The role of dynamic interplay exists between sympathetic and parasympathetic nerve in gastric tumorigenesis

Recent research underscores the significance of the nervous system in tumor progression. Nonetheless, understanding the intricate roles of nerves in gastric tumorigenesis remains limited. Our study aims to elucidate nerve alterations, associated key genes, and signaling pathways during gastric tumorigenesis. We enroll 257 patients with gastric precancerous lesions (GPLs) from four national cohorts. Utilizing immunohistochemistry and digital analysis, we quantified the densities of sympathetic and parasympathetic nerves, as well as Schwann cells in patient slides. We collected three mRNA expression profiles of GPL samples from the gene expression omnibus database and assessed them for nerve expression signatures, differentially expressed genes, enriched pathways, and the immune microenvironment. Parasympathetic nerve and Schwann cell densities display a progressive increase, while sympathetic nerve exhibits an inverse trend along the precancerous spectrums. Notably, CD8+ T cells, natural killer cells, and antigen‐presenting cells (APC) co‐inhibition decrease, while epithelial‐mesenchymal transition increases significantly from early to late premalignancy. Several key nerve‐regulating genes (NTRK3, MYBL2, NRP1, BCL2, CCND1, VEGFA, PLXNA2, ADRB2), and pathways (mitogen‐activated protein kinase [MAPK]/ERK, PI3K/AKT, Wnt) emerge as potential contributors to precancerous progression. Our study reveals a dynamic relationship between sympathetic and parasympathetic nerves, characterized by a gradual increase in parasympathetic nerve and Schwann cell density, contrasting with an inverse trend in sympathetic nerve across precancerous spectrums. Targeting the interaction between tumor cells and nerves, including sympathetic and parasympathetic nerves, has emerged as a promising strategy for the prevention and treatment of gastric cancer.


| INTRODUCTION
Similar to various other solid tumors, human gastric tumorigenesis is a complex and multifactorial process, regulated by the tumor microenvironment (TME). 1,2The TME is an intricate system involving various cell types, such as tumor cells, immune cells, fibroblasts, and vascular endothelial cells.Interactions among these components play a crucial role in tumor initiation, growth, invasion, metastasis, and therapeutic response. 3,4hile immune, endothelial, and epithelial cells have been extensively studied in cancer research and have led to significant advancements in targeted therapies, 5 the nervous system has only recently been recognized as a crucial regulator of cancer progression.The interaction between nerves and cancer, referred to as nerve-cancer crosstalk, has garnered significant attention in the field of cancer neuroscience.
The nervous system plays a pivotal role in regulating tumor growth and metastasis within the TME by releasing various neurotransmitters that stimulate receptors in cancer and stromal cells. 6Additionally, tumor cells can also alter neurological functions and exploit nerves to promote tumor progression. 6,7][10][11][12] However, the involvement of the nervous system in gastrointestinal carcinogenesis has only recently been studied. 13Although perineural invasion has been widely recognized as a prognostic risk factor for several cancers, including gastric cancer (GC), [14][15][16] the distribution of nerve types and density during the gastric tumorigenesis process remains largely unknown.
In this study, we analyze the distribution and density of nerves across the spectrum of gastric neoplasia by immunohistochemical (IHC) staining of gastric precancerous lesions (GPLs), normal and cancerous tissue obtained from endoscopic biopsies or surgically resected specimens in four cohorts.We also utilize gene expression omnibus (GEO) datasets to further validate the distribution patterns of nerves, and to explore key genes and pathways that regulate nerves.We observed a dynamic relationship between sympathetic and parasympathetic innervation during the development of gastric tumorigenesis.The results were validated at the level of nerve expression signatures on mRNA-seq.The key nerveregulating genes and the associated functional pathways are also identified by bioinformatics.
In contrast to prior research, our study, for the first time, provides a detailed description of the changes in various neural signals during the process of GC using multicenter samples, particularly focusing on the dynamic alterations in sympathetic and parasympathetic nerve densities alongside Schwann cell populations across different stages of precancerous lesions.While existing literature acknowledges the involvement of nerves in GC progression, our investigation offers a comprehensive analysis of specific nerve alterations within the gastric microenvironment, integrating data from patient cohorts and mRNA expression profiles.Moreover, we identify novel key genes and signaling pathways implicated in regulating nerve-tumor interactions, shedding light on potential targets for intervention and therapeutic strategies.By elucidating the distinct contributions of sympathetic and parasympathetic nerves in gastric precancerous progression, our study offers critical insights for the development of targeted therapies aimed at disrupting nerve-mediated tumor promotion in GC.

