Research
Comprehensive pan-cancer analysis of FUTs family as prognostic and immunity markers based on multi-omics data
Zexi Jia1 . Pan Liao2 . Bo Yan1 . Ping Lei1,2
Received: 3 July 2024 / Accepted: 11 October 2024
Published online: 16 October 2024
@ The Author(s) 2024 OPEN
Abstract
Background The dysregulation of fucosyltransferases (FUTs) contributes to alterations in fucosylated epitope expression, which serve as distinctive features of cancer cells. Nonetheless, a comprehensive elucidation of the prognostic biological marker and therapeutic target of the FUTs family in pan-cancer remains elusive.
Methods Over 10,000 individuals’ profiling information was examined, including information on 750 small molecule drugs, 33 types of cancer, and 24 types of immune cells. We focused on POFUT2’s function and applied GSVA (Gene Set Variation Analysis) to calculate the FUT score. Survival and cancer pathways were found to be correlated with this score. After deriving a signature via univariate Cox and LASSO regression, we generated and analyzed the ROC curve and developed a nomogram.
Results Our comprehensive analysis revealed epigenetic, genomic, and immunogenomic changes in FUTs, particularly POFUT2, resulting in aberrant expression. Elevated frequencies of CNV (Copy number variation), SNV (Single Nucleotide Variant), and hypermethylation were observed in FUTs. Additionally, the survival of patients with various types of cancers may be predicted by FUT expression. Immune response and prognosis in numerous types of cancer were found to be strongly linked to aberrant POFUT2 expression. Pathway analysis unveiled the role of FUTs in apoptosis, epithelial-to- mesenchymal transition (EMT), cell cycle, DNA damage response, RAS/MAPK, TSC/mTOR, PI3K/AKT, AR, ER, and RTK. A prognostic index for patients diagnosed with adrenocortical carcinoma (ACC) was established by applying a risk model incorporating nine FUTs and based on the findings of the GSVA.
Conclusions FUTs, particularly POFUT2, emerge as candidate targets for improving the outcomes of immune therapy. The significance of aberrant MUC12 expression, cancer immune therapy, and patient survival in the context of diverse malignancies is enhanced by the strong correlation observed among these factors. Our five-gene risk signature provides patients with ACC with an independent prognostic indicator, emphasizing the critical function of these genes in inhibit- ing the immune system’s response in ACC.
Keywords Pan-cancer . FUTs . POFUT2 . Immunogenomic . Genomics . Epigenetic . Risk model
Zexi Jia, Pan Liao and Bo Yan contributed equally to this work.
Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s12672-024- 01447-6.
☒ Ping Lei, leiping1974@163.com | 1Department of Geriatrics, Tianjin Medical University General Hospital, Tianjin, China. 2School of Medicine, Nankai University, Tianjin, China.
Check for updates
(2024) 15:567
| https://doi.org/10.1007/s12672-024-01447-6
Discover
1 Introduction
There has been a considerable upward trajectory in the incidence, morbidity, and mortality related to cancer in recent years. Cancer ranks as the second greatest contributor to mortality globally, after cardiovascular disease. After 2030, malignant tumors are projected to be the leading cause of global mortality, according to the WHO [1]. Cancer genesis and advancement are strongly influenced by the tumor microenvironment (TME). Immunotherapy is exceptionally well-suited for cancer patients due to its remarkable effectiveness. Nevertheless, immunotherapy may not be effective for every patient, and DNA microsatellite instability (MSI) and tumor mutation burden (TMB) are two potential prognostic markers for the success of immunotherapy that have been identified through research. TMB [2] has promising prognostic utility for immunotherapy in several distinct types of cancers. Additionally, Molecular markers such as MSI [3] have shown promise in the prognostic assessment and adjuvant therapy for several solid tumors including colorectal cancer. Unfortunately, our global perspectives on the target gene’s potential mechanism and its various aspects were restricted by cancer research focused on a particular tumor. Researchers in the field of cancer have made extensive use of pan-cancer analysis due to the complexity of tumor development and there has been a significant advancement in the comprehension of several tumor characteristics, such as cancer susceptibility variations, oncogenic pathway co-occurrence and mutual exclusion, and biological regulation network disorder [4-6].
Fucosyltransferases (FUTs) are among the glycosyltransferases (GT) as distinguished by their structural fold archi- tecture, the specificity of the substrate, the mechanism via which they interact with both donor and acceptor sub- strates, and the catalytic activity of the reactions. Additionally, fucosylated epitopes are characteristics of cancer cells, and their expression is changed when FUTs are altered, Among the glycosyltransferases (GT), fucosyltransferases (FUTs) stand out [7, 8]. Their molecular biology and cancer drug development potential have received little attention up until now. While some studies have examined the FUTs family as a prognostic indicator and treatment target, its precise applicability in pan-cancer is still unexplored.
Considering the intricate relationships between FUTs, inflammatory processes, and cancer, this study sought to thoroughly examine the genetic, immunological, and clinical characteristics associated with FUTs in 33 distinct cancer types. With a focus on POFUT2, our findings illustrated how epigenetic alterations, genomic variations, and immunogenomic factors contribute to the aberrant expression of these FUTs. Furthermore, we found a robust cor- relation between infiltration of immune cells, overall survival, and abnormal POFUT2 expression in patients with distinct types of cancer. Moreover, we employed the five FUTs derived from the gene-set expression analysis (GSVA) findings to create a risk model, which was established as an independent indicator of prognosis for adrenocortical carcinoma (ACC) patients. A link between immune cell infiltration and the FUT-related signature risk score was sys- tematically evaluated. Our research shed light on how tumor microenvironment (TME) factors regulate FUTs, and it highlights the possibility that these genes might be used as functional prognostic indicators, providing information on the clinical outcomes of ACC patients and possible areas to focus on to make immunotherapy more effective.
2 Methods and materials
2.1 Tumor types and databases
Assessment of disease diagnosis, disease outcomes, and therapy efficacy in individual patterns relies heavily on alterations in the immune microenvironment and genome, which in turn regulate the onset and progression of malig- nancies. The expression of certain genes may be obscured by substantial background noise in this era of abundant biological data. To get a complete understanding of cancer’s underlying mechanisms, it may be necessary to combine gene expression or gene set scores from many databases for a large number of patients at different carcinogenesis stages. Therefore, a more thorough identification of FUTs’ participation in various malignancies may be feasible via the integration of multi-omics data. Here we investigated gene modifications, including alterations to gene expres- sion, copy number variations (CNVs), alterations in gene methylation status, and single-nucleotide variations (SNVs).
We applied multi-omics data compiled from several reliable sources to conduct this analysis: Patients’ data on clinical characteristics (n = 11,160), disease staging (n = 9,478), gene expression (n = 10,995), immune cell infiltra- tion (ICI, n = 10,995), methylation profiles (450 k level 3) and CNV (n = 11,495) were sourced from the UCSC Xena
Discover
platform (http://xena.ucsc.edu/) and The Cancer Genome Atlas (TCGA; https://portal.gdc.cancer.gov/). Additionally, the Synapse database (https://www.synapse.org/#! Synapse: syn7824274), was screened to compile data on SNVs (n = 10,234), whereas reversed-phase protein arrays (RPPAs, n = 7,876) were derived from The Cancer Proteome Atlas (TCPA; https://tcpaportal.org/tcpa/index.html) [9]. There were a total of 33 distinct cancer types included in the study (Table S1), and 24 distinct types of immune cells evaluated for gene signatures (Table S2) [9].
