Skip to main content
NCBI home page
Frontiers in Physiology logoLink to Frontiers in Physiology
. 2026 Jun 9;17:1870580. doi: 10.3389/fphys.2026.1870580

Decoding isonicotinylation-associated patterns in neutrophil chronic obstructive pulmonary disease: evidence from integrative bioinformatic-driven multi-omics and in vitro validation

Zizhong Wang 1,2, Yuan Li 1,2, Mengqi Zhou 1,2, Min Xiang 1,2, Jiangtao Lin 2,*
PMCID: PMC13286763  PMID: 42344274

Abstract

Background

Chronic obstructive pulmonary disease (COPD) is driven by complex inflammatory processes in which neutrophil dysregulation plays a central role. Isonicotinylation (Kinic), a novel lysine acylation modification linked to cellular metabolism, has not yet been clearly associated with COPD.

Methods

We employed an artificial intelligence (AI)-driven, multi-omics framework. First, the Limma, WGCNA, and CIBERSORT algorithms were used to identify a Kinic- and neutrophil (KN)-associated shared molecular signature in the peripheral blood bulk profiles of COPD patients. Next, NMF and explainable machine learning identified molecular subgroups and developed a diagnostic model for COPD patients based on the KN-associated gene signature. In addition, a KN-associated pathogenic hub factor was identified, and its molecular and immune signatures in COPD were assessed using a cutting-edge analytical framework in spatial and temporal manners. An AI-based drug screening platform (DrugReflector) and molecular docking were used to identify therapeutic candidates. Finally, Lymphotoxin-beta (LTB) expression was validated in vitro using q-RT-PCR assays.

Results

We identified a robust five-gene KN-associated signature that demonstrated diagnostic and patient-stratification potential in COPD. LTB was identified as an upregulated Kinic-associated hub gene mainly distributed in neutrophils and involved in the regulation of COPD pathogenesis. Notably, BRD-K97481123 was identified as a potential LTB-targeting compound for the treatment of COPD.

Conclusion

This study unveils a novel KN-associated molecular axis in COPD pathogenesis, with LTB as a neutrophil-centric pathogenic factor. This axis provides a framework for patient stratification, offers a promising diagnostic biomarker, and identifies a potential therapeutic target, thereby linking a novel metabolic modification to neutrophilic inflammation in COPD.

Keywords: artificial intelligence, chronic obstructive pulmonary disease, isonicotinylation, multi-omics, neutrophil

1. Introduction

Chronic obstructive pulmonary disease (COPD) is a prevalent, progressive respiratory disorder characterized by persistent airflow limitation and chronic inflammation, leading to significant morbidity and mortality worldwide (Ritchie and Wedzicha, 2020). COPD pathogenesis involves a complex interplay of environmental exposures, aberrant immune responses, and tissue remodeling (Hattab et al., 2016). Beyond the classical involvement of neutrophils, macrophages, and lymphocytes, emerging evidence highlights the role of post-translational modifications (PTMs) in modulating protein function and disease progression (Qian et al., 2023).

Isonicotinylation (Kinic) is a recently characterized lysine acylation in which an isonicotinoyl group is transferred onto lysine residues, potentially altering protein stability, interactions, and activity (Jiang et al., 2021). More specifically, for histone proteins, the acetyltransferases CREB-binding protein (CBP) and P300 can mediate Kinic modification of histones, thereby relaxing chromatin structure and promoting the transcription of genes, such as those that regulate the PI3K-AKT-mTOR in liver cancer (Jiang et al., 2021). In non-histone proteins, CBP and Tip60 can mediate Kinic modification of the SMAD3 protein, subsequently activating the TGFβ pathway (Li et al., 2024). In addition, recent studies have suggested that Kinic may serve as an efficient predictive and therapeutic marker for liver cancer patients (Zhou et al., 2026). This modification, which is linked to cellular metabolic states, has been implicated in cancer but remains uninvestigated in COPD (Li et al., 2024). Concurrently, neutrophils are widely recognized as the predominant inflammatory effector cells in the COPD airway, and their dysregulation is a hallmark of disease pathogenesis (Chen et al., 2024). In the lungs of patients with COPD, persistent exposure to cigarette smoke and other noxious particles drives a chronic inflammatory milieu that recruits neutrophils from the circulation into the airways and parenchyma (Kapellos et al., 2023). Once activated, these infiltrating neutrophils exert their pathogenic effects through multiple mechanisms, including the release of serine proteases, matrix metalloproteinases, and reactive oxygen species, which collectively degrade the extracellular matrix, damage the alveolar epithelium, and impair lung repair mechanisms (Yan et al., 2022).