| Patients, samples and study design
The workflow of our study is illustrated in Figure 1.A total of 257 eligible patients were enrolled from four national hospitals: Nanfang Hospital of Southern Medical University (Nanfang cohort), the First Affiliated Hospital of Shantou University Medical College (Shantou cohort), Shantou Longhu People's Hospital (Longhu cohort), and Wuhan Central Hospital (Wuhan cohort).The study collected samples from patients undergoing endoscopic biopsy or surgical resection for various GPLs, including atrophic gastritis (AG), intestinal metaplasia (IM), lowgrade intraepithelial neoplasia (LIN), and high-grade intraepithelial neoplasia (HIN).Additionally, normal and tumor tissues are also collected from patients with intestine-type GC.Three GEO datasets (GSE55696, GSE87666, and GSE160116) containing RNA-seq profiles of human GPLs were obtained for further analysis.A detailed description of the data can be found in Supplementary Methods.

| Immunohistochemistry and digital quantification of nerves densities
IHC is used to visualize nerve-specific protein expression.S100, tyrosine hydroxylase (TH), and ChAT are employed as markers for axons, sympathetic nerve, and parasympathetic nerve, respectively.A standardized IHC protocol was followed, involving tissue sectioning, epitope retrieval, blocking, primary and secondary antibody incubation, and visualization using 3,3'-Diaminobenzidine chromogen and hematoxylin counterstaining.Digital quantification 17 was employed to assess the percentage of immunoreactive area (IRA%) of nerve components.Digital whole-slide images are acquired using a computer-aided slide-scanning system with standardized settings.Two expert pathologists, blinded to the study reviewed all slides after IHC and hematoxylin staining to assess tissue types and lesion grades.The analysis of nerve component focusses only on lesion-related stroma.Fiji ImageJ software was used for immunoreactive staining analysis, employing the Yen algorithm 18 for optimal thresholding (Figure S1 and Table S1).Experiments were conducted at the Guangdong Provincial Key Laboratory of Precision Medicine for Gastrointestinal Tumor.A detailed description of the experimental materials and protocol can be found in Supplementary Methods.
The flowchart of the study design.
CHEN ET AL.

| Nerves signatures, pathways and immune microenvironment assessment
Autonomic nervous system (ANS) signatures are validated using marker genes. 19Differentially expressed genes (DEGs) among GPL subtypes are determined using the "limma" R package with FC >1.4 and FDR <0.05.GC driver genes (Table S2) and nerve-cancer related genes (Table S3) were used to obtain differentially expressed driver-related and nerve-regulated genes.Functional pathway assessments, including gene ontology (GO) enrichment analyses, gene set enrichment analysis (GSEA), and Gene Set Variation Analysis (GSVA), are used to assess tumor hallmarks and pathways.TME signatures were explored using single-sample gene set enrichment analysis (ssGSEA) and specific immunerelated gene sets (Table S4).Details on the bioinformatics methods and other procedures can be found in Supplementary Methods.

| Statistical analysis
Statistical computations and illustrations were performed using IBM-SPSS 25 and GraphPad Prism 8 software.Bioinformatic analysis was performed using the Perl Programming Language (v5.30.0) and R software (4.2.0).The non-parametric Kruskal-Wallis test was employed for statistical testing.Statistical differences are indicated by lowercase letters (p < 0.05) in multiple comparisons, in which variables assigned to the same letter exhibit no statistical significance, whereas different letters indicate significant difference.Statistical significance was determined by a threshold of p < 0.05.

