Decorin evokes reversible mitochondrial depolarization in carcinoma and vascular endothelial cells

Abstract

Decorin, a small leucine-rich proteoglycan with multiple biological functions, is known to evoke autophagy and mitophagy in both endothelial and cancer cells. Here, we investigated the effects of soluble decorin on mitochondrial homeostasis using live cell imaging and ex vivo angiogenic assays. We discovered that decorin triggers mitochondrial depolarization in triple-negative breast carcinoma, HeLa, and endothelial cells. This bioactivity was mediated by the protein core in a time- and dose-dependent manner and was specific for decorin insofar as biglycan, the closest homolog, failed to trigger depolarization. Mechanistically, we found that the bioactivity of decorin to promote depolarization required the MET receptor and its tyrosine kinase. Moreover, two mitochondrial interacting proteins, mitostatin and mitofusin 2, were essential for downstream decorin effects. Finally, we found that decorin relied on the canonical mitochondrial permeability transition pore to trigger tumor cell mitochondrial depolarization. Collectively, our study implicates decorin as a soluble outside-in regulator of mitochondrial dynamics.

INTRODUCTION

Soluble decorin was the first small leucine-rich proteoglycan (SLRP) to be shown to inhibit cell growth in Chinese hamster ovary cells (1), primarily by binding to and blocking transforming growth factor beta (TGFβ) (2). However, we reasoned that TGFβ could not be the main reason for growth inhibition in cancer cells as most of the primary and metastatic carcinomas do not respond to nor are dependent on TGFβ. Thus, we investigated the role of decorin in cancer growth and found that de novo decorin gene expression suppressed the malignant phenotype in human colon cancer cells (3). In addition, we found that ectopic expression of decorin protein core caused a generalized growth suppression in neoplastic cells of various histogenetic origins (4), and this bioactivity required endogenous p21 (4, 5), an inhibitor of cyclin-dependent kinases. As decorin is a secreted proteoglycan, we reasoned that cell-surface receptors should be involved in mediating its outside-in signaling cascade (6). We discovered that decorin bound with relatively high affinity to epidermal growth factor receptor (EGFR) in squamous carcinoma cells overexpressing this receptor (7, 8). In general, soluble decorin or its protein core in a monomeric form (9) binds to a narrow region of EGFR partially overlapping with but distinct from the epidermal growth factor (EGF)-binding epitope (10). This interaction leads to a transient EGFR activation that includes intracellular mobilization of Ca²⁺ (11), followed by a protracted downregulation of the receptor (12) that is internalized via caveosomes and ultimately degraded in lysosomes (13). The discovery of decorin/EGFR interaction has opened this field of research and has led to the discovery of multiple receptor tyrosine kinase interacting with decorin including ErbB2 (14), Met (15, 16), insulin-like growth factor receptor 1 (IGF-IR) (17, 18), vascular endothelial growth factor receptor 2 (VEGFR2) (19, 20) and platelet-derived growth factor receptor (PDGFR) (21). Indeed, the protein interacting network of decorin is quite large, now encompassing over 75 proteins, including growth factors, receptor tyrosine kinases (RTKs), and immune receptors (22). There is strong genetic evidence supporting a role for decorin as an oncosuppressive proteoglycan. For example, cooperative action of germ-line targeting mutations of decorin and p53 accelerates lymphoma tumorigenesis (23). In line with these observations, a genetic background lacking decorin promotes hepatic carcinogenesis (24), favors spontaneous colon tumor formation (25–27), and promotes epithelial-mesenchymal transition and colon cancer metastasis (28). Some of these activities could be pharmacologically targeted (29). Notably, decorin has been implicated in inflammation and fibrosis (30, 31) and in the pathophysiology of several organs and tissues including the heart (32, 33), bone (34), tendon and ligaments (35, 36), cervical remodeling (37), ocular pathology (38), osteoarthritis and cartilage homeostasis (39, 40), and matrix remodeling (41).

Another important function of decorin is its ability to evoke autophagy in endothelial cells (20, 42, 43). Indeed, decorin itself is an autophagy-inducible proteoglycan and is required for proper in vivo autophagy (44). Decorin-evoked signaling through VEGFR2 leads to activation of the energy sensor kinase AMP-activated protein kinase (AMPK) (45), and also to the autophagic degradation of vascular endothelial growth factor A (VEGFA) (46), a mechanism additional to those previously found in support of decorin angiostatic activity (47–51). Notably, biglycan, the closest homolog of decorin, does not cause autophagy in endothelial cells (20); however, it evokes autophagy in macrophages via a novel CD44/Toll-like receptor 4 (TLR4) signaling axis (52, 53). Moreover, biglycan-triggered TLR2 and TLR4 signaling exacerbates the pathophysiology of ischemic acute renal injury (54).

As a mechanistic corollary to decorin to evoked endothelial cell autophagy (42, 51, 55), we found that decorin evokes tumor cell mitochondrial autophagy (mitophagy) (56, 57). This process relies on decorin binding to the Met receptor and initiating a complex signaling cascade involving peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α)-mediated stabilization of mitostatin mRNA and protein (43). Mitostatin has been characterized as a putative tumor suppressor (58, 59) and is required for decorin-evoked tumor cell mitophagy (56, 60). Indeed, loss of mitostatin prevented decorin from inducing mitophagy including loss of mitochondrial fragmentation, mtDNA loss, and voltage-dependent anion channel 1 (VDAC1) clearance. As the most direct result of future mitophagy, mitochondrial depolarization is critical (61–63). Thus, we sought to mechanistically decipher the role of decorin in coordinating this early event as an initiating signal for downstream mitophagy in cancer.

Here, we report on a novel signaling pathway that mechanistically connects decorin-dependent RTK signaling to mitochondrial depolarization in various cancer and vascular models. These findings redefine the stage at which mitostatin functions in the mitophagic cascade and broaden our understanding on the extent of the effect decorin has on tumor cell mitophagy.

