Copper depletion modulates mitochondrial oxidative phosphorylation to impair triple negative breast cancer metastasis

Abstract

Copper serves as a co-factor for a host of metalloenzymes that contribute to malignant progression. The orally bioavailable copper chelating agent tetrathiomolybdate (TM) has been associated with a significant survival benefit in high-risk triple negative breast cancer (TNBC) patients. Despite these promising data, the mechanisms by which copper depletion impacts metastasis are poorly understood and this remains a major barrier to advancing TM to a randomized phase II trial. Here, using two independent TNBC models, we report a discrete subpopulation of highly metastatic SOX2/OCT4+ cells within primary tumors that exhibit elevated intracellular copper levels and a marked sensitivity to TM. Global proteomic and metabolomic profiling identifies TM-mediated inactivation of Complex IV as the primary metabolic defect in the SOX2/OCT4+ cell population. We also identify AMPK/mTORC1 energy sensor as an important downstream pathway and show that AMPK inhibition rescues TM-mediated loss of invasion. Furthermore, loss of the mitochondria-specific copper chaperone, COX17, restricts copper deficiency to mitochondria and phenocopies TM-mediated alterations. These findings identify a copper-metabolism-metastasis axis with potential to enrich patient populations in next-generation therapeutic trials.

Trial registration: ClinicalTrials.gov NCT00195091.

Fig. 1. Copper depletion impacts invasion and metastasis.

a Patient survival with CTR1 or SLC31A1 expression using METABRIC dataset. The graph depicts survival for patients with low expression (first quartile) and high expression (fourth quartile) of the SLC31A1 gene (n = 425 patients/group). Significance was calculated using Log-rank (Mantel–Cox) test. b Distribution of SLC31A1 expression with tumor stage. Significance was calculated using one-way ANOVA with Tukey post-test for comparing multiple groups. N = 433 for Stage 0-I patients, n = 702 for Stage II patients, n = 111 for Stage III–IV patients. p-value (0-I vs. II) < 0.0001, p-value (0-I vs. III–IV) = 0.0011. c Distribution of SLC31A1 mRNA expression in breast cancer subtypes. Significance was calculated using one-way ANOVA with Tukey post-test for comparing multiple groups. n = 290 for ER−/PR− patients, n = 603 for ER+/HER2− High proliferative patients, n = 619 for ER+/HER2− Low proliferative patients, n = 188 for HER2+ patients. p-value < 0.0001. d Event-free survival (EFS) for Stage 4 NED breast cancer patients from TM clinical (Chan et al. 2017). 1 cycle = 4 weeks. e (Top panel) Development of EO771.ML1 model. (Lower panel) Timeline for metastasis assay with or without TM using EO771.ML1 model. f Ceruloplasmin levels in control vs. TM-treated mice in ML1 model, 2 weeks after TM administration (n = 6/group). Analysis was performed by unpaired two-sided t-test. Center lines of box plots denote median values, top whiskers denote maxima and bottom whiskers minima. p-value = 0.0224. g Primary tumor growth in control and TM cohorts in ML1 model, measured with calipers (n = 10 in the control group, n = 9 in TM group). Analysis was performed by unpaired two-sided t-test. p-value = 0.8365. h BLI measurements of lungs in ML1 model showing metastatic burden (n = 4/group). Analysis was performed by unpaired two-sided t-test. p-value = 0.0015. i Measurement of intracellular copper (Cu) in LM2 cells using graphite furnace atomic absorption spectrometry (y-axis refers to Cu content measured at 324.75 nm wavelength). Significance was calculated using one-way ANOVA with Tukey post-test for comparing multiple groups. p-value (0 vs. 0.1 nM) = 0.0003, p-value (rest) < 0.0001. j MTT assay for viability with increasing TM concentrations, 72 h after treatment of LM2 cells (n = 4/group). Significance was calculated using One-way ANOVA with Tukey post-test for comparing multiple groups. p-value (0 vs. 1 nM) = 0.0162, p-value (0.2 vs. 1 nM) = 0.0041. k Matrigel coated transwell assay showing invasion of LM2 cells treated with TM (0.5 µM) for a total of 72 h (48-h pretreatment, followed by 24 h treatment on Matrigel-coated transwells) and rescue with addition of copper (0.5 µM CuCl₂) for the last 24 h in the transwell (along with TM for that cohort). n = 3 replicates/group (7 fields of view/sample). Significance was calculated using one-way ANOVA with Tukey post-test for comparing multiple groups. p-value (Control vs. TM + Cu) = 0.0115. Representative data of at least two independent experiments are depicted. Results are expressed as mean ± SD. *p < 0.05, **p < 0.01, ****p < 0.0001.

