SFXN1 promotes bladder cancer metastasis by restraining PINK1-dependent mitophagy

Abstract

Sideroflexin 1 (SFXN1), a newly identified mitochondrial serine transporter, exhibits great potential to modulate mitochondrial function and promote tumor development. However, its role in bladder cancer (BLCA) remains unclear. Our study revealed that SFXN1 was enriched in clinical BLCA tissues, and high SFXN1 expression in BLCA was positively associated with the progression and poor prognosis. Further, SFXN1 deficiency remarkably suppressed the proliferation and metastasis of BLCA cells in vitro and in vivo, indicating an oncogenic role of SFXN1 in BLCA. Additionally, our results demonstrated that SFXN1 promotes metastasis through its unknown function of restraining PINK1 (PTEN-induced kinase 1)-dependent mitophagy rather than its classical role as a mitochondrial serine transporter to mediate one-carbon metabolism. Mechanistically, SFXN1 acted as a bridge to promote PINK1 degradation by interacting with PARL (presenilin-associated rhomboid-like protein) and MPP-β (mitochondrial processing peptidase-β), leading to mitophagy arrest. Notably, when mitophagy was restrained by highly-expressed SFXN1, mitochondrial reactive oxygen species were considerably enriched, thus activating TGF-β (transforming growth factor-β)-mediated epithelial–mesenchymal transition and promoting metastasis of BLCA cells. This study highlights SFXN1 as a novel promising therapeutic target for BLCA and identifies a new mitophagic modulator to improve our understanding of an association between mitophagy and BLCA progression.

Schematic diagram of the proposed mechanism by which SFXN1 promotes bladder cancer metastasis by restraining PINK1-dependent mitophagy. SFXN1 is upregulated in BLCA tissues, and promotes BLCA metastasis through its unrevealed function of restraining PINK1-dependent mitophagy rather than its classical role as a mitochondrial serine transporter to promote cell proliferation. Specifically, SFXN1 acted as an essential bridging factor to promote PINK1 degradation by interacting with PARL and MPP-β on the IMM, leading to mitophagy arrest and mtROS accumulation, thus activated TGF-β-mediated EMT and promoted BLCA metastasis (This figure was created by Figdraw).

Introduction

Bladder cancer (BLCA) is the tenth most common cancer and the second most prevalent urological malignancy worldwide, accounting for approximately 549,000 new cases and 200,000 deaths annually [1, 2]. Most BLCA cases are classified as urothelial carcinoma, which is further categorized into two clinical stages—non-muscle invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC)—based on the extent of tumor invasion. NMIBC accounts for approximately 75% of all BLCA cases and is typically managed through transurethral resection of the bladder, followed by intravesical chemotherapy instillations, either with epirubicin or mitomycin C, alone or in combination with the Bacillus Calmette–Guérin vaccine [3, 4]. Conversely, the remaining 25–30% of patients with BLCA are diagnosed with MIBC and require a more aggressive treatment approach, including systemic neoadjuvant chemotherapy with cisplatin-based regimens, followed by radical cystectomy [5–7]. Notably, NMIBC exhibits high recurrence and progression rates despite active treatment, whereas MIBC demonstrates rapid local and systemic progression characteristics. Nevertheless, the overall survival rate of patients with BLCA remains unsatisfactory. Therefore, exploring novel carcinogenic mechanisms and therapeutic targets for BLCA is crucial.

Mitochondria, the powerhouse of the cells, facilitate bioenergetic and biosynthetic processes and quickly detect and respond to stress stimuli to maintain cell survival. Under some conditions, functional mitochondria may mediate tumorigenesis by facilitating the import of substrates from the cytoplasm to fuel the electron transport chain, respiration, tricarboxylic acid cycle, fatty acid oxidation, and subsequent macromolecule synthesis [8, 9]. Recent advances in cancer metabolism research have broadened our understanding of metabolic shifts within cancer cell mitochondria [10], highlighting that dysfunctional mitochondria and oxidative stress play crucial roles in tumor progression. Mitophagy, a selective macroautophagy, dominantly maintains mitochondrial homeostasis in cells by eliminating dysfunctional or excess mitochondria via the autophagosome–lysosome system. Notably, deficiency of mitophagy has been reported in various tumor types [11, 12]. Nevertheless, the role of mitophagy in cancer cells is complex, and the specific mitophagy functions and underlying regulatory mechanisms in BLCA cells remain unclear.

SFXN1 (sideroflexin 1) is a member of the mitochondrial transmembrane protein family, which was originally identified as a mutated gene in a flexed-tail mouse model of sideroblastic anemia. Recently, SFXN1 was discovered as a crucial mitochondrial transporter that transports serine into the mitochondria for one-carbon metabolism, thereby playing an important role in promoting cell proliferation [13]. Subsequent studies have revealed other biological functions of SFXN1, such as maintaining the function of Complex III through heme and α-ketoglutarate metabolism, independent of serine transporters [14]. However, despite its emerging novel functions, the role of SFXN1 in BLCA progression and metastasis, and whether and how SFXN1 affects mitophagy remain unexplored.

This study aimed to explore the role of SFXN1 in BLCA. Our findings revealed that SFXN1 expression was markedly upregulated in BLCA tissues compared with paired normal bladder tissues. Notably, increased SFXN1 expression was closely associated with disease progression and an unfavorable prognosis, indicating its oncogenic function in BLCA. Moreover, SFXN1 promoted the metastatic capabilities of BLCA cells in a serine transporter-independent manner through restraining PINK1 (PTEN-induced kinase 1)-dependent mitophagy by mediating PINK1 degradation via PARL (presenilin-associated rhomboid-like protein) and MPP-β (mitochondrial processing peptidase-β). Our findings offer new perspectives on the role and underlying mechanisms of SFXN1 in BLCA, indicating a new strategy targeting SFXN1 to induce mitophagy, thereby suppressing the metastasis of BLCA.

