Large-Scale Analysis and Clinical Samples Reveals the Specic Clinical and Immune Features of KDM4B in UCEC

Background: Uterine Corpus Endometrial Carcinoma (UCEC) ranks fourth among female cancers in the world. Frustratingly, the 5-year survival rates for advanced patients are only 17%. KDM4B is overexpressed or dysregulated in a variety of cancers and could be associated with tumor progression and poor prognosis. Therefore, we performed bioinformatics analysis and in vitro assays to assess the role of KDM4B in UCEC. In addition, its relevance to immune cells in the tumor microenvironment was explored. Methods: The mRNA level and protein level of KDM4B in UCEC was evaluated using the TCGA, HPA and GEO database. Immunohistochemistry and western blotting were used to verify the protein expression level of KDM4B in two batches of clinical samples. Kaplan-Meier curves, Univariate and multivariate analysis were used to assess the correlation between KDM4B expression and prognosis. GO and KEGG were used to predict the function and mechanism of KDM4B, and four immunity related database were used to explore their relevance to the tumor immune microenvironment. Results: Firstly, the present study showed that KDM4B was signicantly overexpressed in UCEC from several databases at the mRNA and protein levels, respectively. Immunohistochemistry and western blotting conrmed the abnormally overexpression of KDM4B. In addition, upregulation of KDM4B was associated with different clinicopathological prognostic factors. Secondly, overexpression of KDM4B was also associated with shorter OS and PFS. Univariate and multivariate analyses conrmed that KDM4B was an independent prognostic factor for poor prognosis. Then, GO and KEGG analysis revealed that KDM4B is enriched in biological processes and cellular signaling pathways closely related to immunity. Finally, KDM4B expression was closely associated with immune cell inltration and immune checkpoints and it may be a new immune-related potential oncogene in UCEC. Conclusions: KDM4B may be a new potential oncogene for UCEC patients. It is of clinical that


Background
Uterine Corpus Endometrial Carcinoma (UCEC) ranks fourth among female cancers in the world. Its incidence is gradually increasing with increasing obesity and aging, affecting 300000 women worldwide each year (1,2). Traditionally, it is divided into two histological types: estrogen-dependent (type I) and nonestrogen-dependent (type II) (3). In the era of molecular biology, it has been classi ed into four distinct molecular subgroups: DNA polymerases ε Hypermutated (POLE ultra-mutated), MSI (hypermutated), low copy number (CN low) and high copy number (CN high). Conventional treatments for UCEC include surgery, radiotherapy, chemotherapy and hormone therapy. It is frustrating that postoperative recurrence, resistance to chemotherapy and low response to hormone therapy in phase III and IV patients resulted in a 5-year survival rate of only 68% and 17% respectively (4). Therefore, it is crucial to determine the prognostic biomarkers of UCEC and develop more effective and novel treatments for patients with advanced stages.
