu(II) binding. Spectra were recorded at 15K. Simulation parameters for WT RAD23B: g = [2.25 2.05], A(^63/65Cu) = [530 40] MHz.

(M) Representative titration of RAD23B-Cu(I) with different concentrations of bicinchoninic acid (BCA).

(N) Measured stoichiometries of RAD23B towards Cu(I) and Cu(II).

WT, wild type; NQ, RAD23B-H274N/H323Q.

All data are means ± SDs except in (M) are means ± SEMs. Welch’s t test was performed. ns, not significant.

Mammalian RAD23B, when complexed with xeroderma pigmentosum group C (XPC), protects the genome from DNA damage through nucleotide excision repair (NER) ³². Human RAD23B contains a ubiquitin-like (UBL) domain, two ubiquitin-associated (UBA) domains and a heat-shock chaperone-binding (STl1) domain, also known as XPC-binding (XPCB) domain ²⁹. In the cytosol, RAD23B functions as a ubiquitin adaptor that recruits polyubiquitinated protein to 26S proteasome for degradation. It is also directly involved in stress and ubiquitylation-dependent phase separation of the proteasome ³³. However, to date, it has not been associated with pathways in copper homeostasis. Immunofluorescence analysis with mouse brain tissue confirms that the RAD23B protein is indeed present in elevated levels in LC neurons compared with other brain cells (Figure 2C).

RAD23B acquires a copper-binding histidine motif in the amniote transition between amphibians and reptiles

Given that Rad23b expression is strongly associated with Ctr1 and intracellular copper concentrations, we asked whether the RAD23B protein could directly bind to copper. To test this hypothesis, we extracted total proteins from various human cell lines, including HEK293T, SH-SY5Y, and A-375, and mouse embryonic fibroblasts (MEFs). Extracted proteins were passed through a panel of metal-charged resins to test their metal-binding affinity and selectivity. Interestingly, we found that RAD23B binds copper-charged resins, but not resins charged with iron, cobalt, nickel, or zinc (Figure 2D).

To identify potential copper-binding residues, we compared RAD23B protein sequences of various vertebrate species from bony fish to mammals (Figure 2E). We found that two histidine residues in the human sequence, H274 and H323, are conserved among amniote species and appear to be evolved from asparagine (N) and glutamine (Q) sites, respectively, between amphibians and reptiles (Figure 2E). This evolutionary transition is coincident with when rapid-eye movement (REM) sleep emerges³⁹ (Figure 2F). The protein structure predicted by AlphaFold^40,41 shows that these two histidine residues are in close proximity with a slightly offset face-to-face display (Figure 2G), suggesting a potential copper-binding site. To test whether H274 and H323 could be involved in copper coordination, we used targeted mutagenesis to assess their effects on RAD23B copper-binding affinity. While mutation on H320 (the conserved histidine residue across all vertebrates, Figure 2E) showed negligible effect, mutations on both H274 and H323 significantly decreased the copper-binding affinity to both ectopically expressed RAD23B in HEK293T cells (Figures 2H and 2I) and recombinant RAD23B proteins (Figures S2E and S2F). Moreover, the double mutation on H274 and H323 further reduced RAD23B binding to the copper-charged resin (Figures 2H and 2I).

We therefore focused on the roles of H274 and H323, and purified the recombinant proteins RAD23B-WT and RAD23B-NQ (RAD23B-H274N/H323Q, NQ), in which these two residues are mutated to asparagine (N) and glutamine (Q) respectively, corresponding to amino acids present in RAD23B protein sequences before Reptilia (Figure 2E). RAD23B-NQ shows negligible binding to a copper-charged resin (Figure 2J) and is incapable of competing for copper against the copper-specific fluorescent probe DP2⁴² (Figure 2K). It also shows no copper-dependent hyperfine signatures in the X-band continuous wave (CW) electron paramagnetic resonance (EPR) spectrum (Figure 2L), further indicating it is unable to bind copper, whereas RAD23B-WT binds Cu(II) and displays typical EPR features. Interestingly, titration experiments^42-44 suggest that RAD23B can bind copper in both Cu(I) (Figure 2M) and Cu(II) (Figures S2G and S2H) oxidation states, with higher affinity towards Cu(I) (Figure S2I). K_d values are 6.3 × 10⁻¹⁴ M for Cu(I) and 5.1 × 10⁻¹¹ for Cu(II). RAD23B binds copper with a 1:1 metal:protein stoichiometry as measured by inductively coupled plasma mass spectrometry (ICP-MS) (Figures 2N and S2J). Taken together, these data establish that key H274 and H323 residues lead to a gain-of-function for RAD23B in the transition of animals from water to land that enable it to bind copper, while its ancestor protein without these two histidine residues is not able to bind copper.

CTR1 binds to RAD23B through direct protein-protein interactions in a copper-dependent manner

Given that Ctr1 loss-of-function decreased Rad23b mRNA, we asked whether RAD23B protein levels are also reduced by CTR1 knockout. To quantify the amount of RAD23B protein, we performed immunoblotting analysis with genetically matched mouse Ctr1^+/+ and Ctr1^−/− MEF cells⁴⁵ (Figure 3A), and human breast cancer LM2 control (SCR) and CTR1 knockout (CTR1KO) cells⁴⁶ (Figure 3B). In both models, RAD23B protein levels are decreased in Ctr1-deficient cells, while its paralog, RAD23A, remains unchanged. This decrease was also observed in other cell types with transient Ctr1-knockdown by short-interfering RNAs (siRNAs) (Figures S3A and S3B). Interestingly, adding copper ionophores Cu-GTSM (Figure S3C) and Cu-ATSM (Figure S3D) which diffuses passively into cytosol bypassing CTR1 has negligible effect on RAD23B protein level. Furthermore, immunofluorescence analysis confirmed the RAD23B protein is reduced in LC neurons marked by TH in Dbh^cre; Ctr1^flox/flox mice (Figure 3C). These collective data suggest a strong correlation between RAD23B and CTR1 expression.

Figure 3. RAD23B binds to high-affinity copper transporter CTR1.

(A and B) Immunoblot analysis and corresponding quantification of RAD23A and RAD23B expression from Ctr1^+/+ or Ctr1^−/− MEFs (A, n = 5), and control (SCR) and CTR1 knockout (CTR1KO) LM2 cells (B, n = 3). VINCULIN, loading control.

(C) Immunofluorescence analysis RAD23B expression in the brain slides from wild-type (WT) and LC-Ctr1^−/− mice. TH, LC marker, DAPI, nuclei marker. Scale bar, 200 μm.

(D) Immunoblot analysis of ectopically expressed CTR1-myc immunoaffinity purified from HEK293T CTR1-Myc cells transfected with indicated plasmids. TTM, bis-choline tetrathiomolybdate. WT, wildtype. NQ, RAD23B-H274N/H323Q. EV, empty vector.

(E) Schematic of CTR1 interaction with RAD23B mediated by copper.

(F) Schematic of probing RAD23B copper occupancy using an activity-based histidine bioconjugation reaction in the presence or absence of CTR1. O5C-TPAC, 2-Propargyloxyethyl thiophosphorodichloridate.

(G) Immunoblot analysis of O5C-TPAC labeling on RAD23B in Ctr1^+/+ or Ctr1^−/− MEFs and the quantification results (n = 3). VINCULIN, loading control.

All data are means ± SDs. Welch’s t test was performed. ns, not significant.

We then tested whether RAD23B could directly bind to CTR1. First, we assessed their direct binding via co-immunoprecipitation from cells ectopically expressing Myc-CTR1 and V5-RAD23B. A stable complex of RAD23B-WT and CTR1 was only detected under treatment with a copper chelator tetrathiomolybdate (TTM) to induce copper deficiency, and not under either basal (i.e., with the presence of endogenous copper) or elevated copper conditions (Figures 3D and S3E-S3G). Moreover, the RAD23B-NQ variant, which lacks copper-binding capacity, does not bind CTR1 regardless of TTM addition (Figures 3D, S3E, S3G and S3H). Mutating the putative copper binding sites on the C-term of CTR1⁴⁷ abolish their interactions as well (Figure S3I). No interaction was observed on the divalent metal transporter 1 (DMT1/SLC11A2)⁴⁸ (Figure S3J). These results show that only apo-RAD23B binds to CTR1 and suggest that holo-RAD23B dissociates upon copper loading (Figure 3E).

Second, we asked whether CTR1 metalates RAD23B through its bis-histidine motif. The status of RAD23B copper occupancy was detected using a chemoselective histidine bioconjugation approach developed by our laboratory that relies on thiophosphorodichloridate reagents⁴⁹ in a competitive metal-binding assay mode. In the absence of copper, the thiophosphorodichloridate reagents can efficiently label histidine residues. Upon copper binding, histidine residues lose their ability to be labeled by the thiophosphorodichloridate reagents due to competition with the metal. If RAD23B metalation is CTR1-dependent, RAD23B in wildtype cells is expected to have diminished histidine labeling, while RAD23B in Ctr1-deficient or other loss-of-function cells is expected to have elevated histidine labeling (Figure 3F). The first-generation thiophosphoro alkyne dichloridate (TPAC) probe, despite the histidine specificity, fails to form a stable adduct after copper-assisted alkyne-azide cycloaddition (CuAAC) chemistry, presumably in an analogous way to palladium catalysts that hydrolyse the propargyl carbamate in biologically relevant conditions^50-52 (Figure S3K). To improve the stability of TPAC-histidine adducts, we designed and synthesized a panel of second-generation TPAC histidine bioconjugation probes (Figure S3L). Of these probes, O5C-TPAC (Figure 3F) outperforms the other members of this family, as well as the original TPAC probe, and can achieve histidine-specific protein bioconjugation with a stable labeling and click-reaction product (Figures S3M and S3N). With this probe, we performed histidine labeling on the lysates from cells with or without CTR1 expression. We observed that O5C-TPAC labeling of RAD23B protein showed a significant increase in Ctr1-deficient cells relative to wildtype controls (Figures 3G, S3O and S3P). Moreover, treatment with the copper chelator TTM to induce copper deficiency also leads to increased RAD23B labeling compared to the control cells (Figure S3Q), while adding copper ionophore Cu-GTSM does not affect RAD23B labeling in either Ctr1-sufficient or Ctr1-deficient cells (Figures S3R and S3S), indicating that the histidine-dependent labeling of RAD23B by TPAC reagents is regulated by intracellular copper availability from CTR1. CTR1 membrane localization is reduced in Rad23b KO cells (Figure S3T). Taken together, these data demonstrate that RAD23B binds to copper in cells in a histidine-dependent manner and support a model where CTR1 metalates RAD23B through direct copper ion transfer.

RAD23B is a master metalloadaptor protein that mediates copper transfer from CTR1 to multiple copper metallochaperones

We then asked whether RAD23B can distribute copper to downstream intracellular copper-trafficking networks. As no significant changes on the total cellular copper concentration (Figures S4A and S4B) nor the labile copper pool (Figures S4C and S4D) were observed in Rad23b-deficient cells, we thus turned our attention to copper chaperones, whose function is to traffic and insert copper into their specific downstream target proteins. The major copper chaperones include ATOX1 (delivering copper to ATP7A and ATP7B)^17-19, CCS (delivering copper to superoxide dismutase)^20-22, and COX17 (delivering copper to cytochrome c oxidase)^23-25.

