Abstract

   [32]graphic file with name ao4c04307_0010.jpg

   In this study, we conducted a comprehensive analysis to identify key
   genes and pathways associated with pulmonary arterial hypertension
   (PAH) and investigated the role of delta-like ligand 4 (DLL4) in PAH
   pathogenesis. Through integrated analysis of multiple data sets, we
   identified 6 candidate differentially expressed genes (DEGs), notably
   DLL4, which showed the highest distinguishing efficiency between PAH
   and control samples. Functional and pathway enrichment analyses
   revealed the involvement of DLL4 in critical biological processes and
   pathways related to PAH, including notch signaling, immune cell
   function, and inflammatory responses. Further investigation
   demonstrated that decreased DLL4 expression correlated with increased
   M2 macrophage polarization, suggesting a potential role for DLL4 in
   preventing M2 differentiation. Additionally, the DLL4/Notch1 axis was
   found to influence the Notch profile and regulate signaling mediators
   during M2 differentiation. These findings highlight DLL4 as a promising
   biomarker and therapeutic target for PAH, shedding light on the
   underlying molecular mechanisms and providing insights for the
   development of novel treatment strategies.

Introduction

   Pulmonary arterial hypertension (PAH) is a severe cardiovascular
   condition characterized by increased pulmonary arterial pressure, often
   leading to right heart failure and mortality.^[33]1 Despite
   advancements in symptomatic management, PAH remains incurable,
   emphasizing the urgent need for a deeper understanding of its
   pathogenesis to develop effective therapies.

   PAH is marked by irreversible pulmonary vascular remodeling, involving
   endothelial cells, smooth muscle cells, and fibroblasts.^[34]2 Notably,
   perivascular inflammation, characterized by the presence of various
   immune cells, plays a crucial role in this process.^[35]2 Specific
   macrophage subtypes, particularly M1 and M2, are implicated in PAH
   progression.^[36]3 M1 macrophages exacerbate inflammation, while M2
   macrophages contribute to tissue repair and vascular remodeling.
   Concurrently, there is an increase in pro-inflammatory cytokines such
   as IL-1β, IL-6, IL-10, TGF-β, and TNF-α.^[37]4,[38]5

   Notch signaling has emerged as a key regulator of macrophage
   polarization.^[39]6 DLL4, a Notch ligand predominantly expressed in
   endothelial and myeloid cells, plays a crucial role in
   endothelial–macrophage interactions during angiogenesis.^[40]7 Notch
   activation in monocytes, facilitated by endothelial DLL4, promotes the
   proinflammatory M1 phenotype.^[41]8 Recent research highlights the
   pivotal role of DLL4/Notch signaling in macrophage polarization. DLL4
   activates Notch receptors on the macrophages, influencing their
   polarization toward either pro-inflammatory M1 or anti-inflammatory M2
   phenotypes, crucial for immune responses and tissue
   homeostasis.^[42]9,[43]10 This pathway regulates cytokine expression,
   phagocytic activity, and metabolic reprogramming, offering promising
   avenues for immunomodulatory therapies.^[44]11 Currently, research on
   the role of DLL4/Notch signaling in PAH and its molecular mechanisms in
   macrophage polarization remain limited. There is a notable gap in
   understanding how this pathway contributes to macrophage polarization
   and its effect on PAH pathogenesis. Further investigation is needed to
   elucidate the specific molecular interactions and signaling cascades
   involved, which could provide insights into potential therapeutic
   strategies targeting immune dysregulation in PAH.

   Our study explores the impact of DLL4 on macrophage differentiation,
   particularly its potential to disrupt M2 polarization in PAH. We
   propose that DLL4 orchestrates a Notch-dependent selection process,
   driving macrophage differentiation toward M1 while inhibiting M2
   differentiation, thus contributing to PAH pathogenesis. This study
   sheds light on the intricate interplay among Notch signaling,
   inflammation, and macrophage polarization in PAH, offering insights
   into potential therapeutic targets for the disease.