The KDM4 (lysine demethylase 4) subfamily consists mainly of four proteins (KDM4A-D), all harboring the Jumonji C structural domain (JMJC), but with different substrate speci cities. The KDM4 protein is overexpressed or dysregulated in a variety of cancers, cardiovascular diseases and mental retardation, and is a potential therapeutic target (5). KDM4B is a newly identi ed member of the KDM4 subfamily, which is characterized by the JMJC domain. KDM4B speci cally targets the trimethylated lysine 9 of histone H3 (H3K9) for demethylation at pericentric heterochromatin and euchromatin(6, 7). Recent studies have suggested that KDM4B facilitates in the regulation of PI3K(8), TGF-β, Notch, and Wnt/βcatenin pathways (9) during malignant transformation of cancer and researches also indicated that KDM4B was a potential biomarker for targeted therapy of these cancers (10,11). Encouragingly, recent studies have reported that KDM4B-HOXC4-PD-L1 axis played an indispensable role in immune evasion of colorectal cancer cells (12). Therefore, the immune function of KDM4B in gynecological malignancies, especially in UCEC, has aroused great interest for us.
Leveraging sequencing technologies and bioinformatics to gain insight into unique molecular and genomic features of UCEC opens a new journey for its targeted therapy, immunotherapy and other precision therapies. In 2018, the US FDA approved anti-PD-1, which may bene t 20-30% of patients with advanced UCEC, thereby achieved milestone progress to immunotherapy in gynecologic tumors. In the era of precision medicine, the new classi cation of UCEC not only provides different prognostic information, but also potentially enables the selection of drug treatments based on the response rate of different subgroups. Due to its high mutation burden and immune invasion, immune checkpoint inhibition strategy is the most attractive candidate for the treatment of UCEC (13). Therefore, combining immunotherapy with gene targeting therapy is a new therapeutic prospect for UCEC, and there is an urgent need to nd new biomarkers for the e cacy and prognosis assessment of immunotherapy and comprehensive treatment of UCEC.
Here, the TCGA and the GEO database were used for the rst time for a pan cancer study of KDM4B.
Meanwhile, we also tried to explore the potential molecular mechanisms of KDM4B in cancer initiation and clinical prognosis of UCEC from the aspects of gene expression, survival status, gene mutation, immune in ltration and related cellular signaling pathways. Most importantly, we collected clinical patient samples for validation and utilized multiple databases for comprehensive analysis of KDM4B gene expression level and immune cells in ltration in UCEC and correlated these data with clinical outcomes and prognosis of UCEC patients. Our results suggested that KDM4B is highly expressed in cancer tissues and affects the clinical prognosis of patients with UCEC. Excitingly, this study found that KDM4B expression was closely associated with the tumor immune microenvironment of UCEC, including immune cell in ltration and immune checkpoints, suggesting that it may serve as a new prognostic biomarker for UCEC patients and provide a basis for immunotherapy.

