RNA-seq data from the TCGA (https://portal.gdc.cancer.gov/) UCEC project and GTEx database describe the differential expression of KNL1 in unpaired and paired samples. The Toil process uniformized the data (19). The TCGA level 3 HTSeq-FPKM (Fragments Per Kilobase Per Million) format was translated to the TPM (transcripts per million reads) format and log2-transformed. All final TCGA-based analyses were conducted using TPM-formatted data. Using GEOquery [version 2.54.1] (20), the differential analysis data for KNL1 in dataset GSE17025 (21, 22) were extracted from the GEO database. These data were obtained by removing probes corresponding to multiple molecules, and when probes corresponding to the same molecule were encountered, only the probe with the highest signal value was retained. The data were then normalized once more using the normalize Between Arrays function of the limma package [version 3.42.2] (23). Using the CPTAC database in UALCAN (http://ualcan.path.uab.edu) (24, 25), differential expression of the KNL1 protein in UCEC and normal adjacent tissues was determined. R was used for all statistical analyses and visualizations (version 3.6.3).

The DESeq2 package [version 1.26.0] and the STAT package [version 3.6.3] were used to conduct single-gene differential analysis and single-gene correlation analysis of KNL1 in the UCEC project utilizing the TCGA database (26). The findings of the single-gene differential analysis were used to generate volcano plots with the ggplot2 software [version 3.3.3]. |log₂ fold change (LogFC)|>1 and p.adj<0.05 were used as the thresholds for differentially expressed genes (DEGs). The STRING database was utilized to show the DEGs (27), the PPI network of DEGs was analyzed using the Cytoscape program, and the MCODE plugin was used to identify the HUB genes. The genes from the single-gene correlation analysis were then sorted by |Pearson value| in descending order, and the top 50 correlations were retrieved. The KNL1 single-gene coexpression heatmap was generated using the top 50 genes and the HUB gene by the ggplot2 [version 3.3.3] package.

In the TCGA UCEC project, gene set enrichment analysis (GSEA) was utilized to investigate the putative signaling pathways based on differential expression analysis (KNL1 high-expression vs. KNL1 low-expression samples). The reference gene set was h.all.v7.2.symbols.gmt [Hallmarks]. An adjusted p value <0.05 was considered significantly enriched. After screening the DEGs based on the threshold (|LogFC |>1 and p.adj<0.05), Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were performed to enhance the pathways associated with KNL1 in UCEC using the R packages “clusterProfiler” and “org.Hs.eg.db”. p.adj<0.05 was considered significantly enriched.

GSVA [version 1.34.0] was used to examine the relative infiltration levels of 24 immune cells (28). For the immune infiltration algorithm, ssGSEA was employed, and Spearman correlation analysis was applied. The markers for twenty-four immune cells were derived from an article in Immunity (29). The samples were then separated into low and high KNL1 expression groups, the enrichment scores of various immune cell infiltrates in the various subgroups were computed, and the analysis was conducted using GSVA software [version 1.34.0]. Finally, the correlation between KNL1 and CD47, CD273, and TNFRSF4 was computed, and ggplot2 software [version 3.3.3] was used to depict it.

The survival data of UCEC patients were statistically analyzed using the survival package [version 3.2-10], and the results were visualized using the survminer package [version 0.4.9] to plot the overall survival (OS), disease-specific survival (DSS), and progression-free interval (PFI) on Kaplan−Meier curves for the UCEC patients. Using the pROC package [version 1.17.0.1], ROC analysis was performed on the data to assess the accuracy of KNL1 for prognostication. All predictive data for the aforementioned survival study were from a Cell article (30). Finally, a dichotomous logistic regression model and clinical baseline datasheet were developed to predict the association between various clinicopathological characteristics and KNL1 expression.

Jilin University’s School of Nursing’s Ethical Review Committee authorized the present study (Changchun, China). Paraffin−embedded specimens were collected at the Second Hospital of Jilin University (Changchun, China) from 108 patients with UCEC and 15 normal controls diagnosed between December 2012 and December 2019. Patients were informed about the UCEC-related study and agreed to participate. The criteria for inclusion were: i) initially diagnosed with UCEC and treated with standard surgery and/or radiotherapy and/or chemotherapy according to the FIGO stage and pathological type of the individual patient; ii) the diagnosis of UCEC was determined by an experienced gynecological pathologist; iii) the postoperative pathology results were interpreted by an experienced gynecological pathologist using FIGO staging criteria (Version 2009); and iv) complete follow-up data were available. The exclusion criteria were: i) a personal history of other malignant tumors; ii) preoperative radiation, chemotherapy, or hormonotherapy; and iii) a secondary uterine tumor. As stated in Supplementary Table 4 , accessible clinical/pathological data were gathered from The Second Hospital of Jilin University’s Medical Record Database. All 108 patients with UCEC were followed up, and their OS was determined.

