Investigation of MANF regulation of glioma stemness via STAT3/TGF-β/SMAD4/p38 pathway based on pan-cancer analysis

Highlights

• •MANF regulates glioma stemness and tumor progression via STAT3/TGF-β/SMAD4/p38 pathways.
• •MANF promotes self-renewal and drug resistance in glioma cells.
• •In vivo experiments confirm MANF’s effects on tumor growth and metastasis.
• •MANF is a potential therapeutic target for glioma, offering new insights into personalized therapy.

Abstract

Background

Glioma, particularly glioblastoma, is a highly aggressive brain tumor with poor prognosis and limited treatment options. Recent research highlights the role of MANF (Mesencephalic Astrocyte Derived Neurotrophic Factor) in tumor biology, yet its specific mechanisms in glioma remain underexplored. This study aims to elucidate the role of MANF in glioma and its underlying mechanisms of action.

Methods

We conducted bioinformatics analysis using TCGA data to identify MANF-related pathways, followed by cellular assays and subcutaneous tumor models for functional validation. Experiments included Western blot and qRT-PCR analysis to investigate the effects of MANF on glioma cell proliferation, migration, and stemness gene expression.

Results

MANF was found to be highly expressed in tumor tissues and associated with poor prognosis in glioma patients. Endogenous MANF regulates tumor cells by modulating the TGF-β/SMAD4/p38 pathway, promoting stemness and enhancing malignant behaviors, including migration and invasion. Exogenous MANF, however, did not significantly affect stemness gene expression but contributed to glioma cell proliferation.

Conclusions

MANF emerges as a promising therapeutic target for glioma. This study clarifies MANF's specific mechanisms, offering insights into its potential for targeted glioma therapies.

Graphical abstract

Introduction

Gliomas represent the most common and aggressive type of primary brain tumors [1], accounting for approximately 80 % of malignant brain tumors in adults. Epidemiologically, the incidence of gliomas varies globally, with high-grade gliomas, such as glioblastoma multiforme (GBM), having a particularly poor prognosis and a median survival of <15 months despite aggressive treatment [2]. Pathologically, gliomas are characterized by their infiltrative growth [3,4], high cellular heterogeneity [4], and genetic alterations [5], including mutations in genes such as IDH1 [6], TP53 [7], and EGFR [8]. These tumors exhibit a high propensity for invasion into surrounding brain tissue [9], making complete surgical resection challenging and contributing to high recurrence rates [10]. Current diagnostic approaches rely heavily on imaging techniques like MRI and CT scans [11], supplemented by histopathological examination and molecular profiling [12]. However, these methods often fall short in accurately predicting patient outcomes due to the complex molecular landscape of gliomas [13]. Treatment options are similarly limited, with standard protocols involving surgical resection followed by radiotherapy and chemotherapy with temozolomide [14]. Unfortunately, resistance to therapy and tumor recurrence are common, resulting in poor overall survival [15]. The prognostic outlook for glioma patients remains grim, particularly for those with high-grade tumors, underscoring the urgent need for novel diagnostic tools, more effective therapeutic strategies, and personalized approaches to improve patient outcomes.

Neurotrophic factors play a crucial role in the development, maintenance, and repair of the nervous system. They regulate neuronal survival, proliferation, differentiation, and repair responses, helping neurons cope with various physiological and pathological challenges. Well-known neurotrophic factors include Brain-Derived Neurotrophic Factor (BDNF), Nerve Growth Factor (NGF), and Glial Cell-Derived Neurotrophic Factor (GDNF), all of which play significant roles in neurodegenerative diseases, tumors, and various pathological conditions [16]. In recent years, another novel neurotrophic factor—mesencephalic astrocyte-derived neurotrophic factor (MANF)—has garnered widespread attention. MANF is a highly conserved protein initially identified for its neuroprotective properties in dopaminergic neurons [17]. Recent studies have expanded our understanding of MANF, revealing its involvement in a broad spectrum of cellular processes, including the regulation of endoplasmic reticulum (ER) [18] stress and the unfolded protein response (UPR) [19]. In the context of tumors, MANF has emerged as a critical regulator of cell survival [20], particularly under conditions of cellular stress commonly encountered in the tumor microenvironment. Its ability to modulate ER stress pathways has positioned MANF as a key player in tumor progression [20], where it can either promote or inhibit tumor growth depending on the cellular context. In gliomas, MANF expression has been associated with both protective and adverse effects on tumor cells [21]. Some studies suggest that MANF may contribute to glioma cell survival by enhancing the cell's ability to manage ER stress [18], thereby promoting resistance to apoptosis. Conversely, other research indicates that ER stress MANF can suppress glioma growth by inducing cytotoxic effects under certain conditions, such as nutrient deprivation or hypoxia [22]. This dual role of MANF in glioma biology underscores the complexity of its function and highlights the potential of MANF as a therapeutic target [23]. Current research is increasingly focused on elucidating the precise mechanisms by which MANF influences glioma progression and exploring its potential as a target for therapeutic intervention [24]. By targeting MANF path ways, there is the potential to disrupt critical survival mechanisms in glioma cells, offering a novel approach to glioma treatment that could complement existing therapies and improve patient outcomes.

Tumor stemness refers to the ability of certain cancer cells to maintain stem cell-like characteristics, including self-renewal, differentiation potential, and resistance to conventional therapies [25,26]. These cancer stem cells (CSCs) play a critical role in tumor initiation, progression, metastasis, and recurrence, contributing to the overall malignancy of tumors such as gliomas [27,28]. Understanding the molecular mechanisms that regulate tumor stemness is vital for developing new therapeutic strategies aimed at targeting CSCs, which are often resistant to standard treatments like chemotherapy and radiotherapy.