Materials and methods

Clinical samples and bioinformatics analysis

We obtained 155 paraffin-embedded BLCA tumor tissues from Nanjing Drum Tower Hospital, including 21 with corresponding normal tissues, and 13 pairs of fresh tissues (Supplementary Tables S1–S3). The study was approved by the hospital’s ethics committee (2022-493-01), and all participants consented. We analyzed SFXN1 expression in normal vs. tumor tissues and assessed survival outcomes for patients with high/low SFXN1 expression using the UALCAN database. (http://ualcan.path.uab/index.hrml).

For the GSEA, the gene expression dataset of BLCA patients was obtaned from TCGA database (https://xenabrowser.net). The patients were stratified into SFXN1-high and low groups based on median expression value of SFXN1. DEGs (fold change >1, P < 0.05) between these groups was performed using the limma package in R software (v4.5.0). The GSEA analysis of Gene Ontology (GO) terms was subsequently conducted using cluster Profiler package in R software. To validate the association between SFXN1 expression and TGF-β signaling pathway, we systematically evaluated DEGs from two independent datasets: (1) RNA-seq data comparing siSFXN1 and siNC groups; (2) TCGA cohort data comparing SFXN1-high and low groups.

Cell lines and cell cultures

The BLCA cell lines T24 and J82, obtained from the Cell Bank of the Chinese Academy of Science (Shanghai, China), were verified through STR DNA profiling. They were cultured in McCoy’s 5A (Wisent Biotech, 317-010) and EMEM (Wisent Biotech, 320-006) media with 10% FBS (Vazyme Biotech, F101-01) and 1% penicillin/streptomycin (Wisent Biotech, 450-201) at 37 °C and 5% CO2. For serine deprivation experiment, cells were grown in serine-free RPMI 1640 medium from Univ Biotech. In formate rescue experiments, 2 mM formate (Solarbio, S6060) was added to the media during seeding. For inhibiting autophagy with CLQ (50 μM, 24 h; Sigma-Aldrich, C6628), inhibiting mitophagy with Mdivi-1(25 μM, 24 h; MCE, HY-15886) and eliminating mtROS with mitoTEMPO (25 μM, 2 h; Sigma-Aldrich, 1334850-99-5), cells were pretreated before experiments. For inducing mitophagy with CCCP, 10 μM CCCP (40333ES60, Yeasen) was added to the media for 24 h before experiments.

Immunohistochemistry

IHC analysis of paraffin-embedded tissue samples was performed in line with the protocols previously outlined by the manufacturer [15]. The area and density of the stained regions were measured with Image-Pro Plus version 6.0 software (Media Cybernetics, Maryland, USA) to calculate the integrated optical density (IOD) value for the IHC sections.

SiRNA and plasmids interference

siRNAs targeting human SFXN1 and PINK1, produced by Generay (Shanghai, China), were used to transfect T24 and J82 cells grown to 60% confluence in 6-well plates with INTERFERin® reagent (Polyplus-Sartorius, 409-01), following the manufacturer’s instructions, with sequences detailed in Supplementary Table S4.

For SFXN1 overexpression, the expressing plasmids of SFXN1 produced by Youbio (Changsha, China) and transfected T24 and J82 cells following the manufacturer’s instructions of Lipofectamine™ 3000 Transfection Reagent (ThermoFisher Scitific, L3000001).

Cell proliferation assay

The MTT assay was used to assess proliferation in T24 and J82 cells. After seeding 96 h, 10 μL MTT (Sangon Biotech, A600799) was added, followed by 2 h of incubation. Medium was removed, and 100 μL DMSO (Sangon Biotech, A460700) was added to dissolve formazan crystals. Absorbance at 490 nm was measured using the Infinite® 200 PRO reader (Tecan, Switzerland) after a 10-min incubation at 37 °C.

Transwell assay

The mentioned cells were seeded in the upper chamber of an 8 μm transwell (Jet Biofil, Guangzhou, China) at 1 ×10⁵ per well and incubated for 6 h in SFXN1 overexpressing cells and 12 h in SFXN1 knockdown cells. The cell that migrated to the lower membrane were stained with crystal violet (Beyotime, C0121), visualized, and quantified in three distinct randomly selected fields to ensure accuracy and reproducibility. A similar procedure was employed for invasion analysis, but with a modification of the transwell apparatus. Notably, the upper chamber was coated beforehand with an extracellular matrix gel (Corning, 56234).

Wound healing assay

Cells were seeded in 6-well plates for uniform growth and confluence. Once confluent, they were scratched with a 200 μL pipette tip, washed with PBS to remove debris, and incubated in serum-free medium. Images were taken at 0 h and 6 h in SFXN1 overexpressing cells and 12 h in SFXN1 knockdown cells to evaluate wound healing and cell migration.

Tumor xenografts in nude mice

Stable SFXN1 (5’-GCTGCTGCTAATTGCATTAAT-3’) knockdown in T24 cells was achieved using shRNA lentiviral vectors from GeneChem (Shanghai, China). Six-week-old nude BALB/c mice were divided into groups for subcutaneous tumor (n = 8) and popliteal lymphatic metastasis models (n = 10). ShNC or shSFXN1 T24 cells were injected subcutaneously (3.0×10⁶ cells/mouse), and after 40 days, tumors were harvested for weight, protein extraction and paraffin embedding, and then analyzed by WB and IHC. For the popliteal lymphatic metastasis model, luciferase-expressing shNC or shSFXN1 T24 cells were injected into footpads, and imaging was performed 40 days later with Tanon’s imaging system (Shanghai, China). Mice were sacrificed, and popliteal lymph nodes (LNs) were evaluated blindly for volume, harvested, and prepared for IHC. All animal procedures followed the Institutional Animal Care and Use Committee guidelines at Nanjing Drum Tower Hospital.

Western blot analyses

For WB, cells and mitochondria (Beyotime, C3601) were lysed and isolated. Lysates were used for immunoblotting with antibodies listed in Supplementary Table S5.

Co-immunoprecipitation analyses

For Co-IP, T24 and J82 cell lysates were incubated with specific antibodies (Supplementary Table S5). After overnight incubation, 20 μL of protein A/G agarose beads (Invitrogen, 80106G) were added for 20 min to form immune complexes. Beads were washed five times to remove nonspecific bindings, and precipitated proteins were analyzed by immunoblotting.