Data collection
The microarray data from the GSE36389 (UCEC = 13, Normal = 7) and GSE115810 (UCEC = 24, Normal = 2) dataset of Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) were used for subsequent studies. Gene expression and related clinical information data for 100 patients with UCEC and 31 normal controls were obtained from the Cancer Genome Atlas database (TCGA, https:// portal.gdc.cancer.gov/) for follow-up studies. Based on the Human Protein Atlas database (HPA, https://www.proteinatlas.org/) which is an immunohistochemical database and the UALCAN database (http://ualcan.path.uab.edu/index.html), the differential expression of KDM4B between UCEC and normal controls were explored from the protein level. Finally, cBioportal (http://www.cbioportal.org/) is a powerful tool for studying different types of genetic alterations in tumors based on TCGA database and more than 200 published studies. We explored the KDM4B genomics data, including mutations and copy alterations, using this database.

Patients, Treatments and Follow-Up
128 patients who were diagnosed with UCEC were collected and analyzed in this research between August 2011 and September 2013 in the Harbin Medical University Cancer Hospital. Inclusion criteria: 1) patients with UCEC diagnosed pathologically, 2) all patients underwent standard staging surgery for UCEC, 3) all of them received standard postoperative adjuvant treatment based on the physician's recommendations. Exclusion criteria: 1) patients with UCEC uncon rmed pathologically, 2) patients who received preoperative radiotherapy or chemotherapy, 3) patients with incompletely documented UCEC.
Specimens were taken from the operating room for surgical resection. This study was reviewed and rati ed by the Ethics Committee of Harbin Medical University Cancer Hospital before starting the study and collecting samples. Written informed consent of all patients was obtained. The end of follow-up time was October 10, 2017, while among these 126 UCEC patients, two was lost to follow-up. The Histological type, FIGO stage, Histological grade, Lymph node metastasis and Muscular layer depth of Invasion of all tumor cases were estimated by skilled pathologists of Harbin Medical University Cancer Hospital. All relevant clinical characteristics were recorded and analyzed in the Table S3. 3. Immunohistochemistry KDM4B expression at the protein level were detected by IHC (Immunohistochemistry). Cut the para nembedded UCEC tissue into 4 µm-thick sections. After that, the tissue sections were depara nized in xylene and gradient ethanol. Rinse the sections with distilled water. Place the slices in a 0.3% hydrogen peroxide solution, shake and incubate for 10 minutes at room temperature to block endogenous peroxidase. The slices were placed in an EDTA solution (0.01mol/L, pH = 9.0), and then autoclaved for 4 minutes (121°C). KDM4B speci c rabbit anti-human monoclonal antibody (dilution 1:200; Abcam, Cambridge) was added to the slices, and then incubated overnight in a refrigerator at 4°C. Then the sections were washed three times with PBS. Then add goat anti-rabbit secondary antibody (dilution 1:5000 Abcam, Cambridge) and incubate for 1 hour at room temperature on a shaker. Put the slices into DAB solution and add hematoxylin dye solution. Put it into a gradient of ethanol for dehydration and seal with a neutral resin. After that, the speci c staining of the sections, the percentage of positively stained tumor cells and the intensity of staining were observed under a light microscope.