We first tested RAD23B-ATOX1 binding by co-immunoprecipitation assays using HEK293T (Figures 4A and 4B) and A-375 (Figure S4E) cells. In both cases, we observed co-immunoprecipitation of RAD23B and ATOX1 under basal conditions, while mutated variants of either RAD23B (RAD23B-NQ) or ATOX1 (ATOX1-C12S/C15S, CS)⁵³ that do not bind copper fail to form this complex (Figure 4A). Along the same lines, intracellular copper depletion using the metal chelator TTM also decreases the interaction between RAD23B and ATOX1 (Figure 4B). Moreover, size-exclusion chromatography (SEC) analysis on purified recombinant ATOX1 and RAD23B proteins demonstrates a stable ternary complex of RAD23B-Cu-ATOX1 and generation of Cu-ATOX1 (Figure 4C), suggesting a copper exchange mechanism in which RAD23B directly binds to and transfers copper to ATOX1. These data are consistent with the measured K_d values for Cu(I) binding to RAD23B and ATOX1 on purified proteins in vitro, with ATOX1 binding Cu(I) much tighter than RAD23B (10⁻¹⁷ M for ATOX1 vs 6.3 × 10⁻¹⁴ M for RAD23B)^54,55. To further test this copper transfer mechanism in cells, we used iodoacetamide-biotin (IA-Biotin) as a cysteine labeling reagent to probe the copper occupancy of ATOX1 cysteine residues in Rad23b-deficient or control cells (Figure 4D). Akin to the copper/TPAC competition assays to probe histidine-dependent copper binding in RAD23B (Figures 3G and S3O-S3Q), copper-bound cysteine residues should have decreased IA-biotin labeling. Indeed, we showed that RAD23B down-regulation significantly increased IA-biotin labeling, suggesting decreased copper occupancy on ATOX1 (Figures 4E and S4F). Notably, this increased IA-biotin labeling signal was also observed in the copper-deficient cells either lacking CTR1 (Figures 4F and S4G) or treated with TTM chelator (Figure S4H). Moreover, overexpressing of ATOX1 by transfection results in decreased TPAC labeling on RAD23B (Figures S4I and S4J), suggesting ATOX1 is able to compete for and strip copper from RAD23B in a cellular context. Together, these data establish that RAD23B directly binds to ATOX1 in a copper-dependent manner and defines ligand sites for the metal on both partners, supporting a model where RAD23B provides a source of copper for ATOX1.

Figure 4. RAD23B is a master metalloadaptor protein delivering copper to multiple copper metallochaperones.

(A and B) Immunoblot analysis of ATOX1-Flag immunoaffinity purified from HEK293T transfected with indicated plasmids. CS, ATOX1-C12S/C15S.

(C) SEC runs of recombinant ATOX1 and RAD23B proteins under indicated treatment. Protein in the indicated fractions was subject to immunoblot analysis.

(D-F) Copper occupancy on ectopically expressed ATOX1-FLAG in HEK293T with siRNAs against Rad23b (E) or Ctr1 (F) was assessed by reactive cysteine targeting probe iodoacetamidebiotin (IA-biotin) (n = 3). Schematic illustration is shown in (D). KD, knockdown.

(G) Immunoblot analysis of CCS-Flag immunoaffinity purified from HEK293T transfected with indicated plasmids. AS, CCS-C22/25A-C244/246S.

(H) Schematic showing copper transport in RAD23B-CCS–SOD1 pathway.

(I and J) SOD1 activity analysis was performed in HEK293T (I, n = 3) and MEF cells (J, n = 6) transfected with indicated siRNAs.

(K and L) Immunoblot analysis of COX17-Flag immunoaffinity purified from HEK293T transfected with indicated plasmids. CG: COX17-C23G/C24G/C26G.

(M and N) SEC analysis of ectopic COX17-Flag from HEK293T cells with indicated siRNAs. Schematic model showing copper-induced COX17 polymerization in (M). The bar graphs show the distribution of COX17-Flag from each fraction.

(O) Cytochrome c oxidase (COX) activity analysis was performed in SCR or CTR1KO LM2 cells with indicated treatment. 10 μM TTM or 40 nM siRNA was used (n = 3). Left, schematic model showing copper transportation in COX17-mitochondria pathway.

(P) Schematic of cellular copper trafficking pathways. The dashed box highlighting the new layer of regulation using RAD23B as a master metalloadaptor as a copper conduit to connect the single copper uptake protein CTR1 with multiple downstream copper chaperones ATOX1, CCS, and COX17 discovered in this work.

WT, wild type. NQ, RAD23B-H274N/H323Q. EV, empty vector. TTM, bis-choline tetrathiomolybdate. IP, immunoprecipitation

All data are means ± SDs. Welch’s t test was performed. ns, not significant.

To explore the generality of RAD23B-metallochaperone interactions, we next tested the interaction between RAD23B and CCS as a second metallochaperone target. Similarly, RAD23B binds to CCS in a copper-dependent manner, where only wildtype RAD23B and CCS form a stable complex in the presence of copper (Figures 4G and S4K). As the IA-biotin copper displacement assay showed insufficient sensitivity to probe CCS copper occupancy under these conditions, presumably due to multiple copper-binding domains and a larger number of non-copper-binding cysteines in CCS relative to ATOX1⁵⁶ (Figure S4L), we performed direct activity measurements of SOD1, the copper delivery target protein of CCS²², as a functional proxy for CCS activity (Figure 4H). Indeed, loss of RAD23B decreases the SOD1 activity in both human (Figures 4I, S4M and S4N) and murine (Figures 4J and S4O) cells, showing a similar loss-of-function in the CCS-SOD1 pathway as observed by copper deficiency induced through CTR1 depletion.

Finally, we established a direct protein-protein interaction between RAD23B and COX17, with similar observations to RAD23B-ATOX1 and RAD23B-CCS complexes (Figures 4K, 4L and S4P). Since copper binding induces COX17 oligomerization⁵⁷ (Figure 4M), we performed size exclusion chromatography (SEC) to analyse oligomeric states of COX17 in cells with or without RAD23B as a functional assay (Figures 4N, S4Q and S4R). We observed that loss of RAD23B significantly decreased the population of tetrameric COX17, consistent with a deficit in copper loading. A similar effect was also observed in Ctr1-deficient cells. As COX17 delivers copper to Cytochrome c Oxidase (COX)^23,25, we tested whether loss of RAD23B affects COX activity. As expected, loss of RAD23B results in decreased COX activity, showing a similar effect to other forms of induced copper deficiency, including genetic loss of CTR1 or copper chelation by TTM. Furthermore, down-regulation of RAD23B in Ctr1-deficient cells does not appear to exert further effects on COX activity, suggesting that RAD23B is downstream of CTR1 in copper-trafficking networks (Figures 4O and S4S). Notably, RAD23B puncta are detected in mitochondria marked by ATP5A (Figure S4T) in A-375 cells, showing that it can localize to this organelle.

Taken together, these findings support a model where RAD23B acts as a copper conduit connecting the single major copper uptake protein CTR1 with an array of intracellular copper metallochaperones. In this model, RAD23B associates with CTR1 under copper-deficient conditions and dissociates from CTR1 under copper-replete conditions upon copper loading, holding copper when this nutrient is abundant. RAD23B then transfers copper to tighter-binding copper metallochaperones (e.g., ATOX1, CCS and COX17) when the need arises to metalate these diverse pathways in the cell. Importantly, we established that the loss of RAD23B impairs the copper-loading efficiency and downstream functional activity of all three major copper trafficking pathways. As such, we term this unique function of RAD23B as a “metalloadaptor” protein that links metalloacquisition and metallochaperone pathways with the metal nutrient copper as a substrate, akin to adapter proteins which bring protein partners in proximity to substrates and their upstream and downstream regulators (Figure 4P). The metalloadaptor function of RAD23B thus adds a new layer of regulation of copper homeostasis in higher organisms.

Copper-bound RAD23B improves cell metabolic fitness while making its ancient nucleotide excision repair function copper-dependent

As a starting point to probe functional consequences of copper-RAD23B interactions, we observed that RAD23B-deficient cells exhibit diminished growth rates compared to wildtype controls, similar to what is observed for Ctr1-deficient cells (Figure S5A). As RAD23B feeds into the COX17-COX pathway to maintain the respiratory electron transport chain, we reasoned that this phenotype could be a result of deficient mitochondrial function. To test this hypothesis, we performed oxygen consumption rate (OCR) assays for mitochondrial activity in RAD23B-deficient versus wildtype cells. We observed a decreased rate of oxygen consumption in both human (Figure 5A) and murine (Figures S5B and S5C) cells lacking RAD23B relative to wildtype congeners. Notably, loss of RAD23B results in a marked 2-fold reduction in basal respiration (Figures 5B and S5C) and a 3-fold reduction in ATP-linked respiration (Figure 5C) compared to control cells. A similar effect was observed in cells lacking CTR1, suggesting that loss of RAD23B disrupts copper homeostasis and cell respiration in a related manner. Interestingly, coordinated dependencies across the Cancer Dependency Map (www.depmap.org) show a strong correlation of RAD23B and genes involved in mitochondrial activities (Figure S5D). Taken together, these data suggest that RAD23B improves cellular metabolic fitness through enhancing copper-dependent mitochondrial respiration.

Figure 5. Holo-Rad23B modulates mitochondrial function through copper-dependent metalloallosteric regulation while maintaining nucleotide excision repair (NER) function.

(A) Oxygen consumption rate (OCR) measurements in SCR or CTR1KO LM2 cells transfected with indicated siRNAs (n = 5/group).

(B and C) Quantification of basal respiration (B) and ATP-linked respiration (C) from metabolic measurements.

(D and E) Immunoblot analysis of XPC immunopurified from HEK293T or A-375 transfected with indicated plasmids upon incubation with 10 μM TTM (D) or 200 μM CuCl₂ (E). IP, immunoprecipitation. WT, wild type. NQ, RAD23B-H274N/H323Q. EV, empty vector. TTM, bis-choline tetrathiomolybdate. ACTIN, loading control.

(F) Representative images of comet assay in Xpc^+/+, Xpc^−/− BI6 mESCs. Scale bar, 50 μm.

(G) Cell viability of Xpc^+/+ BI6 mESCs after UV exposure.

(H) Schematic of RAD23B-XPC complex participating in the UV-damaged double strand DNA repair pathway, with copper-binding evolving as an additional layer of regulation.

(I and J) OCR measurement in Rad23b^+/+, Rad23b^−/− JM8.N4 mESCs (I) and the calculated basal respiration (J, n = 3/group).

(K and L) OCR measurement in Xpc^+/+, Xpc^−/− BI6 mESCs (K) and the calculated basal respiration (L, n = 5/group).

All data are means ± SDs, except in (B), (C), (J) and (L) are means + SDs. Welch’s t test was performed. ns, not significant.

