MANF facilitates breast cancer cell survival under glucose-starvation conditions via PRKN-mediated mitophagy regulation

ABSTRACT

During tumor expansion, breast cancer (BC) cells often experience reactive oxygen species accumulation and mitochondrial damage because of glucose shortage. However, the mechanism by which BC cells deal with the glucose-shortage-induced oxidative stress remains unclear. Here, we showed that MANF (mesencephalic astrocyte derived neurotrophic factor)-mediated mitophagy facilitates BC cell survival under glucose-starvation conditions. MANF-mediated mitophagy also promotes fatty acid oxidation in glucose-starved BC cells. Moreover, during glucose starvation, SENP1-mediated de-SUMOylation of MANF increases cytoplasmic MANF expression through the inhibition of MANF’s nuclear translocation and hence renders mitochondrial distribution of MANF. MANF mediates mitophagy by binding to PRKN (parkin RBR E3 ubiquitin protein ligase), a key mitophagy regulator, in the mitochondria. Under conditions of glucose starvation, protein oxidation inhibits PRKN activity; nevertheless, the CXXC motif of MANF alleviates protein oxidation in RING II-domain of PRKN and restores its E3 ligase activity. Furthermore, MANF-PRKN interactions are essential for BC tumor growth and metastasis. High MANF expression predicts poor outcomes in patients with BC. Our results highlight the prosurvival role of MANF-mediated mitophagy in BC cells during glucose starvation, suggesting MANF as a potential therapeutic target.

Abbreviation: 2DG, 2-deoxy-D-glucose; 5TG, 5-thio-D-glucose; ACSL4/FACL4, acyl-CoA synthetase long chain family member 4; Baf A1, bafilomycin A₁; BRCA, breast cancer; CHX, cycloheximide; DMF, distant metastasis-free; DMFS, distant metastasis-free survival; ECM, extracellular matrix; ER, endoplasmic reticulum; ERS, endoplasmic reticulum stress; F-1,6-BP, fructose-1,6-bisphosphate; FAO, fatty acid oxidation; GSH, reduced glutathione; GSVA, gene set variation analysis; HCC, hepatocellular carcinoma; ICC, intrahepatic cholangiocarcinoma; IF, immunofluorescence; MANF, mesencephalic astrocyte derived neurotrophic factor; Mdivi-1, mitochondrial division inhibitor 1; MFI, mean fluorescence intensity; NAC, N-acetyl-L-cysteine; OCR, oxygen-consumption rate; OS, overall survival; PMI, SQSTM1/p62-mediated mitophagy inducer; PPP, pentose phosphate pathway; PRKN, parkin RBR E3 ubiquitin protein ligase; RBR, RING in between RING; RFS, relapse-free survival; ROS, reactive oxygen species; SAPLIPs, saposin-like proteins; TCGA, The Cancer Genome Atlas; TNBC, triple-negative breast cancer; WT, wild type.

Introduction

High glucose metabolism is a hallmark of cancer, and it involves energy production, biosynthesis of macromolecules, glycosylation of proteins and maintenance of oxidation homeostasis in cancer cells [1]. However, cancer cells often encounter nutrient shortage (particularly glucose shortage) in breast cancer (BC) [2]. For instance, the glucose uptake of BC cells is impaired when they migrate to the lumen of the mammary gland during early stage of tumorigenesis or detaching from the extracellular matrix during cancer dissemination [3]. Also, the glucose supply of BC cells becomes limited due to poor vascularization and rapid nutrient consumption in the bulk tumor [4]. Further studies indicated that BC cells could survive in glucose-deprived conditions [1,5]. Targeting key enzymes of glucose metabolism (including PKM/PKM2 and PHGDH) failed to inhibit tumor proliferation in BC [1]. A fasting-mimicking diet reducing circulating glucose level or antiangiogenic therapy (bevacizumab) limiting nutrient (glucose) supply could only slow tumor progression but did not achieve long-term, progression-free survival in patients with BC [6–8]. Thus, studying the mechanism of BC cells overcoming glucose-starvation condition may point to targetable vulnerabilities. In BC, glucose starvation reduces cellular ATP production but increases reactive oxygen species (ROS) accumulation via disrupting NADPH synthesis by the pentose phosphate pathway (PPP) [3]. Previous studies showed that BC cells compensate for ATP deficiency resulting from glucose starvation by rewiring their metabolism toward other nutrients (such as glutamate and fatty acid) [3,9]. However, the mechanisms by which BC cells cope with the increased oxidative stress caused by glucose shortage remain unclear.

Mitochondria are the major oxidative metabolism site, where energy is released through glucose, fatty acid, and amino acid oxidation; they also act as the principal sites for intracellular ROS production [10]. Overload of mitochondrial damage due to ROS accumulation induces cancer cell death under glucose-starvation conditions [4]. Mitophagy, a selective autophagy process for the elimination of damaged or redundant mitochondria fraction, is required to maintain mitochondrial integrity and cellular oxidative homeostasis [11]. At present, both ubiquitin-dependent (PINK1-PRKN-dependent) and receptor-dependent (PINK1-PRKN-independent) pathways have been described in regulation of mitophagy [12]. Mitophagy mediated by different pathways might exert distinct biological functions. For instance, RIPK1-PGAM5-mediated mitophagy attenuates IDH1/IDH2-mediated NADPH production and subsequently elevates mitochondrial ROS levels [13]. In contrast, HIF1A-BNIP3-BNIP3L-mediated mitophagy limits cellular ROS production and supports cell survival during extracellular matrix (ECM) detachment and cancer metastasis [14]. Because of its multifaceted impact on cellular oxidative homeostasis, the role of mitophagy in cancer cells under nutrient stress warrants further investigation.

MANF (mesencephalic astrocyte derived neurotrophic factor) was initially identified as a neurotrophic factor promoting dopaminergic neuron survival [15]. Moreover, MANF exerts cytoprotective effects in various tissues (including the brain, retinas, pancreas, liver, heart, kidneys, spleen, and bones) by regulating ER stress (ERS) response [15,16]. Under nonstress conditions, MANF is predominantly located in the ER; however, ERS (e.g., hypoxia, inflammation, DNA damage, and nutrient stress) leads to MANF upregulation and secretion [15]. Although uptake of secreted MANF via sulfatide binding prohibits hypoxia-induced cell death [17], the target receptors and the signaling pathways downstream of MANF remain unclear. Moreover, MANF overexpression, but not exogenous MANF supplementation, improves cell survival under cellular stress [18]. In the kidney, inducible tubular overexpression of MANF inhibits kidney fibrosis and protects kidney function by stimulating mitophagy [19]. Thus, intracellular MANF also exerts cytoprotective function. However, the function of intracellular MANF in human cancers remains largely unknown. Although high MANF expression predicts poor prognosis in patients with hepatocellular carcinoma (HCC) [20], Liu et al. reported that MANF translocates into the nucleus to inhibit the NFKB/NF-κB-SNAI1/Snail pathway and thus act as a tumor suppressor [21]. Thus, additional studies investigating the biological functions of intracellular MANF in human cancers are warranted.

In the current study, we assessed the role of MANF-mediated mitophagy in BC cells under glucose-starvation conditions. We noted that SENP1-mediated de-SUMOylation of MANF renders its mitochondrial distribution in BC cells during glucose starvation. MANF facilitated BC cell survival by regulating PRKN-mediated mitophagy. This occurred through the inhibition of PRKN oxidation and restoration of PRKN E3 ligase activity – mediated by the CXXC motif of MANF. MANF-PRKN-mediated mitophagy then induced fatty acid oxidation (FAO) facilitating BC cell survival under conditions of glucose starvation. Taken together, our results reveal a novel function of MANF promoting cell survival in BC during glucose starvation.

Results

MANF facilitates BC cell survival under glucose-starvation conditions

To investigate the role of MANF in BC cells, four BC cell lines (MCF-7, BT474, SKBR3, and MDA-MB-231) were treated with glucose, glutamate, or serum starvation for 24 h. MANF was upregulated in BC cells (particularly MDA-MB-231 and MCF-7) under glucose-starvation conditions (Figure 1A). No significant changes were observed in cell MANF expression under glutamate or serum starvation (Figure S1A,B). MANF expression in BC cells began to increase at 12 h after glucose deprivation, reaching a peak at 24 h (Figure S1C). Next, we stably silenced MANF expression using small hairpin RNA (shRNA) in MDA-MB-231 and MCF-7 cells (both of which exhibited the most significant increases in MANF expression under glucose-starvation conditions; Figure S1D). Cell viability was detected at 0, 12, 24 and 36 h after glucose starvation. Cell viability began to decrease at 12–24 h after glucose starvation (Figure S1E). Compared with the control cells, MANF-knockdown (KD) cells exhibited significantly decreased cell viability at 24 h after glucose starvation (Figure S1E). The differences in cell viability between the two groups gradually became larger with the increasing duration of glucose-starvation conditions (Figure S1E). In addition, live cell imaging was used to dynamically observe the cellular morphology of BC cells in a glucose-deprived environment. After 18 h of glucose-starvation conditions, cells with MANF KD began to show damaged morphology (cytoplasmic vacuoles). With the increasing duration of glucose-starvation conditions, the vacuoles in MANF-KD cells became larger, and in the late stage (24–36 h), most MANF-KD cells became rounded or fragment into small pieces (Figure S1F). In contrast, BC cells expressing wild-type levels of MANF showed a normal cell morphology after 24 h of glucose-starvation conditions (Figure S1F). Considering the above results, we selected 24 h as the representative time point to investigate the biological function of MANF in BC cells under glucose-starvation conditions. Compared with control cells, cells with MANF KD demonstrated a decreased viability under glucose-starvation conditions (Figure 1B); however, MANF KD led to no effect on cell viability under glutamate or serum starvation conditions (Figure S2A,B). These results indicated that MANF is critical for BC cell survival under glucose-starvation conditions.

