Integration of bulk RNA sequencing to reveal protein arginine methylation regulators have a good prognostic value in immunotherapy to treat lung adenocarcinoma

Abstract

Background: Given the differential expression and biological functions of protein arginine methylation (PAM) regulators in lung adenocarcinoma (LUAD), it may be of great value in the diagnosis, prognosis, and treatment of LUAD. However, the expression and function of PAM regulators in LUAD and its relationship with prognosis are unclear.

Methods: 8 datasets including 1798 LUAD patients were selected. During the bioinformatic study in LUAD, we performed (i) consensus clustering to identify clusters based on 9 PAM regulators related expression profile data, (ii) to identify hub genes between the 2 clusters, (iii) principal component analysis to construct a PAM.score based on above genes, and (iv) evaluation of the effect of PAM.score on the deconstruction of tumor microenvironment and guidance of immunotherapy.

Results: We identified two different clusters and a robust and clinically practicable prognostic scoring system. Meanwhile, a higher PAM.score subgroup showed poorer prognosis, and was validated by multiple cohorts. Its prognostic effect was validated by ROC (Receiver operating characteristic curve) curve and found to have a relatively good prediction efficacy. High PAM.score group exhibited lower immune score, which associated with an immunosuppressive microenvironment in LUAD. Finally, patients exhibiting a lower PAM.score presented noteworthy therapeutic benefits and clinical advantages.

Conclusion: Our PAM.score model can help clinicians to select personalized therapy for LUAD patients, and PAM.score may act a part in the development of LUAD.

Keywords: Arginine methylation; Immunotherapy; Lung adenocarcinoma; Tumor microenvironment.

Fig. 1

mRNA expression levels and prognostic value of 9 PAM regulators in LUAD in TCGA-LUAD. (A) The illustration shows the expression distribution of 9 PAM regulators between tumor and normal. (B) Univariate Cox regression analysis of 9 PAM regulators and overall survival in TCGA-LUAD. (C) Correlation heat map of 9 PAM regulators. (D) ROC analysis showed the diagnostic performance 9 PAM regulators in TCGA-LUAD. (E) GSEA-GO showed that 9 PAM regulators related signaling pathways in LUAD. (F) GSEA-KEGG identified 9 PAM regulators related signaling pathways in LUAD. (G) 9 PAM regulators expression are related to immune in LUAD. (*P < 0.05; **P < 0.01; ***P < 0.001).

Fig. 2

Molecular classification based on 9 PAM regulators expression. (A) Kaplan-Meier curve showed a significant overall survival difference between the 2 PAM.clusters. (B) Alluvial diagram showing the relationship between the 2 PAM.clusters and clinical characteristics. (C–F) The relationship between the 2 PAM.clusters and patients' status, patients' stage, patients' age, and patients' gender. (G) GO enrichment analysis, (H) KEGG enrichment analysis for the different expression genes between the 2 PAM.clusters (BP means Biological Process; CC means Cell Component; MF means Molecular Function).

Fig. 3

TME between the 2 PAM.cluster. (A) The relationship between the 2 PAM.clusters and TME. Red is cluster.A and blue is cluster.B. (B) The abundance of each TME infiltrating cell in 2 PAM.clusters. Blue is cluster.A and yellow is cluster.B. (C) The abundance of immune checkpoint condition in 2 PAM.clusters. (cluster.B vs cluster.A. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Fig. 4

PAM.score model was constructed based on 9 PAM regulators. (A) Differences in PAM.score among 2 PAM.clusters. (B) The number of high and low PAM.score patients in 2 PAM.clusters groups. (C) Kaplan-Meier curve showed a significant overall survival difference between high and low PAM.score groups. (D) The prognostic value of PAM.score. (E) GSEA GO identified high PAM.score groups related signaling pathways in LUAD. (F) GSEA KEGG identified high PAM.score groups related signaling pathways in LUAD. Each small cell below the horizontal coordinate represents the number of genes enriched in the pathway.

Fig. 5

TME between the high and low PAM.score groups. (A) The correlation between PAM.score and different immune cells. (B) The abundance of each TME infiltrating cell in high and low PAM.score groups. (C–F) The ESTIMATE score, stromal score, immune score, and tumor immunity levels in high and low PAM.score groups. (G) The expression of immune checkpoint condition between high and low PAM.score groups. The solid line represents the positive correlation and the dotted line represents the negative correlation. The depth of the color represents the size of the correlation, the darker the color, the greater the correlation. (High vs low *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Fig. 6

External validation of PAM.score. (A–H) Kaplan-Meier curve showed a significant overall survival difference between high and low PAM.score groups in GSE13213, GSE31210, GSE37745, GSE41271, GSE42127, GSE50081, GSE72094, and TCGA dataset.

Fig. 7

Evaluation of the PAM.score. (A) Univariate and multivariate Cox regression analysis of PAM.score in the LUAD. (B) The performance of PAM.score was compared with 21 published signatures.

Fig. 8

PAM.score in the role of anti-PD-1/L1 immunotherapy. (A) Differences in PAM.score among non-response and response groups. (B) Differences in TIDE among high and low PAM.score groups. (C) The proportion of non-response and response patients in low or high PAM.score groups. (D) Kaplan-Meier curve showed a significant overall survival difference between low and high PAM.score groups in IMvigor210 cohort. (E) The prognostic value of PAM.score in IMvigor210 cohort. (F) Distribution of PAM.score in CR, PR, SD, and PD groups. (G) Differences in PAM.score among CR/PR and SD/PD groups. (H) The number of CR, PR, PD, and SD patients in high and low PAM.score groups. (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Fig. 9

Sc-RNA sequencing analysis in GSE149655. (A) The clusters of GSE149655 based on tSNE. (B) The annotated cell types of GSE149655 based on tSNE. (C) The top 5 markers genes in these eight cell types. (D) KEGG revealed the enrichment pathways of eight cell types.

Fig. 10

Sc-RNA sequencing analysis reveal the PAM.score on single cell level. (A) The distribution of 9 PAM regulators in eight cell types. (B) The different of 9 PAM regulators between normal cells and LUAD cells. (C) The different of PAM.score between normal cells and LUAD cells.

Fig. S1

(A-C) Unsupervised consensus clustering based on 9 PAM regulators expression in a meta cohort.

Fig. S2

The expression of 9 PAM regulators in PAM.cluster.A group and PAM.cluster.B group. (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Fig. S3

(A-D) The ESTIMATE score, stromal score, immune score, and tumor immunity levels in PAM.cluster.A group and PAM.cluster.B group.

Fig. S4

The expression of 9 PAM regulators in high PAM.score group and low PAM.score group. (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).