SOD2 is a regulator of proteasomal degradation promoting an adaptive cellular starvation response

Abstract

Adaptation to changes in amino acid availability is crucial for cellular homeostasis, which requires an intricate orchestration of involved pathways. Some cancer cells can maintain cellular fitness upon amino acid shortage, which has a poorly understood mechanistic basis. Leveraging a genome-wide CRISPR-Cas9 screen, we find that superoxide dismutase 2 (SOD2) has a previously unrecognized dismutase-independent function. We demonstrate that SOD2 regulates global proteasomal protein degradation and promotes cell survival under conditions of metabolic stress in malignant cells through the E3 ubiquitin ligases UBR1 and UBR2. Consequently, inhibition of SOD2-mediated protein degradation highly sensitizes different cancer entities, including patient-derived xenografts, to amino acid depletion, highlighting the pathophysiological relevance of our findings. Our study reveals that SOD2 is a regulator of proteasomal protein breakdown upon starvation, which serves as an independent catabolic source of amino acids, a mechanism co-opted by cancer cells to maintain cellular fitness.

Keywords: CP: Cancer; CP: Molecular biology; SOD2; UBR1; UBR2; amino acid starvation; cancer; drug resistance; leukemia; protein degradation.

Figure 1.. SOD2 inhibition sensitizes leukemia cells to asparagine depletion

(A) Significance of SOD2 depletion upon asparaginase treatment was assessed by robust ranking aggregation (RRA) score calculated using the MAGeCK analysis. Note that details on the CRISPR-Cas9 screen can be found in. (B) CCRF-CEM cells were transduced with indicated shRNAs, and knockdown efficiency was assessed with RT-qPCR analysis in biological duplicates. (C) CCRF-CEM cells were transduced with shLuc, shSOD2 #3, or shSOD2 #5. Protein levels of SOD2 and GAPDH were assessed by western blot analysis. (D) CCRF-CEM cells were transduced with shLuc, shSOD2 #3, or shSOD2 #5 and treated with indicated asparaginase doses in biological triplicates. Relative viability was assessed after 8 days of treatment. All cell counts were normalized to shLuc-transduced, vehicle-treated cells. (E) CCRF-CEM cells were transduced with indicated shRNAs. Cells expressing shRNAs in the presence or absence of SOD2 cDNA were subsequently treated with the indicated doses of asparaginase in biological triplicates. Relative viability was assessed after 8 days of treatment. Cell counts were normalized as in (D). (F) CCRF-CEM cells were transduced with indicated shRNAs and treated with 100 U/L asparaginase for 48 h, and caspase 3/7 activity was assessed in biological duplicates. All error bars represent SEM. ****p ≤ 0.0001, ***p ≤ 0.001 by one-way ANOVA Dunnett’s adjustment for multiple comparisons (B and F). See also Table S1 and Figure S1.

Figure 2.. SOD2 inhibition sensitizes different cancer entities to asparaginase-induced cytotoxicity

(A–F, left) Jurkat, MOLT4, OCI-AML2, NALM-16, HCT-15, and SW480 cells were transduced with the indicated shRNAs, and the knockdown efficiency of SOD2 was assessed by RT-qPCR analysis in biological duplicates. (A–F, right) Jurkat, MOLT4, OCI-AML2, NALM-16 HCT-15, and SW480 cells were treated with the indicated doses of asparaginase in biological triplicates. After 8 days of treatment, relative viability was assessed. Cell counts were normalized to shLuc-transduced, vehicle-treated cells. (G–J) Jurkat cells were transduced with the indicated shRNAs and treated with the indicated doses of vincristine, dexamethasone, doxorubicin, and 6-mercaptopurine in biological duplicates. After 8 days of treatment, relative viability was assessed. Cell counts were normalized as in (A–F). All error bars represent SEM. ***p ≤ 0.001, **p ≤ 0.01, *p < 0.05 by two-sided Student’s t test with Welch adjustment (A–F, left). See also Figure S1 and Table S2.

Figure 3.. SOD2-regulated asparaginase response is independent of known SOD2-associated pathways