To address this gap, we conducted an integrative multi-omics study leveraging artificial intelligence (AI) to delineate the coordinated role of Kinic-neutrophil biology in COPD. We aimed to identify a shared isonicotinylation-neutrophil (KN)-associated molecular signature in COPD patients, along with its corresponding diagnostic, patient-stratifying, and therapeutic potential.

2. Materials and methods

2.1. Source of data

Bulk RNA-sequencing datasets from the peripheral blood of COPD patients and healthy controls (HC) were retrieved from the Gene Expression Omnibus (GEO) database. GSE42057 served as the internal set 1. GSE44962 served as the internal set 2. GSE77344 was designated as the training set, and GSE148871 served as the independent validation set. GSE42057 comprises 94 COPD samples and 42 HC samples. GSE55962 comprises 82 COPD samples and 24 HC samples. GSE77344 comprises 172 COPD samples and 20 HC samples. GSE148871 comprises 152 COPD samples and 32 HC samples. Batch effects were assessed and corrected using the sva and Limma R packages (Ritchie et al., 2015). In addition, the Kinic-associated signature was obtained from the relevant literature.

2.2. Limma analysis

Differential expression analysis between the COPD and HC groups in GSE42057 was performed using the Limma R package (Ritchie et al., 2015). Genes with a |log2 fold change| > 0.5 and a p-value < 0.05 were defined as differentially expressed genes (DEGs). The intersection of the Kinic-related gene set with the DEGs yielded the Kinic-related DEGs. Next, the Kinic-associated DEGs were analyzed using KEGG and GO enrichment analyses with the clusterProfiler R package, with reference to the KEGG and GO gene sets downloaded from the MSIGDB database (Yu et al., 2012).

2.3. CIBERSORT and WGCNA analyses

The relative abundance of neutrophils in samples from GSE55962 was estimated using the CIBERSORT algorithm in R (Chen et al., 2018). WGCNA was performed on the same dataset using the WGCNA R package to construct a scale-free co-expression network (Langfelder and Horvath, 2008). A soft-thresholding power was selected to achieve a scale-free topology (Langfelder and Horvath, 2008). Modules were identified via hierarchical clustering, and the module whose eigengene showed the highest absolute correlation with the neutrophil infiltration score was selected as the neutrophil-correlated module (Langfelder and Horvath, 2008). Genes from this module were extracted and then intersected with Kinic-associated DEGs to obtain KN-associated DEGs.

2.4. Systemic machine learning model construction

A comprehensive 132-model machine learning workflow was implemented using the caret and glmnet packages in R on the training set (GSE77344) and independent validation set (GSE148871) (Wu et al., 2024). The 132-model machine learning framework typically represents a comprehensive grid-search approach that systematically combines multiple classification algorithms with various feature-selection methods and preprocessing strategies. Combinations of these algorithms were applied to the MetaGSE for signature selection and model construction using 10-fold cross-validation (Wu et al., 2024). The optimal model was selected based on the highest mean area under the curve (AUC) index (Wu et al., 2024). Its performance was rigorously evaluated on the independent validation set (GSE148871) using ROC and PR curves via the pROC package in R (Robin et al., 2011). A nomogram coupled with calibration was constructed for clinical translation using the rms package in R (Lin et al., 2024). SHAP analysis was applied to interpret the model and identify the feature (gene) with the greatest contribution, which was designated as the hub gene (Zhou et al., 2026).

2.5. NMF analysis

To uncover molecular heterogeneity among COPD patients, we performed non-negative matrix factorization (NMF) based on the expression profiles of the KN-associated signature genes in COPD samples from the training set (GSE77344) using the NMF R package (Cai et al., 2023). NMF decomposes the non-negative gene expression matrix into two lower-dimensional, non-negative matrices, revealing latent molecular patterns that define patient subgroups (Cai et al., 2023). The resulting coefficient matrix from the NMF decomposition was used to assign each COPD sample to either Subtype 1 (C1) or Subtype 2 (C2) (Cai et al., 2023). Subsequently, these identified subtypes were compared for differences in the expression levels of the KN-associated gene signature using the ggplot2 package in R (Gustavsson et al., 2022). Immune cell infiltration profiles were estimated using the CIBERSORT package in R (Chen et al., 2018). The enrichment of pathways was assessed using GSEA with the clusterProfiler package in R (Yu et al., 2012).

