Abstract

Background

   Chronic rhinosinusitis with nasal polyps (CRSwNP) is a prevalent
   inflammatory disorder affecting the upper respiratory tract. Recent
   studies have indicated an association between CRSwNP and mitochondrial
   metabolic disorder characterized by impaired metabolic pathways;
   however, the precise mechanisms remain unclear. This study aims to
   investigate the mitochondrial-related signature in individuals
   diagnosed with CRSwNP.

Methods

   Through the integration of differentially expressed genes (DEGs) with
   the mitochondrial gene set, differentially expressed
   mitochondrial-related genes (DEMRGs) were identified. Subsequently, the
   hub DEMRGs were selected using 4 integrated machine learning
   algorithms. Immune and mitochondrial characteristics were estimated
   based on CIBERSORT and ssGSEA algorithms. Bioinformatic findings were
   confirmed through RT-qPCR, immunohistochemistry, and ELISA for nasal
   tissues, as well as Western blotting analysis for human nasal
   epithelial cells (hNECs). The relationship between hub DEMRGs and
   disease severity was assessed using Spearman correlation analysis.

Results

   A total of 24 DEMRGs were screened, most of which exhibited lower
   expression levels in CRSwNP samples. Five hub DEMRGs (ALDH1L1, BCKDHB,
   CBR3, HMGCS2, and OXR1) were consistently downregulated in both the
   discovery and validation cohorts. The hub genes showed a high
   diagnostic performance and were positively correlated with the
   infiltration of M2 macrophages and resting mast cells. Experimental
   results confirmed that the 5 genes were downregulated at both the mRNA
   and protein levels within nasal polyp tissues. Finally, a significant
   and inverse relationship was identified between the expression levels
   of these genes and both the Lund-Mackay and Lund-Kennedy scores.

Conclusion

   Our findings systematically unraveled 5 hub markers correlated with
   mitochondrial metabolism and immune cell infiltration in CRSwNP,
   suggesting their potential to be based to design diagnostic and
   therapeutic strategies for the disease.

   Keywords: Computational biology, Machine learning, Mitochondria, Nasal
   polyps, Rhinosinusitis

Introduction

   Chronic rhinosinusitis with nasal polyps (CRSwNP) is a prevalent
   inflammatory disorder in the upper respiratory tract.[40]^1 Compared to
   chronic rhinosinusitis without nasal polyps (CRSsNP), CRSwNP is
   typically manifested with more severe clinical symptoms, concurrent
   with a higher likelihood of asthma, and accompanied with more
   recurrences after endoscopic sinus surgery.[41]^2 Numerous studies have
   aimed to elucidate the mechanisms of CRSwNP, yet its precise etiology
   and pathophysiology remain unclear. It is imperative to identify
   reliable biomarkers forecasting the onset and progression of the
   disease.

   Mitochondria assist in energy production through oxidative
   phosphorylation (OXPHOS), resulting in the synthesis of adenosine
   triphosphate (ATP).[42]^3 Recent evidence indicates that mitochondria
   intricately regulate various cellular processes, including mitophagy,
   cellular growth, and the generation of reactive oxygen species (ROS)
   triggered by inflammation and stress signaling.[43]^4 As a chronic
   inflammatory disease in the upper airway, CRSwNP is aggravated by
   mitochondrial dysfunction and ROS overproduction.[44]^5 Epithelial
   cells originating from nasal polyps exhibit mitochondrial dysfunction
   in response to environmental contaminants, allergens, and pathogens,
   leading to the release of a range of proinflammatory cytokines.[45]^6
   Compared to normal epithelium, the markers of mitochondrial fusion and
   fission, as well as mitophagy, are elevated in the epithelium of nasal
   polyps.[46]^7 These findings suggest that mitochondria play a crucial
   role in the pathogenesis of CRSwNP and present a potential therapeutic
   target for intervention.

   CRSwNP can be categorized into 3 inflammatory endotypes based on the
   patterns of immune cell infiltration: type 1, type 2, and type 3.[47]^8
   Type 2 CRSwNP is distinguished by significant eosinophilic
   infiltration, correlating with increased allergic symptoms and a poorer
   prognosis. Mitochondrial metabolism disorder contributes to the
   progression of allergic diseases through altering the immune
   microenvironment.[48]^9 In experimental models of allergic rhinitis,
   mice with pre-existing deficiencies in mitochondrial electron transport
   chain exhibit heightened IgE responses, significant airway remodeling
   and hyperresponsiveness.[49]^10 In asthma, type 2 cytokine IL-4
   modulates asymmetric dimethylarginine (ADMA) metabolism, leading to a
   reduction in mitochondrial quantity and an increase in mitochondrial
   ROS, thus exacerbating damage to the airway epithelium.[50]^11 Though a
   growing body of evidence suggests a potential link between immunity and
   mitochondrial function in CRSwNP, the modulatory effects of
   mitochondrial-related genes on type 2 inflammation remain to be
   explored.

   Bioinformatics enables the identification of molecules that exhibit
   differential expression between diseased and healthy individuals.
   Employing this approach, the current study examined the influence of
   mitochondrial-related genes on the progression of CRSwNP and their
   correlation with immune infiltration. In addition, this study explored
   the correlation between the expression of these hub genes and the
   severity of CRSwNP, facilitating the development of personalized
   treatment strategies.