(A) CCRF-CEM cells were transduced with the indicated shRNAs and treated with 100 U/L asparaginase. ROS levels were measured by assessing the percent of PE positivity with flow cytometry in biological duplicates. Antimycin A served as a positive control and N-acetylcysteine (NAC) as a negative control. (B) CCRF-CEM cells were treated with vehicle, 1 mM xanthine, 1 mM xanthine plus 0.01 U/mL xanthine oxidase (XO) in the presence or absence of 100 U/L asparaginase in biological triplicates. Relative viability was assessed after 8 days of treatment. Cell counts were normalized to vehicle-treated cells. (C) CCRF-CEM cells were transduced shLuc or shSOD2 #3 and treated with vehicle or 5 mM NAC in the presence or absence of 100 U/L asparaginase in biological triplicates. Relative viability was assessed after 8 days of treatment. Cell counts were normalized to shLuc-transduced, vehicle-treated cells. (D) CCRF-CEM cells were transduced with the indicated shRNA and treated with either vehicle or 20 μM ebselen in the presence or absence of 100 U/L asparaginase in biological triplicates. Relative viability was assessed as in (C). (E and F) Cells were transduced with shLuc or shSOD2 #3 and treated with vehicle or 100 U/L asparaginase. Protein levels of asparagine synthetase (ASNS), glutamine synthetase (GLUL), and GAPDH were assessed by western blot analysis. Densitometry was performed in biological duplicates for the target bands (ASNS or GLUL) and normalized to their respective GAPDH. (G) Cells transduced with shLuc or shSOD2 #3 were treated with 100 U/L asparaginase after confirmation of an efficient knockdown. Relative expression of amino acid transporters was assessed by RT-qPCR analysis in biological duplicates. (H) Cells transduced with shLuc or shSOD2 #3 were treated with vehicle or 100 U/L asparaginase after confirmation of an efficient knockdown. Protein levels of p-p70S6K, p70S6K, and GAPDH were evaluated by western blot analysis. All error bars represent SEM. ****p ≤ 0.0001, **p ≤ 0.01, *p < 0.05, n.s. p ≥ 0.05 by one-way ANOVA with Dunnett’s adjustment for multiple comparisons (A–F) and two-sided Student’s t test with Welch adjustment (G). See also Figures S2, S3, and Table S3.

Figure 4.. SOD2 promotes cell survival during amino acid starvation by interacting with UBR2

(A) Jurkat cells were transduced with shLuc, or shSOD2 #5, followed by transduction with a V5-tagged pLX304 control vector, pLX304-SOD2 wildtype (WT), or pLX304-SOD2 Y34F vector. After 4 days of treatment, viability was assessed in biological duplicates. Cell counts were normalized to vehicle-treated cells for each condition. (B) Jurkat cells were transduced with shLuc, or shSOD2 #5, and treated as indicated. After 8 days of treatment, viability was assessed. Cell counts were normalized to vehicle-treated cells. (C) Jurkat cells were transduced with the indicated shRNAs and pLX304-SOD2 WT vector and cultured in RPMI medium supplemented with only essential amino acids (EAA) or non-essential amino acids (NEAA) in biological duplicates. Cell counts were normalized to shLuc-transduced cells for each condition. (D) Schematic depiction of the interaction of SOD2 with UBR2. The protein-protein interaction is based on the BioPlex database.^, (E and F) Immunoprecipitation of UBR2 reveals a co-immunoprecipitation with SOD2 in Jurkat cells in a whole cell lysate using a non-denaturing buffer (E) and in a cytosolic cell fraction (F). (G) Jurkat cells were treated with vehicle or 100 U/L asparaginase for 48 h, and immunofluorescence staining was observed using a Zeiss LSM780 microscope. The co-localization of indicated targets was assessed by the Pearson correlation coefficient and ranked based on their degree of co-localization. (H) Representative images using super-resolution microscopy. Arrows indicate co-localization of UBR2 and SOD2. Scale bar, 2 μm (I) Jurkat cells were transduced with shLuc, or shSOD2 #5, followed by transduction with a V5-tagged pLX304 control vector, pLX304-SOD2 WT, or pLX304-SOD2 ΔMTS vector. After 8 days of treatment, viability was assessed in biological duplicates. Cell counts were normalized to vehicle-treated cells for each condition. (J) Jurkat cells were transduced with shUBR2 #2 and #3. Knockdown was confirmed by RT-qPCR in biological duplicates. (K) Cells were transduced with indicated shRNAs in the presence or absence of a UBR2 cDNA and treated with vehicle or 100 U/L asparaginase in biological triplicates. After 4 days of treatment, relative viability was assessed. Counts were normalized to shLuc-transduced cells. (L) Jurkat cells were treated as indicated. After a confirmed knockdown, cells were harvested, and protein levels of total ubiquitin and GAPDH were assessed by western blot analysis. (M) Jurkat cells were transduced with shLuc or shSOD2 and, subsequently UBR2 cDNA. After 4 days of treatment with vehicle or asparaginase (100 U/L) in biological duplicates, relative viability was assessed. Cell counts were normalized to shLuc-transduced, vehicle-treated cells. All error bars represent SEM. ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p < 0.05, n.s., p ≥ 0.05 by one-way ANOVA with Dunnett’s adjustment for multiple comparisons (A–C, I–K, and M). See also Figure S4.