Evaluation of Immunohistochemical Staining
Semi-quantitative analysis was performed on the protein expression level of KDM4B according to the total combined scores of staining intensity and percentage of positive-staining tumor cells. The staining intensity was graded as follows: 0, 1, 2 and 3 respectively represent no staining, weak staining, moderate staining and strong staining. percentage of positive-staining tumor cells was graded as follows: 0, 1, 2, 3 and 4 respectively represent < 5% positive cells, 5%-25% positive cells, 25%-50% positive cells, 50%-75% positive cells and > 75% positive cells. The scoring process of IHC was carried out twice by two experienced pathologists independently. Finally, the percentage of positive cells is added to the intensity score and recorded as the Total Score of the KDM4B protein expression level, which ranged from 0 to 7. All the scores were recorded in the correlation table, and then correlation analysis was done between the clinical characteristics and the KDM4B expression. In addition, the total score greater than or equal to 4 was de ned as high expression, and a total score less than or equal to 3 was de ned as low expression.

Western Blotting
Collected 30 cases of UCEC and 10 cases of normal endometrial tissue samples from the Harbin Medical University Cancer Hospital, from July 2016 to May 2017 and were used to detect and verify the protein expression of KDM4B by Western Blotting. 30 frozen UCEC tissue samples and 10 frozen normal endometrial tissues were lysed to extract the protein suspension. The mixture was then centrifuged at 12,000g for 15 minutes at 4ºC to collect the supernatant. The quanti cation of protein concentration was performed according to Bradford kit (Thermo Scienti c, Waltham, USA). The protein was separated by electrophoresis in a 10% SDS-PAGE gel, and the protein was transferred to a PVDF membrane treated with methanol. Place the PVDF membrane in a TBS-T blocking solution containing 5% skimmed milk powder and shake on a shaker for 1 hour. The membrane was incubated with primary antibody, anti-KDM4B (Abcam, Cambridge, USA) at 4℃ overnight, which was diluted in buffer (Beyotime, CHINA). The membrane was washed, followed by addition of horseradish peroxidase-conjugated rabbit secondary antibodies to incubate for 1h at room temperature. Afterwards, the membrane was washed again. The experiment was conducted in triplicate. The blots were stained using chemiluminescent matrice and imaged with a charge-coupled camera LAS4000 (Fuji lm, Tokyo, Japan).

Immune databases (TIMER, CIBERSORT, TISIDB, EPIC)
Tumor immune estimation resource (TIMER, https://cistrome.shinyapps.io/timer) is a powerful tumor immunology and genetics-related database, including many types of information such as gene expression, mutation and copy number variation. The database is based on a total of 10,897 patient samples from 32 different types of cancers in the TCGA database, and further immune-related extended research has been done (14). In this study, via the TIMER database, the correlation between KDM4B expression and in ltration of six different immune cell types was evaluated (CD8 + T cells, CD4 + T cells, B cells, macrophages, dendritic cells and neutrophils). The survival module can output a Kaplan Meier plot demonstrating the interrelationship of clinical survival outcome and immune cell in ltration or gene expression. Then, the three databases of CIBERSORT (http://cibersort.stanford.edu/.), TISIDB (http://cis.hku.hk/TISIDB/index.php), EPIC (https://gfellerlab.shinyapps.io/EPIC_1-1/) were used to verify, analyze and explore the relationship between KDM4B and immune in ltrating cells and immune checkpoint inhibitors.