Methods and Materials

Microarray Data and Processing

   Microarray data sets, including [45]GSE15197, [46]GSE53408,
   [47]GSE113439, [48]GSE130391, and [49]GSE24988, were acquired from the
   GEO database. [50]GSE15197, [51]GSE53408, [52]GSE113439, and
   [53]GSE130391 were merged for training, while [54]GSE24988 served as
   the test set. Probe names in the data sets were converted to gene names
   using R (version 4.1.0). After integration, the training data set
   comprised 39 control and 71 PAH samples. Data underwent log 2
   transformation for large values and averaging for repeated probes.
   Batch effects were corrected using the “sva” package. Principal
   component analysis (PCA) plots were generated using the ggplot2 package
   to visualize the batch effect correction.

Screening for Candidate Differentially Expressed Genes (DEGs)

   The “limma” package was utilized to identify DEGs from the integrated
   data of [55]GSE15197, [56]GSE53408, [57]GSE113439, and [58]GSE130391.
   DEGs were screened based on criteria of |log 2 FC| > 1 and adjusted-p <
   0.05. Heatmaps and volcano plots of DEGs were generated using the
   “pheatmap” and “ggplot2” packages, respectively. Machine learning
   analysis was conducted on DEGs to identify feature genes for PAH. A
   LASSO model was constructed using the “glmnet” package with parameter
   tuning via 10-fold cross-validation, setting the response type as
   binomial and α as 1. Feature genes were selected based on minimum
   error. The “randomForest” package was employed to establish an RF
   model, determining the number of trees with the minimum error.
   Candidate DEGs were further screened using a Venn diagram generated
   with the “ggvenn” R package.

Verification of the Identified Candidate DEGs

   Violin plots were created using the geom-violin function from the
   ggplot2 R package to visualize expression patterns of identified
   candidate DEGs between PAH and the control samples. The “pROC” package
   was employed to generate receiver operating characteristic (ROC) curves
   for validating the DEG accuracy. Performance metrics, such as
   classification sensitivity, specificity, and area under the curve (AUC)
   were assessed. Additionally, the expression of biomarker DLL4 was
   validated in the external testing cohort, [59]GSE24988.

Functional Analyses

   To explore the functions regulated by DLL4 in PAH, PAH samples were
   divided into two subgroups based on DLL4 expression levels: DLL4-high
   and DLL4-low. The “limma” package was used to identify genes
   differentially expressed between these subgroups, with criteria set at
   |log 2 FC| > 1 and adjusted-p < 0.05. Gene ontology (GO) and Kyoto
   encyclopedia of genes and genomes (KEGG) analyses were conducted using
   the “clusterProfiler” R package to investigate potential molecular
   mechanisms of these genes in PAH. Single-gene gene set enrichment
   analysis (GSEA) was performed using the “clusterProfiler” and
   “enrichplot” packages, classifying samples into high and low expression
   groups based on median expression values. Gene set variation analysis
   (GSVA) was employed with the “c2.cp.kegg.Hs.symbols” subset from the
   MSigDB database to examine pathway variations. Pathway and biological
   function differences between subtypes were analyzed using the “limma”
   and “GSVA” packages.

Infiltration of Immune Cells and Correlation Analysis

   The “CIBERSORT” algorithm was used to estimate immune cell infiltration
   ratios in both the PAH and control samples. Default signature matrix
   permutations were set to 1000, and the standard immune cell expression
   file (LM22.txt) was obtained from the official website. Differences in
   immune cell proportions between the two groups were assessed by using
   the Wilcoxon rank sum test. Spearman correlation analysis was conducted
   to explore the relationships between candidate genes and infiltrating
   immune cells.

PBMC Collection and Small-Interfering RNA (siRNA) Transfection

   Peripheral blood mononuclear cells (PBMCs) were collected from both
   non-PAH and PAH volunteers using EDTA vacutainer tubes. PBMCs were
   isolated using the histopaque density gradient method and a Sepmate
   Tube. Monocytic cells were obtained by seeding PBMCs in 6-well or
   24-well plates at a density of 3 × 10^6 cells/well and culturing them
   in RPMI-1640 medium supplemented with glutamine, sodium pyruvate,
   HEPES, Normocin, and penicillin–streptomycin. The cells were then
   incubated at 37 °C with 5% CO[2] for 3 h. After removing nonadherent
   cells, adherent monocytes were washed with serum-free culture medium
   and further incubated for 24 h in RPMI with 2% fetal bovine serum.