Figure 1.

MANF facilitates BC cell survival under glucose-starvation conditions. (A) BC cells under glucose-starvation conditions exhibit increased MANF expression (n = 3). GAPDH and ACTB/β-actin were the internal controls for qRT-pcr and western blotting, respectively. (B) Morphology (left) and viability (right, n = 3) of MANF-KD and control cells under glucose-starvation conditions. Scale bar: 100 μm. (C) Illustration depicting glucose and 2DG metabolism: glucose metabolism involves not only glycolysis but also other pathways that require glucose (including the PPP, which generates NADPH; the hexosamine pathway, which is required for glycosylation of proteins). F-1,6-bp is a human endogenous metabolite metabolized via glycolysis. Comparatively, 2DG is phosphorylated to form 2DG-6-P, which cannot be metabolized via glycolysis but can be partially metabolized via PPP to form NADPH. In addition, 2DG structurally resembling mannose interferes glycoprotein formation and hence inhibits N-linked glycosylation of proteins. (D) Viability of MANF-KD and control cells treated with 2DG (5 mM), 5TG (5 mM) or F-1,6-bp (5 mM) under glucose-starvation conditions for 24 h (n = 3/condition). (E and F) MANF KD is associated with increased intracellular ROS production (E) and NADP⁺:NADPH ratio (F) in BC cells under glucose-starvation conditions (n = 3). (G) Viability of MANF-KD and control cells treated with 5 mM NAC or GSH under glucose-starvation conditions for 24 h (n = 3/condition). All data are presented as means ± SDs; p values were determined using two-tailed Student’s t test (A) or one-way ANOVA (B and D-G).

Glycolysis is the major metabolic pathway of glucose fulfilling the high energy demand for tumor proliferation. Thus, we first investigated whether glycolysis restoration could rescue MANF-KD cells under glucose-deprivation conditions. We treated control and MANF-KD cells with fructose-1,6-bisphosphate (F-1,6-BP), a human endogenous metabolite metabolized through glycolysis, under glucose-starvation conditions (Figure 1C). F-1,6-BP treatment failed to restore cell viability in MANF-KD cells under glucose-starvation conditions (Figure 1D). Otherwise, glucose is metabolized through the PPP to produce NADPH (Figure 1C). We next investigated whether BC cell survival was affected by disruption of glucose flux to the PPP under glucose-starvation conditions. We treated MANF-KD or control BC cells with two glucose analogs: 2-deoxy-D-glucose (2DG) and 5-thio-D-glucose (5TG). Both 2DG and 5TG cannot be metabolized through glycolysis. 2DG but not 5TG was partly metabolized through the PPP, resulting in NADPH production (Figure 1C). 2DG but not 5TG restored cell viability in MANF-KD cells under glucose-starvation conditions (Figure 1D).

Glucose starvation is associated with intracellular ROS accumulation and mitochondrial damage [5]. MANF KD was noted to be associated with an increase in the ROS level and the NADP⁺:NADPH ratio (Figure 1E,F). Similarly, we detected higher viability and lower oxidative stress in cells with MANF overexpression (Figure S2C–F). Moreover, the antioxidants N-acetyl-L-cysteine (NAC) and reduced glutathione (GSH) increased the viability of MANF-KD cells under glucose-starvation conditions (Figure 1G). MANF, an ERS-responsive protein, was, therefore, secreted in response to cellular stress. We then investigated whether the extracellular addition of MANF (resembling MANF secretion) facilitates BC cell survival under glucose-starvation conditions. However, our results indicated that supplementation of 100 ng/mL MANF did not increase BC cell viability under glucose-starvation conditions (Figure S2G). To investigate whether ERS-responsive signaling is involved in MANF-mediated cytoprotection in BC cells under nutrient stress, we inhibited ERS signaling with either GSK2606414 (EIF2AK3/PERK signaling inhibitor) or 4μ8C (ERN1/IRE1 signaling inhibitor). However, neither of these inhibition approaches abrogated the cytoprotective effect of MANF in BC cells under glucose-starvation conditions (Figure S2H). Our results thus suggested that MANF promotes BC cell survival by mediating oxidative stress under glucose-starvation conditions.

MANF translocates to mitochondria in BC cells under glucose-starvation conditions

To investigate the biological function of MANF, we first determined the intracellular distribution of MANF. In addition to being secreted, SUMOylated MANF can translocate to the nucleus [21]. In both BC and non-cancerous cells (293T) under glucose-starvation conditions, we observed that MANF expression increased in the cytoplasm, not in the nucleus (Figure S3A). Because mitochondria are the major ROS source in the cytoplasm, we investigated whether MANF translocates to mitochondria in BC cells under glucose-starvation conditions. Our immunofluorescence (IF) assay results demonstrated increased MANF staining colocalized with mitochondria in both BC and 293T cells under glucose-starvation conditions (Figures 2A and S3B,C). Similarly, our western blotting results revealed an increased MANF expression in mitochondria under glucose-starvation conditions; this effect was attenuated by antioxidant treatment (Figures 2B and S3D). Therefore, MANF translocates to mitochondria under glucose-starvation conditions.

Figure 2.

MANF translocates to mitochondria in BC cells during glucose starvation. (A) Quantification of manf-mitochondria colocalization (right) in BC cells (n = 3/condition); representative image (left). BC cells were treated with PBS or 5 mM NAC under glucose-starvation or normal conditions for 24 h. Staining of MANF (green), mitochondria (MitoTracker; red), and nucleus (blue) was observed under a confocal microscope. Scale bar: 10 μm. All data are presented as means ± SDs, p values were determined using one-way ANOVA. (B) MANF targeted mitochondria in BC cells under glucose-starvation conditions. BC cells treated with PBS or 5 mM NAC under glucose-starvation or normal conditions were collected. MANF expression was detected in the whole-cell or mitochondria fraction, with HSPD1 and ACTB as the internal controls, respectively. (C) SUMOylation sites in the MANF amino acid sequence were predicted using GPS-SUMO 2.0 (https://sumo.Biocuckoo.cn/). (D) BC cells transfected with SUMO1, SUMO2, or SUMO3 were immunoprecipitated with MANF antibody. Bands of MANF SUMOylation were detected using SUMO1 and SUMO2/SUMO3 antibodies. (E) 293T cells cotransfected with SUMO1 and 3× FLAG-MANF mutants were immunoprecipitated with FLAG antibody. Bands of SUMOylated MANF were detected using the SUMO1 antibody. (F) MANF SUMOylation inhibited mitochondria location of MANF in cells under glucose-starvation conditions. BC cells with SUMO1 overexpression or SENP1 KD were collected, and MANF expression was detected in the whole-cell, mitochondria, or nucleus fraction. ACTB, HSPD1, and histone H3 were the internal controls for whole-cell, mitochondria, and nucleus fractions, respectively. (G) MANF and MANF mutant expression detected using FLAG antibody in mitochondria fraction of BC cells under glucose-starvation conditions. (H) 293T cells cotransfected with SUMO1 and SENP1 shRNA were immunoprecipitated using MANF antibody. Bands of SUMOylated MANF were detected using the SUMO1 antibody.

We next investigated whether SUMOylation of MANF affects its translocation on mitochondria. We predicted four potential SUMOylation sites in the MANF amino acid sequence: Lys94, Lys120, Lys170, and Lys174 by GPS-SUMO 2.0 (Figure 2C). By using a SUMO1 antibody, we detected high-molecular-weight SUMO1 bands in cells immunoprecipitated with a MANF antibody (Figure 2D and Figure S3E, lanes 2 and 5–8). We did not observe high-molecular-weight SUMO2/SUMO3 bands in the same sample (Figure 2D and Figure S3E). We also detected high-molecular-weight SUMO1 bands in 293T cells with cotransfection of MANF and SUMO1 (Figure S3F). Thus, MANF SUMOylation is mediated by SUMO1. We then induced the expression of MANF with mutations at its SUMOylation sites in 293T cells. Mutations at K120 and K170/K174 but not K94 of MANF impaired MANF SUMOylation indicating that K120 and K170/K174 are major sites of SUMOylation in MANF (Figure 2E). MANF SUMOylation induced by SUMO1 overexpression increased MANF expression in nucleus, and simultaneously, decreased MANF expression in mitochondria (Figure 2F and Figure S3G). Also, SUMO1 overexpression significantly reduced mitochondrial expression of wildtype (WT) and K94R-mutant MANF but not mutants resistant to SUMOylation (K120R- and K170R/K174R-mutant MANF) (Figure 2G). Although our results indicated that mitochondrial distribution of MANF is affected by SUMO1-mediated SUMOylation, how BC cells regulated MANF SUMOylation under glucose-starvation conditions remains unknown.

SENP1 and SENP2 are specific SUMO-deconjugating enzymes for SUMO1-mediated SUMOylation. We observed an increase in the expression of SENP1 but not SENP2 in cells under glucose-starvation conditions (Figure S3H, I); however, the increase in SENP1 expression was attenuated by NAC treatment indicated that SENP1 expression is induced by oxidative stress in cells under glucose-starvation conditions (Figure S3H, I). Next, we investigated whether SENP1, upregulated by oxidative stress in BC cells, affects MANF SUMOylation. In SENP1-knockdown cells, we observed decreased mitochondrial MANF expression under glucose-starvation conditions (Figure 2F and Figure S3G). SENP1 KD associated with more intense bands of high-molecular-weight SUMO1 in cells under glucose-starvation conditions (Figure 2H). NAC treatment, which decreased MANF expression in mitochondria (Figure 2B), decreased SNEP1 expression and induced MANF SUMOylation under glucose-starvation conditions (Figure 2H). Moreover, in cells with MANF overexpression, SENP1 inhibition inducing MANF SUMOylation was associated with increased oxidative stress and decreased cell viability under glucose-starvation conditions (Figure S3J, K). Thus, SENP1 induces the mitochondrial distribution of MANF via regulating MANF SUMOylation in BC cells under glucose-starvation conditions.

