Organelle contact reorganization drives calcium-dependent autophagy under proteostatic stress

ABSTRACT

Disruption of proteostasis is a defining feature of cancer and other chronic diseases. The AAA+ ATPase VCP/p97 (valosin containing protein) is a key regulator of proteostasis by disassembling ubiquitinated substrates for degradation. VCP overexpression supports cancer cell survival and correlates with poor prognosis, promoting the development of VCP inhibitors as anti-cancer agents. However, the molecular basis for cancer-selective vulnerability of VCP inhibition remains unclear. Here, we demonstrate that allosteric VCP inhibition triggers cell-type specific macroautophagy/autophagy through dynamic reorganization of organelle contact sites. In human umbilical vein endothelial cells (HUVECs), VCP inhibition induces adaptive autophagy through coordinated reorganization of plasma membrane (PM)-ER-mitochondria contacts. Controlled opening of the mitochondrial permeability transition pore (mPTP) releases calcium into the cytosol, activating AMP-activated protein kinase (AMPK) and TFEB pathways, collectively enhancing autophagic flux and sustaining endothelial survival. Critically, calcium-activated kinase inhibitor or calcium chelators blocked VCP inhibitor-induced autophagy in HUVECs, confirming calcium signaling as the central mediator of adaptive autophagy. In contrast, HCT116 colon cancer cells fail to maintain calcium homeostasis under VCP inhibition, leading to mitochondrial calcium overload, defective autophagy, and cell death. Together, our findings identify organelle contact reorganization and calcium homeostasis as key determinants of cell fate under conditions of proteotoxic stress, revealing how VCP inhibition selectively suppresses tumor progression while preserving vascular integrity that could enhance drug delivery and reduce tumor hypoxia.

Abbreviations: ATF4: activating transcription factor 4; ATG3: autophagy related 3; ATG7: autophagy related 7; ATP: adenosine triphosphate; CAMKK2: calcium/calmodulin dependent protein kinase 2; CETSA: cellular thermal shift assay; CQ: chloroquine; CTS: cryptotanshinone; DDIT3/CHOP: DNA damage inducible transcript 3; DQ-BSA: dye quenched-bovine serum albumin; EIF2A: eukaryotic initiation factor 2A; ER: endoplasmic reticulum; ERAD: endoplasmic reticulum-associated protein degradation; HCT116: human colon carcinoma cell line; HSPA5: heat shock protein family A (Hsp70) member 5; HUVECs: human umbilical vein endothelial cells; ITPR1/IP₃R1: inositol 1,4,5-trisphosphate receptor type 1; MAP1LC3B/LC3: microtubule associated protein 1 light chain 3 beta; mPTP: mitochondrial permeability transition pore; MT-ND1: mitochondrially encoded NADH:ubiquinone oxidoreductase core subunit 1; MTOR: mechanistic target of rapamycin kinase; ORAI1: ORAI calcium release-activated calcium modulator 1; PLA: proximity ligation assay; PM: plasma membrane; PRKAA/AMPK: protein kinase AMP-activated catalytic subunit alpha; ROS: reactive oxygen species; SQSTM1/p62: sequestosome 1; STIM1: stromal interaction molecule 1; TFEB: transcription factor EB; U87MG: human glioblastoma astrocytoma cell line; FAF2/UBXD8: Fas associated factor family member 2; UPR: unfolded protein response; VCP/p97: valosin containing protein; VDAC1: voltage dependent anion channel 1.

Introduction

Proteostasis, the dynamic balance between protein synthesis and degradation, is fundamental for cellular function and survival. Disruption of proteostasis is closely linked to the progression of cancer, neurodegeneration, aging, and metabolic disorders [1,2]. Cancer cells rely on enhanced proteasome activity to sustain high rates of protein turnover, which has led to the development of proteasome inhibitors such as bortezomib and carfilzomib as anti-cancer therapeutics [3]. However, their clinical efficacy is frequently limited by compensatory autophagy and the subsequent drug resistance. To address this challenge, molecular chaperones that facilitate protein folding and degradation have been explored as therapeutic targets for cancer. Among these, the AAA+ ATPase VCP/p97 (valosin containing protein) has attracted interest as a central regulator of proteostasis by extracting ubiquitinated proteins from protein complexes or organelle membranes for proteasomal degradation [4–6]. VCP is also essential for endoplasmic reticulum (ER)-associated protein degradation (ERAD) [7,8], protein aggregate clearance, and the maturation of ubiquitin-containing autophagosomes [9–11], linking the ubiquitin-proteasome and autophagy pathways.

VCP overexpression correlates with poor prognosis in cancer, and its inhibition induces lethal ER stress and cancer cell death. Several VCP inhibitors have been developed as potential anti-cancer agents [12–18]. However, despite their therapeutic promise, clinical translation has faced significant challenges. CB-5083, an inhibitor of VCP ATPase activity, was discontinued in clinical trials due to an off-target effect [19]. NMS-873, an allosteric VCP inhibitor, was shown to impair glycometabolism in HCT116 colon cancer cells through a VCP-independent mechanism [20,21]. These findings underscore that the biological activities of VCP inhibitors cannot be fully explained by canonical proteasomal degradation, highlighting the need to define their broader impact on cellular stress response.

Recent findings have revealed that VCP regulates ER-mitochondria contacts that coordinate lipid exchange, calcium transfer, mitochondria dynamics, and autophagy [22–24]. The VCP- FAF2/UBXD8 complex modulates membrane lipid composition and fluidity at these contacts [25], while VCP-dependent degradation of the tethering protein VAPB via VPS13D further emphasizes its crucial role in regulating ER-mitochondria contacts [26]. Although disrupted ER-mitochondria crosstalk has been increasingly implicated in disease [27], how VCP inhibition alters these contact sites and their downstream signaling to determine cell fate remains poorly understood.

While the anti-cancer effects of VCP inhibition have been extensively studied, its impact on normal cells, particularly vascular cells, remains elusive. The efficacy of anti-cancer agents depends not only on direct tumor cell cytotoxicity but also on preserving vascular integrity, as abnormal vessels impair drug delivery and promote hypoxia-induced angiogenesis [28]. Therefore, understanding how VCP inhibition affects vascular cells is critical for defining its cancer-selective therapeutic potential.

Here, we elucidate a previously unrecognized mechanism by which VCP inhibition suppresses tumor progression while preserving endothelial integrity through calcium-dependent autophagy. We further show that VCP inhibition consistently induces organelle contact reorganization, yet this reorganization produces divergent outcomes between endothelial and tumor cells. Endothelial cells survive VCP inhibition by engaging controlled mPTP opening and activating calcium-dependent adaptive autophagy through AMPK and TFEB pathways. In contrast, HCT116 colon cancer cells fail to maintain calcium homeostasis, resulting in mitochondrial calcium overload, defective autophagy, and subsequent cell death.

Results

Differential autophagy induction by allosteric VCP inhibition in HUVECs and HCT116 cells

To understand the mechanism of cancer-selective VCP inhibitor toxicity, we compared cellular responses between HCT116 colon cancer cells and human umbilical vein endothelial cells (HUVECs). Colorectal cancer (CRC) is among the most prevalent malignancies worldwide, and its progression is often accompanied by persistent ER stress and dysregulation of the unfolded protein response (UPR), which promotes tumor growth, immune evasion, and therapy resistance [29]. Moreover, HCT116 cells are widely used in VCP inhibition studies, and the first orally bioavailable VCP inhibitor, CB-5083, was validated in HCT116 xenograft models [17]. HUVECs represent the critical vascular compartment whose integrity determines drug delivery and hypoxia-driven angiogenesis. To minimize off-target effects, we used the allosteric inhibitor, NMS-873, which stabilizes VCP in ADP-bound conformation by binding between the D1 and D2 ATPase domains [30] (Figure 1(A)).

Figure 1.

Differential autophagy induction by allosteric VCP inhibition in HUVECs and HCT116 cells. (A) Structure of NMS-873 and its binding site between the D1 and D2 domains of VCP. The structures of VCP bound to ADP and ATP were obtained from the Protein Data Bank (PDB IDs: 5FTK and 5FTN). Structural visualizations were generated using PyMOL 3.1 software. (B) VCP mRNA expression levels in colorectal cancer patient cohorts from the Gene Expression Omnibus dataset (GSE33113). Values are means ± SD; *p < 0.05. (C) Western blot analysis of VCP levels in HUVECs and HCT116 cells. (D) Western blot analysis of basal ER stress levels in HUVECs and HCT116 cells. (E) The effect of NMS-873 on cell proliferation was measured by MTT colorimetric assay in HUVECs. Values are means ± SD; n = 3. (F) The effect of NMS-873 on cell viability was assessed by Trypan blue assay after NMS-873 treatment for 48 h in HUVECs. Values are means ± SD; n = 3, **p < 0.01, ****p < 0.0001. ns: not significant. (G) DDIT3/CHOP-promoter-mCherry expression in HEK293 cells after NMS-873 (1 μM) for 24 h. Scale bar: 20 μm. (H) Quantitative data of the mCherry fluorescence intensity/cell graph. Values are means ± SD; n = 3, **p < 0.01. (I) Western blot analysis of the poly-ubiquitinated proteins after NMS-873 (1 μM) treatment in both HUVECs and HCT116 cells. (J) Western blot analysis of UPR markers after NMS-873 (1 μM) and thapsigargin (Tg; 0.05 μM) treatment in both HUVECs and HCT116 cells for 24 h. (K) Schematic diagram of proteostatic compensation between the ubiquitin-proteasome system and autophagy. (L) Western blot analysis of LC3-II and SQSTM1 after NMS-873 (1 μM) treatment in both HUVECs and HCT116 cells.

