PO.BCS01.05 · 生物信息与计算
血清素转录轴塑造肺癌的免疫表型和临床结局
A serotonin transcriptional axis shapes immune phenotypes and clinical outcomes in lung cancer
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
背景
神经调质-代谢物通路是肿瘤-免疫相互作用的新兴调控因子。近期一项研究表明，神经递质信号传导重塑了免疫浸润。在此框架基础上，我们构建了一个血清素转录轴，纳入TPH/HTR合成、MAOA/IDO1降解和SLC6A4转运，以表征肺腺癌（LUAD）和肺鳞癌（LUSC）中的免疫状态。
方法
通过Xena处理TCGA RNA-seq数据。经行z标准化的血清素基因生成了上调（Up）、下调（Down）和综合轴（Axis）评分。k-means定义了血清素表型。使用Wilcoxon检验、相关矩阵、limma差异表达基因（DEG）、Hallmark GSEA以及KM/Cox生存分析，评估了免疫检查点、细胞毒性程序、趋化因子和免疫细胞替代标志物（CD8A、FOXP3、NCAM1、IRF5、CD163、MS4A1、ITGAX）。
AI工具（ChatGPT）用于编辑、语法和文本精简。
结果
在LUAD中，出现了三种表型：C1（MAOA/IDO1高、免疫冷）、C2（血清素耗竭）和C3（TPH/HTR高、免疫热）。
C3显示出CD8A、IFNG、GZMB、PRF1、CXCL9/10/13以及包括PDCD1、CD274、CTLA4、LAG3、TIGIT在内的检查点的强烈激活（p<0.001）。
血清素轴评分与CD8 T细胞、NK细胞、树突状细胞、B细胞和M1标志物呈正相关，与CD163呈负相关，表明一个炎症性、抗原呈递的TME。
GSEA显示C3富集干扰素和炎症程序；C1偏向代谢/MYC通路。
各簇间生存存在显著差异（p=0.0014），C3最佳而C1最差；高血清素轴趋向于改善OS（p=0.055）。在LUSC中，相同的表型再次出现，证实了跨组织学的可重复性。C3再次显示出最强的免疫激活，细胞毒性和检查点表达升高。
血清素激活与CD8A、NK特征、MS4A1、ITGAX和M1标志物相关，而血清素抑制与CD163高的M2状态相关。
GSEA显示C3富集IFN-alpha/gamma和炎症通路；C1富集OXPHOS/MYC程序。
簇OS不显著（p=0.63），但连续血清素轴评分预测改善的OS（p=0.039）。
结论
血清素通路状态在LUAD和LUSC中可重复地塑造免疫景观和生存。血清素激活与细胞毒性浸润、富含趋化因子的炎症、树突状/B细胞特征、M1极化、检查点表达、炎症性GSEA程序以及改善的生存相一致。这些发现将近期研究的神经调质-TME概念加以拓展，并将血清素轴识别为一个跨组织学生物标志物和潜在的免疫调节靶点。
查看英文原文 English abstract
Background
Neuromodulator-metabolite pathways are emerging regulators of tumor-immune interactions. A recent study showed that neurotransmitter signaling remodels immune infiltration. Building on this framework, we developed a serotonin transcriptional axis incorporating TPH/HTR synthesis, MAOA/IDO1 degradation, and SLC6A4 transport to characterize immune states in lung adenocarcinoma (LUAD) and lung squamous carcinoma (LUSC).
Methods
TCGA RNA-seq data were processed via Xena. Row-z-scaled serotonin genes generated Up, Down, and composite Axis scores. k-means defined serotonin phenotypes. Immune checkpoints, cytotoxic programs, chemokines, and immune-cell surrogates (CD8A, FOXP3, NCAM1, IRF5, CD163, MS4A1, ITGAX) were assessed using Wilcoxon testing, correlation matrices, limma DEGs, Hallmark GSEA, and KM/Cox survival.
AI tools (ChatGPT) were used for editing, grammar, and text condensation.
Results
In LUAD, three phenotypes emerged: C1 (MAOA/IDO1-high, immune-cold), C2 (serotonin-depleted), and C3 (TPH/HTR-high, immune-hot).
C3 demonstrated strong activation of CD8A, IFNG, GZMB, PRF1, CXCL9/10/13 and checkpoints including PDCD1, CD274, CTLA4, LAG3, TIGIT (p<0.001).
Serotonin Axis Score correlated positively with CD8 T cells, NK cells, dendritic cells, B cells, and M1 markers, and negatively with CD163, indicating an inflamed, antigen-presenting TME.
GSEA showed C3 enrichment for interferon and inflammatory programs; C1 favored metabolic/MYC pathways.
Survival differed significantly across clusters (p=0.0014), with C3 best and C1 worst; high serotonin axis trended toward improved OS (p=0.055).In LUSC, the same phenotypes recurred, confirming cross-histology reproducibility. C3 again showed the strongest immune activation with elevated cytotoxic and checkpoint expression.
Serotonin activation associated with CD8A, NK signatures, MS4A1, ITGAX, and M1 markers, whereas serotonin suppression correlated with CD163-high M2 states.
GSEA showed C3 enrichment for IFN-alpha/gamma and inflammatory pathways; C1 enriched for OXPHOS/MYC programs.
Cluster OS was not significant (p=0.63), but continuous Serotonin Axis Score predicted improved OS (p=0.039).
Conclusions
Serotonin pathway states reproducibly shape immune landscapes and survival across LUAD and LUSC. Serotonin activation aligns with cytotoxic infiltration, chemokine-rich inflammation, dendritic/B-cell signatures, M1 polarization, checkpoint expression, inflamed GSEA programs, and improved survival. These findings extend neuromodulator-TME concepts from recent study and identify the serotonin axis as a cross-histology biomarker and potential immunomodulatory target.
利益披露 Disclosure
E. Kalemoglu, None..
S. Akgun, None..
A. Caner, None.