MANF mediates mitophagy in BC cells under glucose-starvation conditions

Mitophagy is critical in defense against oxidative stress by removing damaged mitochondria and preventing excess ROS production. We labeled mitochondria colocalized with autophagosome through staining with LC3B (an autophagosome marker) [22] and MitoTracker Red, and evaluated the mitophagy degree. The extent of LC3B staining colocalized with mitochondria increased under glucose-starvation conditions, indicating that glucose starvation induces mitophagy in BC cells (Figure 3A). To evaluate whether mitophagy facilitates BC cell survival under glucose-starvation conditions, we inhibited BC cell mitophagy using bafilomycin A₁ (Baf A1; a lysosomal acidification inhibitor) and mitochondrial division inhibitor 1 (Mdivi-1; a DNM1L/Drp1 inhibitor). Also, we established mitophagy-deficient BC cell lines through stably silencing DNM1L (Figure S4A). Both pharmacological mitophagy inhibition (Baf A1 and Mdivi-1) and genetic mitophagy inhibition (DNM1L-KD) significantly decreased cell viability and inhibited mitophagy under glucose-starvation conditions (Figure 3B and Figure S4B). In contrast, SQSTM1/p62-mediated mitophagy inducer (PMI) facilitated BC cell survival under glucose-starvation conditions (Figure 3B). These results indicated mitophagy has a cytoprotective function in BC cells under glucose-starvation conditions.

Figure 3.

MANF mediates mitophagy in BC cells under glucose-starvation conditions. (A) Quantification of LC3B-mitochondria colocalization (lower, n = 3) assessed in BC cells under glucose-starvation or normal conditions for 24 h; representative image (upper). (B) Both pharmacological mitophagy inhibition (baf A1 and mdivi-1) and genetic mitophagy inhibition (DNM1L-KD) reduced and mitophagy agonist PMI increased BC cell viability under glucose-starvation conditions (n = 3). (C) Mitochondrial and mitophagy-related protein expression was detected in BC cells with or without MANF KD under glucose-starvation or normal conditions. ACTB was the internal control. (D) BC cells with MANF KD displayed decreased mitophagy activity, and BC cells stably expressing mt-Keima were culture under glucose-starvation or normal conditions for 24 h. In living cells, Keima signals under neutral and acidic pH are green (ex = 458 nm/Em = 588–633) and red (ex = 561 nm/Em = 588–633), respectively; representative image (left), scale bar: 50 μm. Cellular mitophagy activity was determined as the red:green fluorescence intensity ratio (right, n = 3). (E) Quantification of LC3B-mitochondria co-localization (right, n = 3) was assessed in the indicated BC cells under glucose-starvation or normal conditions for 24 h; representative image (left). Staining of LC3B (green), mitochondria (MitoTracker, red), and nucleus (blue) was observed under a confocal microscope. Scale bar: 10 μm. (F) LC3B-mitochondria colocalization quantified in MANF-KD (MANFsh1) BC cells with the indicated transfection (n = 3/condition). (G – H) MANF reduces ROS production (left) and facilitates BC cell survival (right) under glucose-starvation conditions via mediating mitophagy (n = 3/condition). Baf A1 was used to inhibit cellular mitophagy. All data are presented as means ± SDs; p values were determined using two-tailed Student’s t test (A) or one-way ANOVA (B – H).

Next, we developed the mitophagy score by using the REACTOME_MITOPHAGY gene set and gene set variation analysis (GSVA). We observed that MANF mRNA expression is positively correlated with mitophagy score in BC tissues (TCGA-BRCA; Figure S4C). We observed a higher abundance of TOMM20 (a mitochondrial outer membrane protein) and HSPD1 (a mitochondrial matrix protein) in MANF-KD cells than in control cells under glucose-starvation conditions indicating an increased content of mitochondria in MANF-KD cells (Figure 3C). By using MitoTracker Red to label mitochondria, we discovered that MANF KD is associated with increased mitochondrial staining intensity in cells under glucose-starvation conditions but not in cells under normal conditions or glucose-starvation conditions with mitophagy inhibition (Figure S4D). MANF overexpression was also associated with low cell mitochondrial content under glucose-starvation conditions (Figure S4F, G). Therefore, MANF might participate in the mitophagy-mediated mitochondrial clearance process.

Furthermore, we assessed the expression of SQSTM1 (an autophagosome-binding protein being degraded during macroautophagy/autophagy) and that of LC3B-II (a lipidated form of LC3B being produced during the autophagy process) [23]. In cells under glucose-starvation conditions, MANF expression was associated with decreased SQSTM1 expression and increased ratio of LC3B-II:LC3B-I (Figure 3C and Figure S4F). To evaluate mitophagy activity in viable cells, we expressed the fluorescent protein mt-Keima in BC cells. Under glucose-starvation conditions, MANF-KD BC cells exhibited a significantly lower ratio of red:green fluorescence, indicating lower mitophagy activity, than did control cells (Figure 3D). To testify whether MANF involved specifically in ROS-induced mitophagy, we treated control and MANF-KD cells with mitochondrial uncouplers (including FCCP and CCCP). FCCP and CCCP have been reported to induce mitophagy via ROS-independent mechanism [24]. MANF inhibition did not reduce mitophagy activity in FCCP/CCCP treated BC cells (Figure S4E). Also, MANF KD was associated with a low proportion of LC3B colocalized with mitochondria (Figure 3E). Similarly, we noted that MANF overexpression was associated with increased mitophagy activity (Figure S3H, I). To validate mitochondrial distribution of MANF is essential for its role in mitophagy, we transfected shRNA-resistant forms of MANF or MANF mutants in cells with MANF knockdown. Mitophagy inhibition resulting from MANF KD could be restored through transfection of MANF WT but not by cotransfection of MANF WT and SUMO1 (Figure 3F). SUMOylation-resistant mutants overexpression restored mitophagy activity in BC cells with MANF KD and SUMO1 overexpression (Figure 3F). Moreover, MANF overexpression did not alleviate oxidative stress and facilitate cell survival in cells with mitophagy inhibition under glucose-starvation conditions (Figure 3G, H). In addition, we also established manf knockout (KO) cell lines to further investigate the role of MANF in the process of mitophagy (Figure S5A, B). manf KO associated with decreased cell viability, increased ROS level and decreased mitophagy activity in BC cells under glucose-starvation conditions (Figure S5C–E). These results indicated that MANF-induced mitophagy confers BC cell survival under glucose-starvation conditions.

MANF-mediated mitophagy induces FAO in BC cells under glucose-starvation conditions

When under glucose-starvation conditions, fatty acids and glutamate are the major molecules involved in cellular NADPH and ATP production. Our results indicated that in mitochondria, the metabolism of fatty acid but not that of glutamate was required for MANF-mediated cell survival under glucose-starvation conditions (Figure S6A – D). FAO, which produces ATP and NADPH, is essential for the adaptation of cancer cells to glucose-starvation conditions [3]. We further investigated whether MANF-mediated mitophagy induces FAO in BC cells under glucose-starvation conditions. MANF expression was positively associated with the expression of FAO regulators, including CPT1A (carnitine palmitoyltransferase 1A) and CPT1B, and ACSL4/FACL4 (acyl-CoA synthetase long chain family member 4) in cells under glucose-starvation conditions (Figure 4A and Figure S6E). MANF KD was noted to reduce FAO-driven oxygen-consumption rate (OCR) in cells under glucose-starvation conditions (Figure 4B). In contrast, its overexpression was associated with increased FAO-driven OCR (Figure S6F). Next, when FAOBlue dye for live cells was used to detect FAO activity under glucose-starvation conditions, MANF KD cells exhibited a decrease in FAOBlue fluorescence intensity compared with the control cells (Figure 4C). In contrast, MANF-overexpressing cells demonstrated an increase in this intensity (Figure S6G). These data indicated that MANF facilitates FAO in BC cells under glucose-starvation conditions.

Figure 4.

Manf-mediated mitophagy induces FAO in BC cells under glucose-starvation conditions. (A) Expression of key FAO regulators decreased in MANF-KD BC cells under glucose-starvation conditions. ACTB was the internal control. (B) MANF KD reduced fao-driven OCR in BC cells under glucose-starvation conditions (right, n = 3/condition). fao-driven OCR was determined by the initial slope from the linear portion of the FAO signal profile (left). FCCP (2.5 µm) treatment dissipating the mitochondrial membrane potential was used to achieve the maximal demand of FAO; 10 µm etomoxir treatment, preventing fatty acid import into mitochondria, was used to inhibit FAO activity. BC cells treated with FCCP and Etomoxir were the positive and negative controls, respectively. (C) Quantification (right panel) of FAOBlue staining (blue fluorescence) in living BC cells with or without MANF KD (MANFsh1) under indicated culture conditions (n = 3); representative image (left). Scale bar: 50 μm. (D) MANF overexpression restored FAO regulator expression in MANF-KD (MANFsh1) cells under glucose-starvation conditions but not in MANF-KD cells treated with baf A1. ACTB was the internal control. (E and F) MANF overexpression increased fao-driven OCR (E) and FAOBlue staining (F) in MANF-KD (MANFsh1) cells under glucose-starvation conditions but not in MANF-KD cells treated with baf A1 (n = 3/condition). Scale bar: 50 μm. All data are presented as means ± SDs; p values were determined using one-way ANOVA (B, C, E, and F).