Materials and methods

Gene expression profiles

   Datasets [51]GSE136825,[52]^12 [53]GSE36830,[54]^13 and
   [55]GSE23552[56]^14 were procured from the Gene Expression Omnibus
   (GEO) database. A total of 42 nasal polyp tissues and 28 control
   samples were selected from [57]GSE136825, and 6 nasal polyp tissues
   along with 6 control samples were chosen from [58]GSE36830. Both
   datasets were combined to create the discovery set. Furthermore, 11
   nasal polyp tissues and 13 control samples from [59]GSE23552 were used
   for validation. For analysis at single-cell levels, we obtained
   scRNA-seq dataset SCP253[60]^15 from the Single Cell Portal, including
   6 nasal polyp samples and 6 control samples. [61]Supplementary Table 1
   provides the information regarding the specifics of each dataset.

Screening of differentially expressed genes (DEGs)

   The R package “sva” was utilized to combine the datasets [62]GSE136825
   and [63]GSE36830, with batch information excluded. Following this, the
   R package “limma” was employed to screen the differentially expressed
   genes (DEGs) between nasal polyps and normal nasal mucosa. In order to
   visually depict the differential gene expression, a volcano plot was
   generated utilizing the R package “ggplot2”. DEGs were determined based
   on the cutoffs of an adjusted P value < 0.05 and |log2FC| > 1.

Identification of differentially expressed mitochondrial-related genes
(DEMRGs)

   By screening the MitoCarta3.0 database,[64]^15 1136 genes localized in
   mitochondria were identified. To identify expressed
   mitochondrial-related genes (DEMRGs), the DEGs were intersected with
   the mitochondrial-related genes using an online tool
   ([65]https://jvenn.toulouse.inra.fr/). By utilizing the R package
   “ComplexHeatmap”, a heatmap was generated to visualize the resulting
   DEMRGs.

Functional enrichment analysis

   GO annotation analysis and KEGG pathway enrichment analysis of DEMRGs
   were conducted by utilizing R package “clusterProfiler”. In addition,
   mitochondrial pathways (MitoPathways) were obtained from MitoCarta3.0
   and analyzed using SangerBox ([66]http://sangerbox.com/). The R package
   “goplot” was employed to visualize all the obtained results.

Extraction of hub DEMRGs

   To identify diagnostic gene candidates, a series of analyses were
   performed on the DEMRGs using different machine learning algorithms.
   Least absolute shrinkage and selection operator (LASSO) regression was
   executed using the R package “glmnt” and minimal lambda value was
   selected as the optimal value.[67]^16 Support vector machine recursive
   feature elimination (SVM-RFE) was carried out via R “kernlab” package,
   and the combination that resulted in the lowest errors was defined as
   optimal genes.[68]^17 Gradient boosting machine (GBM) analysis was
   carried out utilizing the R package “gbm” and the top 10 genes were
   selected as optimal ones.[69]^18 Random forest (RF) model was created
   using the R package “randomForest”.[70]^19 The hub DEMRGs were
   determined by overlapping the diagnostic gene candidates from the 4
   machine learning algorithms, and then visualized in a Venn diagram.

Establishment of a diagnostic model

   To assess the performance and utility of the hub DEMRGs, the package
   “pROC” was utilized to generate receiver operating characteristic (ROC)
   curves. The predictive capability of algorithms was quantified by
   calculating the area under the curve (AUC). Furthermore, a nomogram was
   constructed to predict the risk of CRSwNP by incorporating the hub
   DEMRGs. The effectiveness of the nomogram was evaluated by estimating
   its predictive accuracy using calibration curves.

Creation of network between hub genes, miRNAs and transcription factors (TFs)

   We employed the NetworkAnalyst webserver
   ([71]https://www.networkanalyst.ca/) to explore the regulatory network
   that govern hub genes, miRNAs, and TFs. Following this, we utilized
   Cytoscape software to visualize the network.

Immune infiltration analysis

   To visualize the distribution of immunocytes in the CRSwNP and control
   groups, we utilized the CIBERSORT algorithm, a deconvolution algorithm
   designed for the analysis of tissue-infiltrating immune cell
   composition using signature gene matrix.[72]^20 Additionally, we
   conducted Spearman correlation analysis to explore the relationships
   among various immunocyte pairs or between the hub DEMRGs and
   immunocytes.

Evaluation of MitoPathways

   We obtained the pathways associated with the mitochondria from
   MitoCarta3.0. To explore the relationships between the mitochondrial
   respiratory chain and the hub DEMRGs, we conducted Spearman correlation
   analysis. Furthermore, we employed the single-sample gene set
   enrichment analysis (ssGSEA) algorithm to evaluate the activity of 149
   MitoPathways in each specimen.[73]^21 Among the results, we selected
   the top 10 MitoPathways with the highest variance for further
   investigation.

