Succinyl-CoA ligase ADP-forming subunit beta promotes stress granule assembly to regulate redox and drive cancer metastasis

Significance

Metabolic changes often occur during metastasis when tumors disseminate from primary sites. Therefore, a detailed understanding of these changes will provide new therapeutic strategies. In this work, we used an unbiased RNAi screen to identify succinyl-CoA ligase ADP-forming subunit beta (SUCLA2) as a critical factor that manages redox balance and promotes survival of disseminated cancer cells during metastasis. We show that the metabolic contribution of SUCLA2 during metastasis is independent of its role in the Krebs cycle and involves promoting protein expression of redox-scavenging enzymes through stress granules in the cytosol. Our finding provides an academic basis for the development of new treatment targets against metastasis, which attacks altered mitochondrial metabolic features that occur during tumor cell dissemination.

Abstract

Although recent studies demonstrate active mitochondrial metabolism in cancers, the precise mechanisms through which mitochondrial factors contribute to cancer metastasis remain elusive. Through a customized mitochondrion RNAi screen, we identified succinyl-CoA ligase ADP-forming subunit beta (SUCLA2) as a critical anoikis resistance and metastasis driver in human cancers. Mechanistically, SUCLA2, but not the alpha subunit of its enzyme complex, relocates from mitochondria to the cytosol upon cell detachment where SUCLA2 then binds to and promotes the formation of stress granules. SUCLA2-mediated stress granules facilitate the protein translation of antioxidant enzymes including catalase, which mitigates oxidative stress and renders cancer cells resistant to anoikis. We provide clinical evidence that SUCLA2 expression correlates with catalase levels as well as metastatic potential in lung and breast cancer patients. These findings not only implicate SUCLA2 as an anticancer target, but also provide insight into a unique, noncanonical function of SUCLA2 that cancer cells co-opt to metastasize.

While altered metabolism is a hallmark of cancer, growing evidence indicates that there exists a dynamic change in the metabolism of metastasizing cancer cells that is essential when transitioning through the changes during the metastatic cascade (1–5). Although the requirement for mitochondrial adenosine triphosphate (ATP) production is reduced in glycolytic tumor cells, the demand for TCA cycle-derived biosynthetic precursors and nicotinamide adenine dinucleotide phosphate (NADPH) is unchanged or even increased (6). Studies suggest that metastatic cancer cells exhibit greater utilization of mitochondrial metabolic pathways (4, 5, 7, 8). Nevertheless, recent studies demonstrate very few mitochondrial enzymes that are implicated in metastasis prevention and treatment, and pivotal mitochondrial metabolic pathways responsible for the metastatic switch have not been extensively explored (9).

Reports estimate that primary tumors can shed millions of cancer cells into the circulation every day, most of which are eliminated by anoikis, a programmed cell death induced by extracellular membrane detachment, and thus never metastasize (10, 11). Therefore, cancer cells becoming resistant to anoikis is an essential step during metastasis, but how mitochondrial metabolic reprogramming allows tumor cells to survive during metastasis remains largely unclear. It is widely accepted that reactive oxygen species (ROS) could play distinct roles in metastasis depending on their concentrations, distributions, and subcellular locations. ROS are reported to induce epithelial–mesenchymal transition and mitogen-activated protein kinase (MAPK) activation, thereby playing a crucial role in metastasis (12, 13). However, studies demonstrate that oxidative stress sensitizes cells to anoikis and upregulation of a ROS scavenger superoxide dismutase promotes anoikis resistance and cancer metastasis, suggesting a link between redox metabolism and anoikis resistance (14, 15).

Succinyl-CoA synthetase (SCS) is a mitochondrial enzyme that catalyzes the reversible conversion of succinyl-CoA and ADP or GDP to succinate and ATP or GTP (16). SCS functions as a heterodimer composed of an invariable α subunit (SUCLG1) and a variable β subunit (SUCLA2 or SUCLG2). SUCLA2 encodes for the succinyl-CoA ligase ATP-specific β subunit, which together with the α subunit forms ATP-specific SCS (A-SCS), the enzyme in the TCA cycle that can generate ATP via direct substrate-level phosphorylation. SCS-A can generate two ATPs per molecule of glucose that enters the TCA as two molecules of pyruvate-derived acetyl-CoA. Mutations in SUCLA2 that impair SCS activity led to succinyl-CoA accumulation and global succinylation of proteins in mitochondria (17). The SUCLA2 gene is located 302 kilobase pairs away from the RB1 gene and a strong positive correlation is observed in copy number between RB1 and SUCLA2 genes in prostate cancers (18). Nonetheless, little is known about the biological function of SUCLA2 in human cancers and beyond its classical role in the TCA cycle.

Stress granules and P-bodies are dense aggregations in the cytosol composed of RNAs and proteins that appear to help cells cope with stress stimuli (19, 20). While stress granules assemble key components of the translation machinery, P-bodies assemble the essential enzymes of cytoplasmic RNA degradation (21). Cellular stress leads to mRNA translation reprogramming, and emerging evidence indicates that transcripts localized to stress granules can undergo translation (22). Ras-GTPase-activating protein SH3 domain-binding protein 1 and 2 (G3BP1/2), ubiquitin-specific protease 10 (USP10), and ataxin-2-like (ATXN2L) are components of stress granules that play a role in stress granule formation (23–25). Nuclear fragile X mental retardation-interacting protein 2 (NUFIP2) is known to relocalize to stress granules upon exposure to stress (26). Stress granules have emerged as systems that can integrate oncogenic signals and tumor-related stress stimuli to promote cancer cell fitness (19), and upregulation of stress granules has been proven in many types of cancers including breast and prostate cancer (27, 28). However, the molecular interactions and mechanisms that modulate stress granule assembly and how these may become altered in human diseases remain largely unknown.

To explore the relationship between mitochondrial metabolism and cancer metastasis, we generated a customized RNAi library targeting mitochondrial enzymes. Using this approach, we have obtained evidence that SUCLA2 promotes anoikis resistance. SUCLA2 contributes to stress granule assembly upon extracellular matrix (ECM) detachment, enhances the translation of cytosolic ROS scavengers to manage redox balance and promote survival of disseminated cancer cells during metastasis. These studies uncover the noncanonical function of the Krebs cycle enzyme component SUCLA2 in cancer metastasis that is associated with stress granules and redox metabolism.