METHODS

Cells, Chemicals, and General Reagents

The human triple-negative breast cancer cell line MDA-MB-231 and cervical cancer cell line HeLa were obtained from American Type Cell Culture. Cells were grown at 37°C in a 5% CO₂ atmosphere in Dulbecco’s modified Eagle’s medium (DMEM) containing 4.5 g/L glucose, l-glutamine, and sodium pyruvate from Life Technologies and supplemented with 5% fetal bovine serum (FBS) from ThermoFisher Scientific and 100 U/mL penicillin/streptomycin from Life Technologies. Primary human umbilical vein endothelial cells (HUVECs) were obtained from Lifeline Cell Technology, grown in basal media supplemented with the VascuLife EnGS LifeFactors Kit (from Lifeline Cell Technology), and used within the first five passages. HUVECs were also grown at 37°C in a 5% CO₂ atmosphere. Dimethyl sulfoxide (DMSO) was obtained from ThermoFisher Scientific. Rabbit monoclonal antibodies against mitofusin 2 (64) and GAPDH (65) were obtained from Cell Signaling Technologies. The rabbit polyclonal antibody against Met (66) and α-Tubulin (67) were obtained from Santa Cruz Biotechnology. The rabbit anti-mitostatin affinity-purified antibody was described elsewhere (68). The validation for the custom mitostatin antibody comes from our siRNA experiments (please consult the mitostatin immunoblots in Figs. 4A and 6, A and D). Briefly, relative to siScramble controls, use of a siRNA specific for mitostatin results in the depletion of mitostatin immunoreactivity, thereby validating the mitostatin antibody. Moreover, this same biochemical check is valid for the commercially available mitofusin 2 antibody. Please consult the immunoblots in Figs. 5A and 6, A and D. Recombinant human hepatocyte growth factor (HGF), NIM811, AG1478, and SU11274 were purchased from Sigma. The mitochondrial dyes JC-1, JC-10, TMRE, and TMRM were purchased from Abcam. The mitochondrial dyes used here are cell-permeant, positively charged dyes that readily accumulate in active mitochondria due to their relative negative charge. Depolarized or inactive mitochondria have decreased membrane potential and fail to sequester TMRE and TMRM, resulting in a lower intensity signal. For JC-1 and JC-10, this results in a color change from red (J-aggregates sequestered within the mitochondria) to green (J-monomers) following depolarization. The plasmids encoding myc-Met and 3×FLAG-mitostatin were purchased from Origene. DMSO was used as vehicle where appropriate for all experiments. All primary antibodies were used at 1:1,000 diluted in 1% BSA-TBST except for GAPDH, which was used at 1:10,000. Secondary antibodies for chemiluminescence were used at 1:5,000 in the same buffer as above. The SuperSignal West Pico-enhanced chemiluminescence substrate was purchased from ThermoFisher Scientific.

Figure 4.

Mitostatin, a key downstream effector of decorin, is required for depolarization. A: representative immunoblots confirming the efficacy of mitostatin depletion in 231 (left) or HeLa (right) via siRNA with densitometry values below the loading control. Amounts of siRNA for siScr and siMitostatin were kept constant throughout (100 pM). B: gallery of live cell images depicting 231 (top series of images) or HeLa (bottom series of images) cells following transient knockdown of mitostatin, challenged with decorin (6 h, 100 nM), and then stained with JC-10. C and D: quantification of the JC-10 signal in 231 (C) or HeLa (D) from data in B. E and F: quantification of the resultant TMRE (E) or TMRM (F) signal in 231 cells as per the same experimental conditions as in B. G and H: quantification of the resultant TMRE (G) or TMRM (H) signal in HeLa cells as per the same experimental conditions as in B. I and J: mitostatin rescue experiment done in 231 (I) or HeLa (J) and stained with TMRE as per the conditions listed. Vehicle (pcDNA3.1) and active plasmid (FLAG-mitostatin) were used at equimolar amounts (4 µg). For immunoblots (A), GAPDH served as the loading control. PBS served as vehicle for all experiments. All data presented are expressed as n = 4 (unless otherwise noted) independent biological replicates. Data are expressed as means ± SE. Statistics calculated via one-way ANOVA (***P < 0.001).

Figure 6.

Mitostatin and mitofusin 2 are in the same pathway and required for depolarization. A: representative immunoblots confirming the efficacy of mitostatin and/or mitofusin 2 depletion via siRNA in 231 cells in combination with decorin (6 h, 100 nM) as noted. B: gallery of live cell images depicting JC-10 staining in 231 cells under a variety of experimental conditions as labeled. C: quantification of the JC-10 signal in B. D: representative immunoblots confirming the efficacy of mitostatin and/or mitofusin 2 depletion via siRNA in HeLa cells in combination with decorin (6 h, 100 nM) as noted. E: gallery of live cell images depicting JC-10 staining in HeLa cells under a variety of experimental conditions as indicated. F: quantification of the JC-10 signal in E. For immunoblots (A and D), GAPDH served as the loading control. PBS served as vehicle for all experiments. All data presented are expressed as n = 4 independent biological replicates. Data are expressed as means ± SE. Statistics calculated via one-way ANOVA (***P < 0.001).

Transient DNA Expression and RNAi-Mediated Silencing

We transiently transfected the appropriate cell line with empty vector (pcDNA3.1) or plasmids encoding myc-Met or 3xFLAG-mitostatin using Lipofectamine 2000 (Life Technologies) in OptiMEM reduced serum media (Gibco). Expression was verified by immunoblot using target-specific antibodies. A description of DNA transfection has been described (46). Cells were transiently transfected using Lipofectamine RNAiMAX (Life Technologies) mixed with siRNA against Met, mitostatin, and/or mitofusin 2 that were purchased from Santa Cruz Biotechnology. The siRNA used represents a validated cocktail of 3–5 targeting oligonucleotides for each gene of interest. Scrambled siRNA (sc-37007, Santa Cruz Biotechnology) served as a control for all siRNA experiments presented herein. The full protocol is described in Ref. 69.

Mitochondrial Membrane Potential

At least four individual assays were performed in the appropriate cell line and experimental condition using the mitochondrial dyes JC-1, JC-10, TMRE, or TMRM. Cells were treated as per the experimental conditions described herein. The TMRE and TMRM signals were normalized to FCCP (carbonylcyanide 4-triflouromethoxy phenylhydrazone) for 10 min following treatment to completely collapse the mitochondrial membrane potential. Each chamber or well of cells was incubated with the appropriate dye for 20–30 min. The cells were washed three times with PBS and imaged live using a Leica DM5500B microscope. All the images were procured using the same exposure, gain, and intensity.

Mitochondrial Membrane Potential Assay in Ex Vivo Aortic Rings

All of the animal experiments were performed according to the Guide for Care and Use of Laboratory Animals and the Institutional Animal Care and Use Committee of Thomas Jefferson University.

We initially coated 35-mm plastic bottom plates with a 3D collagen type I, rat tail (Corning) solution. After polymerization of the collagen, we selected male or female C57BL/6 mice between 1 and 3 mo of age and anesthetized the mice with isoflurane. We dissected the mice by removing organs to expose the aorta. We carefully removed the aorta from the spine with a scalpel under a stereoscopic microscope. After trimming excess fat, we then sectioned the aorta into rings that were ∼1 mm in width and placed these rings into the 35-mm dishes, 4 rings per dish. We then covered the rings with a top layer of collagen and VascuLife Basal Medium (Lifeline). Minimal sprouting should be apparent after 2 days, and after 4 days, sprouting should be both dense and widespread.