KDM4B related gene enrichment analysis and Correlation analysis
Completed the Go (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis through DAVID (Database for Annotation, Visualization and Integrated Discovery, https://david.ncifcrf.gov/home.jsp). The positively correlated co-expressed genes and negatively correlated co-expressed genes of KDM4B calculated and obtained from the UALCAN database were uploaded to the DAVID database, and then analyzed by GO and KEGG. The R software (v.3.6.1 version) was used in this analysis. R packages such as "tidyr", "cnetplot" and "ggplot2" are used to visualize the results of GO and KEGG. Among them, Spearman correlation analysis was used to detect the correlation between the KDM4B gene and the enrichment pathway. A p value < 0.05 was considered statistically signi cant.

Statistical analysis
The R software (v.3.6.1 version) was used to perform statistical data analysis. Survival and clinical characteristics data were got from the TCGA database and clinically collected samples. Then, the overall survival of KDM4B was determined by COX regression and Kaplan-Meier. In addition, Wilcox or Kruskal was used to test the correlation between the clinical characteristics and the KDM4B expression level. The Univariate Cox and Multivariate Cox analysis were used to analyze the factors which may impact the prognosis of patients with UCEC. P value < 0.05 was considered statistically signi cant.

1.The pan-cancer expression level of KDM4B and its mutation analysis
To probe into the possible role of KDM4B in tumors, the UALCAN database and the human protein atlas (HPA) database were fully utilized. The results showed that KDM4B was overexpressed in multiple malignancies (ACC, BLCA, BRCA, OV, STAD, TGCT, THCA and UCEC) at either the mRNA level or the protein level ( Fig. 1A and 1B). Further, KDM4B was found to be highly expressed in gynecological malignancies. Then, the cBioPortal database was used to explore the mutation type and mutation site of KDM4B. In UCEC there are 7% of genetic alterations. KDM4B has a somatic mutation rate of 5.85% in UCEC, mainly in some areas: JmjN, JmjC, PHD2 and TUDOR. The above results suggested that KDM4B was abnormally overexpressed in varieties of cancers and may have mutations and is a potential oncogene in patients with UCEC.

2.KDM4B expression level in UCEC and its relationship with the clinical characteristics
A total of 546 samples of UCEC patients with various clinical characteristics were obtained from the TCGA database. These clinical characteristics included Age at diagnosis, Histological type, FIGO stage and Histological grade. For one thing, we used GEO database (GSE36389, n = 20) and (GSE115810, n = 26) and TCGA database to verify the expression level and found that KDM4B is highly expressed in UCEC than normal endometrial tissue, which may play a role as an oncogene ( Fig. 2A and B). Relevant information of the clinically collected samples were statistically analyzed and recorded in Table S3. Distant metastasis and Lymph node metastasis (P < 0.05). The KDM4B expression level increased with he increase of Histological grade and FIGO stage of UCEC (P < 0.0001). As for Muscular layer depth of Invasion, the gene expression level of patients with < 1/2 invasion was lower than that of patients with ≥ 1/2 invasion (p < 0.001). There was a close correlation between the KDM4B expression level and Histological grade, FIGO stage, Muscular layer depth of Invasion and Distant metastasis and Lymph node metastasis. However, there was no signi cant association between KDM4B immunoreactivity and other clinical characteristics (Age at diagnosis and Histological type). In summary, KDM4B was highly expressed in UCECs at both the protein level and nucleic acid level, and the high expression of KDM4B was closely associated with those clinical characteristics that affect the prognosis of patients with UCEC.

3.Validation of KDM4B in Our Clinical Samples
The KDM4B expression was localized in the cytoplasm of tumor cells. Moreover, the differences in expression levels of KDM4B protein in UCEC and normal tissue were assessed by Western blotting (Fig. 3A). The results showed that in 10 cases of normal endometrial tissue, the band of KDM4B was signi cantly shallower than that of 30 cases of cancer tissue (Fig. 3B). The KDM4B expression was lower in normal endometrium than in UCEC tissue (Fig. 3C). Notably, histogram statistical analysis revealed a signi cant difference in the two groups (P < 0.05). Our clinical follow-up information of 126 UCEC cases showed that elevated expression of KDM4B was associated with poor OS or PFS in patients with UCEC (P < 0.001, Fig. 3D-E). The Univariate Cox and Multivariate Cox analysis showed that the high expression of KDM4B was an independent prognostic factor for both PFS and OS (Table S1, Table S2, P < 0.05, respectively). Therefore, our results indicate that KDM4B may serve as an independent adverse prognostic biomarker for UCEC patients. In conclusion, KDM4B can be used as an independent risk factor and a prognostic biomarker in UCEC. Immunohistochemistry and western blot showed that KDM4B was highly expressed and associated with poor prognosis in patients with UCEC.