Results

The Krebs Cycle Enzyme SUCLA2 Is Important for Cancer Cells to Resist Anoikis.

To better picture how mitochondrial metabolism contributes to tumor metastasis in human cancer, we performed an unbiased RNAi screen using a customized shRNA library targeting 119 mitochondrial metabolic enzyme genes represented by 605 lentiviral vector-based shRNA constructs. Lung cancer A549 cells were transduced with pooled virus harboring two to five shRNA clones targeting each individual gene and cultured under detachment conditions. The top candidates whose knockdown induced anoikis by greater than 2.5-fold and are not implicated in cancer metastasis were succinyl-CoA ligase adenosine diphosphate (ADP) forming subunit beta (SUCLA2), thymidine kinase 2 (TK2), and ubiquinol-cytochrome C reductases (UQCRQ and UQCRH) (Fig. 1A and SI Appendix, Table S1). To further substantiate the role of these candidate genes in cancer cell anoikis resistance, we generated diverse human cancer cell lines with a stable knockdown. These cell lines include lung cancer A549 and H460 cells, and breast cancer MDA-MB231 and MDA-MB468 cells. We identified SUCLA2, an enzyme component in the Krebs cycle, as a lead hit that may be commonly important for anchorage-independent survival in lung and breast cancers (SI Appendix, Fig. S1 A–C). Loss of SUCLA2 resulted in enhanced detachment-induced apoptotic cell death in all the cell lines assessed by annexin V staining and caspase 3/7 activity assay (Fig. 1 B and C and SI Appendix, Fig. S1D). However, SUCLA2 knockdown did not significantly affect the cell proliferation rate or apoptotic cell death of these cells under attached culture conditions, suggesting that SUCLA2 is specifically involved in controlling apoptosis mediated by detachment (Fig. 1D and SI Appendix, Fig. S2). We next assessed whether the overexpression of SUCLA2 could confer anoikis resistance potential. We observed that transiently enforced SUCLA2 expression significantly reduced anoikis in all the cancer cells tested (Fig. 1E). Furthermore, the expression of an shRNA-resistant form of human SUCLA2 rescued the anoikis-resistant potential lost by the endogenous SUCLA2 knockdown (Fig. 1F). These data suggest that SUCLA2 is a potential anoikis resistance driver in lung and breast cancer cell lines.

Fig. 1.

SUCLA2 is important for anoikis resistance in cancer cells. (A) RNAi screen evaluating the effect of targeting 119 mitochondrial genes on anoikis induction was carried out using A549 cells on 1% agarose coated dishes for 48 h. Cell death induced by detachment was assessed by annexin V staining. Candidates with viral infection rate less than 25% (gray) or shRNAs that induce cell death greater than 15% under attached condition (blue) were excluded. (B and C) Effect of SUCLA2 target downregulation on anoikis induction in lung and breast cancer cells. Cells were transduced with SUCLA2 shRNA clones followed by anoikis induction by culturing on 1% agarose. Anoikis was assessed after 48 h by annexin V staining (B) and caspase 3/7 activity assay (C). (D) The effect of SUCLA2 knockdown on cell proliferation rates under attached conditions was determined by a bioluminescent cell viability assay. (E) Effect of flag c-terminal tagged SUCLA2 overexpression on anoikis resistance in diverse cancer cell lines. Anoikis rates were determined by annexin V staining or caspase 3/7 activity assay. (F) Anoikis induction changes in cells with rescue expression of shRNA-resistant SUCLA2 in SUCLA2 knockdown cells. Data shown are mean ± SD from two replicates for the screen shown in (A) and three replicates for the rest. One-way ANOVA (B, D, and F) and two-tailed Student’s t test (C and E) were used for statistics (ns: not significant; *: 0.01 < P < 0.05; **P < 0.01).

SUCLA2 Contributes to Anoikis Resistance Through Controlling Redox Status.

We next explored whether SUCLA2 contributes to anoikis resistance by offering metabolic advantages in cancer cells. As SUCLA2 is associated with mitochondrial energy production, we first assessed whether SUCLA2 alters bioenergetics flux, including oxygen consumption rate (OCR) and extracellular acidification rate (ECAR), and the total intracellular ATPs under anchorage-independent conditions. Although SUCLA2 is involved in generating ATP through glucose that enters the TCA cycle, neither the bioenergetics flux nor the intracellular ATP levels of the cell lines tested were significantly changed by the SUCLA2 knockdown in detached cells (Fig. 2 A and B). Moreover, the knockdown of SUCLA2 similarly induced anoikis under hypoxia compared to cells under normoxia conditions (Fig. 2C). These data suggest that SUCLA2 has a non-bioenergetic role during tumor dissemination and metastasis.

Fig. 2.

SUCLA2 contributes to cancer metastasis by controlling redox status. (A) Bioenergetics profiles, including oxygen consumption rate (OCR; Top) and ECAR (Bottom) were obtained over time using extracellular flux analyzer in detached A549, H460, and MDA-MB231 cells with SUCLA2 knockdown or an empty vector. (B) SUCLA2 knockdown effect on ATP levels in attached or detached cells. (C) Outcome of SUCLA2 loss on anoikis resistance under hypoxic (1% oxygen) or normoxic (20% oxygen) conditions. (D) SUCLA2 knockdown effect on cellular ROS levels under attached or detached culture conditions. (E) Detached cancer cells with SUCLA2 knockdown were treated with 0.5 mM N-acetyl cysteine (NAC) for 48 h and cellular ROS and anoikis rates were determined. (F–I) NAC was administered through daily drinking water at 10 mg/mL in A549 xenograft mice with SUCLA2 knockdown and changes in metastasis potential were monitored. SUCLA2 expression in injected cells was confirmed by immunoblotting in (F). Number of metastatic nodules in each mouse (G), number of mice with metastasis in each group (H), and representative images of hematoxylin and eosin (H&E) stained lung section from each group (I) at week 9 are shown. Scale bars shown in (I) represent 100 μm. Data shown are mean ± SEM from three replicates for (A) and ± SD from three replicates of each group for (B–E). Data shown in (G) are mean ± SEM from 10 mice. P values were obtained by one- or two-way ANOVA (A–E, and G) and Chi-square test (H) (ns: not significant; *: 0.01 < P < 0.05; **P < 0.01).