Clinical analysis of publicly available colorectal cancer transcriptomic dataset revealed significantly elevated VCP mRNA levels in CRC tissues compared with normal colon tissues (Figure 1(B)). Consistently, HCT116 cells exhibited higher VCP protein levels (Figure 1(C) and Figure S1A) and basal ER stress than HUVECs, as indicated by increased UPR markers including HSPA5, phospho-EIF2A, and ATF4 (Figure 1(D) and Figure S1B). As such, HCT116 cells might be more vulnerable to VCP inhibition than endothelial cells, due to their higher basal proteostatic stress.

We next examined the cellular response to pharmacological VCP inhibition in both HUVECs and HCT116 cells. HUVECs were treated with different concentrations of NMS-873 (0.01–5 μM), and cell proliferation and viability were assessed using MTT assay and Trypan blue assay. Despite lower VCP expression levels, HUVECs tolerated NMS-873 at concentrations up to 1 μM without significant cytotoxicity after 48 h (Figure 1(E,F)), while the same concentration effectively targeted HCT116 cells, consistent with previously reported IC₅₀ values [13,20,21]. In our experiments using HCT116 cells, the IC₅₀ value of NMS-873 was 0.34 μM at 48 h (Figure S1C), indicating approximately a 4–5 fold greater sensitivity of HCT116 cells to VCP inhibition. Viability was largely preserved up to 1 μM, with a modest decrease at 48 h (Figure S1D). These results indicate the underlying adaptive mechanisms to proteotoxic stress in HUVECs.

As VCP is essential for ERAD by exporting misfolded proteins from the ER to the cytosol, pharmacological inhibition of VCP leads to the accumulation of misfolded proteins in the ER, which in turn triggers ER stress and activates UPR to restore proteostasis [31]. Consistent with this, NMS-873 treatment induced ER stress, as confirmed by increased DDIT3/CHOP-promoter-mCherry fluorescence intensity [32] (Figure 1(G,H)). Both HUVECs and HCT116 cells showed similar initial responses to VCP inhibition, including the accumulation of poly-ubiquitinated proteins (Figure 1(I) and Figure S1E, F). We examined the EIF2AK3-EIF2A-ATF4 signaling axis of the UPR pathways as proteotoxic stress suppresses global protein synthesis while enhancing stress adaptive gene expression [33,34]. NMS-873 increased expression of HSPA5, phospho-EIF2A, and ATF4 in both cell types (Figure 1(J) and Figure S1G, H), similar to the thapsigargin (Tg), an inhibitor of the ATP2A/SERCA (ATPase sarcoplasmic/endoplasmic reticulum Ca²⁺ transporting), used as a positive control. However, only HCT116 cells showed DDIT3 upregulation, indicating progression to pro-apoptotic UPR (Figure 1(J) and Figure S1H). Consistent with this, NMS-873 increased the levels of cleaved CASP3 in HCT116 cells (Figure S1I), indicating activation of pro-apoptotic UPR signaling under sustained ER stress. This divergence in UPR signaling suggests fundamental differences in stress adaptation capacity between normal endothelial cells and colon cancer cells.

Proteasomal degradation and autophagy constitute two complementary and interconnected arms of the cellular proteostasis network [35] (Figure 1(K)). In HUVECs, NMS-873 markedly increased MAP1LC3/LC3-I to LC3-II conversion and SQSTM1 degradation between 12 and 72 h (Figure 1(L) and Figure S1J), indicating enhanced autophagic flux [36]. In contrast, HCT116 cells showed reduced LC3-I to LC3-II conversion and SQSTM1 accumulation (Figure 1(L) and Figure S1K), indicating impaired autophagic flux.

We observed that basal ER stress markers were elevated in HCT116 and HT-29 (another colorectal cancer cell line) compared with CCD-18Co cells (a human colon fibroblast line), suggesting that colorectal cancer cells exhibit higher proteostatic stress (Figure S1L). Notably, in HT-29 cells, NMS-873 consistently induced ER stress (Figure S1M, N) without triggering productive autophagic flux (Figure S1O), mirroring the stress-adaptation uncoupling observed in HCT116 cells. Thus, the qualitative stress response pattern, ER stress induction with impaired autophagic adaptation, is conserved in colorectal cancer cells.

To investigate the mechanistic basis underlying the increased vulnerability of HCT116 to VCP inhibition compared with HUVECs, we examined the contribution of ER stress. Because colorectal cancer cells exhibit elevated basal UPR markers compared with normal cells, we hypothesized that VCP inhibition may exacerbate an already stressed proteostatic environment beyond a tolerable threshold. To test this, we assessed whether alleviating ER stress could modulate sensitivity to VCP inhibition. Pretreatment with 4-phenylbutyrate (4-PBA), a chemical chaperone that alleviates ER stress, partially rescued NMS-873-induced loss of viability in HCT116 and HT-29 cells, whereas it had minimal effect in CCD-18Co cells (Figure S2A). These results indicate that unresolved ER stress could be a major contributor to cancer-selective vulnerability following VCP inhibition.

Consistent with this, these proteotoxic stress and defective autophagy response were also observed in U87MG glioblastoma cells (Figure S2B-G). While our expanded analysis focused primarily on colorectal cancer cell lines, these findings suggest that differential engagement of adaptive autophagy may extend to additional malignant contexts.

Collectively, these results suggest that upon VCP inhibition, normal endothelial cells with low basal ER stress survive through adaptive autophagy, whereas colorectal cancer cells with high basal ER stress fail to induce autophagy, resulting in unresolved proteotoxic stress.

Allosteric VCP inhibition suppresses endothelial proliferation and angiogenesis

As VCP expression was lower in HUVECs than in HCT116 cells, we performed cellular thermal shift assay (CETSA) to confirm NMS-873 directly binds to endogenous VCP in HUVECs. NMS-873 induced a significant thermal stability shift in HUVECs at 72°C (Figure S2H). At 72°C, NMS-873 stabilized VCP in a dose-dependent manner, with an EC₅₀ of 1.4 μM (Figure S2I, J), confirming that NMS-873 binds to endogenous VCP in endothelial cells with high specificity. Despite this direct binding, ANXA5/annexin V and propidium iodide staining revealed that NMS-873 treatment for 72 h did not increase the percentage of ANXA5/annexin V and PI-positive cells compared to untreated controls (Figure 2(A-C)), indicating NMS-873 induced cytostatic rather than cytotoxic stress. Brefeldin A (BFA), which disrupts ER to Golgi trafficking and induces apoptotic ER stress, was used as a positive control.

Figure 2.

Allosteric VCP inhibition suppresses endothelial proliferation and angiogenesis. (A) Representative flow cytometry dot plots of apoptosis. Evaluation of cell death by flow cytometry using ANXA5/annexin V-PI staining after NMS-873 (1 μM) and brefeldin A (BFA; 0.05 μM) treatment for 72 h. (B) Bars represent the percentage of ANXA5/annexin V-positive cells. Values are means ± SD; n = 3, ****p < 0.0001, ns: not significant. (C) Bars represent the percentage of PI-positive cells. Values are means ± SD; n = 3, ****p < 0.0001, ns: not significant. (D) Representative images of in vitro capillary tube formation. HUVECs were stimulated with 30 ng/mL of VEGF and cotreated with NMS-873 (1 μM) for 9 h. Scale bar: 200 μm. (E) Quantitative data of tube junction counts. Tube formation was quantitatively evaluated by measuring tube junction counts. Values are means ± SD; n = 3, ***p < 0.001. (F) Representative images of the wound-healing assay. HUVECs were stimulated with 30 ng/mL of VEGF and cotreated with NMS-873 (1 μM) for 24 h. Scale bar: 200 μm. (G) Quantitative data of the cell-free area. Values are means ± SD; n = 3, ***p < 0.001, ****p < 0.0001.

We further examined the anti-angiogenic effects of NMS-873. Consistent with its cytostatic effects, NMS-873 dose-dependently suppressed VEGF-induced tube formation (Figure 2(D,E)) and reduced HUVECs migration (Figure 2(F,G)). Taken together, these results demonstrate that micromolar concentrations of NMS-873 inhibit angiogenesis without inducing endothelial cell death, revealing a functional separation between anti-angiogenic and cytotoxic effects.