Validation in external dataset

   To examine the expression of hub DEMRGs and MitoPathways, we conducted
   an analysis on the dataset [74]GSE23552. Furthermore, we determined the
   expression levels of these hub DEMRGs in distinct cell types by
   analyzing the scRNA-seq dataset SCP253 using the “Seurat” package
   V4.3.2. Rigorous quality control protocols were implemented to exclude
   cells that did not meet the predefined criteria. Specifically, cells
   with a gene count between 200 and 6000 and a mitochondrial gene content
   below 5% were retained. Subsequently, data normalization was performed
   using the “NormalizeData” function. The 2000 genes with the highest
   variability were identified using the “FindVariableFeatures” function.
   These genes were further analyzed using principal component analysis
   (PCA), cluster analysis, and Uniform Manifold Approximation and
   Projection (UMAP) for dimensionality reduction. Differential expression
   analysis was conducted using the “FindAllMarkers” function, with a
   threshold of |log2FC| > 0.25 and an adjusted P value < 0.05. The
   clusters were annotated based on the previous study.[75]^22 The average
   expression levels of hub DEMRGs in epithelial cell genes were
   calculated across samples and visualized in the boxplot and heatmap.

Clinical data collection

   This study received approval from the Ethics Committee of the First
   Affiliated Hospital of Nanjing Medical University (2020-SR-500). A
   total of 25 participants were included, comprising 15 CRSwNP patients
   and 10 control subjects. Nasal polyp specimens were procured from
   CRSwNP patients, whereas normal tissues from individuals who underwent
   septoplasty without exhibiting any symptoms of rhinosinusitis. The
   diagnosis of CRSwNP was based on the guidelines outlined in the
   European Position Paper on Rhinosinusitis and Nasal Polyps 2020.[76]^23
   Preoperative evaluation included computed tomography (CT) and
   endoscopic endonasal evaluation, while the Lund-Mackay scores and
   Lund-Kennedy scores were recorded by 2 rhinologists independently.
   Details regarding the participants are shown in [77]Supplementary
   Table 2.

Real-time quantitative PCR

   The tissue samples were subjected to total RNA isolation using FreeZol
   Reagent (Vazyme, Nanjing, China). RNA was subjected to reverse
   transcription on the HiScript III All-in-one RT SuperMix Perfect
   (Vazyme, Nanjing, China) to generate cDNA. qPCR analysis was conducted
   on a StepOnePlus System utilizing SYBR qPCR Master Mix (Vazyme,
   Nanjing, China). The relative gene expression, normalized to B2M
   expression, was determined employing the 2^−ΔΔCt methodology. The
   specific primer sequences employed in this study are shown in
   [78]Supplementary Table 3. The experiment was replicated 3 times to
   ensure reproducibility.

Immunohistochemistry staining

   Immunohistochemical staining was performed by deparaffinizing and
   rehydrating tissue sections (5 μm thick) using a gradient concentration
   of ethanol (100-75%). Antigen retrieval was accomplished by 10 mM
   sodium citrate buffer solution. After obstructing nonspecific antigens
   with 5% BSA, the slides were subjected to incubation with the primary
   antibodies including OXR1 (1:200, 13514-1-AP, Proteintech, Wuhan,
   China), BCKDHB (1:100, 13685-1-AP, Proteintech, Wuhan, China), CBR3
   (1:100, 15619-1-AP, Proteintech, Wuhan, China), ALDH1L1 (1:200,
   17390-1-AP, Proteintech, Wuhan, China) and HMGCS2 (1:100, A14244,
   Abclonal, Wuhan, China). After detection with the horseradish
   peroxidase-conjugated secondary antibody (PR30011, Proteintech, Wuhan,
   China), the slides were developed using diaminobenzidine and
   counterstained with hematoxylin. All the results were photographed
   using a microscope and then analyzed utilizing Image J software.

Isolation of primary human nasal epithelial cells (hNECs)

   During the surgical procedure, nasal tissues were obtained from polyps
   in individuals diagnosed with CRSwNP and from inferior turbinate of the
   control subjects. The tissues were enzymatically digested by overnight
   incubation in a mixture of DMEM/F12 fortified with penicillin
   (200 U/mL), streptomycin (200 mg/mL), and 0.2% dispase II (Roche,
   Basel, Switzerland). After digestion, the nasal tissues were sieved
   through a 70 μm cell strainer and then centrifuged at 200 g for 3 min.
   The resultant sediment was suspended in PneumaCult-Ex Medium (Stemcell,
   Vancouver, BC, Canada) and plated onto plastic culture dishes. The
   culture medium was refreshed on the first day post-seeding and
   subsequently every other day. Having reached approximately 80%
   confluence, the cells were subcultured following trypsinization, and
   generations F1 to F3 were used for experiments.

Western blotting

   Following RIPA lysis, the protein concentration of human nasal
   epithelial cells (hNECs) was determined. Approximately 30 μg of protein
   was then separated through the utilization of sodium dodecyl
   sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The proteins
   were subsequently translocated onto membranes composed of
   polyvinylidene difluoride (PVDF). Following the blocking by 5% BSA, the
   membranes were subjected to incubation with primary antibodies,
   including OXR1 (1:4000, 13514-1-AP, Proteintech, Wuhan, China), BCKDHB
   (1:1000, 13685-1-AP, Proteintech, Wuhan, China), CBR3 (1:4000,
   15619-1-AP, Proteintech, Wuhan, China), ALDH1L1 (1:4000, 17390-1-AP,
   Proteintech, Wuhan, China), HMGCS2 (1:1000, A14244, Abclonal, Wuhan,
   China), and β-Actin (1:2000, #4967, Cell Signaling Technology, Danvers,
   MA, USA). After incubation with secondary antibodies, the binding of
   target proteins was visualized via the ECL system, followed by analysis
   utilizing Image J software.