MANF has emerged as a protein of interest in the context of tumor biology, including its potential role in regulating tumor stemness [29]. Recent studies suggest that MANF may influence cancer cell survival, stress response, and resistance to apoptosis, all of which are key features of CSCs [20,30]. In gliomas, MANF has been implicated in promoting cellular survival in the tumor microenvironment, contributing to tumor growth and resistance to treatment [21]. However, the precise role of MANF in regulating the stemness properties of glioma cells remains poorly understood, and research in this area is still in its early stages. Furthermore, MANF has been linked to critical signaling pathways involved in stemness regulation, such as TGF-β, SOX2, Nanog, and c-Myc [[31], [32], [33]]. These pathways are well-known for their roles in maintaining the self-renewal and pluripotency of stem cells. In gliomas, dysregulation of these pathways contributes to the maintenance of the CSC population, promoting tumor growth and therapy resistance [[34], [35], [36]]. However, the specific interactions between MANF and these stemness-related factors (TGF-β, SOX2, Nanog, and c-Myc) in gliomas remain largely unexplored. This gap in knowledge highlights the need for further investigation into how MANF may regulate these pathways and contribute to glioma stemness. Given the potential of targeting CSCs to improve glioma treatment outcomes, MANF represents a promising therapeutic target [37]. Ongoing research aimed at elucidating the mechanisms by which MANF influences tumor stemness could lead to the development of novel therapies that more effectively target glioma stem cells, potentially overcoming resistance to current treatment options and improving patient prognosis.

Materials and methods

Data source

We analyzed RNA-seq expression data from 698 glioma samples obtained from The Cancer Genome Atlas (TCGA), which includes both low-grade glioma (LGG) and glioblastoma multiforme (GBM) cases [38,39]. The dataset comprises samples with varying clinical features, including different histological subtypes, WHO grades, IDH mutation status, and treatment histories.

Data filtering and the standard process

The RNA-seq data underwent preprocessing, including the removal of samples not meeting quality standards, such as those with abnormal expression values or missing clinical data. Genes with low expression, defined as having expression values below the 25th percentile in >50 % of samples, were filtered out. The gene expression matrix was log2-transformed and standardized using z-score normalization to eliminate batch effects. Finally, datasets from different cancer types were integrated into a unified expression matrix for subsequent pan-cancer analysis.

Gene selection

Gene selection was conducted through differential expression analysis, identifying genes specifically expressed in gliomas or those with prognostic significance compared to other cancer types. Gene set enrichment analysis (GSEA) was performed using the clusterProfiler package to conduct gene ontology (GO) and kyoto encyclopedia of genes and genomes (KEGG) enrichment analyses, identifying pathways and biological processes associated with glioma prognosis [40]. Prognostic genes were further screened using the Cox proportional hazards regression model, selecting genes significantly associated with patient survival (adjusted p-value < 0.05) as candidates for the prognostic model.

Gene expression and mutation landscape

The gene expression and mutation landscape analysis was conducted using data obtained from TCGA database. Gene expression profiles were processed using RNA-seq data, normalized by transcripts per million (TPM), and analyzed for differential expression using the DESeq2 package. Mutation data were retrieved from somatic mutation files (MAF format) and processed with the maftools package to visualize and assess mutation frequency and patterns. Key genes of interest were identified based on their expression levels and mutation burden. Further functional annotation and pathway enrichment analysis were performed to explore the biological significance of these genes in the context of the disease.

Relationship between genomic heterogeneity and MANF

To assess genomic heterogeneity and the role of MANF, we utilized whole-exome sequencing (WES) and RNA sequencing (RNA-seq) data from TCGA [41]. Somatic mutations and copy number variations (CNVs) were analyzed using GISTIC 2.0 and the maftools package to evaluate genomic instability. MANF expression was quantified from RNA-seq data and correlated with mutation burden and CNV profiles. Tumor heterogeneity was further assessed through single-cell RNA sequencing (scRNA-seq) data to investigate MANF expression across different tumor subpopulations. Functional enrichment analysis was performed to identify pathways associated with MANF in diverse genomic contexts.

Analysis of correlation between MANF gene and clinical features

To explore the correlation between MANF gene expression and clinical characteristics, RNA-seq data and corresponding clinical information were obtained from TCGA. MANF expression levels were analyzed and stratified based on key clinical variables, including age, gender, tumor grade, and survival outcomes. Correlation analyses were performed using Pearson's or Spearman's rank correlation tests, depending on data distribution. Kaplan-Meier survival analysis was conducted to evaluate the association between MANF expression and patient prognosis, and a multivariate Cox proportional hazards model was applied to adjust for confounding variables. Statistical significance was defined as p < 0.05.

Correlation analysis of immune infiltration and immune cells

To analyze immune infiltration and its correlation with immune cells, we utilized RNA-seq data from TCGA and applied the CIBERSORT algorithm to estimate the proportion of various immune cell types in the tumor microenvironment. Correlations between immune cell infiltration and key genes were assessed using Pearson’s or Spearman’s correlation coefficients. Additionally, immune infiltration scores were calculated using the ESTIMATE algorithm to evaluate the tumor's immune microenvironment. Statistical analyses were performed to identify significant associations between immune cell types and clinical outcomes, with a p-value of <0.05 considered statistically significant.

Correlation analysis between tumor stemness and MANF

To analyze the relationship between tumor stemness and MANF, we obtained RNA-seq data from TCGA and calculated stemness indices using the one-class logistic regression (OCLR) machine learning algorithm. MANF expression levels were then correlated with tumor stemness indices across different samples. Pearson’s or Spearman’s correlation tests were used to evaluate the strength of association between MANF expression and stemness scores. Additionally, functional enrichment analyses were conducted to explore the biological pathways associated with MANF in regulating tumor stemness. A p-value of <0.05 was considered statistically significant.

RNA modification gene analysis

To analyze RNA modification genes, specifically m1A, m5C, and m6A, RNA-seq data were obtained from TCGA and the Genotype-Tissue Expression (GTEx) database. Differential expression analysis of RNA modification-related genes was performed using DESeq2, comparing normal and tumor tissues. The expression levels of m1A, m5C, and m6A modification genes were visualized through heatmaps and boxplots. Correlation analyses were conducted to assess relationships between RNA modification gene expression and clinical characteristics. Functional enrichment analysis was applied to identify relevant pathways, with statistical significance set at p < 0.05.

Immune regulatory genes and immune checkpoint analysis

To analyze immune regulatory genes and immune checkpoints, RNA-seq data were obtained from TCGA [42]. The expression of key immune regulatory genes and immune checkpoint molecules, such as PD-1, PD-L1, and CTLA-4, was assessed using differential expression analysis through the DESeq2 package. Correlation analyses were conducted to explore relationships between immune checkpoint expression and other immune regulatory genes. Additionally, survival analysis was performed to evaluate the impact of immune checkpoint gene expression on patient prognosis. Functional enrichment analyses were carried out to identify pathways associated with immune regulation, with statistical significance defined as p < 0.05.

