Global profiling of N-terminal cysteine–dependent degradation mechanisms

Significance

Our study provides mechanistic insights into how the Cys-Arg/N-degron pathway regulates cellular adaptation to hypoxia. Through a systematic search for substrates of this pathway, we demonstrate that nearly half of the cysteine-commencing proteome can be regulated by this mechanism, and uncover the principles of substrate selectivity. Among the identified substrates, IP6K1 emerged as a key regulator of glucose metabolism during hypoxia, with its deficiency impairing glycolytic ATP production and metabolic adaptation, ultimately reducing cell survival. These findings not only expand our understanding of oxygen-sensing mechanisms but also highlight the therapeutic potential of targeting N-terminal cysteine–bearing proteins in hypoxia-related diseases.

Abstract

Hypoxia, a condition characterized by insufficient oxygen supply, challenges cellular homeostasis and energy production, triggering adaptive responses to promote survival under these stressful conditions. One key strategy involves enzymatic oxidation of N-terminal cysteine residues coupled with proteolysis through the Cys-Arg/N-degron pathway. Despite hundreds of human proteins possessing N-terminal cysteine, very few have been identified as substrates of this pathway, and its substrate selectivity remains unclear. Moreover, the biological role of this pathway in the cellular response to hypoxia is not well defined. Here, by systematically screening protein stability using an N-terminome library, we reveal a broad set of cysteine-initiating proteins regulated by this pathway. Mutagenesis experiments further revealed the specificity of Cys-Arg/N-degron pathway, showing a preference for hydrophobic and positively charged residues following cysteine. Additionally, we uncovered full-length substrates that are regulated by this pathway during hypoxia, including IP6K1. Loss of IP6K1 impaired glucose uptake, glycolytic ATP production, and overall mitochondrial function. Consequently, IP6K1-deficient cells exhibited disrupted metabolic adaptation under hypoxic conditions and reduced survival under stress. These findings underscore the importance of the Cys-Arg/N-degron pathway in regulating metabolic responses and highlight its potential importance in hypoxia-related disorders.

Hypoxia refers to conditions of reduced oxygen availability, which presents a major challenge to cellular function, as oxygen is essential for energy production and maintaining cellular homeostasis. In response, cells activate a range of adaptive mechanisms to adjust to oxygen levels, enabling them to manage metabolic stress, preserve function, and survive under these oxygen-deprived conditions (1). Two key regulatory mechanisms implicated in hypoxic stress response are the hypoxia-inducible factor 1 alpha (HIF-1α) pathway and the Cys-Arg/N-degron pathway of protein degradation (1–3). Among these, the HIF-1α pathway has been more extensively studied. Specifically, HIF-1α regulates the expression of genes involved in cellular adaptation to low oxygen levels, including metabolism, angiogenesis, and erythropoiesis (4).

N-degron pathways refer collectively to proteolytic pathways that selectively promote the degradation of proteins containing specific “destabilizing” N-terminal degradation motifs, known as N-degrons, to maintain cellular proteostasis and control cellular signaling (5–9). There are several types of N-degron pathways, some of which involve modification to the N-terminal residue such as the Arg/N-degron pathway, the Ac/N-degron pathway, and the fMet/N-degron pathway that involve Nt-arginylation, Nt-acetylation, and Nt-formylation, respectively (6). Among N-degron pathways that regulate the turnover of proteins bearing nonmodified N termini are the recently discovered Pro/N-degron pathway (10) and GASTC/N-degron pathway (GASTC = Gly, Ala, Ser, Thr, Cys) (11–14).

Among the various N-degron pathways, Arg/N-degron pathway has been shown to function as an oxygen sensor regulating the turnover of a subset of proteins with cysteine at the N terminus, following the initiator methionine (iMet) (hereafter refers as Cys-Arg/N-degron pathway) (15, 16). This pathway involves a series of processing and modification steps at the N terminus of protein substrates. First, methionine aminopeptidases (MetAPs) cotranslationally remove the iMet from nascent proteins when the residue following iMet is small (e.g., Gly/Ala/Val/Ser/Thr/Pro/Cys) (17). This process exposes cysteine as the new N-terminal residue. Under normoxic conditions, the N-terminal cysteine undergoes dioxygenation by 2-aminoethanethiol dioxygenase (ADO) (2), which activates it for the next step—N-terminal arginylation. This modification is catalyzed by arginyltransferase ATE1, which adds arginine to the N-terminal oxidized cysteine, transforming the protein into a Cys-Arg/N-degron substrate. The modified protein is recognized by E3 ligases containing UBR-box domains, including UBR1, UBR2, and UBR4, which facilitate its ubiquitination and subsequent degradation via the proteasome (16). This process ensures that substrates of the Cys-Arg/N-degron pathway are degraded under normoxic conditions when they are not required, but their turnover is halted under hypoxic conditions, allowing them to accumulate and possibly to play a role in the cellular response to low oxygen.

Which proteins are substrates of the Cys-Arg/N-degron pathway? In plants, three substrates have been identified: VII ETHYLENE RESPONSE FACTOR (ERFVII), a regulator of hypoxia-induced transcriptional reprogramming (18), LITTLE ZIPPER 2 (ZPR2) which modulates the activity of the class-III homeodomain-leucine zipper transcription factors to regulate shoot meristems activity (19), and VERNALIZATION 2 (VRN2), a PRC2 subunit and homolog of the animal Su(z)12 protein (20). In humans, however, the pathway remains understudied, with the only known substrates being the GTPase-activating proteins RGS4, RGS5, and RGS16 (21) and interleukin-32 (IL32) (2). RGS proteins accumulation in ATE1^-/- embryos disrupts Gα_q-mediated cardiac signaling pathways, leading to impaired cardiac function (21–23), suggesting that the Cys-Arg/N-degron pathway plays a crucial role in maintaining low levels of these proteins under normoxic conditions. However, the importance of preserving their degradation under hypoxic conditions is still unclear. A recent study has proposed that RGS5 might have a specific role in hypoxia, where it inhibits pericyte recruitment to blood vessels (24). As for IL32, it has both pro and anti-inflammatory roles and its elevated expression has been linked to increased cell proliferation and cancer progression (25). However, the biological significance of its regulation by Cys-Arg/N-degron pathways is yet to be fully understood.

Met-Cys (MC)-starting proteins comprise approximately 1% of the human protein-coding genome (279 proteins), suggesting the possibility of additional substrates regulated by the Cys-Arg/N-degron pathway. However, to date, no systematic studies have successfully identified new substrates for this pathway. Proteomic analyses following ADO depletion in cancer cells did not reveal significant changes in the proteome (26). RNA sequencing (RNA-seq) and mass spectrometry (MS) analysis conducted in human UBR1/2 knockout (KO) cells revealed significant alterations in the expression of approximately 300 genes, however this approach did not directly identify MC-initiating substrates of the Arg/N-degron pathway (27). In addition, a recent study profiling in vitro 68 human MC-commencing peptides to screen for potential ADO substrates found that while ADO exhibits selectivity for specific N-terminal cysteine–containing motifs, no novel endogenous MC-proteins were identified (28). These limitations in sensitivity and throughput have hindered the discovery of additional protein substrates for this pathway.