Allosteric VCP inhibition induces autophagic flux as a survival mechanism in HUVECs

To validate the adaptive role of autophagy, we conducted comprehensive autophagic flux analyses. NMS-873 increased LC3 puncta (Figure 3(A,D)) and enhanced LysoTracker fluorescence intensity (Figure 3(B,E)). Using the dye-quenched bovine serum albumin (DQ-BSA) assay, a self-quenched substrate that fluoresces upon proteolytic degradation within functional lysosomes, we observed that NMS-873 significantly increased DQ-BSA fluorescence (Figure S3A), indicating enhanced lysosomal proteolysis rather than lysosomal functional blockade. Cryptotanshinone (CTS), a well-established autophagy inducer, was used as a positive control [37,38]. Using mRFP-GFP-LC3 reporters, we observed that NMS-873 significantly reduced yellow puncta (autophagosomes) and increased red puncta (autolysosomes) (Figure 3(C,F)), confirming progression from autophagosome to autolysosome. Furthermore, cotreatment with NMS-873 and chloroquine (CQ), which blocks lysosomal acidification, further elevated SQSTM1 accumulation and LC3-II levels compared with CQ alone (Figure 3(G) and Figure S3B). Consistently, this effect was also evident after 48 h of NMS-873 treatment (Figure S3C). Cotreatment with proteasome inhibitor MG-132 and NMS-873 did not significantly block NMS-873-induced SQSTM1 degradation compared with MG-132 alone (Figure S3D), suggesting that SQSTM1 turnover under NMS-873 treatment is mediated through autophagy. Collectively, these results indicate that NMS-873 induces autophagic flux in HUVECs. Consistently, UPR markers decreased in a time-dependent manner, suggesting that autophagy resolved proteotoxic stress during VCP inhibition in HUVECs (Figure 3(H) and Figure S3E).

Figure 3.

Allosteric VCP inhibition induces autophagic flux as a cellular survival mechanism in HUVECs. (A) Representative confocal images of LC3 puncta after NMS-873 (1 μM) for 24 h. Scale bar: 20 μm. (B) Lysosomal activity was assessed by LysoTracker staining after NMS-873 (1 μM) and CTS (5 μM) treatment for 24 h. Scale bar: 20 μm. (C) Representative confocal images of autophagosome (GFP⁺ RFP⁺) and autolysosome (GFP⁻ RFP⁺) detection. HUVECs were transfected with mRFP-GFP-LC3 and treated with NMS-873 (1 μM) and CTS (5 μM) for 24 h. Scale bar: 10 μm. (D) Quantitative data of LC3 puncta. Values are means ± SD; n = 3, *p < 0.05. (E) Quantitative data of relative LysoTracker fluorescence intensity. Values are means ± SD; n = 3, **p < 0.01, ***p < 0.001. (F) Quantitative data of mRFP-GFP-LC3 puncta. Values are means ± SD; n > 20 cells, **p < 0.01, ****p < 0.0001. (G) Western blot analysis of LC3-II and SQSTM1 after treatment with NMS-873 (1 μM) and CQ (20 μM) for 24 h. (H) Western blot analysis of UPR markers after NMS-873 (1 μM) treatment for 24, 48, and 72 h. (I) Cell proliferation was measured by MTT colorimetric assay after NMS-873 (1 μM) treatment in the presence of chloroquine (CQ; 20 μM) and bafilomycin A₁ (BafA1; 2 nM) for 48 h. Values are means ± SD; n = 3, ***p < 0.001, ****p < 0.0001. (J) Cell proliferation was measured by MTT colorimetric assay after ATG7 knockdown and treatment with NMS-873 (1 μM). Values are means ± SD; n = 3, **p < 0.01, ***p < 0.001, ns: not significant.

To directly determine whether autophagy supports endothelial cell survival, we examined the effects of autophagy inhibition. Pharmacological blockade of autophagy with CQ and bafilomycin A₁ (BafA1) markedly reduced HUVECs survival (Figure 3(I)). To exclude the possibility that lysosomal inhibitors exert cytotoxic effects independent of autophagy inhibition, we genetically suppressed autophagy via knockdown of ATG7 (autophagy related 7), an essential autophagosome formation protein (Figure S3F). ATG7 knockdown significantly reduced HUVEC proliferation (Figure 3(J)), indicating that basal autophagy supports endothelial cell homeostasis and survival. Notably, ATG7 depletion did not further exacerbate growth inhibition in the presence of NMS-873 (Figure 3(J)). Together, our results demonstrate that although VCP inhibition imposes proteotoxic stress in HUVECs, cell survival is maintained through the induction of adaptive autophagy.

In contrast to HUVECs, VCP inhibition produced a markedly different autophagic response in colorectal cancer cells. NMS-873 reduced LysoTracker intensity in HCT116 cells, suggesting impaired lysosomal acidification (Figure S3G). In addition, a decrease in LC3-II levels together with the accumulation of SQSTM1 indicated disruption of autophagic flux (Figure 1(L) and Figure S1K). Moreover, cotreatment with CQ and NMS-873 did not further increase LC3-II or SQSTM1 levels compared with CQ alone (Figure S3H), confirming that autophagic flux was already impaired by NMS-873. These findings suggest that VCP inhibition compromises autophagic processing at multiple steps, thereby limiting adaptive stress response in colorectal cancer cells.

To determine whether impaired autophagy contributes to the increased vulnerability of HCT116 cells to VCP inhibition in HCT116 cells, we genetically suppressed autophagy via knockdown of ATG7 (Figure S3I). ATG7 knockdown alone significantly reduced HCT116 cell proliferation, indicating that basal autophagy supports cancer cell survival. Notably, combined ATG7 knockdown and NMS-873 treatment further decreased cell proliferation compared with ATG7 knockdown alone (Figure S3J). Because NMS-873 impairs lysosomal acidification and disrupts the autophagic degradation step, genetic suppression of autophagosome formation would be expected to further compromise proteostatic capacity. Consistently, ATG3 (autophagy related 3) knockout HeLa cells, which lack the E2-like enzyme required for LC3 lipidation and autophagosome formation, increased sensitivity to NMS-873 (Figure S3K, L), supporting that impaired autophagy enhances cellular vulnerability to VCP inhibition.

Allosteric VCP inhibition reorganizes PM-ER-mitochondria contacts to regulate calcium dynamics

VCP is well-established to be essential for the maturation of ubiquitin-containing autophagosomes [9]. To investigate how autophagy is induced in HUVECs when VCP function is inhibited, we examined whether non-canonical pathways contribute to this process. As ER-mitochondria contacts play a crucial role in ER stress regulation [27,39,40], calcium homeostasis, and autophagy [23,24,41,42], we examined whether NMS-873-induced ER stress reorganizes ER-mitochondria contacts to trigger adaptive autophagy and alleviate proteotoxic stress (Figure 4(A)).

Figure 4.

Allosteric VCP inhibition reorganizes PM-ER-mitochondria contacts. (A) Schematic illustration of ER stress-dependent regulation of ER-mitochondria contacts and calcium dynamics. (B) Representative confocal images of PLA between ITPR1 (ER protein) and VDAC1 (mitochondria protein) after NMS-873 (1 μM) treatment for 24 h. Scale bar: 10 μm. (C) Quantitative data of the PLA analysis showing the interactions between ER and mitochondria. Values are means ± SD; n = 3, **p < 0.01. (D) Representative confocal images of PLA between VAPB (ER protein) and RMDN3 (mitochondria protein) after NMS-873 (1 μM) treatment for 24 h. Scale bar: 10 μm. (E) Quantitative data of the PLA analysis showing the interactions between ER and mitochondria. Values are means ± SD; n = 3, **p < 0.01. (F) ER calcium levels were detected by Mag-Fluo-4 staining after NMS-873 (1 μM) treatment for 24 h. Scale bar: 20 μm. (G) Quantitative data of the Mag-Fluo-4 fluorescence intensity (ER calcium indicator). Values are means ± SD; n = 3, **p < 0.01. (H) Schematic illustration of ER calcium restoration via the plasma membrane and ER contacts. (I) Representative confocal images of PLA between STIM1 (ER protein) and ORAI1 (plasma membrane protein) after NMS-873 (1 μM) treatment for 24 h. Scale bar: 10 μm. (J) Quantitative data of the PLA analysis showing the interactions between ER and plasma membrane. Values are means ± SD; n = 3, **p < 0.01. (K) Mitochondrial calcium levels were detected by Rhod-2 staining after NMS-873 (1 μM) treatment for 24 h. Scale bar: 20 μm. (L) Quantitative data of the Rhod-2 fluorescence intensity (mitochondrial calcium indicator). Values are means ± SD; n = 3, ***p < 0.001. (M) Mitochondrial ROS were detected by MitoSOX staining after NMS-873 (1 μM) and rotenone (Rot; 100 nM) treatment for 24 h. Scale bar: 20 μm. (N) Quantitative data of the MitoSOX fluorescence intensity. Values are means ± SD; n = 3, ****p < 0.0001. ns: not significant. (O) Cytosol calcium levels were detected by Fluo-4 staining after NMS-873 (1 μM) treatment for 24 h. Scale bar: 20 μm. (P) Quantitative data of the Fluo-4 fluorescence intensity. Values are means ± SD; n = 3, ****p < 0.0001.

We first assessed ER-mitochondria contacts by proximity ligation assay (PLA), focusing on two tethering complexes: ITPR1-VDAC1, which facilitates calcium transfer between the two organelles, and VAPB-RMDN3, which stabilizes contacts and mediates lipid exchange [27]. NMS-873 significantly increased PLA signals for both complexes at both 6 and 24 h in HUVECs (Figure S4A and Figure 4(B-E)), indicating enhanced ER-mitochondria contacts. To further validate, we employed the SPLICS (split-GFP-based contact site sensor) reporter system to directly visualize ER-mitochondria proximity. Confocal imaging revealed a significant increase in GFP intensity in NMS-873-treated cells compared with DMSO controls (Figure S4C), providing independent evidence for increased ER-mitochondria contacts. We next assessed the functional consequence of increased contact formation by measuring ER-to-mitochondria calcium transfer. Following histamine-induced ER calcium release, NMS-873-treated cells exhibited significantly enhanced mitochondrial calcium uptake compared with DMSO controls (Figure S4D), indicating strengthened calcium coupling at ER-mitochondria contact sites. Together, these structural and functional analyses independently validate the increased ER-mitochondria contacts detected by PLA.