Mitochondrial depolarization was visualized via a Mitochondrial Membrane Potential Assay Kit (Abcam), which used TMRE (tetramethylrhodamine, ethyl ester) to label active mitochondria with a red fluorescence. TMRE staining was done for 30 min at a concentration of 150 nM (the volume of the collagen was included in the total volume, ∼2.5 mL). For the treatment groups, FCCP was used at 10 μM with a 15-min pretreatment to TMRE staining. Decorin treatment was done at 1 μM with overnight pretreatment to TMRE staining. This higher concentration of decorin was used to adjust for the loss of decorin bound to collagen. Four images of each ring were taken at ×10 magnification and 200 ms exposure time. A Nikon Eclipse Ts2 inverted microscope was used for image capture. Fluorescent intensity was measured using ImageJ, in which the background was subtracted with the Subtract background function with a rolling ball radius of 700 pixels. The mean gray value of sprouting was measured by splitting the channel by color and using the red color channel. Then we selected Analyze → Set Measurements → Mean gray value as well as Limit to threshold. By adjusting the threshold via Image → Adjust → Threshold to highlight the sprouting without picking up any background, we measured the fluorescence intensity of areas with dense sprouting. The FCCP negative control was used to subtract from vehicle- and decorin-treated values as a background fluorescence.

Statistical Analysis

Immunoblots were quantified by scanning densitometry using Scion Image software (NIH). Graphs were generated using Sigma Stat 3.10. Experiments with three or more comparison groups were subjected to one-way ANOVA followed by a Bonferroni post hoc test using the Systat Package of SigmaPlot 13.0. Differences were considered significant at two-sided P < 0.05.

RESULTS

Decorin Triggers Mitochondrial Depolarization in Tumor Cells

To investigate in-depth role of soluble decorin on mitochondrial homeostasis, we used live cell imaging with fluorescence microscopy and TMRE in triple-negative MDA-MB-231 breast carcinoma cells (231 hereafter). We discovered that decorin triggered depolarization in a dose- (Fig. 1A) and time-dependent manner (Fig. 1C). Starting with as little as 1 nM of decorin (Fig. 1, A and B), mitochondria began depolarizing as demonstrated by a visible loss of the TMRE signal. Mitochondrial depolarization progressed with increasing concentrations of decorin until 100 nM where decorin achieved maximal depolarization and was stable up to 1 µM (Fig. 1, A and B). Therefore, unless otherwise noted, we used decorin at 100 nM. Next, we evaluated the temporal effects of decorin and found that as early as 1 h, decorin already significantly decreased the TMRE signal, indicating depolarization (Fig. 1, C and D). Depolarization became more pronounced over time, became maximal at 6 h, and was sustained for up to 24 h (Fig. 1, C and D). Based on these data, unless otherwise noted, we exposed the breast carcinoma cells to soluble decorin for 6 h.

Figure 1.

Decorin triggers depolarization in tumor cells. A: gallery of live cell images depicting 231 cells treated with increasing concentrations of decorin, as noted, for a fixed time (6 h) and stained with tetramethyl rhodamine ethyl ester (TMRE; 150 nM). B: quantification of the normalized TMRE signal as in A. C: gallery of live cell images depicting 231 cells treated over time with decorin, as noted, at a fixed concentration of decorin (100 nM) and stained with TMRE (150 nM). D: quantification of the normalized TMRE signal as seen in C. E: identical experiment as shown in A, but stained with JC-10 (1 µM). F: quantification of the JC-10 signal as seen in E. G: identical experiment as shown in C, but stained with JC-10. H: quantification of the JC-10 signal as seen in G. I and J: gallery of images depicting the depolarization recovery experiments done in 231 (I) or HeLa (J) and stained with JC-10. K and L: quantification of the JC-10 signal from the recovery experiments for 231 (K) or HeLa (L). PBS served as vehicle for all experiments. All data presented are reflective of several (n = 4) independent biological replicates. Data are expressed as means ± SE. Statistics calculated via one-way ANOVA (***P < 0.001; **P < 0.01).

To corroborate these findings, we performed additional experiments using the cationic ratiometric dye, JC-1 (70). We corroborated the temporal dynamics of decorin to induce mitochondrial depolarization over time with this dye (Supplemental Fig. S1, A and B). We then switched to JC-10 for further validation of our model and kinetics of decorin-evoked ΔΨm. JC-10 is a much more soluble and more robust variant of the JC-1 ratiometric dye to measure mitochondrial permeability (71, 72). Using this more vigorous approach, we discovered very comparable effects. Decorin evoked ΔΨm in 231 cells with as little as 1 nM (Fig. 1, E and F) in as quickly as 2 h (Fig. 1, G and H). In both sets of experiments, depolarization was progressive and was maintained up to 1 µM and 24 h, respectively. By using HeLa cells and JC-10, we found that decorin triggered depolarization at 1 nM (Supplemental Fig. S1, C and D) and within 2 h (Supplemental Fig. S1, E and F). Similarly, these effects were progressive and maintained to the endpoints of 1 µM and 24 h (Supplemental Fig. S1, C–F), respectively.

It is important to note that we used highly purified decorin proteoglycan expressed in 293-EBNA cells as a His₆-protein using a Celligen Plus bioreactor and affinity purified on a Ni-NTA column (73). The purity of our decorin preparations was assessed via SDS-PAGE. As colloidal Coomassie has a detection threshold of ∼5-ng protein, and as no additional bands were detectable at 5 µg (Supplemental Fig. S2A), we conclude that our protein-core preparations are >99.9% pure (<5 ng/5,000 ng).

We next evaluated whether decorin required the covalently attached dermatan/chondroitin sulfate glycosaminoglycan chain (74) to trigger depolarization in tumor cells. We found that in either 231 (Supplemental Fig. S2, B–E) or HeLa cells (Supplemental Fig. S2, F–I), in the presence of TMRE or JC-10, the GAG chain was dispensable for mitochondrial depolarization. Quantification of the TMRE or JC-10 signals for either 231 or HeLa revealed no statistically significant differences in the magnitude of depolarization between decorin proteoglycan or decorin protein core. These data are congruent with past and recent findings where the GAG chain is dispensable for most of the biological effects of decorin (46, 75).

To establish specificity of action, we next evaluated the effects of biglycan on triggering depolarization. Biglycan is a class I SLRP and closest relative of decorin with >65% homology at the protein level (76). Biglycan was synthesized in a manner similar to decorin and was found to be contaminant-free via colloidal Coomassie staining (Supplemental Fig. S2J). Interestingly, biglycan had no discernable effect on depolarization in either 231 (Supplemental Fig. S2K) or HeLa (Supplemental Fig. S2L) when assessed via JC-10 (Supplemental Fig. S2M).

