The expression profile of LIFR between tumor and adjacent normal tissues was detected using TIMER2 (http://timer.cistrome.org/) and the GTEx data was obtained from GEPIA2 (http://gepia2.cancer-pku.cn/#analysis).

The relationship between LIFR expression and different pathological stages for various cancers was analyzed via GEPIA2.

The data of the total protein level of LIFR was acquired from UCLCAN web (http://ualcan.path.uab.edu/analysis-prot.html).

We obtained the OS and DFS significance data of LIFR across all tumor types using “Survival Map” module of GEPIA2. We detected the effect of LIFR expression on OS and DFS in various cancer types. The low-expression and high-expression cohorts was splatted using cutoff-low (50%) and cutoff-high (50%) values as the expression thresholds. The survival plots were explored via the “Survival Analysis” module of GEPIA2.

The correlation between LIFR and prognosis including DFI, DSS and PFI was estimated by Sanger box (http://sangerbox.com) (19).

We used Kaplan-Meier analysis to analyze the different survival rates between the high- and low- expression groups to assess the prognostic value of LIFR. After adjusting for age and tumor stage, multivariable Cox regression analysis was used to explore the correlation between LIFR expression and DFI, DSS, and PFI.

The genetic alternation data was collected from cBioPortal tool (https://www.cbioportal.org/) using “Quick Selection” and “TCGA Pan-Cancer Atlas Study” options. Then the mutation type, mutation frequency, and copy number variation were obtained by using “Cancer Type Summary” module. We displayed the mutated site information of LIFR in the schematic diagram of the protein structure or the 3D (Three-dimensional) structure via the “Mutations” module. We acquired the data of OS, DFS, and PFS differences for various cancer types with or without LIFR genetic alternation, and generated Kaplan-Meier plots with log-rank P-value.

The correlation between LIFR expression and ESTIMATE score, immune score and the stromal score was evaluated via Sangerbox using ESTIMATE.

The correlation between immune infiltration and LIFR expression across various tumor types was analyzed by “Immune-Gene” module of the TIMER2 database. The association of LIFR expression with immune infiltration, such as macrophages, endothelial cells, CAFs, CD4⁺ T cells, and CD8⁺ T cells, was estimated applying the TIMER, QUANTISEQ, MCPCOUNTER, CIBERSORT, CIBERSORT-ABS, XCELL, and EPIC algorithms.

The protein-protein interaction network was conducted by STRING (https://string-db.org/) setting the following main parameters: meaning of network edges (“evidence”), active interaction sources (“experiments”), minimum required interaction score [“Low confidence (0.150)”], and max number of interactors to show (“no more than 50 interactors” in 1st shell).

The top 100 LIFR-correlated targeting genes were obtained using the “Similar Gene Detection” module of GEPIA2. Then we performed a pairwise gene Pearson correlation analysis of LIFR and selected genes. Furthermore, we generated the heatmap data of the selected genes using “Gene-Corr” module of TIMER2.

We combined the LIFR-binding and LIFR-correlated genes to performed KEGG pathway analysis using Sanger box database.

UCEC samples were obtained from the Department of Obstetrics and Gynecology in the Zhengzhou Central Hospital. The use of tumor excisions was consented to by the Medical Ethics Committee of Zhengzhou Central Hospital. LIFR antibody was obtained from the Proteintech company (22779-1-AP).

LIFR stable overexpressed UCEC cell line Ishikawa were seeded in 96‐well plates for cell viability assay. CCK8 reagent was added to incubate at 37°C for 2h. The data was calculated according to the reagent instructions. The absorbance of each sample was measured at 450 nm. The CCK8 proliferation assay have been described previously in detail (20).

For Transwell migration assay, LIFR stable overexpressed UCEC cell line Ishikawa were separately placed in the top chamber of transwell chambers (8‐μm BioCoat Control Inserts, Corning Costar). The lower chamber was filled with 500 μl DMEM supplemented with 10% FBS. After 24 hours incubation at 37°C, the cells were fixed and stained. The cells in the top chambers were removed and counted. For invasion assay, cells were plated in the matrigel‐coated chamber and the migration assay was performed. The cell migration and invasion assays has been described previously in detail (21).