   PBMCs were cultured as previously described. Following culture, the
   cells were transfected with siRNAs using linear polyethylenimines
   (PEIs) Reagent according to the manufacturer’s instructions. The siRNAs
   targeted DLL4 or were scrambled (control) siRNA. The efficiency of gene
   knockdown was evaluated using QPCR and Western blot.

Differentiation and Polarization Protocol

   Purified PBMCs were cultured in RPMI1640 medium supplemented with 20%
   fetal bovine serum (FBS), glutamine, penicillin–streptomycin,
   nonessential amino acids, sodium pyruvate, and M-CSF (100 ng/mL) for 7
   days on 9 cm^2 dishes (1.107 cells/cm2) to induce differentiation into
   M0 macrophages. For M1/M2 polarization, M0 macrophages (2.5 × 10^5
   cells/cm^2) were further cultured for an additional 48 h in RPMI
   supplemented with 10% FBS and either IFN-γ plus LPS for M1
   differentiation or with IL-4 for M2a differentiation, or IL-10 for M2c
   differentiation.

Statistical Analysis

   Experiments were performed in triplicate to ensure reproducibility.
   Data are expressed as mean ± standard deviation (SD) and analyzed using
   nonparametric Mann–Whitney test and Kruskal–Wallis test (with Dunn’s
   multiple comparison post-test) for comparisons involving multiple
   conditions. A p < 0.05 was considered statistically significant. In
   figures, (*) indicates p < 0.05, (**) indicates p < 0.01, and (***)
   indicates p < 0.005.

Results

The Candidate DEGs Between PAH and Control Groups in the Training Data Sets

   The [60]GSE15197, [61]GSE53408, [62]GSE113439, and [63]GSE130391 data
   sets were integrated and underwent batch calibration to mitigate batch
   effects. Principal component analysis (PCA) was employed to assess the
   effectiveness of this calibration ([64]Figure [65]1
   A,B). Subsequently, to identify differentially expressed genes (DEGs)
   between PAH and control samples, a Bayesian test was conducted on the
   training data set, yielding a total of 17 DEGs, comprising 2
   upregulated and 15 downregulated genes ([66]Figure [67]1C,D). Machine
   learning analysis was then carried out on these 17 DEGs using R
   software. LASSO feature selection identified 8 genes ([68]Figure
   [69]1E), while 8 genes with importance values exceeding 2 were chosen
   as disease-specific genes in the Random Forest (RF) model ([70]Figure
   [71]1F,G). Finally, Venn analysis yielded 6 candidate DEGs ([72]Figure
   [73]1H).

Figure 1.

   [74]Figure 1
   [75]Open in a new tab

   Candidate DEGs between PAH and control group in the training data sets.
   (A) PCA diagram of [76]GSE15197, [77]GSE53408, [78]GSE113439, and
   [79]GSE130391 data sets before calibration. (B) PCA diagram of these
   training data sets after calibration. (C) Heatmap of DEGs. The red and
   blue areas represent the significantly up- and down-regulated DEGs,
   respectively. (D) Volcano plot of DEGs. These genes consist of 2
   upregulated genes and 15 downregulated genes. (E) Identification of the
   optimal penalization coefficient lambda (λ) in the LASSO model. (F) The
   influence of the number of decision trees on the error rate. The x-axis
   represents the number of decision trees, and the y-axis indicates the
   error rate. (G) Results of the Gini coefficient method in the RF model.
   The x-axis indicates the genetic variable, and the y-axis represents
   the importance index. (H) Venn plot exhibiting reliable biomarkers
   between LASSO and RF.

Verification of the Identified Candidate DEGs

   Compared with control samples, decreased S100A3, DLL4, MAL, SOSTDC1,
   FLJ34503, and CX3CR1 expression were observed in the PAH samples from
   the integrative training cohort ([80]Figure [81]2
   A–F). To estimate the predictive utility, the ROC curve was performed
   and it was found that the S100A3, DLL4, MAL, SOSTDC1, FLJ34503, and
   CX3CR1 illustrated a remarkably distinguishing efficiency with AUC
   values of 0.889 (95% CI: 0.820–0.948), 0.893 (95% CI: 0.824–0.945),
   0.869 (95% CI: 0.801–0.930), 0.850 (95% CI: 0.766–0.921), 0.844 (95%
   CI: 0.751–0.919), and 0.803 (95% CI: 0.711–0.888) in the integrative
   training cohort, respectively ([82]Figure [83]3A–F). It was obvious
   that the DLL4 gene showed the highest distinguishing efficiency in the
   6 candidate DEGs. Line chart in [84]Figure [85]3G showed the expression
   levels of the 6 identified genes in the training set. Interestingly,
   the results were validated in the [86]GSE24988 cohort, and a consistent
   gene expression pattern of DLL4 was obtained ([87]Figure [88]3H).
   Hence, we chose DLL4 as the biomarker for PAH patients.