The mitophagy inhibitor Baf A1 was used to validate whether MANF facilitates FAO through mitophagy mediation. MANF KD was associated with decreased CPT1A/CPT1B and ACSL4 expression, which could then be restored through the transfection of shRNA-resistant forms of MANF (Figure 4D). However, Baf A1 inhibited MANF-induced expression of FAO-related genes under glucose-starvation conditions (Figure 4D). Our FAO assay and FAOBlue staining results also demonstrated that MANF overexpression restored FAO activity in MANF-KD cells under glucose-starvation conditions, but Baf A1 prevented it (Figure 4E, F). Therefore, MANF-mediated mitophagy facilitates BC cell survival through the mediation of FAO activity under glucose-starvation conditions.

MANF interacts with PRKN in BC cells under glucose-starvation conditions

To further investigate the mechanism by which MANF mediates mitophagy, MANF was immunoprecipitated, and proteins interacting with MANF were detected through silver staining and mass spectrometry. One of the specified bands displayed in the glucose-starved cells were excised and used for mass spectrometry (Figure S7A). PRKN, a ubiquitination-dependent mitophagy regulator, interacted with MANF in cells under glucose-starvation conditions (Figure S7A). PRKN, an E3 ubiquitin ligase, is recruited to damaged mitochondria by PINK1 and promotes mitophagy by mediating mitochondrial protein ubiquitination [25]. We used coimmunoprecipitation assays to confirm the endogenous MANF-PRKN interactions in BC cells under glucose-starvation conditions (Figure 5A). In cells transfected with 3× FLAG-MANF and MYC-PRKN, we consistently observed the MANF-PRKN interactions under glucose-starvation conditions (Figure S7B). Our IF staining results demonstrated that the MANF (green)-PRKN (red) colocalization levels were higher in BC cells under glucose-starvation conditions than in BC cells under normal conditions (Figure 5B, C). In cells under glucose-starvation conditions, most staining for MANF-PRKN interaction (yellow) was noted on mitochondria (magenta; Figure 5B). Furthermore, inhibiting mitochondrial translocation of MANF via SUMO1-mediated SUMOylation hindered the MANF-PRKN interaction in cells under glucose-starvation conditions (Figure 5D). Antioxidants inhibiting SENP1-mediated de-SUMOylation of MANF prevented the MANF-PRKN interactions under glucose-starvation conditions (Figure S7C). As mitochondrial recruitment of PRKN is mediated by PINK1 during mitophagy, we also investigated whether PINK1 is required for the MANF-PRKN interactions. We established pink1-knockout (KO) BC cell lines using CRISPR-Cas9 system (Figure S7D-E). Coimmunoprecipitation assay and IF staining showed that pink1-KO abrogated the MANF-PRKN interaction in BC cells under glucose-starvation conditions (Figure S7F, G). Thus, MANF binds to PRKN on mitochondria in BC cells under glucose-starvation conditions.

Figure 5.

MANF interacts with PRKN in BC cells under glucose-starvation conditions. (A) Coimmunoprecipitation assay revealed endogenous interactions between MANF and PRKN in BC cells under glucose-starvation conditions. Asterisk indicates IgG (H) or IgG (L) bands. (B and C) Representative images (B) and quantification (C, n = 3) of MANF (green)-prkn (red) colocalization on mitochondria (magenta) in cells under glucose-starvation conditions. The nucleus was stained with DAPI (blue). Scale bar: 10 μm. (D) SUMO1 overexpression inhibited MANF-PRKN interaction in BC cells under glucose-starvation conditions. BC cells with or without SUMO1 transfection were immunoprecipitated with MANF antibody. Asterisk indicates IgG (H) or IgG (L) bands. (E) Structure analysis and schematic of truncated mutants for MANF (upper) and PRKN (lower). MANF comprises an er-targeting signal sequence (residues 1–24), SAPLIPs domain (residues 29–125), SAP domain (residues 127–169), and RTDL sequence (residues 179–182). PRKN comprises a ubiquitin-like domain (residues 1–76), RING 0 domain (residues 141–225), RING I domain (residues 238–293), IBR domain (residues 313–377), and RING II domain (residues 418–449). (F) SAP domain of MANF is essential for MANF-PRKN interaction. 293T cells cotransfected with MYC-PRKN- and 3× FLAG-MANF-truncated mutants were immunoprecipitated with FLAG antibody. (G) RING II domain of PRKN is essential for MANF-PRKN interaction. 293T cells cotransfected with 3× FLAG-MANF- and MYC-PRKN-truncated mutants were immunoprecipitated with MYC antibody. (H) PRKN is essential for MANF to reduce ROS production (left), induce mitophagy (middle), and facilitate BC cell survival (right) under glucose-starvation conditions (n = 3/condition). All data are presented as means ± SDs (C and H); p values were determined using two-tailed Student’s t test (C) or one-way ANOVA (H).

Next, we established truncated mutants of MANF and PRKN to investigate the domains essential for their interaction (Figure 5E). The MANF structure consists of an ER-targeting signal peptide (N-terminal), saposin-like proteins (SAPLIPs) domain, SAP domain, and RTDL sequence (essential for its ER retention; C-terminal) [15]. Because the N-terminal signal peptide was cleaved after MANF maturation in the ER, we speculated that this peptide is not involved in MANF’s protein interactions. Our immunoprecipitation assays indicated that both the M2 (containing SAP domain and RTDL sequence) and M3 (containing SAP domain alone) mutants of MANF bound to PRKN (Figure 5F), whereas only P5 truncated mutants (RING II-domain) of PRKN bound to MANF (Figure 5G). Under normal conditions, PRKN has an autoinhibited conformation, where the catalytic RING II domain is often blocked by the RING 0 domain [26]. We exogenously overexpressed a PRKN mutant (PRKN^H302A), which remained the autoinhibited conformation [26], in 293T cells and observed that MANF failed to interact with it under glucose-starvation conditions (Figure S7H). These results indicated that the active conformation of PRKN unblocking the RING II domain is essential for MANF-PRKN interactions. PRKN, belonging to the RING in between RING (RBR) domain family of ubiquitin ligases, binds to the lysine of the substrates and conjugates the ubiquitin onto substrates via its RING II domain [27]. Although we discovered two RBR ubiquitination sites (Lys138 and Lys163) on the SAP domain of MANF by using GPS-SUMO 2.0 (Figure S7I), the results of our immunoprecipitation assay using 3× FLAG-MANF demonstrated that PRKN overexpression did not increase MANF ubiquitination levels (Figure S7J). Therefore, MANF is not ubiquitinated by PRKN.

We next investigate whether PRKN is essential for MANF to facilitate BC cell survival under glucose-starvation conditions. PRKN knockdown was associated with increased ROS level and attenuated cell viability in BC cells under glucose-starvation conditions (Figure S7K, L). Further, our results indicated that PRKN expression was required for MANF to reduce ROS production and facilitate mitophagy and cell survival under glucose-starvation conditions (Figure 5H). MANF re-expression significantly reduced ROS levels and facilitated mitophagy-mediated survival in MANF-KD cells under glucose-starvation conditions (Figure S7M). This function is dependent on the SAP domain of MANF, required for PRKN-MANF interactions: overexpression of M1 mutant (lacking the SAP domain) could not facilitate cell survival in MANF-KD cells under glucose-starvation conditions (Figure S7M). These data collectively indicated that MANF facilitates mitophagy by interacting with PRKN.

MANF facilitates mitophagy by inhibiting PRKN oxidation in BC cells under glucose-starvation conditions

PRKN, which mediates the ubiquitination of mitochondrial membrane proteins, becomes degraded during mitophagy [28]. Under normal conditions, MANF KD did not influence PRKN expression significantly (Figure 6A). However, we observed that under glucose-starvation conditions, PRKN expression was lower in control cells than in MANF-KD cells; moreover, mitophagy inhibition recovered the decrease in PRKN expression in BC cells (Figure 6A). Cycloheximide (CHX) chase assay also indicated that MANF KD was associated with PRKN accumulation in BC cells under glucose-starvation conditions (Figure S8A). Therefore, MANF might involve in PRKN-mediated mitophagy.

Figure 6.

MANF facilitates mitophagy by inhibiting PRKN oxidation in BC cells under glucose-starvation conditions. (A) PRKN expression in MANF-KD and control cells with indicated culture conditions. ACTB was the internal control. (B) MANF overexpression induced autoubiquitination of PRKN in HeLa cells overexpressing PINK1 under glucose-starvation conditions. HeLa cells were cultured in a glucose-free medium for 24 h and treated with MG132 for 6 h. Autoubiquitination of PRKN appeared as high-molecular-weight PRKN bands. (C) MANF overexpression induced prkn-mediated ubiquitination of mitochondrial proteins in HeLa cells overexpressing PINK1 under glucose-starvation conditions. HeLa cells were cultured in a glucose-free medium for 24 h and treated with MG132 for 6 h. CCCP treatment (10 µm, 2 h) was used as positive control for PRKN activation. (D) MANF protects the RING-II domain of PRKN from oxidation in BC cells under glucose-starvation conditions. (E – G) CKGC motif is essential for MANF to inhibit PRKN oxidation (E), induce PRKN autoubiquitination (F), and induce prkn-mediated ubiquitination of mitochondrial proteins (G). (H) BC cells with the indicated transfection were cultured under glucose-starvation or normal conditions for 24 h. BC cells were stained with MitoTracker (red) and ubiquitin antibody (green). Ubiquitin – mitochondrial colocalization appeared as yellow fluorescence. Scale bar: 10 μm. (I – K) CKGC motif is essential for MANF to reduce ROS production (I), induce mitophagy (J), and facilitate BC cell survival (K) under glucose-starvation conditions (n = 3/condition). All data are presented as means ± SDs (I – K); p values were determined using one-way ANOVA (I – K).

