Ji Ke1, Jian Cui1, Xingguo Yang1, Xin Du1, Bobo Ma2, Lei Yu1. 1. Department of Thoracic Surgery, Beijing Tongren Hospital, Capital Medical University, Beijing 100730, China. 2. School of Rehabilitation Medicine, Capital Medical University, China Rehabilitation Research Center, Beijing 100068, China.
Abstract
BACKGROUND: m6A RNA methylation modification plays an important role in the occurrence and progression of lung cancer and regulates tumor immunity. Current studies mostly focus on the differential expression of some specific m6A effectors and infiltrating immune cell. m6A methylation modification is the result of mutual adjustment and balance between effectors, and changes in the expression of one or two effectors are far from enough to reflect the panorama of m6A methylation. The role of m6A in the immune microenvironment of lung adenocarcinoma (LUAD) is still poorly understood. The aim of this study is to investigate the effect of different m6A modification patterns in immune microenvironment of LUAD. METHODS: LUAD data was obtained from The Cancer Genome Atlas (TCGA), University of California Santa Cruz Xena (UCSC Xena) and Gene Expression Omnibus (GEO) databases. Gene mutation, differential expression and survival analysis were performed for 24 m6A effectors. The m6A modification pattern was constructed by unsupervised clustering method, and the m6A clusters survival analysis, gene set variation analysis, immune score and immune cell infiltration analysis were performed. The association between LRPPRC protein expression levels and infiltration of CD8+ cytotoxic T lymphocytes and CD68+ macrophages in the tumor microenvironment was validated by immunohistochemistry in LUAD tissue microarray with 68 cases. RESULTS: The mutations of m6A effector were found in 150 of 567 LUAD cases with a frequency of 26.46%. 6 readers and 3 writers were significantly up regulated in LUAD tissues compared with normal tissues. IGF2BP1 and HNRNPC are the independent risk factors for prognosis of LUAD. Abundant cross-talks among writers, erasers and readers were demonstrated. Three m6A modification patterns with different immune cell infiltration characteristics and clinical prognosis were established. Among m6A effectors, LRPPRC was found to be inversely associated with the infiltration of CD8+ cytotoxic T lymphocytes and CD68+ macrophages, and was validated in 68 LUAD tissues. CONCLUSIONS: m6A modification patterns play non-negligible roles in regulating the immune microenvironment. LRPPRC has potential to be a new biomarker for checkpoint inhibitor immunotherapy.
BACKGROUND: m6A RNA methylation modification plays an important role in the occurrence and progression of lung cancer and regulates tumor immunity. Current studies mostly focus on the differential expression of some specific m6A effectors and infiltrating immune cell. m6A methylation modification is the result of mutual adjustment and balance between effectors, and changes in the expression of one or two effectors are far from enough to reflect the panorama of m6A methylation. The role of m6A in the immune microenvironment of lung adenocarcinoma (LUAD) is still poorly understood. The aim of this study is to investigate the effect of different m6A modification patterns in immune microenvironment of LUAD. METHODS: LUAD data was obtained from The Cancer Genome Atlas (TCGA), University of California Santa Cruz Xena (UCSC Xena) and Gene Expression Omnibus (GEO) databases. Gene mutation, differential expression and survival analysis were performed for 24 m6A effectors. The m6A modification pattern was constructed by unsupervised clustering method, and the m6A clusters survival analysis, gene set variation analysis, immune score and immune cell infiltration analysis were performed. The association between LRPPRC protein expression levels and infiltration of CD8+ cytotoxic T lymphocytes and CD68+ macrophages in the tumor microenvironment was validated by immunohistochemistry in LUAD tissue microarray with 68 cases. RESULTS: The mutations of m6A effector were found in 150 of 567 LUAD cases with a frequency of 26.46%. 6 readers and 3 writers were significantly up regulated in LUAD tissues compared with normal tissues. IGF2BP1 and HNRNPC are the independent risk factors for prognosis of LUAD. Abundant cross-talks among writers, erasers and readers were demonstrated. Three m6A modification patterns with different immune cell infiltration characteristics and clinical prognosis were established. Among m6A effectors, LRPPRC was found to be inversely associated with the infiltration of CD8+ cytotoxic T lymphocytes and CD68+ macrophages, and was validated in 68 LUAD tissues. CONCLUSIONS: m6A modification patterns play non-negligible roles in regulating the immune microenvironment. LRPPRC has potential to be a new biomarker for checkpoint inhibitor immunotherapy.