Cell culture and human samples collection

The U-251 and LN229 glioma cell lines were purchased from the American Type Culture Collection (ATCC) and cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10 % fetal bovine serum (FBS) and 1 % penicillin-streptomycin. Cells were maintained in a humidified incubator at 37 °C with 5 % CO₂. The medium was refreshed every 2–3 days, and cells were passaged at 80–90 % confluence using 0.25 % trypsin-EDTA solution. For experimental use, cells were seeded at the appropriate density and allowed to adhere overnight before further treatments or assays. All cell cultures were routinely tested for mycoplasma contamination to ensure experimental reliability.

This study utilized two cohorts for analysis: one comprised glioma tissues and paired normal tissues from 12 patients undergoing surgery, utilized for qPCR analysis; the other included 12 pairs of glioma tissues and matched normal samples for Western blot validation. These tissue samples were pathologically confirmed by the Department of Neurosurgery, First Affiliated Hospital of Anhui Medical University, China, from 2023 to 2024. The study received ethical approval from the Ethics Committee of the Fourth Affiliated Hospital of Anhui Medical University.

Cell transfection

For gene silencing experiments, U-251 and LN229 cells [43] were transfected with short hairpin RNA (shRNA) targeting MANF, EP300, STAT3, and SMAD4, as well as small interfering RNA (siRNA) targeting p38. Transfections were performed using Lipofectamine 3000 (Invitrogen) according to the manufacturer's protocol. For shRNA transfections, cells were incubated with shRNA-containing plasmids and a transfection reagent for 24 h, followed by selection with puromycin. For siRNA transfections, cells were exposed to siRNA and Lipofectamine 3000 for 48 h. Transfection efficiency and gene silencing were validated by quantitative PCR and Western blotting to assess the knockdown of target genes at both mRNA and protein levels. All sequences targeting can be found in Supplementary Table 1.

RT-qPCR and elisa analysis

Reagents for RT-qPCR were purchased from Thermo Fisher Scientific, including the TRIzol reagent for RNA extraction, the High-Capacity cDNA Reverse Transcription Kit for cDNA synthesis, and the PowerUp SYBR Green Master Mix for quantitative PCR. Primers for target genes were designed using Primer-BLAST and synthesized by Sangon Biotech, ensuring optimal specificity and efficiency [44]. Total RNA was extracted from cultured cells using TRIzol, and 1 µg of RNA was reverse transcribed into cDNA. RT-qPCR was performed in triplicate using SYBR Green Master Mix on an ABI 7500 Real-Time PCR System. GAPDH served as an internal control, and relative gene expression was calculated using the 2^{^−ΔΔCt} method. All sequences can be found in Supplementary Table 1, and six experiments were performed.

Reagents for the ELISA assay were purchased from R&D Systems, including the Human MANF and TGF-β ELISA Kits. U-251 and LN229 cell culture supernatants were collected and clarified by centrifugation at 1000 × g for 10 min. ELISA was performed according to the manufacturer’s instructions. Briefly, 100 µL of cell supernatant or standard solutions were added to the pre-coated wells of the ELISA plates and incubated for 2 h at room temperature. Following washes with wash buffer, 100 µL of the respective detection antibodies were added and incubated for 1 hour. After washing, 100 µL of the HRP-conjugated secondary antibody was added and incubated for 1 h [44]. The enzyme-substrate reaction was developed with TMB substrate, stopped with stop solution, and the absorbance was measured at 450 nm using a microplate reader (Bio-Rad). Results were quantified against standard curves, and six experiments were performed.

The experiment of colony formation

Reagents for the colony formation assay were purchased from Sigma-Aldrich, including DMEM, FBS, and penicillin-streptomycin. Cells were seeded in 6-well plates at a low density (500 cells per well) and cultured in DMEM supplemented with 10 % FBS and 1 % penicillin-streptomycin. After 10–14 days, colonies were fixed with 4 % paraformaldehyde and stained with 0.5 % crystal violet. Colonies containing >50 cells were counted using ImageJ software. The experiment was performed in triplicate, and data were analyzed to assess cell proliferation and clonogenic potential.

The experiment of wound healing

U-251 and LN229 cells were seeded in 6-well plates and grown to 90–100 % confluence. A sterile 200 µL pipette tip was used to create a uniform scratch in the cell monolayer. The cells were then washed with phosphate-buffered saline (PBS) to remove debris and cultured in serum-free DMEM to prevent cell proliferation. Images of the scratch were taken at 0 and 72 h using an inverted microscope. The wound area was quantified using ImageJ software to assess cell migration, and experiments were conducted in triplicate.

The experiment of transwell

Reagents for the Transwell migration and invasion assays were purchased from Corning (Transwell chambers with 8.0 µm pores). For the migration assay, 2 × 10^4 U-251 or LN229 cells were suspended in serum-free DMEM and seeded into the upper chamber of the Transwell. For the invasion assay, the upper chamber was pre-coated with Matrigel (Corning). The lower chamber contained DMEM with 10 % FBS as a chemoattractant. After 24 h of incubation at 37 °C with 5 % CO2, cells that migrated or invaded through the membrane were fixed with 4 % paraformaldehyde and stained with 0.1 % crystal violet. Cells were counted under a microscope in five random fields, and the experiment was performed in triplicate.

Apoptotic rate assessed through flow cytometric analysis

Reagents for the apoptosis detection assay using flow cytometry were purchased from BD Biosciences, including the Annexin V-FITC Apoptosis Detection Kit and 7-AAD for dead cell staining. U-251 and LN229 cells were treated according to experimental conditions and harvested by trypsinization. After washing twice with cold PBS, 1 × 10^5 cells were resuspended in 100 µL of binding buffer. Cells were stained with 5 µL of Annexin V-FITC and 5 µL of 7-AAD for 15 min in the dark at room temperature. Stained cells were then analyzed using a BD FACSCanto™ flow cytometer. The proportion of apoptotic cells (early and late apoptosis) was quantified, and data analysis was performed using FlowJo software. The experiment was conducted in triplicate.

CCK-8 assay

Reagents for the Cell Counting Kit-8 (CCK-8) assay were purchased from Thermo Fisher Scientific. U-251 and LN229 cells were seeded in 96-well plates at a density of 2 × 10^3 cells per well in 100 µL of complete medium. After incubation for 24, 48, and 72 h, 10 µL of CCK-8 reagent was added to each well and incubated for 2 h at 37 °C. The absorbance at 450 nm was measured using a microplate reader (Bio-Rad) to assess cell viability. Each condition was tested in five replicates, and the experiment was repeated three times. Data were analyzed to evaluate cell proliferation rate.