2.6. Single-cell analysis

The scRNA-seq data (GSE167295, including three COPD peripheral blood samples) were acquired from the GEO database and processed using the Seurat pipeline in R (Butler et al., 2018). Quality control (QC) filtered out low-quality cells (Butler et al., 2018). Dimensionality reduction was performed using PCA, followed by clustering and visualization with UMAP and t-SNE (Butler et al., 2018). Cell types were annotated using canonical marker genes with the singleR package in R (Cao et al., 2023). The expression distribution of the hub gene was visualized across all cell types. Cell communication patterns among various cell types were assessed using the CellChat package in R (Jin et al., 2021). Next, the metabolic pathway changes among various cell types were assessed using the scMetabolism algorithm in R (Argüello et al., 2020). Pseudotime trajectory analysis of neutrophils was conducted using Monocle2 to infer differentiation states (Fang et al., 2022). A virtual knockout of the hub gene in the neutrophil cluster was simulated using the scTenifoldKnk package in R, and the downstream regulatory network and functional impact of the hub gene were assessed by KEGG and GO enrichment analyses using the clusterProfiler R package, with reference to the KEGG and GO gene sets downloaded from the MSIGDB database and the GENEMINIA database (Yu et al., 2012; Osorio et al., 2022). AUCELL was performed to identify Kinic patterns among various cell types in accordance with the Kinic-associated gene list (Chen et al., 2025).

2.7. Drug screening analysis

The deep learning platform DrugReflector was employed to screen the Connectivity Map (cMAP) database for compounds predicted to reverse the COPD gene expression signature relative to that of HCs derived from GSE77344 (DeMeo et al., 2025). The top candidate compound was selected (DeMeo et al., 2025). Molecular docking simulations between the identified hub protein and the candidate drug were performed using AutoDock Vina software to predict binding affinity and pose, which were visualized with PyMOL software (Liu et al., 2024).

2.8. In vitro estimation

To experimentally validate the functional role of LTB in neutrophils in the context of COPD inflammation, we designed in vitro assays using a neutrophil model system. Given the challenges associated with transfecting primary human neutrophils, we employed the HL-60 cell line (ATCC, USA), a human promyelocytic leukemia cell line that can be differentiated into N1 neutrophil-like cells. HL-60 cells were cultured in an RPMI-1640 medium (Gibco, USA) supplemented with 20% heat-inactivated fetal bovine serum (FBS, Gibco, USA) and 1% penicillin-streptomycin (Gibco, USA). To induce neutrophil differentiation, the cells were treated with 1.25% dimethyl sulfoxide (DMSO, Sigma-Aldrich, USA) for 5–7 days. Differentiation was confirmed by assessing morphological changes. N1-like neutrophil was designated as the COPD group, and the primary neutrophil served as the control group. Total RNA was extracted from HL-60 using TRIzol Reagent (Takara, China). cDNA was then synthesized using a PrimeScript RT Reagent Kit (Vazyme, China). q-RT−PCR was carried out using TB Green Premix Ex Taq II (Vazyme, China) following the manufacturer’s instructions. The primer sequences (5’→3’) employed in this study are provided below.

LTB:

F: 5’-GCCACCACGCTCTTCTC-3’,

R: 5’-TGGTGGCTTTGTTGATGTTC-3’.

GAPDH:

F: 5’-GGAGCGAGATCCCTCCAAAAT-3’.

R: 5’-GGCTGTTGTCATACTTCTCATGG-3’.

2.9. Statistical analysis

Statistical analyses were performed using R (version 4.2.2) or GraphPad Prism (version 9.0). For two-group comparisons, a Student’s t-test or a Mann-Whitney U test was applied, as appropriate. For multiple-group comparisons, one-way ANOVA with a Tukey post-hoc test was used. A p-value < 0.05 was considered statistically significant.

3. Results

3.1. Identification of Kinic-related DEGs in COPD patients

Differential expression analysis of GSE42057 identified 1,037 DEGs between the COPD and HC groups (Figure 1A). Intersection of this data set with the Kinic-related gene set yielded 12 significant isonicotinylation-related DEGs (Figure 1B). Next, the expression patterns of these 12 DEGs in GSE42057, coupled with their molecular functions and gene feature importance, were assessed (Figures 1C–E).

Figure 1.

Panel A shows a volcano plot with red points indicating upregulated genes, blue points denoting downregulated genes, and gray points for genes not significant; downregulated genes total 563 and upregulated 474. Panel B presents a Venn diagram comparing differentially expressed genes (DEGs) and an isonicotinylation gene list, showing 12 genes overlapping, 1,025 unique to DEGs, and 72 unique to the gene list. Panel C contains a clustered heatmap visualizing differential gene expression across groups labeled HC and COPD for several genes. Panel D presents an enrichment dot plot showing top gene ontology terms for selected genes, with dot color indicating adjusted p-value and size indicating count. Panel E is a grouped dot plot comparing gene expression levels for each of twelve genes, separated by group, with corresponding confidence intervals.

Open in a new tab

Identification of Kinic-related DEGs in COPD. (A) Volcano plot displaying DEGs in GSE42057. (B) Venn diagram illustrating the intersection between all DEGs and the Kinic-related gene set, yielding 12 core genes. (C) Heatmap depicting the expression patterns of the 12 Kinic-related DEGs across COPD and HC samples. (D) GO analysis of the 12 Kinic-related DEGs. (E) Functional enrichment analysis of the 12 Kinic-related DEGs.