Figure 2.

   [89]Figure 2
   [90]Open in a new tab

   Expression of the identified candidate DEGs between PAH and the control
   group. (A–F) Violin chart for the differential expression analysis of
   the 6 candidate DEGs in the integrative training cohort. The blue
   indicates the control group, and the red represents the PAH group. p <
   0.05 was considered statistically significant.

Figure 3.

   [91]Figure 3
   [92]Open in a new tab

   Verification of the identified candidate DEGs. (A–F) ROC curves for
   evaluating the diagnostic ability of the 6 candidate DEGs in the
   integrative training cohort. (G) Line chart for the expression of
   candidate DEGs in the integrative training cohort. Each point
   represented a sample. (H) Violin chart for the differential expression
   analysis of DLL4 in the [93]GSE24988 testing cohort.

Function and Pathway Enrichment Analysis on PAH Based on DLL4

   Genes exhibiting differential expression between the DLL4-high and
   DLL4-low subgroups were subjected to functional pathway analysis. This
   analysis revealed involvement in various biological processes,
   including dorsal aorta development, regulation of opsonization,
   complement activation of the lectin pathway, notch signaling in heart
   development, and ventricular trabecula myocardium morphogenesis.
   Regarding cellular components, the focus was on peptidase complexes.
   Molecular function analysis highlighted genes associated with notch
   binding, antigen binding, and carbohydrate binding ([94]Figure [95]4
   A). Additionally, KEGG pathway enrichment identified the notch
   signaling pathway, Th1 and Th2 cell differentiation, endocrine
   resistance, and chemical carcinogenesis-receptor activation, among
   others ([96]Figure [97]4B).

Figure 4.

   [98]Figure 4
   [99]Open in a new tab

   Function enrichment analysis. (A) Histogram exhibiting the results of
   the GO pathway analysis. (B) Histogram exhibiting the results of the
   KEGG pathway analysis. (C,D) Enrichment plots from GSEA analysis in the
   DLL4-high subgroup (C) and DLL4-low subgroup (D). (E) Histogram
   exhibiting the results of the GSVA analysis.

   Further single-gene GSEA analysis delineated enrichment of cell
   adhesion molecules, cytokine receptor interaction, natural killer
   cell-mediated cytotoxicity, neuroactive ligand receptor interaction,
   and the notch signaling pathway in the DLL4-high subgroup ([100]Figure
   [101]4
   C). Conversely, apoptosis, complement, and coagulation cascades, the
   mTOR signaling pathway, the phosphatidylinositol signaling system, and
   the spliceosome were enriched in the DLL4-low subgroup ([102]Figure
   [103]4D). Moreover, GSVA enrichment analysis implicated the mTOR
   signaling pathway, basal transcription factors, cell adhesion
   molecules, natural killer cell-mediated cytotoxicity, and notch
   signaling pathway in the pathogenesis of DLL4-associated PAH
   ([104]Figure [105]4E).

Estimation of Infiltrating Immune Cells and Correlation Analysis

   We began by estimating the proportions of 22 infiltrating immune cells
   using the “CIBERSORT” algorithm on the gene matrix of 28 samples.
   Compared to control samples, PAH samples exhibited significantly lower
   proportions of CD8 T cells, CD4 memory activated T cells, and M2
   macrophages, while proportions of resting dendritic cells, activated
   dendritic cells, eosinophils, and neutrophils were significantly higher
   ([106]Figure [107]5
   A). The composition of infiltrating immune cells in PAH and control
   samples is depicted in [108]Figure [109]5B. Pearson’s correlation
   coefficient analysis revealed strong correlations between different
   types of infiltrating immune cells ([110]Figure [111]5C), emphasizing
   the importance of immune status in PAH.

Figure 5.