| Baseline patient data and nerves distribution characteristics
The baseline data for this study are presented in Table S5.All pathological specimens underwent thorough examination by pathologists to ensure sufficient tissue for immunochemical staining and digital analysis.As shown in Figure 2A, samples from endoscopic biopsies or surgical resections in this study contain, at a minimum, complete mucosal tissue including layers of gastric foveola, glands and muscularis mucosa.Any specimen containing inadequate tissue for analysis was excluded from the study.The grading standard for tissue lesion adheres to the guidelines of the visual analog scale of the updated Sydney system. 20e study comprised 257 participants, with an average age of 57 years (range: 22-89 years).Of these, 158 (61.5%) are male, and 99 (38.5%) are female.The majority of gastric samples originate from the gastric antrum (72.8%), followed by the body (11.3%) and cardia (7.4%).The AG, IM, LIN, and HIN subgroups consist of 46, 52, 55, and 48 patients, respectively.23 tumor-adjacent normal samples and 33 intestinal-type GC samples were also used as controls.Among the participants, 163 patients (63.4%) belong to the Nanfang cohort, 42 to the Shantou cohort (16.3%), and 42 to the Longhu cohort (16.3%).
IHC of nerve components revealed positive staining in both the mucosal and submucosal layers.In normal tissue, the distribution of nerves exhibits an increasing gradient from the gastric foveola, gland region, and muscularis mucosa to the submucosa, with nerve density gradually rising in a stepwise manner (Figure 2B, arrow and square).Notably, nerves are abundant around arterioles within the mucosal and submucosal layers.This stepwise increase in nerve distribution within the gastric wall is also observed in precancerous and tumor tissues (Figure S2).
Stratified analysis in AG and IM subtypes did not reveal significant differences in PN, SN, and S100 density among mild, moderate, and severe lesions, except for ChAT staining between mild and severe IM (p = 0.022).The ChAT-TH ratio, representing the relationship between PN and SN, varies in GPLs.There was no significant change in this ratio among normal, AG, IM, and LIN stages.However, the ratio increases rapidly at the LIN stage and maintains a high level at the tumor stage (Figure 7A).The ChAT-TH ratio has a high clinical value in the diagnosis of HIN from LIN (AUC = 0.986) and other precancerous lesions except tumor (Figure 7B).These findings suggest that changes in the ChAT-TH ratio might indicate the precancerous severity and serve as an indicator of the tendency toward malignancy.
F I G U R E 2 General characteristics of nerve distribution (S100 staining) in gastric tissue.(A) Samples in this study at least contain a complete mucosal layer, including gastric foveola, glands and muscularis mucosa.(B) Densities of nerves are highest in the muscularis mucosa and submucosal layers, moderate in the gastric glands layer, and lowest in the foveolar layer.Nerves were notably abundant around the arterioles (blue square).Scale bar: 250 μm for (A) and upper (B), 50 μm lower in panel (B).

| Validation of autonomic nervous system signatures
RNA-seq profiling of GPL samples from four subtypes (AG: 19 patients, IM: 14 patients, LIN: 26 patients, HIN: 29 patients) was conducted using GEO datasets.Figure S3 shows the gene expression baseline of this cohort, confirming its reliability.We evaluated the gene expression signatures of intratumoral nerve growth (ITNG), sympathetic nervous system (SNS), and parasympathetic nervous system (PNS) across the GPL subtypes (Figure 8).Overall, we observed significantly high levels of ITNG and PNS signatures from AG to HIN (p < 0.05), indicating an increasing trend of ITNG and PNS activity from AG lesions to other GPLs, although the difference among the IM, LIN and HIN groups was not significant.Conversely, the expression level of the SNS signature decreased significantly from AG to HIN (p < 0.05).These findings align with our experimental results of IHC staining on specific nerves, demonstrating a dynamic change in sympathetic and parasympathetic signatures across the GPLs spectrums.

| Identification of DEGs and key nerve-cancer related genes
A total of 700 DEGs (Table S9) were identified between the negative for intraepithelial lesion or malignancy (NILM) and LIN subtypes, with 438 up-regulated and 262 down-regulated genes.Additionally, we identified 297 DEGs (Table S10) between LIN and HIN, including 119 up-regulated and 178 down-regulated genes.The volcano plots (Figure 9A) illustrate the top 5 up-up-and downregulated genes in each comparison.We performed a Venn diagram analysis to identify shared genes among the DEGs, GC driver genes, and nerve-cancer crosstalk related genes.The analysis (Figure 9B) revealed 7 nerverelated genes (NTRK3, NRP1, BCL2, CCND1, VEGFA, PLXNA2, and ADRB2) shared in the DEGs between NILM and LIN.Similarly, two nerve-related genes (NTRK3 and MYBL2) are identified as shared in the DEGs between LIN and HIN.Furthermore, there are 69 DEGs (Figure 9C) present between LIN and HIN that are identified among the DEGs between NILM and HIN, suggesting the presence of premalignancy-like genetic alterations in LIN.Among the DEGs, GC driver genes, and nerve-cancer crosstalk related genes, NTRK3 emerges as the unique key gene.We depicted the expression pattern of eight nerve-related genes in a half violin plot (Figure 9D).Notably, MYBL2, NRP1, and BCL2 show increased levels from NILM to LIN and HIN, while CCND1, VEGFA, PLXNA2, and ADRB2 exhibit the opposite trend.Interestingly, NTRK3 displays an initial rise from NILM to LIN group, followed by a significant drop from LIN to HIN subtypes.