To address these limitations and to identify the full scope of Cys-Arg/N-degron pathway substrates, we utilized here a global protein stability (GPS)-peptidome approach, a high-throughput method designed to characterize degron motifs in human proteins (12, 29). Using this approach, we systemically analyzed the stability of MC-initiating peptides in response to ablation of ATE1 using a library of peptides that encode the N termini of all human proteins. Our results reveal that potentially 45% of all MC-starting proteins are regulated by ATE1. Furthermore, mutagenesis library screens demonstrate the specificity of the Cys-Arg/N-degron pathway, highlighting distinct amino acid preferences downstream of the cysteine residue. Through these screens, we also identified previously unrecognized full-length protein substrates of the Cys-Arg/N-degron pathway, including Primary Cilia Formation (PIFO) and Inositol Hexakisphosphate Kinase 1 (IP6K1), prompting further investigation into the role of IP6K1 in the cellular response to hypoxia. IP6K1 is an enzyme that catalyzes the conversion of inositol hexakisphosphate (IP6) into 5-diphosphoinositol pentakisphosphate (5-PP-IP5), also known as IP7. These IPs play a critical role in cellular signaling, regulating processes such as energy metabolism, cell proliferation, and programmed cell death (30, 31). Additionally, IP6K1 has been shown to regulate insulin sensitivity, promote weight loss, and improve glucose tolerance (32, 33). Although IP6K1 is known to regulate metabolism, its role in hypoxia and the mechanisms controlling its expression remain unclear. In this study, we revealed that IP6K1 is subjected to degradation via the Cys-Arg/N-degron pathway. Under hypoxic conditions IP6K1 protein levels are elevated, due to inhibition of its degradation, and reducing its expression through CRISPR/Cas9-mediated KO led to decreased proliferation under these conditions. This reduced survival can, in part, be attributed to the impaired glucose uptake and disrupted metabolic adaptation observed in IP6K1-deficient cells, offering valuable insights into the role of IP6K1 in oxygen sensing in eukaryotes.

Results

Systematic Identification of Cys-Arg/N-Degron Substrates via N-Terminome Stability Analysis in ATE1 Ablated Cells.

To investigate the role of N-terminal cysteine in promoting protein instability and to determine the substrate selectivity within Cys-Arg/N-degron pathway, we performed a GPS-peptidome screen using N-terminal peptide library in ATE1 KO cells (Fig. 1A and SI Appendix, Fig. S1A). The GPS method (34) monitors protein stability in live cells by employing lentiviral expression of human peptidome libraries fused to GFP (12, 29). The peptide-GFP fusion levels are normalized to a control protein, DsRed, which is expressed from the same transcript. The ratio of GFP to DsRed signals serves as an indicator of the stability of the GFP-fused peptides (12, 29). We recently adapted this approach by incorporating the ubiquitin-fusion technique (7) creating a GPS-peptidome system where the first 24 amino acids (hereafter “N24mer”) of human protein isoforms are fused between ubiquitin and GFP (“Ub-GPS”) (Fig. 1A) (12). Upon expression of the constructs in human cells, proteolytic cleavage of the ubiquitin moiety by endogenous deubiquitinating enzymes leads to the exposure of the peptides at the N-terminus of GFP (Fig. 1A). We introduced the library in duplicates into human embryonic kidney (HEK) 293T control or ATE1 KO cells and used fluorescence-activated cell sorting (FACS) to partition the population into four bins based on the stability of the peptide-GFP fusions. The stability of each fusion was then quantified with Illumina sequencing, with each peptide assigned a protein stability index (PSI) score ranging between 1 (maximally unstable) and 4 (maximally stable) according to the proportion of sequencing reads in each bin from control or ATE1 KO cells (Fig. 1A and SI Appendix, Table S1). Of the 279 MC-starting peptides present in the library, 245 had sufficient reads for analysis. The screen revealed that 108 out of 245 MC- peptides showed significant stabilization in ATE1 KO cells (PSI difference between the ATE1 KO and control KO (ΔPSI) ≥ 0.3, and Cohen’s D < –2; see Materials and Methods) (SI Appendix, Table S1). To determine whether the Cys-Arg/N-degron pathway has a preference beyond the N-terminal cysteine, we calculated the enrichment fold of residues at the third position (MCX-) in ATE1 substrates, comparing them to their frequency at the same position in the human proteome. The results revealed a distinctive amino acid composition in ATE1 substrates, with the most significantly enriched motifs among the stabilized MC- peptides in the ATE1 KO background were those containing hydrophobic (e.g., Phe, Val, Ile), positively charged residues (e.g., His, Arg), as well as Gln following Cys (Fig. 1B). To further explore motif enrichment and depletion related to ATE1, we calculated log₂ enrichment scores for each amino acid at the third position (X) by comparing their frequencies in the ATE1-positive versus the ATE1-negative 24mer N-terminal peptides. Similar to the original comparison against the whole library (Fig. 1B), this refined analysis revealed an enrichment of positively charged and hydrophobic residues, and a depletion of acidic residues as well as serine and threonine, further supporting the sequence preferences underlying ATE1-dependent regulation (Fig. 1C). Cross comparing with a previous N-terminome screen conducted in UBR1/2/4 triple KO cells (UBR KO) (12), revealed that approximately half of ATE1 candidate substrates also scored as UBR substrates (57 out of 108) exhibiting the same trend of N-terminal preference downstream of Cys (SI Appendix, Table S1). Notably, the median PSI of N-terminal cysteine peptides with preferred or nonpreferred ATE1 motifs was significantly lower when the motif was located at the extreme N-terminus, as opposed to internal regions (Fig. 1D). This suggests that N-terminal cysteine motifs, in general, promote protein instability regardless of their regulation by the Cys-Arg/N-degron pathway. However, among the MC- motifs, those recognized by ATE1 exhibited significantly lower PSIs (Fig. 1D), indicating stronger instability conferred by the Cys-Arg/N-degron pathway.

Fig. 1.

Identification of Cys-Arg/N-degron substrates through stability profiling of an N-terminome library in ATE1-deficient cells. (A) Schematic representation of the N-terminome GPS screen, in which the first 24 residues of all human proteins were expressed as N-terminal fusions to GFP in the Ub-GPS vector. To isolate peptide-GFP fusions based on their stability, fluorescence-activated cell sorting (FACS) was utilized to sort cells into four populations (“Bins”) based on their GFP/DsRed fluorescence ratio. Next-generation sequencing was used to identify the peptides enriched in each bin. Illustration was generated using BioRender. (B) The fold-change in amino acid frequency at position 3 (MCX-) of ATE1 substrates was compared to the overall frequency of residue “X” at position 3 in the library. Statistical significance for each N-terminal motif is indicated by Z-score (Materials and Methods). (C) log₂ enrichment scores for each motif by comparing their relative frequencies in ATE1-positive versus ATE1-negative groups. (D) Boxplots showing the distribution of PSI stability scores for all peptides in which the indicated motifs were encoded at the second position (colored boxes) versus any other internal position within the peptide (gray boxes). ATE1 favored motif is presented in blue while ATE1 disfavored motif is shown in red. *P < 0.05; ***P < 0.001; ****P < 0.0001. (E and F) Ub-GPS constructs, in which GFP was fused C-terminally to the first 24 residues of the indicated genes were expressed in control (cells transduced with AAVS1 sgRNA), ATE1 or UBR1/2/4 triple KO cells (“UBR KO”). Stability was then analyzed by flow cytometry for ATE1 preferred motifs (E) or nonpreferred motifs (F). GFP/DsRed ratio represents the stability of the indicated GFP-fusion proteins. Stabilization of the target protein in various genetic background is indicated by a sharp peak to the right side of each panel. For each gene, its name and the first N-terminal three residues are indicated.