We hypothesized that other metabolic advantages such as redox balance may drive the SUCLA2-mediated anoikis resistance. Indeed, we noticed that the loss of SUCLA2 resulted in a significant increase in ROS levels when cells were under detachment conditions but not when cells were adherent, suggesting that SUCLA2 contributes to ROS scavenging in cancer cells upon ECM detachment (Fig. 2D). In line with this observation, a metabolic profiling monitoring levels of 91 metabolites involved in central metabolism including glycolysis, pentose phosphate pathway, TCA cycle, amino acid metabolism, and nucleic acid metabolism in detached lung cancer cells with SUCLA2 loss demonstrated that the knockdown of SUCLA2 resulted in decreases in antioxidants, whereas oxidized forms increased (SI Appendix, Fig. S3). Treatment with N-acetylcysteine (NAC) or L-ergothioneine, pharmacological antioxidants, significantly attenuated both the elevated ROS and increased anoikis that were mediated by SUCLA2 loss in a panel of cancer cell lines including A549, H460, and MDA-MB231 (Fig. 2E and SI Appendix, Fig. S4). We next functionally validated the effect of SUCLA2 on tumor progression in vivo by xenograft mouse study. Mice were intravenously injected with A549 cells that harbor control empty vector or SUCLA2 shRNA (Fig. 2F). The loss of SUCLA2 resulted in a dramatic reduction of metastatic potential, whereas treatment with NAC significantly restored the number of colonized tumor nodules reduced in the mice that harbor cells lacking SUCLA2 (Fig. 2 G–I). These results suggest that SUCLA2 contributes to tumor metastasis predominantly by controlling the redox state of disseminated cancer cells.

SUCLA2-Mediated Anoikis Resistance Is Independent of the TCA Cycle.

To explore the mechanism by which SUCLA2 arranges cancer cells to resist anoikis, we first examined whether SUCLA2 remains in the mitochondria with other subunits of succinyl CoA synthetase (SCS) for the Krebs cycle reaction upon cell detachment. Subcellular fractionation and immunofluorescence staining revealed that detached cell culture resulted in translocation of SUCLA2 from mitochondria to cytosol, whereas another A-SCS subunit, SUCLG1, remained in the mitochondria (Fig. 3 A and B). Knockdown of SUCLG1 or SUCLG2, other components that form the SCS complex, did not induce anoikis as in SUCLA2-deficient cells (Fig. 3 C and D and SI Appendix, Fig. S5). SUCLA2 knockdown did not change the intracellular level of succinate, the metabolic product or substrate of SUCLA2 depending on cells undergoing oxidation or reductive carboxylation. Furthermore, treatment with cell-permeable succinate did not rescue the anoikis resistance lost by SUCLA2 knockdown in A549 and MDA-MB231 cells (Fig. 3 E and F). These data suggest that SUCLA2 confers anoikis resistance through a mechanism that is independent of its known enzymatic role in the Krebs cycle.

Fig. 3.

SUCLA2 translocates from mitochondria to the cytosol upon detachment. (A) Western blot analyses demonstrate the cytosolic and mitochondrial localization of SUCLA2 and SUCLG1 in attached and detached cancer cells. α-tubulin and hsp90α were used as control markers for cytosol. Tom40 and CoxIV were used as markers for mitochondria. c: cytosol, m: mitochondria. (B) Immunofluorescence staining shows the localization of SUCLA2 with mitochondrion-selective probe Mitotracker in A549 cells. Quantitative colocalization analysis is shown on the right. Scale bars represent 5 µm. (C and D) Effect of SUCLG1 (C) and SUCLG2 (D) knockdown on anoikis resistance in A549 and MDA-MB231 cells. (E and F) Effect of succinate supplementation on anoikis resistance in cells with SUCLA2 knockdown. A549 (E) and MDA-MB231 (F) cells were cultured on 1% agarose plates in the presence of 3 mM dimethyl-succinate for 48 h. Intracellular succinate levels (Upper) and anoikis rates (Lower) were determined. Representative images are shown from at least three replicates for (A and B Left). Error bars represent ± SD from three replicates for (B–F). P values were obtained by two-tailed Student’s t test for (B) and one-way ANOVA for the rest (ns: not significant; *: 0.01 < P < 0.05; **P < 0.01).

SUCLA2 Interacts with Stress Granule Factors and Facilitates Stress Granule Assembly in Disseminated Cancer Cells.

To better understand the molecular mechanism of the enzyme-independent role of SUCLA2 in redox management, we uncovered the proteins that interact with SUCLA2 in detached cancer cells by mass spectrometry analysis. Endogenous SUCLA2 was enriched from detached A549 cells, and the identities and quantities of the proteins pulled down with SUCLA2 were compared to those of cells with SUCLA2 knockdown. Proteomics profiling revealed that SUCLA2 binds to a series of stress granule proteins, including NUFIP2, ATXN2L, G3BP1, and G3BP2 (Fig. 4A and SI Appendix, Fig. S6). Co-immunoprecipitation analysis further confirmed that SUCLA2 interacts with stress granule components in detached A549 and MDA-MB231 cells (Fig. 4 B and C).

Fig. 4.

SUCLA2 binds to and promotes the formation of stress granules. (A) Identification of SUCLA2 interacting proteins by proteomics analysis. Detached A549 cells with or without SUCLA2 knockdown were applied to SUCLA2 immunoprecipitation followed by LC-MS/MS analysis. Summed intensities of peptides in each sample were compared. (B and C) Endogenous interaction between stress granule proteins and SUCLA2 was determined by SUCLA2 co-immunoprecipitation in detached cancer cells using whole cell lysates (B) or attached or detached cells using cytosolic fraction (C). (D) Stress granules assembled in detached cells with or without SUCLA2 knockdown are shown by endogenous G3BP1, TIA-1, or TIAR immunofluorescence staining. (E) Effect of SUCLA2 knockdown and rescue expression of shRNA-resistant flag-SUCLA2 on stress granules in attached and detached cells. The number of stress granules per cell was quantified using image J software and shown on the top for (D) and (E). Scale bars represent 5 µm for (D) and (E). Error bars represent ± SD from three replicates for (D) and (E). P values were obtained by two-tailed Student’s t test for (D) and one-way ANOVA for (E) (*: 0.01 < P < 0.05; **P < 0.01).