从癌症基因组图谱数据库(The Cancer Genome Atlas, TCGA)(https://portal.gdc.cancer.gov/)和加州大学圣克鲁兹分校泛癌全基因数据分析工具数据库(University of California Santa Cruz Xena, UCSC Xena)(https://xena.ucsc.edu/)下载肺腺癌队列的基因突变数据、拷贝数变异数据、RNA-seq数据和临床资料。从基因表达综合数据库(Gene Expression Omnibus, GEO)(http://www.ncbi.nlm.nih.gov/geo)下载基因芯片数据集(GSE40419、GSE41271、GSE72094和GSE126044)。GSE40419数据集包含87个肺腺癌肿瘤组织和77个相邻正常肺组织的RNA表达数据。GSE41271和GSE72094数据集分别包含178个和381个肺腺癌肿瘤组织的RNA表达数据,以及完整的临床和生存信息。GSE126044数据集包含16个非小细胞肺癌(non-small cell lung cancer, NSCLC)肿瘤组织的RNA表达数据和抗PD-1免疫治疗的应答信息。将数据进行log2转换及标准化处理。
Mutation, co-occurrence analysis and copy number variation of m6A effector in lung adenocarcinoma. A: The mutation frequency of 24 m6A effectors in 567 patients with LUAD in TCGA-LUAD data sets; B: The co-occurrence of mutations between different m6A effectors; C: CNV of each m6A effectors. TCGA: The Cancer Genome Atlas; LUAD: Lung adenocarcinoma; CNV: copy number variation.
24个m6A效应器列表Information of 24 m6A effectors肺腺癌中m6A效应器的突变、共现分析及拷贝数变异。A:24个m6A效应器在567例TCGA-肺腺癌队列中突变频率;B:m6A效应器的共现分析;C:m6A效应器的拷贝数变异。Mutation, co-occurrence analysis and copy number variation of m6A effector in lung adenocarcinoma. A: The mutation frequency of 24 m6A effectors in 567 patients with LUAD in TCGA-LUAD data sets; B: The co-occurrence of mutations between different m6A effectors; C: CNV of each m6A effectors. TCGA: The Cancer Genome Atlas; LUAD: Lung adenocarcinoma; CNV: copy number variation.为了研究m6A效应器在转录水平上的变化,我们分别比较了TCGA肺腺癌队列和GSE40419队列中肿瘤组织与癌旁正常组织之间m6A效应器的mRNA表达水平(图 2A和图 2B)。与正常组织相比,癌组织中共有6个读取器(HNRNPA2B1、HNRNPC、IGF2BP1、LRPPRC、YTHDF1、YTHDF2)和3个写入器(KIAA1429、METTL5、RBM15)的表达明显上调;相反,癌组织中唯一下调的是擦除器FTO(图 2C和图 2D),表明m6A修饰可能在肺腺癌中具有促癌作用。
Expression of m6A effector in lung adenocarcinoma tissue and adjacent normal tissue. A and B: Heat map for the expression levels of m6A effectors in LUAD and paired normal lung tissues of TCGA-LUAD cohort (A) and GSE40419 cohort (B); C and D: The expression level of m6A effectors in tumor tissues and paired normal lung tissues of TCGA-LUAD (C) and GSE40419 (D) cohorts. T: tumor; N: normal; ns: no statistical significance; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.000, 1.
肺腺癌组织与癌旁正常组织m6A效应器的表达。A、B:TCGA-肺腺癌队列(A)和GSE40419队列(B)中肺腺癌组织与正常组织m6A效应器的表达热图;C、D:TCGA-肺腺癌队列(C)和GSE40419队列(D)中肺腺癌组织与正常组织m6A效应器的表达水平。T:肿瘤组织;N:正常组织;ns:无统计学差异;*P < 0.05;**P < 0.01;***P < 0.001;****P < 0.000, 1。Expression of m6A effector in lung adenocarcinoma tissue and adjacent normal tissue. A and B: Heat map for the expression levels of m6A effectors in LUAD and paired normal lung tissues of TCGA-LUAD cohort (A) and GSE40419 cohort (B); C and D: The expression level of m6A effectors in tumor tissues and paired normal lung tissues of TCGA-LUAD (C) and GSE40419 (D) cohorts. T: tumor; N: normal; ns: no statistical significance; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.000, 1.