Sustained ER-mitochondria contacts are expected to promote calcium influx into mitochondria and consequently reduce ER calcium stores. However, Mag-Fluo-4 fluorescence revealed unexpectedly elevated ER calcium levels following NMS-873 treatment (Figure 4(F,G)). We next examined ER-plasma membrane (PM) contacts as ER calcium replenishment occurs via store operated calcium entry (SOCE), mediated by STIM1 (ER protein) and ORAI1 (PM protein) interactions and subsequent ATP2A/SERCA (ATPase sarcoplasmic/endoplasmic reticulum Ca²⁺ transporting) dependent refilling (Figure 4(H)) [43], NMS-873 increased STIM1-ORAI1 interactions at both 6 and 24 h (Figure S4B and Figure 4(I,J)), indicating active SOCE-mediated ER calcium replenishment.

We next examined mitochondrial calcium levels, as impaired ER to mitochondria calcium transfer could also contribute to elevated ER calcium. Rhod-2 fluorescence revealed that NMS-873 induced a transient increase in mitochondrial calcium at 6 h, followed by a reduction at 24 h (Figure S4E and Figure 4(K,L)), suggesting time-dependent calcium efflux from mitochondria. Importantly, this occurred without mitochondrial stress, as mitochondrial ROS levels remained unchanged (Figure 4(M,N)). Consistent with this redistribution, Fluo-4 fluorescence confirmed elevated cytosolic calcium levels (Figure 4(O,P)), indicating calcium redistribution from mitochondria to the cytosol.

To determine whether these phenotypes were specific to NMS-873, we examined additional VCP inhibitors. The ATP competitive VCP inhibitor ML-240 also induced autophagy (Figure S4F, G) and increased ER-mitochondria contacts (Figure S4H), recapitulating the effects of NMS-873 in HUVECs. In addition, we tested another D1-D2 linker binding allosteric inhibitor, UPCDC-30245. Consistent with NMS-873, UPCDC-30245 induced autophagy (Figure S4I), enhanced lysosomal proteolytic activity (Figure S4J), and increased ER-mitochondria contacts in HUVECs (Figure S4K). Notably, previous studies demonstrated that UPCDC-30245 primarily alkalinized lysosomes and blocked endo-lysosomal degradation in cancer cell lines (HCT116, H1299, and HeLa) [44]. These observations suggest that cellular responses to allosteric VCP inhibition may vary according to cell type.

Calcium-activated AMPK and TFEB pathways mediate autophagy induction

Elevated cytosolic calcium activates autophagy through calcium-dependent kinases (Figure 5(A)). To verify the calcium dependency of this response, we cotreated cells with NMS-873 and BAPTA-AM (an intracellular Ca²⁺ chelator) or EGTA (an extracellular Ca²⁺ chelator). Both chelators significantly reduced NMS-873-induced LC3-I to LC3-II conversion (Figure S5A, B), confirming that NMS-873 induces calcium-dependent autophagy. Notably, NMS-873 enhanced PRKAA/AMPK (protein kinase AMP-activated catalytic subunit alpha) phosphorylation (Figure 5(B) and Figure S5C). However, NMS-873 did not alter MTOR (mechanistic target of rapamycin kinase) phosphorylation or the phosphorylation of the MTOR complex 1 (MTORC1) downstream substrate EIF4EBP1 (Thr37/46) (Figure S5D, E), indicating that canonical MTORC1 signaling remained largely unaffected. To determine whether AMPK activation reflects a calcium-responsive signaling pathway downstream of VCP inhibition, we cotreated HUVECs with NMS-873 and STO-609, a CAMKK2 inhibitor. STO-609 suppressed NMS-873-induced AMPK phosphorylation (Figure 5(B) and Figure S5C), indicating that CAMKK2-AMPK signaling is calcium-dependent and contributes to autophagy induction. Given that mitochondrial calcium is essential for ATP production [45,46] and NMS-873 reduced mitochondrial calcium levels at 24 h (Figure 4(K,L)), we next assessed cellular energy status. NMS-873 significantly decreased intracellular ATP levels (Figure S5F), indicating dual AMPK activation through both calcium-dependent signaling and energy stress pathways.

Figure 5.

Controlled mPTP opening increases cytosolic calcium levels in and induces autophagy in HUVECs. (A) Schematic illustration of calcium-dependent autophagy signaling. (B) Western blot analysis of the AMPK pathway after treatment with NMS-873 (1 μM) and STO-609 (10 μM) for 24 h. (C) Western blot analysis of TFEB nuclear translocation after NMS-873 (1 µM) treatment for 12 h. (D) Schematic illustration of mPTP-mediated calcium dynamics. (E) qPCR analysis of mtDNA levels in mitochondria and (F) cytosol after NMS-873 (1 μM) treatment for 12 h. Values are means ± SD; n = 3, ***p < 0.001, ****p < 0.0001. (G) Mitochondrial calcium levels were detected by Rhod-2 staining after cotreatment with NMS-873 (1 μM) and mPTP inhibitor ER-000444793 (1 μM) for 24 h. Scale bar: 20 μm. (H) Cytosol calcium levels were detected by Fluo-4 staining after cotreatment with NMS-873 (1 μM) and mPTP inhibitor ER-000444793 (1 μM) for 24 h. Scale bar: 20 μm. (I) Quantitative data of the Rhod-2 fluorescence intensity. Values are means ± SD; n = 3, **p < 0.01, ****p < 0.0001. (J) Quantitative data of the Fluo-4 fluorescence intensity. Values are means ± SD; n = 3, *p < 0.05, ****p < 0.0001.

We next investigated TFEB (transcription factor EB), a master regulator of autophagy and lysosomal biogenesis. TFEB dephosphorylation by the calcium-dependent phosphatases promotes TFEB nuclear translocation [47]. Immunoblotting and immunocytochemistry analyses confirmed that NMS-873 promoted TFEB nuclear translocation (Figure 5(C) and Figure S5G, H). Cryptotanshinone (CTS), a well-established autophagy inducer, was used as a positive control [37,38]. These results indicate that NMS-873 activates both AMPK and TFEB signaling pathways downstream of cytosolic calcium, thereby inducing adaptive autophagy in HUVECs.

To determine whether TFEB activation is mediated through CAMKK2-AMPK signaling, we assessed TFEB subcellular localization following cotreatment with STO-609. STO-609 did not significantly attenuate NMS-873-induced TFEB nuclear translocation (Figure S5I), indicating that CAMKK2-AMPK signaling is not the dominant pathway mediating TFEB activation in this context. Instead, TFEB activation likely occurs through calcium-dependent phosphatases signaling.

In HCT116 cells, NMS-873 similarly increased ER-mitochondria (Figure S5J) and ER-PM contacts (Figure S5K). However, this reorganization resulted in calcium overload in both the mitochondrial and the ER (Figure S5L, M), while cytosolic calcium levels were not significantly different from control (Figure S5N). Consequently, TFEB remained cytosolic (Figure S5O), indicating defective activation of the TFEB-dependent autophagy. To determine whether this calcium dysregulation was restricted to a single colorectal cancer cell line, we extended our analysis to another colorectal cancer cell line, HT-29. Consistent with our findings in HCT116 cells, NMS-873 treatment increased mitochondrial Ca²⁺ levels without elevating cytosolic Ca²⁺ (Figure S5P, Q), indicating that mitochondrial Ca²⁺ overload following VCP inhibition is reproducible across colorectal cancer cell lines.

Together, these findings demonstrate that VCP inhibition by NMS-873 reorganizes PM-ER-mitochondria contact sites, thereby redistributing intracellular Ca²⁺. In HUVECs, controlled calcium redistribution from mitochondria to the cytosol activates AMPK and TFEB signaling, inducing adaptive autophagy. In contrast, colorectal cancer cells fail to properly redistribute Ca²⁺, resulting in mitochondrial Ca²⁺ overload, defective autophagy, and cell death.

Controlled mPTP opening increases cytosolic calcium levels in HUVECs

To understand how mitochondria release calcium to the cytosol, we investigated mitochondrial permeability transition pore (mPTP) dynamics. Persistent ER-mitochondria contacts facilitate sustained Ca²⁺ transfer into mitochondria, leading to mitochondrial Ca²⁺ overload and subsequent mPTP opening. The mPTP functions as a calcium-sensitive channel that increases mitochondrial membrane permeability upon opening, allowing the release of mitochondrial contents, including mitochondrial DNA (mtDNA), ions, and small metabolites (Figure 5(D)) [48–50]. The temporal pattern of mitochondrial Ca²⁺ elevation at 6 h followed by a reduction at 24 h (Figure S4E and Figure 4(K,L)) suggested that Ca²⁺ release is mediated by transient mPTP opening.