GPLs subtypes
The GO analysis reveals that the DEGs are primarily involved in granulocyte chemotaxis, granulocyte migration, neutrophil chemotaxis, which relate to neutrophilmediated immunity, and digestive system processes (Figure 10A and Table S11).Through GSEA using Hallmark (Figure 10B,C) and kyoto encyclopedia of genes and genomes (Figure 10D,E) gene sets, we observed that the up-regulated DEGs in NILM versus LIN were enriched in Additionally, GSVA analysis confirmed the underlying biological pathways involved in GPLs progression.Inflammatory and immune signaling pathways, such as inflammatory response, complement and coagulation system, interferon response, and oxidative phosphorylation are significantly enhanced.Classical GC-related pathways, such as epithelial-to-mesenchymal transition (EMT), Ras/Kras, and mTOR, are also involved across the GPLs spectrums (Figure S4).These findings further indicate the occurrence of inflammatory reactions with varied immune responses across GPLs.The activated pathways include epithelial cell signaling in HP infection, MYC, EMT, Ras/Kras, and mTOR.This suggests a correlation among HP-related inflammation, nerve regulation, and immune signaling pathways in gastric tumorigenesis.

| Immune microenvironment exploration
We utilize ssGSEA to estimate immune cell infiltration scores and assess immune-related biomarkers in different risk groups.Significant differences were observed in the infiltration levels of immune cell species or biomarkers (Figure 11A).aDCs, HLA, and T helper cells showed an increasing trend, whereas CD8þ T cells, natural killer cells, and APC co-inhibition decreased significantly from NILM to LIN to HIN, indicating increasing inhibition of anti-tumor immunity as GPLs progressed.This inhibition is particularly evident from NILM to LIN.
The ssGSEA of gene sets related to specific biological processes was performed.The results showed that stromal-activation-relevant signatures, DNA damage and repair-relevant signatures, and immune activationrelevant signatures (Figure 11B) are significantly involved.
EMT, a crucial process where epithelial cells acquire invasive mesenchymal characteristics, 21 shows significantly higher expression in LIN and HIN compared to NILM (p < 0.001), indicating progressive acquisition of mesenchymal attributes with tumor advancement.Activation of the EMT pathway can hinder effector T cell initiation and cytotoxic capabilities, 22 potentially explaining the low CD8þ T cell levels in the HIN group.We examined immune activation/checkpoint and immunomodulator-related gene expression among GPL subtypes.Tumor necrosis factor (TNF), a representative immune activation biomarker, increases gradually from NILM to LIN to HIN, while the immune checkpoint biomarkers CTLA-4 and HAVCR2, increase progressively (Figure 11C).The increased expression of immune checkpoint molecules can lead to the evasion of the immune response by this subtype.These results align with the biological process and pathway enrichment observed through ssGSEA, indicating increasing inhibition of antitumor immunity as GPLs progress.
We analyze immunomodulator expression, including chemokines, interleukins, interferons, and their receptors, in GPLs' immune microenvironments (Figure 11D).LIN and HIN show higher expression of chemokines (CCL20, CCL25, CCR1, CCR3, CCL3, CXCL16, and PF4) and interleukins/cytokines (IL10, IL2RA, IL1A, IL23A, IL34, TNF, CSF2RB, and FAS) compared to NILM.Conversely, other interleukins, cytokines, and interferons (IL1R2, IL10RB, IL4R, IFNGR2, IFNE, VEGFA, TGFBR2, EPOR, EGFR, and TPO) are significantly lower in LIN and HIN groups.Nerves were previously thought to have little impact on tumor progression, leading to a long-term disregard within the TME.With the advent of new nerve biomarkers and advancements in methodologies, there has been a notable increase in research focusing on the regulatory roles of nerves in cancer, attracting widespread attention. 23Despite the recognized impact of nerves in cancer, the roles of nerves in GC are only now emerging. 24he ANS consists of the sympathetic and parasympathetic branches, which work together to regulate physiological functions.Sympathetic nerves initiate the "fight or flight" response, reducing blood flow to the stomach and digestive activity.In contrast, the PN promotes the "rest and digest" response, enhancing digestive function.These opposing roles of SN and PN are also observed in the TME. 6,25,26Adrenergic signaling activates pathways that contribute to tumor growth, invasion, and metastasis, while cholinergic signaling suppresses tumorigenesis and cancer stemness through the MAPK/ EGFR and PI3K/AKT pathways. 12,27ecent studies have illuminated the crucial roles played by nerves in the early onset and development of GC.In a mouse GC model, Y. Hayakawa and colleagues 28 demonstrated that nerves within the gastrointestinal tract regulate tumor stem cells by secreting acetylcholine.Furthermore, the acetylcholine released by Dclk1þ tuft cells and nerves stimulates the production of nerve growth factor (NGF) in gastric epithelial cells.This, in turn, fosters the expansion of neurons and initiates tumorigenesis by activating the Wnt pathway.Inhibition of epithelial proliferation and tumorigenesis can be achieved by blocking downstream signals of NGF/Trk receptors and the muscarinic receptor-3 for acetylcholine.