4.Determination of KDM4B-related cellular signaling pathway and biological process by GO and KEGG
In order to in-depth explore the molecular mechanism that KDM4B may participate in regulation in the occurrence and development of UCEC, we tried to screen out a series of pathway and biological function enrichment analysis by targeting KDM4B binding proteins and KDM4B expression related genes. Figure 4A showed that KDM4B was signi cantly enriched in the following 4 Biological process: GO:0008152(metabolic process), GO:0009987(cellular process), GO:00065007 (biological regulation) and GO:0023052 (signaling). KEGG pathway enrichment analysis indicated that KDM4B was enriched in the following signaling pathways, including ErbB signaling pathway, FoxO signaling pathway, Central carbon metabolism in cancer, Metabolic pathways, Hedgehog signaling pathway and AMPK signaling pathway. Interestingly the Hedgehog signaling pathway is an immunity related pathway. In summary, by GO and KEGG we found that KDM4B is associated with immune-related biological processes and signaling pathways.

5.Genetic alteration analysis data (MSI, TMB, MMR)
Spearman correlation analysis of MSI and KDM4B gene expression can be seen in Fig. 6A. KDM4B is closely related to MSI in a variety of tumors, including UCEC, CESC, GBM, LGG, BRCA and LUAD. In UCEC, the expression level of KDM4B was positively correlated with MSI (P < 0.00001). Spearman correlation analysis of TMB and kdm4b gene expression is shown in Fig. 6B. KDM4B has a close correlation with TMB in multiple tumors, including UCEC, GBM, CESC, COAD, LGG and LUAD. Moreover, KDM4B expression levels were signi cantly correlated with TMB in UCEC (P < 0.00001). Clinically, mutations in MLH1, MSH2, MSH6, PMS2 are commonly found to cause dMMR, and the appearance of MSI in tumor tissue is an important hallmark of DNA mismatch repair de ciency(dMMR). Figure 6C shows that KDM4B is closely associated with MSH2(P < 0.05) and PMS2(P < 0.01). The expression level of KDM4B and the mutation status of MSH2 and PMS2 were signi cantly correlated. Overall, KDM4B is closely associated with MSI, TMB and MMR, and may serve as a promising genetic biomarker. 6.Immune in ltration and immune checkpoint inhibitor analysis data (TIMER, CIBERSORT, TISIDB, EPIC) Tumor in ltrating lymphocytes (TILs) have been reported to be an important component of the tumor microenvironment and in uence tumorigenesis and survival outcomes. To clarify the potential correlation between the expression level of KDM4B and immune cell in ltration in UCEC patients, we performed the following studies utilizing four different databases and algorithms: TIMER, CIBERSORT, TISIDB and EPIC. As shown in Fig. 7A, the timer database was searched to evaluate the correlation of KDM4B mRNA expression in Pan cancer with immune cell in ltration. Six immune in ltrating cells in the Timer database were signi cantly in ltrated in the pan cancer. As can be seen in UCEC, KDM4B expression was correlated to CD4 + T cells (P < 0.01), Neutrophil cells (P < 0.001), Myeloid dendritic cell (P < 0.05) and B cell (P < 0.001). As illustrated in the Fig. 7B, the expression of KDM4B was negatively correlated with immune in ltration of CD4 + T cells (r = − 0.182, P = 1.87e-03) and B cells (r = − 0.135, P = 2.21e-02) and positively correlated with CD8 + T cells (r = 0.2,P = 6.46e-04).No association was discovered between KDM4B expression and Macrophage (r = 0.008,P = 8.92e-01), Eutrophils (r = − 0.105,P = 7.26e-02) or Dendritic cell (r = − 0.008,P = 8.96e-01) in ltration.
In addition, to validate the results of the Timer database, the CIBERSORT tool was further used to explore the relationship between the expression of KDM4B and immune in ltrating cells at the single-cell level. Figure 8(A-B) showed that high KDM4B expression was signi cantly associated with Monocyte cell (P < 0.001), Neutrophil (P < 0.05), B cell (P < 0.001), CD4 + T cell (P < 0.001).
Moreover, a combination of high-throughput screening and genomic pro ling data from the TISIDB database was used to analyze the relevance of KDM4B for T cell killing or immunotherapy. Figure 9A showed that the expression of KDM4B has a signi cant correlation with the Tumor In ltrating Lymphocytes (TILs) of pan-cancer. As for UCEC, the expression of KDM4B is negatively correlated with Act CD4 cell (r = − 0.309, P = 1.84e-13) and B cell (r = − 0.13, P = 0.00238), while it is positively correlated with CD56 (r = 0.105), P = 0.014) and Th17 (r = 0.102, P = 0.0169). This research showed that the expression of KDM4B has a signi cant correlation with pan-carcinoma immune checkpoint (Fig. 9B). In UCEC, the expression of KDM4B was positively correlated with immune checkpoint, including ADORA2A (r = 0.092, P = 0.03), CD96 (r = 0.086, P = 0.04), CD244 (r = 0.039, P = 0.03) and CTLA4 (r = 0.12, P = 0.04).
Eventually, we use the EPIC algorithm to nd (Fig. 10A) that the expression of KDM4B is signi cantly correlated with the in ltration of immune cells in CD8 + T, CD4 + T, B cell, Endothelial cell and Macrophage (P < 0.001). Figure 10B showed that the expression of KDM4B in UCEC is positively correlated with CD4 + T cell (r = 0.337, P = 1.31e-03) and CD8 + T cell (r = 0.427, P = 3.27e-05). As can be seen in Fig. 10B, in UCEC, the expression of KDM4B was positively correlated with immune checkpoint, including SIGLEC15 (P = 0.01) and CTLA4 (P = 0.05).