To confirm mPTP opening, we quantified mtDNA release into cytosol using real-time quantitative PCR (qPCR) with MT-ND1 and MT-ND2 primers, which target mitochondrial NADH dehydrogenase subunit genes. qPCR analysis revealed significantly increased cytosolic MT-ND1 and MT-ND2 levels at 12 h, accompanied by a reduction in mitochondrial MT-ND1 and MT-ND2 levels (Figure 5(E,F)), confirming mtDNA release through mPTP opening. At 24 h, MT-ND1 and MT-ND2 levels increased in both cytosolic and mitochondrial fractions, suggesting sustained mtDNA release with compensatory mtDNA synthesis. At 48 h, cytosolic MT-ND1 and MT-ND2 levels remained elevated while mitochondrial levels stabilized (Figure S6A, B), indicating delayed clearance of cytosolic mtDNA.

Critically, cotreatment with mPTP inhibitor ER-000444793 partially restored mitochondrial Ca²⁺ levels (Figure 5(G,I)) and reduced cytosolic Ca²⁺ levels (Figure 5(H,J)), confirming that NMS-873-induced cytosolic Ca²⁺ elevation is mediated, at least in part, through mPTP opening and contributes to autophagy induction in HUVECs.

VCP knockdown reorganizes PM-ER-Mitochondria contacts and induces calcium-dependent autophagy

To confirm that the biological effects of NMS-873 were specifically attributable to VCP inhibition, we assessed the impact of siRNA-mediated VCP knockdown in HUVECs. VCP knockdown suppressed cell proliferation (Figure 6(A,B)) and induced ER stress, as evidenced by the accumulation of poly-ubiquitinated proteins (Figure 6(C) and Figure S6C) and upregulation of HSPA5, phospho-EIF2A, and ATF4 (Figure 6(D) and Figure S6D). VCP knockdown also increased LC3-I to LC3-II conversion and SQSTM1 degradation, indicating autophagic flux in response to proteotoxic stress (Figure 6(E) and Figure S6E). To directly assess autophagic flux following VCP knockdown, we performed a CQ-based autophagic flux assay. VCP knockdown increased LC3-II levels, and CQ treatment further enhanced LC3-II accumulation, confirming that autophagy induced by NMS-873 is dependent on VCP inhibition (Figure S6F). Furthermore, VCP knockdown increased ER-mitochondria and ER-PM contacts (Figure 6(F-I)). This reorganization altered intracellular Ca²⁺ dynamics, resulting in reduced mitochondrial Ca²⁺ levels (Figure 6(J,K)) and elevated cytosolic Ca²⁺ levels (Figure 6(L,M)). Together, these results demonstrate that both pharmacological and genetic VCP inhibition trigger similar adaptive responses, further supporting that the cellular effects of NMS-873 are mediated through on-target VCP inhibition.

Figure 6.

VCP knockdown reorganizes PM-ER-mitochondria contacts and induces calcium-dependent autophagy. (A) HUVECs were transfected with VCP siRNA for 24 and 48 h. (B) Cell proliferation was measured by MTT colorimetric assay following transfection with VCP siRNA. Values are means ± SD; n = 3. (C) Western blot analysis of poly-ubiquitinated proteins following transfection with VCP siRNA (10 nM). (D) Western blot analysis of UPR markers following transfection with VCP siRNA (10 nM) for 48 h. (E) Western blot analysis of LC3-II and SQSTM1 following transfection with VCP siRNA (10 nM) for 24, 48, and 72 h. (F) Representative confocal images of PLA between ITPR1 (ER protein) and VDAC1 (mitochondria protein) following transfection with VCP siRNA (10 nM) for 48 h. Scale bar: 10 μm. (G) Quantitative data of the PLA analysis showing the interactions between ER and mitochondria. Values are means ± SD; n = 3, **p < 0.01. (H) Representative confocal images of PLA between STIM1 (ER protein) and ORAI1 (plasma membrane protein) following transfection with VCP siRNA (10 nM) for 48 h. Scale bar: 10 μm. (I) Quantitative data of the PLA analysis showing the interactions between ER and plasma membrane. Values are means ± SD; n = 3, ***p < 0.001. (J) Mitochondrial calcium levels were detected by Rhod-2 staining following transfection with VCP siRNA (10 nM) for 48 h. Scale bar: 20 μm. (K) Quantitative data of the Rhod-2 fluorescence intensity. Values are means ± SD; n = 3, *p < 0.05. (L) Cytosol calcium levels were detected by Fluo-4 staining following transfection with VCP siRNA (10 nM) for 48 h. Scale bar: 20 μm. (M) Quantitative data of the Fluo-4 fluorescence intensity. Values are means ± SD; n = 3, **p < 0.01.

To address VCP inhibitor specificity in colorectal cancer cells, we genetically suppressed VCP using siRNA in HCT116 cells. VCP knockdown reproduced key stress phenotypes observed following NMS-873 treatment. Specifically, siRNA-mediated VCP knockdown significantly reduced HCT116 cell proliferation in a dose-dependent manner (10, 20, 50, and 100 nM) (Figure S6G, H). VCP knockdown also induced dose-dependent accumulation of poly-ubiquitinated proteins and upregulation of UPR markers, including HSPA5, phospho-EIF2A, ATF4, and DDIT3 (Figure S6I, J). Furthermore, VCP knockdown increased ER-mitochondria and ER-PM contacts (Figure S6K, L), closely mirroring the phenotype observed with NMS-873 treatment and further supporting an on-target mechanism of action. Regarding autophagic flux in HCT116 cells, acute pharmacological inhibition of VCP with NMS-873 impaired lysosomal acidification and autophagic flux (Figure S3G, Figure 1(L), and Figure S1K), whereas gradual siRNA-mediated VCP suppression permitted partial compensatory activation of autophagy (Figure S6M). This difference likely reflects distinct kinetics and magnitudes of proteotoxic stress induced by acute pharmacological inhibition versus progressive genetic depletion, rather than off-target lysosomal effects of NMS-873.

Allosteric VCP inhibition suppressed HCT116 tumor growth by restraining endothelial proliferation

To investigate whether NMS-873 exerts selective anti-tumor effects through differential proteotoxic stress responses in tumor cells and endothelial cells, we evaluated the pharmacological activity of NMS-873 in an HCT116 xenograft mouse model. Given that NMS-873 exhibits poor metabolic stability and limited in vivo efficacy upon systemic administration [13], we employed intratumoral injection as an alternative approach to assess in vivo pharmacodynamic effects of NMS-873. Nude mice bearing subcutaneous HCT116 xenografts received intratumoral injection of NMS-873 (5 mg/kg) or vehicle, and tumors were analyzed after treatment for 24 h (Figure 7(A)). Body weight remained unchanged in NMS-873-treated mice, indicating no detectable systemic toxicity (Figure S7A). In addition, no significant differences in tumor weight or volume were observed between groups (Figure S7B-D). However, hematoxylin and eosin (H&E) staining revealed increased tumor cell death in NMS-873-treated tumors compared to vehicle controls (Figure S7E).

Figure 7.

Allosteric VCP inhibition suppressed HCT116 tumor growth by restraining endothelial proliferation. (A) Schematic illustration of the HCT116 xenograft mouse experiment. Mice bearing HCT116 tumors received intratumoral injection of either vehicle or NMS-873 (5 mg/kg) for 24 h. Figure created in BioRender.com. (B) Representative immunofluorescence images of MKI67 in tumor sections. Scale bar: 20 μm. (C) Quantification data of MKI67 expression. Values are means ± SD; n > 10, **p < 0.01. (D) Representative immunofluorescence images of SQSTM1 in tumor sections. Scale bar: 20 μm. (E) Quantification data of SQSTM1 expression. Values are means ± SD; n > 10, ****p < 0.0001. (F) Representative immunofluorescence images of LC3 in tumor sections. Scale bar: 20 μm. (G) Quantification data of LC3 expression. Values are means ± SD; n > 10, *p < 0.05. (H) Representative confocal images of PLA between ITPR (ER protein) and VDAC1 (mitochondria protein) in tumor sections. Scale bar: 20 μm. (I) Quantification data of PLA analysis showing the interactions between ER and mitochondria. Values are means ± SD; n > 10, *p < 0.05. (J) Representative immunofluorescence images of PECAM1 in tumor sections. Scale bar: 20 μm. (K) Quantification data of PECAM1 expression. Values are means ± SD; n > 10, ****p < 0.0001.

We examined whether NMS-873 affects tumor cell growth and autophagic flux in vivo. VCP protein levels were unchanged following NMS-873 treatment, consistent with its allosteric mode of action (Figure S7F, J). MKI67 immunostaining demonstrated that NMS-873 treatment significantly suppressed HCT116 tumor cell growth (Figure 7(B,C)). In addition, NMS-873 treatment led to SQSTIM1 accumulation (Figure 7(D,E)) and reduced LC3 levels (Figure 7(F,G)), indicating impaired autophagic flux in tumor tissues.

To determine whether organelle contact reorganization observed in vitro also occurs in vivo, we performed PLA. ER-mitochondria contacts were significantly increased in NMS-873-treated tumors (Figure 7(H,I)), whereas ER-PM contacts remained unchanged (Figure S7G, K). The absence of detectable ER-PM contact reorganization may reflect the relatively short treatment duration (24 h), as ER-PM contact formation likely occurs as a downstream consequence of enhanced ER-mitochondria contacts.