0][31] However, the roles of sympathetic/adrenergic nerves in GC appear to be controversial.Miyato and colleagues 32 conducted a study on surgically resected GC specimens and demonstrated a significant decrease in SN fibers in tumor tissues.The loss of SN was associated with adverse clinical outcomes, including deeper tumor invasion, more lymph node metastasis, and worse patient survival.Similar findings were reported in a recent review. 33However, Tatsuta and colleagues 34 found that chemical sympathectomy using 6-hydroxydopamine, a neurotoxin that destroys catecholaminergic neurons, substantially reduced gastric tumorigenesis in a rat model.Moreover, a recent study observed that a high density of SN was associated with tumor recurrence. 35espite conflicting data and small study cohorts, no definitive conclusions can be drawn regarding the role of sympathetic signaling in gastric tumorigenesis.Our results address the current gap in understanding the dynamic changes in nerve density and distribution across GPLs.
The Correa cascades effectively represent the process of gastric tumorigenesis, 36 and studying changes in nerve types and density within it can offer insights into the role of nerves in GC.While previous studies have identified high-density PN fibers are associated with tumorpromoting effects, as well as conflicting findings regarding SN fiber density in GC, none have compared SN with PN or investigated their presence in gastric precancerous disease using human tissue.By employing IHC staining of paired samples with SN and PN biomarkers, our study reveals a gradual decrease in SN density across the spectrums of gastric precancerous stages, with only a fraction of SN fibers remaining after GC formation.In contrast, PN density increased ninefold in GC compared with normal tissue.Additionally, we observed a concurrent increase in Schwann cell density along with PN fiber density, indicating that Schwann cell proliferation is crucial for supporting neural growth.Interestingly, the density of PN and axons remained low until the late precancerous and early tumor stages, after which a sudden surge in nerve density occurred.This result can be attributed to a distorted form of neoorganogenesis involving angiogenesis for nutrient supply and neurogenesis for coordination and signaling during tumor growth and 37 We propose the PN-SN ratio as an innovative measure of disease severity, considering the contrasting changes of SN and PN in precancerous lesions.This ratio offers a comprehensive reflection of both SN and PN alterations, wherein a smaller ratio indicates a lower likelihood of precancerous lesions advancing to cancer, while a larger ratio suggests a higher likelihood of progression.The PN-SN ratio holds promise as a prognostic indicator of the severity of precancerous lesions and may serve as a reference for GPL treatment.
In genetically engineered mouse models, Zhao and colleagues 30 demonstrated the inhibitory effect of denervation of PN on gastric tumorigenesis.Miyato and colleagues, 32 through the examination of GC specimens, found a significant correlation between low SN density and worse tumor invasion depth, lymph node metastasis, and patient survival outcomes.Based on previous reports and our study's findings, it can be inferred that an imbalance in nerve distribution, resulting from a remodeling process, may contribute to, or at least indicate, the neoplastic transformation of the stomach.Faulkner and colleagues 6 described these opposing yet interconnected functions of SN and PN as a Yin-yang type of neural regulation in cancer.Considering our study's discovery of the dynamic relationship between SN and PN, we infer that this Yin-yang type of neural regulation may also exist in GC progression.
Through bioinformatic analysis of GPL transcriptome data, we identified several key nerve-regulating genes, namely NTRK3, MYBL2, NRP1, BCL2, CCND1, VEGFA, PLXNA2, and ADRB2.MYBL2 functions as an oncogene in GC by promoting cell proliferation, inhibiting apoptosis, and facilitating tumorigenesis and metastasis via the Wnt/β-catenin signaling pathway. 38NRP1 acts as an oncogene by modulating axon guidance, nervous system development, and promoting tumor development through its receptor, VEGFR, and interactions with semaphoring and the VEGF family. 39PLXNA2 plays a crucial role in regulating perineural invasion in prostate cancer, 40 whereas its upregulation reduces tumor migration and invasion, and its downregulation is associated with increased invasive and migratory capacities in melanoma. 41Similar role has been observed for ADRB2, identified as a potential protective gene that regulates immune responses in breast cancer. 42hese core genes might exert their influence on different nerves through complex interactions in the nerve-cancer crosstalk, leading to dynamic changes in nerve distribution.Notably, we observed a gradual increase in MYBL2, NRP1, and BCL2 expression in GPLs, suggesting their involvement in carcinogenesis.Conversely, PLXNA2 and ADRB2 exhibited a decreasing trend in GPLs, indicating their potential tumorsuppressive characteristics.This highlights that nerves, in conjunction with various molecular signals, produce distinct effects, and the differential expression of genes might serve as a regulator behind neurodynamic changes.spectrums.Notably, the literature review (Table S12) confirmed the importance of MAPK/ERK, PI3K/AKT/ mTOR, and Wnt signaling as crucial nerve-regulating pathways involved in GPL progression and nervepremalignancy crosstalk.