We next investigated the mechanism by which MANF induces mitophagy through PRKN. We first evaluated the effects of MANF on the mitochondrial translocation of PRKN. BC cells were treated with Baf A1 to inhibit lysosomal degradation of PRKN during mitophagy. Our western blotting and IF results indicated that MANF KD did not influence the mitochondrial abundance of PRKN (Figure S8B, C). Because our results demonstrated that MANF was specifically bound to the active form of PRKN (Figure S7H), we investigated whether MANF influences the E3 ubiquitin ligase activity of PRKN. We detected the level of PRKN autoubiquitination and mitochondrial protein ubiquitination to evaluate PRKN activity. PINK1-overexpressing HeLa cells were transfected with the ubiquitin and PRKN genes to induce PRKN autoubiquitination [28]. Our western blotting results demonstrated that the MANF induced PRKN autoubiquitination (indicated by PRKN – ubiquitin bands) in cells under glucose-starvation conditions (Figure 6B). Furthermore, MANF overexpression was associated with a relatively high level of mitochondrial protein ubiquitination in cells under glucose-starvation conditions (Figure 6C). These results indicated the role of MANF in increasing the E3 ubiquitin ligase activity of PRKN under glucose-starvation conditions.

Previous studies have reported that the E3 ubiquitin ligase of PRKN is deactivated by ROS-induced oxidation at its cystine residues [29,30]. Thus, we investigated whether glucose starvation-induced oxidative stress (ROS) promotes cystine oxidation of PRKN. Cells under glucose-starvation conditions demonstrated high levels of PRKN oxidation (Figure S8D); moreover, the oxidation level of the P5 mutant (containing RING II domain) demonstrated a particular increase (Figure S8D). In addition, FCCP/CCCP treatment did not induce PRKN oxidation indicating that PRKN oxidation was associated with ROS accumulation (Figure S8E). Thus, PRKN activity might be affected by protein oxidation under glucose-starvation conditions. Similarly, increased MANF oxidation was observed in cells under glucose-starvation conditions (Figure S8F). MANF overexpression decreased PRKN oxidation and that of its P5 mutant in cells under glucose-starvation conditions (Figure 6D). By detecting the oxidation level of MANF truncated mutants, we observed that the oxidation levels of M2/M3 (containing SAP domain) were significantly higher in cells under glucose-starvation conditions than in cells under normal conditions (Figure S8F). M2 or M3 mutant overexpression, nevertheless, reduced PRKN oxidation (Figure S8G). Our structural analysis indicated that a CKGC motif is located in MANF’s SAP domain. CKGC motif acts as antioxidant that it neutralizes ROS via oxidization of its cysteine residues [31]. Thus, we mutated the CKGC motif of MANF (SKGS) by replacing cysteines with serines and investigated whether CXXC motif involves in the regulation of PRKN oxidation. SKGS mutant of MANF demonstrated a decreased level of protein oxidation indicated that cysteine residues of CKGC motif are the major sites for oxidation in MANF (Figure 6E). Importantly, SKGS mutant overexpression could not reduce PRKN (P5) oxidation (Figure 6E). MANF SKGS mutant also could not increase the E3 ubiquitin ligase activity of PRKN (Figure 6F, G). Thus, MANF inhibits PRKN oxidation via its CKGC motif. IF staining using a ubiquitin antibody demonstrated that MANF KD was associated with decreased mitochondrial membrane protein ubiquitination (Figure 6H). However, MANF WT but not SKGS mutant restored ubiquitination of mitochondrial membrane proteins in MANF-KD cells under glucose-starvation conditions (Figure 6H). Overexpression of MANF but not of SKGS mutant restored mitophagy-mediated degradation of PRKN in MANF-KD cells (Figure S8H). Similarly, MANF WT but not SKGS mutant reduced ROS levels and facilitated mitophagy-mediated survival of MANF-KD cells under glucose-starvation conditions (Figure 6I–K). Our results, therefore, indicated that MANF regulates PRKN-mediated mitophagy by inhibiting PRKN oxidation.

MANF induces BC tumor growth and lung metastasis

Cancer cells must overcome challenges related to nutrient stress (particularly glucose shortage) during tumor growth and metastasis [32]. To investigate the effects of MANF on tumor growth, we cultured BC cells under ECM-deprivation conditions, which inhibited glucose consumption of BC cells [3]. Under these conditions, MANF overexpression induced colony growth, reduced intracellular ROS production, and increased viability in BC cells (Figure S9A). This function of MANF was dependent on mitophagy and FAO regulation because the inhibition of mitophagy and FAO was noted to significantly reduce viability and colony formation in MANF-overexpressing cells under ECM-deprivation conditions (Figure S9A). Moreover, MANF-overexpressing cells required PRKN expression to survive under these conditions (Figure 7A and Figure S9B). In vivo, BC tumors with MANF overexpression exhibited increased tumor weight and decreased ROS levels (Figure 7B, C and Figure S9C). PRKN knockdown significantly inhibited tumor growth and induced ROS accumulation in MANF-overexpressing tumors (Figure 7B, C and Figure S9C). Moreover, expression of MANF WT but not SKGS mutant restored colony growth of MANF-KD BC cells through a decrease in ROS production (Figure 7D and Figure S9D). Re-expression of MANF WT significantly reduced ROS levels and induced tumor growth in MANF-KD tumors (Figure 7E, F and Figure S9E). These results indicated that MANF supports BC tumor growth by regulating PRKN oxidation.

Figure 7.

MANF induces tumor growth and lung metastasis in BC. (A – C) PRKN is essential for MANF to facilitate anchorage-independent growth in BC cells (A, n = 3/condition), promote BC tumor growth (B, n = 3/condition), and reduce ROS production in BC tumor tissues (C, n = 3/condition). BC cells with MANF overexpression or MANF overexpression + PRKN-KD (PRKNsh1) were cultured in ultralow attachment dishes for 72 h. (D – F) CKGC motif is essential for MANF to facilitate anchorage-independent growth in BC cells (D, n = 3/condition), promote BC tumor growth (E, n = 3/condition), and reduce ROS production in BC tumor tissues (F, n = 3/condition). MANF KD, MANFsh1. (G) PRKN is essential for MANF to confer lung metastasis in BC cells. Luciferase-expressing MDA-MB-231 cells were injected into the tail veins of mice. Lung metastasis was detected using IVIS; representative IVIS images (left). Luciferase signals were used to quantify tumor burden in mice (right, n = 5/condition). (H) Survival analysis of mice injected with BC cells with MANF overexpression or those with MANF overexpression+PRKN KD (PRKNsh1). (I) CKGC motif is essential for MANF to confer lung metastasis in BC cells (n = 5/condition). (J) Survival analysis of mice injected with the indicated BC cells. All data are presented as means ± SDs (A – G and I); p values were determined using one-way ANOVA (A – G and I) or log-rank test (H and J).

Because MANF increased BC cell viability under ECM-deprivation conditions, we investigated whether MANF induces BC colonization by using a lung metastasis model. The tumor burden was higher in the lungs of mice injected with MANF-overexpressing cells than in those of control mice (Figure 7G and Figure S9F). PRKN KD attenuated cancer metastasis of BC cells overexpressing MANF (Figure 7G and Figure S9F). Mice injected with MANF-overexpressing BC cells exhibited shorter survival than the control mice, whereas PRKN KD in BC cells with MANF overexpression prolonged mouse survival (Figure 7H). Moreover, BC lung metastasis was prevented by MANF KD, which could be restored by MANF WT overexpression but not by SKGS mutant overexpression (Figure 7I and Figure S9G). These results corroborated those of our mouse survival analysis (Figure 7J).

MANF expression predicts BC prognosis

Finally, we investigated the prognostic value of MANF expression in BC tissues. We obtained 513 surgically resected BC tissues and subjected them to IHC. MANF staining was predominantly detected in BC cell cytoplasm; it could also be detected in the cytoplasm and nucleus of some stromal cells (Figure 8A). BC tissues from patients who demonstrated distant metastasis during follow-up exhibited higher MANF expression than those of tissues from distant metastasis-free (DMF) patients (Figure 8A). MANF expression was associated with a higher T and N stage (Figure 8B, C). Tissues of the BC subtype with higher aggressiveness and metastasis potential (including triple-negative BC [TNBC] and ERBB2/Her2-overexpression BC [Her-2 BC]) displayed more intense MANF staining (Figure 8D). Based on TCGA-BRCA database, we also observed that TNBC displayed higher MANF expression than Luminal A/B BC (Figure 8E). Our survival curves analysis showed that the higher the MANF expression, the shorter were the overall survival (OS), relapse-free survival (RFS), and DMF survival (DMFS) among the BC patients within 12.5 years (150 months) of BC diagnosis (Figure 8F). Similar results were observed in our analysis of the Kaplan – Meier Plotter database (Figure 8G). However, we observed that higher MANF expression was associated with decreased mortality risk after 12.5 years of BC diagnosis, while no significant difference between RFS of both groups was observed (Figure 8H). The above results might be caused by the selection biases that a significantly higher proportion of patients with TNBC were identified as MANF high, while the MANF-low group contained a higher proportion of patients with LumA/B BC (Figure 8I). Subgroup analysis indicated that the higher the MANF expression, the higher was the distant metastasis risk in all ages of BC patients or in patients with T2 BC, N2 BC, LumA/B BC, Her-2 BC, and TNBC (Figure 8J).