We next examined whether SUCLA2 contributes to the assembly of stress granules. Knockdown of endogenous SUCLA2 in detached A549 and MDA-MB231 cells significantly attenuated stress granules, monitored using three endogenous markers of stress granules, G3BP1, TIA-1, and TIAR (Fig. 4D). Furthermore, overexpression of the shRNA-resistant SUCLA2 restored stress granules decreased by endogenous SUCLA2 knockdown (Fig. 4E). We noticed that SUCLA2 promotes the formation of stress granules that are induced by detachment culture conditions as well as by oxidative stress which is mediated through sodium arsenite or hydrogen peroxide treatment (SI Appendix, Fig. S7).

To further study whether SUCLA2 controls anoikis through stress granules in detached cells, we next determined whether the stress granule itself is involved in anoikis resistance. Repressing stress granule formation by target downregulation of NUFIP2 or G3BP1 mimicked the effect of SUCLA2 knockdown, leading to reduced stress granules and enhanced anoikis induction (Fig. 5 A–D and SI Appendix, Fig. S8 A and B). Moreover, stress granules enhanced by G3BP1 overexpression reversed the increased anoikis mediated by SUCLA2 knockdown (Fig. 5 E and F). To study whether SUCLA2 translocation from the mitochondria to the cytosol is critical for stress granule formation, we generated a mutant SUCLA2 with deletion of the mitochondria targeting sequence (MTS; 1 to 52 amino acids). The SUCLA2 MTS deletion mutant was located mainly in the cytosol, whereas wild-type (WT) SUCLA2 with the intact MTS was predominantly in the mitochondria (Fig. 5G). Cells harboring cytosolic SUCLA2 successfully formed stress granules in all conditions where stress granules were induced by detached culture conditions, sodium arsenite or hydrogen peroxide treatment, or G3BP1 overexpression. In contrast, stress granules were significantly fewer when cells had mitochondrial SUCLA2 (Fig. 5H). We noticed that phosphorylation of eIF2α on serine 51, a prerequisite for stress granule formation, was unaltered by SUCLA2 knockdown (SI Appendix, Fig. S8C). These data suggest that cytosolic SUCLA2 is critical for phospho-eIF2α-independent stress granule formation and anoikis resistance.

Fig. 5.

Cytosolic SUCLA2-mediated stress granules induce anoikis resistance in cancer cells. (A–D) Effect of stress granule disruption on anoikis. Stress granule formation was inhibited by NUFIP2 (A and B) or G3BP1 (C and D) knockdown and its effect on anoikis was assessed by annexin V staining. Disruption of stress granules was confirmed by staining detached cells with stress granule markers G3BP1 or TIA-1. (E and F) Anoikis rates and stress granule formation in detached cells with SUCLA2 knockdown and rescue of stress granules by G3BP1 overexpression. (G and H) Effect of cytosolic localization of SUCLA2 on stress granules. Mitochondria and cytosolic localization of SUCLA2 WT and SUCLA2 mutant lacking mitochondria targeting sequence (MTS) was assessed using fractionated cell lysates (G). 293T cells with SUCLA2 knockdown were transduced with SUCLA2 WT or SUCLA2 del-MTS. Stress granules were induced by 24 h of detachment, 0.5 mM sodium arsenite or 1 mM hydrogen peroxide treatment for 1 h, or GFP-G3BP1 overexpression for 24 h. Stress granules were assessed by G3BP1 staining or GFP. The number of stress granules per cell was quantified using image J software and shown on the top of (B, D, F, and H). Scale bars represent 5 µm for (B, D, F, and H). Error bars represent ± SD from three replicates for (A–F) and (H). P values were obtained by two-tailed Student’s t test for (A–D) and (H), and one-way ANOVA for (E) and (F) (*: 0.01 < P < 0.05; **P < 0.01).

SUCLA2-Mediated Stress Granules Promote Catalase Expression.

Stress granules are known to promote synthesis of select proteins when translation initiation is inhibited by stress responses. We found that loss of SUCLA2 when cells are under detachment stress results in a significant decrease in global protein biosynthesis (Fig. 6A). Based on our finding that SUCLA2 promotes metastasis by handling intracellular ROS, we next determined whether SUCLA2 knockdown attenuates metabolic enzymes associated with redox scavenging systems including catalase, peroxiredoxin 1/2/4 (PRX1/2/4), and superoxide dismutase 2 (SOD2). Intriguingly, protein levels of the cytosolic enzymes involved in scavenging hydrogen peroxide, specifically catalase, markedly decreased, whereas the levels of mitochondrial SOD2 and other metabolic enzymes such as fatty acid synthase (FASN) were unaltered in detached cells lacking SUCLA2 (Fig. 6B). In agreement with this result, the enzyme activity of catalase was significantly attenuated, and its substrate H₂O₂ accumulated when SUCLA2 was target downregulated in detached cancer cells (Fig. 6 C and D). While SUCLA2 knockdown decreased catalase and PRX1/2/4 expression, the rescue expression of SUCLA2 restored the protein levels of catalase and PRX1/2/4 (SI Appendix, Fig. S9A). Transcription, mRNA stability, and protein stability of catalase and PRX1/2/4 were unaltered by SUCLA2 modulation, suggesting that SUCLA2 promotes translation of these ROS scavenging enzymes (SI Appendix, Fig. S9 B–D). Moreover, induction of stress granules by G3BP1 overexpression recovered the decreased catalase expression, but catalase overexpression did not rescue the decreased stress granule formation in SUCLA2 knockdown cells, which indicates that catalase expression is induced by SUCLA2-mediated stress granules (Fig. 6 E and F).