Immunofluorescence analyses

For immunofluorescence of mitochondrial and LC3B colocalization, T24 and J82 cells were cultured in chambers and labeled with MitoTracker Deep Red (YEASEN Biotech, 40743ES50) and LC3B primary antibodies (Abconal, A5618) overnight at 4 °C, followed by secondary antibodies (Invitrogen, A-11008) with fluorophores for one hour. For mitochondrial and lysosome colocalization, MitoTracker Deep Red and LysoTracker Green DND-26 (YEASEN Biotech, 40738ES50) were used. Confocal images were captured, and colocalization was quantified using ImageJ software (NIH, Maryland, USA).

qPCR assays

Total RNA from T24 and J82 cells was extracted with TRIzol reagent (Vazyme Biotech, R401-01) and 1 µg was reverse transcribed to cDNA using HisyGo RT Red SuperMix (Vazyme Biotech, RT101-01) for qPCR. qPCR was conducted with SupRealQ Purple SYBR qPCR Master Mix (Vazyme Biotech, Q412-02), with primers listed in Supplementary Table S6.

RNA sequencing

Total RNA from T24 and J82 cells was extracted with TRIzol and library prep was done using the KCTM Stranded mRNA Kit for Illumina (catalog no. DR08402, Seqhealth, Wuhan, China). Raw sequencing data were cleaned with Trimmomatic v0.36 to remove low-quality reads and adapters, then aligned to the human genome using STRA v2.5.3a for accurate analysis. Initial data analysis involved meticulous filtering with Trimmomatic v0.36, followed by alignment with STRA v2.5.3a using preset parameters.

Measurement of mitochondrial mass, membrane potential, and ROS production

T24 and J82 cells, with or without SFXN1 knockdown, were seeded and stained with MitoTracker Deep Red (for total mitochondrial mass, YEASEN Biotech, 40743ES50), ΔΨm-specific fluorescent probe JC-1 (for mitochondrial membrane potential, Beyotime, C2003S), and MitoSOX (for mitochondrial ROS, YEASEN Biotech, 40738ES50). Analysis of the cells was performed using the NovoCyte flow cytometer (Agilent Technologies, California, USA).

Transmission electron microscopy

T24 and J82 cells, with or without SFXN1 knockdown or overexpression, were prepared for examination via high-resolution transmission electron microscopy according to previous study [16].

Statistical analysis

Data analysis was conducted with GraphPad Prism software, version 10 (GraphPad, San Diego, USA). Results from at least three separate experiments were evaluated quantitatively, reported as mean ± standard deviation. The Student’s t test was applied to continuous variables, with statistical significance defined as P < 0.05.

Results

SFXN1 expression is associated with the progression and outcomes in patients with BLCA

We compared SFXN1 expression between BLCA tissues and adjacent normal bladder tissues obtained from patients at the Nanjing Drum Tower Hospital using western blotting (WB) and immunohistochemical (IHC) analyses to elucidate the pivotal role of SFXN1 in BLCA progression. The results consistently revealed elevated SFXN1 levels in BLCA tissues compared with their normal counterparts (Fig. 1A, B). We then performed IHC analysis using a larger cohort. The results revealed a significant association between elevated SFXN1 levels and advanced T or N stages in BLCA tissues (Fig. 1C, D). Kaplan–Meier survival analysis based on the IHC scoring of SFXN1 underscored an association between higher SFXN1 expression and reduced overall survival and progression-free survival among the patients with BLCA (Fig. 1E, F). Subsequently, we used The Cancer Genome Atlas database to validate the mRNA expression of SFXN1 in normal bladder and BLCA tissues. Notably, a marked upregulation of SFXN1 mRNA was observed in BLCA tissues (Fig. 1G). Moreover, a similar finding was observed in the survival analysis, indicating an association between elevated SFXN1 expression and an unfavorable prognosis (Fig. 1H). Therefore, our comprehensive findings illustrated a positive association between increased SFXN1 expression and aggressive progression and the adverse prognosis of BLCA, underscoring its potential as a significant biomarker and therapeutic target in this disease.

Fig. 1. SFXN1 expression is associated with the progression and outcomes in patients with BLCA.

A Representative images (left) and quantified results (right) of IHC staining of SFXN1 in matched BLCA tissues and adjacent noncancerous tissues (n = 21). The rectangles in the top row indicate the enlarged areas shown at the bottom. Scale bar: 250 μm (top) and 50 (bottom) μm. B WB analysis of SFXN1 levels in matched BLCA tissues and adjacent noncancerous tissues (n = 13). C Representative images (left) and quantified results (right) of IHC staining of SFXN1 in BLCA tissues with different T stages. (T1 stage: n = 39, T2 stage: n = 54, T3 stage: n = 43, T4 stage: n = 12). The rectangles in the top row indicate the enlarged areas shown at the bottom. Scale bar: 250 μm (top) and 50 (bottom) μm. D Representative images (left) and quantified results (right) of IHC staining of SFXN1 in BLCA tissues with different N stages. (N0 stage: n = 116, N1 stage: n = 25, N2 stage: n = 14). The rectangles in the top row indicate the enlarged areas shown at the bottom. Scale bar: 250 μm (top) and 50 (bottom) μm. The overall (E) and progression-free survival (F) curves of BLCA patients with low and high SFXN1 expression are generated using the Kaplan–Meier survival analysis. G The levels of SFXN1 mRNA in BLCA tissues and normal bladder tissues from TCGA database. (Normal: n = 19, Tumor: n = 408). H The overall survival curves of BLCA patients from TCGA database with low and high levels of SFXN1 mRNA are generated using the Kaplan–Meier survival analysis (n = 406). The P value was assessed using paired, two-tailed Student’s t test (A and B), unpaired, two-tailed Student’s t test (C, D and G) and the log-rank test (E, F and H). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

SFXN1 promotes the proliferation and metastasis of BLCA cells

We performed both in vitro and in vivo experiments to evaluate the functional roles of SFXN1 in BLCA cells using small interfering RNA (siRNA) and overexpression plasmids respectively to knockdown and overexpress SFXN1 in T24 and J82 cells. Quantitative real-time polymerase chain reaction (qPCR) and WB assays confirmed the knockdown efficiency of siSFXN1 (Fig. 2A, B) and the overexpressing efficiency of SFXN1 overexpressing plasmid (Supplementary Fig. S1A, B) in T24 and J82 cells. MTT assay revealed that SFXN1 deficiency significantly reduced the proliferative ability of BLCA cells (Fig. 2C). Wound healing and transwell assays further confirmed the decreased migration ability of cells transfected with siSFXN1 compared with control cells (Fig. 2D, E). The invasion assays also confirmed distinctly impaired invasive functions of T24 and J82 cells following SFXN1 knockdown (Fig. 2F). In contrast, SFXN1 overexpress resulted in significantly increased proliferative, migration and invasion ability of T24 and J82 (Supplementary Fig. S1C–F).

