Monensin suppresses EMT-driven cancer cell motility by inducing Golgi pH–dependent exocytosis of GOLIM4

Significance

Epithelial-to-mesenchymal transition (EMT) is widely recognized as a cell-intrinsic process that augments the metastatic potential of cancer cells by enhancing their motility and invasiveness. In this study, we demonstrate that the Golgi-targeting compound Monensin effectively inhibits EMT-driven cancer cell motility by disrupting the promigratory GOLIM4–TLN1 axis, a process mediated through the elevation of Golgi pH. These findings underscore the critical role of Golgi pH homeostasis in cancer progression, provide mechanistic insights into how targeting Golgi-associated pathways may serve as a promising therapeutic strategy for combating metastatic cancers, and point to Monensin as a drug with potential for repurposing for clinical therapeutic use.

Abstract

Despite extensive efforts to develop strategies to inhibit cancer metastasis—the leading cause of cancer-related deaths—progress has been limited in recent decades. Epithelial-to-mesenchymal transition (EMT) initiates metastasis by enhancing the migratory capacity and plasticity of cancer cells, enabling them to escape the primary tumor site. Identifying vulnerabilities unique to mesenchymal cancer cells is, therefore, critical for developing effective antimetastatic therapies. Our prior research has highlighted the crucial role of the Golgi apparatus in EMT-driven cancer cell motility and metastasis. In this study, we investigated the antimigratory effects of various Golgi-disrupting compounds and identified Monensin, a polyether ionophore antibiotic, as a potent migration suppressor in mesenchymal non–small cell lung cancer (NSCLC) cells. Monensin treatment increases the pH within the Golgi lumen, inducing rapid exocytosis of the promigratory Golgi scaffold protein Golgi Integral Membrane Protein 4 (GOLIM4). GOLIM4 plays a key role in regulating cell motility and adhesion by modulating the post-Golgi trafficking of Talin 1 (TLN1), an essential focal adhesion component. Furthermore, we found that both GOLIM4 and TLN1 are highly expressed in mesenchymal cancer cells and are direct targets of microRNA-200b, a microRNA that is suppressed during EMT. Treatment with Monensin or depletion of GOLIM4 or TLN1 significantly impaired the migratory activity of mesenchymal NSCLC cells. In summary, this study demonstrates that Monensin exhibits potential antimetastatic activity by disrupting the promigratory GOLIM4–TLN1 axis in mesenchymal NSCLC cells.

Non–small cell lung cancer (NSCLC) continues to be the leading cause of cancer-related mortality worldwide, with the majority of lung cancer patients dying from metastatic disease, highlighting an urgent need to identify the potential vulnerabilities of metastatic cancer cells and to develop effective antimetastatic treatments.

Epithelial to mesenchymal transition (EMT) initiates the transition from a proliferative state to a highly migratory phenotype, marking the pivotal first step in the metastatic cascade (1). The Golgi apparatus is a cell organelle that is responsible for the sorting and secretion of proteins and other macromolecules. Our previous work has demonstrated that Golgi proteins play essential roles in modulating directional cell migration and prometastatic secretion in EMT-driven cancer progression (2–5), which underscores the critical role of the Golgi apparatus in cancer metastasis. Thus far, a number of compounds that target the Golgi apparatus exhibit antitumor activity in cell culture and animal models (6). However, the effects of these Golgi-targeting molecules on cancer cell motility and metastasis, as well as their intracellular targets, remain largely unknown.

Golgi proteins are frequently amplified or upregulated in human cancers, driving the secretion of protumorigenic and prometastatic factors. Several commonly amplified Golgi-associated genes, such as GOLPH3 (7), PITPNC1 (8), and PI4KB (9), have been shown to promote cancer growth and metastasis by enhancing the secretion of oncogenic proteins. Recently, we identified Golgi integral membrane protein 4 (GOLIM4), a type II membrane protein located in the Golgi apparatus, as another frequently amplified Golgi gene in human cancers and crucial for the growth and survival of chromosome 3q-amplified cancer cells (10). GOLIM4 has also been reported to function as a prosurvival oncogene in nasopharyngeal carcinoma (11) and head and neck cancer cells (12). However, the role of GOLIM4 in cancer cell motility remains elucidated.

Focal adhesions (FAs) serve as the primary link between the extracellular matrix (ECM) and the actin cytoskeleton, facilitating cell movement (13). Talin (TLN1) is a key regulator in FA maturation, as its interaction with integrin β subunits triggers a conformational change that activates integrins, enhancing their affinity for ECM ligands and initiating FA formation (14). TLN1 also acts as a scaffold, recruiting additional FA proteins such as vinculin, paxillin, and focal adhesion kinase, all of which contribute to FA growth and maturation (15). Despite TLN1’s crucial role in FA maturation and cancer metastasis, its intracellular trafficking remains largely underexplored.

In this study, we explored the antimigratory activity of the Golgi-targeting compound Monensin in NSCLC cells and revealed its regulatory effect on the promigratory GOLIM4–TLN1 axis.

Results

Monensin Exhibits Antimigratory Activity.

To investigate the potential antimigratory activities of Golgi-targeting molecules, we treated human and mouse mesenchymal NSCLC cells with inhibitors that suppress Golgi acidification (Monensin, Nigericin, and Bafilomycin A1[BafA1]), ER-to-Golgi transport (Brefeldin A[BFA] and Golgicide A), or vesicle transportation (BFA and Ilimaquinone) (Fig. 1A) and examine their effect on cell migratory/invasive activities in Boyden chambers and three-dimension (3D) collagen gels. 24 h treatment with these drugs did not significantly affect cell proliferation (SI Appendix, Fig. S1A), while Monensin, Nigericin, and BafA1 robustly suppress cell motility in both models (Fig. 1 B and C and SI Appendix, Fig. S1B). Monensin exhibited a stronger effect than the other two compounds and its antimigratory effect was further confirmed in a panel of mesenchymal but not epithelial human and murine NSCLC cell lines (Fig. 1 D and E and SI Appendix, Fig. S1 C and D). Moreover, Monensin treatment impaired cell adhesion to plastic and ECM including fibronectin and collagen I (Fig. 1F and SI Appendix, Fig. S1E) and disrupted FA formation (Fig. 1G and SI Appendix, Fig. S1F). Monensin is a polyether ionophore widely used in veterinary medicine (16). Recent studies have demonstrated its potential antitumor activity across various cancer types (17–24), highlighting its promise as a candidate for drug repurposing in oncology.