Figure 8.

MANF expression predicts BC prognosis. (A) Tissues of BC patients demonstrating distant metastasis during follow-up displayed increased MANF expression. Quantification of MANF staining in BC tissues (right); representative image of MANF staining (left). Scale bar: 50 μm. All data are presented as means ± SDs. (B – D) MANF expression in subgroups of BC defined by T stage (B), N stage (C), and molecular subtypes (D). (E) MANF expression in luminal A/B BC, Her2 BC and TNBC. Data was obtained from TCGA-BRCA database. (F) Higher MANF expression was associated with shorter OS (left), RFS (middle), and DMFS (right) in BC patients (within 150 months). (G) High MANF expression was associated with shorter OS (left), RFS (middle), and DMFS (right) in BC patients. Data were obtained from the Kaplan–meier plotter database (https://kmplot.com/analysis/). (H) Landmark analysis indicated that high MANF expression was associated with shorter OS (left) and RFS (right) within 150 months of BC diagnosis in BC patients, while it was associated with better OS after 150 months (left). (I) Patients with manf-high BC exhibited higher proportion of TNBC (MANF high vs MANF low: 29.9% vs 21.5% at baseline, 41% vs 20% at 150 months follow-up). (J) Forest plot of the DMFS subgroup analysis with respect to MANF expression: MANF high vs. MANF low. All data are presented as means ± SDs (A-E), p values were determined using two-tailed Student’s t test (A), one way ANOVA (B-E) or log-rank test (F – H).

Discussion

During tumor growth and metastasis, BC cells are continually exposed to oxidative stress due to glucose starvation. The mechanisms by which BC cells maintain intracellular oxidation homeostasis remain unclear. In the present study, we noted that MANF-mediated mitophagy has a major role in BC to overcome lethal levels of glucose starvation-induced oxidative stress (Figure 9).

Figure 9.

Schematic of mechanisms by which MANF facilitates BC cell survival under glucose-starvation conditions by regulating mitophagy. Schematic of mechanisms by which MANF facilitates BC cell survival under glucose-starvation conditions by regulating mitophagy. MANF-mediated mitophagy promotes FAO and cell survival. Under glucose-starvation conditions, SENP1-mediated de-SUMOylation of MANF facilitates its translocation to mitochondria. MANF interacts with PRKN and alleviates PRKN oxidation via its CXXC motif.

MANF is an ERS-responsive protein; it is upregulated under ERS, and it protects cells from ERS-induced death [33]. In human cancers such as BC, HCC, and intrahepatic cholangiocarcinoma (ICC), MANF expression is upregulated and correlated with poor prognoses [20,34,35]. The MANF promoter region contains a functional ER stress element II, which renders its recognition by ERS-responsive transcription factors (including ATF6 and XBP1) [15]. As glucose shortage leads to ERS [36], MANF upregulation might be mediated by ERS-responsive signaling. Notably, we observed that in BC cells under glucose-starvation conditions, MANF upregulation induced cell survival, which is ERS-responsive signaling independent. MANF was also detected in multiple intracellular compartments – including not only ER and Golgi apparatus but also endosomes, nucleus, and mitochondria – indicating that MANF might have a complicated role in cancer cells [21,37,38]. For instance, secreted MANF contributes to radioresistance in melanoma by promoting the cellular DNA damage response [39]. Intracellular MANF induces sorafenib resistance in ICC through ERS response mitigation [35]. In contrast, MANF in cell nucleus acts as a tumor suppressor and binds to RELA/p65 inhibiting the NFKB/NF-κB signaling in HCC [21]. In this study, we observed that MANF translocated to mitochondria to induce mitophagy and thus facilitate survival in BC cells under glucose-starvation conditions. SENP1, which is upregulated in BC cells under glucose-starvation conditions, mediates de-SUMOylation of MANF and facilitates its mitochondrial translocation. Contrarily, a recent study reported that MANF translocated to cell nucleus but not to mitochondria in HCC cells under oxygen-glucose deprivation [21]. This might be attributed to the relatively low SENP1 expression in HCC tissues compared with BC tissues (GEPIA 2: http://gepia2.cancer-pku.cn/#index). As SENP1 regulates MANF intracellular distribution, SENP1 inhibition promoting nuclear localization of MANF and meanwhile reducing its mitochondrial localization might exert synergistic anticancer effects and could be a potential treatment strategy for BC. Although MANF was reported to stimulate mitophagy through AMPK-FOXO3 signaling in the cytoplasm of renal tubule cells [19], we detected an increased mitochondrial abundance of MANF in BC cells under glucose-starvation conditions indicating that MANF might also regulate mitophagy through mechanisms other than AMPK/FOXO3 signaling. Interestingly, AMPK has been shown to activate SENP1 under glucose-starvation conditions [40]; therefore, whether there is a positive feedback loop between MANF and AMPK is worth further study.

Mitophagy is an evolutionarily conserved cellular process, and its role in cancer is complicate. In normal cells, mitophagy eliminates dysfunctional mitochondria, inhibiting malignant cell transformation [12]. Several studies also reported that defective mitophagy activity is advantageous to tumorigenesis [41,42]. Although mitophagy overactivation is lethal to cancer cells, mitophagy can be induced to relieve cellular stress and support cancer cell survival [43]. HIF1A-induced mitophagy can prohibit anoikis and promote cancer metastasis in the lungs [14,44], whereas ULK1-mediated mitophagy attenuates BC bone metastasis under a hypoxic environment through the inhibition of NLRP3 inflammasome activation [45]. The role of mitophagy in cancer varies depending on both the cellular context and the activation mechanism. Our current results revealed that MANF is a critical mitophagy regulator in BC cells under glucose-starvation conditions. In addition to damaged mitochondrial clearance, MANF-mediated mitophagy influences FAO activity in BC cells under glucose-starvation conditions. FAO, attenuated by glucose starvation-induced ROS production, is a critical NADPH source in cancer cells during glucose deprivation [3]. MANF alleviates ROS accumulation via mitophagy; this might be the mechanism by which MANF-mediated mitophagy facilitates cellular FAO and survival in BC cells under glucose-starvation conditions. Similarly, a recent study indicated that the maturation of mitochondrial fatty acid metabolism requires PRKN-mediated mitophagy in cardiac tissues [46]. Thus, in BC cells under nutrient stress, MANF-PRKN-mediated mitophagy is involved in not only mitochondria integrity maintenance but also metabolic reprogramming.

The E3 ligase activity of PRKN is essential for the initiation of PINK1-PRKN-dependent mitophagy, whereas human cancer cells demonstrate PRKN dysfunction [47]. PARK2, which encodes PRKN protein, maps to human chromosome 6q25–q26, and it is frequently deleted in human cancer cells [42]. Moreover, inactivating PARK2 mutations have been noted in glioblastoma, colon cancer, and lung cancer cells [48]. However, PARK2 deletions or mutations are rarely noted in BC cells [49]; therefore, in BC, genome alterations may not play a major role in PRKN activity regulation. Recent accumulating evidence indicates that PRKN activity might be influenced by posttranslational modifications, such as phosphorylation, S-nitrosylation, Neddylation, and oxidation [50]. On exposure to ROS (e.g., hydrogen peroxide), cysteine molecules within PRKN (e.g., those on cysteine-rich RING2 domains) become oxidized, eventually reducing the E3 ligase activity of PRKN [29]. Similarly, we detected PRKN oxidation in BC cells under glucose-starvation conditions, particularly in MANF-KD cells. Tumor tissues often display increased ROS levels compared with normal tissues due to nutrient shortage and hypoxia; thus, PRKN might be more vulnerable to ROS-mediated protein oxidation. This might explain why PRKN is often deactivated in solid tumors. When cancer cells encounter lethal nutrient stress (e.g., glucose and ECM deprivation), MANF becomes transcriptionally activated and translocates to mitochondria to interact with PRKN. MANF alleviates PRKN oxidation via its antioxidative CXXC motif. Thus, PRKN-mediated mitophagy, defective in cancer cells under normal conditions, is restored by MANF to protect BC cells from lethal cellular stress. This result is consistent with the results of studies indicating that PRKN is generally defective in human cancers.

The thiol groups of cysteine residues in proteins are subject to oxidation modifications (including disulfides formation, S-nitrosylation, S-sulfination, and S-sulfination), potentially causing protein deactivation through protein misfolding, aggregation, and degradation [50]. Because the RING and IBR domains of PRKN enrich cysteine residues, protein oxidation plays a critical role in the regulation of PRKN’s E3-ligase activity [29]. The current results demonstrated that the CXXC motif of MANF sustains the E3 ligase activity of PRKN by inhibiting PRKN oxidation. Cysteine residues of the CXXC motif can be oxidized and hence neutralizes ROS [31]. Thus, we assumed that MANF binds to PRKN acting as an antioxidant and protects PRKN from ROS-mediated oxidative attack.

In conclusion, our study revealed that MANF facilitates BC cell survival under glucose-starvation conditions by regulating mitophagy. Under glucose-starvation conditions, MANF-mediated mitophagy promotes FAO and cell survival. SENP1-mediated de-SUMOylation of MANF facilitates its translocation to mitochondria in BC cells under glucose-starvation conditions. MANF interacts with PRKN and alleviates PRKN oxidation via its CXXC motif. In vivo, MANF can promote tumor growth and cancer metastasis by inhibiting PRKN oxidation. Taken together, our results confirm the cytoprotective role of MANF-mediated mitophagy in BC.