To test whether decorin effects were species-specific, we used mouse E0771 cells. These are triple-negative, basal-like tumor cells (77, 78) with a molecular profile similar to that of 231 breast carcinoma cells. Importantly, the E0711 cells responded to exogenous human decorin by triggering depolarization via a significant loss of TMRE signal, and this effect was similar to FCCP (Supplemental Fig. S2N). Collectively, our results demonstrate that soluble decorin specifically triggers depolarization in both human and murine breast carcinoma cells, in a time- and dose-dependent manner. Moreover, this bioactivity is specific for this Class I SLRP as biglycan failed to trigger depolarization.

Decorin-Dependent Mitochondrial Depolarization is Reversible in Carcinoma Cells

It is widely accepted that decorin binds to cell-surface RTKs and triggers internalization with subsequent degradation of the decorin/RTK complex within lysosomes (43, 79, 80). This process is the cornerstone of the mechanism of action for tumorigenic suppression via decorin for protracted biological effects (20, 45, 46, 56, 81). Empirical experimental evidence clearly demonstrates decorin-mediated internalization of RTK complexes following high-affinity binding for EGFR (13) and Met (16). However, it remains unknown whether removal of soluble decorin would lead to a reversal of the downstream biological effects. Therefore, we treated 231 or HeLa cells with decorin (100 nM) for 4 h to trigger depolarization (prewash, Fig. 1, I and J) as determined by JC-10 staining (Fig. 1, K and L). We then washed the cells and monitored depolarization over time in the absence of exogenous decorin. Strikingly, within 30 min of removing the decorin-laden media (e.g., postwash), mitochondrial membrane potential began to recover in both 231 (Fig. 1, I and K) and HeLa cells (Fig. 1, J and L). Within 1 h and by 2 h, mitochondrial membrane potential fully recovered and stabilized in 231 and HeLa cells (Fig. 1, I–L). Therefore, these data indicate that the continued presence of decorin is required to sustain protracted depolarization. Furthermore, these data began to substantiate that once internalized, the putative decorin/RTK complex is not capable of intracellular signaling, consistent with a wholly degradative role for decorin to silence oncogenic signals and curb tumorigenesis.

Decorin Requires the Met Receptor Kinase to Trigger Mitochondrial Depolarization in Carcinoma Cells

Given that decorin predominantly uses RTKs (60) to transduce its biological signals (51, 82), we assessed the requirement of Met or EGFR. As measured by JC-10 staining, decorin triggered depolarization in 231 (Fig. 2, B and I) and HeLa cells (Fig. 2, K and R) vis-à-vis their respective vehicles (Fig. 2, A and J). Treatment with a SU11274, a tyrosine kinase inhibitor (TKi) of the Met tyrosine kinase (83), alone had no discernable effect on depolarization at basal conditions (Fig. 2, C, L, I, and R); however, in combination with decorin SU11274 significantly blocked (Fig. 2, D and M) decorin-triggered depolarization (Fig. 2, I and R). Next, using an EGFR-specific inhibitor, AG1478 (84, 85), we discovered that blocking EGFR under basal conditions had no effect on depolarization (Fig. 2, E, N, I, and R) akin to the Met Tki. Similarly, combination treatment of AG1478 and decorin results in a significant abrogation of mitochondrial depolarization in 231 Fig. 2, F and I) and HeLa cells (Fig. 2, O and R).

Figure 2.

Decorin requires the Met receptor kinase to trigger depolarization in carcinoma cells. A–H: gallery of live cell images of 231 cells treated in various combinations of decorin (100 nM), SU11274 (1 µM), or AG1478 (1 µM) and stained with JC-10. I: quantification of the JC-10 signal from A–H. J–Q: identical experiment as in A–H performed in HeLa cells. R: quantification of the JC-10 signal from J–Q. S and T: identical experiments as performed above with combinations of decorin and TKi but stained with TMRE in 231 (S) or HeLa cells (T). DMSO served as vehicle to account for SU11274 and AG1478 for all experiments. All data presented are reflective of several (n = 4) independent biological replicates. Data presented are expressed as means ± SE. Statistics calculated via one-way ANOVA (***P < 0.001).

Next, we sought to rule out any possible contributions from other RTKs that decorin might use to trigger depolarization, as decorin uses Met and EGFR to achieve the majority of antitumorigenic outcomes. Therefore, we used both inhibitors either in combination with or without decorin. We found that treatment with both inhibitors alone had no effect on basal mitochondrial membrane potential in 231 (Fig. 2, G and I) or HeLa (Fig. 2, P and R). Adding both inhibitors, in the presence of decorin, blocked depolarization in a magnitude akin to single-agent alone (Fig. 2, H, I, Q, and R). Furthermore, there was no observed compensation or redundant receptor for decorin to evoke ΔΨm in 231 or HeLa cells.

We repeated these experiments in 231 cells with JC-1 and obtained identical results. Decorin-mediated depolarization was significantly blocked with single-agent Tki with no significant change in magnitude or additional redundancy found with dual TKi (Supplemental Fig. S3, A and B).

We used TMRE to validate our findings with the ratiometric dyes (e.g., JC-10, JC-1). Importantly, we found highly similar patterns using TMRE. Blocking either Met or EGFR individually significantly blocked decorin-mediated depolarization in 231 (Fig. 2S) and HeLa (Fig. 2T). Lastly, dual agent in the presence of decorin recapitulated the concept that there was no further change in the magnitude of depolarization when both receptors were blocked and no additional redundancy conveyed by a different receptor complex (Fig. 2, S and T).

Collectively, our findings indicate that decorin used Met and/or EGFR to evoke an intracellular signaling cascade that leads to a reversible but profound mitochondrial depolarization.

The Met Receptor is Required for Decorin-Mediated Tumor Cell Mitochondrial Depolarization

We decided to sharpen our studies on only the Met receptor as it represents the primary receptor (15) for a multitude of antioncogenic activities imbued by decorin (86) against cancer. To this end, we used siRNA and significantly depleted Met from 231 (approximately eightfold, Fig. 3A) and HeLa cells (∼3.6 folds, Fig. 3B). Using JC-1, we discovered that physical depletion of Met completely blocked the ability of decorin to induce depolarization in both cell types (Fig. 3, C–E).

Figure 3.