Fig. 2. Copper depletion targets SOX2/OCT4+ metastatic cells.

a (Top panel) schematic of lentivirus-based SOX2/OCT4-GFP promoter-reporter system. (Bottom panel) Flow-sorted GFP+ and GFP− cells from bulk ML1 cells with CMV control or SOX2/OCT4 reporter system. b Impact of TM (0.5 µM) on the invasion of SOX2/OCT4+ GFP+ and SOX2/OCT4− GFP− cells across a matrix coated transwell in LM2 cells. Significance was calculated using Two-way ANOVA (n = 3 replicates/group, 5 fields of view/sample). c Intracellular copper (Cu) levels in SOX2/OCT4+ GFP+ vs. SOX2/OCT4− GFP− cells measured by atomic absorption spectrometry (y-axis refers to Cu content measured at 324.75 nm wavelength) in ML1 cells. Analysis was performed by unpaired two-sided t-test. p-value = 0.0040. d Tandem mass tags (TMT) isobaric label multiplexed quantitative proteomic analysis of vehicle and TM (0.5 µM) treated ML1 SOX2/OCT4+ GFP+ reporter cells. Proteins with absolute log2 Fold Change (FC) > 1, p < 0.05 and FDR < 0.05 are shown in green and labeled, n = 5/group. Significance was calculated using multiple t-tests, p values were adjusted for multiple comparisons. e Tandem mass tags (TMT) isobaric label multiplexed quantitative proteomic analysis of vehicle and TM (0.5 µM) treated ML1 SOX2/OCT4− GFP− cells. Proteins with log2 > 1 or < −1 and p < 0.05 are shown, n = 5/group. Significance was calculated using multiple t-tests, p values were adjusted for multiple comparisons. f and g Gene ontology (GO) analysis of downregulated proteins (p < 0.05, average log2 < −0.6) after 48 h TM treatment in ML1 SOX2/OCT4+ GFP+ (f) and ML1 SOX2/OCT4− GFP− g cells (log2 fold change from −2.6 to −0.7). log2 fold change is log2 FC of TM 48 h/Control. Count refers to how many genes/proteins showed up in that GO term. GO analysis was performed using the clusterProfiler R package, p values were adjusted using Benjamin–Hochberg method. dsCop destabilized copepod. Representative data of two independent experiments are depicted. Results are expressed as mean ± SD. **p < 0.01 ****p < 0.0001.

Fig. 3. TM treatment impacts mitochondrial Complex IV.

a Western blot showing respiratory complexes in LM2 cells following 72-h TM treatment (0, 0.1, 0.2, 0.5 µM). Complexes I–V are referred to as CI (mol. wt. ~19 kDa), CII (mol. wt. ~30 kDa), CIII (mol. wt. ~48 kDa), CIV (mol. wt. ~40 kDa), and CV (mol.wt. ~55 kDa), respectively. NDUFA4 is a subunit of Complex IV. b qPCR analysis showing transcript levels of human Complex IV subunit 4I1 (COX4I1) in control and TM (0.5 µM) treated LM2 cells, treated with TM for 72 h. Analysis was performed by unpaired two-sided t-test. p-value = 0.079. c Protein expression of cytochrome c with and without TM treatment (0.5 µM) (n = 4/group). d Human cytochrome c oxidase activity (Abcam) kinetically measured over 2 h after 72 h of treatment with TM (0.5 µM) in LM2 cells. The graph represents the difference in Complex IV activity at 110th minute during measurement. Analysis was performed by unpaired two-sided t-test (n = 4/group). p-value = 0.0063. Center lines of box plots denote median values, top whiskers denote maxima and bottom whiskers minima. e Mitochondrial DNA content in LM2 cells as determined by qPCR (p = 0.060). Analysis was performed by unpaired two-sided t-test (n = 4/group). f–h MitoTracker deep red analysis in control and TM (0.5 µM) treated LM2 (f), ML1 (g) and MDA-MB-468 (h) cells, treated for 72 h (n = 3/group). i Structure of cristae in the inner mitochondrial membrane as determined by transmission electron microscopy (×50,000), red arrows indicate normal/folded and perturbed cristae morphology in LM2 cells. Scale bar, 500 nm. j Quantification of perturbed cristae per 100 units of mitochondrial area in pixels (n = 30 mitochondria in controls, n = 36 mitochondria for TM group). Analysis was performed using Mann–Whitney two-sided test. Representative data of two independent experiments are depicted. Results are expressed as mean ± SD. **p < 0.01 ****p < 0.0001.