Dual-Luciferase® reporter assay

Reagents for the dual-luciferase reporter assay were purchased from Promega, including the Dual-Luciferase® Reporter Assay System. U-251 and LN229 cells were co-transfected with 100 ng of the firefly luciferase reporter plasmid containing the promoter region of interest and 10 ng of the Renilla luciferase control plasmid using Lipofectamine 3000 (Thermo Fisher Scientific). After 48 h of transfection, cells were lysed, and luciferase activity was measured using a luminometer. Firefly luciferase activity was normalized to Renilla luciferase activity to control for transfection efficiency. Each experiment was performed in triplicate to ensure reproducibility.

ChIP-qPCR

Reagents for the chromatin immunoprecipitation (ChIP)-qPCR assay were purchased from Millipore (ChIP Kit) and Abcam (anti-EP300 antibody). U-251 and LN229 cells were cross-linked with 1 % formaldehyde for 10 min, followed by quenching with 125 mM glycine. Cells were then lysed, and chromatin was sonicated to obtain DNA fragments of 200–500 bp. Immunoprecipitation was performed using the EP300, SMAD4 antibody, with normal IgG as a control. After reversal of cross-linking, DNA was purified and analyzed by qPCR to determine the binding of EP300 to the MANF promoter region and SMAD4 to the SOX2, NANOG and C-MYC promoter region. Primers targeting the MANF, SOX2, NANOG and C-MYC mRNA promoter were designed using Primer3 software. Data were normalized to input DNA, and experiments were conducted in triplicate.

Western blotting

Reagents for Western blotting (WB) were purchased from Abcam, including primary antibodies against MANF, EP300, SMAD4, SOX2, NANOG, C-MYC, and β-ACTIN, as well as secondary antibodies conjugated with horseradish peroxidase (HRP) [32]. Samples were lysated and prepared in RIPA buffer supplemented with protease inhibitors. Protein concentrations were quantified using the BCA assay (Thermo Fisher Scientific). Equal amounts of protein (30 µg) were separated by SDS-PAGE and transferred to PVDF membranes. Membranes were blocked with 5 % non-fat milk in TBST for 1 hour, then incubated overnight at 4 °C with primary antibodies. After washing, membranes were incubated with HRP-conjugated secondary antibodies for 1 h. Protein bands were visualized using the ECL detection system (GE Healthcare) and imaged with a chemiluminescence imaging system (Bio-Rad). The experiment was performed in triplicate, and β-ACTIN was used as a loading control.

Animal experiment

Nude mice (Balb/c) were subcutaneously injected with 1 × 10^6 LN229 and U-251 cells mixed with Matrigel (1:1) into the flanks. Tumor growth was monitored weekly, and tumor volumes were measured using calipers with the formula: volume = (length × width²) / 2. Mice were euthanized after 4 weeks, and tumors were excised and measured. The experimental setup included three independent experiments, each with a minimum of six mice per group. The total number of mice used across all experimental conditions was 18 per group (6 per experiment), to ensure reproducibility and statistical power. Tumor growth and size data were analyzed to assess the effects of recombinant human MANF (rhMANF), short hairpin MANF (shMANF), and shSTAT3 on tumorigenesis. Statistical analysis was conducted using one-way ANOVA with Tukey’s post hoc test to compare tumor volume differences between groups. All procedures were approved by the institutional animal care and use committee [18].

The in vivo imaging and pulmonary metastasis

Reagents for the in vivo imaging and pulmonary metastasis assay were purchased from Thermo Fisher Scientific (DMEM, fetal bovine serum [FBS], and luciferase reagent). U-251 cells genetically modified to stably expressing luciferase were used. Nude mice (Balb/c) were tail-vein injected with 1 × 10^6 luciferase-expressing cells in 200 µL of DMEM. After 4 weeks, pulmonary metastases were assessed using an in vivo imaging system (IVIS Spectrum, PerkinElmer) following the administration of d-luciferin substrate. Mice were anesthetized, and bioluminescent imaging was performed to visualize and quantify lung metastases. Tumor burden and metastasis data were analyzed to evaluate the effects of rhMANF, shMANF, and shSTAT3 on lung metastasis. The experiment was conducted with a minimum of six mice per group, and all procedures adhered to institutional animal care guidelines [45].

Statistical analysis

Statistical analysis was performed using SPSS software (version 26.0, IBM). Data were expressed as mean ± standard deviation (SD). For comparisons between two groups, Student's t-test was used, whereas one-way ANOVA followed by post hoc Tukey's test was applied for comparisons among multiple groups. Correlations between variables were assessed using Pearson’s correlation coefficient. Survival analysis was conducted using Kaplan-Meier curves and compared with the log-rank test. p-values < 0.05 were considered statistically significant. All statistical tests were two-sided.

Results

MANF correlates with poor clinical outcomes and gene mutations

The disease-free interval (Supplementary Fig. 1A), overall survival (Supplementary Fig. 1B), disease-specific survival (Supplementary Fig. 1C), and progression-free interval (Supplementary Fig. 1D) were assessed. Kaplan-Meier survival curves for GBMLGG-specific survival (Supplementary Fig. 1E), GBMLGG-overall survival (Supplementary Fig. 1F), GBMLGG-disease-free interval (Supplementary Fig. 1G), and GBMLGG-progression-free interval (Supplementary Fig. 1H) demonstrated that patients with elevated MANF expression had significantly shorter disease-free intervals compared to those with lower expression levels. The hazard ratio (HR) for recurrence was higher in the high MANF expression group, indicating an increased risk of recurrence. Multivariate Cox proportional hazards analysis confirmed that MANF expression is an independent prognostic factor for disease-free interval (Supplementary Fig. 1), reinforcing its association with poor prognosis.

In a comprehensive analysis of gene expression and mutation profiles in glioma samples, significant variations in MANF expression were observed, with notably higher levels in tumor tissues. Mutation analysis revealed recurrent TP53 mutations, with 38.1 % of samples also showing alterations in MANF (Supplementary Fig. 2A). A significant association was found between high MANF expression and IDH1 mutations, suggesting a potential interaction (Supplementary Fig. 2B). Co-occurrence analysis further demonstrated that mutations in additional key genes were frequently present in samples with altered MANF expression, implying possible synergistic roles in tumorigenesis.