Materials and methods

Cell culture and plasmid construction

The culture media used for the human embryonic kidney cell line (293T; ATCC, CRL-3216), HeLa cell line (ATCC, CRM-CCL-2), and BC cell lines (MCF-7, BT474, SKBR3, and MDA-MB-231; ATCC, HTB-22, HTB-20, HTB-30, CRM-HTB-26) included Dulbecco’s modified Eagle’s medium (Invitrogen 11,965,118), RPMI-1640 (Invitrogen 11,995,073), or McCoy’s 5a Medium Modified (Invitrogen 16,600,082) – all with 10% FBS and 1% penicillin-streptomycin. All cells were maintained at 37°C under 5% CO₂. For serum, glutamine, or glucose starvation: cells (1 × 10⁵/mL) were cultured for 12 h and then incubated in a serum-, glutamine-, or glucose-free culture medium for 24 h.

A 3× FLAG tag was added between the N-terminal ER targeting signal sequence and the shRNA-resistant form of human MANF, as reported previously [48]. MANF, PRKN, and the deletion mutants were expressed using pCDH-CMV-MCS-EF1 (Addgene 72,266; deposited by Kazuhiro Oka) with a 3× FLAG or MYC tag. We amplified and cloned SUMO1, SUMO2, and SUMO3 cDNA into pcdna3.1 (Invitrogen, V79020); ubiquitin-encoding cDNA into pCDEF (Solarbio, VT021949-1EA); and PINK1 cDNA into the PLVX-IRES vector (Qiyunbio, QP1452). The pLKO.1 vectors (Sigma, SHC001) containing shRNA were used to inhibit MANF, SENP1, DNM1L and PRKN expression. All the shRNA target sequences used here are listed in Table S3. Lentiviral transduction system was used for stable transfection of pCDH-CMV-MCS-EF1, PLVX-IRES and pLKO.1-based plasmids. Cells were added with lentivirus generated by transfecting 293T cells with the lentiviral vector and the plasmids using Lipofectamine 3000 (Invitrogen, L3000008). Culture medium containing lentivirus was collected at 48 h and filtered using a 0.45-mm filter. Cells were cultured with lentivirus for 24 h and selected with puro or hygromycin.

For CRISPR-Cas9-mediated pink1 and manf knockout in BC cells, the specific sgRNA sequences “5-CGTCTCGTGTCCAACGGGTC-3” targeting the PINK1 gene (NM_032409.3) and “5-ACGAACTTATAAAGTTCTGC-3” targeting the MANF gene (NM_006010.6) were cloned into the pXPR_001 plasmid (Addgene 49,535; deposited by Feng Zhang). For CRISPR-Cas9-mediated knockout (KO) of pink1/manf, human PINK1/MANF targeting sgRNAs were transfected into BC cells using Lipofectamine 3000 (Invitrogen, L3000008). After 36 h, single GFP-positive cells were sorted using fluorescence-activated cell sorting and seeded in 96-well plates. pink1-KO and manf-KO clones were identified via western blotting and sequencing.

Chemicals and compounds

2DG (Sigma, D8375), 5TG (Aladdin, T107913), FBP (Santa Cruz Biotechnology, sc -214,805), NAC (Sigma, V900429), and GSH (Solarbio, G8180) were used at a concentration of 5 mM. The EIF2AK3/PERK inhibitor GSK2606414 (Selleck, S7307) and the ERN1/IRE1 inhibitor 4μ8C (Selleck, S7272) were used at concentrations of 2 and 25 µM, respectively. Human MANF recombinant Protein (PeproTech, 450–06) was used at a concentration of 100 ng/mL. Baf A1 (Selleck, S1413), Mdivi-1 (Selleck, S7162), and PMI (MCE, HY-115576) were used at a concentration of 100 nM, 50 µM, and 10 µM, respectively. CCCP (MCE, HY-100941) and FCCP (Selleck, S8276) were used at a concentration of 10 µM and 1 µM, respectively. Etomoxir (MCE, HY-50202) and BPTES (MCE, HY-12683) were used at concentrations of 40 and 10 µM. MG132 (Sigma, M7449) was used at 10 µM, whereas CHX (Sigma 239,763-M) was used at 100 µg/mL.

NADP⁺:NADPH measurement

We measured the relative ratio of NADP⁺:NADPH by using an NADP/NADPH Quantification Kit (Sigma, MAK038), according to the manufacturer’s instructions. Cells were incubated with 800 µL of the NADP/NADPH extraction buffer for 10 min on ice. For NADPH detection, 200 µL of NADP/NADPH-extracted samples were heated at 60°C for 30 min and cooled on ice. Next, to convert NADP to NADPH, the Master Reaction Mix was added, followed by incubation at room temperature for 5 min. The NADPH developer was added, and NADP/NADPH levels were measured using a colorimetric assay (at 450 nm).

Live cell imaging

The plasmid encoding the fluorescent protein mt-Keima (Addgene 72,342; deposited by Richard Youle) was used to assess mitophagy in living cells [51]. Cells (1 × 10⁵/mL) stably expressing mt-Keima were plated for 12 h and cultured in glucose-free medium for 24 h. Keima fluorescence signals under neutral and acidic pH are green (Ex = 458 nm/Em = 588–633) and red (Ex = 561 nm/Em = 588–633), respectively. The fluorescence intensity was evaluated using ImageJ. The cellular mitophagy degree was determined as the ratio of red fluorescence intensity to that of green fluorescence intensity. The experiments were performed in triplicate.

FAOBlue (Funakoshi, FDV-0033) is a coumarin dye used to measure FAO activity in viable cells [52]. FAOBlue is detected only after metabolization through FAO. After being degraded during FAO, the resulting coumarin dye demonstrates blue fluorescence (Ex = 405 nm/Em = 430–480 nm). BC cells were washed twice with HEPES-buffered saline (HBS; dextrose, 2.0 g/L, HEPES, pH 7.4, 10 g/L, KCl, 0.74 g/L, NaCl, 16 g/L, Na₂HPO₄.2 H₂O, 0.27 g/L) and incubated with serum-free medium containing 10 μM FAOBlue at 37°C for 45 min. Then, cells were washed with HBS and observed under live conditions with blue fluorescence (Ex = 405 nm/Em = 430–480 nm).

FAO assay

FAO activity was determined using an FAO Complete Assay Kit (Abcam, ab222944). In total, 2 × 10⁴ cells were seeded in a 96-well plate for 12 h and then cultured in a glucose-free medium for 24 h. These cells were washed with FA-Free Measurement Media (Base Measurement Media, 0.5 mM L-Carnitine, and 4 mM glutamate) and then incubated with 90 µL of FA Measurement Media (FA-free Measurement Media +150 µM FAO Conjugate) and 10 µL of O₂ consumption reagent each well. Then, 2.5 µM FCCP, 10 µM Etoxomir, or BSA (negative control; Sigma, A1933) were added to the FA Measurement Media. Each well was sealed with prewarmed high-sensitivity mineral oil, and the resulting fluorescence was read on a fluorescence microplate reader every 1.5 min. FAO activity was determined using the initial slope from the linear portion of the FAO signal profile.

Protein extraction and western blotting

To extract nuclear proteins from the cells, we used the Nuclear and Cytoplasmic Extraction Kit (Thermo Fisher Scientific; 78833). In brief, the cells were trypsinized and centrifuged at 500 g for 5 min. The cell precipitate was then incubated in ice-cold CER-I and CER-II reagents for cytoplasmic protein extraction. The remaining pellet fraction was incubated in ice-cold NER reagent for nuclear protein extraction.

To extract mitochondrial proteins from the cells, we used the Mitochondria Isolation Kit (Thermo Fisher Scientific 89,874). In brief, the cells were harvested and incubated in Reagents A, B, and C to obtain cell lysates. Next, sequential centrifugation was performed at 700 g and 37°C for 10 min, followed by that at 12,000 g and 37°C for 15 min. The mitochondria-containing precipitate was boiled with SDS-PAGE sample buffer and then analyzed through western blotting.

Antibodies used in western blotting assays were listed in Table 1. Bands of western blotting assays were quantified using ImageJ software and the protein expression levels were assessed by calculating the ratio of the target protein density to the internal reference protein density.

Table 1.

Antibodies with the applications in which they were used.

Assay/Protein Name	Catalog No.	Company	Concentration	Source
Western blotting
MANF	SAB1406636	Sigma-Aldrich	1 μg/mL	Mouse
MANF	PA5 -96,542	Invitrogen	1:500	Rabbit
ACTB/β-actin	3700	CST	1:1000	Mouse
PRKN	PA5 -13,399	Invitrogen	1:1000	Rabbit
PRKN	4211	CST	1:1000	Mouse
HSPD1	12165	CST	1:1000	Rabbit
SUMO1	SAB1402954	Sigma-Aldrich	5 μg/mL	Mouse
SUMO2/SUMO3	4971	CST	1:1000	Rabbit
SENP1	11929	CST	1:1000	Rabbit
SENP2	SAB1407830	Sigma-Aldrich	1 μg/mL	Mouse
FLAG	F7425	Millipore	1 μg/mL	Rabbit
Histone H3	4499	CST	1:2000	Rabbit
TOMM20	42406	CST	1:1000	Rabbit
VDAC	4661	CST	1:1000	Rabbit
SQSTM1	5114	CST	1:1000	Rabbit
LC3B	2775	CST	1:1000	Rabbit
DNM1L/DRP1	8570	CST	1:1000	Rabbit
CPT1A	ab234111	abcam	1:1000	Rabbit
CPT1B	ab134135	abcam	1:1000	Rabbit
ACSL4/FACL4	SAB2100035	Sigma-Aldrich	1 μg/mL	Rabbit
MYC	SAB1305535	Sigma-Aldrich	1:2000	Mouse
PINK1	6946	CST	1:1000	Rabbit
Ubiquitin	701339	Invitrogen	1:1000	Rabbit
Anti-Rabbit Secondary Antibody	31460	Invitrogen	1:10000	Rabbit
Anti-Mouse Secondary Antibody	31430	Invitrogen	1:10000	Mouse
Immunofluorescence
MANF	PA5 -96,542	Invitrogen	1:50	Rabbit
MANF	SAB1402974	Sigma-Aldrich	40 μg/mL	Mouse
PRKN	PA5 -13,399	Invitrogen	1:20	Rabbit
LC3B	L7543	Sigma-Aldrich	5 μg/mL	Rabbit
Ubiquitin	701339	Invitrogen	1 μg/mL	Rabbit
Alexa Fluor 488 Secondary Antibody	A-11008	Invitrogen	4 μg/mL	Rabbit
Alexa Fluor 488 Secondary Antibody	A-11001	Invitrogen	1 μg/mL	Mouse
Alexa Fluor 647 Secondary Antibody	A-21245	Invitrogen	2 μg/mL	Rabbit
Immunoprecipitation
MANF	PA5 -96,542	Invitrogen	1:200	Rabbit
PRKN	4211	CST	1:50	Mouse
Immunohistochemistry
MANF	PA5 -96,542	Invitrogen	1:100	Rabbit

CST: Cell Signaling Technology.