Detection of branched chain amino acids (BCAA)

   Following the manufacturer's guidelines, the BCAA levels were assessed
   using the BCAA kit (MAK003, Sigma-Aldrich, St Louis, MO, USA). The
   nasal polyps and normal mucosal tissues were weighed and homogenized,
   then the BCAA concentration was detected using 200 μL of the Assay
   buffer.

Statistical analysis

   In this investigation, we utilized R software (version 4.3.1) for
   statistical computations and visualization of bioinformatics datasets.
   Additionally, GraphPad Prism (San Diego, CA, USA) and SPSS (IBM SPSS,
   Chicago, IL, USA) were used to analyze our experimental data delineated
   as mean ± standard deviation (SD). Binary comparisons were addressed
   using the t-test, while multiple group assessments using one-way ANOVA.
   A threshold of P < 0.05 was established for statistical significance.

Results

Identification of DEGs between CRSwNP and control samples

   The study protocol is outlined in the flow chart ([79]Fig. 1). After
   mitigating the influence of batch effects, 48 samples from CRSwNP
   patients and 34 control samples were combined for analysis
   ([80]Fig. 2A). A total of 1322 genes were identified as DEGs,
   comprising 737 upregulated and 585 downregulated ([81]Fig. 2B). A Venn
   diagram ([82]Fig. 2C) and a heatmap ([83]Fig. 2D) were utilized to
   visualize the 24 DEMRGs obtained from the intersection of DEGs with
   mitochondrial-related genes. The majority of these genes exhibited a
   lower expression level in CRSwNP samples, compared to control samples.
   To assess the functional enrichment of DEMRGs, we performed GO, KEGG,
   and MitoCarta3.0 pathway enrichment analyses, revealing that the DEMRGs
   were involved in diverse biological processes, including “Mitochondrial
   matrix”, “Arginine and proline metabolism” and “Small molecule
   catabolic process” ([84]Fig. 2E). Additionally, the KEGG analysis
   unveiled the enrichment of DEMRGs in various pathways associated with
   metabolism and biosynthesis ([85]Fig. 2E). Furthermore, using the
   MitoCarta3.0 database, the pathways associated with DEMRGs were
   explored. The findings indicated a significant connection between the
   DEMRGs and such pathways as “Amino acid metabolism”, “Folate and 1-C
   metabolism” and “Vitamin metabolism” ([86]Fig. 2F).

Fig. 1.

   [87]Fig. 1
   [88]Open in a new tab

   The flowchart illustrating the investigative procedure. CON, control;
   CRSwNP, chronic rhinosinusitis with nasal polyps.

Fig. 2.

   [89]Fig. 2
   [90]Open in a new tab

   Identification of the differentially expressed mitochondrial-related
   genes (DEMRGs) between CRSwNP and control groups. (A) PCA revealing a
   batch effect between the integrated datasets both before and after
   de-batching. (B) The volcano plot of the differentially expressed genes
   (DEGs). (C) The Venn diagram showing intersection genes between DEGs
   and Mitochondrial-related genes from MitoCarta3.0. (D) Clustered
   heatmap of DEMRGs. (E) GO and KEGG pathway enrichment analyses of
   DEMRGs. (F) MitoPathway enrichment analysis of DEMRGs.

Extraction of hub DEMRGs

   In order to extract the hub DEMRGs, we utilized the aforementioned set
   of 24 genes as inputs in 4 distinct machine learning algorithms,
   including LASSO regression ([91]Fig. 3A and B), SVM-RFE algorithm
   ([92]Fig. 3C), GBM algorithm ([93]Fig. 3D), and RF algorithm
   ([94]Fig. 3E). By intersecting the outputs of these algorithms, we
   successfully identified 6 hub DEMRGs, consisting of ALDH1L1, BCKDHB,
   C15orf48, CBR3, HMGCS2, and OXR1 ([95]Fig. 3F). C15orf48 was found
   highly expressed in the CRSwNP samples, whereas the others in the
   control samples ([96]Fig. 3G). The ROC curves for ALDH1L1, BCKDHB,
   C15orf48, CBR3, HMGCS2 and OXR1 exhibited AUCs of 0.858, 0.886, 0.807,
   0.869, 0.873, and 0.912, respectively ([97]Fig. 3H). Furthermore, the
   combined model achieved an AUC of 0.958, significantly higher than that
   of any single gene. Subsequently, a nomogram was developed to further
   investigate the clinical significance of these biomarkers
   ([98]Fig. 3I). The calibration curves showed that the probability
   predicted by the nomogram was closely aligned with that of the ideal
   model, suggesting that these biological markers possessed a high
   predictive accuracy ([99]Fig. 3J).

Fig. 3.

   [100]Fig. 3
   [101]Open in a new tab

   Detection of hub DEMRGs utilizing a comprehensive methodology. (A, B)
   Implementation of the LASSO regression. (C) Tenfold cross-validation
   error estimated using SVM-RFE. (D) The importance of features according
   to GBM algorithm. (E) Biomarker screening using the RF algorithm. (F)
   Venn diagram illustrating the intersection among the 4 above-mentioned
   machine learning outputs. (G) The expression of hub DEMRGs in CRSwNP
   and control samples. (H) The ROC curve of the diagnostic efficacy of
   the hub DEMRGs. (I) A nomogram model based on the 6 diagnostic
   biomarkers. (J) The calibration curve of the nomogram assessing the
   prognostic accuracy of the model. ^∗P < 0.05; ^∗∗P < 0.01;
   ^∗∗∗P < 0.001; ^∗∗∗∗P < 0.0001; ns, not significant.