Association of MANF with enhanced tumor stemness and methylation modifications

To investigate the potential involvement of MANF in glioma stemness maintenance, we analyzed the correlation between MANF expression and stemness indices derived from both transcriptomic (mRNAsi) and epigenetic (DNAsi) features across TCGA-GBMLGG, TCGA-LGG, and TCGA-GBM cohorts. As shown in Fig. 1A–C, MANF expression was positively associated with mRNAsi across all datasets, indicating that high MANF levels are linked to increased transcriptional stemness features. Similarly, MANF also showed a consistent positive correlation with DNAsi in each cohort (Fig. 1D–F), suggesting a role in epigenetic plasticity associated with tumor dedifferentiation.

Fig. 1.

Correlation between MANF expression and tumor stemness indices and RNA modification genes in glioma. (A–C) Correlation between MANF expression and mRNA-based stemness index (mRNAsi) in the TCGA-GBMLGG (A), TCGA-LGG (B), and TCGA-GBM (C) cohorts. (D–F) Correlation between MANF expression and DNA methylation-based stemness index (DNAsi) in the TCGA-GBMLGG (D), TCGA-LGG (E), and TCGA-GBM (F) cohorts. (G) Heatmap showing the correlation between MANF expression and RNA modification genes, including m1A, m5C, and m6A regulators, across three glioma subtypes. Rows represent individual RNA modification genes, and columns represent cohorts. The intensity and hue of red indicate the strength of positive correlation; blue indicates negative correlation. Dot size reflects p-value significance, and color sidebars annotate gene function as "writer", "reader", or "eraser".

Given the known interaction between RNA modifications and stemness regulation, we further assessed the correlation between MANF and key RNA modification-related genes, including m1A, m5C, and m6A regulators. As illustrated in the heatmap (Fig. 1G), MANF expression was positively correlated with a wide range of RNA-modifying enzymes, including m6A “writers” such as METTL3 and METTL14, “readers” such as YTHDF1/2, and m5C regulators such as NSUN2 and DNMT3A. Notably, these correlations were particularly strong in the LGG subgroup, indicating that MANF may influence RNA modification-mediated post-transcriptional regulation in glioma stem-like cells.

These findings suggest that MANF contributes to glioma stemness by promoting transcriptional and epigenetic programs characteristic of stem-like phenotypes, potentially via modulation of RNA modification networks.

MANF expression is associated with increased genomic instability and intratumoral heterogeneity in glioma

To further explore the role of MANF in glioma genomic heterogeneity, we conducted a comprehensive correlation analysis between MANF expression and several genomic instability-related features across multiple TCGA cohorts, including TCGA-GBM, TCGA-LGG, and TCGA-GBMLGG. As shown in Fig. 2A, MANF expression exhibited a modest positive correlation with tumor mutational burden (TMB), particularly in the GBMLGG cohort, suggesting a potential association between MANF expression and the accumulation of somatic mutations.

Fig. 2.

Correlation between MANF expression and genomic heterogeneity features in glioma. (A) Correlation with tumor mutational burden (TMB). (B) Correlation with copy number variation (CNV) burden. (C) Correlation with the number of tumor subclones. (D) Correlation with predicted neoantigen load. (E) Correlation with intratumoral heterogeneity (ITH). (F) Correlation with homologous recombination deficiency (HRD) score. (G) Correlation with aneuploidy score. (H) Correlation with stemness index (mRNAsi). Each dot represents a TCGA glioma subtype (GBM, LGG, GBMLGG), and dot size indicates sample size. Color gradient represents p-values, and the x-axis shows the correlation coefficient (Spearman's R).

A similar trend was observed between MANF and chromosomal instability, as measured by global copy number variation (CNV) burden (Fig. 2B). In addition, MANF expression was positively associated with the number of tumor subclones (Fig. 2C), implicating a role in promoting clonal diversification and intratumoral heterogeneity.

Further analysis revealed weak but notable correlations between MANF expression and other genomic heterogeneity indicators, including neoantigen load (Fig. 2D), intratumoral heterogeneity score (ITH) (Fig. 2E), and homologous recombination deficiency (HRD) score (Fig. 2F), the latter of which is suggestive of impaired DNA repair capacity in MANF-high gliomas.

Interestingly, MANF expression also showed a moderate positive correlation with aneuploidy score (Fig. 2G), indicating its potential involvement in large-scale chromosomal alterations. Finally, consistent with previous findings, MANF expression remained significantly associated with mRNA-based stemness index (mRNAsi), further supporting its role in maintaining stem-like transcriptional programs (Fig. 2H).

MANF immunological analysis

The analysis of immune cell infiltration within the tumor microenvironment revealed a distinct pattern in tumors with high MANF expression. Specifically, tumor-infiltrating scores were significantly positively correlated with MANF levels in gliomas compared to the LUAD group (p < 0.01). This immune profile is associated with poor patient outcomes and suggests that MANF may play a role in modulating immune evasion mechanisms (Fig. 3A–C). Gene expression profiling of immune regulatory markers indicated that MANF positively correlates with the expression of multiple immunosuppressive genes, which are involved in dampening effector immune responses and promoting tumor immune escape (Fig. 3D). Additionally, analysis of immune checkpoint molecules revealed a significant upregulation of PDCD1 and CTLA-4 in MANF-high tumors, demonstrating a strong association with checkpoint-mediated immune suppression. Moreover, a correlation analysis between MANF and VEGFA/B expression showed a robust positive relationship, indicating that MANF may enhance immune evasion through the upregulation of immune checkpoints (Fig. 3E).

Fig. 3.

MANF and tumor immunology. (A-C) High MANF expression is associated with increased tumor-infiltrating immune cells and poorer patient outcomes compared to LUAD. (D) MANF positively correlates with the expression of immunosuppressive genes involved in immune evasion. (E) Increased expression of immune checkpoint molecules PDCD1 and CTLA-4 in MANF-high tumors indicates enhanced immune suppression. (F-L) MANF-high tumors show reduced cytotoxic T cell infiltration and increased levels of Tregs and MDSCs, with a strong negative correlation between MANF expression and CD8+ T cells and a positive correlation with Tregs.