2.2 Analysis of copy number variation
A summary of CNVs in selected malignancies and the genetic changes that correspond to them were revealed using the CNV Summary module. The CNV data used to compile these summaries came from the TCGA’s extensive cohort of 11,495 participants. The GISTIC2.0 algorithm was utilized to identify regions within patient genes that were significantly amplified or deleted in our analysis. The GISTIC score is a crucial metric for assessing the association between a CNV and a gene: A deep deletion, represented by -2, signifies a substantial loss or homozygous deletions, -1 denotes a superfi- cial deletion, which is equivalent to a mild heterozygous deletion or a mild loss, and 0 signifies a diploid condition. In addition, a GISTIC score of 1 or higher signifies a negligible increase, often associated with the acquisition of a small number of additional duplicates through extensive gains or heterozygous amplification. A high level of amplification is indicated by a score of 2 or greater, which signifies the existence of additional copies, possibly resulting in homozygous amplification. The meticulous summary of sample distribution across 4 classifications (GISTIC scores: - 2, -1, 1, 2) of CNV in distinct cancer types was achieved.
2.3 Analysis of single nucleotide variant
The SNV summary module functions in GSCA (https://guolab.wchscu.cn/GSCA/#/) as a useful tool for comprehending SNVs that are present in particular types of cancer. Our TCGA dataset comprised data on 10,234 patients who were diag- nosed with 33 different types of cancer. Our evaluation was specifically directed toward deleterious mutations, which are of specific relevance to cancer research within this extensive dataset. As is common knowledge, deleterious mutations might lead to the onset and progression of cancer by interfering with normal gene function.
On the other hand, our analysis did not consider non-deleterious mutations, such as intergenic region (IGR) muta- tions, intronic mutations, silent mutations, as well as mutations occurring within 3’ and 5’ untranslated regions (UTRs), and flanking regions at 3’and 5’. Non-deleterious mutations are generally regarded as having a lesser impact on cancer- related mechanisms and gene function.
2.4 Methylation analysis
The differential methylation module in GSCA (https://guolab.wchscu.cn/GSCA/#/) serves as a valuable instrument in elucidating the methylation circumstances exhibited by cancer patients in comparison to healthy controls. For the com- pilation of this dataset, we acquired Illumina HumanMethylation 450 k level 3 data from more than ten paired tumors and adjoining non-cancerous samples from TCGA. These samples represented a wide range of cancer types, including THCA, BLCA, ESCA, COAD, KIRP, LIHC, LUAD, BRCA, STAD, HNSC, KIRC, PRAD, LUSC, and KICH. Many different kinds of methylation sites are present in every gene, and multiple tags are used to store data regarding the level of methylation at each site.
2.5 Differential mRNA expression analysis
An essential component of our investigation into gene expression patterns associated with cancer was the execution of differential mRNA expression analysis. The clinical characteristics (n= 11,160) and RNA-Seq (n = 10,995) of patients were extracted from TCGA datasets. Normalized and batch-corrected RSEM gene expression data were utilized in the differ- ential expression analysis. The dataset comprised information on 13 paired tumor and healthy samples representing a range of cancer types, including KIRC, LIHC, LUAD, BRCA, HNSC, KICH, KIRP, ESCA, THCA, STAD, BLCA, COAD, PRAD, and LUSC. The following formula was used to determine the fold change (FC) in expression: FC=mean (tumor)/mean (normal).
2.6 Analysis of pathway activities
We analyzed the data gene expression and pathway scores to determine the presence of variations in the enrichment of pathways among sample types. In this study, we calculated median pathway scores to determine whether samples showed pathway inhibition or activation. We determined the activity scores of 10 pathways linked to cancer in 7,876 patients with 32 distinct types of tumors using TCGA-based RPPA data [9]. Additionally, we investigated signaling pathways linked to epithelial-to-mesenchymal transition (EMT), DNA damage response, apoptosis, cell cycle, AR, RTK, RAS/MAPK, TSC/mTOR, ER, and PI3K/AKT (Table S3).
Subsequently, we employed median-centered RPPA-RBN data to compare each sample’s protein expression, after which standard deviation calculations were performed to normalize the data. Using the procedure outlined in a previ- ous study [10], the pathway score was calculated by adding the protein expression of positive regulatory components in a particular pathway and subtracting the expression of negative regulatory components.
Two patient groups were distinguished based on the median gene expression: the high-gene expression group (HRG) and the low-gene expression group (LRG). We implemented the Student’s t-test to contrast the two groups’ pathway activity scores (PAS), and then we corrected the P-value using FDR. FDR ≤ 0.05 was employed as the sig- nificance criterion. Gene A may be able to activate a pathway if PAS (high gene A expression) > PAS (low gene A expression). In contrast, the inhibition of a pathway by gene A was signified by PAS (high-gene A expression) < PAS (low-gene A expression), as demonstrated previously [10, 11].
2.7 Survival analysis
Our clinical data set included 33 different cancer types, and these data were used to analyze gene expression and survival. If a patient’s data were missing or it was determined that they had co-morbid conditions, they were not included in the following analyses, involving progression-free survival (PFS), disease-specific survival (DSS), disease- free survival (DFS), and overall survival (OS).
We then integrated gene expression with survival data using sample barcodes. The median gene expression value served as the cutoff value for patients to be included in either the HRG or the LRG. Patients’ survival durations and sta- tuses were determined utilizing the “survival” R package(https://guolab.wchscu.cn/GSCA/#/). Lastly, we implemented Kaplan-Meier (KM) analysis and analyzed the selected genes’ impact on cancer prognosis via Cox proportional haz- ard analysis and a log-rank test. Additional analysis was conducted on genes that had log-rank test p values < 0.05.
2.8 POFUT2-related differential expression among cancerous, normal, and risk groups
Initially, an Open Target Platform bubble graph (https://platform.opentargets.org/) was utilized to illustrate the dis- eases or phenotypes linked to POFUT2. TIMER2.0’s Gene_DE module [12] (http://timer.cistrome.org/) was utilized to quantify the variations in POFUT2 mRNA expression between adjoining normal tissues and cancer tissues. GTEx and TCGA data were adopted and analyzed by the Box Plot module in the Expression DIY function of GEPIA2.0 (http:// gepia2.cancer-pku.cn/); the absolute log2FC cutoff and p-value cutoff were set as 0.585 and 0.01, respectively. This study adopted GEPIA2 “survival analysis” to investigate the association between POFUT2 expression and the prognosis of various TCGA-derived malignancies [13].
2.9 Analysis of immune cell infiltration
To get further insight into the function and pathways of the POFUT2 in pan-cancer, we searched different databases. The Cancer Single-Cell State Atlas (CancerSEA) database (http://biocc.hrbmu.edu.cn/CancerSEA/) provides diverse functional states of specific genes at the single-cell level in different cancer types, allowing researchers to bypass the limitation of tumor heterogeneity. Correlations between POFUT2 and functional states in various cancers were performed based on the CancerSEA database. The immune checkpoint markers to analyze their correlations with POFUT2. To investigate cytokine treatment’s effects on POFUT2 expression, the Tumor Immune Syngeneic MOuse (TISMO, http://tismo.cistrome.org/) web tool was used to compare gene expression levels across cell lines between
Discover
pre- and post-cytokine treated samples. Following that, the CIBERSORT algorithm was utilized to calculate the immu- nocyte infiltration correlations of POFUT2 [14].