Fig. 1.

Monensin suppresses cancer cell motility and adhesion. (A) Golgi-targeting compounds. BFA: Brefeldin A; GCA: Golgicide A; Ilimaq: ilimaquinone; BafA1: Bafilomycin A1. (B) three-dimensional invasion of H1299 cell spheres treated with indicated compounds in collagen gel. Invaded single cells were quantified (graph). (C) Transwell migration assay using Boyden chambers. (D) Boyden chamber transwell assays on mesenchymal NSCLC cells treated with indicated doses of Monensin. Migrated and invaded cells were quantified. (E) Scratch wound healing assays on H1299 and A549 cells treated with or without 500 nM Monensin. Wound closure (%) was quantified and normalized to the 0 h time point. (F) Cell adhesion assay on indicated matrix. (G) Confocal images of FAs in A549 cells treated with or without 500 nM Monensin and stained with antibody against phospho-Paxillin (pPXN) (red). (Scale bar: 20 µm.) FA number (Left, n = 50) and size (Right, n = 150) were quantified. Data indicate the mean ± SD from a single experiment incorporating biological replicate samples (n = 3, unless otherwise indicated) and are representative of at least two independent experiments. P values were determined using the one-way ANOVA test (for B and C) or two-tailed Student’s t test (for D–G).

Monensin Decreased GOLIM4 Protein Levels through Elevating Golgi pH.

To identify the Golgi targets of Monensin, we isolated Golgi fractions using the Golgi immunoprecipitation (IP) method (25) from H1299 cells treated with or without Monensin and subjected them to liquid chromatography–mass spectrometry (LC-MS) analysis. The levels of three Golgi-resident proteins (GOLIM4, GLG1, and GGA2) were decreased after Monensin treatment (Fig. 2A and SI Appendix, Table S1). By western blot (WB) analysis, we confirmed that GOLIM4 but not GLG1 or GGA2 protein levels were dramatically decreased by Monensin treatment in a dose-dependent manner (Fig. 2B). Notably, Monensin treatment caused a rapid and sustained decrease in GOLIM4 protein levels (Fig. 2 C and D and SI Appendix, Fig. S2A) without affecting its messenger ribonucleic acid (mRNA) levels (SI Appendix, Fig. S2B).

Fig. 2.

Monensin decreased GOLIM4 protein levels through Golgi deacidification. (A) Schema: Golgi IP assay. Golgi fractions were isolated from H1299 cells treated with or without 500 nM Monensin and subjected to LC/MS for protein quantification. Volcano plot: Proteins identified by LC-MS analysis of Golgi fractions. Results are expressed as a log2 ratio (DMSO/Monensin). y axis: P values; x axis: fold change. Three Golgi-resident proteins that decreased following Monensin treatment are indicated. (B) WB analysis of the indicated Golgi proteins in H1299 cells treated with varying doses of Monensin. Protein levels of GOLIM4, GLG1, GGA2, and GM130 were quantified (graph). (C and D) WB analysis of GOLIM4 protein levels in H1299 cells treated with 500 nM Monensin for the indicated durations (C) and at various time points following drug washout (D). (E) Top: A schematic illustration of the Golgi pH sensor. Bottom: Confocal images of H1299 cells transfected with pH senor and treated with or without Monensin. (F and G) Golgi pH of H1299 cells treated with Monensin (F) or the other two compounds (Nigericin or BafA1) (G). (H) WB analysis of GOLIM4 protein levels in H1299 cells treated with the indicated compounds. Protein levels were quantified by densitometry analysis (graph). (I) Golgi pH in H1299 cells treated with or without 25 mM NH₄Cl. (J) WB analysis of GOLIM4 protein levels in H1299 and A549 cells treated with 500 nM Monensin or 25 mM NH4Cl or vehicle for 8 h. (K) Schema: Golgi pH regulators. Blot: WB analysis of GOLIM4 protein levels in H1299 cells transfected with indicated siRNAs. (L) Pearson correlation between Golgi pH and GOLIM4 protein levels. Data indicate the mean ± SD from a single experiment incorporating biological replicate samples (n = 3, unless otherwise indicated) and are representative of at least two independent experiments. P values were determined using the one-way ANOVA test (for J and H) or two-tailed Student’s t test (for B, F, and I).

Monensin acts as an ionophore to dissipate the proton gradient and increase the pH in the Golgi lumen (26) and GOLIM4 has been reported to be sensitive to Golgi pH changes (27). Thus, we hypothesized that Monensin treatment decreases GOLIM4 protein levels by increasing Golgi pH. To measure intra-Golgi pH values, we utilized a pH sensor that expresses a pH-sensitive green fluorescent protein (pHluorin) and a pH-insensitive mCherry protein in the Golgi lumen (28) (Fig. 2E). The fluorescence intensity of pHluorin, but not mCherry, increased in Monensin-treated cells (Fig. 2E). Using this assay, we confirmed that Monensin treatment elevated Golgi pH (Fig. 2F). In further support of this hypothesis, nigericin and BafA1, which can also deacidify the Golgi (29, 30) (Fig. 2G), decreased GOLIM4 protein levels to a similar extent as Monensin did (Fig. 2H). Furthermore, elevating Golgi pH by treating the cells with weak base ammonium chloride (NH₄Cl) or siRNAs against Golgi pH regulator GPR89A and AE2 mimicked the effect of Monensin on GOLIM4 protein levels (Fig. 2 I–K and SI Appendix, Fig. S2 C and D). By treating H1299 cells with varying doses of Monensin to modulate Golgi pH, we observed an inverse correlation between Golgi pH and intracellular GOLIM4 protein levels (Fig. 2L), suggesting that GOLIM4 protein levels are regulated by Golgi pH changes.