The human UCEC cell lines Ishikawa and HEC-1-A were purchased from iCell Bioscience Inc., Shanghai. Ishikawa cells were cultured with Minimum Essential Medium (MEM, product code iCell-0012, iCell) supplemented with 10% fetal bovine serum (product code FS301-02; TransGen), 1% nonessential amino acids (NEAA, product code iCell-01000, iCell) and 1% penicillin‐streptomycin (product code P1400, Solarbio). HEC-1-A cells were cultured with McCoy’s 5A medium (product code iCell-0011, iCell) supplemented with 10% fetal bovine serum and 1% penicillin‐streptomycin. The two cell lines were cultured at 37°C in a humidified atmosphere with 5% CO₂.

The lentiviral vector plasmid pLKO.1-Puro (product code FH1717; Hunan Fenghui Biotechnology Co., Ltd.) was utilized to construct the pLKO.1-Scramble and pLKO.1-shKNL1 plasmids. The interference sequences were 5’-GGUAAAAGUCCCAUAGAAATT-3’ for shKNL1 and 5’-GTATAAGTCAACTGTTGAC-3’ for shScramble. The lentiviruses used in this study were packaged using the 3 plasmid packaging system. After combining the lentiviral vector plasmids with the packaging plasmid PMD2.G (product code BR037, Fenghui), psPAX2 (product code BR036, Fenghui) and Lipofectamine™ 3000 transfection reagent (product code L3000150, Thermo Fisher), the complexed solution was introduced to HEK-293T cells (product code iCell-h237, iCell). The medium was collected and filtered using a 0.22 μm filter after 48 and 72 hours. The medium was then kept at 4°C for up to one week before use.

To generate stably transfected cell lines, Ishikawa and HEC-1-A cells were seeded into 6-well plates (300,000 cells/well). Subsequently, 24 hours later, 1 ml of medium containing the above lentivirus was added to each well. After 48 hours, the medium was changed. Cells infected with viruses encoding the puromycin resistance gene were selected in 2 μg/mL puromycin. One week of puromycin selection was continued prior to cell collection and subsequent analysis.

Similar to an earlier study (31), immunohistochemical (IHC) staining was conducted. After 24 hours of fixation in 10% formalin at room temperature, the samples were embedded in paraffin and sectioned to a thickness of 3 µm. The sections were immersed in EDTA retrieval buffer (catalog number AR0023; Wuhan Boster Biological Technology, Ltd.) and cooked in a microwave. Then, 5% bovine serum albumin (product code AR1006; Boster Biological Technology, Inc.) was applied at room temperature for 20 minutes to prevent nonspecific binding. The histological sections were stained overnight at 4°C with rabbit anti-KNL1 antibody (product code DF13491; 1:100; Affinity), rabbit anti-CD56 antibody (product code GB112671; 1:750; Servicebio), rabbit anti-CD4 antibody (product code GB11064; 1:1000; Servicebio), and rabbit anti-B3GAT1 antibody (product code GB113461; 1:1000; Servicebio). The secondary antibody was goat antirabbit IgG coupled with horseradish peroxidase (product code GB23204; 1:200; Servicebio), and the staining technique was performed at 37°C for 30 minutes. Reactive products were observed using 3,3’diaminobenzidene (Boster Biological Technology, Inc.) as the chromogen, and the sections were counterstained for 2 minutes at room temperature with 0.1% hematoxylin (Boster Biological Technology, Inc.). Under a light microscope (AE2000, Motic) with an objective magnification of x200 or x400, images of the stained sections were recorded. The positive cell density was evaluated with Image-Pro Plus 6.0 (Media Cybernetics, Inc.), and the findings are reported as the average optical density (AOD) values. Two experienced pathologists from the Pathology Department of the Second Hospital of Jilin University graded the IHC staining independently under double-blind conditions.