Fig. 4. Copper depletion disrupts mitochondrial respiration in SOX2/OCT4+ metastatic cells.

a Impact of TM (0.5 µM) on oxygen consumption rate of sorted GFP+ and GFP− ML1 cells as measured by Seahorse assay (n = 8/group). Quantification of basal respiration (b) and ATP-linked respiration (c) in GFP+ vs. GFP- ML1 cells with and without TM (0.5 µM). Significance was calculated using one-way ANOVA with Tukey post-test for comparing multiple groups. Center lines of box plots denote median values, top whiskers denote maxima and bottom whiskers minima. d Ratio of mitochondrial to glycolytic ATP production rate in GFP+ vs. GFP− ML1 cells, measured by Seahorse ATP production rate assay (n = 8/group). Significance was calculated using one-way ANOVA with Tukey post-test for comparing multiple groups. Representative data of two independent experiments are depicted. e Representative IF microscopy images from control and TM-treated (0.7 mg/day) primary LM2 tumors in the mammary gland (n = 5/group). Scale bar, 20 µM. f Flow cytometry analysis showing the percentage of GFP+ cells in primary tumors in the LM2 model (n = 12/group) of total mCherry+ tumor cells. Analysis was performed by unpaired two-sided t-test. p-value = 0.0111. g Disseminated tumor cells in early lungs of control and TM (0.7 mg/day) treated LM2 cohorts. Analysis was performed by unpaired two-sided t-test. p-value = 0.0150. Representative data of two independent experiments are depicted. Results are expressed as mean ± SD. Results are expressed as mean ± SD. **p < 0.01, ****p < 0.0001.

Fig. 5. COX17 knockdown yields TM phenotypes.

a qPCR showing shRNA-mediated COX17 knockdown in LM2 cells. Analysis was performed by unpaired t-test (n = 5/group). b Oxygen consumption in shCOX17#1 vs. SCR-sh (n = 8 for SCR-sh, n = 6 for shCOX17#1). c, d Basal and ATP-linked respiration in shCOX17#1 vs. SCR-sh cells using mitostress assay. Analysis was performed by unpaired two-sided t-test. Center lines of box plots denote median values, top whiskers denote maxima and bottom whiskers minima. e Invasion assay through matrix-coated transwell. Analysis was performed by unpaired two-sided t-test (n = 3/group, 5 fields of view/sample). f Representative images of metastatic nodules in lungs using shCOX17#1 and SCR-sh LM2 in vivo model (n = 7/group). g Quantification of lung metastases from panel f. Analysis was performed by unpaired two-sided t-test. p-value = 0.0236. h Oxygen consumption rate in ML1 SCR-sh vs. shCOX20 cells (n = 8/group). Representative data of two independent experiments are depicted. Results are expressed as mean ± SD. *p < 0.05, ****p < 0.0001.

Fig. 6. Copper depletion activates AMPK.

a Glucose consumption and lactate production measured by nutrient analysis after 72 h of TM (0.5 µM) treatment in LM2 cells. Supernatant was replaced 24-h before collection in both groups. Significance was calculated using Two-way ANOVA (n = 3/group). b ECAR for LM2 cells after 72 h of TM (0.5 µM) treatment (n = 8/group). c Levels of nucleotide precursors measured by LC–MS after 72 h of TM (0.5 µM) treatment in LM2 cells. Significance was calculated using Two-way ANOVA, followed by Bonferroni’s multiple comparisons test (n = 5/group). d and e Mitochondrial ATP production rate with TM (0.5 µM) treatment in LM2 (d) and ML1 (e) cells. Significance was calculated using Two-way ANOVA, followed by Tukey’s multiple comparisons test (n = 8/group). f Western blot of p-AMPK (Thr172), p-ACC (Ser79), and Complex IV (CIV) in control and TM (0.5 µM) treated LM2 cells. Complex IV is referred to as CIV. g Mitochondrial ATP production rate in LM2 SCR-sh and shCOX17#1, using Seahorse ATP production rate assay. Significance was calculated using Two-way ANOVA, followed by Tukey’s multiple comparisons test (n = 3 for SCR-sh, n = 8 for shCOX17#1). h Western blot of p-AMPK in shCOX17 and SCR-sh LM2 cells. Complex IV is referred to as CIV. i mTOR pathways (p-Raptor Ser792, p-p70S6K Thr389) downstream of phospho-AMPK in LM2 cells. Representative data of two independent experiments are depicted. Results are expressed as mean ± SD. **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 7. AMPK is a mediator of downstream TM phenotypes.

a, b Impact of pharmacologic activation of AMPK with AICAR (1.25 mM) and A-769662 (12.5 µM) on the invasion of LM2 cells. Representative data from two independent experiments. Analysis was performed by unpaired two-sided t-test (n = 3/group, 5 fields of view/sample). c Western blot showing activation of AMPK by AICAR (1.25 mM) in LM2 cells. d Western blot showing activation of p-ACC by A-769662 (12.5 µM) in LM2 cells. e Control and 0.5 µM TM-treated cells (after 24 h) were transfected with siRNA against a control (C) sequence or AMPK (A). AMPK expression by western blot in LM2 cells. f Rescue of TM-mediated loss of invasion with siAMPK in LM2 cells. Significance was calculated using Two-way ANOVA, followed by Tukey’s multiple comparisons test (n = 3/group, 5 images/sample). Representative data of two independent experiments are depicted. Results are expressed as mean ± SD. **p < 0.01, ****p < 0.0001.