Monensin Induced the Exocytosis of GOLIM4.

It has been reported that GOLIM4 is transported to the lysosome for degradation when cells are exposed to high levels of manganese (10, 31). We hypothesized that Monensin decreases GOLIM4 protein levels by inducing its degradation. However, neither proteasome nor lysosome restored GOLIM4 protein levels in Monensin-treated cells (Fig. 3A). Upon Monensin treatment, GOLIM4 exited the Golgi (SI Appendix, Fig. S3A) but did not accumulate in lysosomes (SI Appendix, Fig. S3B), arguing that GOLIM4 was degraded in the cells. We thus speculated that Monensin triggers the exocytosis of GOLIM4 proteins. Indeed, 1 h after Monensin exposure, GOLIM4 proteins levels started to decrease in the cell lysates and accumulated in the conditioned medium (CM) (Fig. 3B).

Fig. 3.

Monensin triggers GOLIM4 exocytosis. (A) WB analysis of GOLIM4 protein levels in H1299 and A549 cells treated with proteasome (MG-132) or lysosome (Leupeptin and BafA1) inhibitors and Monensin or vesicle. (B) WB analysis of GOLIM4 protein levels in the CM and lysate from H1299 and A549 cells treated with Monensin for the indicated time. (C) WB analysis of GOLIM4 protein levels in the CM and lysate from H1299 cells treated with Golgi secretion (BFA) or exosome (GW4869) inhibitors and Monensin or Vehicle. (D) Confocal micrographs of H1299 cells transfected with mCherry-TGN46 and stained with GOLIM4 antibody and treated with or without Monensin for 1 h. Plots of signal intensity (y axis) against distance (x axis) in single cells. (E) Shema: Isolation of TGN46 positive vesicles from H1299 cells stably transfected with mCherry-TGN46 expression vectors and treated with or without Monensin for 1 h using anti-mCherry antibody and protein G agarose beads. Blot: WB analysis of GOLIM4 and mCherry-TGN46 protein levels in indicated fractions. (F and G) WB analysis of GOLIM4 protein levels in the CM and lysate from H1299 cells transfected with siRNAs against TGN46 (F) or RAB8A (G). (H) Schematic model: Monensin treatment elevates Golgi pH, causing GOLIM4 to be missorted into CARTs, leading to its exocytosis. Data indicate the mean ± SD from a single experiment incorporating biological replicate samples (n = 3, unless otherwise indicated) and are representative of at least two independent experiments.

To determine whether GOLIM4 exocytosis is Golgi- or exosome-mediated, we treated H1299 cells with Golgi secretion inhibitor BFA or exosome inhibitor GW4869 and examined their effect on Monensin-induced GOLIM4 exocytosis. We found that BFA but not GW4869 treatment restored intracellular GOLIM4 protein levels and prevented GOLIM4 secretion, indicating GOLIM4 exocytosis relied on the Golgi secretory pathway (Fig. 3C). The trans-Golgi network (TGN) protein TGN46 plays an essential role for cargo sorting and loading into nascent carriers, such as CARriers of the TGN to the cell Surface (CARTS), at the TGN (32, 33). 30 min after Monensin exposure, GOLIM4 proteins were present in TGN46 positive puncta (Fig. 3D), which was further confirmed by analyzing GOLIM4 protein levels in TGN46-positive vesicles (Fig. 3E). Depletion of TGN46 or another CARTs protein RAB8A with siRNAs partly restored GOLIM4 protein levels and reduced GOLIM4 exocytosis in Monensin-treated cells (Fig. 3 F and G), suggesting that Monensin induced GOLIM4 exocytosis was dependent on the TGN46-mediated exocytosis pathway (Fig. 3H).

GOLIM4 Mediates the Antimetastatic Function of Monensin.

To determine whether GOLIM4 mediates Monensin’s antimigratory functions, we first examined the role of GOLIM4 in lung cancer cell motility and metastasis. GOLIM4 expression levels are higher in metastatic sites than in primary tumors across several cancer types, including lung, colon, kidney, and liver tumors (Fig. 4A) and inversely correlated with LUAD patients’ survival time (Fig. 4B). In orthotopic lung cancer models, GOLIM4 deficient 344LN cells generated smaller tumors and fewer metastasis (Fig. 4 C and D). To exclude the possibility that reduced metastasis is caused by impaired tumor growth, we generated lung tumors from A549 cells transfected with doxycycline (Dox)-inducible GOLIM4 shRNAs and started Dox treatment to induce GOLIM4 depletion 4 wk after tumor cell injection. Depletion of GOLIM4 in a late time point did not affect tumor growth but reduced metastasis to the contralateral lungs by more than twofold (Fig. 4 E and F). In addition, CRISPR/Cas9-mediated GOLIM4 knockout (KO) in A549 cells had minimal impact on primary tumor growth but significantly reduced contralateral lung metastasis in the orthotopic lung cancer model and metastatic colonization in the intravenous injection model (SI Appendix, Fig. S4 A–C). Similar to Monensin treatment, GOLIM4 depletion reduced cell migration in Boyden chamber transwell and scratch wound healing assays (Fig. 4 G and H and SI Appendix, Fig. S4 D and E) and decreased cell adhesion to fibronectin and collagen I (Fig. 4I), with minimal to moderate impact on cell proliferation (SI Appendix, Fig. S4F). siRNA-mediated GOLIM4 depletion in mesenchymal murine NSCLC cells also resulted in a decrease in cell motility. Conversely, overexpression of GOLIM4 enhanced the migratory and invasive activity of epithelial 393P cells (SI Appendix, Fig. S4 I and J). The ectopic expression of GOLIM4 in GOLIM4-null H23 cells resensitized the cells to Monensin treatment (Fig. 4 J and K), while GOLIM4 KO A549 cells were less sensitive to Monensin (Fig. 4 L and M). Moreover, Monensin-induced reduction of GOLIM4 protein levels and inhibition of cell migration were dose-dependent and closely correlated (Fig. 4N and SI Appendix, Fig. S4K). These findings suggest that GOLIM4 is a downstream mediator of Monensin in suppressing cancer cell motility.