To identify the possible role of LIFR in carcinogenesis, we first assessed the differential expression of LIFR in different cells from normal and tumor tissues. LIFR showed the highest expression in midbrain, followed by basal ganglia and thalamus, with low tissue specificity ( Figure S1A ), while low expression of LIFR was detected in most tumor cell lines, with higher expression in AF22 (brain), SCLC-21H (lung) and HHSteC lines (mesenchymal) ( Figure S1B ). Next, we analyzed LIFR expression across 33 tumor types in TCGA via the TIMER database. As shown in Figure 1A , compared with normal tissues, we observed significant downregulation of LIFR in BLCA (Bladder Urothelial Carcinoma), BRCA (Breast invasive carcinoma), CESC (Cervical squamous cell carcinoma and endocervical adenocarcinoma), CHOL (Cholangiocarcinoma), COAD (Colon adenocarcinoma), ESCA (Esophageal carcinoma), HNSC (Head and Neck squamous cell carcinoma), KICH (Kidney Chromophobe), KIRC (Kidney renal clear cell carcinoma), KIRP (Kidney renal papillary cell carcinoma), LIHC (Liver hepatocellular carcinoma), LUAD (Lung adenocarcinoma), LUSC (Lung squamous cell carcinoma), PRAD (Prostate adenocarcinoma), READ (Rectum adenocarcinoma), STAD (Stomach adenocarcinoma), THCA (Thyroid carcinoma), and UCEC (Uterine Corpus Endometrial Carcinoma) ( Figure 1A ).

Because data on normal tissue distribution was not available for some tumor types in TCGA, we next analyzed differential expression of LIFR between normal and tumor tissues via the GTEx database. Downregulated expression of LIFR in tumor tissues was seen in ACC (Adrenocortical carcinoma), LAML (Acute Myeloid Leukemia), OV (Ovarian serous cystadenocarcinoma), TGCT (Testicular Germ Cell Tumors), and UCS (Uterine Carcinosarcoma), while there was a significant upregulation in LGG (Brain Lower Grade Glioma) and THYM (Thymoma) ( Figure 1B ). For other tumors, such as DLBC (Lymphoid Neoplasm Diffuse Large B-cell Lymphoma) and SARC (Sarcoma), there were no significant differences ( Figure S1C ). Overall, LIFR was significantly more weakly expressed in multiple cancers than in the corresponding normal tissues.

We also assessed LIFR expression at the protein level and found significant downregulation in BLCA, BRCA, CHOL, COAD, ESCA, HNSC, KICH, KIRC, KIRP, LIHC, LUAD, LUSC, PRAD, READ, STAD, THCA and UCEC tumor tissues ( Figure 1C ).

Finally, we sought correlations between LIFR expression and tumor pathological staging, which suggested stage-specific alternations in LIFR for a few tumors, such as CESC, KIRC and THCA ( Figure 1D ), while in most tumor types we found no significant correlations ( Figure S2 ).

We investigated the prognostic value of LIFR and found that higher levels correlated with better overall survival (OS) for tumors of KIRC, KIRP, and LUAD within the TCGA datasets, but the opposite result for STAD( Figure 2A ). Analysis of disease-free survival (DFS) showed a correlation between highly expressed LIFR and a worse prognosis for ACC, BLCA and STAD, although it was a protective factor for KIRC, PRAD and THCA ( Figure 2B ).

To further investigate the prognostic potential of LIFR, we analyzed some other prognostic indicators, including disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI) via the Kaplan–Meier method and univariate Cox regression. Cox proportional hazards analysis indicated that high LIFR expression led to a poor DFI in CESC and ACC. On the contrary, elevated LIFR resulted in better DFI in BRCA, KIRP, PRAD, THCA and LIHC ( Figure 3A ). Patients with low LIFR had better DFI in ACC, CEAC, while in patients with BRCA, KIRP and LIHC, low LIFR expression was related to poorer DFI ( Figure 2C ).

Moreover, higher LIFR was significantly related to a poorer DSS in STAD and ACC. In contrast, low LIFR corelated with poorer DSS in KIRC, KIRP, LUAD and BRCA ( Figure 3B ). Kaplan-Meier analysis indicated that low LIFR was significantly associated with better DSS in ACC and PAAD, but high LIFR expression correlated with a better DSS for patients with BRCA, KIRC, KIRP, LUAD and SKCM-M ( Figure 2D ).