2.10 Analyses of POFUT2-targeting compounds, and molecular docking
The anti-POFUT2 chemical compounds were screened employing the “query” function of cMap [15] (https://clue.io/). A heat- map was created to illustrate the top 30 drugs and how they relate to POFUT2-related differentially expressed gene (DEGs) signature. Additionally, the heatmap displays the compounds’ mechanisms of action (MoA). Using ChemBioDraw Ultra 17.0, we designed 3D models of all potential medicinal substances. Afterward, we optimized their energy utilizing the MMFF94 force field. The PDB (http://www.rcsb.org/pdb/home/home.do) was searched for the 3D model of POFUT2. To prepare for docking analysis, AutodockTools v1.5.6 was employed to transform all molecular and protein data into PDBQT format. The molecular docking analysis was conducted utilizing Autodock Vina v1.1.2. The ‘exhaustiveness’ docking parameter was speci- fied as ‘20’ whereas all other parameters were set as their default values. The highest-scoring conformation was chosen for additional analysis utilizing Free Maestro 11.9. The 2D representations were drawn utilizing MOE software 2019, and the model visualization was done with Pymol software 2.3 [16].
2.11 Pathway exploration for gene set variation analysis (GSVA)
Our analysis sought to enhance our comprehension of the pathways and functions related to FUTs in different types of tumors by investigating the correlations of FUTs with functional status using sourced data. By employing this method, we were able to discern correlations between FUTs and particular functional states in diverse cancers, thereby enhancing our comprehension of their mechanisms and pathways (Table S4).
2.12 Development and verification of the FUT-related prognostic model
LASSO Cox regression analysis was employed to develop a risk model from the complete dataset (n = 169) by utilizing prog- nosis-associated FUTs. As determined by the median risk score, individuals were classified as either high- or low-risk. Survival analyses were executed utilizing log-rank tests and KM survival curves to contrast the rates of survival between the low- and high-risk patients. The prediction power of the FUT-related risk model was verified via time-dependent receiver operating characteristic (ROC) curves. Additionally, the areas under the curves (AUC) were established to evaluate the accuracy of the model in anticipating patient outcomes.
We performed univariate and multivariate Cox regression analyses to verify that the FUTs association signature could accurately predict outcomes. Furthermore, we created a nomogram and calibration plots to assess the model’s capability to provide a comprehensive prognostic prediction for ACC. The integration of clinicopathological characteristics into this nomogram enhances its prognostic utility. We then examined the associations between the OS and the risk score, along with a range of clinical factors exhibited by patients with ACC. This was accomplished by conducting survival analyses on subgroups of patients, which provided further insight into the link between the risk score and patient features and outcomes.
2.13 Statistical analysis
For data analysis, the R software v4.2.2 was employed. We implemented Spearman’s correlation analysis to evaluate correla- tions. We implemented a Cox proportional hazard model to evaluate the HR values-based patient survival risk. We examined and contrasted the GSVA scores of the patients via the Wilcoxon test for two-stage groups and ANOVA for groups with more than two stages. The Mann-Kendall trend test was employed to execute the trend analysis. The variation in the PAS scores between the two patient subgroups was established utilizing the Student’s t-test. Unless otherwise specified, the compara- tive analysis involving two groups was accomplished utilizing the rank-sum test. p < 0.05 or FDR ≤ 0.05 was considered to be suggestive of a significant difference.
Discover
A CNV percentage in each cancer
ACC
DLBC
LAML
COAD
GBM
UVM
KICH
THCA
THYM
TGCT
SARC
PCPG
READ
KIRP
UCEC
STAD
KIRC
LGG
CESC
HNSC
BRCA
PRAD
BLCA
PAAD
MESO
UCS
ESCA
LIHC
LUSC
SKCM
CHOL
LUAD
3
FUT9
FUT11
FUT8
FUT4
FUT10
FUT7
Hete. Amp.
POFUT2
Homo. Amp.
Hete. Del.
FUT3
Homo. Del.
None
FUT6
FUT5
FUT2
FUT1
POFUT1
B
SNV percentage heatmap
UCEC (n=531)
SKCM (n=468)
COAD (n=407)
READ (n=149)
STAD (n=439)
LUSC (n=485)
LUAD (n=567)
BLCA (n=411)
DLBC (n=37)
CESC (n=291)
GBM (n=403)
HNSC (n=509)
ACC (n=92)
ESCA (n=185)
LIHC (n=365)
SARC (n=239)
KIRP (n=282)
OV (n=412)
UCS (n=57)
LGG (n=526)
KICH (n=66)
BRCA (n=1026) UVM (n=80)
PAAD (n=178)
PRAD (n=498)
MESO (n=82)
THCA (n=500) LAML (n=85)
KIRC (n=370)
TGCT (n=151)
CHOL (n=36)
PCPG (n=184)
THYM (n=123)
C
Altered in 623 (89.13%) of 699 samples.
D
18 -
Methylation difference in each cancer
FDR
.
0
164
FUT3 ☐
☐
☐
☐
☐
☐
☐
☐
⇐ 0.05
0
No af samples
FUT9
23%
FUT6
☐
☐ ☐
☐
☐
FUT8
FUT5 ☐
☐
☐
☐
-Log10(FDR)
16%
FUT8 ☐
☐
☐
☐
☐
10
FUT3
14%
FUT7
☐
☐
☐
☐
☐
20
FUT5
☐
11%
Symbol
FUT2
☐
☐
☐
☐
☐
☐
☐
☐
30
FUT6
11%
POFUT1
☐
☐
☐
☐
40
POFUT2
10%
FUT4 ☐
☐ ☐
☐
☐
50
☐
☐
☐
FUT10
10%
FUT10
☐
☐
FUT1
9%
FUT11 ☐
☐
☐
Methy. diff(T-N)
FUT2
9%
POFUT2
☐
☐
☐
☐
☐
-1
POFUT1
8%
FUT1 ☐
☐
☐
☐
Cancer_type
FUT9 ☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
0
Missense_Mutation
In_Frame_Del
Cancer_type
PRAD
BRCA
COAD
ESCA
HNSC
LUAD
PAAD
LIHC
UCEC
LUSC
KIRC
Splice_Site
THCA
BLCA
KIRP
Nonsense Mutation
Frame_Shift_Del
. Multi_Hit
ACC
BLCA
COAD
HNSC
KICH
LAML
LUSC
READ SARC
TGCT
THCA UCEC
UVM
Frame_Shift_Ins
DLBC
ESCA
KIRC
LGG
LIHC
OV
1
BRCA
= PAAD
SKCM
CESC
GBM
KIRP
LUAD
PRAD
STAD
UCS
Cancer type
| FUT9 | 22 | 51 | 12 | 5 7 | 22 | 15 | 3 | 0 | 3 | 2 | 10 | 0 | 2 | 4 | 2 | 0 | 1 | 0 | 0 | 0 1 1 | 1 | 0 | 0 | 0 | 0 | 0 | ||||