Fig. 4.

GOLIM4 mediates the antimigratory functions of Monensin. (A) GOLIM4 mRNA levels in primary lung tumors and distant metastases (https://tnmplot.com/). Box plots represent 33% (lower box) and 66% (upper box). (B) Kaplan–Meier survival analysis of TCGA LUAD cohorts based on GOLIM4 expression levels. (C) WB analysis of GOLIM4 in murine NSCLC 344LN cells transfected with indicated shRNAs. The arrow indicates the GOLIM4 band. (D) Tumor weight (Left) and lung metastasis (Right) generated by the cells in (C). (E) WB analysis of GOLIM4 protein levels in A549 cells transfected with doxycycline (Dox)-inducible shRNAs against GOLIM4 (ishGOLIM4) or control shRNA (ishCTL) and treated with or without Dox. (F) Primary tumor size (Left) and metastasis (Right) generated by the cells in (E). (G) Transwell migration and invasion assays in mesenchymal NSCLC cells transfected with siRNAs against GOLIM4 (siGOLIM4) or control siRNA (siCTL). (H) Scratch wound healing assay. (I) Cell adhesion assay on fibronectin (Left) and collagen I (Right). (J) WB analysis of GOLIM4 protein levels in H23 transfectants treated with or without Monensin. (K) Transwell migration assay in cells in (J). (L) WB analysis of GOLIM4 protein levels in A549 transfectants treated with or without Monensin. (M) Transwell migration assay in cells in (L). (N) Pearson correlation between Monensin concentration and GOLIM4 protein level or relative cell migration in H1299 cells. Data indicate the mean ± SD from a single experiment incorporating biological replicate samples (n = 3, unless otherwise indicated) and are representative of at least two independent experiments. P values were determined using Dunn’s test (B) or one-way ANOVA test (for D, F–I, K, and M).

Previous studies have shown that Monensin suppresses multiple cancer-related pathways, including NFκB and Wnt signaling, and reduces EGFR expression in ovarian and colorectal cancer cells (34, 35). To determine whether Monensin exerts its effects through these pathways, we assessed EGFR protein levels and the activities of NFκB (36) and Wnt/β-catenin (37) reporters. At the doses used in our study, Monensin did not produce significant changes in these pathways (SI Appendix, Fig. S5), suggesting that its inhibitory effect on cell motility is primarily mediated through targeting GOLIM4 rather than modulating these signaling cascades.

GOLIM4 Interacts with TLN1 and Regulates Its Trafficking.

We have previously demonstrated that GOLIM4-regulated secretion promotes the growth and survival of chromosome 3q-amplified NSCLC tumors (10), yet the mechanism by which GOLIM4 regulates the motility of 3q-diploid NSCLC cells remains unclear. CM from control A549 cells did not restore the migratory activity of GOLIM4-depleted A549 cells (SI Appendix, Fig. S6A), suggesting that GOLIM4 enhances cell motility through mechanisms other than secretion modulation. To identify potential mediators associated with GOLIM4, we used a specific antibody to pull down the endogenous GOLIM4 complex and conducted LC/MS analysis to profile its binding partners (Fig. 5A). Several proteins were coimmunoprecipitated with GOLIM4 in control, but not in GOLIM4-deficient H1299 cells (Fig. 5A), indicating that these proteins are potential GOLIM4 interactors. Among them, TLN1 is of particular interest due to its critical role in FA maturation and cell motility (38). The interaction between GOLIM4 and TLN1 was confirmed through an IP-WB analysis (Fig. 5B). In GOLIM4 KO cells, TLN1 failed to localize to FAs (Fig. 5C) or interact with integrin β1 (Fig. 5D). GOLIM4 depletion dramatically reduced the protein levels of TLN1 in the vesicle fractions (Fig. 5E). Importantly, TLN1 is highly expressed in metastatic NSCLC (SI Appendix, Fig. S6B) and TLN1 depletion replicated the effects of GOLIM4 deficiency and Monensin treatment on cell migration/invasion, adhesion, and metastasis (Fig. 5 F–H and SI Appendix, Fig. S6 C–F). Based on these findings, we conclude that GOLIM4 promotes cancer metastasis by regulating TLN1 trafficking (Fig. 5I).

Fig. 5.

GOLIM4 promotes cell motility and adhesion through modulating TLN1 trafficking. (A) Schema: IP using a GOLIM4 antibody in control (siCTL) or GOLIM4-deficient (siGOLIM4) H1299 cell lysates, with coimmunoprecipitated (IP) proteins identified by LC/MS analysis. Graph: Fold changes of identified GOLIM4-binding proteins. (B) IP/WB validation of the interaction between GOLIM4 and TLN1. (C) Confocal images of parental and GOLIM4 KO H1299 cells costained with antibodies against TLN1 and pPXN. Graphs: Colocalization between TLN1 and pPXN (Left, n = 15) and average FA size (Right, n = 50) were quantified. (D) IP/WB analysis of the interaction between TLN1 and ITGB1 in parental and GOLIM4 KO H1299 cells. (E) WB analysis of TLN1 and GOLIM4 protein levels in vesicle and Golgi fractions and whole cell extracts from H1299 cells transfected with indicated siRNAs. (F) WB analysis of GOLIM4 protein levels in A549 cells stably transfected with shRNAs against TLN1 (shTLN1) or shCTL. (G) Primary tumor size (Left) and number of contralateral metastases (Right) generated from the cells in (F). (H) Boyden chamber transwell migration and invasion assay in H1299 and A549 cell transfectants. (I) Schematic model: GOLIM4 regulates cell adhesion and migration by modulating TLN1 trafficking. Data indicate the mean ± SD from a single experiment incorporating biological replicate samples (n = 3, unless otherwise indicated) and are representative of at least two independent experiments. P values were determined using two-tailed Student’s t test (for C) or one-way ANOVA test (for G and H).