We also sought to interrogate the contributions of copper binding to canonical activities of RAD23B. As one of the ancient functions of RAD23B is to participate in the ultraviolet (UV)-induced DNA damage repair pathway through forming a heterodimer structure with xeroderma pigmentosum group C protein (XPC) for nucleotide excision repair (NER)^32,58, we sought to probe how copper affects this process underlying cell integrity. Interestingly, co-immunoprecipitation assays showed that, while the copper binding mutation variant, RAD23B-NQ, has strong binding affinity to XPC, only the holo form, but not the apo form of RAD23B-WT, binds to XPC (Figures 5D and 5E). Copper chelation by TTM decreases RAD23B-XPC binding (Figure 5D), while copper supplementation increases their interaction (Figure 5E). Consequently, a marked increase in DNA damage (Figure 5F) and lower viability was observed in copper-chelated cells (Figures 5G and S5E) after UV irradiation. Similarly, copper also modulates RAD23B binding to ubiquitinated proteins when it works as a ubiquitin receptor to stimulate protein degradation by proteasome^30,31 (Figures S5F-S5K) and accumulation of more ubiquitinated proteins was observed in Rad23b-deficient cells under copper stressed conditions (Figure S5L). Finally, inspired by pioneering work on copper reductase activity of histone H3⁵⁹, we also tested whether RAD23B would bind this potential target in a copper-dependent manner but found that both RAD23B and RAD23B-NQ bind to it, in a copper-independent manner (Figure S5M).

These results suggest that the evolutionary acquisition of copper-binding capacity by RAD23B through acquisition of this special histidine pair not only adds new functions to RAD23B in boosting mitochondrial respiration and metabolic fitness, but also enables copper to tune its ancient function in DNA repair to maintain cell integrity (Figure 5H). To provide further support for this model, we performed oxygen consumption assays in murine embryonic stem cells (mESCs) with either knockout of Rad23b or Xpc⁶⁰. Cell respiration is dramatically lower in Rad23b^−/− mESCs compared to wildtype control cells (Figures 5I, 5J and S5N). In contrast, cell respiration capacity remains unchanged in Xpc^−/− mESCs (Figures 5K, 5L and S5O), indicating that the contribution of RAD23B to functional mitochondrial activity is independent of XPC and separate from its roles in copper-dependent DNA repair.

Probing copper-dependent modulation of RAD23B structure using nuclear magnetic resonance (NMR) spectroscopy

To gain structural insights into the copper-dependent modulation of RAD23B function, we performed solution NMR analysis on a truncated form of RAD23B containing the H274/H323 copper-binding motif, noting that this sequence is also within the XPC binding domain (Figure S6A). We confirmed that this truncated protein RAD23B_272-342 can bind copper by a DP2 competition assay (Figure S6B). The structures of RAD23B_272-342-WT and RAD23B_272-342-NQ were determined using solution NMR (Figures 6A, S6C and S6D, and Table S3). To calibrate our results, we note that the RAD23B_272-342-WT structure we obtained showed similar features to the published NMR structure²⁹ (PDB: 1PVE, Figure S6E). When superimposed with RAD23B_272-342-WT on Helix2 and 3 (α2 and α3), the RAD23B_272-342-NQ variant shows significant movement of the N-terminus and Helix4 (α4) (Figure 6A), forming a more compact conformation than RAD23B_272-342-WT. This notable difference is supported by the long-range nuclear Overhauser effect (nOe) maps, in which more interactions between α2 and α3, as well as α2 and α4, are detected in RAD23B_272-342-NQ relative to RAD23B_272-342-WT (Figure S6F). When NQ and WT NMR models are fitted onto the XPC-RAD23B complex EM map⁶¹, helices α2, α3, and α4 participate in the interactions with the long helix in XPC (Figure S6G). More residues in the contact interface (distance < 4 Å) in RAD23B-XPC models were seen in RAD23B_272-342-NQ over RAD23B_272-342-WT (Figures S6H and S6I), supporting that the amniote RAD23B-WT is a weaker binder towards XPC compared to the ancient RAD23B_272-342-NQ variant.

Figure 6. Solution nuclear magnetic resonance (NMR) structures of RAD23B.

(A) Ribbon diagram of superimposed NMR structures of RAD23B_272-342-WT and RAD23B_272-342-NQ. Structures were superimposed relative to residues 25 to 50.

(B) Ribbon diagram of Ag(I) bound RAD23B_272-342-WT showing Ag coordination to the H274/H323 bis-histidine motif.

(C and D) Structure comparison between Ag(I)-bound and apo RAD23B_272-342-WT (C), or RAD23B_272-342-NQ (D). Structures were superimposed relative to residues 25 to 50.

(E) Cell viability curves of A-375 cells with indicated siRNAs. Lines are best-fit curves. Data are means ± SDs, n = 5.

(F) Schematic model showing how the CTR1-RAD23B axis regulates copper-dependent sleep-wake cycle.

We then performed an NMR chemical shift perturbation assay to identify the residues directly involved in binding Cu(I), the diamagnetic oxidation state of this metal. Backbone chemical shift assignments for the apo and Cu(I)-bound forms of RAD23B_272-342-WT reveal that H274 and α4 are strongly affected by Cu(I) addition, suggesting direct binding (Figure S6J). To overcome oxygen-dependent stability issues of Cu(I) during NMR data collection, we used Ag(I) as an isoelectronic metal substitute for Cu(I) and determined the Ag(I)-bound NMR structure of metal-bound RAD23B. The chemical shift perturbation assay shows addition of Ag(I) and Cu(I) have similar profiles (Figures S6J and S6K). The NMR structure of Ag(I)-bound RAD23B_272-342-WT (Figures 6B and S6L, and Table S3) reveals Ag(I)-nitrogen coordination through H274 and H323 in the metal center (Figure 6B). Although no drastic global conformational changes were observed in the Ag(I)-bound structure, Ag(I)-binding shifts the RAD23B_272-342-WT structure into a conformation similar to RAD23B_272-342-NQ. In this metal-bound form, α4 is in a position that aligns well with NQ (Figures 6C and 6D) and α2 shows more interactions with α3 and α4 (Figures S6M and S6N). The collective structural data suggest that changes from N274/Q323 ancient form to H274/H323 amniote form of RAD23B loosen the conformation in its XPCB domain, while metal binding restores the tighter conformation suited for XPC helix binding. Thus, because it has distinct interfaces for metal and XPC binding, RAD23B provides another biochemically well-characterized example of metalloallostery^62,63 for endowing copper-dependent regulation to its canonical NER function. As only holo-RAD23B binds to XPC, which then initiates NER, we hypothesized that RAD23B-XPC protects cell integrity under elevated copper conditions. To test this hypothesis, we measured the cell viability under different copper concentrations. Indeed, knockdown of RAD23B decreases the IC₅₀ value of A-375 cells under a copper challenge relative to wildtype controls, showing that RAD23B-deficient cells are particularly vulnerable to an excess of this nutrient (Figure 6E).

DISCUSSION

The identification of RAD23B as a foundational component for recruiting copper to increase cell metabolic fitness provides a missing link to how animals meet rising energy demands in the transition from water to land environments. In dividing cells, copper regulates growth, and in differentiated cells, it boosts activity and function. We showed that by acquiring a key histidine pair to evolve the ability to bind copper in the amphibian-to-reptile transition, RAD23B gained new cellular functions in enhancing respiration and metabolic fitness while adding copper as a modulator to its ancient functions in DNA repair and protein quality control (Figure 6F). CTR1 in the LC imports and enriches copper, which is vital for promoting active/wake states in mice, and that loss of CTR1 results in lower expression of RAD23B. Further cell-based studies support a model in which RAD23B functions as a metalloadaptor, serving as a copper conduit connecting a single major copper-acquiring protein CTR1 at the cell membrane with multiple intracellular metallochaperone partners. In this model, apo-RAD23B binds to CTR1, and upon receiving copper from CTR1 it dissociates and then distributes copper to various copper chaperones (e.g., ATOX1, CCS, and COX17). RAD23B not only directly interacts with the copper acquisition protein CTR1 at the cell membrane, but it also interacts with the conventional intracellular metallochaperones ATOX1, CCS, and COX17 that deliver copper to downstream target proteins across different locations in the cell. The protein-protein interactions between RAD23B and CTR1 and RAD23B with ATOX1, CCS, and COX17 are copper-dependent. As such, we term its unique function as a “metalloadaptor” protein that links metalloacquisition and metallochaperone pathways using copper as a nutrient substrate, akin to the biochemical definition of adaptor proteins which bring protein partners in proximity to substrates and their upstream and downstream regulators.

This metal-dependent gain-of-function for RAD23B occurs through acquisition of two key histidine residues in amniotes, between amphibians and reptiles, at an evolutionary transition that coincides with the emergence of REM sleep³⁹ (Figure 2F). As such, we posit that this additional layer of copper regulation primes the wake-promoting LC neurons to maintain circadian rhythms. By acquiring this bis-histidine motif for copper binding, RAD23B enables enhanced metabolic fitness by boosting mitochondrial activity. At the same time, copper was added a metalloallosteric modulator for ancient RAD23 functions in DNA repair and protein quality control to mitigate copper toxicity, simultaneously protecting cell against reactive oxygen species (ROS)-mediated oxidative stress and damage. Thus, copper serves as a central hub to integrate foundational cell activities spanning genome and proteome stability to metabolism by metalloallostery.

The copper-dependent gain-of-function for RAD23B has broader implications for evolutionary biology. For example, a rise in tissue copper concentrations in mosaic patterns through development of increasingly specialized cell types during evolution could be the direct selection pressure for retaining histidine mutations in RAD23B, giving an advantage for storing excessive copper and ameliorating DNA damage, which in turn augments resistance to nutrient and oxidative stress and maintains cell integrity. A fascinating related question is understanding what is the nature of this selection pressure between amphibians and reptiles, which ultimately gave rise to the copper metalloadaptor function of RAD23B. One possible reason is a change in reproductive strategy, as fecundity in reptiles is significantly lower than in amphibians while indirect cost for reproduction escalates. For the former, survival rate of their offspring is even more critical for the survival of the species and developing more sophisticated maternal behaviours confer an advantage.

A final functional consequence of copper metalloallostery regulating RAD23B is that it protects cells against copper toxicity while enhancing metabolic fitness, making it a central player in copper-dependent cell death and proliferation pathways, termed cuproptosis^64,65 and cuproplasia⁶⁶, respectively. Indeed, RAD23B has long been known as a risk factor for cancer^32,67 and copper is emerging as a key regulator of cancer biology⁶⁶. Along these lines, both Rad23b^−/− and Ctr1^−/− cells in culture show poor growth relative to wildtype controls. Furthermore, Rad23b KO and Ctr1 KO mice exhibit similar phenotypes, where homozygous pups are mostly embryonic lethal with defects in facial structure associated with connective tissue issue. If any pup survives birth, it always fails to thrive and dies quickly after birth⁶⁸. We speculate that these early developmental defects could in part be due to compromised cellular respiration and deficiencies in metabolic fitness. Given that the copper-RAD23B interaction also regulates DNA repair and protein quality control, it provides a novel metal-dependent disease vulnerability that could be exploited for diagnostics and therapeutic intervention. Indeed, the growing connections between copper and diseases such as cancer⁶⁶, neurodegenerative diseases^6,69, and metabolic disorders^70,71, add to a broader body of work in transition metal signaling via metalloallostery^63,72,73.