Regarding associations between LIFR expression and PFI, the results demonstrated that higher expression of LIFR was related to a poorer PFI of patients with ACC, STAD, and BLCA, but in patients with KIRC, KIRP, PRAD and THCA, LIFR expression exhibited the opposite correlation with prognosis ( Figure 3C ). Kaplan–Meier survival curves with were shown in Figure 2E . These data suggested that LIFR was an independent prognostic marker for many cancers.

The accumulation of genetic alternations influences human cancer development. Thus, the genetic alternations of LIFR were investigated in various human cancer samples. Figure 4A showed that the highest frequency of LIFR alternation (>13%) was observed in SKCM for “mutation”, whereas DLBC exhibited the highest incidence of “amplification” (>9%) of copy number alteration (CNA). The types, case numbers and location of LIFR genetic alternations were depicted in Figure 3B . Missense mutation of LIFR was the main type of genetic alternation, with E391K detected in 3 cases of SKCM, 1 case of GBM, 1 case of STAD and 1 case of COAD ( Figure 4B ). This mutation induced a frame shift E (Glutamic acid) to K (Lysine) at the position 391 of LIFR protein. Next, we examined the 3D structure of the LIFR protein with this mutation ( Figure 4C ). Further, we investigated potential correlations between genetic alternations to LIFR and the clinical prognosis of cancers ( Figure 4D ). ACC cases with LIFR alternations had a poor prognosis in terms of OS, DSS and PFS (progression free survival). LIFR alternations were also correlated with poor DSS in CESC, with poor DSS in COAD, with poor OS in KIRC, and with poor DSS in STAD. However, LIFR alternations correlated with better DFS and PFS in UCEC. TMB (Tumor mutational burden) and MSI (Microsatellite instability) are crucial factors reflecting prognosis and immune response. Correlations between LIFR expression and TMB and MSI may be a potential biomarker for clinical immunotherapy response in patients with tumors with different LIFR expression patterns. As presented in Figure S3 , we found strong correlations between LIFR expression and TMB and MSI in most cancer types ( Figures S3A, B ). We then explored the correlation of LIFR expression with genetic alterations in various tumor types. For Simple Nucleotide Variation (SNV), we found no significant correlation in most tumor types, with the exception of UCEC ( Figure S3C ). In terms of Copy Number Variation (CNV), we detected correlations of LIFR expression with genetic alterations for a few tumors, such as LGG, LUAD, BRCA, SARC, PRAD, MESO, OV, and BLCA ( Figure S3D ).

DNA methylation directly affects cancer occurrence and progression. 11 probes were used to explore DNA methylation of LIFR ( Figure 5A ). We observed decreased DNA promotor methylation of LIFR in BLCA, BRCA, CESC, HNSC, LIHC, LUSC, PRAD, TGCT and UCEC tumor tissues. In contrast, significantly increased methylation of LIFR DNA promotor was observed in COAD, LUAD and READ tumor tissues according to the UALCAN database ( Figure 5B ). Further, we Further, we applied Kaplan-Meier survival analysis to explore relationships between LIFR methylation and patient prognosis (22). LIFR methylation was a protective factor for BLCA, KIRC and UCEC. Moreover, we observed a correlation between low LIFR methylation level and poor prognosis for KIRP and LUAD ( Figure 5C ). We then examined the relationship between DNA methylation of LIFR and LIFR expression. LIFR methylation correlated significantly with gene expression at multiple probes in many cancer types ( Figure S4 ). These data indicated that the DNA methylation might be not the only reason of abnormal LIFR expression. Other possibilities may contribute to the abnormal expression of LIFR, which need further exploration.