Prometastatic GOLIM4–TLN1 Axis Is Activated in Mesenchymal Tumor Cells.

Given the pivotal role of EMT in promoting cancer cell motility and metastasis, we investigated whether GOLIM4 and TLN1 are regulated as part of the EMT program. Interestingly, we found that GOLIM4 and TLN1 expression levels were positively correlated with EMT scores in TCGA LUAD (Fig. 6A), implying that they are highly expressed in mesenchymal lung cancer cells. GOLIM4 and TLN1 mRNA levels were positively correlated with ZEB1 or each other in the TCGA LUAD and LUSC tumors as well as NSCLC cell lines (Fig. 6 B–D and SI Appendix, Fig. S7A). TLN1 and GOLIM4 proteins were highly expressed in mesenchymal human NSCLC cell lines (Fig. 6E). In a panel of murine NSCLC cell lines, GOLIM4 and TLN1 protein levels were higher in mesenchymal compared to epithelial cell lines (SI Appendix, Fig. S7B) and positively correlated with the metastatic and migratory capacities of the cell lines (2) (SI Appendix, Fig. S7 C and D).

Fig. 6.

GOLIM4–TLN1 axis is upregulated in mesenchymal lung cancer cells. (A) Pearson correlations between the expression levels of GOLIM4, TLN1, or EMT markers (CDH1, CDH2, ZEB1, and SNAI1) and EMT scores. (B–D) Pearson correlation between GOLIM4 and ZEB1 (B), TLN1 and ZEB1 (C), or GOLIM4 and TLN1 (D) expression levels in TCGA LUAD and LUSC datasets. (E) WB analysis of TLN1, GOLIM4, and ZEB1 protein levels in human NSCLC cell lines classified as E or M. (F) Top: MicroRNA (miR)-200a and -200b binding sites in the 3′-UTR of GOLIM4. Bottom: miR-182 and miR-200b binding sites in the 3′-UTR of TLN1. (G and H) WB analysis of GOLIM4 protein levels in H1299 cells transfected with miR-200a/b mimics (G) or miR-182/200b (H) mimics or negative control oligos (miR-NC). (I and J) Luciferase activities in H1299 cells cotransfected with GOLIM4 (I) or TLN1 (J) 3′-UTR luciferase reporters and miR mimics or miR-NC. (K and L) qPCR analysis of miR-200b levels (K) and WB analysis of GOLIM4 protein levels (L) in indicated transfectants. (M) Schematic model: GOLIM4 and TLN1 are regulated by the ZEB1–miR-200b axis. Data indicate the mean ± SD from a single experiment incorporating biological replicate samples (n = 3, unless otherwise indicated) and are representative of at least two independent experiments. P values were determined using the one-way ANOVA test (for H and J) and two-tailed Student’s t test (I).

To investigate how GOLIM4 and TLN1 are upregulated in mesenchymal lung cancer cells, we examined their three prime untranslated regions (3′-UTRs) and identified potential miR-200b binding sites in both (Fig. 6F). MiR-200b is an epithelial-specific miRNA whose expression is silenced by mesenchymal transcription factors such as ZEB1/2 and SNAI1/2 (39). Ectopic expression of miR-200b decreased GOLIM4 and TLN1 protein levels (Fig. 6 G and H and SI Appendix, Fig. S7E). miR-200b suppressed GOLIM4 and TLN1 3′-UTR activity, and mutating the miR-200b binding sites abolished these suppressive effects (Fig. 6 I and J), indicating that GOLIM4 and TLN1 are direct targets of miR-200b. Moreover, ectopic expression of EMT inducer ZEB1 in epithelial 393P cells decreased miR-200b levels and increased the expression of GOLIM4 and TLN1, which can be reversed by ectopic expression of miR-200b (Fig. 6 K and L). Thus, both GOLIM4 and TLN1 are up-regulated during EMT (Fig. 7M), potentially explaining the increased reliance on this promigratory GOLIM4–TLN1 axis and the heightened sensitivity of mesenchymal cancer cells to Monensin. To assess whether GOLIM4 or TLN1 contributes to the maintenance of the mesenchymal phenotype in cancer cells, we used siRNAs to deplete their expression and examined EMT marker levels. However, no evidence of a mesenchymal-to-epithelial transition was observed (SI Appendix, Fig. S7 F and G), suggesting that GOLIM4 and TLN1 are not key regulators of EMT.

Fig. 7.

Monensin exhibits potential antimetastatic activity in vivo. (A) Schema: Subcutaneous tumor metastasis model. Murine NSCLC 344SQ cells were injected into syngeneic 129/sv mice to generate subcutaneous tumors and the tumor-bearing mice were treated with indicated doses of Monensin. Graphs: Primary tumor weight (Left) and number of lung metastasis (Right). (B) Hematoxylin and eosin staining of mouse lung sections. Lung metastasis indicated (arrows). (C) Schema: Orthotopic lung tumor metastasis model. Graph: Primary tumor size (Left) and number of metastasis (Right) generated from intrathoracically injected 344SQ cells in syngeneic mice treated with or without Monensin. (D) Primary tumor size (Left) and number of metastasis (Right) generated from intrathoracically injected A549 cells in nude mice treated with or without Monensin. (E) WB analysis of GOLIM4 protein levels in tumor tissues treated with or without Monensin. (F and G) Body weight changes of 129/sv (F) and nude (G) mice during the Monensin or vehicle treatment. Data indicate the mean ± SD from a single experiment incorporating biological replicate samples (n = 3, unless otherwise indicated) and are representative of at least two independent experiments. P values were determined using the one-way ANOVA test (for B, C, and F) or log-rank test (for G).

Monensin Treatment Suppresses Metastasis.