Total RNA was isolated from fresh frozen tissue and stably transfected cells using an EasyPure RNA kit (product code ER101-01; TransGen), and first-strand cDNA was synthesized using a cDNA synthesis kit (product code AT311-02; TransGen) following the manufacturer’s instructions. KNL1 transcription levels were determined by real-time PCR using the SYBR Green qPCR kit (product code AQ132; TransGen) according to the manufacturer’s instructions, with GAPDH serving as the internal reference gene. The following primers were used: KNL1 gene, 5’-GATGGGGTGTCTTCAGAGGC-3’ for forward and 5’-AGAGGACTCCTTGGGGGTTT-3’ for reverse; GAPDH gene, 5’‐GAAGGTGAAGGTCGGAGTC‐3’ for forward and 5’‐GAAGATGGTGATGGGATTTC‐3’ for reverse. An ABI-Q3 was used to conduct PCR at 94°C for 30 seconds, followed by 45 cycles of amplification at 94°C for 5 seconds, 51°C for 15 seconds, and 72°C for 10 seconds (Thermo Fisher Scientific, Inc.). The expression levels of the mRNA were measured using the 2^-ΔΔCt method (31).

Ishikawa, HEC-1-A, Ishikawa-shScramble, HEC-1-A-shScramble, Ishikawa-shKNL1 and HEC-1-A-shKNL1 cells were seeded into 96-well plates (3,000 cells/well). CCK-8 reagent (10 μl/well; product number BA00208, Bioss) was then added to each well 24, 48, and 96 hours later. After 1.5 hours of culture at 37°C, the absorbance of each well was measured at 450 nm with a microplate reader (E0226; Detie, Inc.

For the invasion experiment, a Transwell chamber (Labselect, product code 14342) was used to determine the invasive potential of the UCEC cells listed above. The chamber was covered with Matrigel (BD Biosciences, product code 356234) per the manufacturer’s instructions. A total of 3x10⁴ cells in 100 μL serum-free MEM or McCoy’s 5A were placed in the upper chamber, while 600 μL 10% FBS media-based medium was placed in the lower chamber. After 30 hours of treatment at 37°C, the residual cells on the top surface were removed with a cotton swab, and the invasive cells were stained with 10% Giemsa. An optical microscope was used to record the images (AE2000, Motic).

A wound-healing experiment was performed to assess the migratory capacity of the UCEC cells described above. Cells seeded in six-well plates (3x10⁵ cells/well) were scratched with a 200 μl pipette tip to create a linear wound. The dislodged cells were washed and removed with PBS. Photographs were obtained using a digital camera and an optical microscope (Motic Corporation) to observe the movement of cells into the wounded region at 24 and 48 hours. All micrographs were obtained at the same magnification at the same time for each cell type.

HEC-1-A and Ishikawa cells were seeded onto 6-well plates at a density of 100 cells/well. Ten days later, the formation of typical colonies was observed. The cells were fixed with methanol and stained with 10% Giemsa (Biotopped, China). The number of visible colonies was counted to evaluate the colony formation ability of the cells. All experiments were conducted in three replicates.

The statistical analyses were conducted with the mean of three independent tests plus the standard deviation (SD). Statistical analyses were conducted utilizing SPSS 23.0 or R version 3.6.3; differences between groups were examined using one-factor analysis of variance (ANOVA) followed by Dunnett’s post hoc test, Kruskal−Wallis test, or Student’s t test. When p<0.05, differences were judged statistically significant. The Spearman correlation coefficient was calculated to determine the correlation between KNL1 and CD4, CD56 and B3GAT1.

As shown in Supplementary Figure 1A , we found that in unpaired samples, the expression of KNL1 was higher in the following tumors than in normal tissues: adrenocortical carcinoma (ACC), bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), colon adenocarcinoma (COAD), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), brain lower grade glioma (LGG), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), skin cutaneous melanoma (SKCM), stomach adenocarcinoma (STAD), thyroid carcinoma (THCA), thymoma (THYM), uterine corpus endometrial carcinoma (UCEC), and uterine carcinosarcoma (UCS). Similarly, the expression of KNL1 was decreased in kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), acute myeloid leukemia (LAML), and testicular germ cell tumors (TGCTs) compared to normal tissues.