We next investigated how NMS-873 suppressed tumor progression through modulation of tumor angiogenesis. NMS-873 treatment resulted in a modest but statistically significant reduction in PECAM1-positive microvessel density (Figure 7(J,K)), indicating decreased tumor-associated vascularization. Consistently, VEGFA levels were reduced in NMS-873-treated tumors, supporting attenuation of pro-angiogenic signaling (Figure S7H, L). Finally, CDH1 levels were reduced in NMS-873-treated tumors, indicating altered epithelial marker expression (Figure S7I, M).

Collectively, these findings demonstrate that VCP inhibition by NMS-873 suppresses tumor progression through combined disruption of tumor autophagy and modulation of the tumor vasculature. Consistent with our in vitro observations, NMS-873 selectively suppresses endothelial proliferation, yet stabilizes the tumor vasculature at a lower density. This dual mechanism effectively suppresses tumor progression while preserving functional vascular structures, which could potentially enhance chemotherapeutic efficacy in a clinical setting.

Discussion

Here, we identify a previously unrecognized vascular protective mechanism that explains the cancer-selective VCP inhibitor toxicity. In endothelial cells, coordinated reorganization of PM-ER-mitochondria contacts enables adaptive responses that resolve proteostatic stress while maintaining viability. ER-PM contacts restore ER calcium levels via SOCE, while controlled mPTP opening releases mitochondrial calcium into the cytosol to activate calcium-dependent AMPK and TFEB signaling pathways, thereby inducing autophagy.

During the early phase of ER stress, ER-mitochondria contacts facilitate Ca²⁺ transfer to support mitochondrial ATP production [39]. In endothelial cells, prolonged ER stress induced by VCP inhibition led to sustained ER-mitochondria contacts, which triggered reversible mPTP opening to restore mitochondrial Ca²⁺ homeostasis. While sustained mPTP opening canonically triggers apoptosis via cytochrome c release, transient or reversible opening can promote release of Ca²⁺ and ROS, contributing to intracellular signaling and protecting mitochondria from oxidative stress [45,50]. Consistent with this emerging understanding of mPTP regulation, our findings demonstrate that controlled and reversible opening of mPTP induces Ca²⁺ redistribution and survival signaling in endothelial cells.

Cancer cells are characterized by chronically elevated proteotoxic stress driven by oncogenic signaling, increased protein synthesis, hypoxia, and metabolic imbalance. Although persistent ER stress activates adaptive survival UPR signaling, cancer cells operate near the upper limit of proteostatic tolerance, rendering them particularly vulnerable to additional perturbations [51,52]. This vulnerability has led to the development of ER stress inducers and autophagy regulators as anti-cancer agents. Most studies have focused on tumor cell cytotoxicity. However, successful clinical translation requires understanding how normal cells, particularly vascular endothelial cells, adapt to prolonged proteotoxic stress as abnormal vessels promote hypoxia, impair drug delivery, and accelerate therapy resistance [53].

Classical ER stress inducers such as tunicamycin or thapsigargin broadly disrupt protein glycosylation or ER calcium homeostasis, thereby perturbing ER function in both cancer and normal cells and limiting therapeutic selectivity. Proteasome inhibitors such as bortezomib block ubiquitin-dependent protein degradation but frequently induce compensatory autophagy, which can mitigate proteotoxic stress and contribute to adaptive resistance. Similarly, autophagy inhibitors, including chloroquine derivatives, primarily interfere with lysosomal acidification without directly affecting upstream protein folding and ER-associated protein degradation. Moreover, the role of autophagy in cancer is highly context-dependent, and autophagy inhibition alone does not universally suppress tumor growth, producing divergent effects depending on tumor stage and microenvironment [54].

In contrast, VCP functions as a central protein unfolding ATPase that integrates ER-associated protein degradation, ubiquitin-mediated proteostasis, organelle contact regulation, and autophagosome maturation. Thus, VCP inhibition does not merely induce ER stress or block autophagy in isolation. Rather, VCP inhibition simultaneously collapses multiple interconnected arms of the proteostasis network [5,9]. However, how VCP inhibitors differentially affect tumor cells and endothelial adaptation to proteotoxic stress has remained unclear. Our data suggest that cancer-selective vulnerability to VCP inhibition arises from the convergence of elevated basal ER stress, impaired autophagic flux due to lysosomal dysfunction, and progression to DDIT3-dependent apoptosis. Importantly, endothelial and normal colon-derived cells exhibit greater tolerance to VCP inhibition, consistent with their lower basal proteostatic burden. Accordingly, VCP inhibition suppresses endothelial proliferation while preserving vascular integrity through autophagy induction, thereby enabling tumor-selective cytotoxicity.

Despite similar organelle contact reorganization in both endothelial and cancer cells upon VCP inhibition, the downstream Ca²⁺ fate diverged markedly. In endothelial cells, VCP inhibition induces tolerable ER stress, accompanied by increased ER-mitochondria contacts and controlled opening of the mPTP, leading to elevated cytosolic Ca²⁺ levels that activate autophagy pathways. In parallel, partial ER Ca²⁺ depletion may activate SOCE through ER-PM contacts, facilitating Ca²⁺ replenishment and adaptive stress resolution. In contrast, colorectal cancer cells already experience elevated basal proteotoxic stress. Upon VCP inhibition, further accumulation of misfolded proteins may enhance ER stress and increase ER luminal osmotic pressure, promoting ER Ca²⁺ efflux [34]. Sustained Ca²⁺ release at ER-mitochondria contacts results in persistent mitochondrial Ca²⁺ uptake. We also observed increased AKT phosphorylation in HCT116 cells following NMS-873 treatment (Figure S7N), suggesting that oncogenic signaling may sustain protein synthesis and thereby exacerbate proteotoxic stress.

Moreover, overexpression of mitochondrial Ca²⁺ transport channels such as MCU (mitochondrial calcium uniporter) and VDAC1 in cancers [55,56] may further amplify Ca²⁺ influx into mitochondria and prevent adaptive redistribution. Under these conditions, sustained mitochondrial Ca²⁺ accumulation is associated with bioenergetic dysfunction and reduced cell viability. Collectively, these findings suggest that mitochondrial Ca²⁺ homeostasis is a critical determinant of cell fate under proteostatic stress.

Our study also extends previous research showing that VCP-FAF2/UBXD8 complexes regulate ER-mitochondria contacts by modulating membrane lipid saturation and composition. Whereas proteinopathy-associated VCP mutations chronically increase these contacts [25], we demonstrate that transient pharmacological VCP inhibition reorganizes ER-mitochondria contacts under adaptive contexts. PLA analysis confirmed that NMS-873 reduces VCP-FAF2/UBXD8 interactions (Figure S7O), potentially remodeling membrane lipid composition to transiently enhance ER-mitochondria contacts and induce calcium-dependent autophagy.

Beyond oncology, our findings may have broader implications for diseases associated with proteostasis imbalance. In retinitis models, a degenerative retinal disorder characterized by rhodopsin misfolding and elevated ER stress [57], VCP inhibition has been shown to preserve rhodopsin homeostasis [58–60]. Although those studies did not directly examine autophagy, our results suggest that calcium-dependent autophagy induction could contribute to the cytoprotective effects observed in retinal cells.

In summary, we demonstrate that organelle contact reorganization and Ca²⁺ redistribution determine cell fate under VCP inhibition. By maintaining functional tumor vasculature while selectively targeting cancer cells, VCP inhibition may enhance drug delivery and reduce tumor hypoxia, positioning VCP inhibitors as dual-action therapeutics.

Materials and methods

Cell culture

Human umbilical vein endothelial cells (HUVECs, passages up to 9) were purchased from Lifeline Cell Technology (FC-0003) and cultured in EBM-2 Basal Medium (Lonza Bioscience, CC-3156) with supplements (Lonza Bioscience, CC-4176) and 1% antibiotic-antimycotic (Gibco, 15240062) at 37°C in a humidified incubator with 5% CO₂ and pH 7.4. DDIT3/CHOP-mCherry reporter cells, provided by Dr. Hae-Ryung Park (University of Rochester, USA) and ATG3 knockout HeLa cells, provided by Dr. Seong-Woon Yu (DGIST, Republic of Korea), were cultured in DMEM (Gibco, 11995065) with 10% fetal bovine serum (FBS; Gibco, 16000044) and 1% antibiotic-antimycotic. HCT116 cells (10247) and HT-29 cells (30038) were purchased from the Korean Cell Line Bank (KCLB) and cultured in RPMI 1640 (Gibco, 11875093) with 10% fetal bovine serum and 1% antibiotic-antimycotic. U87MG cells (30014) were purchased from the KCLB and cultured in EMEM (ATCC, 30–2003) with 10% FBS and 1% antibiotic-antimycotic. Cells were tested for mycoplasma contamination every two months.

Plasmids and siRNAs

mRFP-GFP-MAP1LC3B/LC3 plasmid was provided by Prof. Jaewhan Song (Yonsei University, Republic of Korea), and SPLICS vectors were provided by Dr. Dong-Hyung Cho (Kyungpook National University, Republic of Korea). ON-TARGETplus siRNA VCP SMARTPool (Horizon Discovery, L-008727-00-0005) and ON-TARGETplus siRNA ATG7 SMARTPool (Horizon Discovery, L-020112-00-0005) were purchased.