| FUT8 | 24 | 16 | 8 | 3 10 | 7 | 7 | 7 | 1 | 5 | 2 | 2 | 1 | 1 | 1 | 1 | 5 | 4 | 0 | 2 | 1 | 1 | 0 | ||||||||
| FUT3 | 26 | 21 | 6 | 3 4 | 4 | 6 | 4 | 2 | 4 | 1 | 2 | 2 | 0 | 1 | 3 | 1 | 0 | 3 | 1 | 2 | ||||||||||
| POFUT2 | 23 | 10 | 8 | 1 9 | 2 | 6 | 0 | 1 | 0 | 0 | 2 | 1 | 1 | 2 | 1 | 2 | 0 | 1 | 1 1 | 10 | ||||||||||
| FUT6 | 24 | 10 | 3 | 1 6 | 6 | 4 | 2 | 1 | 5 | 2 | 2 | 1 | 0 | 1 | 2 | 0 | 1 | 2 | 1 | 01 (%) | ||||||||||
| FUT1 - | 12 | 4 | 11 | 2 5 | 6 | 3 | 3 | 2 | 2 | 1 | 3 | 4 | 2 | 1 | 1 | 0 | 1 | 1 | 0 | 0 freq. | ||||||||||
| FUT10- | 22 | 7 | 1 | 3 5 | 5 | 7 | 3 | 1 | 1 | 3 | 0 | 1 | 1 | 1 | 1 | 3 | 2 | 1 | 1 0 | 0 | ||||||||||
| FUT5 | 14 | 16 | 5 | 1 4 | 8 | 2 | 3 | 2 | 5 | 4 | 1 | 1 | 1 | 0 | 0 | 3 | 5 | 0 | Mutation | |||||||||||
| FUT4. | 17 | 9 | 3 | 1 3 | 3 | 0 | 4 | 1 | 2 | 2 | 0 | 0 | 0 | 3 | 1 | 0 | 1 | 1 | 1 0 | 0 | ||||||||||
| FUT2. | 15 | 11 | 7 | 1 1 | 3 | 1 | 6 | 2 | 3 | 2 | 1 | 1 | 1 | 1 | 1 1 | 1 | 1 | 0 | ||||||||||||
| POFUT1- | 13 | 10 | 6 | 4 | 4 | 2 | 2 | 1 | 1 | 0 | 1 | 2 | 1 | 1 | 1 | 1 | 4 | 0 | 1 | 1 | ||||||||||
| FUT11 - | 9 | 5 | 3 | 2 3 | 5 | 2 | 0 | 0 | 0 | 0 | 1 | 3 | 0 | 1 3 | 0 | 1 | ||||||||||||||
| FUT7 | 6 | 6 | 4 | 4 | 0 | 4 | 1 | 1 | 1 | 1 | 2 | 1 | 0 | 1 |
Fig. 1 Copy number variation (CNV) distribution, single nucleotide variation (SNV) frequency and methylation distribution of FUTs in 33 tumors. A CNV pie chart showing the combined heterozygous/homozygous CNVs of each gene in each cancer. A pie chart representing the proportions of different types of CNVs of one gene in one cancer, where different colors represent different types of CNVs. Hete Amp=het- erozygous amplification; Hete Del=heterozygous deletion; Homo Amp=homozygous amplification; Homo Del =homozygous deletion; None =no CNV. B SNV oncoplot. An oncoplot showing the mutation distribution of FUTs and a classification of SNV types. C Mutation fre- quency of FUTs. The numbers represent the number of samples that have the corresponding mutated gene for a given cancer. A ‘0’ indicates that there was no mutation in the gene coding region, and no number indicates that there was no mutation in any region of the gene. D Differential methylation in FUTs between tumor and normal samples in each cancer type. Blue indicates decreased methylation in tumors, and red indicates increased methylation in tumors; the darker the color is, the larger the difference in methylation level
Discover
3 Results
3.1 FUTs and gene mutation analysis
We utilized TCGA patient data to study CNVs in FUTs as part of our analysis. As demonstrated by a CNV distribution pie chart, heterozygous amplifications and deletions constituted the most prevalent forms of CNVs identified in patients (Fig. 1A).
In addition, an examination of single nucleotide polymorphisms (SNPs) in FUTs was undertaken to assess the frequency and variants of genes within each tumor subtype. Patients with UCEC, READ, LUSC, STAD, LUAD, COAD, SKCM, and BLCA exhibited a frequency range of 0% to 51% for SNVs in FUTs, as illustrated in Fig. 1B. An additional finding indicated that SNVs were present in regulators of 89.13% (623/699) of patients (Fig. 1C). Of these SNVs, SNP types with the highest prevalence among patients were missense mutations. Particularly, a high proportion of these mutations were accounted for by the top 10 mutated genes, namely FUT9, FUT8, FUT3, FUT5, FUT6, POFUT2, FUT10, FUT1, FUT2, and POFUT1, with frequencies varying from 8 to 23%.
3.2 Analysis of FUT-related differential methylation across various types of cancer
We investigated the epigenetic regulation of FUTs by examining the methylation state of these genes. Significant patient heterogeneity was observed in the FUTs methylation status (Fig. 1D). Hypermethylation of FUTs appears to be more prevalent in PRAD and BRCA, according to our findings. Conversely, an elevation in hypomethylation in FUTs was observed among patients with KIRP, BLCA, THCA, KIRC, LUSC, UCEC, LIHC, PAAD, and LUAD. In most cancers, genes such as FUT9 and FUT1 exhibited excessive methylation (FDR ≤ 0.05, Fig. 1D). Additionally, FUT3, FUT6, FUT5, FUT8, FUT7,and FUT2 exhibited reduced methylation in most cancers (FDR ≤ 0.05, Fig. 1D).
3.3 Differential expression of FUTs across cancers and their impact on pathway activity and prognosis
FUT expression variations among cancer patients were investigated. Differences in FUT expression were found to be significant among patients diagnosed with the following solid tumor types: THCA, LIHC, HNSC, LUAD, KIRC, KICH, COAD, STAD, PRAD, LUSC, KIRP, and BRCA (FDR ≤ 0.05, Fig. 2A). An analysis BLCA and ESCA individuals did not reveal a statistically significant variation in FUT expression (FDR ≤ 0.05, Fig. 2A).
The critical involvement of FUTs in cancer-associated pathways was established through pathway activity analysis. The signaling pathways, hormone-related pathways including ER and AR, EMT, the cell cycle, programmed cell death, PI3K/AKT, RAS/MAPK, RTK, response to DNA damage, and TSC/mTOR were among the pathways involved (Fig. 2B).
Furthermore, a strong link was observed between the FUTs’ expression and the survival status of the patients (PFS, OS, DSS, and DFI) (Cox p < 0.05, Fig. 2C). These data indicated that aberrant FUT expression might play a role in the occurrence of tumors. Additionally, POFUT2 overexpression was negatively correlated with OS in 10 cancer types, DFS in 4 cancer types, DSS in 11 cancer types, and PFS in 9 cancer types (Cox P < 0.05, Fig. 2C). Based on this finding, POFUT2 has the potential to function as a carcinogenic gene in various types of cancer.