Based on the finding that Monensin suppresses the promigratory GOLIM4–TLN1 axis, we speculate that it may serve as a potential antimetastatic therapy. To evaluate this in vivo, we generated tumors with mesenchymal murine (344SQ) and human (A549) NSCLC cells via subcutaneous, intrathoracic, or intravenous injection, and treated the tumor-bearing mice with Monensin. Subcutaneous 344SQ tumor growth and metastasis were reduced after Monensin treatment in a dose-dependent manner (Fig. 7 A and B). At a lower dose (5 mg/kg), Monensin treatment reduced the metastatic capacity of orthotopic lung tumors generated by 344SQ and A549 cells without affecting tumor growth (Fig. 7 C and D). Consistent with Monensin’s effect on tumor cells, intratumoral GOLIM4 protein levels decreased following Monensin treatment (Fig. 7E). Further reducing the Monensin dosage to 2.5 mg/kg failed to decrease intratumoral GOLIM4 protein levels or suppress metastasis, confirming a dose-dependent effect (SI Appendix, Fig. S8 A and B). In addition, the metastatic capacity of GOLIM4-null H23 tumors was minimally affected by Monensin treatment (SI Appendix, Fig. S8C), suggesting that GOLIM4 is a primary target of Monensin in inhibiting metastasis. Importantly, Monensin treatment demonstrated minimal toxicity, as evidenced by little to no decrease in mouse body weight (Fig. 7 F and G).

Discussion

Despite extensive efforts to develop strategies for inhibiting cancer metastasis, the leading cause of cancer-related deaths, progress has remained unsatisfactory in recent decades. EMT initiates metastasis by enhancing cancer cell migratory capacity (1). Identifying vulnerabilities specific to mesenchymal cancer cells is therefore crucial for developing effective antimetastatic therapies. Our findings reveal that mesenchymal cancer cells express high levels of GOLIM4 and TLN1, supporting the activation of this promigratory GOLIM4–TLN1 axis during EMT-driven metastasis. This axis represents a potential therapeutic vulnerability in mesenchymal cancer cells.

TLN1 regulates integrin interactions with the ECM and plays a crucial role in cancer cell invasion and metastasis (40). Despite its essential role in FA maturation and cell motility, the mechanisms underlying TLN1 transport to FAs remain largely unexplored. In this study, we found that GOLIM4 interacts with and may regulate the intracellular trafficking of TLN1. GOLIM4 depletion reduced TLN1’s incorporation into vesicles and hindered its binding to ITGB1 at FAs. Previously, we demonstrated that GOLIM4 regulates the secretion of prosurvival proteins (such as APP and GGH) by forming a complex with ATP2C1 and GOLPH3 on the trans-Golgi (10). ATP2C1 regulates calcium (Ca)-dependent cargo sorting by pumping Ca into the Golgi lumen, while GOLPH3 drives vesicle budding and secretion. It is likely that GOLIM4 interacts with TLN1 and facilitates its sorting into vesicles, although further experimental validation is required.

Our findings suggest that, at lower doses, Monensin exhibits significant antimetastatic activity. Similarly, Nigericin, another ionophore antibiotic commonly used in veterinary medicine, disrupts the pH balance in the Golgi lumen and reduces GOLIM4 protein levels. Nigericin has also been shown to possess antimetastatic properties in colorectal (41) and ovarian (42) cancers. Therefore, these Golgi-targeting compounds hold promise as potential antimetastatic therapies. Monensin has long been used as an antibiotic in veterinary medicine (43). Recent studies have highlighted its antitumor activity in various cancer types, particularly in EMT-like prostate cancer cells (44) and gastric cancer stem-like cells (45). Despite its therapeutic potential, the known toxicity of Monensin warrants careful consideration. At high doses, Monensin disrupts cellular ion homeostasis, leading to mitochondrial dysfunction, adenosine triphosphate depletion, and cell death—particularly in cardiac and skeletal muscle cells (46). It is highly toxic to certain species, such as horses, and has been linked to muscle weakness, cardiotoxicity, and even fatal outcomes in both animals and humans (47, 48). Therefore, optimizing the dosage and developing targeted delivery strategies will be critical to minimizing off-target effects and achieving a favorable therapeutic window in future translational applications.

The Golgi apparatus is essential for cellular homeostasis, yet the regulation of Golgi protein homeostasis remains largely unexplored. In this study, we found that Monensin reduces intracellular GOLIM4 protein levels by promoting its exocytosis. While we have demonstrated that TGN46 may facilitate the Golgi exit and exocytosis of GOLIM4 following Monensin exposure, the mechanisms by which GOLIM4 is sorted into vesicles remain mysterious. An intriguing question is how changes in Golgi pH trigger the exocytosis of GOLIM4, a protein with a transmembrane domain. One possibility is that in a deacidified Golgi environment, GOLIM4 is cleaved by a protease, producing a fragment that lacks the transmembrane domain and is subsequently incorporated into post-Golgi vesicles for secretion. Supporting this hypothesis, GOLIM4 has been identified as a substrate of the proprotein convertase (PC) PCSK7, and cleavage by PCSK7 leads to its release into the extracellular medium (49). Similarly, another Golgi protein, GP73, which shares structural similarities with GOLIM4, is secreted after cleavage by PC Furin (50). PCs such as Furin are known to be pH-sensitive (51), suggesting that the pH alterations induced by Monensin could potentially activate specific PCs, facilitating the cleavage of GOLIM4 and its release from the Golgi. However, further investigations are required to validate this hypothesis.

There are several limitations to this study. First, the subcutaneous and intravenous injection models are not ideal for evaluating lung cancer metastasis. Employing a spontaneous lung tumor model to assess the impact of disrupting the GOLIM4–TLN1 axis on distant organ metastasis would provide stronger evidence for its prometastatic role. Second, although we demonstrated that Monensin inhibits cancer cell motility by inducing aberrant exocytosis of GOLIM4, additional mediators involved in this process remain to be explored. In summary, our findings suggest that Monensin exerts its effects by modulating Golgi pH and inducing abnormal exocytosis of GOLIM4, which decreases GOLIM4 levels at the Golgi. This disruption impairs TLN1 trafficking, resulting in defective focal adhesion maturation and reduced cancer cell motility. This study identified a promigratory GOLIM4–TLN1 axis and provides a rationale for developing Monensin as a promising antimetastatic therapy.