Limitations of the Study

Our experiments determining how copper modulates RAD23B binding to XPC relied on solution NMR analysis, which has a limited size range, so our studies were performed on truncated, soluble forms of the protein. Structural studies on the full-length RAD23B and its copper-dependent binding to interaction partners using other techniques warrant further investigation. Our analysis comparing copper-dependent RAD23B and copper-independent RAD23-NQ variants was primarily conducted in cell-based models, with animal models containing either RAD23B or RAD23-NQ. The study did not address comparative studies in the same animal species with RAD23B and RAD23B-NQ variants—another area that warrants further investigation.

STAR METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell lines and culture conditions

HEK293T, SH-SY5Y and A-375 cells were from ATCC (CRL-3216, CRL-2266, CRL-1619), MEF Ctr1^+/+ and Ctr1^−/− cells were from the Donita Brady laboratory (University of Pennsylvania), LM2 SCR and CTR1KO cells were from the Vivek Mittal laboratory (Weill Cornell Medicine). HEK-293T and A-375 cell lines stably expressing N-terminal c-Myc epitope tagged hCTR1 were generated as previously described^74,75. All cells except SH-SY5Y cells were cultured in DMEM (Gibco, Cat# 11995073) supplemented with 10% fetal bovine serum (FBS) (Corning, Cat# 35-010-cv). SH-SY5Y cells were cultured in DMEM (Gibco, Cat# 11995073) supplemented with 10% fetal bovine serum (FBS) (Corning, Cat# 35-010-cv), 1 mM sodium pyruvate (Gibco Cat# 11360070) and MEM non-essential amino acid (Gibco, Cat# 11140050). Rad23b^+/+, Rad23b^−/− JM8.N4 mESCs and Xpc^+/+, Xpc^−/− BI6 mESCs are from Tjian lab (University of California, Berkeley). mESCs were cultured on 0.1% gelatin-coated plates in ESC media (Knockout D-MEM (Gibco, Cat# 10829018) with 15% FBS, 0.1 mM MEM non-essential amino acids (Gibco, Cat# 11140050), 2 mM GlutaMAX (Gibco, Cat# 35050061), 0.1 mM 2-mercaptoethanol (Sigma) and 1000 units mL⁻¹ of LIF (Tjian lab)). All cells were cultured at 37 °C with 5% CO₂.

Generation of Rad23b knockout cells

sgRNA targeting human Rad23b were designed using CRISPick (https://portals.broadinstitute.org/gppx/crispick/public) and were cloned into LentiCRISPR-V2 (Addgene #52961). The cloning and virus production were performed according to Feng Zhang lab’s protocol⁷⁶. Briefly, LentiCRISPR-sgNT and LentiCRISPR-sgRad23b lentiviruses were produced in HEK293T cells co-expressing the packaging vectors (psPAX2, Addgene #12260, and pVSVG, Addgene#8454). Virus was filtered using 0.45 μm syringe filter and transduced into A-375 cells with 8 μg/mL polybrene (Millipore Sigma Cat#TR-1003). Next day after infection, cells were treated with 2 μg/mL puromycin (Gibco, Cat# A1113803) for 7-day selection.

Animals

All animal experiments were conducted in accordance with procedures approved by the IACUC at University of California, Berkeley (AUP-2019-04-12038-1). Mice were housed in 12-hour light-dark cycle with free access to food and water. The following mice were obtained from Jackson Laboratory: Ctr1^Flox/Slc31a1tm2Djt/J (RRID:IMSR_JAX:025651); EGFP-L10a/129S4-Gt(ROSA)26Sortm9(EGFP/Rpl10a)Amc/J (RRID:IMSR_JAX:024750); Ai14/B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J (RRID:IMSR_JAX:007914). TH^cre(RRID: IMSR_EM:00254) was obtained from European Mouse Mutant Archive. Dbh^Cre/ Tg(Dbh-cre)KH212Gsat/Mmucd (RRID:IMSR_JAX:007914) was obtained from the Mutant Mouse Resource & Research Centers (MMRRC). Experiments were performed on adult animals (2–6 months) of both genders.

METHOD DETAILS

Laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS)

Mice were deeply anesthetized and transcardially perfused using 1× DPBS (Gibco, Cat# 14190144). Brain tissue was embedded in Tissue-Tek OCT compound (Sakura finetek), flash frozen in Isopentane/liquid nitrogen bath, and cut into 20 μM sections. Laser ablation was performed on an NWR213 laser with a TV2 sample chamber (ESI, Bozeman, MT) using the following parameters: spot size: 6 μm; fluence: 2.3 J cm⁻²; stage speed: 15 μm s⁻¹; firing rate: 20 Hz; He flow: 800 mL min⁻¹; pattern spacing: 6 μm. Using these parameters, the tissue was fully ablated, but the glass slide remained undamaged. The ablated material was introduced by gas flow into an iCAP-Qc ICP-MS (Thermo Fisher) and analyzed for ⁶³Cu, ⁶⁶Zn or ⁵⁷Fe content using a 0.4 s dwell time in standard acquisition mode. The resulting mass spectrometry traces and laser log files were processed in Igor Pro using the Iolite application. The trace elements data reduction scheme was used in semi-quantitative mode using ⁶³Cu, ⁶⁶Zn or ⁵⁷Fe as the reference trace and a custom matrix-matched standard to convert mass spectrometer counts to metal concentration. The matrix-matched LA-ICP-MS standards were prepared as described previously ³⁴.

Fluorescence in-situ hybridization

Fluorescence in-situ hybridization was performed using RNAscope assays according to the manufacturer’s instructions (Advanced Cell Diagnostics). The following probes were used: Mm-Dbh-C3 (400921-C3); RNAscope Probe - Mm-Atp7a (464931); and Mm-Slc31a1-C2 (300031-C2).

Surgical procedures

Adult TH^Cre; Ctr1^flox/flox or their littermate TH^Cre; Ctr1^+/+ mice were anesthetized with 1.5-2% isoflurane and placed on a stereotaxic frame. Body temperature was kept stable throughout the procedure using a heating pad. After asepsis, the skin was incised to expose the skull, and the overlying connective tissue was removed. To implant electroencephalogram (EEG) and electromyogram (EMG) recording electrodes, one stainless steel screw was inserted into the skull 1.5 mm from midline and 1.5 mm anterior to the bregma, and one inserted 2.5 mm from midline and 3 mm posterior to the bregma. Two EMG electrodes were inserted into the neck musculature. A reference screw was inserted into the skull on top of the right cerebellum. Insulated leads from the EEG and EMG electrodes were soldered to a pin header, which was secured to the skull using dental cement.

Sleep recordings and scoring

Mice were habituated for at least three days before a 3-day recording period, with food and hydrogel provided in their home cage that was placed in a sound-attenuation box. EEG and EMG electrodes were connected to flexible recording cables via a mini-connector, and signals were recorded and amplified with TDT RZ5 at 1017 Hz.

For sleep scoring, spectral analysis was carried out using fast Fourier transform (FFT), and brain states were classified into wake (desynchronized EEG and high EMG activity) and sleep state (NREM: synchronized EEG with high delta power (1-4 Hz) and low EMG activity; REM: high EEG theta power (6-10 Hz) and low EMG activity, with an epoch length of 5 s. Short periods of EEG desynchronization during NREM sleep that lasted for less than 15 s were also considered NREM sleep. The classification was automatically achieved and manually checked by using a custom-written graphical user interface.

Translating ribosome affinity purification (TRAP)

mRNA purification by TRAP was performed following the protocol in³⁸. All experiments were performed on ice and all solutions were chilled. Mice brains were dissected, cut in half, and each sample moved to 10 mL of tissue lysis buffer (20 mM HEPES (pH 7.3), 150 mM KCl, 10 mM MgCl₂, EDTA-free protease inhibitors (one mini tablet per 10 mL), 0.5 mM DTT, 100 μg mL⁻¹ cycloheximide, 10 μL mL⁻¹ rRNasin and Superasin). The samples were homogenized with glass-Teflon homogenizers with 3 strokes at low speed and 10 strokes at high speed. The samples were centrifuged for 10 min in 4 °C at 2,000 g. The supernatant was then transferred to another tube and 1/9 sample volume of 10% NP40 and 1/9 sample volume of 300 mM DHPC were added. The samples were mixed and incubated on ice for 5 min and subsequently centrifuged for 10 min in 4 °C at 20,000 g. The supernatant was moved to new tubes and used for immunopurification.

Preparation of the Affinity matrix used for immunopurification occurred concurrently with the sample preparation. Streptavidin MyOne T1 Dynabeads were resuspended in the original bottle through mixing, and the appropriate amount (300 μL per sample) was collected by a magnet and washed with and resuspended in RNase free PBS. The appropriate amount (120 μL per sample) of 1 μg μL⁻¹ Biotinylated Protein L was added to the solution containing Dynabeads and incubated for 35 min at room temperature using a tube rotator. The Protein L-coated Dynabeads were collected with a magnet and washed with a 3% IgG-free and Protease-free BSA in PBS 5 times, each wash incubated for 5-8 min and beads were collected with a magnet in between. The beads were then added to (1 mL per sample) primary antibody solution in a low salt buffer (50 μg mL⁻¹ GFP antibody 19C8, 50 μg mL⁻¹ GFP antibody 19F7, 20 mM HEPES, 150 mM KCl, 10 mM MgCl₂, 1% NP-40, 0.5 mM DTT, 100 μg mL⁻¹ cycloheximide) after the last wash and was incubated for one hour at room temperature using a tube rotator. The beads were then washed with a low salt buffer 3 times and resuspended in 30 mM DHPC in low salt buffer. The sample supernatant prepared concurrently was added to the resuspended beads and incubated for 5 hours at 4 °C on a tube rotator.

After incubation completed, the beads were collected with a magnet and washed 4 times with high salt buffer (20 mM HEPES, pH 7.3, 350 mM KCl, 10 mM MgCl₂, and 1% NP-40, 0.5 mM DTT, 100 μg mL⁻¹ cycloheximide). After the last wash, beads were added to the Nanoprep lysis buffer (Agilent, Cat# 400753) with β-mercaptoethanol, vortexed, and incubated for 10 min at room temperature. The beads were collected with a magnet, and RNA samples in solution were cleaned up using the manufacturer’s instructions from the RNA Nanoprep kit.

cDNA libraries were constructed from purified RNA samples with Smartseq and sequenced using Illumina Next Generation Sequencing. Reads below a quality and length threshold were dropped and the sequences were cleaned with Trimmomatic⁷⁷. RSEM⁷⁸ was used to quantify transcripts using paired end data. Differential analysis of the data was performed using the EBSeq package⁷⁹. All subsequent data analysis was performed in R. Transcript with TPM (Transcripts Per Million) > 0.05 was used in the analysis. Transcripts with PPEE (Posterior Probably of Equal Expression) < 0.05 and PostFC (Posterior Fold Change) > ∣1.5∣ were considered differentially expressed in its comparison group.