3.4 POFUT2 exhibits differential expression in pan-cancer cells and potentially predicts patient survival
OpenTarget was utilized to investigate the disease associated with POFUT2; the bubble graph illustrates that POFUT2 was linked to cancer and benign tumors (Fig. 3A). The disparities in POFUT2 mRNA levels between pan-cancerous and normal tissues were subsequently examined with TIMER2.0; POFUT2 mRNA expression was substantially increased in 10 distinct cancer types (GBM, LUAD, KIRC, LIHC, HNSC, LUSC, COAD, READ, CHOL, and STAD) (Fig. 3B). The disparities in POFUT2 mRNA levels between pan-cancerous and normal tissues were subsequently examined with GEPIA2.0; POFUT2 mRNA expression was substantially decreased in 9 distinct cancer types (CESC, LAML, LUSC, OV, READ, TGCT, THCA, and UCEC) and increased in GBM (Fig. 3C). For the prognostic significance of POFUT2, TCGA data were utilized to generate KM curves. We discovered that the POFUT2 upregulation was linked to lower OS percentages in ACC, BRCA, CESC, COAD, LGG, LIHC, and SARC; and lower DFS percentages of ACC, LIHC, and UVM (Fig. 3D). According to
A
DEGs in selected cancer types
FUT9
☐
☐
☐
☐
☒
☐
☐
☐
☒
☒
☒
FUT8
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☒
FDR
POFUT1 ☒
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
0.05
FUT1
☒
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
0.01
☐
☐
0.001
FUT2
☐
☐
☒
☒
☐
☐
☐
☐
☐
☐
☒
☒
☐
☒
☐
⇐ 0.0001
Gene symbol
FUT5 ☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
log2(FC)
8
FUT7
☐
☐
☐
☒
☐
☐
☐
☐
☐
☐
☒
☐
☐
☐
POFUT2
Q1
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
FUT11 ☒
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
-10
FUT4 ☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
FDR
FUT10 ☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐
☐ ⇐ 0.05
☐ >0.05
FUT3
☒
☒
☒
☐
☐
☐
☐
☐
☐
☐
☒
☒
FUT6
☒
☐
☒
☐
☐
☐
☐
☐
☒
☒
KICH-
PRAD
HNSC
KIRC
KIRP
STAD
COAD
BLCA
ESCA
LIHC
BRCA
LUSC
THCA
LUAD
Cancer
B
POFUT2
9
16
6
6
12
6
16
3
6
28
3
19 12
0
9
6
9
16
9
3
POFUT1
9
19
16
9
12
3
9
6
16
12
9
9
6
3
6
6
6
3
FUT9
6
19
3
3
6
12
16
9
9
9
3
9
6
3
12
0
16
3
6
0
FUT8
12
6
6
9
6
9
9
0
12
6
6
0
0
9
3
9
3
6
Symbol
FUT7
34
0 6
0
12
0 3
9
28
0
0
12
22
9 9
6 6
12
0
12
FUT6
6
3
19
16
6
19
16
16
3
6
16
3
12
3
6
9
FUT5
3
3
3
6
9
3
3
6
9
6
6
6
6
3
0 12
FUT4
12
0 9
12
3
6
9
22
6
9
25
6
9 9
9
9
12
3
12
9
9
9
FUT3
6
3
25
6
12
3
22
9
9
3
6
12
3
3
9
6
3
6
FUT2
16
3
3
12
3
12
6
19
3
12
3
19
6
6
16
0
12 3
12
6
FUT11
3
9
0
22
3 3
19
28
0
0
19
3
16
9
6
25
0 19
0
9
3
FUT10
0
9
3
9
3
3
6
9
0 19
0 6
FUT1
12
6
0 ☐
9
6
3
6
9
0 ☐
6
3
6
6
12 ☐
0
Apoptosis_A
Apoptosis
CellCycle_A
CellCycle,
DNADamage_
DNADamage,
EMT_A
EMT
Hormone AR
Hormone AR
Hormone ER_A
Hormone ER
PI3KAKT_A
PI3KAKT
RASMAPK_A
RASMAPK
RTK_A
RTK
TSCmTOR A TSCmTOR
Pathway (A: Activate; I: Inhibit)
Survival difference between high and low gene expression
Cancer type
| Percent 28 | 0 ☐ | 34 |
| Inhibit | Activate | |
| C | DFI | DSS | OS | PFS | |
|---|---|---|---|---|---|
| POFUT2 | 000 ☐ 00000 ☐ 000000 ☐ | ☐ .0000 ☐ 00000000 ☒ ☐ 00 | 0000 ☒ ☐ ☒ o ☐ ☐ 000 | ☐ ☐ 00000000 PC ☐ ☐ | |
| POFUT1-000000000000 ☐ ☒ 000 0 0 | ☐ ☐ | ☐ ☐ | ☐ ☒ ☐ ☒ 0 | ||
| FUT& | 000000000000 ☐ 1000 | ☐ ☐ ☐ ☐ | ☐ ☒ ☐ ☐ ☐ ☐ | ☐ 0 ☒ 000 ☐ 0.0000 ☐ ☐ ☐ 0 | -Log10(FDR) ☐ 2.5 |
| FUT11 | ☐ 0-0-0 ☒ 000 000000. 0-0 | ☐ 000 ☐ | ☐ | 000 ☐ 00000.000 ☐ 0000 ☐ | ☐ 5.0 7.5 |
| FUT4 | 00000 0000000 0-0 | ☐ ☐ 0000 ☒ 000 0.000000 ☐ 0 ☐ | ☐ ☐ ☐ 59 ☐ | ☐ 10000000 ☐ 0000000 ☐ 0 ☐ 1000000 | |
| FUT2 | ☐ 000000000000 ☐ 00 | ☒ 0000000000 ☐ 00 ☐ ☐ | ☐ ☐ ☐ ☐ | ☐ ☐ 100 ☐ 900000 ☐ ☐ | Hazard ratio |
| Symbol FUT9 | 000000000000 ☐ 0.0-0 ☐ 0000000 0.0 | ☐ 0- ☐ -0-0-0-0 ☐ 000 | 0.0- ☐ 0000 ☐ | ☐ 1000 ☐ 0.00 ☐ 00000000000 ☐ | 9 |
| FUT5 | ☐ 0.000000 ☐ 0.0 | 000000000000 ☐ ☐ ☐ | ☐ 00000000000000000000 | ☐ 00 ☐ 00000000000000 ☐ | 5 |
| FUT10 | 0.000000 ☐ 1000 ☐ ☐ 0-9 | 0.00 ☐ ☐ -0-0-00 ☐ | ☐ ☐ 10.000 ☐ | 0-0-0 ☐ 100 ☐ 00.0000 000000000 ☐ 0.000 ☐ 0 | 10 |
| FUT1 | ☐ 0.0000 ☒ ☐ -0-0-6 0-0 | ☒ 0-0-0-0-0+ 000 ☐ | 0-0-0- ☐ 0 ☒ 00 ☐ 100.000 | ☐ 00000 00.0000 ☐ 000 0.000.000 | Cox P value |
| FUT7 | ☐ 00.00 ☐ -0.00 0000000 0-6 | ☐ ☐ ☐ ☐ | ☐ ☐ ☐ ☐ ☐ ☐ | ☐ 00 ☐ ☐ 0 ☐ 9000000000 ☐ | <= 0.05 >0.05 |
| FUT3 | 0.0000 0.000 ☐ 0-0 | ☐ 0000000000 ☐ ☐ | 0.0-0 ☐ ☐ 20000 ☐ | 00000000000000 ☐ 0.0-0 ☐ 2.0 | |
| FUT6 | ☐ 0 ☐ | 00000000099900 ☐ 00000600 0 ☐ 1000 ☐ 00 | ☐ 00000000000 ☐ 0000 | 00000000000000 ☐ ☐ 9.0.0 0 ☐ 0.0000 ☐ |
Fig. 2 Differential expression of FUTs across cancers and their impact on pathway activity and prognosis. A The mRNA differences between normal samples and tumor samples. B The combined percentage of the effect of FUTs on pathway activity. C Differences in survival between patients with high and low gene expression. Red points indicate poor survival in the high-expression group, and blue points indi- cate poor survival in the low-expression group. The size of the point represents the statistical significance, where the larger the dot size is, the greater the statistical significance
these findings, POFUT2 may serve as a cancer driver gene in ACC and LIHC and enhance the progression of BRCA, CESC, COAD, LGG, and SARC.