   Figure 5
   [112]Open in a new tab

   Infiltration of immune cells and correlation analysis. (A) Boxplots
   comparing the proportions of 22 major immune cell subsets between PAH
   and control samples. (B) The histogram showed the distribution
   characteristics of immune cells in PAH. (C) Pearson’s correlation
   coefficient revealed astrong link between each type of infiltrating
   immune cells. (D) Boxplots comparing the immune-related function
   between the DLL4-high subgroup and DLL4-low subgroup.

   To delve deeper into the immune–DLL4 gene relationship in PAH, we
   analyzed the immune functions between two PAH subgroups stratified by
   DLL4 expression. This analysis identified 10 significant immune
   functions, including APC coinhibition, type II IFN response, T cell
   coinhibition, pCDs, inflammation promoting, CCR, cytolytic activity,
   macrophages, Th1 cells, and Tregs ([113]Figure [114]5
   D). [115]Figure [116]6A–F further confirmed positive correlations
   between DLL4 expression levels and CD8 T cells ([117]Figure [118]6A),
   M2 macrophages ([119]Figure [120]6C), and resting NK cells ([121]Figure
   [122]6E), while negative correlations were observed with plasma cells
   ([123]Figure [124]6B) and M0 macrophages ([125]Figure [126]6D) in PAH.

Figure 6.

   [127]Figure 6
   [128]Open in a new tab

   Correlation analysis of DLL4 and immune cells. (A–E) Spearman
   correlation analyses between DLL4 expression levels in the PAH samples
   and the infiltrations of five kinds of immune cells, including CD8 T
   cells (A), plasma cells (B), M2 macrophages (C), M0 macrophages (D),
   and resting NK cells (E). (F) Immune infiltrating cells associated with
   DLL4 in PAH.

DLL4 May Drive the Notch-Dependent Process and Prevent M2 Macrophage
Differentiation in PAH

   We collected the peripheral blood of 47 patients diagnosed with PAH and
   46 healthy controls from Jiangsu Provincial Hospital of Traditional
   Chinese Medicine from August 2021 to July 2023. In order to further
   clarify the expression of DLL4 in PAH patients and its effect on the
   notch pathway in M2 macrophages, we detected the RNA expression levels
   of DLL4, notch pathway, and the molecular markers on the surface of M2
   macrophages in peripheral blood mononuclear cells of PAH and control
   groups. We observed significantly lower expression levels of DLL4 (**p
   = 0.0076) and Notch1 (**p = 0.0087) in the PAH group compared to those
   in the healthy control group ([129]Figure [130]7
   A and [131]Figure [132]7B). Furthermore, our investigation included the
   assessment of molecular markers, such as CD40, CD64, CD80, and CD86 on
   the surface of M1 macrophages, as well as CD11b (**p = 0.0065), CD200R
   (**p = 0.0089), CD163 (*p = 0.0223,) and CD206 (*p = 0.0342) on the
   surface of M2 macrophages. Our findings revealed a notably higher
   presence of M2 macrophages in the PAH group compared with the control
   group ([133]Figure [134]7C).

Figure 7.

   Figure 7
   [135]Open in a new tab

   DLL4 may drive the Notch-dependent process and prevent M2 macrophage
   differentiation in PAH. (A) and (B) DDL4 and Notch expression levels in
   PAH patients (n = 47) and healthy controls (n = 46). (C) The expression
   levels of the molecular markers (CD11b, CD200R, and CD206) of M2
   macrophages in PAH and healthy controls.

Decreased DLL4 Facilitates IL-4-Driven Differentiation into M2 Macrophages

   To explore the interaction between DLL4 and the macrophage phenotype,
   monocytes from healthy volunteers were isolated and differentiated into
   M0 macrophages using M-CSF for 7 days. Subsequently, polarization into
   M1 macrophages with IFN-γ and LPS, or into M2 macrophages (specifically
   M2a subtype) using IL-4 for 48 h, was conducted as previously outlined.
   The phenotype of M1 and M2 macrophages at 24 h postdifferentiation,
   compared to M0 macrophages harvested at the end of M-CSF treatment, is
   depicted in [136]Figure [137]8
   A.

Figure 8.