Notably, the N-terminome screen also revealed an additional 99 non-MC starting peptides whose stability is regulated by ATE1, 31 of which were also identified in the UBR screen (SI Appendix, Table S1). The regulation of these substrates by both ATE1 and UBRs, despite the absence of an MC- motif, suggests that they may undergo cleavage by proteases such as dipeptidyl peptidases (DPPs) (35), calpains, separases, and secretases (36–38) which expose neo–N termini that are recognized by the Arg/N-degron pathway. For example, CEP290, ARID1B, and PCK1 are among DPP8/9 substrates (35) and contain MPPN- and MPPQ- N-terminal motifs. After cleavage, these substrates expose N- and Q- at the neo–N-terminus, which are then subjected to amidation and arginylation, converting them into Arg/N-degron substrates (35). Alternatively, some of these non-MC peptides may undergo internal arginylation (39, 40), although we cannot rule out potential indirect effects on the stability of this subset of substrates due to ATE1 KO.

To validate the results from the ATE1 screen, example candidate ATE1/UBR peptide substrates containing various enriched N-terminal Cys motifs were cloned individually in the GPS vector, expressed in ATE1 and UBR KO cells and stability was analyzed using flow cytometry. In all cases, peptide stability was increased in ATE1 and UBR KO cells confirming the validity of the N-terminome screen and indicating that these peptides are regulated by the Cys-Arg/N-degron pathway (Fig. 1E). In contrast, peptides containing N-terminal Cys motifs that were found to be depleted among ATE1 substrates (MCA-, MCD-, MCP-, and MCS-) (Fig. 1C) showed no change in stability in ATE1 or UBR-deficient cells (Fig. 1F).

Uncovering the Specificity of Cys-Arg/N-Degron Substrate Recognition Through Mutagenesis Libraries.

Next, we conducted mutagenesis experiments on a set of MC-initiating ATE1 substrate peptides to assess the role of MC- motif in driving substrate instability. In this set of experiments, we generated peptide libraries wherein single or triple amino acid mutations were introduced across all positions within the 24mer sequence (see Materials and Methods). The scanning mutagenesis libraries were cloned into Ub-GPS vector and the GPS screen of mutagenesis libraries was performed as described earlier. Notably, our findings revealed that mutating Cys and the adjacent 1 to 2 residues downstream resulted in a significant stabilization of the tested peptides (Fig. 2A and SI Appendix, Table S2). This suggests that the motif composed of the N-terminal Cys and the immediately adjacent residues is relatively short and critical for substrate instability. Additionally, residues beyond the N-terminal fourth position do not appear to play a role in substrate turnover.

Fig. 2.

Specificity of the Cys-Arg/N-degron pathway revealed through mutagenesis screening. (A) Scanning mutagenesis experiment of various N-terminal cysteine commencing peptides. For each of the indicated genes, data are presented for mutagenesis of single residues (Top) or three consecutive residues (Bottom). (B) Saturation mutagenesis of four representative substrate examples with N-terminal cysteine. For both (A) and (B), darker colors represent a greater degree of stabilization conferred by the mutation. Gene’s name is indicated at the Top of each panel and a universal scale of stabilization is provided.

To gain further insight into the specific amino acids and positions contributing to peptide stability, we conducted saturation mutagenesis experiments, replacing each amino acid with the remaining 19 amino acids (SI Appendix, Table S2). Our results elucidated two key observations: First, Cys at the second position is most critical for substrate instability and cannot be substituted by any other residues (besides Gly in some cases, as Gly also serves as a potent N-degron (12)); second, the presence of residues such His, Phe, Val, Ile, Leu, Lys, and Trp at the third position significantly contributed to peptide instability and can often be substituted for one another without disrupting stability (Fig. 2B). Importantly, these tolerated substitutions are enriched in peptides identified as ATE1 substrates (Fig. 1 B and C), supporting the idea that they are likely regulated by the ATE1 pathway. In contrast, substitutions that maintain peptide instability but correspond to disfavored ATE1 motifs—such as MCG and MCY—may be targeted by alternative degradation pathways.

Overall, our findings underscore the significance of N-terminal Cys and neighboring residues, particularly the residue immediately adjacent to Cys, in regulating peptide instability and susceptibility to degradation pathways.

Oxygen- and ADO-Dependent versus ADO-Independent Mechanisms of Stability Regulation.

Previously, it was demonstrated that following iMet cleavage, N-terminal cysteine residue can undergo N-terminal arginylation following chemical or enzymatic oxidation by ADO (2, 16, 21). To confirm whether the identified MC- substrates are regulated by ADO, peptide-GFP fusions with varying amino acid compositions at the third position were tested in ADO KO cells (SI Appendix, Fig. S1B). The results revealed that all ATE1-positive substrates (“ATE1 positives”), including the positive controls RGS5 and IL32, were stabilized in ADO-deficient cells, with the exception of NADSYN1 (MCC-, Fig. 3A). In contrast, MC-commencing peptides that were not identified as ATE1 substrates in the screen (“ATE1 negatives”) showed no stability change in response to ablation of ADO (Fig. 3A). In line with previous observations for established substrates like RGS5, the newly identified peptide examples also exhibited substantial stabilization under hypoxic conditions as revealed by western blots (Fig. 3B). Notably, in these assays, HIF1α serves as another positive control substrate, stabilized under oxygen-deprived conditions (Fig. 3B).

Fig. 3.

Analysis of oxygen-dependent and -independent regulation of N-terminal Cysteine substrates. (A) Stability analysis of the indicated ATE1 negative and positive N24mer–GFP fusions assessed in control and ADO KO cells by flow cytometry. (B) HEK293T cells expressing N24mer–GFP fusions were incubated in hypoxic chamber for 16 h. Protein samples were then collected and analyzed by immunoblot with antibodies to GFP to detect peptide-GFP and β-actin was used as a loading control. Stabilization of endogenous HIF1α served as a positive marker for hypoxia stress response activation. RGS5 N24mer-GFP was used as positive control for Cys-commencing peptide-GFP fusion. (C) Stability analysis of N24mer–GFP fusions starting with MCC- motif in control, ADO, and ATE1 KO cells by flow cytometry. (D) Stability analysis by flow cytometry of WT (MCX-) or mutant (MHX-) peptide-GFP fusions expressed in control or UBR KO cells. (E) The stability of the indicated peptide-GFP fusions was analyzed in control, ADO, ATE1, CDO1, and ADO/CDO1 double KO cells. (F) Protein levels were analyzed by immunoblotting using the indicated antibodies in cells expressing the specified peptide–GFP fusions in response to hypoxia. All immunoblots are representative of at least three independent experiments.

We were specifically intrigued by the fact that NADSYN1 (MCC-) did not stabilize in ADO1 KO cells. Upon testing additional GFP-fusion peptides containing the MCC- motif (ATP5S, WDR36), it became clear that although all are substrates of the Arg/N-degron pathway, with their stability regulated by ATE1 and UBR proteins, they are not subjected to regulation by ADO (Fig. 3C). Notably, the two adjacent Cys residues are the only ones shown to play a critical role in peptide stability (Fig. 2; see NADSYN1, ATP5S). To confirm that the degradation of these peptides depends on the N-terminal cysteine, we mutated Cys2 to histidine. While the wild type (WT) peptide was stabilized upon UBR ablation, the mutant peptide displayed substantial increased stability in control KO cells with no further stabilization observed in UBR KO cells (Fig. 3D). These data underscore the critical role of the N-terminal MCC- degron for these substrates instability. To explore the possibility of another thiol dioxygenase catalyzing the oxidation of N-terminal Cys in these substrates, we generated KO cells of the only other known mammalian thiol dioxygenase, CDO1 (SI Appendix, Fig. S1C). However, similarly to ADO KO, CDO1 deficiency as well as double KO of both dioxygenases ADO and CDO1 (SI Appendix, Fig. S1D) failed to induce any noticeable impact on stability of NADSYN1 (MCC-) (Fig. 3E). Altogether the data suggest that MCC- commencing peptides are not subjected to regulation by ADO or CDO1. It is noteworthy that N-terminal cysteine can undergo spontaneous oxidation to sulfinic or sulfonic acid forms in the presence of potent oxidants like hydrogen peroxide and nitric oxide (NO). While this nonenzymatic modification could theoretically be inhibited in anaerobic conditions, intriguingly, hypoxia had no noticeable effect on MCC- peptide stability, while it stabilized IP6K1, an MCV-containing peptide (Fig. 3F). This finding suggests that while ADO predominantly catalyzes cysteine oxidation to allow regulation of MC- substrates by ATE1 and Arg/N-degron pathway, alternative mechanisms may facilitate ATE1 recognition of unoxidized cysteine residues in MCC- commencing substrates. These pathways highlight additional complexity in Cys-Arg/N-degron mediated regulation.