3.5 POFUT2 is involved in cancer immune infiltration and cytokine-mediated immune modulations
We analyzed the correlation between POFUT2 and 14 functional states in different tumors using CancerSEA data. The POFUT2 was positively correlated with angiogenesis, differentiation, hypoxia, inflammation, metastasis, and quiescence in OV, CRC, and RB et al. (P < 0.05) (Fig. 4A). However, POFUT2 was negatively associated with DNAdam- age, DNArepair, EMT, invasion in OV, RB, UM et al. The cancers showing positively POFUT2 correlations with most
Discover
A
00
POFUT2 Expression Level (log2 TPM)
*
**
anci squamous carcinoma
gastric adenocarcinoma
Cancer
₼
esophageal cancer
D
neurodegenerative
smoking measured
N
ACC.Tumor (n=79)
BLCA. Tumor (n=408)
BLCA.Normal (n=19)
BRCA. Tumor (n=1093)
BRCA.Normal (n=112)
BRCA-Basal. Tumor (n=190)
BRCA-Her2.Tumor (n=82)
BRCA-LumA. Tumor (n=564)
BRCA-LumB. Tumor (n=217)
CESC. Tumor (n=304)
CESC.Normal (n=3)
CHOL. Tumor (n=36)
CHOL. Normal (n=9)
COAD. Tumor (n=457)
COAD.Normal (n=41)
DLBC. Tumor (n=48)
ESCA. Tumor (n=184)-
ESCA.Normal (n=11)-
GBM. Tumor (n=153)-
GBM.Normal (n=5)
HNSC. Tumor (n=520)
HNSC.Normal (n=44)
HNSC-HPV+.Tumor (n=97)
HNSC-HPV -. Tumor (n=421)-
KICH. Tumor (n=66)
KICH.Normal (n=25)
KIRC. Tumor (n=533)
KIRC.Normal (n=72)
KIRP. Tumor (n=290)
KIRP.Normal (n=32)
LAML. Tumor (n=173)
LGG. Tumor (n=516)
LIHC. Tumor (n=371)
LIHC. Normal (n=50)
LUAD. Tumor (n=515)
LUAD.Normal (n=59)
LUSC. Tumor (n=501)
LUSC.Normal (n=51)-
MESO. Tumor (n=87)
OV. Tumor (n=303)
PAAD. Tumor (n=178)
PAAD.Normal (n=4)
PCPG. Tumor (n=179)
PCPG.Normal (n=3)
PRAD. Tumor (n=497)-
PRAD.Normal (n=52)
READ. Tumor (n=166)
READ.Normal (n=10)
SARC. Tumor (n=259)
SKCM. Tumor (n=103)
SKCM.Metastasis (n=368)
STAD. Tumor (n=415)
STAD.Normal (n=35)
TGCT.Tumor (n=150)
THCA. Tumor (n=501)
THCA.Normal (n=59)-
THYM. Tumor (n=120)
UCEC. Tumor (n=545)
UCEC.Normal (n=35)
UCS. Tumor (n=57)
UVM.Tumor (n=80)
P
gastric adenocarcinoma
esophageal
C
M
₡
Expression-log:(TPM+1)
₪
¥
₥
₪
0
—
CESC
(num[T]=306; num(N)=13)
(num(T)=163; num(N)=207)
GBM
LAML
(num(T)=173; num(N)=70)
LUSC
(num(T)=486; num[N]=338)
OV (num[T]=426; num(N)=88)
READ (num(T)=92; num(N)=318)
TGCT (num(T)=137; num(N)=165)
THCA (num(T)=512; num(N)=337)
UCEC
(num[T]=174; num(N)=91)
D
log10(HR)
ENSG00000186866.16 (POFUT2)
0.3
0.0
OS
ACC
BLCA
BRCA
CESC
CHOL
COAD
DLBC
ESCA
GBM
HNSC
KICH
KIRC
KIRP
LGG
LIHC
LUAD
LUSC
MESO
OV
PAAD
PCPG
PRAD
READ
SARC
SKCM
STAD
TGCT
THCA
THYM
UCEC
UCS
UVM
-0.3
log10(HR)
0.6
ENSG00000186866.16 (POFUT2)
0.3
0.0
DFS
ACC
BLCA
BRCA
CESC
CHOL
COAD
DLBC
ESCA
GBM
HNSC
KICH
KIRC
KIRP
LGG
LIHC
LUAD
LUSC
MESO
OV
PAAD
PCPG
PRAD
READ
SARC
SKCM
STAD
TGCT
THCA
THYM
UCEC
UCS
UVM
-0.3
-0.6
ACC
LIHC
Overall Survival
Disease Free Survival
Overall Survival
Disease Free Survival
0
Low POFUT2 Group
Low POFUT2 Group
1.0
High POFUT2 Group
Low POFUT2 Group
1.0
Low POFUT2 Group
Logrank p=0.0052
High POFUT2 Group
‘Logrank p=1.8e-05 HR(high)=4.7
High POFUT2 Group
Logrank p=0.0087
High POFUT2 Group
Logrank p=0.0017
0.8
HR(high)=3.1
p(HR)=8.46-05
HR(high)=1.6
p(HR)=0.0092
HR(high)=1.6
p(HR)=0.008
0.8
0.8
0.8
p(HR)=0.0018
Percent survival
n(high)=38
n(high)=38
n(high)=182 n(low)=182
n(high)=182
0.6
n(low) 38
Percent survival
Percent survival
0.6
n[low) 38
Percent survival
0.6
0.6
n(low)=182
0.4
0.4
0.4
0,4
0.2
0.2
0.2
0.2
0.0
0.0
0.0
0.0
0
50
100
150
0
50
100
150
0
20
40
60
80
100
120
0
20
40
60
80
100
120
Months
Months
Months
Months
immune checkpoint genes, including UVM, READ, DLBC, THYM, THCA, and SKCM et al. (Fig. 4B). However, POFUT2 was negatively associated with most immune checkpoint genes in TGCT. Finally, we compared the POFUT2 expres- sion differences between pre- and post-cytokine treatment in cancer cell lines on web tool TISMO (Fig. 4C). We discovered that the POFUT2 expression decreased after the IFN-g treatment in four cell lines, and it also decreased in one IFN-b and one TNF-a posttreatment cell line. The results from multiple perspectives demonstrated that POFUT2 is a critical factor in immunosuppressive environment construction in many cancers, probably via suppressing
Discover
A
Angiogenesis
Apoptosis
CellCycle
Differentiation
DNAdamage
DNArepair
Hypoxia
Inflammation
Invasion
Metastasis
Proliferation
Quiescence
=
Stemness
B
Type
EMT
.
EDNRB
correlation coefficient
3.2
IL12A
“ON
significant
datasets
positive
I negtive
CD276
-1.0-0.5 0.0 0.5 1.0
VEGFA
pValue
0
ARGI
AML
1.0
.0
KIR2DL1
0.0
0.5
1.0
Blood
KIR2DL3
Type:
ALL
-0.8
0.8
VTCN1
Inhibitory Stimulaotry
-0.5
0.5
HAVCR2
GEM
-0.3
0.3
IL10
Glioma
0.0
.
0.0
CD274
Correlation
LAG3
CNS/brain
AST
PDCDI
CTLA4
HCG
TIGIT
ODC
IL13
IL4
LUAD
BTLA
Lung
IDO1
NSCLC
SLAMF7
VEGFB
Skim
MEL
ADORA2A
Kidney
RCC
TGFB1
C10orf54
Breast
BRCA
HMGBI
HNSCC
ENTPDI
Head and neck
TLR4
Ovary
OV
BTN3A1
BTN3A2
Bowel
CRC
TNFSF4
CD27
RR
TNFRSF18
Eye
UM
TNFRSF4
IFNA2
TNFSF9
C
IFNA1
4T1_GSE110912(n=6)
-
**
ILIA
4T1_XW33589424(n=15)
-
IFNG
4T1_RTM28723893(n=12)
P
IL2
B16_GSE149824(n=8)
CXCL10
B16_SSG33589424(n=16)
CXCL9
SELP
B16_GSE110708(n=6)
-
-
CD70
B16_GSE107670(n=6)
I
GZMA
B16_GSE106390(n=6)
CCL5
B16_GSE85535(n=7)
Baseline
IFNb
CX3CL1
B16_RTM28723893(n=8)
ICOSLG
-
**
IFNg
CT26_RTM28723893(n=12)
TGFb1
TNFRSF14
m
.