Fig. 6.

SUCLA2-stress granule signaling controls redox in part by mediating catalase expression. (A) Effect of SUCLA2 loss on global protein biosynthesis. Cells with or without SUCLA2 cultured under detachment conditions were applied to fluorescent-based protein synthesis assay. (B) Levels of metabolic enzymes catalase, PRX1/2/4, SOD2, and FASN in detached SUCLA2 knockdown cells. (C and D) Effect of SUCLA2 knockdown on the enzyme activity of catalase (C) and intracellular hydrogen peroxide levels (D) in A549, H460, and MDA-MB231 cells. (E and F) Effect of stress granule rescue on catalase expression (E) and effect of catalase overexpression on stress granule formation (F) in SUCLA2 knockdown cells. Scale bars represent 5 µm for (E) and (F). Error bars represent ± SD from three replicates for (A, and C–F). P values were obtained by two-tailed Student’s t test for (A, C, and D), and one-way ANOVA for (E and F) (ns: not significant; **P < 0.01).

Overexpression of a cytoplasmic H₂O₂ scavenger catalase, but not mito-TEMPO, a mitochondria-targeted antioxidant, reversed the elevated ROS and anoikis induced by SUCLA2 knockdown in detached cancer cells (Fig. 7 A and B). These data suggest that SUCLA2 contributes to anoikis-resistant cell survival through catalase by removing cytosolic H₂O₂ rather than managing mitochondrial ROS. We next validated SUCLA2-catalase signaling in tumor metastasis using the H460 xenograft mouse model. Mice bearing SUCLA2 knockdown H460 cells showed reduced catalase expression and metastatic tumor formation in the liver and kidney compared with the control group. However, overexpression of catalase significantly restored anoikis resistance and tumor metastasis potentials, which were diminished by SUCLA2 knockdown (Fig. 7 C–E). These data suggest that SUCLA2 promotes metastatic potential, at least in part, through catalase expression.

Fig. 7.

SUCLA2 induces anoikis resistance and tumor metastasis in part through catalase expression. (A and B) Effect of flag-catalase overexpression (A) or mito-TEMPO (B) on the cellular ROS or H₂O₂ levels (Left) and anoikis induction (Right) in A549, H460, or MDA-MB231 cells with SUCLA2 knockdown. Ten millimolar mito-TEMPO was supplemented in the culture media, or flag-catalase was transiently overexpressed followed by 48 h of detachment. SUCLA2 knockdown and total catalase level change by flag-catalase expression are shown by immunoblotting analysis. (C) Anoikis rates of H460 cells with SUCLA2 knockdown and catalase overexpression. (D and E) Effect of SUCLA2 knockdown and catalase overexpression on tumor metastasis potential in H460 xenograft mice. NSG mice were injected with H460 cells with SUCLA2 knockdown and catalase overexpression. At the experimental endpoint, 22 d after injection, catalase levels in the liver metastasized tumors (D), the number of metastasized tumors in livers and kidneys (E; Left), and representative liver and kidney images of each group (E; Right) are shown. Bars represent 50 µm for catalase IHC, 5 mm for organ morphology, and 100 µm for H&E staining. Error bars represent ± SD from three replicates for (A–C). n = 5 for (D) and n = 8 mice for (E). P values were obtained by one-way ANOVA for all panels (ns: not significant; *: 0.01 < P < 0.05; **P < 0.01).

Expression of SUCLA2 Positively Correlates with Catalase Protein Level and Metastatic Potential in Cancer Patients.

To demonstrate the clinical significance of our findings, we first examined whether SUCLA2 positively correlates with metastatic progression and with protein expression of catalase in primary human lung cancer and breast cancer tumor samples. A total of 221 cases of paired primary tumors and metastatic lymph nodes were applied to immunohistochemistry (IHC) staining to assess SUCLA2 and catalase expression. The IHC study revealed that the protein levels of both SUCLA2 and catalase were significantly higher in lymph node metastases compared to the matched primary tumor samples (Fig. 8 A and B). Furthermore, there were weak but significant positive correlations between the staining scores of SUCLA2 and catalase in these lung and breast cancer patient tumor specimens (Fig. 8C). In line with our results from the histological study of patient tumors, analyses of a publicly available transcriptomic database of a larger pool of human cancers including lung, breast, bladder, cervical, sarcoma, and ovarian cancers further demonstrated that high levels of SUCLA2 correlated with poor outcome in patients with various types of cancers (Fig. 8D). Our studies support the role of SUCLA2 signaling as a common metastasis driver in cancers, which may be linked with SUCLA2 supporting cancer cells to acquire metastatic potential in part by protein expression of catalase.

Fig. 8.

SUCLA2-catalase signaling is associated with metastatic progression in human cancers. (A and B) The levels of SUCLA2 (A) and catalase (B) in paired primary and metastatic tumors from patients with breast cancer (Left) and lung cancer (Right). Representative IHC images for 0, +1, +2, and +3 scores are shown at the bottom for each cancer type. Scale bars represent 50 µm. (C) The correlation between protein expression levels of SUCLA2 and catalase in tumor samples analyzed in (A and B). (D) Kaplan-Meier survival analysis demonstrating high expression of tumor SUCLA2 mRNA is associated with poor overall survival in patients with diverse types of cancer. RNA-seq data were downloaded from the TCGA and patients were stratified as SUCLA2 high (≥ upper quartile) and SUCLA2 low (≤ lower quartile). P values were determined by Wilcoxon signed-rank test for (A and B), Spearman’s Rho for (C) and log-rank test for (D) (**P < 0.01). (E) Proposed working model of enzyme-independent SUCLA2-mediated anoikis resistance and cancer metastasis. SUCLA2 dissociates from the succinyl-CoA synthetase complex and moves from mitochondria to the cytosol upon cancer cell detachment from the ECM. SUCLA2 binds to and facilitates stress granule assembly that mediates protein translation of antioxidant enzymes, including catalase and peroxiredoxin, and this provides metabolic advantages for cancer cells to survive during detachment and metastasize.