Network between hub genes, miRNAs and TFs

   By analyzing the relationship between hub genes, miRNAs and
   transcription factors (TFs), we conducted a network to identify the
   essential miRNAs and TFs implicated in the mechanism of CRSwNP
   ([102]Fig. 4). This network consisted of 93 miRNAs, 6 mRNAs, 43 TFs,
   142 nodes, and 181 edges. OXR1 was found to be the top targeted gene,
   modulated by 71 miRNAs and 13 TFs. Hsa-mir-182-5p was the miRNA
   regulating the highest number of DEMRGs. Furthermore, PPARG and STAT3
   were identified as the TFs regulating the largest number of DEMRGs.

Fig. 4.

   [103]Fig. 4
   [104]Open in a new tab

   Target gene-miRNA and target gene-transcription factor (TF) networks
   are depicted in the diagram. Genes are represented by purple circled
   nodes, miRNAs by blue triangle nodes, and TFs by green diamond nodes.

Immune cell infiltration analysis

   Five groups of immunocytes demonstrated notable variations between the
   nasal polyps and normal nasal tissues ([105]Fig. 5A). Specifically,
   resting mast cells, M2 macrophages and resting dendritic cells (DCs)
   were significantly enriched in CRSwNP samples. On the other hand,
   plasma cells and M0 macrophages were much more abundant in control
   samples ([106]Fig. 5B). Further analysis of immune cell infiltration
   revealed several significant associations among various cellular
   subtypes ([107]Fig. 5C). Spearman correlation analysis revealed a
   consistently positive correlation of the expression levels of ALDH1L1,
   BCKDHB, CBR3, HMGCS2 and OXR1 with the abundance of M2 macrophages and
   resting mast cells. Conversely, their expression levels showed a
   coherently negative correlation with the infiltration of plasma cells
   and M0 macrophages. C15orf48 displayed a markedly positive correlation
   with memory activated CD4 T cells and M2 macrophages, but a negative
   correlation with M0 macrophages ([108]Fig. 5D).

Fig. 5.

   [109]Fig. 5
   [110]Open in a new tab

   Comparison of immune cell infiltration between CRSwNP and control
   groups. (A) The stacked bar chart illustrating the distribution of
   immune cell types. (B) The box plot depicting the contrasting
   distribution of 22 distinct types of immune cells between the CRSwNP
   and control groups. (C) The heatmap displaying the correlation patterns
   among the 22 types of infiltrating immune cells. (D) The correlation
   map showing the connection between differentially infiltrating immune
   cells and hub DEMRGs. ^∗P < 0.05; ^∗∗P < 0.01; ^∗∗∗P < 0.001;
   ^∗∗∗∗P < 0.0001; ns, not significant.

Mitochondrial characteristics in CRSwNP

   The top 10 MitoPathways were selected based on their log2FC values
   using the ssGSEA algorithm, all of which displayed a decreased activity
   in CRSwNP ([111]Fig. 6A). By integrating the top 10 MitoPathways with
   the MitoPathways via MitoCarta3.0 pathway enrichment analysis, 3 hub
   MitoPathways including “Amino acid metabolism”, “Phospholipid
   metabolism” and “Branched-chain amino acid dehydrogenase complex” were
   identified ([112]Fig. 6B). An assessment was carried out to investigate
   the correlation between DEMRGs and genes linked to the respiratory
   chain complexes (I–V) ([113]Fig. 6C). A robust positive correlation was
   observed between the expression levels of respiratory chain complex
   genes and HMGCS2.

Fig. 6.

   [114]Fig. 6
   [115]Open in a new tab

   MitoPathway-related characteristics in CRSwNP. (A) Distinct ssGSEA
   scores of the top 10 MitoPathways between CRSwNP and control samples.
   (B) Hub MitoPathways by intersecting the top 10 MitoPathways with the
   DEMRGs enrichment results. (C) Correlation between hub DEMRGs and
   respiratory chain complex (I–V) genes.

Expression of hub DEMRGs in validation datasets

   The analysis of dataset [116]GSE23552 revealed that the expression
   levels of ALDH1L1, BCKDHB, CBR3, HMGCS2, and OXR1 were significantly
   lower in nasal polyps compared to normal nasal mucosa ([117]Fig. 7A).
   In contrast, the expression levels of C15orf48 remained unchanged
   between the 2 cohorts. Furthermore, activities of major MitoPathways,
   specifically “Amino acid metabolism” and “Branched-chain amino acid
   dehydrogenase complex”, were significantly repressed in CRSwNP samples
   ([118]Fig. 7B). Subsequent scRNAseq analysis highlighted that C15orf48
   was predominantly expressed in myeloid cells, but the other hub genes
   in epithelial cells ([119]Fig. 7C–D). The downregulation of these genes
   in nasal polyp epithelial cells was consistent with findings from
   previous analysis ([120]Fig. 7E–F).

Fig. 7.