Immunohistochemistry

Mice were deeply anesthetized and transcardially perfused using 1× DPBS followed by 4% paraformaldehyde in 1× DPBS. Brains were dissected and post-fixed in 4% paraformaldehyde for 24-48 hours and stored in 30% sucrose in PBS solution for 48 hours for cryoprotection. Brains were embedded with Tissue-Tek OCT compound (Sakura finetek) and 20 μm sections were cut using a cryostat (Leica). Immunohistochemistry staining was performed as described previously ³⁴ Anti-GFP (Abcam, Cat# ab13970) was used at 1:2,000; anti-Rad23B (Proteintech, Cat# 12121-1-AP) was used at 1:200 for frozen-section immunohistochemistry. The following secondary antibodies were used: goat anti-chick Alexa Fluor 488 (Thermo Fisher, Cat# A-11008) was used 1:250; goat anti-rabbit Alexa Fluor 647 (Thermo Fisher, Cat# A-21244) was used 1:250.

Phylogenetic analysis

The trees were built based on Ensembl methods⁸⁰. Data were exported and plotted with Matlab (Mathworks).

Transfection

For transient plasmid transfection, HEK293T cells were transfected using lipofectamine 2000 (Invitrogen, Cat# 11668027) while MEFs, A-375 and LM2 cells were transfected using lipofectamine 3000 (Invitrogen, Cat# L3000008) according to the manufacture’s protocols with cDNAs expressed from either pCDNA3.1 (Invitrogen) or pCMV6 (origene) vectors. siRNA transfections were performed at 40 nM with Lipofectamine RNAiMAX (Invitrogen, Cat# 13778075) according to the manufacture’s protocols. All siRNAs were ordered from Dharmacon/Horizon.

Metal binding assays

Profinity IMAC resin (Biorad, Cat# 1560121) was washed and loaded with metals (0.1 M of either FeSO₄, CoCl₂, NiSO₄, CuSO₄, and ZnSO₄) based on the manufacturer’s protocol. Cleared cell lysates were loaded on the column and binding for 1 hour at room temperature. Columns were washed three times with PBS and eluted with PBS containing 300 mM imidazole. Eluted proteins were subject to immunoblot analysis using indicated antibodies.

Copper binding with RAD23B variants

Cell lysates

HEK293T cells transfected with indicated plasmids encoding corresponding RAD23B variants were lysed after 24 hours transfection. The cleared cell lysates were subject to Cu charged IMAC resin as described above. The elution solution was subject to immunoblot analysis using anti-V5 antibody.

Recombinant proteins

20 μg recombinant RAD23B protein was loaded to 20 μL uncharged or Cu-charged resin. After washing and elution steps as described above, the elution solution was subject to SDS-PAGE analysis.

RAD23B protein sequence analysis

RAD23B protein sequences from indicated species were downloaded from Ensembl (www.ensembl.org) ⁸¹ and alignment was performed using Clustal Omega ⁸². The corresponding transcript IDs for protein sequence alignment were listed below:

Species	Transcript ID
Danio rerio	ENSDARP00000107646.2
Gallus gallus	ENSGALP00010020255.1
Gopherus evgoodei	ENSGEVP00005028255.1
Homo sapiens	ENSP00000350708.3
Latimeria chalumnae	ENSLACP00000004049.1
Leptobrachium leishanense	ENSLLEP00000001279.1
Mus musculus	ENSMUSP00000030134.9
Pelodiscus sinensis	ENSPSIP00000016820.1
Podarcis muralis	ENSPMRP00000035037.1
Pseudonaja textilis	ENSPTXP00000026776.1
Taeniopygia guttata	ENSTGUP00000006568.1
Xenopus tropicalis	ENSXETT00000012666.5

Protein expression and purification

Full-length RAD23B and its variants were transformed into BL21 DE3 rossetta cells for overexpression. The cells were grown at 37 °C in LB media containing 50 mg mL⁻¹ kanamycin until an OD₆₀₀ of 0.6 was reached. The cultures containing RAD23B were induced by adding 1 mM isopropyl-1-thio-β-D-galactopyranoside (IPTG) at 37 °C and growth was continued for 3-4 hours. Cells were harvested by centrifugation and purified using three different purification steps, (a) affinity chromatography, (b) anion exchange, and finally (c) size exclusion chromatography. In the first step the protein was passed through a 5 mL HisTrap FF column (Cytiva) at 4 °C according to the manufacturer’s instructions. The proteins were washed with 40 mM imidazole and eluted with a linear imidazole gradient from 40 to 500 mM in buffer containing 20 mM sodium phosphate, pH 7.4, 300 mM NaCl. The protein containing fractions were concentrated, the salt concentration was reduced to 50 mM NaCl, and the protein was immediately loaded on a HiTrap Q HP column (Cytiva) equilibrated in 20 mM Tris/HCl, pH 8, 50 mM NaCl. The protein was eluted with a linear NaCl gradient (50-500 mM). After concentrating the protein using Amicon Ultra 30 kDa MWKO centrifugal filter (Millipore, USA), His6 tag was removed by incubation with TEV protease (protein:TEV= 20:1 (w/w)) at 4 °C overnight. The flow-through was concentrated and loaded on a HiLoad 16/600 Superdex 200pg size exclusion column (Cytiva) equilibrated in 20 mM sodium phosphate, pH 7.4, 250 mM NaCl. The corresponding protein fractions were pooled, concentrated by using Amicon Ultra 30 kDa MWKO centrifugal filter (Millipore, USA) and then flash-frozen into aliquots in liquid nitrogen, and stored at −80 °C.

¹³C-, ¹⁵N-labeled RAD23B_272-342 expression and purification

RAD23B_272-342 and RAD23B_272-342-NQ were transformed into BL21 DE3 rossetta cells for overexpression. 50 mL overnight culture in LB was centrifuged at 5,000 g for 8 min, and washed once with M9 media and re-suspended with M9 media supplemented with 1g L^{−1 15}N-NH₄Cl (Cambridge Isotope Laboratories, Cat# NLM-467) and 2g L^{−1 13}C₆-glucose (Cambridge Isotope Laboratories, Cat# CLM-1396). Protein was expressed at 25 °C for 12 hours by adding 1 mM IPTG when an OD₆₀₀ of 0.6-0.8 was reached. Cells were harvested by centrifugation and purified using affinity chromatography and size exclusion chromatography. For affinity chromatography, the protein was passed through a 5 mL HisTrap FF column (Cytiva) at 4 °C according to the manufacturer’s instructions. The proteins were eluted with a linear imidazole gradient from 0 to 500 mM in buffer containing 20 mM sodium phosphate, pH 7.4, 300 mM NaCl. The protein containing fractions were concentrated and dialysis overnight at 4 °C in 20 mM Tris/HCl, pH 7.5, 100 mM NaCl, 1 mM β-mercaptoethanol with TEV protease (protein:TEV= 20:1 (w/w)) to remove the His6 tag. Next day, the dialysis solution was loaded to 10 mL HisPure resin (Thermo, Cat# 88221) equilibrated with 40 mM imidazole containing buffer (20 mM sodium phosphate, pH 7.4, 300 mM NaCl). The flow-through was concentrated and loaded on a HiLoad 16/600 Superdex 75pg size exclusion column (Cytiva) equilibrated in 20 mM sodium phosphate, pH 7.4, 160 mM NaCl. The corresponding protein fractions were pooled, concentrated by using Amicon Ultra 3 kDa MWKO centrifugal filter (Millipore, USA) and then flash-frozen into aliquots in liquid nitrogen, and stored at −80 °C. The protein concentration was determined by protein BCA kit (Thermo, Cat# 23227).

Determination of solution structures by nuclear magnetic resonance (NMR) spectroscopy

Uniformly ¹³C-, ¹⁵N- enriched protein samples were prepared and supplemented with 5% (v/v) D2O. All NMR experiments were performed at 300 K on Bruker NEO 600 and 800 MHz NMR spectrometers with Z-gradient 5mm TCI Cryoprobes. Triple resonance HNCA, CBCANH, CBCA(CO)NH spectra were recorded and analyzed to obtain backbone assignments. HBHA(CO)NH, CC(CO)NH and HCCH-TOCSY spectra were used for the side-chain chemical shift assignment. A simultaneous 120 ms ¹³C/¹⁵N-edited NOESY spectrum was recorded and used to extract distance restraints in structural calculation/validation. Spectra were acquired and processed with Bruker TopSpin software (v.4.0 & v.4.3). Proton chemical shifts were referenced externally to a DSS standard while ¹³C and ¹⁵N chemical shifts were referenced indirectly to this value⁸³. Temperature calibration was according to the method of ⁸⁴.

A deep neural network based peak picking of the spectra, resonance assignment using the FLYA algorithm⁸⁵ and automated NMR structure calculation was performed using CYANA⁸⁶ within the automatic ARTINA NMR structure determination pipeline hosted on the NMRtist server which implements the ARTINA approach⁸⁷. Structures were further refined in the presence of explicit water molecules using the Xplor-NIH (3.7.0.1) wrefine scripts^88,89. In the case of the Ag(I)-bound structure determination, structures were first calculated in CYANA (3.98.15) and refined using a modified Cu-His₂ geometry and explicit water molecules applying modified Xplor-NIH (3.7.0.1) wrefine scripts^88,89. The coordination bond between the H274 and H323 Nε₂ atoms and Ag(I) was set to 2.3 Å.

For the chemical shift perturbation (CSP) assay, 1.5 equivalent metal ions (Cu(I) stock solution: 100 mM CuSO₄, 250mM TCEP in 20 mM sodium phosphate buffer, 160 mM NaCl; Ag(I) stock solution: 100 mM AgNO₃) were added for analysis. The combined CSP values were calculated using the following equation⁹⁰: Δδ = ∣Δδ(¹H)∣ + 0.2∣Δδ(¹⁵N)∣.

Competition assay with the Cu(II)-responsive fluorescent probe DP2

Fluorescence spectra were recorded using a Photon Technology International Quanta Master 4 L-format scan spectrofluorometer equipped with an LPS-220B 75-W xenon lamp and power supply, A-1010B lamp housing with integrated igniter, switchable 814 photocounting/analog photomultiplier detection unit, and MD5020 motor driver. The apparent affinity constants for Cu(II) of RAD23B-WT was determined by competition titrations with fluorescent probe DP2 (synthesized by Biomatik) in HEPES buffer (20 mM HEPES, 150 mM NaCl, pH 7.4). 2 μM of DP2 was incubated with 2 μM CuCl₂ for 30 min, and aliquots of DP2-Cu were incubated with different equivalent RAD23B protein for 1 hour before the fluorescence emission spectra were recorded. Dissociation constant of 10^−10.1 was used for Cu(II)-DP2 under this pH condition ⁴².

