Downloading and processing of datasets

Three atherosclerosis datasets (GSE100927, GSE21545 and GSE159677) were downloaded from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) [15]. GSE100927 consisted of 69 AS patients and 35 normal samples. The GSE21545 dataset encompasses gene expression profiles of peripheral blood mononuclear cells (PBMC) from 97 patients with atherosclerosis (AS). Additionally, prognostic data were also sourced for these patients. After accounting for missing values, follow-up data were available for 96 patients, 21 of whom experienced an ischemic event during the follow-up period. The follow-up spanned 1,159 ± 631 days on average, with ischemic events defined as either myocardial infarction or ischemic stroke. Additionally, the GSE159677 dataset focuses on single-cell transcriptome data from atherosclerotic core (AC) plaque and patient-matched proximal-adjacent (PA) tissue. Post-download, the dataset was preprocessed using R and Perl, involving steps such as background calibration and normalization [16]. By reviewing the literature, we identified 69 genes associated with NETosis (NETosis-related genes, NRGs) [17]. We extracted the expression profiles of these genes from the AS dataset and conducted a differential expression analysis using the “limma” package [18]. This analysis revealed 24 differentially expressed NRGs (DE-NRGs), determined under screening conditions of an adjusted p-value < 0.05 and an absolute log2 fold change (FC) ≥ 0.5.

Functional enrichment analysis

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed using the “clusterProfiler” R package to investigate the functional pathways enriched by the differentially expressed NETosis-related genes (DE-NRGs) [19]. Only GO terms and KEGG pathways achieving an adjusted p-value < 0.05 were considered statistically significant, ensuring that the findings highlight the most relevant biological insights associated with DE-NRGs in the context of atherosclerosis.

Screening of hub NETosis- related genes

In this study, two machine learning techniques, random forest (RF) and support vector machine-recursive feature elimination (SVM-RFE), were employed to identify key genes among the DE-NRGs. RF is a robust machine learning algorithm renowned for its high accuracy and minimal requirement for model tuning, making it highly effective across various applications [20]. The “randomForest” package [21] was used to develop RF models for the DE-NRGs. The importance of each gene was assessed using the Gini coefficient method, which measures the decrease in model accuracy when a variable is omitted. Genes with importance values exceeding 2 were selected for further analysis. SVM-RFE models involves constructing models using the “kernlab” package and evaluating them based on their 10-fold cross-validated mean errors. The gene corresponding to the minimum error was identified as the disease signature gene [22]. The final set of key NRGs was determined by intersecting the genes identified by both RF and SVM-RFE methods. This dual-method approach ensures that the genes selected are both impactful and robust, significantly contributing to understanding the genetic drivers of atherosclerosis.

Prognostic analysis of hub NETosis- related genes

To understand the impact of these key NETosis-related genes on the prognosis of AS patients, we conducted a prognostic analysis of these genes in another dataset containing patient follow-up data. We performed univariate and multivariate COX regression on these genes and age. Genes with a P-value < 0.05 in the COX model were considered to have independent prognostic value. Furthermore, for genes significant in both univariate and multivariate COX regression, we further plotted their ROC curves in GSE100927 and GSE21545 to determine if they can accurately distinguish AS patients from healthy individuals and differentiate between AS patients who have and have not experienced ischemic events.

Evaluation of NETosis and PTAFR expression levels in AS animal models

6-week-old male C57BL/6 mice were fed with normal diet (control group) and 6-week-old male APOE-/mice were fed with high-fat diet (AS group) for 5 months. First, we assessed the levels of serum NETs in these two groups, including the levels of serum MPO-DNA and cell-free DNA (cfDNA), as previously described [23]. In addition, assessment of histological changes in the aorta between the two groups. The aorta was processed into paraffin sections for hematoxylin and eosin (HE) staining, immunofluorescence, and immunohistochemistry. All animal experiments in this study were approved by the Institutional Animal Ethics Committee of the Second Affiliated Hospital of Chongqing Medical University and conducted in accordance with the relevant guidelines.

Cellular experiment

HL-60 cells (Procell, CL-0110) were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum. Cells were induced to differentiate into neutrophil-like cells by exposure to 1.25% dimethyl sulfoxide (DMSO) and 25 ng/ml granulocyte colony-stimulating factor (G-CSF) for 6 days [24].Subsequently, the cells were divided into groups including LPS, LPS + siPTAFR, and LPS + BN52021 groups. The LPS group was stimulated with LPS (100 ng/ml) alone for 6 h. The LPS + siPTAFR group used Lipofectamine RNAiMax transfection reagent (Invitrogen, USA), transfecting HL-60 cells with 100 nM siRNA specific for PTAFR (siPTAFR) according to the manufacturer’s protocol, followed by stimulation with LPS for 6 h after 48 h of transfection. The siRNA used in this study is listed in Supplementary table S1. The LPS + BN52021 group pre-treated the cells with the PTAFR inhibitor BN52021 (MedChemExpress, HY-N0784) for 30 min before adding LPS for 6 h of stimulation. After the experiment, cell samples were collected and analyzed using immunofluorescence and Western blotting.