TNFa
CD40
E0771_XW33589424(n=4)
.
ICAMI
EMT6_XW33589424(n=6)
ITGB2
KPC_RTM28723893(n=12)
* IFNb vs. Baseline
PRFI
-
m
IFNg vs. Baseline
LLC_RTM28723893(n=44)
* TGFb1 vs. Baseline
CD40LG
MC38_GSE112251(n=12)
TNFa vs. Baseline
ICOS
Underscored if left side is larger
CD28
MC38_RTM28723893(n=48)
CD80
MOC1_RU31562203_LZ5733(n=18)
IL2RA
MOC2_RU31562203(n=7)
TNFRSF9
MOC22_RU31562203(n=4)
ILIB
2
TNF
Panc02_RTM28723893(n=12)
-
4
**
Renca_RTM28723893(n=11)
4
5
6
7
Pofut2 log(TPM)
COADR
GBM
D
T cell regulatory (Tregs)
*
*
…
**
*
T cell gamma delta
*
**
*
**
*
**
**
*
*
**
T cell follicular helper
**
*
**
**
T cell CD8+
**
*
**
*
**
**
**
*
*
**
**
*
T cell CD4+ naive
…
*
..
T cell CD4+ memory resting
**
*
**
**
*
T cell CD4+ memory activated
**
*
*
**
**
*
**
*
Neutrophil
**
**
*
*
*
* p <0.05
NK cell resting
.
.
.
..
*
..
..
*
..
** p < 0.01
NK cell activated
…
..
..
..
..
Một
.
M
…
·
CIBERSORT
*** p<0.001
Myeloid dendritic cell resting
*
*
**
*
*
Correlation
Myeloid dendritic cell activated
*
*
0,50
Monocyte
**
**
*
*
**
*
0.25
0.00
Mast cell resting
…
-0.25
Mast cell activated
..
.
·
.
…
Macrophage M2
*
*
.
*
.
**
Macrophage MI
*
**
*
*
*
**
Macrophage MO
*
*
**
**
**
*
**
*
Eosinophil
*
…
B cell plasma
.
-
.
*
..
B cell naive
*
**
B cell memory
*
*
*
*
*
*
*
**
ACC
BLCA
BRCA
3533
CHOL
COAD
DLBC
ESCA
GBM
HNSC
KICH
KIRC
KIRP
LAML
007
LIHC
avni
LUSC
MESO
OV
PAAD
PCPG
PRAD
READ
SARC
SKCM
STAD
TOCT
THCA
THYM
UCEC
UCS
UVM
Discover
A
pc
cell_id
PC3
VCAP
A375
A549
HA1E
HCC515
HT29
MCF7
HEPG2
summary
#
ts_pc
name
description
1
oligomycin-a
ATP synthase inhibitor, ATPase inhibitor
2
rhapontin
apoptosis inducer
3
damnacanthal
src inhibitor
4
antimycin-a
antibiotic, electron transfer inhibitor
5
GW-7647
PPAR receptor agonist
6
ziprasidone
dopamine receptor antagonist, serotonin receptor antagonist, norepinephrine reuptake inhibitor, serotonin receptor agonist, serotonin reuptake inhibitor
7
TCS-359
8
VU-0400195-3
FLT3 inhibitor
glutamate receptor modulator
9
W-12
BRD-A61599461
calmodulin antagonist
10
thyroid-stimulating hormone receptor inverse agonist
11
PKCbeta-inhibitor
PKC inhibitor
12
navitoclax
BCL inhibitor, apoptosis stimulant
13
mesulergine
BRD-K86682249
dopamine receptor agonist
14
tyrosine phosphatase inhibitor
15
buphenine
adrenergic receptor agonist
16
cyclopenthiazide
17
BRD-K28452084
diuretic, inhibitor of sodium chloride symporter
GSK-0660
opioid receptor agonist
18
PPAR receptor inhibitor
19
carmofur
thymidylate synthase inhibitor, pyrimidine antagonist
20
FPL-64176
calcium channel agonist, L-type calcium channel activator
21
entinostat
HDAC inhibitor, cell cycle inhibitor
22
CGP-13501
GABA receptor modulator
23
tadalafil
phosphodiesterase inhibitor
24
quinine
cytochrome P450 inhibitor, hemozoin biocrystallization inhibitor, P glycoprotein inhibitor
25
loreclezole gibberellic-acid BRD-K07872006
GABA receptor agonist
26
NFKB pathway modulator
27
CGP-54626
lipoxygenase inhibitor
28
GABA receptor antagonist
29
5’-guanidinonaltrindoleopioid receptor antagonist
30
norcyclobenzaprine
adrenergic receptor ligand, serotonin receptor ligand
-100.00
-50.00
0.00
50.00
100.00
B
oligomycin-a
C
rhapontin
D
damnacanthal
E
antimycin-a
Binding Affinity -9.5 kcal/mol
Binding Affinity -8.7 kcal/mol
Binding Affinity -8.9 kcal/mol
Binding Affinity
-9.1 kcal/mol
PRO
VAL A 357
PRO A:95
ALA
ASP
A:368
ARG A:88
A
VAL
A:367
PHE
A:389
:29
LYS
ASP
A:292
A 292
A 296
A:364
Interactions
Attanctive Charge
Coonestienil Mpdrnem Bond
Interactines
Conventional Myileogen Bond
Fi-Caties
immunostimulator function and immune checkpoint effects. Our analysis of POFUT2 expression in immunocytes allowed us to delve further into its role in cancer immunity. We first applied the CIBERSORT algorithm to obtain 22 immunocyte correlations with POFUT2. In several malignancies, we found that POFUT2 had a negative association with CD4 + memory resting T cells and resting NK cells, but a strong positive association with gamma delta T cells, CD8 +T cells, CD4 + memory active T cells, activated NK cells, and memory B cells. Furthermore, in six tumors, Tregs exhibited a positive association with POFUT2 (Fig. 4D).
Discover
3.6 Mechanism of molecular docking-based drug binding to their targets
In light of the suboptimal therapeutic outcomes observed in patients with high-POFUT2 cancer who are undergoing standard chemotherapy, our objective was to identify prospective anti-POFUT2 drugs that could increase cancer sensi- tivity to existing chemotherapy while exerting greater effects. Compounds that induced transcriptional changes in nine distinct tumor cell lines that were contrary to those upregulated by high-POFUT2 expression were filtered with the cMap tool. The thirty compounds with the greatest potential to target POFUT2 were then presented (Fig. 5A). To determine whether these compounds are capable of binding to the POFUT2 protein, molecular docking was undertaken between the POFUT2 protein and potential drugs. The analysis encompassed the determination of the binding energy associated with every target-drug interplay by considering the binding modes exhibited by the targets and their potential drugs (Fig. 5). Strong hydrogen bonding and electrostatic interactions were the primary means by which each potential drug bound to its protein target, as shown by the results. Moreover, these potential drugs successfully occupied each target’s active site. The binding energy for the POFUT2-Oligomycin-a complex is -9.5 kcal/mol, for the POFUT2-Rhapontin complex
A
UVM
*#
B
Survival between high and low GSVA score group
UCS
UCEC
*#
UVM
THYM
*
*#
*#
*#
STAD
☐
☐
☐
THCA
*#
*#
*#
*#
*#
*#
*#
*#
READ
TGCT
*
KIRC
☐
☐
STAD
.