Cu(I) binding assays

UV-Vis absorption spectra were recorded in a 1-cm quartz cuvette (Starna) using an Agilent Technologies Cary 60 UV-Vis. The apparent affinity constant for Cu(I) of RAD23B-WT was determined by competition titrations with colorimetric ligands BCA in HEPES buffer (20 mM HEPES, 150 mM NaCl, pH 7.4). Aliquots of 10 mM BCA stock were titrated into solutions containing 30 μM RAD23B protein and 15 μM CuCl₂ to establish the exchange equilibrium. The formation constant of 1.6 × 10¹⁷ was used for Cu(BCA)₂^43,44. The concentration of Cu(BCA)₂ complex was calculated from its absorbance at 562 nm ( ε = 7900 M⁻¹ cm⁻¹)⁴⁴.

Continuous Wave Electron Paramagnetic Resonance (CW EPR) experiments

100 μL 150 μM recombinant protein was incubated with 120 μM CuSO₄ at room temperature for 30 min and passed through Zeba spin desalting columns (7K MWCO, Thermo Scientific, Cat#89882). 20% glycerol was added before transferred into X-band EPR sample tube and frozen in liquid nitrogen.

X-band (9.37 GHz) CW EPR spectra were recorded using a Bruker (Billerica, MA) EleXsys E500 spectrometer equipped with a super high Q resonator (ER4122SHQE). Cryogenic temperatures were achieved and controlled using an ESR900 liquid helium cryostat in conjunction with a temperature controller (Oxford Instruments ITC503) and gas flow controller. CW EPR spectra were recorded at 15 K by using 0.06325 mW power under slow passage conditions. The spectrometer settings were as follows: conversion time of 40.00 ms, modulation amplitude of 0.8 mT and modulation frequency of 100 kHz. Simulations of CW spectra were performed using EasySpin 6.0.0 and Matlab (The Mathworks Inc., Natick, MA)⁹¹.

Copper-RAD23B stoichiometry measurements by inductively coupled plasma mass spectrometry (ICP-MS)

A sample of 50 μL of 10 μM protein was incubated with 15 μM Cu(I) or Cu(II) at room temperature for 30 min and excess metal was removed by desalting using Zeba spin desalting columns (7K MWCO, Thermo Scientific, Cat# 89882). Protein was digested with 50 μL 4% trace-metals-grade nitric acid (Aristar Ultra, Cat# 87003-228) at room temperature. Digested samples were diluted in 2% nitric acid with 20 p.p.b. gallium solution as an internal control and analyzed with iCAP-Qc ICP-MS (Thermo Fisher).

Co-immunoprecipitation assays

Anti-Flag IP

Cells were lysed with Flag IP lysis buffer (50 mM Tris-HCl, 150 mM NaCl, pH 7.4, 1% TRITON X-100) on ice and cleared by centrifugation at 16,000 g for 10 min. Anti-Flag M2 magnetic beads were washed three times with TBS buffer (50 mM Tris-HCl, 150 mM NaCl, pH 7.4) and then incubated for 1 hour with soluble lysates. Beads were washed four times with TBS buffer and eluted with 3× Flag peptide (200 ng μL⁻¹).

Anti-V5 IP

Cells were lysed with RIPA buffer on ice and cleared by centrifugation at 16,000 g for 10 min. Anti-V5 agarose affinity gel were washed five times with PBS buffer before incubation with lysates at room temperature for 1.5 hours with agitation. After washing twice with RIPA buffer and twice with PBS, proteins were eluted from the agarose gel by adding 2× LDS sample buffer (Invitrogen Cat# NP0007) (diluted with PBS) and boiled at 95 °C for 5 min.

Anti-HA IP

Cells were lysed with RIPA buffer on ice and cleared by centrifugation at 16,000 g for 10 min. Anti-HA magnetic beads (Pierce, Cat# 88836) were washed with TBS buffer containing 0.05% Tween-20 according to the manufacturer’s protocol before incubation with lysates at room temperature for 1 hour with agitation. After washing with TBS buffer containing 0.05% Tween-for three times and once with ultrapure water, proteins were eluted from the beads by adding 2× LDS sample buffer (Invitrogen Cat# NP0007) (diluted with PBS) and boiled at 95 °C for 5 min.

Histidine labeling assays using the O5C-TPAC reagent

Histidine labeling using the O5C-TPAC (0.5 mM) reagent was conducted on freshly prepared cell lysates (20 mM HEPES, 150 mM NaCl, pH 8.0, 1% TRITON X-100) at room temperature for 1 hour. CuAAC reactions were performed using 200 mM Biotin-PEG3-Azide probe (Click Chemistry Tools, Cat# AZ104-25), followed by anti-V5 immunoprecipitation. After elution, the biotinylated level and eluted protein amount were analyzed by immunoblotting using streptavidin-HRP conjugated antibody and anti-V5 antibody respectively.

Cysteine labeling assays using the Iodoacetyl-biotin (IA-biotin) reagent

EZ-Link Iodoacetyl-PEG2-Biotin (Thermo Scientific, Cat# 21334) (100 μM) was added into cell lysis buffer (50 mM Tris-HCl, 150 mM NaCl, pH 7.4, 1% TRITON X-100) to make modified lysis buffer. Cells were lysed with this modified lysis buffer and incubated at room temperature for 30 min. After centrifugation, the cleared cell lysates were subject to anti-Flag immunoprecipitation. After elution, the biotinylated level and eluted protein amount were analyzed by immunoblotting using streptavidin-HRP conjugated antibody and anti-flag M2 antibody respectively.

SOD1 activity assay

SOD1 activity was measured by superoxide dismutase assay kit (Cayman Chemical, Cat# 706002) according to the manufacturer’s protocol. Briefly, cells were lifted by cell lifter (Corning, Cat# 3008) and collected by centrifugation at 300 g for 4 min at 4 °C. Cells were homogenized in cold lysis buffer (20 mM HEPES buffer, pH 7.2, containing 1 mM EGTA, 210 mM mannitol and 70 mM sucrose) and pellets were removed by centrifugation at 1,500 g for 5 min at 4 °C. The supernatant was further centrifuged at 10,000 g for 5 min at 4 °C to obtain the cytosolic SOD1 in the resulting supernatant. The SOD1 activity was normalized to protein amount which is determined by protein BCA kit (Thermo, Cat# 23227).

Size exclusion chromatography analysis

Oligomerization of COX17-Flag

Cells were lysed with native lysis buffer (Abcam Cat# ab156035) on ice and cleared by centrifugation at 16, 000 g for 10 min. Then ~ 1 mg of cellular protein was separated on a Enrich SEC 650 10 × 300 column (Bio-Rad) in PBS buffer (50 mM sodium phosphate, 150 mM NaCl, pH 7.2). Fractions were concentrated using Amicon Ultra-0.5 mL Centrifugal Filters (Millipore, Cat# UFC501024) and subject to immunoblotting analysis. Enrich SEC 650 10 × 300 column was calibrated using gel filtration standard (Bio-Rad, Cat# 1511901).

Complex of recombinant RAD23B and ATOX1 proteins

A 100 μL solution containing 20 μM apo ATOX1 protein with 20 μM Cu(I), apo-RAD23B protein, or Cu-RAD23B was incubated at room temperature for 30 min before separation on a Enrich SEC 650 10 × 300 column (Bio-Rad) in PBS buffer (50 mM sodium phosphate, 150 mM NaCl, pH 7.2). Fractions corresponding to the peak were pooled and concentrated using Amicon Ultra-0.5 mL Centrifugal Filters (Millipore, Cat# UFC501024), and subject to immunoblotting analysis. Enrich SEC 650 10 × 300 column was calibrated using gel filtration standard (Bio-Rad, Cat# 1511901).

Cytochrome c oxidase (COX) activity assays

COX activity was measured using Cytochrome c Oxidase Assay kit (Abcam, Cat# ab239711). LM2 cells with indicated treatment were collected and the mitochondrial fraction were extracted using the Cell Fractionation kit-Standard (Abcam, Cat# ab109719). After protein quantification by BCA, 5 μg of total purified mitochondrial protein was used for COX activity measurement. The activity of the enzyme was determined by measuring the oxidation of reduced cytochrome c as an absorbance decrease at 550 nm. The rate of the enzyme reaction was calculated in the linear range.

Seahorse assays

Oxygen consumption rate (OCR) measurements were performed using the Seahorse XF Cell Mito Stress Test kit (Agilent, Cat# 103015-100) on a Seahorse XFe24 Analyzer according to the manufacturer’s protocol. Seahorse Wave Desktop Software (v2.6) was used for measurements. For LM2 and MEF cells, 1 μM oligomycin, 1 μM carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), 0.5 μM Antimycin A/Rotenone were used. For mESCs cells, 1 μM oligomycin, 1.25 μM carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), 1 μM Antimycin A/Rotenone were used. After the analysis, cells were fixed in 4% paraformaldehyde (Electron Microscopy Sciences, Cat# 15710-S) in PBS for 10 min, washed with PBS, and stained overnight with 0.5% methylene blue (Sigma-Aldrich, Cat# M9140). The next day, cells were washed with ddH₂O for three times, de-stained with 4% acetic acid in 40% methanol, and the absorbance was measured at 668 nm on Microplate Reader (BioTek). OCR values were normalized to total cellular DNA content. Basal respiration, ATP-linked respiration, maximal respiration, and non-mitochondrial respiration were then calculated from the OCR Mitostress test profile.

Copper toxicity assays

3,000 A-375 cells were seeded in clear 96-well plates (Corning, Cat# 3595) and incubated with indicated siRNAs 48 hours before adding CuSO₄. After overnight incubation with CuSO₄, cell number was measured by using CCK8 kit (Dojindo, Cat# CK04-11) and read at 450 nm on Microplate Reader (BioTek). IC₅₀ was calculated by Prism Graphpad.

Cell proliferation assays

3,000 A-375 cells incubated with indicated siRNAs were seeded in clear 96-well plates (Corning, Cat# 3595) and the viability was measured every 24 hours by using CCK8 kit (Dojindo, Cat# CK04-11) and the readings at 450 nm on Microplate Reader (BioTek) were used. Readings for Day0 were recorded 6 hours after cells were seeded.

Treatments with UV irradiation

Cells were treated with UV (254 nm) irradiation and recovered in complete growth media with or without 100 μM TTM overnight. Cells were then collected for comet assay analysis or stained with 0.5% methylene blue for cell number measurement.

Comet assay

After UV treatment, cells were washed with cold PBS and processed using the comet assay kit (Abcam, cat# ab238544) according to the manufacture’s instructions. Briefly, cells were resuspended in PBS and deposited on the slides with 1% low-melting agarose. Cells were lysed in the lysis buffer for 45 minutes, and subject to alkaline treatment for 30 minutes. Electrophoresis was performed under alkaline conditions for 25 minutes at 300 mA. Stained slides were imaged by laser scanning confocal microscopy (Zeiss) using 488 nm channel.