IP6K1 and PIFO are Substrates of the Oxygen-Dependent N-Degron Pathway.

Next, we aimed to identify full-length proteins regulated by oxygen-dependent mechanisms. The MC-peptidome encompasses all classes of proteins, including extracellular, membrane-bound, organelle-specific, and cytoplasmic proteins. Among the substrates identified in the ATE1 and UBR screens, we specifically focused on cytoplasmic proteins. We selected 10 open reading frames (ORFs) encoding proteins with diverse MCX motifs, corresponding to peptides identified as ATE1 substrates. These ORFs were individually cloned into the GPS plasmid as GFP fusions, and their stability was assessed by flow cytometry in ADO, ATE1, and UBR KO cells. IL32, a known full-length substrate, served as a positive control. Of the 10 candidates, 8 were stabilized in ATE1-, UBR-, and ADO-deficient cells, confirming regulation by the Cys-Arg/N-degron pathway (Fig. 4A). Notably, all validated hits contained preferred MCX motifs recognized by ATE1, while the two proteins not stabilized (HMCES and GAS2) harbored nonpreferred motifs (MCG and MCD, respectively). Consistent with the behavior of peptides containing the MCC motif (Fig. 3C), the full-length MCC-bearing protein ATP5S did not show stabilization in ADO-deficient cells (Fig. 4A). These findings confirm that the degradation behavior of MC-containing peptides is recapitulated in full-length proteins, underscoring that the Cys-Arg/N-degron pathway governs substrate stability at both the peptide and ORF levels. Notably, this analysis reveals previously uncharacterized substrates of the pathway, expanding its known substrate repertoire.

Fig. 4.

IP6K1 And PIFO are substrates of oxygen-dependent N-degron pathway. (A) Stability analysis by flow cytometry of the indicated full-length GFP-fusion proteins in control, ADO, ATE1 and UBR KO cells. IL32-GFP was used as a positive control substrate. (B) Immunoblot of endogenous IP6K1 (Top), or PIFO-HA (Middle) and IL32-HA (Bottom) stably expressed in HEK293T following 16 h exposure to hypoxia. Vinculin serves as a loading control. (C) HEK293T and HeLa cells were analyzed by immunoblot for endogenous IP6K1 (Top) and stably expressed PIFO-GFP (Bottom). The stabilization of HIF1α serves as a positive indicator of hypoxia stress response activation. GAPDH serves as a loading control. (D) Representative western blots of Cycloheximide (chx) chase assays done to monitor the turnover rate of endogenous IP6K1 (Top) and PIFO-GFP (Bottom) in control or UBR KO cells by immunoblot (Left). TSPYL1 was used as a positive control for a short-lived protein during chx treatment. Quantification of IP6K1 and PIFO-GFP chx assays normalized to β-actin (Right). Data presented as mean. *P < 0.05; ***P < 0.001; ****P < 0.0001. (E) Immunoblot of HA–ubiquitin conjugates and full-length IP6K1 (Left) or PIFO (Right) GFP fusions in total cell lysate (TCL) or GFP immunoprecipitates (IP:GFP) using anti-HA or -GFP antibodies, respectively. All immunoblots are representative of three independent experiments.

From the validated substrate list, we selected two candidates for further investigation: IP6K1 and PIFO. To rule out potential artifacts from the GFP fusion, we used effective endogenous antibodies for IP6K1 and demonstrated increased protein levels across all KO cells by western blotting (Fig. 4B). For PIFO, which lacks available antibodies, we cloned it as a C-terminal HA fusion and generated single-integrant expressing cells via lentiviral transduction. Similar to IP6K1, PIFO-HA also showed increased protein levels in all KO cells by immunoblotting (Fig. 4B). To assess oxygen-dependent regulation of these substrates, HEK293T and HeLa cells were exposed to hypoxia and the results demonstrated that both endogenous IP6K1 and PIFO-GFP proteins accumulated in cells exposed to hypoxia (Fig. 4C) indicating regulation by hypoxia and Arg/N-degron of these substrates.

Using a Cycloheximide (chx) chase assay to assess protein turnover rates, we observed that IP6K1 undergoes slow degradation, with approximately 60% of its initial protein levels remaining after 16 h of treatment. However, in UBR KO cells the degradation of IP6K1 was attenuated (Fig. 4D). PIFO, tagged with GFP, exhibited a more rapid degradation profile, which was also reduced in UBR KO cells (Fig. 4D). Last, in vivo ubiquitination state of IP6K1 and PIFO was investigated in WT and UBR KO cells, revealing notably reduced levels of ubiquitination for both substrates in the UBR KO background (Fig. 4E). These findings confirm that IP6K1 and PIFO are substrates of Arg/N-degron pathway, regulated by oxygen-dependent modulation of their turnover rates.

IP6K1 Plays a Role in the Hypoxia Stress Response.

The next step was to investigate the biological significance of N-terminal Cys containing proteins during hypoxia. PIFO is known to play a crucial role in ciliary assembly and function, as well as in signaling pathways related to the cell cycle and stress responses (41, 42). According to the Human Protein Atlas (HPA) (43), PIFO is expressed in cells involved in the formation and function of cilia, including embryonic node, endometrial ciliated cells, and spermatids. However, as previously reported (41), ectopic expression of PIFO in HEK293T or U2OS cells, both of which naturally lack PIFO expression, resulted in the activation of target genes GLI1 and PTCH1, key components of the Hedgehog signaling pathway under normoxic condition (SI Appendix, Fig. S2 A and B). In HEK293T under hypoxic conditions, increased levels of PIFO due to inhibition of its degradation (Fig. 4C) led to a further increase in the accumulation of GLI1 and PTCH1 (SI Appendix, Fig. S2B). The ATP5S-GFP protein, which is expressed at higher levels than PIFO as indicated by flow cytometry (SI Appendix, Fig. S2C), was used as a negative control MC- motif containing protein and did not induce GLI1 or PTCH1 expression under the same settings. These findings suggest that hypoxia-induced PIFO stabilization boosts Hedgehog signaling. The effects of this activation on cellular adaptation and metabolism during hypoxia are beyond the scope of this study and require further investigation to assess the functional impact of PIFO stabilization.

IP6K1 is a versatile enzyme that impacts various cellular processes through its role in generating inositol pyrophosphates (IPs) (31). IPs, such as IP7 and IP8, contain one or more energy-rich pyrophosphate moieties with a free energy of hydrolysis comparable to that of ATP. IP7 is also involved in several physiological functions, including insulin secretion and AKT signaling (44, 45). A link between hypoxia and the regulation of IPs levels was demonstrated recently. It was found that while IP7 levels gradually decline in cultured cells under normoxic conditions, they remain stable during hypoxia (46). This observation suggests that the hypoxia-induced increase in IP6K1 protein levels plays a role in maintaining IP7 levels under low-oxygen conditions. Building on these findings—and considering both the hypoxia-dependent regulation of IP6K1 turnover and its role in cellular signaling—we investigated its potential involvement in the cellular hypoxic response.