Immune infiltration and gene enrichment analysis

To further elucidate the immune landscape of atherosclerosis (AS), this study employed single-sample gene set enrichment analysis (ssGSEA) to assess and compare immune infiltration in AS patients versus normal samples, aiming to delineate the unique immune microenvironmental characteristics of AS patients. Additionally, we explored the relationship between key atherosclerosis-related genes (ARGs) and immune cell types by calculating correlations using the Spearman method. Our analysis particularly highlighted the prognosis-related gene PTAFR. To visually represent the association between PTAFR and immune cells, lollipop diagrams were constructed. These diagrams effectively illustrate how variations in PTAFR correlate with different types of immune cells within the context of AS. Further, we conducted Gene Ontology (GO) analysis to identify biological pathways potentially regulated by PTAFR. For this purpose, the reference gene list “c5.go.symbols.gmt” was obtained from the Molecular Signature Database (MsigDB). Using this gene list, both Gene Set Variation Analysis (GSVA) and Gene Set Enrichment Analysis (GSEA) were performed to pinpoint pathways significantly enriched in relation to PTAFR activity. Pathways were considered significantly enriched based on a p-value threshold of < 0.05, helping to ensure the biological relevance of the findings. This integrative approach provides a comprehensive view of how PTAFR may influence the immune dynamics and pathological processes in atherosclerosis.

Single-cell analysis

Single-cell analysis was used to validate and evaluate PTAFR expression at the single-cell level. Quality control, dimensionality reduction and clustering of AS scRNA-seq data were performed using “Seurat” package, as described in previous studies [25]. “singleR” R package was used to annotate the clusters, and then “CellMarker” was used for manual correction [26]. Based on the identified single-cell clusters, we further evaluated the expression of PTAFR in each cell subset.

Flow cytometry

Following the protocol by Elena et al., atherosclerotic plaques were first cleaned, then minced into small fragments, and digested overnight at 37 °C with 250 units/ml of Type I collagenase in 5 ml of culture medium containing 10% fetal bovine serum. The resulting cell suspension was filtered through a 150-mesh nylon net, washed three times in PBS, and analyzed using the CytoFLEX flow cytometer (Beckman, USA). We used Ly6g to label neutrophils, Mac2 to label macrophages, and also labeled PTAFR (Immunway, YT3566). Data were analyzed using the Listmode Analysis feature of the Epics XL software.

Magnetic bead separation of neutrophils

Suspend up to 10^8 cells in MACS buffer (PBS + 0.5% BSA + 2 mM EDTA) as described above. Add anti-Ly6g biotin beads (Miltenyi, 130-092-332) to each sample, incubate at 4 °C, then add MACS buffer and anti-biotin microbeads (Miltenyi, 130-092-332), incubate at 4 °C again. Wash the cells with MACS buffer, centrifuge, discard the supernatant, and resuspend in MACS buffer. Add the cell suspension to the LS column (Miltenyi, 130-042-401) placed on a magnet, wash the column three times with MACS buffer, and collect the negative cell fraction that flows through. Take the column off the magnet and place it into a new centrifuge tube. Wash the column with MACS buffer, then use the plunger to flush out the cells. Centrifuge, resuspend the cells in PBS, and complete the separation. Ly6g immunofluorescence staining was used to determine the purity of neutrophils (almost 90% of the cells were positive for Ly6g+).

Immunofluorescence and immunohistochemistry

Immunofluorescence was used with anti-MPO (AF3667, R&D) and anti-cit-H3 (ab5103, Abcam) double-label staining to assess the level of tissue NETosis [27], immunohistochemistry for assessing the expression level of PTAFR (ab104162, Abcam) in tissues. To validate single-cell analysis results, aortic plaques were stained with triple immunofluorescence for neutrophils (Ly6g, BioLegend, 127614), macrophages (Mac2, BioLegend, 125401), and PTAFR (GeneTex, GTX105822). Furthermore, following LPS stimulation, the three groups were also subjected to dual immunofluorescence staining for MPO and citH3, and the staining results were quantitatively analyzed using ImageJ.

Western blot

The WB analysis method involved homogenizing arterial tissue, separating it on a 10% SDS-PAGE gel, transferring it to a polyvinylidene fluoride membrane, and detecting with antibodies for PTAFR, MPO, and citH3 diluted in 5% bovine serum albumin.

Statistical analysis

All statistical analyses were performed using R software (4.0.1) and GraphPad Prism 9 (GraphPad Software Inc., San Diego, CA). Wilcoxon test was used to compare the differences between the two groups. One-way ANOVA was utilized for comparisons among multiple groups. P-value < 0.05 was considered statistically significant.