Fig. 2. Knocking down SFXN1 inhibits proliferation and metastasis of BLCA cells in vitro and in vivo.

The knock down efficiency of siSFXN1 in BLCA cells were confirmed by WB (A) and qPCR (B) analysis. C Proliferation of BLCA cells transfected with siSFXN1 or siNC analyzed by MTT assays (n = 3). D Representative images (left) and quantified results (right) of wound healing assays in BLCA cells transfected with siSFXN1 or siNC (n = 3). E Representative images (left) and quantified results (right) of transwell assays in BLCA cells transfected with siSFXN1 or siNC (n = 3). F Representative images (left) and quantified results (right) of invasion assays in BLCA cells transfected with siSFXN1 or siNC (n = 3). G, H The knock down efficiency of shSFXN1 lentivirus in T24 cells were confirmed by WB (G) and qPCR (H) analysis. I Representative images (left) and quantified results (right) of enucleated subcutaneous tumors (n  = 8 per group). J Representative images of HE and IHC staining of SFXN1 or Ki-67 in subcutaneous tumors (n  = 5 per group). Scale bar: 250 and 50 μm. K Representative images of the nude mouse model of popliteal LN metastasis. L Representative bioluminescence images (left) and quantified results (right) of popliteal LN metastasis (n = 10 per group). M Representative images (left) and quantified results (right) of enucleated popliteal LNs (n = 10 per group). N Representative images of HE and IHC staining of lusiferase in popliteal LN (n = 10 per group). The rectangles in the left row indicate the enlarged areas shown at the right. Scale bar: 500 μm (left) and 125 μm (right). The P value was assessed using two-tailed Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Furthermore, we used BLCA xenografts to investigate the effects of SFXN1 on tumor growth and metastasis in vivo. We used shSFXN1 and shNC lentiviruses to individually construct a stable SFXN1-deficient and negative control T24 cell line. The knockdown efficiency is shown in Fig. 2G, H. In subcutaneous tumor models, SFXN1 silencing restrained tumor growth in mice in the shSFXN1 group compared with that in the shNC group (Fig. 2I). Further pathological analysis of subcutaneous tumors revealed decreased SFXN1 levels and attenuated signals for cell proliferation (Ki-67 index) in tumors of the shSFXN1 group compared with those of the shNC group (Fig. 2J). We also constructed a popliteal lymphatic metastasis model by implanting luciferase-labeled shSFXN1 or shNC T24 cells into the footpads of nude mice (Fig. 2K). In vivo imaging revealed decreased luminescence intensity in the popliteal space of mice in the shSFXN1 group compared with the shNC group (Fig. 2L). Additionally, smaller popliteal lymph nodes (LNs) and lower LNs metastasis rates (Fig. 2M, N) were observed in the shSFXN1 group than in the shNC group, indicating that SFXN1 knockdown could inhibit LN metastasis in BLCA cells. Overall, these findings demonstrated that SFXN1 plays a critical role in promoting the proliferation and metastasis of BLCA cells in vitro and in vivo.

SFXN1 promotes the metastasis of BLCA cells independent of its serine transporter function

SFXN1 was recently identified as the main mitochondrial serine transporter in human cells, and it is crucial for one-carbon metabolism and plays a pivotal role in cancer cell proliferation [13]. In the present study, we treated SFXN1-deficient BLCA cells with formate, a product of the mitochondrial one-carbon pathway, in culture medium supplemented with or without serine to investigate whether SFXN1 promotes the proliferation and metastasis of BLCA cells based on its function as a mitochondrial serine transporter. Aligning with previous studies [13, 17, 18], SFXN1 deficiency inhibited the proliferation of BLCA cells, and this suppressive effect on proliferation was greatly enhanced in serine-free medium (Fig. 3A, B), as cell proliferation may be more dependent on the mitochondrial serine metabolism in the absence of the cytosolic one-carbon pathway in serine-free media. In addition, the attenuated proliferative abilities of SFXN1-deficient cells were significantly rescued by the addition of formate, indicating that the promotive effect of SFXN1 on proliferation was primarily dependent on its mitochondrial serine transporter function (Fig. 3A, B). However, the suppressive effects of siSFXN1 on the migration and invasion abilities of BLCA cells were not altered by removing exogenous serine and could not be restored after simultaneous formate supplementation (Fig. 3C–H). Therefore, SFXN1 may enhance the metastasis of BLCA cells through a mechanism other than its classical role as a mitochondrial serine transporter.

Fig. 3. SFXN1 promotes the metastasis of BLCA cells independently of its serine transporter function.