3.2. Identification of neutrophil-related DEGs in COPD patients

CIBERSORT analysis of GSE55962 confirmed a significant increase in neutrophil infiltration in COPD samples compared to HC samples (Figure 2D). WGCNA constructed a co-expression network and identified a specific module (black) that showed the strongest positive correlation with the neutrophil score (Figures 2A–C, E, G). This module contained 496 genes functionally enriched in neutrophil activation and immune responses. After intersection with Kinic-associated DEGs, we identified five KN-associated DEGs, namely, HDAC6, ZEB2, MCM3AP, LMAN1, and LTB (Figures 2F, H).

Figure 2.

Panel A displays a hierarchical clustering dendrogram with colored bars indicating clusters, Panel B and C show scatterplots of network analysis parameters against soft threshold values, Panel D is a violin plot comparing CIBERSORT scores between HC and COPD groups, Panel E is a heatmap of correlation coefficients and p-values for gene modules, Panel F is a Venn diagram comparing gene overlap between two modules, Panel G displays a dendrogram for gene clustering grouped by disease status and neutrophil presence, and Panel H is a circular ideogram mapping specific genes on human chromosomes.

Open in a new tab

WGCNA identifies a neutrophil-correlated module and defines the KN signature. (A) Cluster dendrogram of genes from GSE44962, with colors indicating assigned modules. (B, C) Scale-free topology analysis and mean connectivity analysis for selecting the soft-thresholding power (β). (D) Box plot comparing the CIBERSORT-estimated neutrophil infiltration score between the COPD and HC groups in GSE44962. (E) Heatmap of module-trait relationships. (F) Venn diagram of the intersection between the genes in the neutrophil-correlated black module and the 12 Kinic-related DEGs, defining the five-gene KN-associated signature. (G) Tree dendrogram of neutrophil infiltration in the black module. (H) Chromatin localization of the five KN-associated DEGs.

3.3. Identification of a KN-associated diagnostic model for COPD patients

In the COPD training set and validation set, 132 machine learning model combinations confirmed that the Stemglm(both)+plsRglm model built on this signature achieved an AUC of 0.9300 (Figure 3A). This performance robustly distinguished the pathogenesis of COPD in both the training set (AUC = 0.971) and the validation set (AUC = 0.803) (Figures 3B, C). The calibration curve of the associated nomogram showed excellent agreement (Figures 3D, E). The SHAP summary plot unequivocally identified LTB as the most influential feature driving the model’s predictive power, establishing it as the central hub gene of the KN-associated axis for further analysis (Figure 3F).

Figure 3.

Panel A presents a heatmap comparing AUC values of multiple machine learning methods across GSE77344 and GSE148871 datasets, with higher AUC in red and lower AUC in blue. Panel B shows ROC and precision-recall curves for the GSE77344 training set, indicating good model performance. Panel C displays corresponding ROC and precision-recall curves for the GSE148871 validation set with slightly lower performance. Panel D contains a nomogram summarizing model risk prediction. Panel E presents a calibration plot comparing actual versus predicted probabilities. Panel F features a SHAP summary plot indicating the impact and value of top five model features.

Open in a new tab

Development and validation of a KN-associated gene signature diagnostic model for COPD. (A) Performance metrics of 132 machine learning algorithm combinations during 10-fold cross-validation on the COPD bulk profiles. (B, C) ROC and PR curves of the optimal model. (D, E) Nomogram and calibration plot for the diagnostic model. (F) SHAP summary plot ranking the contribution of each KN-associated signature gene to the model output.

3.4. Identification of KN-associated molecular subgroups in COPD patients

NMF consensus clustering based on the five KN-associated gene signatures in GSE77344 robustly classified COPD patients into two molecular subtypes: C1 and C2 (Figures 4A, B). PCA confirmed a clear separation between the two subtypes (Figure 4C). All five KN-associated gene signatures were significantly downregulated in the C1 subtype compared to the C2 subtype (Figure 4E). GSEA revealed that the C1 subtype was markedly enriched in pathways related to neutrophil degranulation, IL-4 signaling, and epigenetic modification in C1 compared to the C2 subtype (Figure 4D). CIBERSORT analysis further indicated that C1 had a higher estimated infiltration of activated T cells and neutrophils (Figure 4F).

Figure 4.