7.Correlation analysis of KDM4B with immune-related copy number variation and immune in ltration-related survival rates
In UCEC patients, alteration in the copy number of KDM4B at different degrees results in altered in ltration of CD8 + T cells, Macrophages, Neutrophils, and Dendritic cells (Fig. 11A). This suggests that the expression of KDM4B is associated with immune in ltration in UCECs. In UCEC, the high expression of KDM4B is associated with poor OS. It is speculated that KDM4B affected the prognosis of patients through immune in ltration. Therefore, we further use Kaplan Meier curves to test the above hypothesis (Fig. 11B). The results showed that expression levels of KDM4B was associated with poor prognosis in patients with CD8 + T cell and B cell in ltration, but not with the other four kinds of cells. In conclusion, the above results indicated that the high expression of KDM4B gene could affect the prognosis of UCEC patients through the immune in ltration of some immune cells and the immunosuppression of other immune cells.

Discussion
Patients with late-diagnosed and refractory UCEC including high-grade, recurrent and metastatic have a poor prognosis and are treated with traditional surgery and radiotherapy with limited success. Fortunately, harnessing the immune system through checkpoint blockade has greatly expanded the treatment options for advanced UCEC. The continued exploration of the oncogenic role of immune checkpoints and immune cell in ltration in the tumor immune microenvironment has provided new ideas for the development of new UCEC treatment options, such as a combination of immunotherapy and gene targeted therapy. Although signi cant advances have been made in immunotherapy and gene targeted therapies for UCEC, improvements in patient prognosis are yet to be achieved. Therefore, this study attempted to explore the relationship between the expression of KDM4B and the prognosis and tumor immune microenvironment of UCEC, using the potential oncogene KDM4B as an entry point. This study is dedicated to nding prognostic-related predictive biomarkers that can be used to identify subgroups that are particularly sensitive to immunotherapy, which is of great signi cance for improving the survival outcome of UCEC (15,16).
Firstly, the expression level of KDM4B in UCEC and its value for prognostic assessment were explored. As seen in Fig. 2A and 2B, the use of the UALCAN database and the HPA database revealed signi cant overexpression of KDM4B at the protein and nucleic acid levels in various malignancies. In the light of the above pan-cancer results, we further explored the effect of KDM4B on the malignant biological process and prognosis of UCEC and explored its possible mechanisms. First of all, this study has been validated by TCGA data mining and GEO data. KDM4B was found to be signi cantly highly expressed in UCEC both at the protein level and at the mRNA level and mutated at multiple loci (Fig. 1C). Meanwhile, the use of IHC and WB further con rmed the respective expression levels of KDM4B in UCEC and normal control ( Fig. 3A-C).Secondly, our experimental results also found that KDM4B is related to a variety of clinical factors that affect the prognosis of UCEC, such as the histological grade of the tumor, the FIGO staging, Muscular layer depth of Invasion, Lymph node metastasis and Distant metastasis, as shown in Fig. 2 (E-I) .Numerous studies have con rmed that FIGO stage III and IV, lymph node metastasis and deep muscle in ltration are high risk factors for poor prognosis in UCEC patients. Therefore, we boldly hypothesized that KDM4B contributed to the poor outcome of UCEC patients. Then, survival prognosis was explored by collecting clinical samples from 126 UCEC patients. Kaplan-Meier survival curves showed that patients in the high KDM4B expression group had shorter overall and progression-free survival than those in the low KDM4B expression group (Fig. 3D-E).However, in order to exclude the in uence of accidental factors, we continue to rigorously conduct univariate and multivariate analysis to con rm that KDM4B can be used as an independent risk factor for poor prognosis in UCEC patients (Table S1and Table S2).A large number of previous studies have demonstrated that KDM4B contributes to the progression of various malignancies such as gastric, prostate and colorectal cancers, and is strongly associated with poor prognosis, which is highly consistent with our study (17)(18)(19). In summary, it is not di cult to nd that KDM4B can act as a new oncogene in UCEC leading to poor prognosis of patients, but the possible mechanism of its oncogenesis needs to be further explored.
To further understand the pathological mechanism of poor prognosis of UCEC due to KDM4B, GO annotation analysis and KEGG cell signaling pathway were used to perform enrichment analysis of KDM4B. In our study, many biological processes related to metabolism and immunity were found to be closely associated with high expression of KDM4B. (Fig. 5). The KEGG results suggested that KDM4B is signi cantly enriched in signaling pathways such as metabolic pathways and Hedgehog signaling pathways in UCEC. Interestingly, metabolic-related cellular signaling pathways play a pivotal role in the regulation of the tumor immune microenvironment (20). Of interest is that Hedgehog signaling pathway plays a crucial role in the progression of multiple tumors as well as in immune in ltration. Studies have shown that the Hedgehog signaling pathway is thought to be responsible for the formation of squamous cell carcinoma of the head and neck and leads to tumor formation (21). Second, Hedgehog signaling pathways and Innate Lymphocytes (ILCs) have important roles in immune responses to infection, cancer and autoimmunity (22). It has also been demonstrated that Hedgehog signaling pathway affect CD8 + T cell in ltration in primary squamous cell carcinoma of the head and neck (21). The above study gave us signi cant insight that KDM4B in UCEC may contribute to malignant progression and poor prognosis by altering the tumor immune microenvironment through immune-related biological processes and Hedgehog signaling pathways.