Among the identified genes, we discovered a unique key gene called neurotrophic-tropomyosin receptor tyrosine kinase 3 (NTRK3), also known as tropomyosin receptor kinase C (Trkc).NTRK3 belongs to the neurotrophin receptor family (NTRK 1-3) and plays a vital role in the embryological development of the central nervous system. 43NTRK3 interacts with its ligand NT-3 and activates downstream targets and pathways, including mTOR and Wnt/β-catenin, which regulate cell differentiation, proliferation, and survival in GC. [44][45][46] NTRK3 can function as either an oncogene or a tumor suppressor gene.While NTRK3 gene fusions are oncogenic drivers in various solid tumors, 47 they can also exhibit tumorsuppressive effects.For instance, NTRK3 is commonly inactivated by epigenetic mechanisms in colorectal cancer, and its reconstitution induces apoptosis in colorectal cancer cells. 484][45] Pathway analysis indicates the enrichment of these pathways in GPLs subtypes, underscoring the crucial role of NTRK3 in GPLs progression.In short, the pathway analysis supports the notion that the initiation of GPLs is attributed to inflammation-induced damage of the gastric mucosa, particularly in the context of HP infection-induced gastritis.This inflammatory response triggers immune reactions and facilitates interactions with the nervous system.Notably, the nerves actively engage in specific signaling pathways such as ERK, mTOR, and Wnt, exerting regulatory control over the TME.Consequently, these complex neurobiological processes contribute to the precancerous progression.However, since these findings are solely based on bioinformatics analysis, further basic experiments and molecular mechanism studies are warranted.
Our study has some limitations.Primary among these is the fact that it is a purely quantitative IHC study with bioinformatic validation.In signal and pathway exploration using bioinformatic methods, only a preliminary analysis of potentially relevant genes and signaling pathways involved in nerve regulation is performed, without providing a detailed elucidation of the specific molecular mechanisms underlying the dynamic changes in nerve densities during the progression of GPLs.Additionally, if the number of samples is still not large enough, there might be a risk of selection bias influencing the results.Finally, since all the samples were collected retrospectively, the standard of specimen collection is not uniform, resulting in different sizes and thicknesses of tissue layers.Considering that nerves are mainly enriched in muscularis mucosa and submucosal layer, there may be a variability in immunostaining calculation of nerve densities.
Our study enhances understanding of nerves' roles in GPLs and aids in identifying therapeutic targets.Targeting the nervous system and its interaction with cancer cells holds promise in the field of cancer research.Our findings have three theoretical implications.Firstly, HP infection contributes to a large population exhibiting various degrees of precancerous lesions.Nerve distribution pattern aid in stratifying patients and high-risk individuals for prompt treatment.The combined effect of both parasympathetic and sympathetic nerves in GC may function as a more comprehensive prognostic indicator of tumor characteristics and clinical outcomes.Lastly, given the limited treatment options currently available for GPLs, targeting specific neural activity might offer a novel therapeutic approach to address precancerous lesions and prevent the onset of GC.
Future research on nerves and GPLs is crucial for innovative therapies and patient outcomes.One promising avenue for future investigation lies in elucidating the specific molecular mechanisms underlying the dynamic changes in nerve densities observed during the progression of GPLs.Advanced techniques such as single-cell RNA sequencing could reveal nerve population heterogeneity and their roles in disease progression.Additionally, understanding signaling pathways between nerves and cancer cells, particularly focusing on the roles of key nerve-regulating genes identified in our study, such as NTRK3, MYBL2, and NRP1, is necessary.Moreover, prospective clinical studies involving larger, wellcharacterized patient cohorts are needed to validate the clinical value of sympathetic and parasympathetic nerves, as well as the proposed PN-SN ratio and their potential for guiding personalized treatment approaches in patients with GPLs and GC.Furthermore, investigations into novel therapeutic strategies targeting the nervous system and disrupting the nerve-cancer crosstalk hold promise for preventing the onset of GC and improving survival outcomes in patients with GC.Embracing these research directions may deepen our understanding of neurobiological processes in gastric tumorigenesis, leading to transformative cancer therapeutics.
In conclusion, this study enhances our understanding of the density and distribution of sympathetic and parasympathetic nerves in different stages of GPLs.We observed a progressive decrease in sympathetic nerve density and an increase in parasympathetic nerve density from normal tissue to precancerous lesions and cancer.The dynamic relationship between sympathetic and parasympathetic nerves suggests a Yin-yang type of neural regulation in GC.We have identified several key nerve-regulating genes and their involvement in MAPK/ ERK, PI3K/AKT/mTOR, and Wnt signaling pathways during the progression of GPLs.These findings underscore the importance of comprehending the interaction between nerves and tumor cells for effective prevention and treatment of GC.They also provide potential avenues for future clinical translation, focusing on targeting the nervous system and disrupting the crosstalk between nerves and cancer.