A comprehensive immune cell analysis showed that MANF-high tumors exhibited reduced infiltration of cytotoxic T cells, alongside increased levels of regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs), both of which are known to promote immune tolerance. Correlation analysis revealed a strong negative association between MANF expression and CD8+ T cell levels, while a positive correlation was observed with Treg levels. These findings suggest that MANF contributes to the suppression of anti-tumor immune responses (Fig. 3F–L).

Endogenous and exogenous MANF mediate malignant behavior in glioma cells

To investigate the functional role of MANF in glioma malignancy, we first generated stable MANF knockdown cell lines (sh1# and sh2#) in LN229 and U251 glioma cells. Western blot and qRT-PCR analysis confirmed successful silencing of MANF expression at both protein and transcript levels (Fig. 4A–C).

Fig. 4.

Effects of endogenous and exogenous MANF on glioma cell stemness, proliferation, migration, and apoptosis. (A) Western blot validation of MANF knockdown (sh1#, sh2#) in LN229 and U251 cells. (B–C) qRT-PCR confirming decreased MANF mRNA levels after knockdown. (D–E) Expression of stemness genes (SOX2, NANOG, C-MYC) after treatment with high-dose rhMANF (2500 ng/mL); no significant changes observed. (F, I) Colony formation assay showing no significant difference in colony numbers with 2500 ng/mL rhMANF. (G–H) Dose-dependent increase in stemness gene expression following treatment with 0, 50, 250, and 500 ng/mL rhMANF. (J–K) Colony formation significantly reduced in MANF-knockdown cells compared to shCtrl. (L–M) Wound healing and Transwell assays demonstrate reduced migratory capacity in MANF-deficient LN229 and U251 cells. (N) Flow cytometric analysis reveals increased apoptosis following MANF knockdown. (O) CCK-8 assays show reduced cell viability over 5 days in both LN229 and U251 cells upon MANF knockdown. ns, not significant; *p < 0.05; **p < 0.01.

We then evaluated the effects of recombinant human MANF (rhMANF) on the expression of glioma stemness genes. In high-dose treatment (2500 ng/mL), rhMANF showed no significant effect on SOX2, NANOG, or C-MYC expression in either LN229 or U251 cells (Fig. 4D–E), nor did it promote colony formation (Fig. 4F, I). However, when rhMANF was administered at lower doses (50–500 ng/mL), stemness gene expression was upregulated in a dose-dependent manner, peaking at 250 ng/mL (Fig. 4G–H), suggesting a biphasic regulatory effect of exogenous MANF.

Endogenous MANF knockdown significantly reduced colony formation in both cell lines (Fig. 4J–K), indicating its requirement for glioma cell proliferative capacity. In wound healing assays, MANF-deficient cells showed impaired wound closure after 72 h (Fig. 4L–M), consistent with reduced migratory potential. Similarly, Transwell assays revealed that silencing MANF drastically inhibited cell migration.

To assess the effect of MANF on cell survival, flow cytometry showed increased apoptotic cell populations upon MANF knockdown (Fig. 4N). CCK-8 assays further confirmed reduced cell viability over a 5-day time course in MANF-deficient cells compared to controls (Fig. 4O).

Collectively, these data suggest that endogenous MANF promotes glioma cell stemness, proliferation, and migration, and protects against apoptosis. Exogenous MANF, in contrast, exerts a dose-dependent, limited regulatory effect, most prominent at intermediate concentrations.

Regulation of glioma stemness genes by MANF through the TGF-β/SMAD4 pathway

To elucidate the specific mechanism by which MANF affects stemness in glioma cells, we first screened transcription factors associated with stemness-related genes using the TFDB database. This was followed by validation through a dual-luciferase reporter assay, which demonstrated that SMAD4 could bind to the promoter regions of stemness genes SOX2, NANOG, and C-MYC, thereby regulating their transcriptional activity and contributing to the maintenance of cellular stemness (Fig. 5A–F).

Fig. 5.

MANF regulates stemness genes via the TGF-β/SMAD4 signaling pathway. (A-F) ChIP and luciferase reporter assays demonstrate that SMAD4 binds to the promoter regions of stemness-related genes SOX2, NANOG, and C-MYC, regulating their transcriptional activity. (G, H) ChIP assays show that MANF treatment increases SMAD4 binding to these promoters, while inhibition of TGF-β signaling reduces this binding, indicating the involvement of the TGF-β/SMAD4 pathway in MANF-mediated stemness regulation. (I) Knockdown of SMAD4 diminishes the MANF-induced upregulation of stemness genes, confirming SMAD4 as a critical mediator in this regulatory pathway.

Since TGF-β is a well-known activator of SMAD signaling, we conducted a ChIP assay to further investigate the interaction. The results showed that SMAD4 directly binds to predicted binding sites within the promoter regions of SOX2, NANOG, and C-MYC. Upon MANF treatment, the binding of SMAD4 to these regions was upregulated, whereas blocking TGF-β signaling led to a downregulation of this binding in MANF-treated cells (Fig. 5G, H). To further confirm the critical role of SMAD4 in MANF-mediated regulation of glioma stemness genes, we performed SMAD4 knockdown and observed a significant attenuation of MANF-induced upregulation of SOX2, NANOG, and C-MYC expression (Fig. 5I). These results indicate that SMAD4 is a key mediator of MANF-driven transcriptional regulation, and the loss of SMAD4 impairs MANF’s ability to activate stemness-related genes.

Reciprocal regulation of MANF and TGF-β secretion via the STAT3 pathway

To further investigate the regulatory interaction between MANF and TGF-β, we employed siRNA targeting STAT3 and treated the culture system with recombinant human MANF and TGF-β proteins. ELISA was used to measure protein concentrations in the supernatant, revealing that MANF and TGF-β can regulate each other's secretion. However, this mutual regulation was abolished upon STAT3 silencing, indicating that the interplay between MANF and TGF-β is likely mediated through a STAT3-dependent mechanism (Fig. 6A, B).

Fig. 6.

MANF and TGF-β mutual regulation via STAT3 and p38 pathways. (A, B) STAT3 knockdown via siRNA abolishes the mutual regulation of MANF and TGF-β secretion, indicating that their interaction depends on STAT3 signaling. (C) MANF regulates the stemness genes SOX2, NANOG, and C-MYC through SMAD4; silencing SMAD4 impairs MANF’s regulatory effects on these genes. (D) Silencing of p38 MAPK inhibits MANF-induced regulation of stemness genes, suggesting that MANF modulates SMAD4 partially through the p38 pathway.