Since IP6K1 protein levels increase under hypoxia, we examined the effects of its depletion. To this end, we generated two independent IP6K1 KO cell lines in HeLa cells (SI Appendix, Fig. S1 E and F) and assessed their proliferation under normoxic and hypoxic conditions using colony formation assay. While no significant difference in colony formation was observed under normoxia, IP6K1 KO cells showed a marked reduction in colony formation under hypoxia compared to control cells (Fig. 5 A and B). These findings suggest that IP6K1 deficiency markedly impairs cell proliferation and inhibits colony formation in hypoxic environments.

Fig. 5.

IP6K1 regulates metabolic reprogramming during hypoxia. (A and B) Colony formation assay of two independent IP6K1 KO cell lines (KO1, KO2) compared to control KO HeLa cells grown in normoxic or hypoxic conditions for 12 d. (A) Representative images and (B) Quantitative analysis of colony formation assay from (A). (C) Intracellular ATP level measured in IP6K1 KO and control KO HeLa cells cultured in normoxic or hypoxic conditions for 16 h. (D and E) Mitochondria morphology was analyzed in IP6K1 and control KO HeLa cells under normoxic or hypoxic conditions. n > 50 cells were imaged and quantified per condition. (D) Representative images of mitochondria visualized by immunostaining with COX-IV antibody. The zoom-in area represents a magnified view of the dashed box in the COX-IV image, highlighting details of the mitochondrial phenotype. DAPI (blue) marks the nucleus. Scale bar, 20 μM. (E) Quantitative analysis of mitochondrial morphology with ImageJ. (F) Intracellular ATP level in IP6K1 and control KO HeLa cells treated with 10 μM oligomycin for 16 h. (G and H) Glucose uptake (G) and mRNA level of GLUT1 normalized to β-actin (H) in control or IP6K1 KO cells cultured in normoxic versus hypoxic conditions. Data presented as mean. Ns- no significance; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

IP6K1 plays a crucial role in maintaining cellular metabolic balance and ATP production (47), presumably by regulating phosphate export through XPR1, a phosphate exporter. Inhibition of IP6K1 reduces phosphate export, leading to increased intracellular ATP levels (48). Consistent with these findings, IP6K1 KO cells exhibited elevated ATP levels under normoxic conditions (Fig. 5C). However, during hypoxia, ATP levels decreased by approximately 30%, in control cells, whereas IP6K1 KO cells experienced a more pronounced reduction of 50% (Fig. 5C).

Given the link between hypoxia and mitochondria functions (49, 50), we speculated that the effect of IP6K1 on energy homeostasis might be elicited by mitochondrial dysfunction and disruption of oxidative phosphorylation. To explore this further, we examined mitochondria morphology by confocal microscopy. Hypoxia was found to reduce the overall mitochondrial area, with mitochondria accumulating in the perinuclear region (Fig. 5 D and E). In the absence of IP6K1, there was a marked decrease in mitochondrial area, accompanied by an even greater perinuclear clustering (Fig. 5 D and E), indicating that the loss of IP6K1 contributes to mitochondrial dysfunction.

Building on these morphological changes, we next evaluated the impact of mitochondrial dysfunction on cell survival by culturing cells in galactose, a substrate that can only be metabolized by respiring mitochondria, and analyzed colony formation over 12 d. IP6K1 KO cells formed fewer colonies under these conditions, suggesting that cells lacking IP6K1 have compromised mitochondrial functionality (SI Appendix, Fig. S3A).

This mitochondrial dysfunction prompted us to further investigate how metabolic adaptation might be affected in the absence of IP6K1. Hypoxia was shown to induce a metabolic shift from mitochondrial respiration to anaerobic glycolysis (51). To assess glycolytic ATP production in the absence of IP6K1, cells were treated with oligomycin, an inhibitor of ATP synthesis via oxidative phosphorylation (OXPHOS). The results revealed that IP6K1 KO cells exhibited lower ATP levels in response to oligomycin treatment (Fig. 5F), mirroring the decrease seen during hypoxia conditions (Fig. 5C), suggesting that IP6K1 plays a role in glycolysis during hypoxia. Additionally, glucose uptake, which is typically enhanced by hypoxia, was approximately 30% lower in IP6K1 KO cells compared to control cells (Fig. 5G), indicating that IP6K1 depletion restricts glucose consumption under hypoxic conditions. Since hypoxia stimulates glucose transport through increased transcription of the GLUT1 glucose transporter gene (52), we evaluated GLUT1 expression and found it was more strongly induced in IP6K1 KO cells (Fig. 5H), likely reflecting a compensatory feedback mechanism to counteract reduced glucose uptake in these cells (Fig. 5G). The altered mitochondria function was also evident from the expression patterns of OXPHOS and glycolytic genes, which were upregulated in IP6K1 KO (SI Appendix, Fig. S3 B and C) likely due to impaired glucose uptake under hypoxia.

Taken together these findings suggest that IP6K1 is critical for maintaining mitochondrial function and facilitating metabolic adaptation during hypoxia. Thus, increasing IP6K1 protein levels during hypoxia by preventing its degradation via the Cys-Arg/N-degron pathway could enhance its function under these conditions.

Finally, we ruled out the possibility that the Arg/N-degron pathway functions primarily to limit IP6K1 abundance under normoxia and that reduced IP6K1 levels in this state confer a survival advantage. To test this, we overexpressed IP6K1-GFP and IP6K1-HA using high-titer lentiviral transduction (SI Appendix, Fig. S4A). Functionally, IP6K1 deficiency impaired cell survival under hypoxia, whereas IP6K1 overexpression shows comparable levels of survival to WT cells (SI Appendix, Fig. S4 B and C). Similarly, under mitochondrial stress (oligomycin treatment), ATP levels were maintained in IP6K1-overexpressing cells but reduced in IP6K1 KO cells (SI Appendix, Fig. S4D). These findings suggest that IP6K1 is protective under low-oxygen conditions; however, elevating its levels beyond those achieved by endogenous stabilization does not further enhance survival, consistent with the idea that hypoxia naturally protects IP6K1 from Arg/N-degron pathway-mediated degradation. Under normoxia, IP6K1 overexpression did not affect cell proliferation (SI Appendix, Fig. S4B), despite a modest but significant decrease in ATP levels (SI Appendix, Fig. S4D), suggesting that increased IP6K1 levels alone are not harmful. Together, these results support a model in which the Cys-Arg/N-degron pathway does not primarily function to restrict IP6K1 under normoxic conditions but its inhibition during hypoxic stress facilitates IP6K1 accumulation and promotes metabolic adaptation.

Discussion

Hypoxia activates signaling pathways that enable cell survival and adaptation under stress (53, 54). A recent study by Tian et al. (55), compared two oxygen-sensing regulatory systems that are linked to proteolytic mechanisms and serve as first-line responses to hypoxia: HIF1α regulation via prolyl hydroxylation by prolyl hydroxylases, and protein oxidation by ADO. While HIF1α is rapidly stabilized under hypoxia due to inhibition of its proteasomal degradation, the accumulation of its target genes is delayed, as transcriptional responses require time to take effect. In contrast, the ADO-dependent Cys-Arg/N-degron pathway regulates proteins that are not transcription factors, but instead are controlled directly through oxygen-sensitive proteolysis. This enables a faster and more direct response to hypoxia. Furthermore, unlike HIF1α-dependent regulation, protein substrates of the N-degron pathway are not regulated by feedback inhibition, enabling them to remain stable and functional during prolonged hypoxic stress (55). Here, we systematically searched for additional proteins regulated at their turnover rate in response to hypoxia as part of the Cys-Arg/N-degron pathway.