As shown in Supplementary Figure 1B , in paired samples, the expression of KNL1 in BLCA, BRCA, COAD, ESCA, HNSC, LIHC, LUAD, LUSC, PRAD, STAD and UCEC was higher than that in adjacent tissues. The expression of KNL1 in KIRC and KIRP was lower than that in adjacent tissues. The number of tumor samples used in the pancancer analysis is presented in Supplementary Table 1 . As shown in Figures 1A, B , in both unpaired and paired UCEC samples, the expression of KNL1 in tumors was higher than that in normal tissues. This point was validated by utilizing the mRNA and protein expression data of KNL1 in the GSE17025 and UALCAN databases, which were compatible with the results from the TCGA database, as shown in Figures 1C, D .

The AOD of 108 UCEC clinical samples and 15 normal tissues was measured by immunohistochemical staining. As shown in Figure 2A , KNL1 expression was different in tissues with different degrees of differentiation. The expression of KNL1 protein in normal tissues is low, and the expression of KNL1 in tumor tissues gradually increases with a gradual decrease in tumor differentiation. The expression levels of KNL1 protein in 108 UCEC samples and 15 normal samples are shown in Figure 2B . The expression of KNL1 was significantly increased in tumor tissues. As shown in Figures 2C–F , KNL1 expression was different in patients with different FIGO stages, different tumor invasion statuses, different histologic grades, and different lymphatic metastases. A KNL1 expression box diagram of patients with other clinical features is shown in Supplementary Figure 2 . Moreover, ROC curves of the protein expression data of KNL1 are shown in Figure 2G . The AUC=0.764, suggesting that KNL1 may be closely related to the occurrence and development of tumors.

The results of single-gene differential analysis are shown in the volcano plot in Figure 3A . There were 850 genes that satisfied the threshold of |LogFC|>1 and p.adj<0.05, under which 243 genes were highly expressed and 607 genes were poorly expressed. These 850 genes were imported into the STRING database to construct differential protein interaction networks, and a total of 46 HUB genes were identified (MELK, E2F7, SMC2, ANLN, HMMR, OIP5, CDCA2, PBK, RAD51AP1, CENPI, CKAP2L, KIF14, FBXO5, FAM83D, CENPF, DLGAP5, CCNE2, FOXM1, TOP2A, NCAPG, SGOL2, DEPDC1, ASPM, KIF23, KIF15, BUB1, KIF11, MCM10, BUB1B, KIF18A, ERCC6L, NEK2, ECT2, NEIL3, ATAD2, NUSAP1, E2F8, DEPDC1 B, SMC4, MAD2L1, CENPE, KIF20B, CCNA2, CLSPN, ESCO2, and ARHGAP11A), as shown in Figure 3B . After that, we performed a correlation analysis and created a coexpression heatmap utilizing the 50 genes with the strongest connection with KNL1, as shown in Figure 3C . The heatmap of coexpression between the HUB genes and KNL1 is shown in Figure 3D .

KNL1 and its differentially expressed genes were used for GO and KEGG functional enrichment analyses. GO functional enrichment analysis showed that in terms of “biological process”, pathways such as acute inflammatory response, humoral immune response, hormone metabolic process, chromosome organization involved in meiotic cell cycle, and meiotic cell cycle process were enriched. In terms of “molecular function”, significant enrichment occurred in the pathways of G protein-coupled receptor binding, serine-type endopeptidase inhibitor activity, cytokine activity, hormone activity, and cysteine-type endopeptidase inhibitor activity involved in the apoptotic process. In terms of “cellular component”, keratin filament, catenin complex, kinesin complex, mitotic spindle, condensed chromosome and other pathways were enriched, and the results are shown in Figures 4A, C and Supplementary Table 2 .

The results of KEGG functional enrichment analysis showed that steroid hormone biosynthesis, metabolism of xenobiotics by cytochrome P450, drug metabolism - cytochrome P450, pentose and glucuronate interconversions, etc., were enriched, as shown in Figures 4B, D and Supplementary Table 2 .

Finally, GSEA functional enrichment analysis was used to predict the function of KNL1 in the development of endometrial carcinoma, and it was found that KNL1 was closely associated with hallmark allograft rejection, hallmark complement, hallmark Kras signaling up, hallmark g2m checkpoint, hallmark mitotic spindle, hallmark mtorc1 signaling, hallmark e2f targets, hallmark myc targets v1 and other pathways, as shown in Figures 4E, F .