F I G U R E 4
Stratified analysis of parasympathetic nerve signals across normal, AG, IM, LIN, HIN, and tumor lesions.Representative pictures of immunohistochemical staining in precancerous subtypes, and statistical comparisons of ChAT staining among the GPLs spectrums.The statistical differences in nerve densities were performed by Kruskal-Wallis test.Statistical difference is denoted by the letter notation with p < 0.05, in which different letters (a, b, c) denote significant differences between groups (p < 0.05), whereas groups with the same letter indicate not significantly different.AG, atrophic gastritis; GPL, gastric precancerous lesion; HIN, high-grade intraepithelial neoplasia; IM, intestinal metaplasia; LIN, low-grade intraepithelial neoplasia.complementand coagulation cascades, inflammatory response, RAS/KRAS signaling, and neuroactive ligand receptor interaction pathways.The down-regulated DEGs in NILM versus LIN are enriched in interferon-alpha response, Notch signaling, and DNA replication.In LIN versus HIN, the up-regulated DEGs are enriched in Myc, VEGF, mTORc1, PI3K/AKT/mTOR, pancreatic cancer, and epithelial cell signaling in Helicobacter pylori (HP) infection pathway.The down-regulated DEGs in LIN versus HIN are enriched in neuroactive ligand receptor interaction, fatty acid metabolism, and Hedgehog signaling.These DEG groups share common biological pathways related to inflammatory and immune responses, nerve regulation, and cancer-related signaling, including Ras/Kras and PI3K/AKT/mTOR pathways.