Discussion

The long-standing belief that mitochondrial metabolism was inessential for cancer progression has been disputed by recent human and animal studies, indicating that mitochondria support essential cellular processes in cancer. Although mitochondria have been shown to be associated with metastasis, the precise molecular mechanism by which mitochondria foster cancer cell survival during metastasis is unclear. Our study identified a Krebs cycle enzyme component, SUCLA2, as the critical factor that sustains cell survival during tumor cell dissemination. We show that SUCLA2 uncouples from another complex subunit, SUCLG1, in the Krebs cycle and moves from the mitochondria to the cytosol upon cell detachment, where it binds to and promotes the formation of stress granules that are composed of proteins, including NUFIP2 and G3BP2. The stress granules mediated by SUCLA2 promote redox homeostasis by augmenting the translation of ROS-eliminating enzymes, including catalase and peroxiredoxins, which supports cells resisting anoikis and eventually metastasizing (Fig. 8E). Our study deciphers a unique model for SUCLA2 action in stress granules in the cytoplasm that modulates redox metabolism for survival, which occurs without altering the metabolic function of the SCS complex that is its canonical home.

Protein–protein interactions by RNA providing a scaffold are thought to drive stress granule assembly. For instance, the interaction of G3BP with 40S ribosomal subunits is required for G3BP2-mediated stress granule formation (24). Ataxin-2 interacts with the DEAD/H-box RNA helicase DDX6 in the stress granules, and the level of ataxin-2 is important for the assembly of stress granules and P-bodies (29). The stress granule protein NUFIP2 is known to interact with DDX6 (26). It is plausible that the interaction of SUCLA2 with proteins in the stress granules such as G3BP and NUFIP2 facilitates the assembly of stress granules to perform their function as protein translation machinery. Detailed future structural dynamics analyses on interactions among these molecules are warranted.

Our study indicates that SUCLA2 contributes to redox metabolism and eventually anoikis resistance by enhancing the protein levels of catalase and peroxiredoxins 1, 2, and 4, which are cytosolic enzymes that scavenge hydrogen peroxide. Manganese superoxide dismutase, an antioxidant enzyme that regulates hydrogen peroxide fluxes, is known to mediate anoikis resistance and cancer metastasis in nasopharyngeal cancer (15). Glutamate dehydrogenase 1 is reported to suppress hydrogen peroxide via the product a-ketoglutarate, protecting human mammary epithelial cells against anoikis (30). However, hydrogen peroxide was noted to inhibit a human large cell lung carcinoma cell line from anoikis by inhibiting caveolin-1 degradation, which was performed in a cell line with intact SUCLA2 that controls hydrogen peroxide by making ROS scavenging enzymes (31). These studies together suggest that the homeostasis of hydrogen peroxide is the key factor that determines cell fate in disseminated tumors. On the other hand, antioxidants NAC and L-ergothioneine or overexpression of catalase partially but not fully restored the antianoikis potential in cancer cells lacking SUCLA2. Thus, although cytosolic ROS scavengers may be the predominant downstream factors in SUCLA2 signaling, there may exist additional proteins beyond redox enzymes that are generated by SUCLA2-induced stress granules involved in the antianoikis signaling. Global proteomics analysis using cells with SUCLA2 modulation is further warranted.

Few studies to date have identified molecules as SUCLA2 inhibitors. Vanadate and vanadyl ions attenuate the activity of ATP-dependent SCS, composed of SUCLA2 and SUCLG1, isolated from rat brain mitochondria (32). Tartryl-CoA and propionate have been identified to inhibit GTP-specific SCS (33, 34). However, our study delineates that the role of SUCLA2 in anoikis resistance is enzyme-independent and not associated with the functional SCS complex in the Krebs cycle but with the stress granules in the cytosol. Therefore, molecules that interfere with the interaction between SUCLA2 and the stress granule complex would offer a strategy to target anoikis resistance and metastasis that is mediated by SUCLA2 in human cancers including breast and lung carcinomas.

Methods

Constructs.

Short hairpin RNAs (shRNAs) for SUCLA2, SUCLG1, SUCLG2, and mitochondria metabolic enzymes in pLKO.1-based lentiviral vector were obtained from Horizon Discovery. NUFIP2 shRNA clones were from GeneCopoeia. The sense strand sequences of the shRNAs were CCAGCCAACTTCCTTGATGTT (SUCLA2 #1), CCCAGGAGAGAATACTA CTTT (SUCLA2 #2), TGGAATGGATCACGTAGACAT (SUCLG1 #1), CGGCAACATCTCT ATGTTGAT (SUCLG1 #2). CCTGCTTCATTTACAAGAATT (SUCLG2 #1), AGGTGTCTT CAATAGTGGTTT (SUCLG2 #2), CCAATCATCAAGTCGCTTATC (NUFIP2 #1), GCTAAT ACTCTAACACCCATC (NUFIP2 #2), CCTGTATAGAAGTGGGAAGAT (TK2 #1), CCTGT GTCCAAGTGGAGAAAT (TK2 #2), CGCATTCGGGAGTCTTTCTTT (UQCRQ #1), GTGA TCAGCTACAGCTTGTCA (UQCRQ #2), AGCTCTGTGATGAGCGTGATT (UQCRH #1), and GTGATTCCTCTCGATCACATA (UQCRH #2). SUCLA2, G3BP1, and catalase were flag tagged at the C (SUCLA2) and N (G3BP1 and catalase) terminus, respectively by PCR and cloned into pLHCX-Gateway vector (3). pENTR-del MTS (1 to 52 amino acids)-SUCLA2 and pENTR-SUCLA2 shRNA #1-resistant SUCLA2 constructs were made by site-directed mutagenesis. pENTR223-SUCLA2 (HsCD00622837) was purchased from DNASU. A lentiviral vector for G3BP1-GFP expression (119950) was obtained from Addgene.

Antibodies.