   [121]Fig. 7
   [122]Open in a new tab

   The expression of biomarkers between CRSwNP and control groups in
   validation datasets. (A) The expression of hub DEMRGs in [123]GSE23552.
   (B) The expression of hub MitoPathways in [124]GSE23552. (C) The UMAP
   map showing annotation results of cell subgroups in the scRNAseq data.
   (D) The distribution of hub DEMRGs in 6 cell subgroups. (E, F) The mean
   expression of hub DEMRGs in epithelial cells for each individual
   sample. ^∗P < 0.05; ^∗∗P < 0.01; ^∗∗∗P < 0.001; ^∗∗∗∗P < 0.0001; ns,
   not significant.

Negative correlation between the expression of hub DEMRGs and CRSwNP severity

   The results of RT-qPCR indicated that the expression levels of ALDH1L1,
   BCKDHB, CBR3, HMGCS2, and OXR1 were significantly decreased in CRSwNP
   samples ([125]Fig. 8A). To investigate the relationship between the hub
   DEMRGs and the severity of CRSwNP, the Lund-Mackay and Lund-Kennedy
   scores of patients with CRSwNP were evaluated ([126]Supplementary
   Table 4). The Lund-Mackay score had significantly positive correlations
   with the ΔCt values of ALDH1L1 (R^2 = 0.45; P = 0.005), BCKDHB
   (R^2 = 0.48; P = 0.042), CBR3 (R^2 = 0.47; P = 0.011), HMGCS2
   (R^2 = 0.30; P = 0.032), OXR1 (R^2 = 0.41; P = 0.006) ([127]Fig. 8B).
   Similarly, The Lund-Kennedy score displayed significantly positive
   correlations with the ΔCt of ALDH1L1 (R^2 = 0.47; P = 0.006), BCKDHB
   (R^2 = 0.28; P = 0.004), CBR3 (R^2 = 0.40; P = 0.005), HMGCS2
   (R^2 = 0.31; P = 0.033), OXR1 (R^2 = 0.45; P = 0.010) ([128]Fig. 8B).

Fig. 8.

   [129]Fig. 8
   [130]Open in a new tab

   Confirmation of ALDH1L1, BCKDHB, CBR3, HMGCS2, and OXR1 expression and
   their association with the severity of CRSwNP. (A) mRNA expression of
   hub DEMRGs in CRSwNP and control samples. (B) Correlations between the
   ΔCt in PCR of hub DEMRGs and the severity of CRSwNP, including
   Lund-Mackey score and Lund-Kennedy score. (C, D) Immunostaining of
   protein expression. Scale bars = 200 μm. (E, F) Representative
   expression levels of hub DEMRGs in Western blotting. (G) BCAA
   concentrations in nasal mucosa from CRSwNP and control groups. ΔCt = Ct
   (DEMRGs) – Ct (B2M); ^∗P < 0.05; ^∗∗P < 0.01; ^∗∗∗P < 0.001;
   ^∗∗∗∗P < 0.0001; ns, not significant.

Nasal polyps exhibit reduced protein expression levels of DEMRGs

   Immunohistochemistry and Western blotting showed a significant decrease
   in the protein levels of ALDH1L1, BCKDHB, CBR3, HMGCS2, and OXR1 in
   both nasal polyp tissues and hNECs derived from these polyps
   ([131]Fig. 8C–F). Additionally, the concentrations of BCAA elevated
   significantly in nasal polyps, compared to the normal nasal mucosa
   ([132]Fig. 8G).

Discussion

   CRSwNP, a multifactorial condition, has an etiology involving genetic,
   anatomical, and environmental components. Prior investigations have
   demonstrated that dysregulation of mitochondrial function in nasal
   epithelial cells may induce the pathogenesis of CRSwNP.[133]^5 However,
   the molecular mechanisms underlying this phenomenon remain undefined.
   In this investigation, we utilized public datasets from the GEO
   database to identify the DEMRGs between CRSwNP patients and control
   subjects. By applying integrated machine learning methods, we
   pinpointed 5 genes: ALDH1L1, BCKDHB, CBR3, HMGCS2, and OXR1. The
   downregulation of these genes was subsequently validated through
   external datasets and laboratory experiments. Furthermore, we observed
   a significant association between the downregulation of these genes and
   the heightened severity of CRSwNP.

   Cytosolic 10-formyltetrahydrofolate dehydrogenase (ALDH1L1) serves as
   an essential enzyme in the folate metabolic pathway. By breaking down
   harmful intracellular aldehyde compounds, ALDH1L1 effectively reduces
   oxidative stress, a major contributor to the development of CRSwNP.
   Additionally, ALDH1L1 regulates the expression of cell cycle-related
   gene via its metabolic capabilities. Recent investigations have
   revealed that the re-expression of ALDH1L1 not only hampers cellular
   proliferation, but also the migration and invasion of cancer
   cells.[134]^23^,[135]^24 Given the critical role of tissue
   restructuring in the pathogenesis of CRSwNP, ALDH1L1 emerges as a
   promising indicator for the progression of nasal polyp growth.