Panel A shows four heatmaps for non-negative matrix factorization (NMF) consensus clustering at ranks two through five with corresponding silhouette scores. Panel B presents six line graphs evaluating NMF rank survey metrics, including cophenetic, dispersion, evar, residuals, rss, silhouette, and sparseness, plotted against factorization rank. Panel C displays a principal component analysis (PCA) scatter plot illustrating two groups, c1 and c2, with ellipses highlighting group separation. Panel D lists enriched pathways or processes with enrichment scores, p-values, and density plots for each pathway. Panel E features a volcano plot showing log fold change versus rank of differentially expressed genes, highlighting selected genes with varying false discovery rates. Panel F shows a violin plot comparing CIBERSORT scores for multiple immune cell types between groups C1 and C2, with some statistically significant differences marked by asterisks.

Open in a new tab

Molecular subtyping of COPD patients based on the KN-associated gene signature. (A, B) Consensus matrix and cophenetic correlation coefficient plot from NMF analysis supporting k=2 as the optimal number of clusters. (C) PCA plot showing clear separation between the two identified subtypes: C1 and C2. (D) GSEA plots showing representative pathways enriched in the C1 subtype compared with C2. (E) Heatmap comparing the expression levels of the five KN-associated gene signatures between the C1 and C2 subtypes. (F) CIBERSORT infiltration levels of key immune cells between the C1 and C2 subtypes.

3.5. Single-cell atlas of neutrophils in COPD

Analysis of scRNA-seq data (GSE167295) identified 26 major cell clusters, annotated as nine cell types (Figures 5A–C). T cells and NK cells constituted the largest proportions of the identified cell populations (Figure 5D). Cell-cell communication analysis suggested strong interactions between neutrophils and macrophages (Figure 5E). In addition, metabolic heterogeneity revealed a unique metabolic pathway in neutrophils (Figure 5F).

Figure 5.

Panel A shows a tSNE plot with clusters labeled by color and number, while Panel B presents a UMAP plot with similar clusters. Panel C displays UMAP and tSNE plots colored by cell type labels. Panel D illustrates a stacked bar chart of cell type proportions across three sample origins. Panel E features a circular chord diagram representing gene connections among different cell types. Panel F shows a dot plot of metabolic pathways by cell type, with dot size and color indicating enrichment values.

Open in a new tab

Single-cell characterization of the immune microenvironment in COPD peripheral blood. (A, B) UMAP and t-SNE visualizations of 26 cell clusters. (C) UMAP and t-SNE visualizations of 9 major cell types. (D) The proportion of each cell type across all samples. (E) Circle plot of cell-cell communication networks inferred by CellChat. (F) Heatmap of metabolic pathway activity scores across different cell types, analyzed using scMetabolism.

3.6. KN-associated signature in neutrophils in COPD patients

AUCELL analysis revealed the activation of Kinic patterns in neutrophils (Figure 6A). Expression analysis at single-cell resolution confirmed that LTB expression was predominantly and specifically localized to the neutrophil cluster, with minimal expression in other cell types (Figure 6B). Pseudotime trajectory analysis of neutrophils revealed a differentiation continuum from a resting state to an activated state (Figure 6G). The expression of LTB progressively increased along the pseudotime trajectory, correlating with the acquisition of an activated phenotype (Figure 6I). In silico knockout of LTB in neutrophils perturbed a gene network significantly enriched in functions related to inflammation and epigenetic modifications (Figures 6C–F).

Figure 6.

Panel A displays a t-SNE plot of Kinic scores for cellular clustering. Panel B shows a dot plot of LTB gene expression across cell types by percent expressed and average expression. Panel C depicts a horizontal bar chart of the top ten differentially regulated genes after LTB knockout. Panel D presents a volcano plot with Z-score versus negative log ten p-value after knockout of LTB. Panel E features a grouped bar chart highlighting enriched gene ontologies and pathways following LTB knockout. Panel F illustrates a network diagram showing gene interactions of LTB-associated genes. Panel G provides trajectory analyses, including a pseudotime plot and dimension reduction visualization for neutrophils, with additional plots for state and cell type annotation. Panel I has a scatter plot of relative LTB expression over pseudotime for neutrophils.

Open in a new tab

Functional characterization of the KN-associated axis in neutrophils of COPD patients. (A) AUCell scores for the Kinic gene set across major cell types. (B) Feature plot demonstrating specific expression of LTB in the neutrophil cluster. (C–F) Results of the in silico knockout of LTB in neutrophils. (G, I) Plots showing the expression dynamics of LTB along the neutrophil pseudotime trajectory.

3.7. Drug enrichment targeting the KN-associated signature for the treatment of COPD patients

The DrugReflector screening of the COPD signature against the cMAP database identified BRD-K97481123 as the top candidate compound predicted to reverse the disease-associated gene expression profile (Figure 7A). Molecular docking simulations predicted stable and favorable binding interactions between BRD-K97481123 and the LTB protein, with a binding energy of -8.7 kcal/mol, suggesting high affinity and potential inhibitory activity (Figure 7C). In addition, in vitro analysis demonstrated increased expression patterns of LTB in the COPD group compared to the HC group (Figure 7B).