Materials and Methods

Animal Husbandry.

All mouse studies were conducted in accordance with the guidelines and regulations approved by the Institutional Animal Care and Use Committee at Tulane School of Medicine. Mice received standard care and were killed according to the established protocols of the Institutional Animal Care and Use Committee. For subcutaneous tumor generation, 129/sv and nu/nu mice (5 to 10 mice per group) were subcutaneously injected with 1 × 10⁶ murine and human lung cancer cells in 100 μL phosphate-buffered saline (PBS), respectively. Orthotopic lung tumors were generated by intrathoracic injection of 1 × 10⁶ human lung cancer cells into nu/nu mice. Mice were treated with Monensin (5 or 10 mg/kg, daily) or vehicles (10% DMSO + 90% corn oil) via intraperitoneal injection for a period of 2 or 3 wk. Tumor size and mouse body weight were monitored daily. Necropsies were performed to quantify primary tumor size and assess the numbers of metastases. For orthotopic lung tumors, the long (L) and short (S) diameters of primary lung tumors in the left lungs will be measured using a digital caliper, and tumor size will be calculated using the formula: tumor size = π (L × S)/4 (22). Lung metastases in the contralateral (right) lung will be visually counted. Lung metastasis was confirmed through hematoxylin and eosin staining of lung tissue sections. For intravenous injections, 1 × 10⁵ RFP-tagged A549 cells suspended in 50 µL PBS were injected into the tail vein of nu/nu mice. Four weeks later, lungs were harvested, and pulmonary micrometastases were examined using fluorescence microscopy.

Reagents.

We purchased SYBR Green, fetal bovine serum (FBS), Dulbecco’s minimal essential medium, RPMI Media 1640, Alexa Fluor-tagged secondary antibodies, and DAPI from Life Technologies; puromycin from InvivoGene; paraformaldehyde from Electron Microscopy Sciences; Transwell and Matrigel-coated Boyden chambers from BD Biosciences; G418 from Corning; Monensin (HY-N0150), Nigericin (HY-127019), Bafilomycin A1 (HY-100558), Brefeldin A (HY-16592), Golgicide A (HY-100540), Ilimaquinone (HY-119500), GW4869 (HY-19363) from MedChemExpress; shRNAs against human GOLIM4 (TRCN0000143576 and TRCN0000140441), human TLN1 (TRCN0000123105 and TRCN0000299020), mouse GOLIM4 (TRCN0000126034, TRCN0000126037, and TRCN0000126035); siRNAs against human GOLIM4 (SASI_Hs02_00345751 and SASI_Hs01_00148816), human TLN1 (SASI_Hs01_00226540 and SASI_Hs01_00226541), human TGN46 (SASI_Hs01_00080765), human RAB8A (SASI_Hs02_00339466), mouse GOLIM4 (SASI_Mm01_00026737 and SASI_Mm01_00026738), and siRNA Universal Negative Control #2 (SIC002) from Sigma. We purchased primary antibodies against GOLIM4 (ALX-804-603-C100) from Enzo Life Sciences; GM130 (#560066) from BD Transduction Laboratories; α-tubulin (#T9026) from Sigma; hemagglutinin (HA tag) (#3724 and #2367), β-actin (#4970), GM130 (#12480), phospho-paxillin (Tyr118) (#2541), phospho-EGF Receptor (#2234), Antibody #2234 and Golgin-97 (#13192) from Cell Signaling; GOLIM4 (31083-1-AP), TLN1 (14168-1-AP), ITGB1 (12594-1-AP), TGN46 (13573-1-AP), mCherry (26765-1-AP), EGFR (18986-1-AP), and RAB8A (55296-1-AP) from Proteintech.

Cell Lines.

Human lung cancer cell lines (H292, H322, H2122, H1650, HCC827, A549, H1299, CALU1, CALU6, HCC44, H358, H23, H441, H358, and H1792) were obtained from the American Type Culture Collection. Murine lung cancer cell lines (713P, 393P, 307P, 344P, 412P, 393LN, 531P1, 344P, 344SQ, 344LN, 531LN1, 531LN2, and 531P2) were previously described (52). All cancer cell lines were cultured in RPMI 1640 medium containing 10% FBS. Cells were maintained at 37 °C in a humidified atmosphere with 5% CO₂. Cell transfections were carried out using the jetPRIME Versatile DNA/siRNA transfection reagent (Polyplus). Stable cell transfectants were selected using puromycin (for pLKO.1 vectors) or G418 (for pcDNA3.1 and pEGFP-C3 vectors). GOLIM4 KO A549 cells were generated using the CRISPR-CAS9 system in the Cell-Based Assay Screening Service Core Facility at Baylor College of Medicine. Two guide RNAs (gRNA-1: 5′-ATCTTTGCAGAGCCAACACG-3′; gRNA-2: 5′-CAAGAACTTTCTAAGCTAAA-3′) were used. GOLIM4 KO clones were confirmed by WB analysis.

Vector Construction.

The human GOLIM4 expression vector was previously described (10). TGN46-mCherry was a gift from Michael Davidson (Addgene plasmid # 55145) (National MagLab, Tallahassee, FL). pGL3Luc-5XNF-kappaB was a gift from Esther López-Bayghen (Addgene plasmid # 185695) (Center for Research and Advanced Studies, National Polytechnic Institute of Mexico, Mexico City, Mexico). M50 Super 8× TOPFlash was a gift from Randall Moon (Addgene plasmid # 12456) (University of Washington School of Medicine, Seattle, WA). GalT-mCherry-pHluorin was kindly provided by Cosimo Commisso (Sanford Burnham Prebys Medical Discovery Institute, La Jolla, CA). TMEM115-3 × HA was kindly provided by Monther Abu-Remaileh (Department of Genetics, Stanford University, Stanford, CA). The mouse GOLIM4 coding sequence was isolated by PCR from cDNA prepared from 344SQ cells and then cloned into the pcDNA3.1(-) vector (Invitrogen). The 3′ UTRs of GOLIM4 and TLN1 were amplified by PCR from genomic DNA extracted from H1299 cells and cloned into the pRL-neo vector as previously described (2). Mutations were introduced using the PCR method. The PCR primers used are listed in SI Appendix, Table S2.