Proteins starting with an MC- sequence are underrepresented in both plant and human genomes (12, 56), likely due to the robustness of Cys-Arg/N-degron pathway toward these substrates and evolutionary pressure against these sequences, as also seen for other terminal degrons (12, 29). Our scanning mutagenesis experiment demonstrated that substrate selectivity in Cys-Arg/N-degron pathway extends beyond N-terminal Cys and that there is a pronounced preference for hydrophobic and positively charged residues immediately following the N-terminal Cys. These ATE1-dependent motifs closely resemble the motifs recognized by ADO in vitro (28); peptides exhibiting the highest ADO activity typically contained a basic (Lys, Arg, His) or aromatic (Phe, Trp) residue immediately following the N-terminal Cys. In contrast, ADO showed low activity toward sequences with acidic residues in this position. In addition, it has been shown that ADO and the N-terminal acetyltransferase NatA have distinct preferences for N-terminal Cys substrates, protecting against opposing modifications in vitro (28). NatA displayed the highest activity toward peptides containing acidic (Asp, Glu) or polar (Ser, Thr) residues following the N-terminal Cys, and minimal activity toward those with basic or aromatic residues. Our findings align well with these trends: substrates stabilized upon ATE1 loss are enriched for motifs that match ADO-preferred sequences, such as MCH, MCK, MCR, MCF, and MCW. Conversely, motifs such as MCD, MCE, MCS, and MCT—which align with NatA preference—are depleted among ATE1-regulated substrates. Finally, some enriched motifs in our dataset—such as MCV, MCI, and MCQ—were not examined by Heathcote et al., and therefore their susceptibility to ADO remains to be determined. These observations highlight the complexity of protein stability regulation, which involves a balance of stabilizing and destabilizing posttranslational modifications.

Intriguingly, our data reveal that ATE1 can regulate the stability of a specific class of N-terminal Cys-containing proteins—those bearing the MCC motif—independently of ADO. Notably, MCC is enriched among ATE1 substrates but was not identified as a preferred ADO target in the Heathcote et al. study (28). Although N-terminal cysteine can be oxidized directly by NO (57) evidence further suggests that the MCC degron is not regulated by hypoxia. Whether the presence of two vicinal cysteine residues can increase the likelihood of oxidation by NO or whether ATE1 might recognize unmodified cysteine residues remains an open question.

Our study extends beyond revealing the substrate specificity of Cys-Arg/N-degron pathway to explore additional substrates of the pathway and their biological implications. We identified IP6K1 as a key metabolic enzyme that requires tight regulation of its degradation to maintain cellular homeostasis. Previously, it was shown that elevated IP6K1 activity, linked to increased IP7 production, disrupts DNA repair and protein phosphorylation (33, 58). Additionally, IP6K1 impairs insulin signaling (32), which may contribute to insulin resistance and metabolic disorders such as obesity and diabetes (59). In contrast, IP6K1 deletion or inhibition reduces cell invasiveness and migration, offering protection against carcinogenesis (30). This altogether highlights the need for its tight regulation and position IP6K1 as a potential therapeutic target. Given the role of IPs in regulating metabolic balance between glycolysis and oxidative phosphorylation (47, 60), we investigated the role of IP6K1 in metabolic control under low-oxygen conditions. Depletion of IP6K1 resulted in reduced glucose uptake in hypoxia, impairing glycolytic ATP production consistent with the fact that cells rely heavily on glycolysis for ATP production when oxidative phosphorylation is restricted during hypoxia. This reduced glucose uptake also limits the availability of key metabolic intermediates necessary for nucleotide, lipid, and amino acid biosynthesis, thereby impairing cell proliferation during hypoxia (51, 53, 54). To cope with metabolic stress, hypoxia induces the transcription of glycolytic enzymes and glucose transporters. Although IP6K1 KO cells show increased transcript levels of GLUT1, OXPHOS, and glycolytic genes, their failure to uptake glucose, normally enhanced during hypoxia, disrupts this adaptive mechanism. Hypoxia-mediated adaptations involve changes in the regulatory control of key molecular components of metabolic pathways centered on the mitochondria, as well as dynamic changes in the morphology, mass, and subcellular localization of mitochondria themselves (61, 62). Consistent with this, we observed that IP6K1-deficient cells exhibited altered mitochondrial morphology. Moreover, these cells showed reduced survival, reinforcing the importance of IP6K1 in cellular homeostasis.

Oxygenation levels vary across mammalian tissues due to variations in metabolic demand, vascularization, and environmental conditions. While “normoxia” refers to atmospheric oxygen (about 20%), “physioxia” in the body ranges from 11% to 1% depending on the tissue (63). This variability underscores the importance of considering tissue-specific oxygen levels in cellular studies. Future research comparing the turnover of MC-starting proteins across tissues could reveal their roles in oxygen homeostasis and uncover tissue-specific regulatory mechanisms, enhancing our understanding of cellular adaptation to oxygen fluctuations in health and disease.

Materials and Methods

Cell Culture.

HEK293T (ATCC CRL-3216), HeLa cells (a gift from Kimchi A., Weizmann Institute of Science), and U2OS cells (A gift from Shav-Tal Y., Bar-Ilan university) were grown in Dulbecco’s Modified Eagle’s Medium (DMEM) (Life Technologies) supplemented with 10% fetal bovine serum (FBS) (Gibco) and penicillin/streptomycin (Life Technologies). Cells were maintained at 37 °C under an atmosphere of 5% CO2 in air. Hypoxia was induced by incubations within a Hypoxia Incubator Chamber (# 27310; STEMCELL). The proteasome inhibitor Bortezomib (#A2614; APExBio) was used at a final concentration of 1 μM and Cycloheximide (#A8244; APExBio) was used at a final concentration of 100 μg/ml.

Transfection and Lentivirus Production.

Lentivirus was generated through the transfection of HEK293T cells using PolyJet In Vitro DNA Transfection Reagent (#SL100688; SignaGen Laboratories). Cells seeded at ∼80% confluency were transfected as recommended by the manufacturer with the lentiviral transfer vector plus four plasmids encoding Gag-Pol, Rev, Tat, and VSV-G. The lentiviral supernatants were collected 48 h later. Transduction of target cells was achieved by adding the virus in the presence of 8 μg/ml hexadimethrine bromide (Polybrene) (# H9268; Sigma-Aldrich).

Plasmids.

ORFs encoding IP6K1, PIFO, IL32, TOM1L1, GMDS, TRMT44, HOXA10, PPIA, HMCES, GAS2, and ATP5S were obtained from the Ultimate ORF Clone collection (Thermo Fisher Scientific). ORFs were amplified by PCR to include XhoI and BstBI sites and were cloned using the Gibson assembly method (NEBuilder HiFi Cloning Kit) into pHAGE-GFP-IRES-DsRed vector N-terminal to GFP. To monitor ORFs levels by western blot, ORFs were amplified by PCR to include an HA epitope at their C-terminus and subcloned into the lentiviral pHAGE vector that also contains IRES-DsRed cassette to monitor equal expression of the constructs in various cell lines by flow cytometry.