Labile copper analysis by confocal microscopy imaging

Cells were seeded at 20-40% confluency in 8-well chamber slide (Nunc Lab-Tek) with growth medium and were allowed to grow to 70% confluency before performing cell imaging. For CD649.2 ⁹² imaging, cells were washed once with HBSS (+Ca, Mg) and incubated with 1 μM CD649.2 in HBSS (+Ca, Mg) with 2% DMSO for 1 hour. Cells were washed once with HBSS (+Ca, Mg) and imaged. For CF4 and Ctrl-CF4-S2 ³⁴ imaging, cells were washed once with HBSS (+Ca, Mg) and incubated with 2 μM CF4 or Ctrl-CF4-S2 in HBSS (+Ca, Mg) with 1% DMSO for 30 minutes. Cells were washed once with HBSS (+Ca, Mg) and imaged.

Confocal fluorescence imaging was performed with a Zeiss laser scanning microscope LSM880 with a 20x dry objective lens using Zen 2015 software (Carl Zeiss, Zen 2.3 black). CD649.2 was excited using a 633 nm diode laser and emission was collected using a META detector between 650 to 750 nm. CF4 and Ctrl-CF4-S2 were excited using a 488 nm diode laser and emission was collected using a META detector between 550 to 650 nm. Average cellular fluorescence intensity for fluorescent sensors was determined using Fiji (National Institutes of Health). The area of stained cells was selected by setting the appropriate threshold with a Gaussian blur filter (sigma = 1). The “Create Mask” function followed by the “Create Selection” function were then used to create a selection from this threshold. Using this selection, the mean fluorescence was measured. For each biological replicate, three images in different fields of view were analyzed and the values were combined for statistical analysis. Statistical comparison of fluorescence intensities in different conditions was performed using an unpaired t-test (assume Gaussian distribution, no Welch’s correction) in Prism 10 (GraphPad). For representative confocal images, the maximum and minimum brightness was kept consistent for all images.

In silico protein structure prediction

To predict the XPCB structure, we leveraged ColabFold, a Google Colab-based implementation of AlphaFold2^40,41. Sequence used for prediction is “SGGHPLEFLRNQPQFQQMRQIIQQNPSLLPALLQQIGRENPQLLQQISQHQEHFIQMLNEPVQEAGGQGGGG” and the calculation was done with default settings. Runs were supported by Google Cloud virtual machines running NVI DIA Tesla A100 GPUs. The resulting structures were visualized and recolored with Pymol.

Synthesis of thiophosphorodichloridate (TPAC) reagents

General synthetic methods

Unless otherwise noted, all commercial reagents were used without further purification. All reactions utilizing air- or moisture-sensitive reagents were performed in dried glassware under an atmosphere of dry N₂. 1-Pentyne-5-ol was purchased from Oakwood Chemicals (Estill, SC). Fmoc-His-OH was purchased from Ark Pharm (Arlington Heights IL). Bombesin was purchased from Alfa Aesar (Tewksbury, MA). 3-Azido-1-propanol was synthesized according to published procedure⁹³. All other reagents were purchased from Sigma-Aldrich. ¹H NMR, ¹³C NMR and ³¹P NMR spectra were collected in CDCl₃, or d₆-acetone (Cambridge Isotope Laboratories, Cambridge MA) at 25 °C on AVB-400 or AVQ-400 spectrometers at the College of Chemistry NMR Facility at the University of California, Berkeley. All chemical shifts in ¹H NMR and ¹³C NMR are reported in the standard δ notation of p.p.m. relative to residual solvent peak (CDCl₃ δH = 7.26, δC = 77.16; d₆-acetone: δH = 2.05, δC = 29.84), and for ³¹P NMR 85% phosphoric acid in sealed capillary tube is used as internal standard (δP = 0.00). Splitting patterns are indicated as follows: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; br, broad. Preparative HPLC was performed on 1260 Infinity LC (Agilent, Santa Clara CA) equipped with a Prep-C18 column (Agilent, 30 × 250 mm, 10 μm). Low resolution mass spectral analysis (ESI-MS) was carried out with LC/MS using 1220 Infinity LC (Agilent, Santa Clara CA) coupled with Expression-L Compact Mass Spectrometer (Advion, Ithaca NY).

General procedure for thiophosphorodichloridate synthesis

Alcohol starting material was dissolved in dry THF under N₂ and cooled in an ice/water bath. To this solution was added dropwise nBuLi (2.5 M solution in hexanes, 1.1 equiv.) solution and stirred for 10 min. In a separate flask PSCl₃ (1.2 equiv.) was dissolved in dry THF under N₂ and cooled in a dry-ice/acetone bath. The lithium salt suspension in THF was then added dropwise. The mixture was warmed up to room temperature and stirred for 1 hour. THF was evaporated and the residue was dissolved in CH₂Cl₂, filtered to remove LiCl and purified by column chromatography (1:100 to 1:50 ethyl acetate/hexanes) to give the product as a clear to light yellow oil.

But-3-yn-1-yl thiophosphorodichloridate (1, 4C-TPAC)

Following the general procedure for thiophosphorodichloridate synthesis, but-3-yn-1-ol (0.62 mL, 8.2 mmol) was reacted with nBuLi (3.6 mL of 2.5 M in hexanes, 9.0 mmol) and PSCl₃ (1.0 mL, 9.8 mmol) to provide compound 1 as a clear oil (1.2 g, 71%). ¹H NMR (400 MHz, CDCl₃) δ 4.41 (dt, J = 12.0, 6.9 Hz, 2H), 2.72 (td, J = 6.9, 2.6 Hz, 2H), 2.09 (t, J = 2.6 Hz, 1H). ¹³C NMR (101 MHz, CDCl₃) δ 78.18, 71.39, 69.01 (d, J = 10.1 Hz), 20.23 (d, J = 10.8 Hz). ³¹P NMR (162 MHz, CDCl₃) δ 58.72.

(±)-Pent-4-yn-2-yl thiophosphorodichloridate (2, 4+1C-TPAC)

Following the general procedure for thiophosphorodichloridate synthesis, (±)-buta-3-yn-1-ol (0.77 mL, 8.2 mmol) was reacted with nBuLi (3.6 mL of 2.5 M in hexanes, 9.0 mmol) and PSCl₃ (1.0 mL, 9.8 mmol) to provide compound 2 as a light-yellow oil (0.7 g, 39%). ¹H NMR (400 MHz, CDCl₃) δ 5.09 – 4.98 (m, 1H), 2.73 – 2.57 (m, 2H), 2.10 (t, J = 2.7 Hz, 1H), 1.57 (d, J = 6.3 Hz, 3H). ¹³C NMR (101 MHz, CDCl₃) δ 79.29 (d, J = 10.0 Hz), 78.14, 72.00, 27.01 (d, J = 7.0 Hz), 20.47 (d, J = 4.6 Hz). ³¹P NMR (162 MHz, CDCl₃) δ 58.28.

Pent-4-yn-1-yl thiophosphorodichloridate (3, 5C-TPAC)

Following the general procedure for thiophosphorodichloridate synthesis, pent-4-yn-1-ol (0.35 mL, 4.1 mmol) was reacted with nBuLi (1.8 mL of 2.5 M in hexanes, 4.5 mmol) and PSCl₃ (0.50 mL, 4.9 mmol) to provide compound 3 as a clear oil (0.27 g, 27%). ¹H NMR (400 MHz, CDCl₃) δ 4.42 (td, J = 11.0, 6.0 Hz, 2H), 2.34 (td, J = 6.8, 2.6 Hz, 2H), 2.03 – 1.90 (m, 3H). ¹³C NMR (101 MHz, CDCl₃) δ 81.88, 70.50 (d, J = 10.5 Hz), 69.92, 28.27 (d, J = 10.0 Hz), 14.69 (s). ³¹P NMR (162 MHz, CDCl₃) δ 58.25.

3-Propargyloxy-1-propanol (4)

This compound was synthesized according to published procedure ⁹⁴.

3-Propargyloxy-1-propanyl thiophosphorodichloridate (5, O6C-TPAC)

Following the general procedure for thiophosphorodichloridate synthesis, 3-propargyloxy-1-propanol (4, 504 mg, 4.4 mmol) was reacted with nBuLi (1.94 mL of 2.5 M in hexanes, 4.9 mmol) and PSCl₃ (0.90 mL, 8.8 mmol) to provide compound 5 as a clear oil (0.63 g, 58%). ¹H NMR (400 MHz, CDCl₃) δ 4.45 (dt, J = 11.1, 6.2 Hz, 2H), 4.15 (d, J = 2.4 Hz, 2H), 3.65 (t, J = 5.9 Hz, 2H), 2.44 (t, J = 2.4 Hz, 1H), 2.11 – 2.04 (m, 2H). ¹³C NMR (101 MHz, CDCl₃) δ 79.52, 74.79, 69.40 (d, J = 10.5 Hz), 65.26, 58.41, 29.87 (d, J = 9.9 Hz). ³¹P NMR (162 MHz, CDCl₃) δ 58.67

Propargyl tosylate (6)

This compound was synthesized according to published procedure for the preparation of ¹³C₃-propargyl tosylate⁹⁵.

2-Propargyloxyethanol (7)

Ethylene glycol (373 μL, 6.7 mmol) was dissolved in dry THF (10 mL) containing NaH (357 mg 60% dispersion in mineral oil, 8.9 mmol). Tiny bubbles formed while stirring for 30 min. To this mixture was added tetrabutylammonium iodide (TBAI, 165 mg, 0.45 mmol) and propargyl tosylate (6, 937 mg, 4.46 mmol) in THF (10 mL). The mixture was stirred at refluxing temperature for 16 hours to form a light brown suspension. It was then diluted in H₂O (25 mL), saturated with NaCl and extracted with 4×ethyl ether. The organic phase was dried (Na₂SO₄), concentrated and purified by silica column chromatography (1:3 ethyl acetate/hexanes) to give compound 7 as a yellow oil (90 mg, 20%). ¹H NMR (400 MHz, d₆-acetone) δ 4.18 (d, J = 2.4 Hz, 2H), 3.68 – 3.61 (m, 3H), 3.60 – 3.54 (m, 2H), 2.94 (t, J = 2.4 Hz, 1H). ¹³C NMR (101 MHz, d₆-acetone) δ 80.95, 75.65, 72.22, 61.77, 58.55.

2-Propargyloxyethyl thiophosphorodichloridate (8, O5C-TPAC)

Following the general procedure for thiophosphorodichloridate synthesis, 2-propargyloxyethanol (7, 600 mg, 6.0 mmol) was reacted with nBuLi (2.88 mL of 2.5 M in hexanes, 7.2 mmol) and PSCl₃ (1.22 mL, 12.0 mmol) to provide compound 8 as a yellow oil (0.86 g, 62%). ¹H NMR (400 MHz, CDCl₃) δ 4.51 – 4.44 (m, 2H), 4.24 (d, J = 2.1 Hz, 2H), 3.88 – 3.84 (m, 2H), 2.48 (t, J = 2.1 Hz, 1H). ¹³C NMR (101 MHz, CDCl₃) δ 78.86, 75.37, 70.37 (d, J = 10.2 Hz), 67.41 (d, J = 10.2 Hz), 58.56. ³¹P NMR (162 MHz, CDCl₃) δ 59.67.