The Met receptor is required for decorin-dependent tumor cell depolarization. Representative immunoblots confirming the efficacy of Met depletion in 231 (A) or HeLa (B) cells via siRNA with densitometry values below the loading control. Amounts of siRNA for siScr and siMet were kept constant throughout (100 pM). C: gallery of live cell images depicting 231 or HeLa cells following transient knockdown of the Met receptor, challenged with decorin (6 h, 100 nM), and then stained with either JC-1 (top) or JC-10 (bottom). D and E: quantification of the JC-1 signal in 231 (D) or HeLa (E) from data in C. F and G: quantification of the JC-10 signal in 231 (F) or HeLa (G) from data in C. H and I: identical Met knockdown experiment as in C, but stained with TMRE in 231 (H) or HeLa (I). J and K: quantification of the resultant TMRE signal in 231 (J) or HeLa (K) after being stimulated with a combination, as noted, of decorin (100 nM) or HGF (50 ng/mL) for 6 h. L and M: Met rescue experiment done in 231 (L) or HeLa (M) and stained with TMRE as per the conditions listed. Vehicle (pcDNA3.1) and active plasmid (myc-Met) were used at equimolar amounts (4 µg). For immunoblots A and B, α-tubulin served as the loading control. PBS served as vehicle for all experiments. All data presented are reflective of several (n = 4, unless otherwise noted) independent biological replicates. Data are expressed as means ± SE. Statistics calculated via one-way ANOVA (***P < 0.001).

Next, we complemented the JC-1 data with a more robust mitochondria dye (JC-10) and found similar patterns for 231 and HeLa (Fig. 3, C, F and G). Finally, we used TMRE as an additional verification of decorin-driven depolarization. We found that loss of decorin and staining with TMRE revealed a similar mechanism whereby depolarization was abrogated following Met silencing in 231 (Fig. 3H) and HeLa cells (Fig. 3I). Interestingly, similar to TKi alone, depletion of Met itself had no observable effects on depolarization in neither 231 nor HeLa cells.

Next, we evaluated whether decorin could interfere with Met signaling conveyed by hepatocyte growth factor (HGF), the endogenous agonist of Met (87), widely considered to be pro-oncogenic and proangiogenic (88–94). Mechanistically, decorin displaces HGF from binding the Met ectodomain (15) resulting in decreased breast cancer metastasis (95) and decreased angiogenesis (48). Moreover, prooncogenic growth factors can increase mitochondrial polarization (96–98). Assaying with TMRE, we found that HGF was indeed capable of mitochondrial hyperpolarization in 231 (Fig. 3J) and HeLa (Fig. 3K) cells. As expected, decorin-induced depolarization, and strikingly, was able to significantly blunt HGF-evoked mitochondrial hyperpolarization (Fig. 3, J and K) in both cell types. These data reinforce the role of decorin as a critical antagonist of HGF in cancer.

We wanted to further establish the sufficiency of Met-dependent signaling to induce mitochondrial depolarization in response to decorin. Therefore, we performed rescue experiments by reexpressing siRNA-resistant MET cDNA (which has an N-terminal myc-tag) in 231 and HeLa cells followed by TMRE staining. We reproduced our results insofar as that loss of Met via siRNA prevented decorin-induced mitochondrial depolarization (Fig. 3, L and M) in the presence of empty vector (pcDNA3.1). Notably, supertransfecting 231 or HeLa with myc-Met alone did not have an adverse effect on basal mitochondria membrane potential, as the resting magnitude is comparable with experiments reported herein.

We observed two important effects after reintroducing myc-Met into cells that have been depleted of the receptor. First, overexpressing Met in cells that only received scramble siRNA enhanced the effect of decorin-mediated depolarization in 231 and HeLa (compare condition 2 with condition 5, Fig. 3, L and M). Second, restoring decorin in Met-depleted cells significantly restored the effect of decorin to induce depolarization (compare condition 4 with condition 8, Fig. 3, L and M). Akin to TKi and siRNA, overexpression of Met had negligible effects on basal mitochondrial membrane potential (Fig. 3, L and M).

Collectively, we have demonstrated a mechanistic dependency of Met to convey signal required for decorin-mediated depolarization. Moreover, Met is both necessary and sufficient for this process and complements the pharmacological approach underscoring the importance of the decorin/Met interaction in cancer cells.

Mitostatin, a Key Downstream Effector of Decorin, is Required for Mitochondrial Depolarization

We next turned our attention to potential downstream effectors that could be critical for decorin-mediated depolarization. A key modulator of decorin bioactivity is mitostatin. We reasoned that mitostatin may have a functional role in the early stages of mitophagy (63, 99) and potentially in decorin-mediated depolarization. To this end, we used siRNA and significantly depleted mitostatin in both 231 and HeLa (Fig. 4A) cells. Using JC-10, we discovered that loss of mitostatin severely abrogated the ability of decorin to evoke ΔΨm in either cell type (P < 0.001 for both, Fig. 4, B–D). These data were recapitulated using JC-1 in 231 whereupon mitostatin silencing significantly prevented decorin-mediated depolarization (Supplemental Fig. S4, A and B).

We then sought to further authenticate our findings. Using 231 and HeLa and different mitochondrial membrane potential dyes [TMRE and TMRM (100)], we found a strikingly similar pattern in decorin-mediated depolarization following mitostatin depletion. In 231 cells loaded with either TMRE (Fig. 4E) or TMRM (Fig. 4F), loss of mitostatin significantly impaired decorin-mediated depolarization. In HeLa cells, we found an identical pattern where cells loaded with TMRE (Fig. 4G) or TMRM (Fig. 4H) following mitostatin silencing significantly abrogated decorin-mediated depolarization. These sets of data corroborate the JC-10 and JC-1 staining above and illustrate a critical role of mitostatin in conveying information for decorin-mediated depolarization. Interestingly, in all cases independent of staining method or cell type, loss of mitostatin alone did not perturb basal mitochondrial membrane potential (Fig. 4B). These data indicate that mitostatin is an active molecule in transducing signals given the appropriate stimulus (e.g., decorin) in tumor cells.

As we did with Met (cf. Fig. 3, L and M), we wanted to establish the sufficiency of mitostatin-dependent depolarization in the presence or absence of decorin. Thus, we executed rescue experiments by reexpressing siRNA-resistant mitostatin cDNA (which harbors an N-terminal 3xFLAG tag) in 231 and HeLa cells followed by TMRE staining. We were able to reproduce our depolarization results to the degree that loss of mitostatin prevented depolarization in cells supertransfected with equal amounts of empty plasmid (231, Fig. 4I; HeLa, Fig. 4J).

Surprisingly, supertransfecting 231 or HeLa cells with 3xFLAG-mitostatin alone was sufficient to evoke significant mitochondrial depolarization (fifth and seventh condition, Fig. 4, I and J). Depolarization occurred without the use of any external stimulant (e.g., decorin) and seemingly independent of whether endogenous mitostatin was present. Concurrent treatment with decorin and 3x-FLAG-mitostatin had a marginally combined effect on depolarization (sixth and eighth condition, Fig. 4, I and J). These data substantiate and confirm that mitostatin alone is capable of mediating depolarization in tumor cells.