   [138]Figure 8
   [139]Open in a new tab

   Reduced DLL4 promotes IL-4-mediated differentiation into the M2
   macrophage. (A) Phenotype analysis of macrophages before (M0) and after
   polarization in M1 or M2 macrophages for 24 h with LPS and IFN-γ or
   IL-4, respectively. (B) The mRNA and protein expression levels of DLL4
   upon IL-4 stimulation following DLL4 knockdown. (C) The mRNA and
   protein expression of CD11b, CD206, and CD200R upon IL-4 stimulation
   following DLL4 knockdown. (D) The mRNA and protein expression of Notch1
   upon IL-4 stimulation followed DLL4 knockdown.

   Under these conditions, polarized M1 macrophages exhibited significant
   phenotypic changes compared to M0, including notably increased
   expression of CD40 and CD86, along with a trend toward elevated CD80.
   Conversely, LPS/IFN-γ treatment induced no changes in M2 markers.
   Polarization in the presence of IL-4 resulted in increased expression
   of three of four tested M2-specific markers (CD11b, CD200R, and CD206),
   with no impact on the M1 markers.

   To investigate the impact of DLL4 on macrophage polarization, we
   evaluated the expression of surface molecular markers on each
   macrophage subtype after DLL4 knockdown. As shown in [140]Figure [141]8
   B, DLL4 was knocked down successfully (**p = 0.0017). We observed a
   significant increase in the expression of CD11b (**p = 0.0085), CD206
   (**p = 0.0039), and CD200R (**p = 0.0057) upon IL-4 stimulation
   following DLL4 knockdown (**p = 0.0035, [142]Figure [143]8C) and a
   significant decrease in the expression of Notch1 (**p = 0.0031,
   [144]Figure [145]8D). Importantly, DLL4 knockdown did not alter the
   basal expression of CD11b, CD206, or CD200R on M0 or M1 macrophages.
   Furthermore, manipulating DLL4 expression during M1 polarization did
   not induce significant changes in M1-specific markers (CD40, CD64,
   CD80, and CD86).

   These results suggest an inhibitory function of DLL4 and Notch
   signaling in the IL-4 signaling pathway. In summary, our data imply
   that Notch signaling mediated by DLL4 acts as a regulatory mechanism in
   macrophage differentiation, crucially preventing M2 polarization.

The DLL4/Notch1 Axis Impacts Notch Profile and Modulates Signaling Mediators
During M2 Differentiation

   Quantitative RT-PCR and Western blot were employed for gene expression
   analysis to elucidate the regulation mediated through DLL4/Notch in M2
   macrophages. Macrophages polarized with IL-4 (M2) with or without DLL4
   knockdown were analyzed. Thirteen transcripts encoding Notch receptors,
   ligands, effectors, and signaling molecules were detected. Our findings
   indicate that DLL4 selectively influences RNA transcription and protein
   expression during M2 differentiation. Knockdown of DLL4 during M2
   polarization resulted in decreased mRNA levels for Notch1 (**p =
   0.0065), Notch3 (**p = 0.0077), and HES1 (**p = 0.0089) ([146]Figure
   [147]9
   A). Further analyses showed that the protein levels of Notch1, Notch3,
   and HES1 were also decreased when the DLL4 was knocked down
   ([148]Figure [149]9A). Hence, DLL4 appears to selectively activate the
   Notch pathway in macrophage subsets.

Figure 9.

   [150]Figure 9
   [151]Open in a new tab

   DLL4/Notch1 axis influences the Notch profile and regulates signaling
   mediators upon M2 differentiation. (A) Upon M2 polarization
   knocked-down DLL4 decreased mRNA and protein levels for Notch1, Notch3,
   and HES. (B) The levels of inflammatory factors (IL-10 and TGF-β)
   released by M2 cell. (C) The cell proliferation level increased after
   knocking down DLL4 detected by the CCK-8 kit.

   Considering that Notch signaling pathway was inhibited after DLL4 was
   knocked down, promoting the polarization of M2 cells, we further
   examined the changes in the levels of inflammatory factors released by
   M2 cells and found that the concentrations of IL-10 (**p = 0.0087) and
   TGF-β (**p = 0.0066) were significantly increased ([152]Figure [153]9
   B). DLL4/Notch signaling pathway affects the polarization of M2 and the
   release of downstream inflammatory factors. We detected by the CCK8 kit
   that the cell proliferation level increased after knocking down DLL4
   (**p = 0.0087, [154]Figure [155]9C).