Proliferation of BLCA cells transfected with siSFXN1 or siNC cultured in full (A) and serine-free (B) media supplemented with or without 2 mM formate analyzed by MTT assays (n = 3). Representative images (left) and quantified results (right) of transwell assays in BLCA cells transfected with siSFXN1 or siNC cultured in full (C) and serine-free (D) media supplemented with or without 2 mM formate (n = 3). Representative images (left) and quantified results (right) of wound healing assays in BLCA cells transfected with siSFXN1 or siNC cultured in full (E) and serine-free (F) media supplemented with or without 2 mM formate (n = 3). Representative images (left) and quantified results (right) of invasion assays in BLCA cells transfected with siSFXN1 or siNC cultured in full (G) and serine-free (H) media supplemented with or without 2 mM formate (n = 3). The P value was assessed using two-tailed Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

SFXN1 promotes the metastasis of BLCA cells by inhibiting mitophagy

We performed transcriptome sequencing using SFXN1-deficient (siSFXN1) and SFXN1-wild-type (siNC) T24 cells cultured in complete or serine-free media to investigate the specific biological processes and mechanisms through which SFXN1 promotes the metastasis of BLCA cells. SFXN1 knockdown could change the gene profiles of T24 cells, and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of the differentially expressed genes (DEGs) between the siSFXN1 and siNC groups revealed the mitophagy pathway as the most enriched pathway, regardless of whether the cells were cultured in complete or serine-free media (Fig. 4A, B). Notably, mitophagy is associated with tumor metastasis [19, 20]. As SFXN1 is a mitochondrial inner membrane protein [13, 14], we hypothesized that SFXN1 might participate in regulating cancer cell mitophagy to promote BLCA metastasis. Accordingly, we first examined the colocalization of mitochondria and autophagosome marker LC3B-II or lysosomes (Fig. 4C, D). SFXN1 deficiency distinctly increased the area of colocalization of the mitochondria and LC3B-II (Fig. 4C, E), as well of the mitochondria and lysosomes (Fig. 4D, F). Further investigation of the LC3B-II levels in the presence or absence of chloroquine (CLQ), an autophagy inhibitor that blocks the fusion of autophagosomes with lysosomes, revealed an elevated LC3B-II to LC3B-I ratio in cells following SFXN1 knockdown (Fig. 4G). While, overexpressing SFXN1 inhibited the CCCP-induced accumulation of LC3B-II (Supplementary Fig S2A). Furthermore, LC3B-II accumulated in the intact mitochondrial fraction isolated from SFXN1-deficient BLCA cells (Fig. 4G), suggesting that SFXN1 deficiency boosted the total mitophagic flux in BLCA cells.

Fig. 4. SFXN1 deficiency promotes mitophagy of BLCA cells.

KEGG pathway enrichment analysis of the differently expressed genes between T24 cells transfected with siSFXN1 or siNC cultured in full media (A) and serine-free media (B). C, D Representative images (left) and the pixel intensity of red (mitochondria) and green (LC3B-II or lysosome) (right) in BLCA cells transfected with siSFXN1 or siNC. Scale bar: 35 and 2 μm. Quantified results of the co-localization mitochondria and LC3B (E) or lysosomal (F) in BLCA cells transfected with siSFXN1 or siNC (n = 3). G WB analysis of LC3B levels in whole-cell lysates (WCL) (left) and mitochondrial fractions (right) of BLCA cells transfected with siSFXN1 or siNC treated with or without chloroquine (CLQ). The numbers below the lanes represent the densitometry analysis of band intensity, relative to the ratio of LC3B-II to LC3B-I. H Representative images of transmission electron microscopy of BLCA cells transfected with siSFXN1 or siNC. Arrows indicate mitochondria (blue), autolysosome (red) and autophagosome (green). Scale bar: 400 nm. I Representative images (left) and quantified results (right) of mitochondrial mass of BLCA cells transfected with siSFXN1 or siNC measured by flow cytometry (n = 3). J Representative images (left) and quantified results (right) of mitochondrial ROS of BLCA cells transfected with siSFXN1 or siNC measured by flow cytometry (n = 3). K Representative images (left) and quantified results (right) of mitochondrial membrane potential of BLCA cells transfected with siSFXN1 or siNC measured by flow cytometry (n = 3). The P value was assessed using two-tailed Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We further investigated mitochondrial morphology using transmission electron microscopy and found that damaged mitochondria, characterized by mitochondrial matrix swelling and vacuolar structures, accumulated in BLCA cells following SFXN1 knockdown (Fig. 4H). Accordingly, overexpressing SFXN1 inhibited the CCCP-induced changes of mitochondrial morphology (Supplementary Fig S2B). Moreover, the total mitochondrial mass was significantly reduced in SFXN1-deficient BLCA cells (Fig. 4I). Mitochondrial reactive oxygen species (mtROS) and mitochondrial membrane potential are key indicators of mitochondrial activity [21], SFXN1 deficiency significantly reduced mtROS and mitochondrial membrane potential in BLCA cells (Fig. 4J, K), indicating that SFXN1 deficiency may induce mitophagy to clear abnormal mitochondria, thus reducing mtROS levels to maintain mitochondrial homeostasis. Subsequently, we treated BLCA cells with CLQ and Mdivi-1, a mitophagy-specific inhibitor [22], to confirm whether SFXN1 promotes BLCA metastasis by inhibiting mitophagy. Notably, SFXN1 knockdown did not restrain the migration and invasion of BLCA cells when mitophagy was blocked by CLQ and Mdivi-1 treatment (Fig. 5A–C and Supplementary Fig. S2C–E), confirming that SFXN1 promoted BLCA metastasis by inhibiting mitophagy.

Fig. 5. SFXN1 promotes the metastasis of BLCA cells by inhibiting mitophagy.

A Representative images (left) and quantified results (right) of wound healing assays in BLCA cells transfected with siSFXN1 or siNC treated with or without Mdivi-1 (n = 3). B Representative images (left) and quantified results (right) of transwell assays in BLCA cells transfected with siSFXN1 or siNC treated with or without Mdivi-1 (n = 3). C Representative images (left) and quantified results (right) of invasion assays in BLCA cells transfected with siSFXN1 or siNC treated with or without Mdivi-1 (n = 3). The P value was assessed using two-tailed Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

SFXN1 inhibits PINK1-dependent mitophagy by mediating PINK1 degradation via PARL/MPP-β