Collectively, these data provide a mechanistic basis for the role of mitostatin in acting at the earliest step in the mitophagic process by evoking mitochondrial depolarization. Moreover, decorin relies on mitostatin for depolarization to occur.

Mitostatin and Mitofusin 2 Are in the Same Pathway and Required for Mitochondrial Depolarization

A key binding partner for mitostatin was identified as mitofusin 2 (101) as a regulator of mitochondria-associated membranes. Mitofusin 2 is a critical mediator of overall mitochondrial health (61, 101), ER/mitochondrial calcium homeostasis (102, 103), and mitophagy (104–106). The fact that mitostatin binds mitofusin 2 (a finding we recapitulated in 231 and HeLa cells, data not shown) provided a rationale to test two hypotheses: 1) does decorin-mediated depolarization also require mitofusin 2? and 2) do mitofusin 2 and mitostatin reside in the same pathway to promote depolarization?

To begin testing these possibilities, we depleted mitofusin 2 from 231 and HeLa (Fig. 5A) cells via siRNA. Following authentication that mitofusin 2 was substantially depleted, we loaded 231 and HeLa with JC-10. We discovered that loss of mitofusin 2 phenocopied the loss of mitostatin insofar as blocking the ability of decorin to mediate depolarization in 231 and HeLa cells (Fig. 5B) (Fig. 5, D and F). We also performed this experiment with JC-1 and found that it operates identically (Supplemental Fig. S4C). Loss of mitofusin 2 blocked decorin when stained with JC-1 (Supplemental Fig. S4D). Next, we loaded 231 or HeLa cells with TMRE and found a very similar response that loss of mitofusin 2 rendered decorin unable to mediate depolarization (Fig. 5, C and D). These results mechanistically link decorin to a critical and well-established mitochondrial mediator.

Figure 5.

Decorin requires mitofusin 2 for depolarization in tumor cells. A: representative immunoblots confirming the efficacy of mitofusin 2 depletion in 231 (left) or HeLa (right) via siRNA with densitometry values below the loading control. Amounts of siRNA for siScr and siMitostatin were kept constant throughout (100 pM). B: gallery of live cell images depicting 231 (top series of images) or HeLa (bottom series of images) cells following transient knockdown of mitostatin, challenged with decorin (6 h, 100 nM) and then stained with JC-10. C and D: quantification of the JC-10 signal in 231 (C) or HeLa (D) from data in B. E and F: quantification of the resultant TMRE signal in 231 (E) or HeLa (F) cells as per the same experimental conditions as in B. For immunoblots (A), GAPDH served as the loading control. PBS served as vehicle for all experiments. All data presented are expressed as n = 4 (unless otherwise noted) independent biological replicates. Data are expressed as means ± SE. Statistics calculated via one-way ANOVA.

Next, we tested whether mitostatin and mitofusin, two physically interacting proteins (107) (and unpublished results), could be in the same signaling pathway that leads to depolarization. We approached this by performing single- or double-agent siRNA against mitostatin, mitofusin 2, or both in the presence or absence of decorin. First, we documented proper and significant knockdown of both proteins in 231 cells (Fig. 6A and Supplemental Fig. S5, A and B). Importantly, we found that decorin increased mitostatin levels in 231 cells (Fig. 6A and Supplemental Fig. S5A) in line with our previous findings (56, 58). We found that decorin significantly increased mitofusin 2 levels (Fig. 6A and Supplemental Fig. S5B). Using JC-10, we independently and fully reproduced our earlier findings (cf. Figs. 4 and 5) that loss of either mitostatin or mitofusin 2 alone blocked decorin-mediated depolarization (Fig. 6, B and C). Further loss of single agent alone without decorin had no adverse effect on basal depolarization (Fig. 6, B and C). Interestingly, siRNA-mediated depletion of both mitostatin and mitofusin 2 did not perturb basal mitochondrial membrane potential (Fig. 6, B and C). Notably, dual depletion of mitostatin and mitofusin 2 in the presence of decorin did not result in either an attenuated blockade or a more severe blockade of decorin activity (Fig. 6, B and C). Indeed, the magnitude of preventing decorin from driving depolarization was visually and statistically similar to either single agent knockdown alone (Fig. 6, B and C).

We then fully recapitulated these critical set of experiments in HeLa cells. We determined that single and dual knockdown of mitostatin and/or mitofusin 2 was robust and significant (Fig. 6D and Supplemental Fig. S5, C and D). We found that akin to 231 cells, decorin increases mitostatin and mitofusin 2 levels (Fig. 6D and Supplemental Fig. S5 C and D). Importantly, in a manner that mimics the situation in 231 cells, we independently reproduced that loss of either mitostatin or mitofusin 2 singly abrogated the prodepolarization effect of decorin (Fig. 6, E and F). Moreover, dual knockdown of mitostatin and mitofusin 2 did not overtly affect basal mitochondrial membrane potential, which is in line with single knockdown as we have demonstrated in Fig. 6, E and F. Dual depletion of mitostatin and mitofusin 2 did not alter the effects of decorin (Fig. 6, E and F). Collectively, these data implicate mitofusin 2 as a key mediator of decorin-evoked ΔΨm in two different cancer cell lines. Mechanistically, it appears that mitostatin and mitofusin 2 are functionally within the same signaling pathway as there were neither additive nor synergistic effects when single versus dual knockdowns were performed. Because we used only one concentration of decorin (100 nM) and one time point (6 h) for these studies, as empirically determined (cf. Fig. 1), it is possible that differential combinations of concentrations and time points may promote a different response.

Decorin Relies on the Canonical mPTP to Evoke Tumor Cell Mitochondrial Depolarization

Canonical mitochondrial depolarization proceeds via a specialized pore widely known as the mitochondrial permeability transition pore (mPTP) (108). The formation of the highly conductive and long-lasting mPTP is often the result of a damaging stimulus or tightly regulated signal such as an overload of Ca²⁺ (108) or reactive oxygen species (ROS) (109). Indeed, long-term opening of the mPTP leads to oxidative phosphorylation (OXPHOS) decoupling, mitochondrial injury, and cellular death (109). Although the molecular composition of the mPTP is still hotly contested (110, 111), there remains critical physiological relevance for its function (112).

One vital molecular component that aids in the opening of the mPTP is cyclophilin D (CypD) (110). Inhibition of cyclophilin D with cyclosporine A (CsA) (113, 114) provides a useful tool to investigate the molecular function of the mPTP (115). We, therefore, investigated whether decorin uses the canonical mPTP to mediate depolarization in tumor cells.