Antibodies against SUCLA2 (A-9/sc-374107, RRID AB_10917237), SUCLG2 (C-1/sc-393756), α-tubulin (B-5-1-2/sc-23948, RRID AB_628410), Tom40 (H-300/sc-11414, RRID AB_793274), PRX (B-11/sc-137222, RRID AB_2168215), and FASN (A-5/sc-55580, RRID AB_2231427) were purchased from Santa Cruz Biotechnology. Anti-SUCLG1 antibody (NBP1-32728, RRID AB_2286802) was obtained from Novus Biologicals. Antibodies against CoxIV (3E11/4850, RRID AB_2085424), G3BP2 (31799, RRID AB_2920540), phospho-eIF2α Ser51 (119A11/3597, RRID AB_390740), eIF2α (D7D3/5324, RRID AB_10692650), catalase for immunoblotting and immunofluorescence staining (D4P7B/12980, RRID AB_2798079), G3BP1 (17798, RRID AB_2884888), and TIAR (D32D3/8509, RRID AB_10839263) were obtained from Cell Signaling Technology. Anti-β-actin antibody (AC-15/A1978, RRID AB_476692) and antiflag antibody (M2/F3165, RRID AB_259529) were purchased from Sigma-Aldrich. Antibody against Hsp90α (ab2928, RRID AB_303423) was from Abcam. Anti-TIA-1 antibody (PA518699, RRID AB_10986154) was from Thermo Fisher Scientific. Anti-NUFIP2 antibody (17752-1-AP, RRID AB_2878433) and anticatalase antibody for immunohistochemistry staining (19792-1-AP, RRID AB_10640433) were from Proteintech. Antirabbit IgG Alexa Fluor 568 (A21069, RRID AB_10563601), antimouse IgG Alexa Fluor 488 (A11001, RRID AB_2534069), and antigoat IgG Alexa Fluor 568 (A11057, RRID AB_142581) were from Invitrogen.

Reagents.

The general oxidative stress indicator CM-H₂DCFDA (C6827) was from Invitrogen. Mito-TEMPO (SML0737), N-acetyl-L-cysteine (NAC; A7250), and dimethyl-succinate (W239607) were obtained from Sigma-Aldrich. Q5 Site-Directed Mutagenesis Kit (E0554) was from New England BioLabs. Annexin V staining kit (556547) was from BD Biosciences. Mitochondria isolation kit for cultured cells (89874) was obtained from Pierce. TransIT-LT1 (MIR2304) was from MirusBio. ROS-Glo H₂O₂ Assay (G8820) and Caspase-Glo 3/7 Assay Systems (G8091) were obtained from Promega. EZClick Global Protein Synthesis Assay (K459/ab239725) and Catalase Activity Colorimetric/Fluorometric Assay (K773/ab83464) were purchased from Biovision. Succinate Assay Kit (ab204718) was obtained from Abcam. Primers for qRT-PCR were obtained from Integrated DNA Technologies. PROTEOSTAT Thermal Shift Stability Assay Kit (ENZ-51027-K100) was from Enzo Life Sciences. All other chemicals not specified were obtained from Sigma-Aldrich.

Cell Culture.

A549, H460, MDA-MB231, and MDA-MB468 cell lines were purchased from American Type Culture Collection and Lenti-X 293T cell line was obtained from Takara Bio, USA. A549 and H460 cells were cultured in RPMI1640 media supplemented with 10% fetal bovine serum (FBS). MDA-MB231, MDA-MB468, and Lenti-X 293T cells were cultured in Dulbecco modified Eagle media supplemented with 10% FBS. Lentivirus shRNA production, viral infection, and stable knockdown cell selections were performed using psPAX2 and pMD2.G packaging system as described (35).

RNAi Screen.

The customized shRNA library targeting 119 human mitochondria metabolic genes was generated using the pLKO.1 lentiviral vectors designed by The RNAi Consortium (TRC). Lentivirus from each shRNA clone was produced and pooled to infect 1 × 10⁵ A549 cells for 48 h in six well-plates to target each mitochondrial gene. Cells were then split and reseeded on 1% agarose coated six well-plates or into four replicates in 96 well-plates and half of the replicates were treated with 2 µg/mL puromycin. After 48-h incubation, detachment-induced cell death was assessed by annexin V staining and cell proliferation rate was determined using CellTiter Glo^® Luminescent Cell Viability Assay. Candidate genes that resulted in <15% cell death by knockdown alone or had >25% shRNA virus infection efficacy by puromycin selection were selected for analysis.

Anoikis and Cell Proliferation Assays.

Anoikis was induced by culturing cells on 1% agarose-coated plates for 48 h unless specified. Apoptotic cell death was determined by FITC-conjugated annexin V/propidium iodide staining or Caspase Glo 3/7 activity assay based on the manufacturer’s instructions. Cell viability was determined at indicated time points using CellTiter-Glo^® Luminescent Cell Viability Assay. Numbers of viable cells were estimated using cell number standard curves.

Bioenergetic Flux Analysis of Nonadherent Cells.

To detect energetic flux using suspended cells, cells were cultured on 1% agarose-coated plates 24 h before the experiment. To immobilize the suspended cells on Seahorse XF24 Cell Culture Microplates, each well of the plates was coated with 50 μL of Cell-Tak™ Cell and Tissue Adhesive (22.4 µg/mL in 0.1 M sodium carbonate, pH 8.0). The suspended cells were then transferred into the plates at a density of 1 × 10⁵ cells/well and centrifuged at 200 × g for 5 min. OCR and ECAR were acquired in cells with or without SUCLA2 knockdown by sequential treatment with 500 nM oligomycin, 500 nM FCCP, and 1 µM antimycin A/rotenone and analyzed using Agilent Seahorse XFe24 flux analyzer (Agilent Technologies).

Other Metabolic Assays.