Antibodies

Anti-VCP (Santa Cruz Biotechnology, sc57492); anti-FAF2/UBXD8 (Proteintech, 16251–1-AP); anti-ubiquitin (Santa Cruz Biotechnology, sc8017); anti-HSPA5/GRP78/BiP (Proteintech, 66574–1-Ig); anti-phospho-EIF2A/eIF2α (Cell Signaling Technology, 3398s); anti-EIF2A/eIF2α (Cell Signaling Technology, 2103s); anti-ATF4 (Cell Signaling Technology, 11815s); anti-DDIT3/CHOP (Proteintech, 15204–1-AP); anti-VDAC1 (Abclonal, A19707); anti-ITPR1/IP₃R1 (Santa Cruz Biotechnology, sc271197); anti-ITPR1/IP₃R1-ITPR2/IP₃R2-ITPR3/IP₃R3 (Santa Cruz Biotechnology, sc377518); anti-VAPB (Proteintech, 14477–1-1AP); anti-RMDN3/PTPIP51 (Proteintech, 20641–1-AP); anti-STIM1 (Santa Cruz Biotechnology, sc166840); anti-ORAI1 (Novus, NBP1-85463); anti-phospho-PRKAA/AMPKα (Thr172; Cell Signaling Technology, 2535s); anti-PRKAA1/AMPKα1 (Abcam, ab110036); anti-phospho-MTOR (Ser2448; Cell Signaling Technology, 5536s); anti-MTOR (Invitrogen, 215Q18); anti-TFEB (Proteintech, 13372–1-AP); anti-LMNA/lamin A/C (Santa Cruz Biotechnology, sc376248); anti-GAPDH (Cell Signaling Technology, 2118s); anti-SQSTM1 (Abcam, ab56416); anti-LC3B (Abcam, ab48394); anti-LC3 (MBL Life Science, PM036); anti-PECAM1/CD31 (Abclonal, A19014); anti-MKI67/Ki67 (Abcam, ab16667); anti-VEGFA (Abcam, ab1316); anti-phospho-AKT (Ser473; Cell Signaling Technology, 4060s); anti-AKT (Cell Signaling Technology, 4685s); anti-phospho-EIF4EBP1/4E-BP1 (Thr37/46; Cell Signaling Technology, 9459s); anti-EIF4EBP1/4E-BP1 (Cell Signaling Technology, 9452s); anti-CDH1/E-cadherin (Cell Signaling Technology, 3195s); anti-ACTB/beta actin (Abcam, ab6276, ab8227); secondary antibodies (GenDEPOT, SA001-500, SA002-500); anti-mouse Alexa Fluor 488 (Invitrogen, A11008); anti-rabbit Alexa Fluor 594 (Invitrogen, A11012).

Chemicals and reagents

Chemicals used were: NMS-873 (Xcessbio, M60165); brefeldin A (MedChemExpress, HY-16592); thapsigargin (GLPBIO, GC11482); ER-000444793 (MedChemExpress, HY-100852); cryptotanshinone (MedChemExpress, HY-N0174); chloroquine (MedChemExpress, HY-17589A); rapamycin (Tokyo Chemical Industry, 53123–88-9); STO-609 (MedChemExpress, HY-19805); BAPTA-AM (Sigma-Aldrich, A1076); EGTA (Sigma-Aldrich, E3889); ML-240 (Tocris, 5153); UPCDC-30245 (Sigma-Aldrich, SML1674); 4-phenylbutyric acid (Sigma-Aldrich, P21005).

Reagents used were: DMSO (Sigma-Aldrich, D2650); 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; VWR Chemicals, 0793); DMSO (SAMCHUN, 67–68-5); Trypan Blue solution (Sigma-Aldrich, T8154); ANXA5/annexin V-FITC (BioLegend, 640906); propidium iodide (PI) (BioLegend, 421301); Matrigel (Corning, 356234); Matrigel (Corning, 356231); VEGF/vascular endothelial growth factor (R&D Systems, 11458-VE); 10X PBS (CureBio, P0213); trypsin-EDTA solution (Gibco, 12605036); Accutase (Gibco, A11105-01); Protease and Phosphatase inhibitor Cocktail (Thermo Fisher Scientific, 78441); NP-40 (Merck Millipore, 492016); enhanced chemiluminescence kit (Enzynomics, EOE003S); formaldehyde (Sigma-Aldrich, 252549); 10% neutral buffered formalin solution (Sigma-Aldrich, HT501128); Triton X-100 (Sigma-Aldrich, T8787); mounting medium (Abcam, ab104139); antifade mounting medium (VECTASHIELD, H-1200); Lipofectamine 3000 (Invitrogen, L3000015); Lipofectamine LTX (Invitrogen, 15338100); Lipofectamine RNAiMAX (Invitrogen, 13778150); Rhod-2, AM (Invitrogen, R1244); Fluo-4, AM (Invitrogen, F14201); Mag-Fluo-4, AM (Invitrogen, M14206); MitoTracker Red (Invitrogen, M7512); LysoTracker (Invitrogen, L7528); DQ-BSA (Invitrogen, D12051); Hoechst33342 (Invitrogen, H3570); ATPlite 1-step Luminescence Assay Kit (Revvity, 6016736); NaveniFlex Cell MR RED (Navinci, 60025); Nuclear and cytoplasmic extraction reagents (Thermo Fisher Scientific, 78833); BCA protein assay (GLPBIO, GK10009); Mitochondria isolation kit (Thermo Fisher Scientific, 89874); Genomic DNA extraction kit (Bioneer, K-3032); FastStart SYBR Green Master (Roche, 68711120); bovine serum albumin (BSA; Sigma-Aldrich, A2153); skim milk (BD Difco, 232100); normal goat serum (Abcam, ab7481); NaveniFlex Tissue MR RED (Navinci, 60027).

Cell proliferation assay

Cells were seeded onto 96-well plates and treated with the drugs. Cell proliferation was measured in triplicate using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) at 0.4 mg/mL. MTT formazan was solubilized in DMSO, and absorbance was measured at a wavelength of 595 nm using a Victor Nivo Multimode Plate Reader (PerkinElmer, Waltham, MA).

Trypan Blue assay

Cells were seeded onto 24-well plates and treated with the drugs. Cell viability was measured in triplicate using 0.4% Trypan Blue solution. The cell suspension was mixed with Trypan Blue solution, and viable cells were counted based on dye exclusion.

Fluorescence-activated cell sorting (FACS)

Cells were seeded onto 6-well plates and treated with the drugs. Cells were then harvested and stained with ANXA5/annexin V-FITC and propidium iodide (PI) according to the manufacturer’s instructions. Apoptotic cell populations were quantified by flow cytometry using a CytoFLEX analyzer (Beckman Coulter, Brea, CA).

In vitro capillary tube formation assay

HUVECs were serum-starved overnight, and then seeded onto Matrigel (Corning, 356234)-coated 48-well plate. Cells were treated with the drugs and vascular endothelial growth factor (VEGF). Tubular structures were observed by a microscope with a DP74 camera (Olympus, Tokyo, Japan, IX71). Tube formation was quantitatively evaluated by measuring tube junction counts.

Wound-healing assay

HUVECs were seeded onto SPLScar Blocks (SPL Life Sciences, 201935). After overnight incubation, SPLScar Blocks were detached, and cells were treated with the drugs and VEGF. The migration of cells was analyzed using a microscope (Olympus, Tokyo, Japan, IX71). Quantification of wound closure was performed by measuring the gap distance at 24 h and normalizing it to the initial wound width.

Cellular thermal shift assay (CETSA)

HUVECs were harvested using Accutase solution. Cell suspensions containing 3 × 10⁷ cells/15 mL were aliquoted into two separate conical tubes. DMSO was used as a control, and drugs were added to separate tubes. The cells were incubated in a CO₂ incubator at 37°C for 1 h with gentle shaking. After centrifugation of each sample, the pellet was washed with PBS, resuspended in 1 mL of PBS with protease and phosphatase inhibitor cocktail, aliquoted into PCR tubes (~100 μL per tube) and heated between 55–72°C for 3 min cool-down at 25°C for an additional 3 min in the thermal cycler. The tubes were centrifuged to pellet the cells, and the supernatant was discarded and replaced with 0.4% NP-40 in PBS with protease inhibitors. The cell suspension underwent two cycles of freezing and thawing in liquid nitrogen and was centrifuged at 20,000 g for 20 min at 4°C. The supernatants (soluble proteins) were obtained and used for western blotting.

Western blotting

Cells were seeded onto 12-well plates and treated with the drugs. Soluble proteins were harvested using 2×SDS lysis buffer (0.12 M Tris-Cl, pH 6.8, 3.3% SDS, 10% glycerol, 3.1% DTT). The lysates were separated by SDS-PAGE, transferred to PVDF (Merck Millipore, IPVH00010) membranes, blocked, and incubated overnight at 4°C with primary antibodies at the indicated dilutions. Next, the membranes were probed with HRP-conjugated secondary antibody and detected using an enhanced chemiluminescence kit according to manufacturer’s instructions and a ChemiDoc XRS+ imaging system (Bio-Rad, Hercules, CA). Western blot images were acquired using Image Lab Software version 6.1 and quantified using ImageJ software.

