Analysis of expression data for LSD1
In our study, we analyzed the expression levels of human LSD1 across various TCGA tumors using the TIMER2 approach. As shown in Fig. 1A, The expression of LSD1 in the tumor tissues of BLCA (Bladder transitional cell cancer), BRCA (Breast invasive cancer), CESC (Cervical and endocervical carcinomas), CHOL (Cholangiocarcinoma), CRC (Colorectal carcinoma), EC (Esophageal cancer), GBM (Glioblastoma multiforme), SCCHN (Squamous cell carcinoma of the head and neck), LIHC (Liver hepatocellular cancer), LUAD (Lung adenomatous carcinoma), LUSC (Lung squamous cell cancer), PCA (prostatic cancer), READ (Rectum adenomatous carcinoma), STAD (Stomach adenomatous carcinoma) and CECU (Corpus endometrial carcinoma of the uterus) is higher than the adjacent control tissues.
In the GTEx dataset, we included the normal tissue as controls, the expression of LSD1 was compared between normal tissues and tumor tissues of BRCA (Breast invasive carcinoma), DLBC (Lymphoid tumor diffuse large B-cell lymphoma), LGG (Low grade glioma of the brain), THYM (Medullary thymoma), and UCS (Uterine Carcinosarcoma). The results showed a significant difference for these tumors (Fig. 1B).
We investigated the expression levels of LSD1 total protein using the CPTAC dataset, the results showed that the expression of LSD1 total protein was higher in the primary tissues of breast carcinoma, colon carcinoma, ovarian carcinoma, clear cell RCC, LUAD and UCEC (Fig. 1C) than in normal tissues. With HEPIA2, the "Pathological Stage Plot" module, it was observed that LSD1 expression correlated with pathological stages in cancers including HNSC, LIHC, OV (Ovarian serous cystadenocarcinoma), and SKCM (Melanoma of the skin and cutaneous) (Fig. 1D, all P < 0.05) but not in other types of cancer.
Survival analysis data
According to the expression levels of LSD1, we classified cancer cases into high expression group and low expression group, and mainly used TCGA and GEO datasets respectively to study the relationship between LSD1 expression and prognosis of patients with different tumors. As shown in Fig. 2A, high expression of LSD1 genes was associated with poor overall survival for cancers of ACC (P = 0.003), LIHC (P = 0.0058), and SARC (P = 0.012), while a low expression of LSD1 genes was associated with poor survival for COAD (Fig. 2A, P = 0.025) and KIRC (Fig. 2A, P = 0.026). According to the DFS (Disease-free Survival) analysis data (Fig. 2B), high LSD1 expression predicts a poor prognosis for ACC (P = 0.00015), KICH (P = 0.045), LGG (P = 0,017), and LIHC (P = 0.021) cases in TCGA. In addition, low expression of LSD1 gene was associated with poor prognosis of KIRC OS (Fig. 2B, P = 0.016).
Genetic alteration analysis data
We investigated the LSD1 genetic alteration status in different tumor samples from the TCGA cohorts in our study. As shown in Fig. 3A, the most frequent abnormalities of LSD1 (> 3%) were noted in patients with uterine corpus endometrial carcinoma with a primary diagnosis of "mutation". "Amplification" was the predominant type of CNA in uterine carcinosarcomas, which has a prevalence of ~ 2% (Fig. 3A). A copy number deletion of LSD1 was noted in cholangiocarcinoma patients with genetic alteration (~ 3% frequency) (Fig. 3A). Figure 3B shows the 3D structure of LSD1 protein. Figure 3C shows the types, sites, and number of LSD1 genetic changes. We noticed the most frequent alteration was the E477K modification in the amino_oxidase of LSD1.
Phosphorylation analysis data of proteins
A comparison of LSD1 phosphorylation levels between normal tissues and primary tumor tissues was also performed in our study. The CPTAC dataset was used to analyze six types of tumors (breast carcinoma, UCEC, clear cell RCC, ovarian carcinoma, colon carcinoma, and LUAD). In all primary tumor tissues, the S166 locus of LSD1 shows differentially expressed phosphorylation levels in comparison to normal tissues (Fig. 4A-F, all P < 0.05), followed by a differentially expressed phosphorylation level of the S69, S131, S131S137, S131Y135, S137, S849, Y135S137, Y135Y136 and Y136 locus for breast cancer (Fig. 4A, P < 0.05), the S131, S137, S855, Y135 locus for UCEC (Fig. 4B, P < 0.05), the S131, S131Y135 locus for clear cell RCC (Fig. 4C, P < 0.05), the S131, S131Y135 locus for clear cell RCC (Fig. 4C, P < 0.05), the S69, S131 locus for colon cancer (Fig. 4E, P < 0.05) and the S131, S131Y135 locus for LUAD (Fig. 4F, P < 0.05).
Analysis of immune infiltration data
Tumor-infiltrating immune cells are prominent components of the tumor microenvironment and are closely linked to the initiation, progress, and metastasis of cancer. The results showed that various tumor-infiltrating immune cells were modulated by cancer-associated fibroblasts in the stroma of the tumor microenvironment. In this study, the TIMER, CIBERSORT, CIBERSORT-ABS, QUANTISEQ, XCELL, MCPCOUNTER and EPIC algorithms were used to investigate the potential association between immune cell infiltration levels and LSD1 gene expression in diverse carcinoma types of the TCGA. After a series of analyses, we concluded that cancer-associated fibroblast_TIDE infiltration and LSD1 expression were statistical positively correlated in BRCA-LumA, ESCA, KIRC, PAAD and PCPG tumors, and statistical negatively correlated in THYM tumors (Fig. 5). In addition, we found a statistically significant relationship between LSD1 expression and the approximated infiltration value of cancer-associated fibroblasts_EPIC in TCGA tumors from CESC, HNSC, HNSC-HPV-, LUAD and LUSC (Fig. 5).
Analyse of partners associated with LSD1
We tried to look for LSD1-binding proteins and the expression-related genes of LSD1 for a series of enrichment analyses of the signal pathways to investigate the molecular and cellular mechanisms of LSD1 involvement in tumor development. By using the STRING tool, we identified 51 LSD1-binding proteins with experimental support. Figure 6A shows the interaction network of these proteins. Figure 6B showed that LSD1 expression levels were positively correlated with those of the DHX9 (R = 0.58), HNRNPR (R = 0.63), MRTO4 (R = 0.59), RCC2 (R = 0.59) and SNRNP40 (R = 0.59) genes (all P < 0.001). In addition, the heatmap data showed a positive correlation between LSD1 and the above five genes in most cancer types (Fig. 6C). The two datasets were combined to conduct KEGG and GO enrichment analyses. The KEGG results indicated that “RNA Transport”, “mRNA Surveillance Pathway” and “spliceosome” could all play a role in the effect of LSD1 on tumor pathogenesis (Fig. 6D). As shown in Fig. 6E, the GO enrichment analysis further showed that most of these genes are related to RNA metabolism or pathways, such as mRNA splicing (via spliceosome), RNA splicing, mRNA 3’-UTR binding, RNA transport, and Mrna surveillance pathway.