Cell Proliferation, Colony Formation, Apoptosis, Migration, and Invasion Assays.

Cell proliferation assays were conducted using the Cell Proliferation Kit II (Sigma), following the manufacturer’s instructions. Migration and invasion assays were performed using Transwell and Matrigel-coated Boyden chambers, respectively, as previously described (3). Scratch wound healing assays and cell adhesion assays were performed as previously described (53). 3D collagen invasion assays were performed as previously described (2).

WB Analysis and IP Assays.

WB analysis was performed as previously described (3). For IP, H1299 cells were transfected with indicated expression vectors, lysed after 48 h in 1× radioimmunoprecipitation assay buffer (RIPA buffer, Cell Signaling), and incubated with antibodies at 4 °C overnight. The immune complex was captured with protein G agarose beads (Cell Signaling), washed with 1 × RIPA buffer once and 1 × PBS three times. The beads were subjected to LC/MS analysis or boiled in 1× sodium dodecyl sulfate loading buffer at 98 °C for 10 min for WB analysis.

qPCR Analysis.

To isolate the total RNA from cells, we utilized the RNeasy Mini Kit (Qiagen). Reverse transcription was carried out using the qScript cDNA superMix (Quanta Biosciences). Genomic DNA was extracted from cells using the DNeasy Blood & Tissue Kits (Qiagen). Gene copy numbers and mRNA levels were assessed using SYBR® Green Real-Time PCR Master Mixes (Thermo Fisher Scientific) and normalized to ribosomal protein L32 (Rpl32) mRNA. The specific PCR primers used in this study are listed in SI Appendix, Table S2.

Golgi and Vesicle Fraction Isolation.

As previously described (54), we enriched cell lysates for Golgi and vesicle fractions using the Minute Golgi Apparatus Enrichment Kit (GO-037, Invent Biotechnologies), following the manufacturer’s instructions.

CM Sample Preparation and Transfer.

Following a previously described protocol (4), CM samples were isolated, filtered using a 0.45-μm filter, and combined with an equal volume of complete growth medium, resulting in a final concentration of 5% FBS. This mixture was then applied to cells and cell migration assays were performed as described above.

Luciferase Reporter Assay.

A549 and CALU1 cells were transfected with NF-κB or Wnt/β-catenin luciferase reporters and treated with the indicated concentrations of Monensin for 24 h. Luciferase activity was measured using the Dual-Luciferase® Reporter Assay System (Promega), according to the manufacturer’s instructions.

LC-MS Analysis.

To identify Monensin-regulated Golgi proteins, 2 × 10⁷ H1299 cells stably transfected with TMEM115-3 × HA were treated with or without 500 nM Monensin and then homogenized using Dounce homogenizer tissue grinders. Golgi fractions were isolated with anti-HA beads (Cell Signaling) following previously described protocols (25). In order to identify proteins that interact with GOLIM4, the endogenous GOLIM4 protein complex was immunoprecipitated using an anti-GOLIM4 antibody (HuaBio) and protein G agarose beads. The proteins bound to the beads were then identified through LC-MS analysis, following the established protocol (4).

Golgi pH Assays.

H1299 cells were stably transfected with GalT-mCherry-EGFP. For pH calibration, the cells were incubated for 15 min with calibration buffers (130 mmol/L KCl, 1 mmol/L MgCl₂, 10 μmol/L Nigericin, 10 μmol/L Valinomycin, and 30 mmol/L MES, pH 5.5 to 6.5). In the experimental conditions, cells were treated with specified compounds or siRNAs. Fluorescence intensities of mCherry and EGFP were measured using the Synergy H1 Multimode Microplate Reader (BioTek), and the ratios of fluorescence intensity (EGFP/mCherry) were calculated. A best-fit curve for the pH standards and fluorescence intensity ratios was generated using nonlinear regression analysis with GraphPad software. The pH values of the experimental samples were then calculated based on this curve.

Microscopy and Image Analysis.

Immunofluorescence procedures were carried out following previously described (9). Cells were imaged using an Eclipse Ti inverted microscope with an A1+ confocal scanner (Nikon, Japan), equipped with diode lasers of 405, 488, 561, and 640 nm wavelengths, high sensitivity Gallium arsenide phosphide and photomultiplier tube detectors, and either a 60× 1.4 NA Oil or 100× 1.45 NA Oil objective. NIS-Elements software (Nikon) version 4.40 (Build 1084) was utilized for image acquisition. Nyquist sampling criteria were followed, and laser power was adjusted to minimize bleaching. Postacquisition, images were processed and deconvolved using Huygens Professional version 18.04 (Scientific Volume Imaging, The Netherlands) with the Classic Maximum Likelihood Estimation algorithm. Image analysis was performed using Fiji (ImageJ version 1.51 s, NIH), Huygens Professional, or NIS-Elements. The number and size of focal adhesions were analyzed as previously described (2).

Statistical Analysis.

Unless stated otherwise, the results shown are representative of replicated experiments and are the means ± SD from triplicate samples or randomly chosen cells within a field. When conducting the correlation analysis and comparing mRNA levels with EMT scores in human lung cancers, the EMT score was calculated as previously described (32). Boxplots represent 5% (lower whisker), 25% (lower box), 50% (median), 75% (upper box), and 95% (upper whisker). Statistical evaluations were carried out with Prism 6 (GraphPad Software, Inc.). Unpaired 2-tailed Student t tests were used to compare the mean values of 2 groups. ANOVA with Dunnett’s test was used for comparing multiple treatments to a control. P values < 0.05 were considered statistically significant. Plots were generated for the respective groups using GraphPad Prism version 9.0.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.