The PINK1–PARKIN-dependent mitophagy removes the dysfunctional mitochondria [23–25], wherein PINK1 stabilizes on the outer mitochondrial membrane (OMM) and recruits PARKIN, an E3 ubiquitin ligase that ubiquitinates multiple OMM proteins, leading to mitophagy [26, 27]. We investigated whether SFXN1 suppressed mitophagy via the PINK1–Parkin pathway. The WB assays revealed increased PINK1 abundance in the whole-cell lysate following SFXN1 knockdown (Fig. 6A), along with the accumulation of PINK1 and PARKIN in the mitochondrial fraction (Fig. 6B). In addition, the total ubiquitination of mitochondrial proteins and LC3B-II levels in the mitochondria were significantly enhanced following SFXN1 knockdown (Fig. 6B, C). PINK1–PARKIN - dependent mitophagy can be induced by CCCP [28]. As shown in Supplementary Fig. S3A, B, overexpressing SFXN1 inhibited the CCCP-induced accumulation of PINK1, PARKIN and the total ubiquitination of mitochondrial proteins. Moreover, reduced SFXN1 expression did not alter mitophagy-associated protein levels in BLCA cells lacking PINK1 (Fig. 6C and Supplementary Fig. S4A), suggesting that SFXN1 deficiency might facilitate PINK1–PARKIN mitophagy by stabilizing PINK1 on the OMM. Furthermore, IHC confirmed an increase in PINK1 and PARKIN abundance in subcutaneous tumor tissues following SFXN1 knockdown in a mouse subcutaneous tumor model (Fig. 6D). WB analyses also revealed an increase of PINK1 abundance in SFXN1-knockdown tumors (Supplementary Fig. S4B).

Fig. 6. SFXN1 inhibits PINK1-dependent mitophagy by mediating the degradation of PINK1 via PARL/MPP-β.

A WB analysis of PINK1 levels in whole-cell lysates (WCL) of BLCA cells transfected with siSFXN1 or siNC. B WB analysis of protein levels in the mitochondrial fractions of BLCA cells transfected with siSFXN1 or siNC. C WB analysis of protein levels in the mitochondrial fractions of BLCA cells transfected with siNC, siSFXN1, siPINK1, or siSFXN1 + siPINK1. D Representative images of IHC staining of PINK1 and PARKIN in subcutaneous tumors (n  = 5 per group). Scale bar: 250 and 50 μm. WB analysis of PINK1 and SFXN1 levels in the products pulled down by IgG (as a control), anti-SFXN1 (E), or anti-PINK1 (F) antibodies in BLCA cells. G WB analysis of Tim23, Tom40, MPP-β, PARL and SFXN1 levels in the products pulled down by IgG (as a control), and anti-SFXN1 antibodies in BLCA cells. H WB analysis of MPP-β, PARL and SFXN1 levels in the products pulled down by IgG (as a control), or anti-MPP-β (left), or anti-PARL (right) antibodies in BLCA cells. I WB analysis of MPP-β, PARL and PINK1 levels in the products pulled down by IgG (as a control), or anti-PINK1 antibodies in BLCA cells transfected with siSFXN1 or siNC. Right (Input), WB analysis of MPP-β, PARL and PINK1 levels in cells as above, without co-IP. J WB analysis of full-long and cleaved PINK1 levels in the mitochondrial fractions of BLCA cells transfected with siSFXN1 or siNC. The P value was assessed using two-tailed Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Under normal conditions, PINK1 is regularly degraded by translocating from the OMM to the inner mitochondrial membrane (IMM) by TOM40 (translocase of the outer mitochondrial membrane 40) and TIM23 (translocase of the inner mitochondrial membrane 23). Thereafter, PINK1 is cleaved by MPP-β (mitochondrial processing peptidase-β) and PARL and subsequently degraded via the N-terminal regular degradation pathway. Owing to the localization of SFXN1 on IMM, SFXN1 may play a role in PINK1 translocation from OMM to IMM or may mediate the PINK1 degradation by interacting with PARL or MPP-β. Co-immunoprecipitation (Co-IP) results confirmed the interaction between endogenous SFXN1 and PINK1 (Fig. 6E, F). However, TOM40 and TIM23, the core components of translocase complexes, were not enriched in the products pulled down by SFXN1 antibody, whereas endogenous MPP-β and PARL were highly precipitated by SFXN1 antibody (Fig. 6G). Further Co-IP assays performed using MPP-β or PARL antibodies confirmed SFXN1 binding to MPP-β and PARL (Fig. 6H). Overall, these results suggested that SFXN1 might affect PINK1 degradation by interacting with MPP-β and PARL.

We assessed PINK1 interaction with PARL or MPP-β under SFXN1-deficient conditions to further investigate whether SFXN1 was indispensable to PINK1 degradation. Notably, SFXN1 knockdown considerably reduced PINK1 interaction with MPP-β or PARL but did not alter their protein levels (Fig. 6I). We also isolated mitochondrial protein and detected PINK1 levels before or after being cleaved by PARL and MPP-β (Fig. 6J). The full-long PINK1 protein was significantly accumulated in SFXN1-deficient mitochondria, and cleaved PINK1 correspondingly decreased following SFXN1 knockdown. Conversely, SFXN1 overexpression rescued the CCCP-induced disruption of PINK1 interaction with MPP-β/PARL, was well as attenuated the mitochondrial accumulation of full-length PINK1 and restored the depletion of cleaved PINK1 in mitochondria. (Supplementary Fig. S5A, B). These findings collectively demonstrated that SFXN1 acted as an essential bridging factor to promote PINK1 processing and degradation by interacting with PARL and MPP-β on IMM, and SFXN1 depletion stabilized PINK1 in mitochondria, thus recruiting Parkin and leading to mitophagy.