Figure 7.

Panel A shows a scatter plot ranking the top ten predicted drugs for COPD to healthy control (HC) conversion by logit score and probability, with labeled drug IDs and one outlier at higher probability. Panel B is a bar chart displaying significantly elevated LTB expression in COPD compared to normal, indicated by two asterisks and error bars. Panel C presents a protein structure illustration with a zoomed-in view identifying key residues at a binding pocket; a table lists CurPocket IDs, Vina scores, and cavity volumes for five pockets.

Open in a new tab

AI-driven drug screening and clinical validation. (A) Workflow and result of the DrugReflector screening. (B) q-RT-PCR) validation showing significantly elevated LTB mRNA levels in peripheral blood mononuclear cells (PBMCs) from COPD patients (n = 6) compared with healthy controls (HC, n = 6). (C) Molecular docking model depicting the predicted binding pose and interaction between BRD-K97481123 and the LTB protein. ** p < 0.01.

4. Discussion and conclusion

This integrative, AI-augmented multi-omics study potentially deciphers a novel molecular axis linking KN-associated biology in COPD. We identified five KN-associated gene signatures that could potentially guide the molecular stratification and diagnosis of COPD patients. In addition, LTB was identified as the central pathogenic factor among the five KN-associated gene signatures in COPD patients. Furthermore, we nominated a potential therapeutic compound, thereby providing a novel avenue for COPD clinical translation.

LTB is a well-established member of the tumor necrosis factor (TNF) superfamily, primarily recognized for its crucial role in the development and organization of secondary lymphoid organs (Saeki and Yokomizo, 2017). Its signaling through the LTβR receptor is pivotal in structuring immune responses (Subramanian et al., 2017). In the context of COPD, LTB has been implicated in the formation of tertiary lymphoid structures, such as inducible bronchus-associated lymphoid tissue (iBALT), which is a hallmark of advanced COPD and contributes to persistent local inflammation and disease progression (Larsson, 2008). Kinic, a recently characterized form of lysine acylation in which an isonicotinoyl moiety is transferred, has emerged as a novel metabolic sensor (Zhou et al., 2026). Previous studies have identified that LTB can undergo Kinic modification during liver cancer progression; however, the mechanisms underlying Kinic in COPD have not yet been clarified (Li et al., 2024). Moreover, there is a notable lack of evidence demonstrating significant LTB expression within neutrophils themselves in any disease context, let alone in COPD. This gap underscores a potential blind spot in the current understanding of neutrophil biology and their contribution to inflammatory networks.

In conclusion, we constructed a novel KN-associated molecular axis involved in COPD pathogenesis through an AI-driven multi-omics approach. The five-gene KN-associated signature serve as a potential diagnostic tool and reveal patient molecular and immune heterogeneity. In addition, LTB was established as a neutrophil-centric pathogenic factor in COPD. The identification of BRD-K97481123 as a potential LTB-targeting therapy offers a new translational avenue. These findings illuminate previously unrecognized KN-associated mechanisms in COPD, providing fresh insights for biomarker development and targeted therapeutic strategies. However, this study has several limitations. For instance, the direct Kinic mechanism in COPD has not yet been clarified. Future studies should focus on addressing the direct role of Kinic mechanisms in the pathogenesis of COPD using preclinical studies. In addition, although LTB was identified as a protein that can undergo Kinic modification in a previous study, the specific mechanisms of Kinic-modified LTB in neutrophils and the differential expression of LTB should be further investigated in preclinical COPD models. Furthermore, as LTB is a secreted protein, its differential expression in pulmonary tissues and peripheral blood should be evaluated in clinical trials to enhance its corresponding clinical translational potential as a liquid biopsy biomarker for COPD patients. Moreover, the diagnostic model and molecular subgroups of COPD patients identified in our study should be validated in clinical trials to assess patient clinical characteristics and improve the model’s authenticity. Finally, preclinical studies and clinical trials should be performed to evaluate the efficacy and safety of BRD-K97481123 in the treatment of COPD and to clarify its corresponding pharmacological mechanism.

Funding Statement

The author(s) declared that financial support was not received for this work and/or its publication.

Footnotes

Edited by: Tzong-Shyuan Lee, National Taiwan University, Taiwan

Reviewed by: Jose Luis Fachi, Washington University in St. Louis, United States

Mathew John, Jubilee Mission Medical College and Research Institute, India

Data availability statement

The datasets analyzed during the current study are publicly available in the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) under the accession numbers: GSE42057, GSE44962, GSE77344, GSE148871, and GSE167295. The isonicotinylation (Kinic)-related gene signature was obtained from published literature. All code and analysis scripts used in this study are available from the corresponding author upon reasonable request.