For individual CRISPR/Cas9-mediated gene disruption experiments, the lentiCRISPR v2 vector was used (#52961; Addgene). Oligonucleotides encoding the top and bottom strands of the sgRNAs were synthesized (IDT), annealed, and cloned into the lentiCRISPR v2 vector as previously described (64).

Nucleotide sequences of the sgRNAs used were:

• sg-AAVS1: 5′-GGGGCCACTAGGGACAGGAT-3′

• sg-ADO: 5’-ATGGACAAGCTAGACGCGGG-3’

• sg-CDO1: 5’-GTAGCATCTTCAGAAAGCAG-3’

• sg1-ATE1: 5’-GTATCAGGATCTCATAGACCG-3’

• sg2-ATE1: 5’-GCCACAATCTGGCCCATTAGG-3’

• sg1-IP6K1: 5’-GGAGCTGCACAGCCACTCAG-3’

• sg2-IP6K1: 5’-GCAGTGGCAGTGACCACAAGG-3’

Generation of CRISPR/Cas9 Knockout Cells.

Lentivirus was generated through the transfection of HEK293T with lentiCRISPR v2 as explained before. 48 h following transduction cells were selected with puromycin to eliminate nontransduced cells. Seven days post transduction, genomic DNA of transduced cells was extracted, PCRs were performed to amplify ~500 base pairs flanking the edited site followed by Sanger sequencing. Inference of CRISPR Edits (ICE) CRISPR Analysis Tool was used to analyze efficiency of editing (Synthego Performance Analysis, ICE Analysis. 2019. v3.0.). UBR1/2/4 KO clone #2 (12) is used in this study as it gave the strongest stabilization effect of known UBR substrates.

Global Protein Stability (GPS) Screen.

The generation of a GPS lentiviral vector expressing a N-terminal peptide library was described previously (12). Briefly, the oligonucleotide pool synthesized by Agilent Technologies, was cloned into Ub-GPS vector using Gibson assembly downstream to ubiquitin and followed by GFP. For the scanning mutagenesis screen, oligonucleotides were synthesized to introduce mutations at individual residues or sets of three consecutive residues. Amino acids were replaced with ones possessing different characteristics from the original residues to disrupt potential degrons (e.g., large nonpolar to small polar, acidic to basic) (35). For saturation mutagenesis, each amino acid across all positions in selected peptides was mutated to all other 19 amino acids. The library was packaged into lentiviral particles and introduced into HEK293T cells at a multiplicity of infection of ~0.2 (achieving approximately 20% DsRed+ cells). After selection with Hygromycin (100 μg/ml) cells were partitioned into four bins based on the stability of the GFP fusion (GFP/DsRed ratio). Genomic DNA was extracted from cells collected from each of the bins using the Gentra Puregene Cell Kit (#158767; Qiagen) and the fusion peptides were amplified by PCR (Q5 Hot Start Polymerase; NEB) using primers binding in constant regions flanking the N24mer peptide for the first PCR, followed by a second PCR to add Illumina indexes and P5 and P7 adaptors. Samples to be multiplexed were then pooled, purified on an agarose gel (QIAEXII Gel Extraction Kit, #20051; Qiagen), and sequenced on an Illumina NextSeq instrument.

Data analysis was performed as described previously (12, 35). Raw Illumina reads derived from each GPS bin were first trimmed of constant sequences derived from the Ub-GPS vector backbone using Cutadapt (65) and count tables were generated from reads that aligned perfectly to the reference sequence. Following correction for sequencing depth, and based on the read counts of each peptide in every bin, the protein stability index (PSI) metric was calculated for each peptide-GFP fusion. The PSI score is given by the sum of multiplying the proportion of reads in each bin by the bin number thus yielding a stability score between 1 (maximally unstable) and 4 (maximally stable):

$$\mathrm{PSI}={\displaystyle \sum _{i=1}^4}Ri\ast i,$$

where $i$ represents the number of the bin and R $i$ represents the proportion of Illumina reads present for a peptide in that given bin $i$.

Read counts and associated stability score for each peptide-GFP fusion in the ATE1 KO screen are detailed in SI Appendix, Table S1. Candidate ATE1 substrates were defined based on three criteria: 1) Effect size analysis of Cohen’s D < –2 indicative of a very large difference between ATE1 KO and WT groups. 2) PSI_ATE1-PSI_Control (ΔPSI) ≥ 0.3. Peptides meeting the Cohen’s D < –2 criterion show a median ∆PSI of 0.2. To ensure high confidence, we adopted a more stringent ∆PSI cutoff of 0.3, supported by prior GPS screening studies (12, 35), in which candidates exceeding this threshold consistently validated as true positives. 3) Validation by histogram shifts: For each candidate meeting the above criteria, we examined PSI histograms and observed a consistent rightward shift in geometric mean in ATE1-deficient cells—further supporting substrate stabilization.

For classification of UBR substrates from the Ub-GPS screen detailed by Timms et al. (12):

a ΔPSI score was generated for each peptide-GFP fusion reflecting the difference in raw PSI scores between control KO sample and UBR KO clones #1–3. Peptide-GFP fusions were defined as UBRs substrates if they were stabilized ≥0.3 PSI units in at least one UBR KO clone. In addition, GPS screen profiles for each of representative Ub-GPS N24mers commencing with Cys showing the distribution of Illumina sequencing reads across the bins in control cells versus three independent UBR1/2/4 KO cells were plotted. Those with a positive median geometric mean stability of the peptide in UBR KO cells compared to WT cells (indicating a stabilization shift in UBR KO cells) were filtered as substrates.

For the mutagenesis screens- a ΔPSI score was generated for each peptide-GFP fusion reflecting the difference in raw PSI scores between wild type peptides and mutants. These values are detailed in SI Appendix, Table S2 and presented as heatmaps in Fig. 2.

Flow Cytometry.

Analysis of HEK293T cells by flow cytometry was performed on CytoFlex (Beckman Coulter) instrument and the resulting data were analyzed using FlowJo software. Cell sorting was performed on BD FACS Aria II (Becton Dickinson).

Immunoblotting.

Protein samples were collected in lysis buffer (10 mM NaPO4, 100 mM NaCl, 5 mM EDTA pH 8, 1% Triton X-100, 0.5% Deoxycholic acid sodium salt, 0.1% SDS) supplemented with Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific) and lysed for 30 min in +4 °C. Protein concentration was determined by a standard Bradford assay (#500-0006; Bio-Rad), a linear bovine serum albumin (BSA) calibration curve, and an Epoch microplate spectrophotometer. 30 μg of total cell extract media were subsequently resolved by SDS-PAGE (Mini-PROTEAN TGX Precast Protein Gels; Bio-Rad) and transferred to a nitrocellulose membrane (Trans-Blot Turbo System; Bio-Rad) which was then blocked in 10% nonfat dry milk in PBS + 0.1% Tween-20 (PBS-T). The membrane was incubated with primary antibody overnight at 4 °C, and then, following three washes with PBS-T, HRP-conjugated secondary antibody was added for 1 h at room temperature. Following a further three washes in PBS-T, reactive bands were visualized using SuperSignal West Femto chemiluminescence substrate (#34095; Pierce) or an EZ-ECL (#20-500-171; Biological Industries) for 5 min using the ImageQuant TL software v8.2 on Amersham Imager 680 (Cytiva). Primary antibodies used: HIF-1α (# A300-286A, FORTISLIFE - Bethyl), GFP (Abcam, ab290), IP6K1 (A305-628A-T; FORTISLIFE - Bethyl) TSPYL1 (#ab95943; Abcam), HA-Tag (C29F4) (#3724; Cell Signaling Technology), Vinculin (#V9131; Sigma-Aldrich), β-actin (#4967; Cell Signaling Technology), GAPDH (14C10) (#2118; Cell Signaling Technology). HRP-conjugated goat anti-rabbit and anti-mouse IgG secondary antibodies were obtained from Jackson ImmunoResearch (#111-035-003, #115-035-003, respectively). Densitometric analysis was performed using ImageJ software (NIH) and values were presented relative to loading control protein.