To this end, we treated 231 cells with decorin in the presence or absence of CsA. As before, decorin potently mediated depolarization when stained with JC-10 relative to vehicle-treated controls (Fig. 7, A and B) as quantified (Fig. 7J). The magnitude of depolarization was similar to that of FCCP (Fig. 7, C and J). Strikingly, we found that application of CsA in combination with decorin significantly blocked decorin-mediated depolarization (Fig. 7, E and J). Interestingly, CsA alone did not significantly affect mitochondrial membrane polarization (Fig. 7, D and J). Consistent, however, with the mechanism of action for an uncoupler such as FCCP (116), CsA did not block the ability of FCCP to induce mitochondrial depolarization (Fig. 7, F and J). These data are important as it provides a point of inflection between the RTK (and mPTP)-dependent mechanism used by decorin and that of a potent protonophore.

Figure 7.

Decorin relies on the canonical mPTP to trigger tumor cell depolarization. A–I: gallery of live cell images depicting 231 cells treated in a variety of combinations with decorin (100 nM), carbonylcyanide 4-triflouromethoxy phenylhydrazone (FCCP; 1 µM), cyclosporin A (5 µM), or NIM811 (5 µM) and stained with JC-10. J: quantification of the JC-10 signal as in (A–I). K–S: gallery of live cell images depicting an identical experiment as in (A–I) carried out in HeLa cells. T: quantification of the JC-10 signal as in K–S. U and V: quantification of the resultant tetramethyl rhodamine ethyl ester (TMRE) signal under identical conditions (e.g., A–I; K–S) performed in 231 (U) or HeLa (V). Dimethyl sulfoxide (DMSO) served as vehicle for all experiments to account for cyclosporin A and NIM811. All data presented are expressed as n = 4 independent biological replicates. Data are expressed as means ± SE. Statistics calculated via one-way ANOVA (***P < 0.001).

Next, we used a more specific CypD inhibitor with less potential off-target effects, NIM811 (117). NIM811 functions similarly to CypD in which it inhibits the transition of the mPTP but does not bind to calcineurin, making it nonimmunosupressive (117). Akin to CsA, NIM811 alone had no effects on basal mitochondrial membrane polarization (Fig. 7, G and J), but did significantly blunt the ability of decorin to trigger depolarization (Fig. 7, H and J). Moreover, NIM811 did not have any effect on FCCP-mediated uncoupling and resulting depolarization (Fig. 7, I and J).

To provide additional mechanistic robustness to this mechanism, we recapitulated these experiments in HeLa cells and stained with JC-10. As shown, decorin evoked a strong depolarization relative to vehicle treatments (Fig. 7, K, L, and T) that was comparable in responsiveness to FCCP (Fig. 7, M and T). Similar to 231, CsA had no substantive effect on resting mitochondrial membrane potential (Fig. 7, N and T), but potently blocked decorin-evoked ΔΨm (Fig. 7, O and T). Furthermore, CsA was functionally unable to block FCCP (Fig. 7, P and T). We then applied NIM811 and treated it in combination with decorin or FCCP in HeLa cells. We observed similar results. NIM811 had no effect on basal mitochondrial membrane potential (Fig. 7, Q and T), but completely abrogated decorin-mediated depolarization (Fig. 7, R and T). As for CsA, NIM811 did not impair FCCP from evoking potent depolarization in HeLa cells (Fig. 7, S and T).

We next verified these findings with TMRE in 231 (Fig. 7U) and HeLa cells (Fig. 7V) insofar as that CsA and NIM811 potently block decorin-mediated depolarization in both cell types without compromising FCCP-mediated depolarization. These independently obtained data corroborate our JC-10 staining.

Collectively, these data substantiate a mechanistic role for the mPTP in significantly and specifically mediating decorin-mediated depolarization. This mechanism is different from that of a well-established uncoupler (FCCP) in triggering depolarization. Therefore, an undefined signaling pathway involving communication from a cell-surface RTK exists that serves as the link between decorin and mitochondrial depolarization.

Decorin Triggers Mitochondrial Depolarization in Vascular Endothelial Cells and in Ex Vivo Aortic Ring Assays

In an effort to validate our profound mitochondrial depolarization findings, we used an ex vivo approach vis-à-vis aortic ring assays that have recently been improved upon in our laboratory (118). The aortic rings are carefully harvested and embedded in a 3D collagen type I matrix.

Therefore, TMRE staining in aortic rings isolated from C57BL/6 mice revealed that an 18-h decorin treatment (1 µM) significantly induced depolarization in the sprouts relative to vehicle conditions (Fig. 8, A and B). Interestingly decorin used at half this concentration (0.5 µM) had no discernable effect on sprout mitochondrial depolarization (Fig. 8, A and B).

Figure 8.

Decorin induces depolarization in ex vivo aortic ring assays in 3D collagen I and in vascular endothelial cells. A: serialized aortic rings treated overnight with the indicated concentrations of decorin followed by tetramethyl rhodamine ethyl ester (TMRE) staining (150 nM). Live bright field images were acquired to demonstrate the presence of sprouts from the aortic rings. All images were captured with a 200-ms exposure at ×10 magnification. Bar ∼ 90 µm. B: quantification of the TMRE signal from A. C: live cell images of human umbilical vein endothelial cells (HUVECs) treated with decorin (6 h, 200 nM) and stained with JC-1. D: quantification of the resultant TMRE signal in HUVEC following decorin or CCCP (1 µM). PBS served as control vehicle for all experiments. All data presented are reflective of several (n = 4, unless otherwise indicated here or in the text) independent biological replicates. Data are expressed as means ± SE. Statistics calculated via one-way ANOVA (***P < 0.001).

Considering the cellular similarity between aortic endothelial cells (comprising the aortic ring sprouts) and HUVECs, which are a target of decorin (20, 46), we assayed depolarization in this system. Akin to the robust depolarization we discovered in tumor cells and aortic rings, decorin triggered significant depolarization in HUVECs as per JC-1 staining (Fig. 8C) and TMRE (Fig. 8D).

Taken together, these data provide the first evidence that decorin can evoke mitochondrial depolarization in an ex vivo aortic ring system as well as in HUVECs. These data expand on the already exciting mechanism of RTK-mediated mitochondrial depolarization seen in tumor cells.

DISCUSSION

One of the earliest, if not the earliest, step in the mitophagic cascade is the rapid and widespread loss of the mitochondrial membrane potential (57, 63). We previously identified that decorin evokes mitochondrial depolarization in 231 cells immediately before mitochondrial network fragmentation (42, 56). However, despite the early evidence that decorin triggers mitochondrial depolarization, the mechanism, molecular factors, general applicability, and biological function were unknown.