Ethics statement

Ethical approval was not required for the studies on humans in accordance with the local legislation and institutional requirements because only commercially available established cell lines were used.

Author contributions

ZW: Conceptualization, Data curation, Formal Analysis, Methodology, Software, Visualization, Writing – original draft. YL: Data curation, Formal Analysis, Validation, Visualization, Writing – review & editing. MZ: Investigation, Methodology, Resources, Validation, Writing – review & editing. MX: Data curation, Formal Analysis, Software, Visualization, Writing – review & editing. JL: Conceptualization, Methodology, Project administration, Supervision, Writing – review & editing.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

  1. Argüello R. J., Combes A. J., Char R., Gigan J. P., Baaziz A. I., Bousiquot E., et al. (2020). SCENITH: a flow cytometry-based method to functionally profile energy metabolism with single-cell resolution. Cell Metab. 32, 1063–1075.e7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Butler A., Hoffman P., Smibert P., Papalexi E., Satija R. (2018). Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420. doi:  10.1038/nbt.4096 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Cai Y., Gu H., Kenney T. (2023). Rank selection for non-negative matrix factorization. Stat. Med. 42, 5676–5693. doi:  10.32614/cran.package.dbnmfrank [DOI] [PubMed] [Google Scholar]
  4. Cao G., Xuan X., Li Y., Hu J., Zhang R., Jin H., et al. (2023). Single-cell RNA sequencing reveals the vascular smooth muscle cell phenotypic landscape in aortic aneurysm. Cell. Communication Signaling: CCS 21, 113. doi:  10.1186/s12964-023-01120-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chen B., Khodadoust M. S., Liu C. L., Newman A. M., Alizadeh A. A. (2018). Profiling tumor infiltrating immune cells with CIBERSORT. Methods Mol. Biol. (Clifton NJ) 1711, 243–259. doi:  10.1007/978-1-4939-7493-1_12 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Chen J., Wang T., Li X., Gao L., Wang K., Cheng M., et al. (2024). DNA of neutrophil extracellular traps promote NF-κB-dependent autoimmunity via cGAS/TLR9 in chronic obstructive pulmonary disease. Signal Transduction Targeted Ther. 9, 163. doi:  10.1038/s41392-024-01881-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chen Y., Lin J., Shi D., Miao Y., Xue F., Liu K., et al. (2025). Identification of WDR74 and TNFRSF12A as biomarkers for early osteoarthritis using machine learning and immunohistochemistry. Front. Immunol. 16, 1517646. doi:  10.3389/fimmu.2025.1517646 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. DeMeo B., Nesbitt C., Miller S. A., Burkhardt D. B., LipChina I., Fu D., et al. (2025). Active learning framework leveraging transcriptomics identifies modulators of disease phenotypes. Sci. (New York NY) 390, eadi8577. doi:  10.1126/science.adi8577 [DOI] [PubMed] [Google Scholar]
  9. Fang Z., Li J., Cao F., Li F. (2022). Integration of scRNA-Seq and bulk RNA-Seq reveals molecular characterization of the immune microenvironment in acute pancreatitis. Biomolecules 13 (1), 78. doi:  10.3390/biom13010078 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Gustavsson E. K., Zhang D., Reynolds R. H., Garcia-Ruiz S., Ryten M. (2022). ggtranscript: an R package for the visualization and interpretation of transcript isoforms using ggplot2. Bioinf. (Oxford England) 38, 3844–3846. doi:  10.1093/bioinformatics/btac409 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Hattab Y., Alhassan S., Balaan M., Lega M., Singh A. C. (2016). Chronic obstructive pulmonary disease. Crit. Care Nurs. Q. 39, 124–130. doi:  10.1097/cnq.0000000000000105 [DOI] [PubMed] [Google Scholar]
  12. Jiang Y., Li Y., Liu C., Zhang L., Lv D., Weng Y., et al. (2021). Isonicotinylation is a histone mark induced by the anti-tuberculosis first-line drug isoniazid. Nat. Commun. 12, 5548. doi:  10.1038/s41467-021-25867-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Jin S., Guerrero-Juarez C. F., Zhang L., Chang I., Ramos R., Kuan C. H., et al. (2021). Inference and analysis of cell-cell communication using CellChat. Nat. Commun. 12, 1088. doi:  10.1038/s41467-021-21246-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kapellos T. S., Baßler K., Fujii W., Nalkurthi C., Schaar A. C., Bonaguro L., et al. (2023). Systemic alterations in neutrophils and their precursors in early-stage chronic obstructive pulmonary disease. Cell Rep. 42, 112525. doi:  10.1016/j.celrep.2023.112525 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Langfelder P., Horvath S. (2008). WGCNA: an R package for weighted correlation network analysis. BMC Bioinf. 9, 559. doi:  10.1186/1471-2105-9-559 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Larsson K. (2008). Inflammatory markers in COPD. Clin. Respir. J. 2, 84–87. doi:  10.1111/j.1752-699x.2008.00089.x [DOI] [PubMed] [Google Scholar]