Discussion

   Recent studies indicate that PAH is closely associated with pulmonary
   vascular remodeling^[156]12 and immune cell^[157]13 involvement. Immune
   cells, such as macrophages contribute to vascular remodeling by
   promoting inflammation and releasing growth factors that stimulate
   smooth muscle cell proliferation.^[158]14

   The purpose of this study is to identify potential key genes and
   underlying mechanisms of PAH using various bioinformatics approaches.
   Five data sets derived from PAH patients were obtained from GEO, with
   one data set designated as a validation set. Utilizing multiple
   bioinformatics techniques, including DAVID, STRING, R language, and
   Cytoscape, we investigated the DEGs between PAH patients and healthy
   controls and conducted GO annotation and KEGG enrichment analysis.
   Compared with existing studies, our bioinformatics analysis results are
   more reflective of the actual environment. Furthermore, our immune
   infiltration analysis revealed that macrophages constitute the largest
   proportion of immune cells, differing from some previous reports that
   highlighted monocytes,^[159]15 and aligning with certain other
   studies.^[160]16,[161]17 Preliminary identification of candidate DEGs
   between the PAH and control groups revealed 17 DEGs, with DLL4 emerging
   as a promising biomarker due to its consistently reduced expression in
   PAH samples across multiple data sets. Subsequent validation of these
   candidate DEGs confirmed the downregulation of DLL4 in PAH, consistent
   with sporadic reports in the existing literature.^[162]18 Studies have
   indicated that restoring DLL4/Notch1/PPARγ signaling may be beneficial
   in preventing or reversing the pathological vascular remodeling in PAH,
   offering new hope for PAH treatment.^[163]19

   Through further KEGG and immune cell infiltration analyses, we found
   that DLL4 is associated with the downstream Notch signaling pathway and
   macrophages. While other studies have shown that DLL4/Notch signaling
   promotes pro-inflammatory macrophage activation and accelerates
   atherosclerosis in chronic kidney disease,^[164]10,[165]20 some reports
   indicate that DLL4 inhibitors can alleviate chronic atherosclerosis,
   venous graft disease, vascular calcification, insulin resistance, and
   fatty liver in mice.^[166]21 However, there are few studies
   investigating the mechanisms by which DLL4 and Notch signaling affects
   macrophage polarization in PAH.

   The investigation into infiltrating immune cells in PAH revealed
   significant alterations in immune cell composition with implications
   for disease pathogenesis. The observed decrease in CD8 T cells, CD4
   memory activated T cells, and M2 macrophages, coupled with increased
   proportions of dendritic cells, eosinophils, and neutrophils, suggests
   a dysregulated immune response in PAH.^[167]22,[168]23 Importantly,
   correlation analysis revealed associations between DLL4 expression and
   immune cell subsets,^[169]24−[170]26 highlighting the potential
   interplay between DLL4 and immune dysregulation in PAH.

   Further characterization of DLL4’s role in PAH pathogenesis elucidated
   its impact on macrophage polarization and inflammatory signaling.
   Reduced DLL4 expression was associated with enhanced M2 macrophage
   differentiation, which implicated DLL4 as a negative regulator of M2
   polarization. Mechanistically, DLL4-mediated Notch signaling was shown
   to modulate the expression of key inflammatory mediators,^[171]27
   thereby influencing macrophage phenotype and function in PAH.

   The study results also suggest that targeting the DLL4/Notch axis in
   PAH holds a potential therapeutic significance. Notably, manipulating
   the expression of DLL4 influences macrophage polarization and leads to
   significant changes in the release of inflammatory factors and cell
   proliferation, which are closely related to the mechanisms of vascular
   remodeling in PAH.^[172]23,[173]28,[174]29 This underscores the
   therapeutic potential of modulating DLL4-mediated signaling pathways in
   the treatment of PAH.

   In conclusion, this study provides comprehensive insights into the
   molecular mechanisms underlying PAH pathogenesis with a specific focus
   on the role of DLL4 in immune dysregulation and macrophage
   polarization. The findings underscore the clinical relevance of DLL4 as
   a diagnostic biomarker and potential therapeutic target for PAH,
   offering promising avenues for further research and therapeutic
   intervention in this devastating disease.

Acknowledgments