   Branched chain keto acid dehydrogenase E1 subunit beta (BCKDHB)
   performs an important role in metabolizing BCAA within the human body.
   The role of BCKDHB in BCAA metabolism is associated with inflammatory
   responses and immune function. Impairment in BCKDHB function can
   elevate oxidative stress, leading to excessive generation of free
   radical. Such oxidative imbalance can cause cellular damage and tissue
   inflammation, thereby promoting the advancement of CRSwNP. Previous
   studies have indicated that individuals with severe asthma exhibit
   higher plasma BCAA levels, compared to those with mild asthma.[136]^25
   Considering the close association between CRSwNP and asthma, BCKDHB
   might contribute to the development of CRSwNP by promoting abnormal
   accumulation of BCAA and their metabolites.

   Carbonyl reductase 3 (CBR3) is an NADPH-dependent enzyme that primarily
   catalyzes the reduction of various endogenous and exogenous carbonyl
   compounds. Bioinformatic analysis has revealed an association between
   CBR3 and both chronic obstructive pulmonary disease (COPD) and exposure
   to cigarette smoke, indicating that CBR3 could serve as a potential
   diagnostic biomarker for airway inflammatory diseases.[137]^26
   Furthermore, genetic variations in CBR3 have been associated with the
   progression of several inflammatory conditions, including rheumatoid
   arthritis, chronic nephritis, and inflammatory bowel disease.[138]^27
   Consequently, dysfunction in CBR3 may impair the body's antioxidant
   defenses, thereby heightening the risk of developing CRSwNP.

   3-Hydroxy-3-methylglutaryl-CoA synthase 2 (HMGCS2) plays a crucial role
   in mitochondrial function, significantly impacting both energy and
   lipid metabolism. In the ketogenesis pathway, HMGCS2 catalyzes the
   production of β-hydroxybutyrate, a ketone body known to inhibit the
   NLRP3 inflammasome.[139]^28 This inhibition reduces the secretion of
   pro-inflammatory cytokines like IL-1β and IL-18, thereby alleviating
   oxidative stress and inflammation within nasal epithelial cells.
   Additionally, HMGCS2 is involved in the differentiation of epithelial
   cells and the regulation of tight junction proteins, which are vital
   for maintaining the integrity of the epithelial barrier.[140]^29

   Oxidation resistance 1 (OXR1) is fundamental to the stabilization of
   intracellular redox states, providing a robust defense against
   oxidative stress. Genetic studies have revealed that polymorphisms in
   the OXR1 gene are associated with the development of allergic asthma in
   children, suggesting a potential link to airway inflammatory
   diseases.[141]^30 Recent research indicates that the expression levels
   of OXR1 are reduced in epithelial cells of nasal polyps compared to
   healthy nasal mucosa which increase susceptibility to oxidative stress
   and exacerbate inflammatory responses.[142]^31

   Immune dysregulation plays a pivotal role in the pathogenesis of
   CRSwNP. Our current research utilized the CIBERSORT algorithm to assess
   immune cell infiltration in nasal tissues, revealing increased
   abundances of M2 macrophages, resting DCs, and mast cells. Notably, M2
   macrophages emerge as a pivotal player in the tissue remodeling in
   CRSwNP, particularly through the release of growth factors that promote
   fibroblast proliferation and the degradation of the extracellular
   matrix.[143]^32^,[144]^33 Additionally, M2 macrophages are instrumental
   in modulating anti-inflammatory responses and promoting type 2 immune
   responses.[145]^34 Recent findings highlight a significant increase in
   mast cells within nasal polyps, predominantly localized to the
   epithelium, where they facilitate tissue remodeling by releasing
   carboxypeptidase A3.[146]^35 Furthermore, the presence of mast cells
   may enhance the recruitment of eosinophils, thereby amplifying type 2
   inflammatory responses.[147]^36 Flow cytometry studies by Zhai
   et al[148]^37 have substantiated a positive correlation between the
   presence of eosinophils and mast cells in patients with CRSwNP,
   corroborating the outcomes of our analysis on immune infiltration. In
   the microenvironment of nasal polyps, a marked increase in DC
   infiltration significantly elevates Th2 cytokines levels, thereby
   disturbing the Th1/Th2 balance.[149]^38^,[150]^39 Aligning with our
   results, a recent scRNA-seq analysis by Wang et al[151]^40 has
   demonstrated high expression levels of DC markers CD163 and NR4A1 in
   early CRSwNP, potentially contributing to T cell activation.
   Furthermore, research by Lin et al[152]^41 has identified that DCs
   expressing CD80 and CD86 exacerbate CRSwNP inflammation through
   interactions with CD40L on T cells, promoting T cell maturation and the
   release of inflammatory molecules. Previous research has highlighted
   the critical role of mitochondrial metabolism in shaping the function
   and destiny of immune cells.[153]^42 Our correlational studies revealed
   that the hub DEMRGs are inversely related to M2 macrophages and mast
   cells, suggesting that mitochondrial dysregulation may amplify type 2
   inflammation by facilitating the infiltration of these cells in the
   nasal epithelium. Additionally, we observed a positive correlation
   between the expression levels of all 5 hub genes with M0 macrophages
   and plasma cells. M0 macrophages, initially in a quiescent state, have
   potential to differentiate into either M1 or M2 macrophages in response
   to environmental stimuli. Our study demonstrated that in nasal polyps,
   the loss of function of the mitochondrial-related genes facilitated the
   transition from M0 to M2 macrophages. Plasma cells, which are the
   terminal differentiation stage of B cells, exhibit high metabolic
   activity and require substantial energy to maintain their immune
   functions.[154]^43 Reduced expression of mitochondrial-related genes
   may impair the survival and function of plasma cells by limiting energy
   supply and weakening antioxidant defenses, ultimately affecting their
   abundances in nasal polyps.