Therefore, in this study, we delved deeper into the mechanism of action that governs decorin-dependent depolarization in multiple tumor cell types and increased scientific robustness surrounding this biological process. We have demonstrated that there is a dose- and time-dependent pattern in which decorin promotes extensive depolarization of the tumor mitochondrial network. The depolarization evoked by decorin is also conserved among various types of human and mouse tumor cells including MDA-231, HeLa, and E0771. This effect is also seen in normal, genomically stable HUVECs, and ex vivo murine aortic ring assays. Moreover, using a combination of up to four different mitochondrial-specific dyes (TMRE, TMRE, JC-1, and JC-10), we validated that decorin is capable of mediated depolarization under a variety of detection methods. Interestingly, biglycan, the closest relative of decorin, was unable to initiate depolarization in our tumor models, most likely owing to the lack of a specific receptor.

Starting from the outside-in signaling paradigm (51), decorin takes on average ∼1–2 h to trigger mitochondrial depolarization. This is consistent with decorin-transducing signals via a cell-surface receptor tyrosine kinase. In this study, we found that both EGFR and Met are equally capable to communicate the signal for depolarization. This lends additional credence to decorin integrating the signal among several receptors for biological effects (43). Considering that Met represents the primary biological receptor for the observed antiangiogenic and antioncogenic activities of decorin (15), we pursued more in-depth studies with the decorin/Met interaction under the construct of depolarization. We discovered the physical presence of Met is not only necessary but also sufficient for decorin bioactivity.

As we found that mitostatin was crucial for executing decorin-evoked mitophagy (56), we redefined the role of mitostatin that it carries out in the mitophagic program. Indeed, mitostatin is not only required early for decorin-mediated depolarization but also it is capable of triggering depolarization alone (cf. Fig. 4, I and J), without external stimulation. This opens new avenues for understanding mitostatin biology as a tumor suppressor and regulation of mitochondrial health and dynamics. Furthermore, in studies done by our group in endothelial cells (119) and unpublished data in tumor cells, and others (107), mitostatin physically binds mitofusin 2. Mitofusin 2 is a key regulator of mitochondrial Ca²⁺ signaling and overall homeostasis (101, 103, 106). We discovered a functional role for mitofusin 2 in coordinating decorin-mediated depolarization in tumor cells. This is important given the role of calcium flux into and out of the mitochondria. Moreover, via dual knockdown experiments, we discovered that mitostatin and mitofusin 2 are within the same signaling pathway, giving more robustness and functionally corroborating the biochemical data of these two proteins physically interacting with each other. It will be interesting to determine if the innate GTPase activity of mitofusin 2 (102, 105) is required for either mitostatin binding and downstream depolarization and/or coordinating external signals (such as those conveyed by decorin) to depolarize the mitochondria.

Considering that decorin engages a Met/mitostatin/mitofusin 2 signaling axis to affect depolarization, the connections to PINK1/Parkin-mediated mitophagy are substantially strengthened. Damaged mitochondria (99, 104), mitochondria in highly metabolic tissues (120), or mitochondria that have received a signal for depolarization (62, 121) are “marked” to undergo a disconnection from the tubular mitochondrial network and are eventually degraded by mitophagosomes (63, 122–126). Deciphering this mechanism would be directly relevant to fully comprehending decorin-evoked tumor cell mitophagy.

Using various inhibitors, we have determined that decorin engages and “opens” the mitochondria permeability transition pore (mPTP) for extracellular signal-dependent depolarization. Although the responsible members of signaling apparatus currently remain unknown, several candidates may be involved. There is evidence that energy-sensing kinases such as AMPK and ULK1 may be responsible in coordinating mitophagy (127, 128). It is not known, however, if these kinases can also induce ΔΨm. Furthermore, it is also unknown where these kinases, if suitable for mediating depolarization, fall in the decorin/Met/mitostatin/mitofusin 2/mPTP axis. Whether there is a connection between these kinases and the molecular components of the mPTP is unknown as well but is currently the subject of future studies.

It has been shown before that a cocktail of three chicken brain hyalectans (aggrecan, phosphacan, and versican) mediates depolarization in embryonic avian axons (129). In this work, the mixture of the three proteoglycans was primarily bound to the substrate. In another study, soluble aggrecan used at double the concentration was able to induce a higher magnitude of mitochondrial depolarization in sensory neurons (130). Thus, although the mechanism of action was not delineated in these abovementioned studies, it is plausible to infer that minimal aggrecan harbors the depolarizing effect on growing axons.

It has been previously shown in an ex vivo assay using cartilage explants that following a single impact load, calcium is released from the endoplasmic reticulum via the ryanodine receptor and is taken up by the mitochondria via the uniport transporter, causing depolarization and caspase 9 activation (131). Notably, soluble decorin is capable of increasing intracellular calcium, [Ca²⁺]_I, in EGFR-overexpressing squamous carcinoma cells (11). Decorin effects persisted in the absence of extracellular Ca²⁺ but were blocked by specific inhibitors of EGFR tyrosine kinase, by physical downregulation of the EGFR, and were not mimicked by biglycan (11), as in the present report for depolarization.

Finally, we previously found that decorin-evoked mitophagy is required for decreased VEGFA (56). This is consistent with recent findings that decorin catabolizes intracellular endothelial VEGFA via autophagy (46) and in this study that decorin mediates depolarization in ex vivo aortic rings and HUVEC (cf. Fig. 8). This lays the foundational framework to investigate whether ΔΨm is needed for decorin-mediated angiostasis. Using either Ca²⁺ release antagonists or mPTP inhibitors (pharmacological or genetic) will help us understand the underlying connection to this important aspect of decorin biology in cancer.

The potential future use of decorin as a next generation protein therapy against human cancer is bright. This assertion is based on and reinforced by an overwhelming number of studies conducted over the past couple of decades. The supporting data and rationale demonstrate that reduced DCN expression is significantly associated with a poor clinical outcome (132–136). Conversely, high DCN expression within the stromal compartment is a positive prognosticator for patients with breast cancer (137). These clinical relations are further buttressed by a plethora of data documenting the antioncogenic properties of decorin in preclinical and various genetic studies (23–28, 138–148).

In conclusion, we have delineated a molecular axis focused on delivering prodepolarization signals to the mitochondrial network in tumor cells. These findings constitute a framework by which decorin could potentially exert promitophagic programs in an effort to curb tumorigenesis.

SUPPLEMENTAL DATA

Supplemental Figs. S1–S5: https://doi.org/10.6084/m9.figshare.20352915.

GRANTS

This original research was supported in part by the National Institutes of Health Grants RO1 CA39481 and RO1 CA245311 (to R.V.I).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTION

T.N. and R.V.I. conceived and designed research; T.N., C.X., and R.V.I. performed experiments; T.N., C.X., and R.V.I. analyzed data; T.N., C.X., and R.V.I. interpreted results of experiments; T.N., C.X., and R.V.I. prepared figures; T.N. and R.V.I. drafted manuscript; T.N., C.X., and R.V.I. edited and revised manuscript; T.N., C.X., and R.V.I. approved final version of manuscript.