SFXN1 promotes the epithelial–mesenchymal transition (EMT) of BLCA cells by inhibiting PINK1-mediated mitophagy

We performed transwell and wound-healing assays in T24 and J82 cells transfected with siSFXN1 and/or siPINK1 to confirm whether SFXN1 promotes BLCA metastasis by inhibiting PINK1-mediated mitophagy. Notably, suppressed migration and invasion abilities of BLCA cells by siSFXN1 could be restored under PINK1-deficient conditions (Fig. 7A, B and Supplementary Fig. S6A), indicating that SFXN1 promoted BLCA metastasis by inhibiting PINK1-mediated mitophagy. Cells lose epithelial characteristics and acquire mesenchymal features via EMT, and this process is a prerequisite for metastasis in various tumors [29–31]. In this study, EMT inhibition, including loss of mesenchymal morphology (Fig. 7C), upregulation of the epithelial marker E-cadherin, and downregulation of the mesenchymal marker N-cadherin (Fig. 7D, E), was observed in T24 and J82 cells transfected with siSFXN1. Furthermore, SFXN1 knockdown suppressed cytoskeletal reorganization, as evidenced by attenuated α-actinin expression and promoted activation of F-actin dynamics marked by a significant upregulation of phosphorylated cofilin (Supplementary Fig. S6B). The EMT is predominantly facilitated by a set of EMT-activating transcription factors, including prototypical EMT transcription factors such as SLUG [32–34]. Notably, both mRNA and protein expression of SLUG were reduced following SFXN1 knockdown (Fig. 7D, E). The TGF-β (transforming growth factor-β) signaling pathway typically plays a dominant role in promoting the expression of EMT-activating transcription factors [35–37]. Therefore, we hypothesized that SFXN1 promotes BLCA metastasis by suppressing mitophagy, thereby altering intracellular mitoROS levels to activate the TGF-β signaling pathway and induce EMT in BLCA cells. To validate this hypothesis, we first performed Gene Set Enrichment Analysis (GSEA) on the DEGs from RNA sequencing data of NC and siSFXN1 groups. The results revealed significant suppression of the TGF-β signaling pathway upon SFXN1 knockdown (Supplementary Fig. S6C). Further supporting this observation, we stratification of BLCA samples in the TCGA database into SFXN1-high and low groups based on median expression levels of SFXN1. The GSEA analysis also demonstrated that genes upregulated in the SFXN1-low group were negatively correlated with TGF-β signaling pathway activity (Supplementary Fig. S6C). The WB assays revealed significantly reduced expression of proteins involved in the TGF-β–EMT signaling pathway in SFXN1-deficient BLCA cells, including TGF-β, SMAD2/3 and p-SMAD2/3 (Fig. 7F). Besides, IHC analyses confirmed a significant increase in E-cadherin expression and concomitant downregulation of N-cadherin, SLUG, TGF-β and Phospho-SMAD2/3 in SFXN1-knockdown subcutaneous tumor tissues compared to NC (Supplementary Fig. S6D). Overall, these findings revealed that SFXN1 deficiency restrained BLCA metastasis by blocking EMT promoted by TGF-β signaling pathway.

Fig. 7. SFXN1 promotes the EMT of BLCA cells by inhibiting PINK1-mediated mitophagy.

A Representative images (left) and quantified results (right) of transwell assays in BLCA cells transfected with siNC, siSFXN1, siPINK1, or siSFXN1 + siPINK1 (n = 3). B Representative images (left) and quantified results (right) of invasion assays in BLCA cells transfected with siNC, siSFXN1, siPINK1, or siSFXN1 + siPINK1 (n = 3). C Representative brightfield image of BLCA cells transfected with siSFXN1 or siNC. Scale bar: 100 μm. WB (D) and qPCR (E) analysis of SLUG, E-cadherin and N-cadherin levels of BLCA cells transfected with siSFXN1 or siNC (n = 3). F WB analysis of proteins levels of TGF-β/SMADs pathway of BLCA cells transfected with siSFXN1 or siNC. G WB analysis of proteins levels of TGF-β/SMADs and EMT pathway of BLCA cells transfected with siNC, siSFXN1, siPINK1, or siSFXN1 + siPINK1. H Representative images (left) and quantified results (right) of mitochondrial ROS of BLCA cells transfected with siSFXN1 or siNC treated with or without mitoTEMPO measured by flow cytometry (n = 3). I WB analysis of proteins levels of TGF-β/SMADs and EMT pathway of BLCA cells transfected with siSFXN1 or siNC treated with or without mitoTEMPO. The P value was assessed using two-tailed Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We determined whether SFXN1 accelerated EMT in BLCA cells by suppressing PINK1-mediated mitophagy. Therefore, we tested the above proteins in TGF-β–EMT signaling in BLCA cells transfected with siPINK1 and/or siSFXN1 (Fig. 7G). Notably, siSFXN1 could not alter TGF-β, SMAD2/3, p-SMAD2/3, SLUG, E-cadherin, and N-cadherin levels in cells simultaneously transfected with siPINK1. Cells are highly enriched in mtROS when mitophagy is inhibited. Previous studies have revealed the interaction between the TGF-β signaling pathway and ROS [37–39]. As previously described, ROS could not only activate TGF-β to transform it from its latent form, but also stimulate its expression and secretion, thus playing a vital role in both classical and non-classical TGF-β signaling pathways [38]. Therefore, we used mitoTEMPO to clear mtROS in BLCA cells. Notably, mitoTEMPO and siSFXN1 clearly reduced mtROS levels separately, whereas siSFXN1 did not further reduce mtROS levels following mitoTEMPO treatment (Fig. 7H). Consistently, the WB results revealed that mitoTEMPO could decrease the expression of TGF-β signaling pathway-related proteins, SLUG, and N-cadherin in T24 and J82 cells and increase the expression of E-cadherin (Fig. 7I). However, SFXN1 knockdown in cells treated with mitoTEMPO could not alter the levels of these proteins (Fig. 7I), suggesting that mtROS removal by siSFXN1-promoted mitophagy plays an important role in suppressing TGF-β-mediated EMT. Overall, these results revealed the underlying mechanism by which SFXN1 promoted the EMT of BLCA cells via inhibiting PINK1-mediated mitophagy.

Discussion

This study provides clinical and experimental evidence supporting the crucial oncogenic role of SFXN1 in BLCA progression and metastasis. SFXN1 was highly enriched in clinical BLCA tissues, and its high expression in cancer cells was closely associated with the progression and poor prognosis of BLCA. Furthermore, in vitro and in vivo assays demonstrated that SFXN1 deficiency significantly suppressed BLCA cell proliferation and metastasis. Moreover, SFXN1 promoted metastasis via an unknown function of restraining PINK1-dependent mitophagy rather than via its classical role as a mitochondrial serine transporter that mediates one-carbon metabolism. Mechanistically, SFXN1 promoted PINK1 degradation by acting as an essential bridging factor and interacting with PARL and MPP-β on IMM, thus leading to mitophagy arrest. Cells were considerably enriched with mtROS when mitophagy was restrained by high SFXN1 expression, thus activating TGF-β-mediated EMT and promoting BLCA metastasis.