Analysis of stability against hydrolysis for thiophosphorodichloridate-labeled histidine adducts

Copper-catalyzed azide-alkyne cycloaddition (CuAAC) reactions between thiosphorodichloridate-labeled Fmoc-His-OH and 3-azido-1-propanol were carried out under the following condition: 100 μM labeled Fmoc-His-OH, 200 μM 3-azido-1-propanol, 500 μM THPTA, 100 μM CuSO₄ and 5 mM sodium ascorbate (all added from their 100× stock solution) in 25 mM HEPES, pH 7.5. The mixture was incubated at 37 °C for 1 hour. After filtration, 30 μL of the solution was injected for immediate analysis by a LC/MS using 1220 Infinity LC (Agilent, Santa Clara CA) coupled with a SB-C18 Zorbax rapid resolution cartridge (Agilent) and an Expression-L Compact Mass Spectrometer (Advion, Ithaca NY). Solvent A was water + 0.05% formic acid and solvent B was methanol + 0.05% formic acid. The linear gradient employed was 5-100% B in 8 min and 100% B for 4 min, with mass spectrometer connected 2 min after injection.

Stability of thiophosphorodichloridate-histidine adducts in HeLa cell lysate was tested by adding 100 μM labeled Fmoc-His-OH into 1 mg mL⁻¹ HeLa cell lysate in 50 mM HEPES, pH 7.5. The solution was incubated in a 37 °C water bath and after indicated time, 10 μL of the solution was added to 30 μL of cold methanol in a PCR tube. Proteins were removed by cooling at −80°C for at least 1 hour followed by centrifugation. All supernatant was injected for analysis by HPLC using an Agilent 1100 series LC system, with the same column and mobile phase gradient as above.

QUANTIFICATION AND STATISTICAL ANALYSIS

Biochemical experiments in vitro were routinely repeated at least three times. The indicated “n” in figure legends represents biological replicates. Statistical tests are defined in the figure legends. No statistical method was used to predetermine sample size. No animals or samples were excluded from any analysis. Animals were randomly assigned groups for in vivo studies; no formal randomization method was applied when assigning animals for treatment. Western blotting results were analyzed and quantified by ImageJ 1.52q. Statistical analyses were performed using GraphPad Prism 8.

Supplementary Material

2

Table S1. Differentially expressed genes in male and female locus coeruleus during active phase, related to Figure 2

Documents S1. Figures S1-S6 and Table S2 and S3

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
VINCULIN	Cell Signaling Technology	Cat# 4650; RRID:AB_10559207
TUBULIN	Cell Signaling Technology	Cat# 2148; RRID:AB_2288042
FLAG M2 tag	Sigma-Aldrich	Cat# F1804; RRID:AB_262044
FLAG M2 tag	Cell Signaling Technology	Cat# 14793; RRID:AB_2572291
Streptavidin-HRP conjugate	Jackson ImmunoResearch	Cat# 016-030-084; RRID:AB_2337238
Streptavidin-HRP	Cell Signaling Technology	Cat# 3999; RRID:AB_10830897
V5-tag	Abcam	Cat# ab206566; RRID:AB_2819156
CTR1	Proteintech	Cat# 67221-1-lg; RRID:AB_2882512
CTR1	Proteintech	Cat#27499-1-AP; RRID:AB_291812
RAD23B	Cell Signaling Technology	Cat# 13525; RRID:AB_2798247
RAD23B	Santa Cruz Biotechnology	Cat# sc-137088; RRID:AB_2300766
RAD23B	Proteintech	Cat# 12121-1-AP; RRID:AB_10641048
RAD23A	Cell Signaling Technology	Cat# 24555; RRID:AB_2750888
XPC	Sigma-Aldrich	Cat# HPA035707; RRID:AB_2674743
XPC	Bethyl Laboratories	Cat# A301-122A; RRID:AB_2288476
MYC-tag	Cell Signaling Technology	Cat# 2276; RRID:AB_331783
MYC-tag	Invitrogen	Cat# MA5-35831; RRID:AB_2849731
Ubiquitin	Cell Signaling Technology	Cat# 43124; RRID:AB_2799235
Ubiquitin	Santa Cruz Biotechnology	Cat# sc-8017; RRID:AB_628423
ACTIN	Cell Signaling Technology	Cat# 8456; RRID:AB_10998774
HA-tag	Cell Signaling Technology	Cat# 3724; RRID:AB_1549585
GAPDH	Cell Signaling Technology	Cat# 5174; RRID:AB_10622025
DMT1	Cell Signaling Technology	Cat# 15083, RRID:AB_2798699
Histone H3	Cell Signaling Technology	Cat# 4499, RRID:AB_10544537
GFP	Abcam	Cat# ab13970, RRID:AB_300798
goat anti-chick Alexa Fluor 488	Thermo Fisher	Cat# A-11008, RRID:AB_143165
goat anti-rabbit Alexa Fluor 647	Thermo Fisher	Cat# A-21244, RRID:AB_2535812
Biological samples
Mouse brain tissues	This study	N/A
Chemicals, peptides, and recombinant proteins
Anti-FLAG M2 magnetic beads	Sigma-Aldrich	Cat# M8823
3× Flag peptide	Sigma-Aldrich	Cat# F4799
Native lysis buffer	Abcam	Cat# ab156035
Paraformaldehyde	Electron Microscopy Science	Cat# 15710-S
Lipofectamine RNAiMAX transfection reagent	Invitrogen	Cat# 13778075
Lipofectamine 2000 transfection reagent	Invitrogen	Cat# 11668027
Lipofectamine 3000 transfection reagent	Invitrogen	Cat# L3000008
EZ-Link Iodoacetyl-PEG2-Biotin	Thermo Scientific	Cat# 21334
Biotin-PEG3-Azide	Click Chemistry Tools	Cat# AZ104-25
anti-V5 agarose affinity gel	Millipore/Sigma	Cat# A7345-1ML
Goat anti-V5 Tag antibody agarose immobilized	Bethyl	Cat# S190-119
methylene blue	Sigma-Aldrich	Cat# M9140
DP2	Young et al.42; Synthesized by Biomatik	N/A
profinity IMAC resin	Biorad	Cat# 1560121
Pierce anti-HA magnetic beads	Thermo Scientific	Cat# 88836
ATN-224	Cayman	Cat# 23553
Bicinchoninic acid disodium salt hydrate	Sigma-Aldrich	Cat# D8284
Tris(2-carboxyethyl)phosphine hydrochloride	Sigma-Aldrich	Cat# C4706
trace metals grade nitric acid	Aristar Ultra	Cat# 87003-228
TPAC	This study	N/A
Cu-GTSM	Cayman	Cat# 35455
Cu-ATSM	MedChemExpress	Cat# HY-139827
Critical commercial assays
Protein BCA kit	Thermo	Cat# 23227
superoxide dismutase assay kit	Cayman Chemical	Cat# 706002
Cytochrome c Oxidase Assay kit	Abcam	Cat# ab239711
Seahorse XF Cell Mito Stress Test kit	Agilent	Cat# 103015-100
CCK8 kit	Dojindo	Cat# CK04-11
Absolute RNA Nanoprep kit	Agilent	Cat# 400753
Comet assay kit	Abcam	cat# ab238544
Deposited data
Atomic model of apo RAD23B272-342-WT	This study	PDB: 9VFE
Atomic model of RAD23B272-342-NQ	This study	PDB: 9VFF
Atomic model of Ag-RAD23B272-342-WT	This study	PDB: 9VFG
NMR experimental data of apo RAD23B272-342-WT	This study	BMRB: 31131
NMR experimental data of RAD23B272-342-NQ	This study	BMRB: 31132
NMR experimental data of Ag-RAD23B272-342-WT	This study	BMRB: 31133
Experimental models: Cell lines
Human: HEK293T	ATCC	Cat# CRL-3216; RRID:CVCL_0063
Human: SH-SY5Y	ATCC	Cat# CRL-2266; RRID:CVCL_0019
Human: A-375	ATCC	Cat# CRL-1619; RRID: CVCL_0132
Mouse: MEF Ctr1+/+ and Ctr1−/− cells	Lee et al.45	Brady Lab
Human: LM2 SCR and CTR1KO cells	Ramchandani et al.46	Mittal Lab
Mouse: Rad23b+/+, Rad23b−/− JM8.N4 mESCs	Cattoglio et al.60	Tjian Lab
Mouse: Xpc+/+, Xpc−/− BI6 mESCs	Cattoglio et al.60	Tjian Lab
Human: A-375 sgNT and sgRad23b cells	This study	N/A
Experimental models: Organisms/strains
Mouse: Ctr1Flox/Slc31a1tm2Djt/J	Jackson Laboratories	RRID:IMSR_JAX:025651
EGFP-L10a/129S4-Gt(ROSA)26Sortm9(EGFP/Rpl10a)Amc/J	Jackson Laboratories	RRID:IMSR_JAX:024750
Ai14/B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J	Jackson Laboratories	RRID:IMSR_JAX:007914
THcre	European Mouse Mutant Archive	RRID:IMSR_EM:00254
DbhCre/ Tg(Dbh-cre)KH212Gsat/Mmucd	Mutant Mouse Resource & Research Centers	RRID:IMSR_JAX:007914
Oligonucleotides
ON-TARGETplus human Rad23B siRNA	Dharmacon	Cat# L-011759-00-0005
ON-TARGETplus human Slc31a1 siRNA	Dharmacon	Cat# L-007531-02-0005
ON-TARGETplus mouse Rad23B siRNA	Dharmacon	Cat# L-058837-01-0005
Nontargeting gRNA (sgNT): CTGAAAAAGGAAGGAGTTGA	Joung et al.76	N/A
gRNA targeting Rad23b (sgRad23b): AGAAGATTGAATCTGAAAAG	This study; Using CRISPick as the design tool (https://portals.broadinstitute.org/gppx/crispick/public)	N/A
Recombinant DNA
RAD23B-V5-pCDNA3.1(+)	This study	N/A
CCS-MYC-FLAG-pCMV6-entry	Origene	Cat# RC211311
ATOX1-FLAG-pCDNA3.1(+)	This study	N/A
COX17-MYC-FLAG-pCMV6-entry	Origene	Cat# RC210756
RAD23B-pNIC28-bsa4	This study	N/A
RAD23B272-342-pNIC28-bsa4	This study	N/A
Software and algorithms
ImageJ	National Institutes of Health	https://imagej.nih.gov/ij/; RRID:SCR_003070
GraphPad Prism	GraphPad	http://www.graphpad.com/; RRID:SCR_002798
ChemBioDraw 13.0	PerkinElmer	https://www.perkinelmer.com/category/chemdraw, RRID:SCR_016768
Trimmomatic	Bolger et al.77	http://www.usadellab.org/cms/index.php?page=trimmomatic, RRID:SCR_011848
RSEM	Li and Dewey78	http://deweylab.biostat.wisc.edu/rsem/; RRID:SCR_000262
EBSeq package	Leng et al.79	https://www.biostat.wisc.edu/~kendzior/EBSEQ/, RRID:SCR_003526
Other

Highlights.

• Transcriptome analysis in mouse LC identifies Rad23b as a copper binding protein

• RAD23B acquires copper binding motif in Amniota

• RAD23B metalloadaptor function improves cellular metabolism fitness

• Copper-allostery regulation of RAD23B canonical function in NER