MANF regulates stemness gene expression via the p38 signaling pathway

Western blot analysis demonstrated that MANF regulates the expression of stemness genes SOX2, NANOG, and MYC through modulation of SMAD4. Upon addition of recombinant human MANF protein to the culture system, both SMAD4 and the stemness genes SOX2, NANOG, and MYC were upregulated. However, silencing SMAD4 abolished the regulatory effects of MANF on these stemness genes (Fig. 6C). To further elucidate the mechanism by which MANF regulates SMAD4 and the stemness genes, we treated cells with siRNA targeting p38. Western blot results showed that silencing p38 negated the effects of MANF on stemness regulation, indicating that MANF likely modulates SMAD4 expression indirectly through the regulation of p38 (Fig. 6D).

EP300 as a transcription factor regulating MANF

To investigate the regulatory mechanisms controlling MANF expression in glioma cells, we conducted a dual-luciferase reporter assay. Initially, we screened transcription factors for MANF from TFDB and constructed a dual-luciferase reporter system to validate the interactions. The results demonstrated that EP300 specifically binds to the promoter region of MANF, regulating its transcription (Fig. 7A–C, G). ChIP assays further confirmed that EP300 directly binds to the predicted binding sites on the MANF promoter (Fig. 7D, E). Additionally, Western blot analysis revealed a significant reduction in MANF protein expression following EP300 knockdown (Fig. 7F).

Fig. 7.

EP300 as a transcriptional regulator of MANF in glioma cells. (A-C) Dual-luciferase reporter assays reveal that EP300 binds specifically to the promoter region of MANF and regulates its transcriptional activity. (D, E) ChIP assays confirm the direct binding of EP300 to the MANF promoter in glioma cells. (F) Western blot analysis shows a decrease in MANF protein levels after EP300 knockdown, indicating EP300 positively regulates MANF expression. (G) Knockdown of EP300 results in suppressed glioma cell proliferation as assessed by cell viability assays.

Assessment of MANF regulation and mechanisms in model animals

To investigate the in vivo effects and mechanisms of MANF, we conducted subcutaneous tumor formation experiments using LN229 and U251 cell lines treated with PBS, rhMANF, MANF knockdown, and STAT3 knockdown. The results revealed that MANF knockdown led to a significant reduction in tumor volume, while rhMANF treatment counteracted this effect. Additionally, STAT3 knockdown mitigated the adverse effects of rhMANF treatment. Compared to the rhMANF + shMANF group, tumors in the rhMANF + shSTAT3 group were significantly smaller (Fig. 8A, C). This suggests that solely knocking down MANF without blocking external MANF stimulation may not substantially reduce tumor burden in vivo.

Fig. 8.

In vivo functional analysis of MANF in glioma progression and metastasis. (A, C) Tumor growth curves and volume measurements following MANF knockdown or recombinant MANF (rhMANF) treatment in glioma xenograft models. STAT3 knockdown attenuates the rhMANF-induced tumor growth promotion. (B, D) Quantification of lung metastatic nodules in mice injected with glioma cells under different treatments shows modulation of metastatic potential by MANF and STAT3. (E) Rate of tumor formation escape in groups treated with combinations of shMANF, shSTAT3, and siP38. (F) Representative in vivo bioluminescence imaging illustrates the regulatory effects of endogenous and exogenous MANF on tumor metastasis. A total of 6 mice per group (n = 6) were used for each in vivo experiment. Each experiment was repeated three times independently. Error bars represent mean ± SD.

To assess metastatic potential, we established a lung metastasis model and performed in vivo imaging. The results demonstrated that rhMANF treatment significantly enhanced lung metastasis, whereas MANF knockdown markedly reduced the metastatic capacity of glioma cells. However, even with MANF knockdown, rhMANF treatment restored lung metastatic potential. STAT3 knockdown slightly reduced the metastasis (Fig. 8B, D).

Further in vivo studies involved subcutaneous tumor formation with cells treated with rhMANF, shMANF, shSTAT3, and siP38. The results indicated that shMANF alone did not significantly increase tumor formation escape rate compared to the control. In contrast, combined shMANF, shSTAT3, and siP38 treatments significantly enhanced the tumor formation escape rate in nude mice (Fig. 8E). This suggests that MANF may involve both intracellular regulation and external MANF stimulation in modulating cell stemness and tumor malignancy. In vivo imaging further confirmed that tumor metastasis characteristics are influenced by both intracellular MANF regulation and external MANF stimulation (Fig. 8F).

Validation of MANF expression characteristics in clinical samples

We performed Western blot analysis to assess MANF protein expression in tumor tissues and adjacent non-tumor specimens from 12 glioma patients. The results revealed a significant increase in MANF protein levels in the tumor tissues compared to the adjacent non-tumor tissues (Fig. 9A–F). Complementary qRT-PCR analysis confirmed that MANF mRNA levels were significantly elevated in tumor tissues relative to adjacent non-tumor samples in all 12 cases. These findings suggest that MANF may play a critical role in the development and progression of gliomas (Fig. 9G–I).

Fig. 9.

MANF expression profile in clinical glioma specimens. (A-F) Western blot analysis comparing MANF protein levels in glioma tumor tissues versus adjacent non-tumor tissues from patients. (G-I) qRT-PCR evaluation of MANF mRNA expression in paired glioma tumor and adjacent non-tumor tissue samples, demonstrating significant upregulation in tumors.

Discussion

The treatment of gliomas, particularly high-grade gliomas, remains a significant challenge in neuro-oncology due to their aggressive nature and resistance to conventional therapies [46]. Despite advances in precision medicine and targeted therapies for other cancers, the progress in glioma treatment has been notably slow [47]. This is largely attributed to the heterogeneity of gliomas, the complexity of their molecular landscape, and the limited understanding of key therapeutic targets [48]. MANF has recently emerged as a promising target in glioma therapy [49]. Preliminary studies suggest that MANF plays a role in tumor progression and survival, but the precise mechanisms by which it influences glioma biology remain unclear [18]. Further elucidating the function and regulatory pathways of MANF in gliomas could pave the way for the development of novel targeted therapies, offering hope for improved clinical outcomes in patients with this devastating disease.