Since the discovery of the SFXN1 role in mitochondrial serine transport, its valuable functions and mechanisms in tumor progression have been gradually identified. For example, in lung cancer, SFXN1 promotes tumorigenesis and increases T regulatory cell infiltration in tumor tissues by upregulating CCL20 expression [17]. Furthermore, SFXN1 promotes breast cancer progression by inhibiting the TOLLIP-mediated autophagic degradation of CIP2A [18]. Although the oncogenic phenotype of SFXN1 in several cancers has been reported in recent studies, research on its role in BLCA and the specific molecular mechanism remains unclear. In this study, we demonstrated, for the first time, the oncogenic role of SFXN1 in BLCA and revealed a novel mechanism through which SFXN1 promotes the metastasis of BCLA cells by restraining PINK1-dependent mitophagy.

The PINK1–PARKIN ubiquitin system is the most extensively studied mitophagy pathway [40, 41]. An increasing number of studies have reported an association between decreased Parkin activity and enhanced cancer development, proposing a new concept of Parkin activity or mitophagy as a tumor suppression mechanism [42]. In the present study, SFXN1 knockdown activated PINK1–PARKIN-mediated mitophagy. PINK1 accumulation on the OMM is a critical step in initiating PINK1–PARKIN-mediated mitophagy [27, 43, 44]. Under normal conditions, PINK1 undergoes regular degradation [45], and alterations in the regulators of this degradation process can directly affect PINK1 accumulation on OMM, thus disrupting the balance of mitophagy in cells. However, whether and how SFXN1 is involved in PINK1 degradation remains unexplored. Our results demonstrated that SFXN1 acts as an essential regulator to promote PINK1 processing and degradation by interacting with PARL and MPP-β on IMM, and SFXN1 knockdown stabilizes the full-long PINK1 on OMM, thereby promoting mitophagy. Therefore, SFXN1 in BLCA cells may be a potential therapeutic target for inducing mitophagy to suppress BLCA metastasis.

Mitophagy maintains cellular homeostasis by eliminating dysfunctional or excess mitochondria. However, the precise role of mitophagy in tumorigenesis and metastasis remains unclear. The tumor promoter or suppressor function of mitophagy is highly contingent on the context, including cancer type, tumor stage, genetic factors, and the specific microenvironment in which it occurs. Generally, during tumor initiation, decreased mitophagy allows for the persistence of tumorigenic mitochondrial signals, thereby accelerating tumorigenesis. In line with this, the mitophagy modulator Parkin is generally considered a tumor suppressor gene that is often lost in many human cancers. In established tumors, mitophagy may be required for stress adaptation and survival by removing pro-apoptotic mitochondria. For example, silencing of PINK1 inhibited proliferation and blocked cell cycle of lung cancer cells [46]. Another study demonstrated that LC3-II and Pink1 expression increased after chemotherapeutic treatment in the ESCC cells, and inhibition of autophagy or mitophagy restored chemosensitivity [47]. However, growing evidence suggests that the ablation of mitophagy accelerates metastasis. For instance, BRCA1 deficiency impairs stress-induced mitophagy by blocking ATM–AMPK–DRP1-mediated mitochondrial fission, thereby facilitating breast cancer [48]. Furthermore, MAPK1/3 kinase-dependent ULK1 degradation promotes breast cancer bone metastasis by attenuating mitophagy [12]. Therefore, further investigation is warranted to elucidate the intricacies of dysregulated mitophagy modulators in cancer and gain an improved understanding of the complicated interplay between mitophagy and cancer development. In this study, we demonstrated that SFXN1 deficiency-induced mitophagy suppressed BLCA metastasis and uncovered that mitophagy promoted by mtROS removal following SFXN1 knockdown plays an important role in suppressing TGF-β-mediated EMT.

In conclusion, our findings demonstrated the oncogenic role of SFXN1 in BLCA for the first time and revealed a novel mechanism through which SFXN1 promotes BCLA metastasis by restraining PINK1-dependent mitophagy (Fig. 8). This study reveals SFXN1 as a novel promising therapeutic target for BLCA, identifies a new mitophagic modulator, and provides insights on the association between mitophagy and BLCA progression.

Fig. 8. Schematic diagram of the proposed mechanism by which SFXN1 promotes bladder cancer metastasis by restraining PINK1-dependent mitophagy.

SFXN1 is upregulated in BLCA tissues, and promotes BLCA metastasis through its unrevealed function of restraining PINK1-dependent mitophagy rather than its classical role as a mitochondrial serine transporter to promote cell proliferation. Specifically, SFXN1 acted as an essential bridging factor to promote PINK1 degradation by interacting with PARL and MPP-β on the IMM, leading to mitophagy arrest and mtROS accumulation, thus activated TGF-β-mediated EMT and promoted BLCA metastasis (This figure was created by Figdraw).

Author contributions

BZ, MD, QZ and HG conceived and designed the experiments. BZ and MD wrote, reviewed and revised the manuscript; GD, XG, HL, WC and WD developed the methodology; QL and GX analyzed and interpreted the data. All authors read and approved the final manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (81972388 to HG, 81902571 to MD, 82072822 and 82373266 to QZ, and 82203765 to QL); Nanjing Medical Science and technology development Foundation (ZKX22024 to MD); Jiangsu Province Capability Improvement Project through Science, Technology and Education, Jiangsu Provincial Medical Key Discipline (Laboratory) Cultivation Unit (JSDW202221).

Data availability

The data and resources used during this study are available from the corresponding authors on reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

All study participants provided informed consent, and the study design was approved by Institutional Ethics Committee of the Nanjing Drum Tower Hospital approved this study (2022-493-01). All authors consent to the publication of the manuscript.

Supplementary information

The online version contains supplementary material available at 10.1038/s41388-025-03460-7.