Cycloheximide Chase Assay.

Following treatment with 100 μg/ml Cycloheximide, cells were harvested at the indicated time points and subjected to Western blot as explained before.

Analysis of Ubiquitination.

HEK293T cells stably expressing ORF fused to GFP were grown in 10 cm plates and transfected with HA–ubiquitin plasmid (29). 48 h post transfection, cells were treated with bortezomib (1 μM, 5 h), and then were lysed in ice-cold lysis buffer (10 mM NaPO4, 100 mM NaCl, 5 mM EDTA pH 8, 1% Triton X-100, 0.5% Deoxycholic acid sodium salt, 0.1% SDS) supplemented with HaltTM Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific) and 50 µM of the deubiquitinating enzyme inhibitor PR-619 (#662141; EMD Millipore) for 30 min on ice. Protein concentration was determined by Bradford assay and equal amounts were taken for immunoprecipitation experiment by incubation for 2 h with 20 μl beads coated with anti-GFP (GFP-Trap_MA magnetic agarose beads (ChromoTek GmbH)). The beads were then washed three times with stringent washes in wash buffer (10 mM NaPO4, 300 mM NaCl, 5 mM EDTA pH 8, 1% Triton X-100, 0.5% Deoxycholic acid sodium salt, 0.1% SDS) supplemented with HaltTM Protease and Phosphatase Inhibitor Cocktail and 50 µM PR-619 before bound proteins were eluted upon incubation with SDS-PAGE sample buffer (95 °C, 10 min). SDS-PAGE and immunoblot was done as explained before.

Immunocytochemistry.

Cells were fixed with 4% paraformaldehyde and then permeabilized with 0.2% Triton X-100. After blocking with 3% BSA in PBS, mitochondria were stained with COX-IV antibody (#4850; Cell Signaling Technology). Following three washes with PBS-T, DAPI staining (#D9542; Sigma-Aldrich) was added for 1 min. Finally, coverslips were mounted onto slides using mounting media (#F6182; Sigma-Aldrich) prior to imaging. Cells were imaged with the Leica Stellaris 5 confocal microscope using LASX software and a 63× oil lens/1.4 N.A. UPlanSApo objective (Olympus). Analysis was done in ImageJ using the Mitochondria Analyzer plugin (66).

Measurement of Intracellular ATP and Glucose Uptake.

ATP level was quantified using CellTiter-Glo assay (#G7571; Promega) according to the manufacture protocol. Briefly, for ATP detection, for each well containing 500 μl media, 500 μl of CellTiter-Glo reagent was added. The luminescence levels from 200 μl per well were recorded after 10 min of incubation.

Glucose uptake was quantified using Glucose Uptake-Glo (#J1341; Promega) according to manufacture protocols. Briefly, for glucose uptake detection, cells were rinsed in warmed PBS and placed in 100 μL PBS in a 24-well culture plate. 2-Deoxyglucose (1 mM) was added for 10 min uptake. After stopping and neutralization, 2-deoxyglucose-6-phosphate detection reagent was added, supernatant was transferred to white 96-well plate (Costar) and luminescence was recorded at 30 min.

Normalization of cell number was done with Hoechst 33342 stain (62249, ThermoFisher Scientific) of 1 ng/ml final concentration. Fluorescence signal intensity was detected in plate reader at excitation ~350 nm and emission ~460 nm wavelength.

Colony Formation Assay.

To evaluate the ability of single cancer cells to form a colony, cells were plated at a concentration of 2,000 cells/well on six-well plates (day 0) in replicates and were grown in normoxic or hypoxic conditions. At day 10, cells were fixed with methanol:acetic acid (3:1) for 5 min. Colonies were stained with crystal violet working solution (0.1% in methanol) for 15 min. Colonies areas were measured for each well using ImageJ software (version 1.54 h; NIH) and the ratio between hypoxia and normoxia was calculated.

RT-qPCR.

mRNA was extracted using the RNEASY PLUS MINI KIT (#74104; QIAGEN) and equal amounts were used for cDNA synthesis using the qScript cDNA Synthesis Kit (#101414; Quantabio). qPCR analysis was performed using Applied Biosystems™ Power SYBR™ Green PCR Master Mix, on a CFX96 Touch Real-Time PCR Detection System (BioRad) using the ΔΔCt method. Levels of the housekeeping gene β-actin (ACTB) were used as a reference. Sequences for the primers used are as follows (FP- forward primer, RP- reverse primer):

1. GLUT1 FP:5’- CTGTCGTGTCGCTGTTTGTG-3’

2.  RP:5’- CTAGCGCGATGGTCATGAGT-3’

3. MCTO1 FP:5’- CCGTCCTAATCACAGCAGTCCTA-3’

4.  RP:5’- TGAGGTTGCGGTCTGTTAGTAGT-3’

5. HK2 FP:5’- CAAGTGCAGAAGGTTGACCA-3’

6.  RP:5’- CTCCGTGTTCTGTCCCATCT-3’

7. LDHA FP:5’- GACCTACGTGGCTTGGAAGA-3’

8.  RP:5’- TCCATACAGGCACACTGGAA-3’

9. PDK1 FP:5’- TTCGGATCAGTGAATGCTTG-3’

10.  RP:5’- AGCATCCTCAGCACTTTTGTC-3’

11. GLI1 FP:5’- AGCCCAGATGAATCACCAAAAAG-3’

12.  RP:5’- CAGCATGTCCAGCTCAGACTTC-3’

13. PTCH1 FP:5’- GAAGGTGCTAATGTCCTGAC-3’

14.  RP:5’- TACCTAGGAGGTATGCTGTC-3’

15. ACTB FP:5’- CACCTTCTACAATGAGCTGCGTGTG-3’

16.  RP:5’- ATAGCACAGCCTGGATAGCAACGTAC-3’

Statistical Analysis.

To evaluate statistical significance of the enrichment of residues in position 3 in MC$\underset{\_}{X}$-containing proteins, a Z-score was calculated with the formula:

$$Z-score=\frac{\mathrm{Enrichment} \mathrm{factor}}{\sigma }.$$

The Enrichment factor is the ratio of CX- occurrences in ATE1 substrates to the total occurrences of X residue in position 3 in the proteome. σ is the SD of CX- positive occurrences. This data is presented in Fig. 1B.

To evaluate the differential enrichment of CX- motifs between ATE1-positive and ATE1-negative groups, we calculated the log₂ enrichment score for each motif as

$${\mathit{log}}_2\frac{\text{Frequency} \text{in} \text{ATE}1 \text{positive}}{\text{Frequency} \text{in} \text{ATE}1 \text{negative}}.$$

Motifs with higher positive values are enriched in ATE1-positive substrates, while negative values indicate depletion. A heatmap was generated to visualize the distribution of log₂ enrichment scores across motifs, as presented in Fig. 1C. The log₂ enrichment/depletion score is shown for each motif, and the motifs are ordered based on the statistical significance of the log₂ comparisons between groups, from highest to lowest. Values were scaled and centered at zero.

Unless specified otherwise, data are expressed as means ± SEM.

Comparisons between groups were made via unpaired two-tailed Student’s t test and two-way ANOVA. Statistical significance was set at P ≤ 0.05; ns indicates no significance, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.