  17. Li Y., Jiang Y., Yan H., Qin Z., Peng Y., Lv D., et al. (2024). Global isonicotinylome analysis identified SMAD3 isonicotinylation promotes liver cancer cell epithelial-mesenchymal transition and invasion. iScience 27, 110775. doi:  10.1016/j.isci.2024.110775 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lin G., Gao Z., Wu S., Zheng J., Guo X., Zheng X., et al. (2024). scRNA-seq revealed high stemness epithelial Malignant cell clusters and prognostic models of lung adenocarcinoma. Sci. Rep. 14, 3709. doi:  10.1038/s41598-024-54135-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Liu X., Sun P., Bao X., Cao Y., Wang L., Wang Q. (2024). Potential mechanisms of traditional Chinese medicine in treating insomnia: a network pharmacology, GEO validation, and molecular-docking study. Medicine 103, e38052. doi:  10.1097/md.0000000000038052 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Osorio D., Zhong Y., Li G., Xu Q., Yang Y., Tian Y., et al. (2022). scTenifoldKnk: an efficient virtual knockout tool for gene function predictions via single-cell gene regulatory network perturbation. Patterns (New York NY) 3, 100434. doi:  10.1016/j.patter.2022.100434 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Qian Y., Cai C., Sun M., Lv D., Zhao Y. (2023). Analyses of factors associated with acute exacerbations of chronic obstructive pulmonary disease: a review. Int. J. Chronic Obstructive Pulmonary Dis. 18, 2707–2723. doi:  10.2147/copd.s433183 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Ritchie M. E., Phipson B., Wu D., Hu Y., Law C. W., Shi W., et al. (2015). limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 43, e47. doi:  10.1093/nar/gkv007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Ritchie A. I., Wedzicha J. A. (2020). Definition, causes, pathogenesis, and consequences of chronic obstructive pulmonary disease exacerbations. Clinics Chest Med. 41, 421–438. doi:  10.1016/j.ccm.2020.06.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Robin X., Turck N., Hainard A., Tiberti N., Lisacek F., Sanchez J. C., et al. (2011). pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinf. 12, 77. doi:  10.1186/1471-2105-12-77 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Saeki K., Yokomizo T. (2017). Identification, signaling, and functions of LTB(4) receptors. Semin. Immunol. 33, 30–36. doi:  10.1016/j.smim.2017.07.010 [DOI] [PubMed] [Google Scholar]
  26. Subramanian B. C., Majumdar R., Parent C. A. (2017). The role of the LTB(4)-BLT1 axis in chemotactic gradient sensing and directed leukocyte migration. Semin. Immunol. 33, 16–29. doi:  10.1016/j.smim.2017.07.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Wu X., Guo B., Chang X., Yang Y., Liu Q., Liu J., et al. (2024). Identification and validation of glycolysis-related diagnostic signatures in diabetic nephropathy: a study based on integrative machine learning and single-cell sequence. Front. Immunol. 15, 1427626. doi:  10.3389/fimmu.2024.1427626 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Yan Z., Chen B., Yang Y., Yi X., Wei M., Ecklu-Mensah G., et al. (2022). Multi-omics analyses of airway host-microbe interactions in chronic obstructive pulmonary disease identify potential therapeutic interventions. Nat. Microbiol. 7, 1361–1375. doi:  10.1038/s41564-022-01196-8 [DOI] [PubMed] [Google Scholar]
  29. Yu G., Wang L. G., Han Y., He Q. Y. (2012). clusterProfiler: an R package for comparing biological themes among gene clusters. Omics: A. J. Integr. Biol. 16, 284–287. doi:  10.1089/omi.2011.0118 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Zhou J., Jiang Y., Yu M., Wang M., Li Y., Ji D., et al. (2026). Kinic index: an artificial intelligence-driven predictive model and multitarget drug discovery framework for hepatocellular carcinoma patients. NPJ Precis. Oncol. 10 (1). doi:  10.1038/s41698-026-01324-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The datasets analyzed during the current study are publicly available in the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) under the accession numbers: GSE42057, GSE44962, GSE77344, GSE148871, and GSE167295. The isonicotinylation (Kinic)-related gene signature was obtained from published literature. All code and analysis scripts used in this study are available from the corresponding author upon reasonable request.

Articles from Frontiers in Physiology are provided here courtesy of Frontiers Media SA

RESOURCES