Figure 5.. UBR1 inhibition phenocopies the defective response to amino acid starvation

(A) Jurkat cells were transduced with shUBR1 #1 and #4. Knockdown was confirmed by RT-qPCR in biological duplicates. (B) Jurkat cells were transduced with shUBR1 and treated with vehicle or 100 U/L asparaginase in biological triplicates. After 4 days of treatment, viability was assessed and normalized to shLuc-transduced cells. (C) Jurkat cells were transduced with shLuc or shUBR1 in addition to a UBR1 cDNA. After 4 days of treatment with vehicle or asparaginase (100 U/L) in biological duplicates, relative viability was assessed. Cell counts were normalized to shLuc-transduced, vehicle-treated cells. (D) Jurkat cells were transduced with shLuc or shSOD2 and subsequently UBR1 cDNA. After 4 days of treatment with vehicle or asparaginase (100 U/L) in biological duplicates, relative viability was assessed as in (C). (E–G) Jurkat cells were transduced with indicated shRNAs and cultured in a medium supplemented with all amino acids, or in a medium deficient in essential or non-essential amino acids in biological triplicates. Cell counts were normalized to shLuc-transduced cells. All error bars represent SEM. ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, n.s., p ≥ 0.05 by one-way ANOVA with Dunnett’s adjustment for multiple comparisons (A–G). See also Figure S5.

Figure 6.. SOD2 is not a target of UBRs and does not appear to regulate the ATE1- dependent axis of the Arg/N-degron pathway

(A) Jurkat cells were transduced with the indicated shRNA and treated with vehicle or asparaginase (100 U/L) for 48 h. SOD2 and GAPDH levels were assessed by western blot analysis. (B) Jurkat cells were transduced with the indicated constructs and treated with vehicle or asparaginase (100 U/L) for 48 h. SOD2 mRNA levels were assessed by RT-qPCR in biological duplicates and normalized to shLuc control cells. (C) Jurkat cells were treated with vehicle or 100 U/L asparaginase in the presence or absence of 1 μM tannic acid (TA). Relative viability was assessed after 4 days of treatment in biological duplicates. Cell counts were normalized to vehicle-treated cells. (D) Jurkat cells were transduced with indicated shRNAs, and knockdown efficiency was assessed by RT-qPCR in biological duplicates (left). Upon successful knockdown, cells were treated with vehicle or 100 U/L asparaginase (right). Relative viability was assessed after 4 days of treatment in biological triplicates. (E) Jurkat cells were transduced with indicated shRNAs. Levels of RGS4 and alpha-tubulin were assessed by western blot analysis. All error bars represent SEM. ****p ≤ 0.0001; n.s., p ≥ 0.05 by one-way ANOVA with Dunnett’s adjustment for multiple comparisons (B–E). See also Figure S5.

Figure 7.. SOD2-mediated protein breakdown promotes cancer cell fitness upon amino acid starvation and reflects an adaptive proteasomal degradation machinery

(A and B) Jurkat cells were transduced with the indicated shRNAs and treated with 100 U/L of asparaginase. Western blot analysis was performed for K48-linked ubiquitin (A) or K63-linked-ubiquitin (B) together with GAPDH. Densitometry analysis was performed for target protein levels and normalized to the respective GAPDH signal in biological triplicates. (C) Jurkat cells were transduced with the indicated constructs and treated with 1,000 U/L of asparaginase for 48 h in the presence or absence of bortezomib (20 nM for 5 h). Western blot analysis was performed for K48-linked ubiquitin. Note that a very short time of exposure was chosen due to the striking K48 accumulation in bortezomib-treated shLuc cells, resulting in a lack of signal in samples not treated with bortezomib. (D) Jurkat cells were transduced with shLuc or shSOD2 in addition to a pLX304-UBR2 RING wildtype (WT) or mutant (mut) vector, and viability was assessed after 4 days of treatment in biological duplicates. Viability was normalized to each vehicle condition. (E) Jurkat cells were transduced with shLuc or shSOD2 in addition to a pLX304-GFP control vector, or the hyperactive proteasomal subunit pLX304-ΔN-PSMA4. After 8 days of treatment with vehicle or asparaginase (100 U/L) in biological triplicates, relative viability was assessed. Cell counts were normalized to shLuc-transduced, vehicle-treated cells. (F) Jurkat cells were transduced as indicated and treated with vehicle or 10 U/L asparaginase for 48 h. Asparagine content in cells was quantified by liquid chromatography tandem mass spectrometry. (G) (Inset) Jurkat cells were first transduced with shLuc, shSOD2 #3, shUBR1 #1, shUBR2 #3, or a combination of shUBR1+2. Cells were then treated with vehicle, or asparaginase (100 U/L) in a complete growth medium (RPMI-1640 + 10% fetal bovine serum), or treated with asparaginase in a complete growth medium supplemented with 10× of L-asparagine or 10× of L-glutamine. Fifty percent of the media was removed every 12 h and replaced with fresh growth medium, supplemented with the appropriate concentration of asparaginase, asparagine, or glutamine. (Bottom) The viability was assessed after 48 h by counting viable cells and normalized to the viability in vehicle controls. (H) Schematic depiction of generation of B-ALL PDX models used in this study. (I–M) ALL PDX specimens were transduced with the indicated constructs, and treated with asparaginase (100 U/L) for 48 h. Viability was normalized to shLuc cells. Note that treatment could not be extended beyond 48 h due to the limited life span of PDX cells. (N) Proposed model. All error bars represent SEM. ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p < 0.05, n.s., p ≥ 0.05 by one-way ANOVA with Dunnett’s adjustment for multiple comparisons (A, B, D–G, and I–M). See also Figure S5–S10.