Biotin labeling of oxidized protein

MANF and PRKN oxidation were detected using biotin labeling of oxidized protein assay [53]. Cells in a 100-mm dish were lysed using biotin-labeling lysis buffer (BLLB; 50 mM Tris-HCl, pH 7.0, 5 mM EDTA, 120 mM NaCl, and 0.5% IGEPAL CA-630 [Sigma, I8896]) with a protease inhibitor cocktail (Thermo Fisher Scientific 87,786) and 100 mM maleimide (Sigma 129,585) on ice for 15 min. The cell lysates were centrifuged at 20,000 g for 5 min, and the precipitate was discarded. The protein concentration in the supernatant was adjusted to 1 µg/µL using BLLB.

Next, the protein lysates were mixed with 1% SDS through rotation at room temperature for 2 h. The redundant maleimide was removed through incubation with 5 volumes of preequilibrated acetone at − 20°C for 20 min. This was followed by centrifugation at 20,000 g at 4°C for 10 min to obtain a protein precipitate. After air drying, the proteins were resuspended using 200 µL of BLLB containing 10 nM DTT, 1% SDS, and 100 µM biotin-maleimide (Sigma, B1267; dissolved in dimethylformamide) to reduce and label the oxidized sulfhydryl groups of the proteins. This was then followed by incubation with 5 volumes of acetone at − 20°C for 20 min. After centrifugation and air-drying, the protein precipitate was resuspended using 0.5 mL of BLLB containing 10 µL of streptavidin – sepharose resin (Cytiva 17,511,301) at 4°C for 2 h and mixed by rotating. The resin was washed four times with BLLB and boiled in an SDS-PAGE buffer. The protein samples were then assessed through western blotting.

Animal experiments

We obtained 4–5-week-old BALB/c-nude female mice from GemPharmatech. To induce a xenograft tumor model, 1 × 10⁶ MDA-MB-231 cells with the indicated transfection were injected into the left mammary fat pads of the mice. After 42 days of injection, we sacrificed the mice and recorded their tumor volumes and weights. ROS levels in fresh tumor tissues were detected using a DCF ROS/RNS Assay Kit (Abcam, ab238535). A 50-mg tissue sample was homogenized in 1 mL of PBS on ice. The obtained homogenate was then centrifuged at 10,000 g for 5 min, and the precipitate was removed. Next, 50 µL of the supernatant was incubated with 50 µL of a catalyst in each well of a 96-well plate at room temperature for 5 min. DCFH solution (100 µL) was then added to the supernatant, followed by incubation at room temperature for 45 min. Then, the samples were read on a fluorescence plate reader at Ex = 480 nm/Em = 530 nm. Another 50 µL of tissue supernatant was used for protein concentration detection. ROS levels in the tumor tissues were calculated as [H₂O₂ concentration (RFU)/protein content (µg)] per 50 mg of tissue.

To induce a lung metastasis model, 5 × 10⁵ MDA-MB-231 cells with the indicated transfection were suspended in a volume of 100 µL and injected in the tail vein of mice. Lung metastasis was detected weekly using an in vivo imaging system (IVIS). The lungs were collected and fixed in picric acid. Tumor burden in the lungs was assessed by measuring the number of lung surface nodules and staining paraffin-embedded lung tissues with hematoxylin and eosin (H&E).

Anchorage-independent growth assay and extracellular matrix-deprived culture

In total, 1 × 10⁴ cells were cultured for 3 weeks. Spheroid size was measured under an inverted microscope. The frequency of spheroid formation (>50 μm) per millimeter-squared was calculated microscopically (magnification, 100×). For suspension culture, 1 × 10⁶ cells were cultured in a Corning ultralow attachment dish for 72 h.

Quantitative real-time reverse transcription PCR (qRT-pcr)

QRT-PCR was performed using TB Green Fast qPCR Mix (TaKaRa, RR820A), according to the manufacturer’s instructions. The sequences of the primer used for qRT-PCR are listed in Table S3.

Immunofluorescence (IF)

Cells were fixed on a cover glass with 4% paraformaldehyde at room temperature for 30 min and then blocked with 1% BSA at room temperature for 30 min. Next, they were incubated with appropriate concentrations of antibodies at 4°C overnight. The cells were then incubated with secondary antibodies conjugated with Alexfluor-488 (Invitrogen, A-11008/A-11001) or Alex Fluor 647 (Invitrogen, A-21245) in the dark for 1 h; then, their nuclei were stained with DAPI. Images were taken under a confocal fluorescence microscope. For mitochondrial staining, the cells were incubated with 200 nM MitoTracker Red (Cell Signaling Technology, 9082) at 37°C for 45 min. Then, they were fixed with methanol at − 20°C for 15 min. The fixed sample was washed with PBS and used for other IF staining procedures. Antibodies used in IF assay were listed in Table 1.

Immunoprecipitation (IP) and protein interaction identification

Cells were lysed with IP lysis buffer (Thermo Fisher Scientific 87,787) containing 1 mM PMSF (Sigma 78,830) on ice for 20 min. The obtained cell lysates were incubated with relevant antibodies, anti-FLAG agarose beads (Thermo Fisher Scientific, A36798) or anti-MYC agarose beads (Thermo Fisher Scientific 20,168) at 4°C overnight. The cell lysates containing antibodies were then mixed with protein A/G magnetic beads (Thermo Fisher Scientific 88,802) at room temperature for 1 h. The protein-bound beads were heated or incubated with 3× FLAG or MYC peptide (Sigma, F4799/M2435) in SDS-PAGE protein loading buffer at 96°C–100°C for 10 min. Antibodies used in IP assay were listed in Table 1.

To evaluate proteins bound to MANF, 3× FLAG-MANF was expressed in 293T cells cultured in a glucose-free or -containing medium and then immunoprecipitated using anti-FLAG agarose beads (Thermo Fisher Scientific, A36798). Proteins immunoprecipitated with MANF were assessed through western blotting and silver staining (Thermo Fisher Scientific 24,612). Silver-stained bands specifically noted in glucose-starvation cells were excised and analyzed through mass spectrometry.

Immunohistochemistry (IHC)

In total, 513 formalin-fixed paraffin-embedded (FFPE) breast cancer (BC) tissues were collected from Sun Yat-Sen University Cancer Center (SYSUCC). The clinicopathological characteristics of these BC patients are listed in Table S1. Antibody used in IHC assay was listed in Table 1. Patient prognoses were recorded through outpatient follow-ups or telephone interviews. MANF expression in FFPE tissues was detected through IHC. MANF expression levels were determined using the staining index (SI), calculated as staining intensity × proportion of positive staining cells. Staining intensity was defined as follows: 1, negative; 2, weak (light yellow); 3, medium (yellow brown); and 4, intense (brown). Proportions of positive-stained cells were defined as follows: 0, <1%; 1, 1%–10%; 2, 10%–35%; 3, 35%–75%; and 4, >75%. BC tissues with an SI of ≤ 4 and > 4 were considered to have low and high MANF expression, respectively. Comparison of baseline characteristics between MANF-high and MANF-low groups was listed in Table S2. Our human study was approved by the SYSUCC Review Board (GZR2022–273) and performed in accordance with the Declaration of Helsinki.

Statistical analyses

SPSS (version 22) was used for all data analyses. Landmark analysis was performed using R package (“jskm”). The statistical methods used in each assay and their precise p values are shown in their respective figure legends. A p value of < 0.05 was considered to indicate statistical significance.

Study approval

Our human study was approved by the SYSUCC Review Board (GZR2022–273) and performed in accordance with the Declaration of Helsinki. Animal experimental procedures were approved by the IACUC of Sun Yat-sen University (No. L102012022007D).

Funding Statement

This work was supported by the National Natural Science Foundation of China (grant numbers 82202850, 82072609, 81972459), National Key Research and Development Program of China (grant number 2020YFA0509400), Natural Science Foundation of Guangdong Province (grant number 2023A1515012746), Guangdong Basic and Applied Basic Research Foundation (grant numbers 2021A1515111191), and Science and Technology Projects in Guangzhou (grant numbers 2023A04J2392).

Disclosure statement

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

Data availability statement

The mass spectrometry data regarding the MANF-interacting proteins have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the iProX partner repository with the data set identifier PXD046689. The main data supporting the findings of this study are available within Data S1. Uncropped images for all blots and gels are available within Data S2.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2024.2392415