Our study provides compelling evidence that MANF plays a significant role in glioma progression and patient outcomes. The results demonstrate that elevated MANF expression is associated with poor clinical outcomes, including shorter disease-free intervals and overall survival. This finding suggests that MANF serves as a critical prognostic marker for glioma. Kaplan-Meier survival curves and multivariate Cox proportional hazards analysis further reinforce MANF's role as an independent prognostic factor, indicating that higher MANF levels correlate with increased risk of recurrence and poor prognosis.

In line with our findings, previous studies have also highlighted the importance of MANF in cancer biology [29,50], although its specific role in gliomas has not been extensively characterized. Our results reveal that high MANF expression is associated with key gene mutations, including recurrent TP53 mutations and IDH1 mutations. These associations suggest that MANF may interact with these genetic alterations, potentially contributing to glioma pathogenesis. The observed co-occurrence of mutations in other key genes with altered MANF expression further supports the notion of MANF's involvement in synergistic mechanisms driving tumorigenesis.

Our study also identifies a significant association between MANF expression and tumor stemness. Elevated MANF levels are correlated with increased stemness indices, indicating that MANF promotes a stem-like phenotype in glioma cells. This aligns with recent literature suggesting that MANF plays a role in maintaining cancer stem cells and promoting tumor progression. The positive correlation between MANF and RNA modification-related genes, such as m1A, m5C, and m6A regulators, suggests that MANF may influence tumor progression through RNA modification processes. This finding adds a new dimension to our understanding of MANF's role in glioma biology.

The analysis of genomic heterogeneity reveals that tumors with high MANF expression exhibit increased mutational burden and chromosomal instability. This supports the hypothesis that MANF contributes to genomic instability and intratumoral heterogeneity, potentially facilitating tumor evolution and resistance to therapy. The positive correlation between MANF expression and the number of tumor subclones highlights MANF's role in sustaining a heterogeneous tumor microenvironment.

Immunological analysis further underscores the multifaceted role of MANF in gliomas. We observed that high MANF expression is associated with increased immune cell infiltration, particularly immunosuppressive cells such as Tregs and MDSCs. This suggests that MANF may modulate immune evasion mechanisms, potentially contributing to tumor immune escape. The upregulation of immune checkpoint molecules like PDCD1 and CTLA-4 in MANF-high tumors further supports this hypothesis, indicating that MANF may enhance immune evasion through checkpoint-mediated immune suppression.

Endogenous MANF promotes the malignant phenotype of glioma cells, including the enhanced expression of stemness genes and increased cell migration and invasion abilities. In contrast, exogenous MANF does not significantly affect the expression of stemness genes, but it plays a role in promoting tumor cell proliferation. We hypothesize that the lack of effect from exogenous MANF may be due to the inability of the recombinant protein to effectively engage with the cellular machinery or signaling networks in a manner comparable to the endogenous form. Endogenous MANF may interact more effectively with intracellular pathways and molecular partners that are critical for regulating stemness genes. In this context, endogenous MANF might activate key downstream signaling pathways through effective receptor binding within the cell, thereby regulating the expression of tumor stemness genes. In contrast, exogenous MANF may fail to trigger the same downstream signaling due to improper cellular integration or receptor engagement. This may be because the exogenous protein is not sufficiently processed or internalized by the cells, preventing it from effectively activating the signaling cascades involved in regulating stemness gene expression. As a result, exogenous MANF may not induce the same cellular signals as endogenous MANF, leading to its limited effect on stemness gene expression.

Further investigation into the regulatory mechanisms of MANF reveals that it influences the expression of stemness genes through the TGF-β/SMAD4 pathway. Dual-luciferase reporter assays and ChIP analysis confirmed that SMAD4 binds to the promoter regions of stemness genes and is upregulated under the regulation of MANF. Additionally, the regulation of stemness genes by MANF is also modulated by the p38 signaling pathway, indicating a complex signaling network governing the malignant behavior of glioma cells. More importantly, the reciprocal regulation between MANF and TGF-β via the STAT3 pathway further highlights the intricate roles of these factors in glioma biology, offering new therapeutic insights for targeting this signaling axis.

In vivo studies corroborate our in vitro findings, showing that MANF knockdown reduces tumor volume and metastatic potential, while exogenous MANF enhances these properties. The results indicate that MANF's effects on glioma malignancy involve both intracellular regulation and external stimulation. Validation of MANF expression in clinical samples confirms its significant upregulation in tumor tissues compared to adjacent non-tumor tissues, underscoring its potential as a biomarker and therapeutic target.

While our study provides valuable insights, it is not without limitations. First, the use of cell lines and animal models may not fully replicate the complexity of human gliomas. Second, the mechanisms underlying MANF's role in tumorigenesis warrant further investigation. Additionally, although we utilized TCGA data for analysis, the dataset's clinical heterogeneity and the underrepresentation of certain tumor subtypes may affect the generalizability of our results. Future research should explore the therapeutic potential of targeting MANF and its associated pathways, as well as investigate the broader implications of MANF in other cancer types.

Conclusions

Overall, this research underscores MANF's multifaceted role in glioma and its promise as a biomarker for prognosis and a target for novel therapeutic strategies. Future studies should focus on validating these findings in larger cohorts and exploring MANF-targeted therapies to improve glioma treatment outcomes.

Ethical statement

The study was approved by the Ethics Committee of The Second Hospital of Hebei Medical University (Approval number: 2023-R-216; Title of the approved project: Investigation of MANF Regulation of Glioma Stemness; Date of approval: Sep 12, 2023). Written informed consent was obtained from a legally authorized representative(s) for anonymized patient information to be published in this article.

Data availability

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

Contributor information

Yang Hong, Email: hongyangcc@163.com.

Huabao Cai, Email: ahmudrtsai@163.com.

Xin Liu, Email: LiuxinHB2H@163.com.

CRediT authorship contribution statement

Shi Feng: Writing – original draft, Formal analysis. Ming Yang: Writing – original draft, Visualization, Software. Pengfei Dong: Writing – original draft, Validation, Methodology, Investigation. Fangfang Ding: Writing – review & editing, Validation, Project administration, Data curation. Yang Hong: Methodology, Project administration, Validation, Writing – review & editing. Huabao Cai: Project administration, Supervision, Validation, Writing – review & editing. Xin Liu: Writing – review & editing, Resources, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.