The prognostic value of m6A effectors in LUAD. A: The forest plot of the univariate Cox regression analysis of 24 m6A effectors in TCGA-LUAD cohorts; B-E: The Kaplan-Meier curves of METT5 (B), HNRNPA2B1 (C), HNRNPC (D) and IGF2BP1 (E); F: The forest plot of the multivariate Cox regression analysis of the clinicopathological characteristics and the above four effectors.
m6A效应器对肺腺癌预后的影响。A:TCGA-肺腺癌队列中24个m6A效应器的单变量Cox回归分析;B-E:METT5(B)、HNRNPA2B1(C)、HNRNPC(D)和IGF2BP1(E)的Kaplan-Meier生存曲线;F:临床病理学特征和上述4个效应器的多变量Cox回归分析。The prognostic value of m6A effectors in LUAD. A: The forest plot of the univariate Cox regression analysis of 24 m6A effectors in TCGA-LUAD cohorts; B-E: The Kaplan-Meier curves of METT5 (B), HNRNPA2B1 (C), HNRNPC (D) and IGF2BP1 (E); F: The forest plot of the multivariate Cox regression analysis of the clinicopathological characteristics and the above four effectors.m6A修饰是甲基化和去甲基化的动态平衡过程,受效应器的表达和活性的影响。因此,我们在m6A效应器之间进行Spearman相关性分析,以进一步发现不同m6A效应器之间的相互作用。发现某些写入器和读取器之间存在显著的正相关关系,如KIAA1429和YTHDF3、METTL14和YTHDC1、ZC3H13和YTHDC1、KIAA1429和LRRPRC(图 4A-图 4D)。部分写入器也表现出显著的正相关,如KIAA1429和CBLL1、METTL14和ZC3H13,这反映了它们的功能协同作用(图 4E、图 4F)。
Correlation analysis between m6A effectors in lung adenocarcinoma. A: KIAA1429 and YTHDF3; B: METTL14 and YTHDC1; C: ZC3H13 and YTHDC1; D: KIAA1429 and LRPPRC; E: CBLL1 and KIAA1429; F: ZC3H13 and METTL14.
肺腺癌m6A效应器间相关性分析。A:KIAA1429和YTHDF3;B:METTL14和YTHDC1;C:ZC3H13和YTHDC1;D:KIAA1429和LRPPRC;E:CBLL1和KIAA1429;F:ZC3H13和METTL14。Correlation analysis between m6A effectors in lung adenocarcinoma. A: KIAA1429 and YTHDF3; B: METTL14 and YTHDC1; C: ZC3H13 and YTHDC1; D: KIAA1429 and LRPPRC; E: CBLL1 and KIAA1429; F: ZC3H13 and METTL14.
Survival analysis, enrichment pathway and immune infiltration analysis in distinct m6A modification pattern. A: Survival analyses of three m6A clusters; B-C: GSVA enrichment analysis in distinct m6A modification patterns; D: The stromal score and immune score of three m6A clusters; E: Abundance of the infiltration of 28 immune cells in the three m6A clusters. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.000, 1.
不同m6A修饰模式的生存分析、富集通路及免疫浸润分析。A:m6A聚类的生存分析;B-C:不同m6A修饰模式中的GSVA富集分析;D:m6A聚类的基质评分和免疫评分;E:m6A聚类中28种免疫细胞的浸润丰度。*P < 0.05;**P < 0.01;***P < 0.001;****P < 0.000, 1。Survival analysis, enrichment pathway and immune infiltration analysis in distinct m6A modification pattern. A: Survival analyses of three m6A clusters; B-C: GSVA enrichment analysis in distinct m6A modification patterns; D: The stromal score and immune score of three m6A clusters; E: Abundance of the infiltration of 28 immune cells in the three m6A clusters. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.000, 1.