Proximity ligation assay (PLA)

Cells were seeded onto 8-well chambered cell culture slides and treated with the drugs. PLA experiments were performed using the NaveniFlex Cell MR RED according to manufacturer’s instructions. Briefly, cells were fixed with 4% formaldehyde for 10 min and permeabilized with 0.2% Triton X-100 for 20 min at room temperature. Then, cells were blocked with blocking solution for 1 h and then incubated with diluted primary antibodies overnight at 4°C. PLA probes (anti-rabbit PLUS and anti-mouse MINUS) were applied and incubated for 1 h at 37°C. Cells were incubated with ligase for 30 min, then with DNA polymerase for 100 min at 37°C. Samples were mounted with mounting medium. PLA signals were observed using an LSM980 confocal microscope (Zeiss, Oberkochen, Germany). PLA signals quantification was performed by the number of red dots or red fluorescence intensity per cell area.

Calcium imaging

Cells were seeded onto 8-well chamber slides and treated with the drugs. Cells were incubated with Rhod-2 AM, Fluo-4 AM, or Mag-Fluo-4 AM in Krebs-Ringer-HEPES (KRH) buffer (pH 7.4) for 30 min in a CO₂ incubator at 37°C. The cells were then washed with Ca²⁺-free KRH buffer three times (pH 7.4) to de-esterify the dye. Images were captured using an LSM980 confocal microscope (Zeiss, Oberkochen, Germany). Fluorescence intensity per cell area was quantified using ImageJ software.

ATP monitoring luminescence

HUVECs were seeded onto 96-well plates and treated with the drugs. Cellular ATP levels were measured using the ATPlite 1-step Luminescence Assay Kit (Revvity, 6016736) according to the manufacturer’s instructions. Luminescence was measured using a Victor Nivo Multilabel Plate Reader (PerkinElmer, Waltham, MA).

Nuclear fraction

Cells were seeded onto 6-well plates and treated with the drugs. Cells were harvested using trypsin-EDTA or Accutase solution. After centrifugation of each sample, the pellet was washed with PBS. Each sample was then pelleted again, and the supernatant was discarded. Nuclear and cytoplasmic extraction reagents (Thermo Fisher Scientific, 78833) were used according to the manufacturer’s instructions. Nuclear and cytoplasmic total proteins were measured by BCA protein assay (GLPBIO, GK10009) according to the manufacturer’s protocol.

Mitochondrial DNA (mtDNA) and cytosol DNA extraction and qPCR

HUVECs were seeded onto 100-mm culture dishes and treated with the drugs. Cultured cells were harvested using Accutase solution. After centrifugation of each sample, the pellet was washed with PBS. Each sample was then pelleted again, and the supernatant was discarded. A mitochondria isolation kit (Thermo Fisher Scientific, 89874) was used to isolate mitochondria and cytosol according to the manufacturer’s instructions. The mitochondrial pellet was resuspended in PBS, and mtDNA and cytosol DNA were extracted using a genomic DNA extraction kit (Bioneer, K-3032). qPCR was performed using the FastStart SYBR Green Master. Primers used were: Human MT-ND1 forward (5’- CTCTTCGTCTGATCCGTCCT −3‘) and reverse (5’- TGAGGTTGCGGTCTGTTAGT −3‘); Human MT-ND2 forward (5’- GTAGACAGTCCCACCCTCAC −3‘) and reverse (5’- TTGATCCCGTTTCGTGCAAG −3‘); Human RNA18S forward (5’- GTAACCCGTTGAACCCCATT −3‘) and reverse (5’- CCATCCAATCGGTAGTAGCG −3’).

Immunocytochemistry (ICC)

Cells were seeded onto 8-well cell culture slides and treated with the drugs. Cells were fixed with 4% formaldehyde for 10 min and permeabilized with 0.2% Triton X-100 for 10 min at room temperature. Then, cells were blocked with 1% BSA for 1 h and then incubated with diluted primary antibodies overnight at 4°C. After washing with PBS, cells were incubated with Alexa Fluor-conjugated secondary antibodies for 1 h at room temperature and mounted using mounting medium with DAPI. Images were obtained using an LSM980 confocal microscope (Zeiss, Oberkochen, Germany). All fluorescence signal intensity was quantified using ImageJ software.

LysoTracker red staining

Cells were seeded onto 8-well cell culture slides and treated with the drugs. Cells were treated with 100 nM LysoTracker in PBS for 30 min in a CO₂ incubator at 37°C and fixed with 4% formaldehyde. Then, the cells were stained with Hoechst33342. Washing with PBS, samples were mounted with mounting medium. Images were obtained using an LSM980 confocal microscope (Zeiss, Oberkochen, Germany). All fluorescence signal intensity was quantified using ImageJ software.

DQ-BSA staining

Cells were seeded onto 8-well cell culture slides. Cells were pretreated with DQ-BSA (10 μg/mL) in PBS for 1 h in a CO₂ incubator at 37°C. After drug treatments, the cells were fixed with 4% formaldehyde and stained with Hoechst33342. Washing with PBS, samples were mounted with mounting medium. Images were obtained using an LSM980 confocal microscope (Zeiss, Oberkochen, Germany). All fluorescence signal intensity was quantified using ImageJ software.

mRFP-GFP-LC3 plasmid transfection

Cells were seeded onto 6-well plates and transfected with mRFP-GFP-LC3 plasmids using Lipofectamine LTX with Plus reagents for 4 h. After drug treatment, cells were fixed with 4% formaldehyde, stained with Hoechst33342, and imaged using an LSM980 confocal microscope (Zeiss, Oberkochen, Germany).

siRNA-mediated gene knockdown

Cells were seeded onto 12-well plates and transfected with siRNA targeting VCP or ATG7 using Lipofectamine RNAiMAX for 4 h according to the manufacturer’s protocol.

In vivo mouse tumor xenograft assay

The Institutional Animal Care and Use Committee of Yonsei University approved the study protocol (Permission number: IACUC-A-202510-2143–01). Mice were handled in accordance with the Institutional Animal Care and Use Committee and international guidelines for the ethical use of animals. Four-week-old female BALB/c nu/nu mice (Nara-Biotec., Seoul, Republic of Korea) were purchased and acclimated for 1 week to a 12-h light/12-h dark cycle at 23 ± 1°C and 40–60% humidity.

HCT116 cells (5 × 10⁶ cells) suspended in 200 μL of PBS:Matrigel (1:1) were subcutaneously implanted into the dorsal flank of BALB/c nu/nu mice. Once the tumors became palpable (500 mm³), the mice were randomly selected and separated into two groups (four mice per group) and received intratumoral injections of vehicle or NMS-873 (5 mg/kg) solutions. Twenty-four hours after the injections, the mice were sacrificed, and tissue samples were obtained. The tumors were surgically removed, and tissue sections were prepared for histological analysis.

Histological analysis and immunohistochemistry

Tumor tissues were fixed in 10% neutral buffered formalin solution at room temperature for 24 h, rinsed, and embedded in paraffin. Paraffin-embedded tissue sections were processed for hematoxylin and eosin (H&E) staining, immunofluorescence (IF) staining, and proximity ligation assay (PLA). H&E staining was performed by Obenbio, Inc. (Seoul, Republic of Korea). For IF and PLA, paraffin-embedded tissue sections were deparaffinized and rehydrated using xylenes and a graded ethanol series. For cryosectioning, tumor tissues were equilibrated in 30% sucrose (Ducksan Neolux Co., Ltd., 57–50-1), embedded in OCT (Sakura Finetek USA, 4583), and sectioned by Obenbio, Inc. (Seoul, Republic of Korea). For IF, cryosectioned tissues were fixed in 4% formaldehyde.

Tissue sections were permeabilized with 0.1% Triton X-100 and blocked with 3% goat serum at room temperature for 2 h. The sections were then incubated overnight at 4°C with primary antibodies at the indicated dilutions. After washing with PBS, sections were incubated with Alexa Fluor-conjugated secondary antibodies for 1 h at room temperature and mounted using mounting medium with DAPI. Images were using an LSM980 confocal microscope (Carl Zeiss AG, Oberkochen, Germany).

Statistics and reproducibility

All data were analyzed using GraphPad Prism 9.0 (GraphPad Software Inc., San Diego, CA) and presented as mean ± standard deviation (SD) values. Comparison between two groups was done by the unpaired, two-tailed Student’s t-test. For analyses involving three or more groups, one-way or two-way analysis of variance (ANOVA) was used. A minimum of three independent experiments was conducted to allow for statistical comparisons. All immunoblots were quantified from independent biological replicates, and uncropped images are available (Data S1). p-values less than 0.05 were considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; n.s., not significant).

Funding Statement

This work was supported by the National Research Foundation of Korea and was funded by the Government of the Republic of Korea (MSIP; 2021R1A3B1077371), the Brain Korea 21 Plus Project of the Republic of Korea and ICONS (Institute of Convergence Science) and YFL (Yonsei Frontier Laboratory), supported by Yonsei University. This research also received support from the Regional Innovation System & Education (RISE) through the Seoul RISE Center, funded by the Ministry of Education (MOE) and the Seoul Metropolitan Government (2025-RISE-01–022-05).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2026.2677184