We quantified tumor purity based on the estimation score, which was used to estimate stomal and immune components in tumor tissues. LIFR was significant positively related to the estimate score in several tumors, such as ESCA, COAD, STAD, HNSC, LIHC, SKCM, BLCA, READ, PAAD, DLBC and CHOL, but there was a negative correlation in GBM, UCEC, KIRP, THCA, ACC ( Figure S5 ). Regarding the immune score, there was a significantly positive correlation in ESCA, COAD, STAD, HNSC, BLCA, READ, PAAD and CHOL, while there was a clear negative correlation in GBM, UCEC, KIRP, THCA and ACC ( Figure S6 ). Additionally, LIFR was positively correlated with the stromal score in BRCA, ESCA, COAD, STAD, HNSC, LIHC, SKCM, BLCA, READ, PAAD, USC, LAML and CHOL ( Figure S7 ). These results indicated that elevated LIFR was usually accompanied by immune infiltration in the tumor microenvironment.

Next, we determined the relationship between LIFR and immunoregulation-related genes, and saw clear correlations in most cancer types ( Figure 6 ). The correlation between LIFR and immune checkpoints was shown as Figure 7 . In most tumors, except LAML, UCS, SKCM, UVM, MESO and SARC, significant correlations were present between the expression of LIFR and recognized immune checkpoints such as EDNRB (Endothelin receptor type B), C10orf54 (V-type immunoglobulin domain-containing suppressor of T-cell activation), HMGB1 (High mobility group protein B1), CX3CL1, TNFSF4 (Tumor necrosis factor ligand superfamily member 4), BTN3A1 (Butyrophilin subfamily 3 member A1), ENTPD1 (Ectonucleoside triphosphate diphosphohydrolase 1) and TLR4 (toll-like receptor 4), suggesting a potential correlation between LIFR and known immune checkpoints. Notably, LIFR was positively correlated with EDNRB in all tumors ( Figure 7 ).

The infiltration of immune cells in tumors intimately affects their initiation, progression and metastasis (23, 24). Therefore, we sought a correlation between LIFR expression and various immune infiltrates across human cancers ( Figure 8 ). Overall, LIFR was positively related to many kinds of cells in the microenvironment, including endothelial cells, cancer-associated fibroblasts, monocytes, mast cells and hematopoietic stem cells. However, LIFR was negatively correlated with natural killer T cell (NKT) abundance. This profile suggested that LIFR has a critical role in immune-oncological interactions. Interestingly, in some tumors, the trend of the relationship was subtly different because of different immune infiltration ratios in various tumors.

To identify the mechanism of LIFR action in tumorigenesis and cancer progression, we explored targeting LIFR-interacting proteins and LIFR-expression related genes and performed functional enrichment analysis. We identified several experimentally detected LIFR-binding proteins from the STRING dataset. The interaction network was shown in Figure 9A . Based on the GEPIA2 tool, we acquired genes related to LIFR expression. In Figure 9B , LIFR was seen to be positively correlated with BBS2 (Bardet-Biedl syndrome 2 protein), RANBP3L (Ran-binding protein 3-like), SPARCL1 (SPARC-like protein 1), WASF3 (Actin-binding protein WASF3) and ZHX3 (Zinc fingers and homeoboxes protein 3) ( Figure 9B ). The heatmap showed positive correlations in most tumor types ( Figure 9C ). We then conducted functional enrichment analyses by combining the two datasets. The results suggested that “JAK-STAT signaling pathway”, “signaling pathways regulating pluripotency of stem cells” and “Wnt signaling pathway” might be related to the tumor pathogenesis function of LIFR ( Figure 9D ).

To clarify whether LIFR is an anti-tumorigenic factor in UCEC, we investigated its expression in 7 pairs of UCEC clinical samples by Western blot. LIFR was expressed at a low level in UCEC tumor tissues ( Figure 10A ). We overexpressed LIFR in the human UCEC cell line Ishikawa and the efficiency of LIFR expression was confirmed as shown in Figure 10B . CCK8 assay showed that the overexpression of LIFR decreased the viability of Ishikawa cells in vitro ( Figure 10C ).

​To determine whether LIFR is a functional gene in UCEC cells, we estimated cancer cell migrative and invasive abilities by Trans-well migration/invasion assays. We investigated whether LIFR overexpression repressed the migrative and invasive capacities of Ishikawa. The migration and invasion of UCEC cells were remarkably suppressed upon LIFR overexpression in Ishikawa ( Figure 10D ), indicating that LIFR represses the progression and aggression of UCEC.