The ATP present in cells was determined using ATP luminescent somatic cell assay kit (FLASC, Sigma-Aldrich). Cells were cultured under hypoxic conditions using sealed hypoxia incubation chambers (StemCell Technologies). General ROS and hydrogen peroxide levels in cells were quantified using chloromethyl derivative of H₂DCFDA (Invitrogen) and ROS-Glo™ H₂O₂ Assay (Promega), respectively. Levels of intracellular succinate, catalase activity, and global protein synthesis were monitored in detached cancer cells with or without SUCLA2 loss using commercial assay kits indicated in the Reagents section according to the manufacturers’ protocols. In brief, fixed and permeabilized cells were treated with O-propargyl-puromycin to stop translation and synthesized peptides were quantified by reaction with the fluorescent azide at Excitation 494/Emission 521 nm for protein synthesis assay. To detect activity of catalase in cells, unconverted substrate H₂O₂ that reacts with OxiRed probe was measured at 570 nm. Quantitative analysis of 116 hydrophilic and ionic metabolites involved in central energy metabolism in A549 cells with SUCLA2 knockdown was performed by Human Metabolome Technologies C-SCOPE using capillary electrophoresis and mass spectrometry (CE-MS) platform.

Immunofluorescence Staining.

First, 5 × 10⁴ cells cultured in suspension were washed, diluted in 100 mL of PBS, and placed on sample chambers made with Cytoslide and Cytofunnel. Then, the cells were centrifuged using Shandon Cytospin 4 (Thermo Scientific) at 800 rpm for 5 min and applied to immunofluorescence staining. In brief, cells on glass slides were stained with MitoTracker for 30 min at 37 °C, fixed in PHEMO buffer (25 mM HEPES, 68 mM PIPES, 0.5% Triton X-100, 15 mM ethylene glycol-bis(beta-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA), 3 mM MgCl₂, 3.7% formaldehyde, 0.05% glutaraldehyde), blocked in 10% goat serum, incubated with antibodies against SUCLA2, G3BP1, TIA-1, TIAR, or catalase followed by antimouse, rabbit, or goat IgG antibodies conjugated with Alexa 488 or 568. For adherent cells, cells cultured on the glass slides were stained with MitoTracker for 30 min at 37°C and applied to immunofluorescence staining as described for suspension cells. Cells were applied with antifade mountant with DAPI and imaged on an SP8 confocal microscope. Stress granules were counted by particle analysis using ImageJ software, and catalase expression was analyzed by the quantification of the stained area compared to the control group. All quantification was performed for each group in three randomly chosen sections.

Mass Spectrometry-Based Proteomics.

SUCLA2 interacting proteins were identified by SUCLA2 co-immunoprecipitation followed by microcapillary LC-MS/MS analysis. In brief, 3 mg of cell lysates obtained from detached A549 cells with or without SUCLA2 knockdown were pre-cleared with Protein G Sepharose 4 Fast Flow (Millipore Sigma). Precleared lysates were incubated with 10 μg of monoclonal anti-SUCLA2 antibody followed by protein G Sepharose incubation. The bead bound proteins were extracted and denatured in 0.12 M Tris (pH 6.8), 3.3% SDS, 10% glycerol, and 3.1% dithiothreitol, followed by precipitation using ProteoExtract Protein Precipitation Kit (Calbiochem). The precipitates were applied for trypsin digestion and mass spectrometry analysis at the Taplin Mass Spectrometry Facility at Harvard Medical School.

mRNA and Protein Stability Assays.

For mRNA stability assay, detached cells were incubated with 4 mg/mL of actinomycin D for the indicated time. Quantitative RT-PCR was conducted with cDNA Reverse Transcription Kit (Applied Biosystems) and SYBR Green Supermix (Bio-Rad). For protein stability assay, catalase or PRX were immunoprecipitated from cell lysates precleared with Protein G Sepharose beads. Unbound proteins were washed, and the beads-bound catalase and PRX were subjected to PROTEOSTAT Thermal Shift Assay. Fluorescence levels were measured with excitation/emission of 500/550 nm at increasing temperatures and analyzed based on the manufacturer’s recommendations.

Animal Study.

In vivo xenograft mouse model study was performed according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) of Emory University. For Fig. 2 F–I, athymic nude mice (Hsd:Athymic Nude Foxn1^nu, female, 5-wk–old, Envigo) were intravenously injected with 1 × 10⁶ A549 cells with or without SUCLA2 shRNA through the tail vein. The mice harboring SUCLA2 knockdown cells were randomly divided into two groups, and antioxidant was administered to one group by supplementing the drinking water with 10 mg/mL NAC from 1 d after xenograft for 9 wk. For Fig. 7 D and E, NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ, female, 5-wk-old, Jackson Laboratory) were intravenously injected with 2 × 10⁶ H460 cells with SUCLA2 knockdown or catalase overexpression. After 22 d post-injection, the livers and the kidneys bearing tumor nodules were extracted. Formalin-fixed paraffin-embedded tissues were stained with hematoxylin-eosin (H&E) and tumor nodules were counted.

Immunohistochemistry (IHC).

Usage of de-identified human tissue specimens was approved by the Institutional Review Board (IRB) of Emory University. Clinical samples were obtained with informed consent under the protocols that are approved by the Health Insurance Portability and Accountability Act. Formalin fixed-paraffin-embedded breast and lung cancer tissue samples composed of primary and matched lymph node metastasized tumors were obtained from US Biomax (BR1005b, BR20837a, LC817, and LC814). Immunohistochemistry staining was performed as previously described using anti-SUCLA2 antibody (1:1,000) and anticatalase antibody (1:500) (36). The staining intensities were scored in the range of 0 to +3.

Publicly Available Database Analysis.

Expression of SUCLA2 mRNA (RNAseq RSEM) and clinical data for each cancer type were downloaded from cBioPortal (http://cbioportal.org/) or firebrowse (http://firebrowse.org/) and Kaplan–Meier plots were generated in R environment using survival packages (https://cran.r-project.org). Patients were dichotomized into two groups which are upper and lower quartiles by SUCLA2 expression.

Statistics.

Statistical analyses and graphic presentations were conducted using GraphPad Prism version 9. Data presented are from a representative experiment of multiple biological replicates. Error bars indicate mean ± SEM for Fig. 2 A and G and mean ± SD for all other graphs. Statistical analyses of significance were based on two-tailed Student's t test for Figs.1 C and E, 3B, 4D, 5 A–D, and H, and 6 A, C, and D, chi-square test for Fig. 2H, Wilcoxon signed-rank test for Fig. 8 A and B, Spearman’s Rho for Fig. 8C, log-rank test for Fig. 8D, and one-way or two-way ANOVA for all other data figures.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.