m6A效应器的表达与免疫细胞浸润
由于m6A修饰模式影响了肿瘤微环境中的免疫细胞浸润,我们通过Spearman相关性分析进一步研究了24个m6A效应器的表达与肿瘤微环境中免疫细胞浸润程度的相关性(图 6A)。发现读取器LRPPRC与免疫细胞浸润程度呈负相关。通过对LRPPRC表达进行进一步的分组研究,结果表明,与LRPPRC低表达组相比,LRPPRC高表达组的基质评分和免疫评分均较低(图 6B)。信号通路富集分析显示,LRPPRC低表达组的免疫激活通路明显富集(图 6C)。我们进一步比较了28种免疫细胞的浸润情况,发现其中24种在LRPPRC低表达组的肿瘤组织中浸润更多,特别是树突细胞、CD8+细胞毒性T淋巴细胞以及巨噬细胞(图 6D)。同时,在LRPPRC低表达组的肿瘤细胞中树突细胞相关共刺激因子和黏附因子的表达也显著上调(图 6E)。我们进一步评估了LRPPRC对肺腺癌患者(GSE126044)抗程序性死亡受体1(programmed death 1, PD-1)免疫治疗应答率的影响,发现在8例LRPPRC高表达患者中仅1例应答,而在8例LRPPRC低表达患者中发现4例应答(图 6F)。因此,我们推测LRPPRC的表达可能通过抑制肿瘤微环境中免疫细胞的浸润和激活,从而影响了免疫治疗的应答,是一种潜在的免疫治疗标记物。
Correlation analysis of the infiltration of immune cells and each m6A effector. A: Heatmap of the spearman correlation coefficients between immune cells infiltration and the expression level of m6A effectors; B: Stromal and immune score between LRPPRC high and low expression group; C: The enrichment score of immune associated signal pathways between LRPPRC high and low expression group; D: The infiltration of immune cells in LRPPRC high and low expression group; E: The expression level of DCs-related costimulatory factors and adhesion factors on tumor cells with LRPPRC high or low expression; F: The relative expression level of LRPPRC and response to anti-PD1 immunotherapy of LUAD. ns: no statistical significance; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.000, 1.
免疫细胞浸润与m6A效应子的相关性分析。A:m6A效应器表达与免疫细胞浸润的相关性分析热图;B:LRPPRC高、低表达组的基质评分和免疫评分;C:LRPPRC高、低表达组免疫相关信号通路的富集评分;D:LRPPRC高、低表达组的免疫细胞浸润情况;E:树突细胞相关共刺激因子和黏附因子在LRPPRC高、低表达组间的表达水平;F:肺腺癌患者LRPPRC的相对表达水平与对抗PD1免疫治疗的应答。ns:无统计学差异;*P < 0.05;**P < 0.01;***P < 0.001;****P < 0.000, 1。Correlation analysis of the infiltration of immune cells and each m6A effector. A: Heatmap of the spearman correlation coefficients between immune cells infiltration and the expression level of m6A effectors; B: Stromal and immune score between LRPPRC high and low expression group; C: The enrichment score of immune associated signal pathways between LRPPRC high and low expression group; D: The infiltration of immune cells in LRPPRC high and low expression group; E: The expression level of DCs-related costimulatory factors and adhesion factors on tumor cells with LRPPRC high or low expression; F: The relative expression level of LRPPRC and response to anti-PD1 immunotherapy of LUAD. ns: no statistical significance; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.000, 1.
Analyses of LRPPRC expression, CD8+ cytotoxic T lymphocytes and CD68+ macrophages infiltration in TME by immunohistochemistry. A: Immunohistochemical staining of LRPPRC, CD8 and CD68 in LUAD tissues (original magnification, x100); B-C: The number of CD8+ cytotoxic T lymphocytes (B) and CD68+ macrophage (C) infiltration per filed was calculated in different group according to LRPPRC expression with 68 LUAD patients. *P < 0.05; **P < 0.01; ***P < 0.001.
免疫组化分析肿瘤微环境中LRPPRC表达、CD8+细胞毒性T细胞和巨噬细胞浸润。A:肺腺癌组织中LRPPRC、CD8和CD68的免疫组织化学染色(原始放大倍数,x100);B-C:根据LRPPRC表达分组,每视野CD8+细胞毒性T细胞(B)和CD68+巨噬细胞(C)的浸润数量。*P < 0.05; **P < 0.01; ***P < 0.001。Analyses of LRPPRC expression, CD8+ cytotoxic T lymphocytes and CD68+ macrophages infiltration in TME by immunohistochemistry. A: Immunohistochemical staining of LRPPRC, CD8 and CD68 in LUAD tissues (original magnification, x100); B-C: The number of CD8+ cytotoxic T lymphocytes (B) and CD68+ macrophage (C) infiltration per filed was calculated in different group according to LRPPRC expression with 68 LUAD patients. *P < 0.05; **P < 0.01; ***P < 0.001.