The tumor microenvironment (TME), consisting of tumor cells, mesenchymal cells and immune cells, is
constantly being recognized, and the alteration of the TME is a fertile ground for malignant transformation of tumors. Among them, the immunologic suppression and immunologic escape play a decisive role in the unrestricted survival and development of tumor cells. By exploring the TME and new molecular biomarkers, the combination of both for clinical bene t is signi cant for the diagnosis and treatment of UCEC (23,24). Therefore, we integrated four immune databases for the study of the TME. On the one hand, the molecular classi cation theory based on sequencing analysis has, to a certain extent, compensated for the limitations of traditional binary classi cation and opened up new horizons for prognosis evaluation and treatment guidance of patients. The correlation between KDM4B, a new potentially oncogene of UCEC discovered by the above research, and immune cell in ltration was explored nding that KDM4B expression was correlated to CD4 + T cells, Neutrophil cells, Myeloid dendritic cell and B cells (Fig. 7B). The relationship between the expression level of KDM4B and immune checkpoint was also explored showing that KDM4B was associated with SIGLEC15, ADORA2A, CD96, CD244 and CTLA4 (Fig. 8B and 9B). The present study demonstrated a signi cant positive correlation between KDM4B and CTLA4 and SIGLEC15 respectively, which may inhibit the immune response of UCEC patients through excessive immunosuppression leading to immunologic escape of malignant tumor cells resulting in poor prognosis of malignant tumor. Mesenchymal stem cell-derived extracellular vesicles overexpressing miR-15a limit immunologic escape in colorectal cancer (CRC) via the KDM4B/Hoxc4/PD-L1 axis, and this study con rms our ndings (12). In addition, immune checkpoint blockade using anti-PD1/PD-L1/CTLA4 antibodies improved the prognosis of patients with refractory solid tumors. As a new player in the eld of cancer immunotherapy, SIGLEC15 may be able to act as a novel immunosuppressant with potential impact on anti-PD-1/PD-L1 resistant patients (25)(26)(27)(28). On the other hand, since the Cancer Genome Atlas group (TCGA) performed extensive molecular genetic analysis, signi cant progress has been made in exploring the underlying molecular biology of UCEC. The molecular classi cation theory based on sequencing analysis has, to a certain extent, compensated for the limitations of traditional binary classi cation and opened up new horizons for prognosis evaluation and treatment guidance of patients. This study showed that the expression level of KDM4B in UCEC was signi cantly associated with representative signature genes evaluating immunotherapy response rates including high tumor mutation load, high microsatellite instability and mismatch repair deletion genes by bioinformatics analysis (Fig. 5A-C)(28). Since the crosstalk between the tumor and the immune system is complex and profound, the low response rate of cancer immunotherapy can be explained (29). Therefore, this study provides a solid foundation for further comprehensive study of the interaction between oncogenes (KDM4B) and immune cells, which will help to elucidate the pathogenesis of UCEC and improve the effectiveness of immunotherapy.
Immunotherapy is a promising approach for the treatment of gynecologic cancers. Current and ongoing researches were attempting to improve clinical prognosis through immunotherapeutic strategies. The newly identi ed potential oncogene KDM4B in this study is inextricably linked to immunity and may be a new hope for immunotherapy of UCEC. Recent advances in gene targeted therapy suggested that with appropriate biomarkers, gene targeted therapy has the potential to improve long-term survival of UCEC patients. However, the median progression-free survival of patients receiving single gene targeted therapy is less than 5 months (30). Thus, the KDM4B gene explored in this study is not only a potential oncogene that can be used for prognostic assessment, but also a possible therapeutic target for immunity (31). The cancer genetic sequencing data were analyzed in depth using advanced bioinformatics analysis, thus summarizing the genetic background of UCEC and opening new pathways to improve the survival outcome of patients with refractory advanced UCEC in the future. It also provides a theoretical background for the combination of gene targeted therapy and immunotherapy.
However, our study still has some limitations. Firstly, some clinical features were incomplete when studying the clinical correlation between KDM4B and UCEC in TCGA and GEO databases, such as the lack of data on TNM staging, Lymph node metastasis and Muscular layer depth of Invasion. But the lack of some speci c clinical features is an inevitable drawback of public databases. So we compensated for the above limitation by collecting clinical samples. Secondly, the detailed mechanism about KDM4B and the alternation of TME needs more time and effort to continue our exploration at a later stage. However, it is gratifying to note that we conducted studies related to KDM4B at multiple levels (genomics and proteomics) and in multiple databases to ensure the comprehensiveness and reliability of our study. It is a solid foundation for further studies in the future to create more clinical bene ts for UCEC patients.

Conclusions
Our study suggested that aberrantly expressed KDM4B may be a potential prognostic marker for UCEC and, more importantly, it may be associated with tumor immune microenvironment. Of clinical signi cance, KDM4B may be used to assess the clinical prognosis of UCEC patients and may also serve as a target for immunotherapy or as a potential marker for checkpoint inhibitor-based immunotherapy.