   In this study, we employed 2 algorithms to identify differentially
   activated metabolic pathways between nasal polyps and normal mucosa.
   Our findings demonstrated a downregulation of BCAA in the CRSwNP group.
   Previous research indicates that BCAA is implicated in modulating the
   metabolism of glucose, lipids, and proteins.[155]^44 Furthermore, BCAA
   can facilitate interorgan metabolic communication, suggesting that
   abnormalities in BCAA catabolism may contribute to the progression of
   diverse metabolic disorders.[156]^45 However, the correlation between
   BCAA metabolism and airway inflammatory diseases has been
   underexplored. Comhair et al[157]^25 have reported that the level of
   plasma BCAA increases in asthmatics with elevated levels of fractional
   exhaled nitric oxide (FENO), compared to those with low FENO levels.
   Conversely, Li et al[158]^46 have revealed the upregulation of
   branched-chain amino acid transaminase 1 (BCAT1) in neonatal asthmatic
   mice, suggesting that inhibiting BCAT1 may attenuate airway
   inflammation and remodeling by reducing autophagy. The present study
   indicated that the metabolic disruption of BCAA might serve as a
   trigger for CRSwNP, thereby providing novel insights into the mechanism
   of upper airway inflammatory diseases.

   To the best of our knowledge, this study is the first to comprehensive
   explore the relationship between mitochondrial dysfunction and the
   immune microenvironment in CRSwNP. We identified that 5 genes, namely
   ALDH1L1, BCKDHB, CBR3, HMGCS2, and OXR1, as potential molecular markers
   for the diagnosis and treatment of CRSwNP. However, our study has
   several limitations. First, the analysis was constrained by a
   relatively small dataset. Moreover, though we validated the expression
   of hub DEMRGs in human tissues, the intricate regulatory mechanisms
   governing these genes in CRSwNP require further elucidation through
   in vivo and in vitro studies.

Conclusion

   This comprehensive bioinformatics study identifies substantial
   variations in mitochondrial gene expression between patients diagnosed
   with CRSwNP and control subjects. Specifically, ALDH1L1, BCKDHB, CBR3,
   HMGCS2, and OXR1 regulate immune infiltration in CRSwNP. These genes
   might serve as prognostic and therapeutic strategies for CRSwNP.

Abbreviations

   CRSwNP, chronic rhinosinusitis with nasal polyps; CRSsNP, chronic
   rhinosinusitis without nasal polyps; OXPHOS, oxidative phosphorylation;
   ATP, adenosine triphosphate; ROS, reactive oxygen species; ADMA,
   asymmetric dimethylarginine; GEO, Gene Expression Ominibus; DEGs,
   differentially expressed genes; DEMRGs, differentially expressed
   mitochondrial-related genes; GO, gene ontology; KEGG, Kyoto
   encyclopedia of genes and genomes; MitoPathways, mitochondrial
   pathways; LASSO, absolute shrinkage and selection operator; SVM-RFE,
   support vector machine recursive feature elimination; GBM, gradient
   boosting machine; RF, random forest; ROC, receiver operating
   characteristic; AUC, area under the curve; TFs, transcription factors;
   CIBERSORT, cell-type identification by estimating relative subsets of
   RNA transcripts; ssGSEA, single-sample gene set enrichment analysis;
   CT, computed tomography; hNECs, human nasal epithelial cells; SDS-PAGE,
   sodium dodecyl sulfate-polyacrylamide gel electrophoresis; PVDF,
   polyvinylidene difluoride; BCAA, branched chain amino acids; DCs,
   dendritic cells; N-ERD, nonsteroidal anti-inflammatory drug-exacerbated
   respiratory disease; COPD, chronic obstructive pulmonary disease; FENO,
   fractional exhaled nitric oxide; SEB, staphylococcal enterotoxin B.

Funding

   This work was supported by the National Natural Science Foundation
   (82171117) and Jiangsu Province Capability Improvement Project through
   Science, Technology and Education (JSDW202203) of China.

Availability of data and materials

   The datasets utilized in this study are accessible in online
   repositories. [159]GSE136825, [160]GSE36830, and [161]GSE23552 can be
   accessed at [162]https://www.ncbi.nlm.nih.gov/geo/. SCP253 can be
   accessed at [163]https://singlecell.broadinstitute.org/single_cell/.

Author contributions

   M.-P.L. and L.C. designed the study. M.G., C.H., J.-Q.Z., Y.Y., and
   X.-Y.Z. carried out subject enrollment and data collection. B.Y.
   conducted experimental work, analyzed the data, and prepared the
   manuscript. C.-Y.Q., M.-P.L., and L.C. interpreted the results,
   reviewed, and edited the manuscript. All authors provided their consent
   for the submission of the final version.

Ethics statement

   This study obtained ethical approval from the Ethical Committee of the
   First Affiliated Hospital of Nanjing Medical University (2020-SR-500)
   for research involving human participants.

Consent for publication

   The authors obtained informed consent for publication from all
   participants in the study. The participants understand that their
   personal information will not be published.

Declaration of competing interest

   The authors declare that they conducted the research without any
   potential conflict of interest arising from commercial or financial
   relationships.

Acknowledgments