The relationship between the expression of KNL1 and the degree of infiltration of 24 immune cells was analyzed, and the results are shown in Figures 5A, C . The results showed that the expression of KNL1 had a significant positive correlation with the infiltration degree of Th2 cells, T helper cells, and Tcm cells, while the expression of KNL1 showed a significant negative correlation with the infiltration degree of pDCs, NK CD56bright cells, iDCs, and NK cells.

To validate the results of ssGSEA, we analyzed the correlation between the expression of KNL1 and the expression of various immune cell surface marker proteins and plotted a heatmap, shown in Figure 5B . The heatmap showed a strong correlation between KNL1 and CR2, CD1A, TPSAB1, B3GAT1, IL3RA, CD3D, and PTPRC, consistent with the previous analysis. Finally, as shown in Figures 5D–F , we analyzed the correlation between some common immunotherapeutic targets and KNL1 and found that the expression of KNL1 showed a significant positive correlation with the expression of CD47.

To verify the relationship between KNL1 expression and immune infiltration in patients with UCEC, immunohistochemical analysis of CD4, CD56 and B3GAT1 was performed using samples from 108 patients, shown in Figures 6A–D . The immunohistochemical results were used to analyze the correlation between KNL1 and CD4, CD56 and B3GAT1, and it was found that the expression of KNL1 was negatively correlated with the expression of CD56 and B4GAT1 and positively correlated with the expression of CD4, as shown in Figures 6E–G .

To determine the relationship between KNL1 expression and the prognosis of UCEC patients, we performed survival analysis using the prognostic data of UCEC in TCGA, and the results are shown in Figures 7A–C . We found that high expression of KNL1 was correlated with worse overall survival (OS), disease-specific survival (DSS) and progression-free interval (PFI). As shown in Figure 7D , we also performed a ROC analysis to test the accuracy of KNL1 expression in predicting patient outcome and found that the AUC=0.952, suggesting that KNL1 expression is highly accurate in predicting the outcome of UCEC patients.

After that, we analyzed the relationship between KNL1 expression and various clinical characteristics of UCEC patients, for which the baseline data table is shown in Supplementary Table 3 , and the results of the logistic analysis are shown in Table 1 . The results in both Supplementary Table 3 and Table 1 suggest that the expression of KNL1 is closely related to the histologic grade of UCEC patients.

Finally, in addition to analyzing the correlation between KNL1 expression and the prognosis of UCEC patients, we also used the prognostic data of GBMLGG, LGG, BRCA, KIRP, KIRC, and PAAD in the TCGA database to analyze the association of KNL1 expression with the prognosis of these tumors, and the results are shown in Supplementary Figures 3A–F . The results showed that high expression of KNL1 led to worse OS of patients with these tumors, suggesting that KNL1 may be closely related to tumor progression.

We then analyzed the relationship between KNL1 expression and clinical features using the clinical information of 108 previously collected samples to verify the relationship between KNL1 and the prognosis of UCEC patients, and the results are shown in Figures 7E, F . The results in Figure E show that high expression of KNL1 was correlated with a poor prognosis in these 108 clinical patients, which further verifies the relationship between KNL1 expression and the OS of patients. Figure F shows the results of logistic regression analysis, indicating that there is a relationship between KNL1 expression and FIGO stage and histologic grade. The results in Supplementary Table 4 also show that the expression of KNL1 is significantly related to the degree of histologic grade, tumor invasion, FIGO stage and the expression of Ki67 protein, which is consistent with the previous analysis.

After knocking down the expression of KNL1 in HEC-1-A and Ishikawa cell lines, the expression of KNL1 was confirmed to be significantly reduced, as shown in Figure 8A . We then performed CCK-8 assays and found that cell proliferation was significantly reduced after knockdown, as shown in Figures 8B, C . We also performed a wound-healing assay and found that the metastatic ability of cells with KNL1 knockdown was significantly weaker than that of the control group over time, as shown in Figures 8H–J . We also performed Transwell experiments and the invasive ability of cells with KNL1 knockdown was also significantly weakened, as shown in Figures 8F, G . Finally, we performed a colony formation assay and found that knockdown of KNL1 expression in cells was followed by a decrease in their colony formation ability Figures 8D, E ).