F I G U R E 5
Stratified analysis of sympathetic nerve signals across normal, AG, IM, LIN, HIN, and tumor lesions.Representative pictures of immunohistochemical staining in precancerous subtypes, and statistical comparisons of TH staining among the GPLs spectrums.The statistical differences in nerve densities were performed by Kruskal-Wallis test.Statistical difference is denoted by the letter notation with p < 0.05, in which different letters (a, b, c) denote significant differences between groups (p < 0.05), whereas groups with the same letter indicate not significantly different.AG, atrophic gastritis; GPL, gastric precancerous lesion; HIN, high-grade intraepithelial neoplasia; IM, intestinal metaplasia; LIN, low-grade intraepithelial neoplasia; TH, tyrosine hydroxylase.

F I G U R E 6
Stratified analysis of Schwann cell signals across normal, AG, IM, LIN, HIN, and tumor lesions.Representative pictures of immunohistochemical staining in precancerous subtypes, and statistical comparisons of S100 staining among the GPLs spectrums.The statistical differences in nerve densities were performed by Kruskal-Wallis test.Statistical difference is denoted by the letter notation with p < 0.05, in which different letters (a, b, c) denote significant differences between groups (p < 0.05), whereas groups with the same letter indicate not significantly different.AG, atrophic gastritis; GPL, gastric precancerous lesion; HIN, high-grade intraepithelial neoplasia; IM, intestinal metaplasia; LIN, low-grade intraepithelial neoplasia; TH, tyrosine hydroxylase.

F I G U R E 7
The ChAT-TH ratio across the gastric spectrums.(A) ChAT-TH ratio comparison in different groups.(B) Diagnostic performance of ChAT-TH ratio for LIN and HIN, HIN and tumor.The statistical differences were performed by Kruskal-Wallis test.Statistical difference is denoted by the letter notation with p < 0.05, in which different letters (a, b, c) denote significant differences between groups (p < 0.05), whereas groups with the same letter indicate not significantly different.ChAT, choline acetyltransferase; HIN, high-grade intraepithelial neoplasia; IRA, immunoreactive area; LIN, low-grade intraepithelial neoplasia; ns, not significant; TH, tyrosine hydroxylase.

F I G U R E 8
Validation of gene expression in nerve signatures across the gastric precancerous spectrums.Gene expression signatures of overall ITNG (A) and PNS (B) significantly increase from AG to HIN, whereas SNS (C) decreases from AG to HIN.AG, atrophic gastritis; HIN, high-grade intraepithelial neoplasia; ITNG, intratumoral nerve growth; PNS, parasympathetic nervous system; SNS, sympathetic nervous system.The GSEA and GSVA analyses revealed significant enhancement of inflammatory and immune signaling, epithelial-to-mesenchymal transition and epithelial cell signaling in HP infection pathway.Additionally, wellknown GC-related pathways, such as VEGF, Ras/Kras, mTOR, and Wnt signaling, are implicated across the GPL F I G U R E 9 The differentially expressed genes (DEGs) across the gastric precancerous spectrums.(A) Volcano plots of DEGs clustering with top 5 up-regulated and down-regulated genes between NILM, LIN and HIN groups.(B, C) Venn diagram analysis of shared between the DEGs, GC driver genes, and nerve-cancer crosstalk-related genes.(D) Half violin plot of expression patterns of the key 8 nerverelated genes.GC, gastric cancer; HIN, high-grade intraepithelial neoplasia; LIN, low-grade intraepithelial neoplasia; NILM, negative for intraepithelial lesion or malignancy.F I G U R E 1 0 Pathway enrichment analysis.(A) Gene ontology (GO) enrichment analyses in terms of biological processes among GPLs subtypes.Gene set enrichment analysis of DEGs with Hallmark gene set in (B) NILM versus LIN and (C) LIN versus HIN, and with KEGG gene sets in (D) NILM versus LIN and (E) LIN versus HIN.DEGs, differentially expressed genes; GPLs, gastric precancerous lesions; HIN, high-grade intraepithelial neoplasia; KEGG, kyoto encyclopedia of genes and genomes; LIN, low-grade intraepithelial neoplasia; NILM, negative for intraepithelial lesion or malignancy.

F I G U R E 1 1
Immune microenvironment analysis.(A) Box plot by single-sample gene set enrichment analysis showing the difference in infiltration of immune cells among NILM, LIN, and HIN subtypes; (B) gene clusters were distinguished by different immune relevant signature, TMEscore, and other pertinent biological processes; (C) difference in the expression of immune activation-related and immune checkpoints genes among GPLs subtypes; (D) the heat map illustrates difference in expression of chemokines, interleukins, interferons and their receptors among the GPLs subtypes.GPLs, gastric precancerous lesions; HIN, high-grade intraepithelial neoplasia; LIN, low-grade intraepithelial neoplasia; NILM, negative for intraepithelial lesion or malignancy.