*#
*#
*
*#
HNSC
SKCM
*#
*#
*#
*#
UCS
SARC
*#
UCEC
Hazard ratio 0
*#
*#
*#
*#
*#
READ
*#
*#
THYM
PRAD
*#
*#
*#
*#
THCA
1
PCPG
TGCT
PAAD
*#
*#
SKCM
2
OV
Cancer type
SARC
MESO
PRAD
☐
LUSC
*#
*#
*#
*#
LUAD
Cancer type
PCPG
4
*#
*#
LIHC
PAAD
*
*#
*#
*#
OV
Cox P value
LGG
*
*#
*#
*#
KIRP
*#
⇐ 0.05
*
*
*
*
*#
LUSC
KIRC
*#
*#
*#
*#
LUAD
>0.05
KICH
*
LIHC
HNSC
*#
*#
LGG
GBM
LAML
-Log10(FDR)
ESCA
*#
*#
*#
KIRP
1
DLBC
KICH
COAD
*#
*#
GBM
☐ 2
CHOL
ESCA
☐ 3
CESC
.
*
DLBC
☐ 4
BRCA
*#
*#
*#
*#
COAD
☐ 5
BLCA
*#
*#
*#
*#
*#
CHOL
ACC
*#
*#
*
Apoptosis
CellCycle
DNADamage
EMT
Hormone AR
Hormone ER
PI3KAKT
RASMAPK
RTK
TSCmTOR
CESC
BRCA
BLCA
Cancer related pathway
MESO
ACC
Spearman cor.
OS
PFS
DSS
DFI
-0.5
0.0
0.5
C
OS survival of GSVA score in ACC
D ☐
PFS survival of GSVA score in ACC
E
DSS survival of GSVA score in ACC
1.00
1.00
1.00
0.75
0.75
0.75
OS probability
PFS probability
DSS probability
0.50
Higher GSVA, n=39
Lower GSVA, n=40
0.50
Higher GSVA, n=39
0.50
Higher GSVA, n=38
Lower GSVA, n=40
Lower GSVA, n=39
0.25
Logrank P value
0.011
0.25
Logrank P value = 0.021
0.25
Logrank P value =0.0049
0.00
0.00
0.00
0
50
100
150
0
50
100
150
50
Time (month)
0
100
Time (month)
Time (month)
150
Discover
is -8.7 kcal/mol, for the POFUT2-Damnacanthal complex is -8.9 kcal/mol, and for the POFUT2-Antimycin-a complex is -9.1 kcal/mol, demonstrating exceptionally stable binding (Fig. 5B-5E). Taken together, Oligomycin-a, Rhapontin, Dam- nacanthal, and Antimycin-a were determined to be candidate POFUT2-targeted drugs and may be effective in cancer as alternative therapies.
3.7 Single-cell examination of the function of the FUT GSVA score
Ten functional states in various tumors were evaluated for their correlation with the FUT GSVA in our analysis. In ACC, the FUT GSVA scores were correlated positively with apoptosis, EMT, and hormone AR (p <0.05, Fig. 6A).
A
0
10
12
12
12
13
13
13
B
13
13
13
13
13
12
12
11
11
9
7
7
4
4
4
3
FUT6
10
Coefficients
Partial Likelihood Deviance
1
0
PUTtra
FUT3
2
-20
9
a
-40
FUT9
00
0
20
40
60
-6
-5
-4
-3
-2
L1 Norm
Log(2)
C
D
4
RiskType
High_risk
groups=High groups
groups=Low groups
Low_risk
1.00
Cumulative hazard
1.5
3
..
IL
0
Riskscore
0.75
15
Overall survival probability
2
000
0
2.5
5
7.5
10
12.5
Time
0.50
1
…
0.25-
Log-rank P = 0.000354 HR(High groups)-5.211 95%CI(2.106, 12.894)
Status
Alive
0.00
Median time: 3.3
Dead
10
groups=High groups
39
22
7
4
2
1
groups-Low groups
40
28
17
6
2
1
0
2.5
5
7.5
10
Time
E
Time (years)
12.5
5
1.00
0.75
0
FUT11
True positive fraction
0.50
FUT6
FUT4
0.25
FUT2
Type
2-Years,AUC=0.826,95%CI(0.733-0.918)
FUTI
4-Years,AUC=0.813,95%CI(0.717-0.909)
6-Years,AUC=0.852,95%CI(0.746-0.957)
0.00
- 8-Years,AUC=0.78,95%CI(0.596-0.963)
z-score of expression
0.00
0.25
0.50
False positive fraction
0.75
1.00
-2-1 0 1 2
Discover
To investigate the association of FUT expression with the survival of cancer patients, OS, PFS, DSS, and DFI were evaluated in patients stratified by FUT scores. A univariate analysis was implemented to ascertain the prognostic value of the FUTs (Fig. 6B). Patients exhibiting a high FUTs score had an unfavorable prognosis, as shown by the OS results for ACC (Fig. 6C) and MESO; this variable was determined to be a risk factor (all p <0.05). According to the PFS findings, an increased FUTs score was linked to an unfavorable prognosis in ACC (Fig. 6D) and served as a patient risk factor (all p <0.05). Additionally, an elevated FUT score was linked to an unfavorable DSS in ACC (Fig. 6E) and was determined to be a risk factor (all p < 0.05). Also, an elevated FUT score was linked to a dismal DFI in LUAD patients and was determined to be a risk factor (all p <0.05). The results suggest that the FUTs score may potentially function as a conventional prog- nostic predictor and may be beneficial in anticipating outcomes for patients diagnosed with diverse tumor types, with particular emphasis on ACC.
3.8 Development of a risk prognostic model utilizing FUTs in ACC
By employing LASSO Cox regression analyses, FUTs with a positive impact on prognosis that were erroneously predicted were eliminated. As illustrated in Fig. 7A, B, the acquired FUTs comprised FUT11, FUT6, FUT4, FUT2, and FUT1. Five selected FUTs were utilized in the development of prognostic risk models via LASSO Cox regression analysis. FUT11, FUT6, FUT4, FUT2, and FUT1 were determined to be risk factors by this analysis. Risk scores were calculated as indicated below: risk score=(0.1203) * FUT1+(0.3332) * FUT2+(0.517) * FUT4+(0.7169) * FUT6+(0.1858) * FUT11.
| A Uni_cox | Pvalue | Hazard Ratio(95% CI) |
|---|---|---|
| FUT1 | 0.00058 | 1.34081(1.13453,1.58459) |
| FUT11 | 0.00077 | 2.03151(1.34418,3.07031) |
| FUT2 | 0.00043 | 1.86589(1.31839,2.64075) |
| FUT4 | 0.00039 | 2.86876(1.60155,5.13864) |
| FUT6 | 0.18503 | 24.98763(0.21422,2914.73603) |
| Age | 0.37934 | 1.01097(0.98667,1.03587) |
| Gender | 0.99857 | 1.00069(0.46855,2.13721) |
| pT_stage | <0.0001 | 3.37759(2.11006,5.40653) |
| pM_stage | le-05 | 6.15025(2.70974,13.95908) |
| pTNM_stage | le-05 | 2.62824(1.71405,4.03) |
1
TTTTTTTTTTm
0.21422 100017502500 Hazard Ratio
| B Mult_cox | p.value | Hazard Ratio(95% CI) |
|---|---|---|
| FUT1 | 0.44413 | 1.08768(0.87702,1.34895) |
| FUT11 | 0.23696 | 1.66778(0.71448,3.89302) |
| FUT2 | 0.00205 | 2.78995(1.45311,5.35665) |
| FUT4 | 0.00352 | 4.33368(1.61854,11.60353) |
| FUT6 | 0.00167 | 13898183.4244(488.87048,395113862171.108) |
| Age | 0.08787 | 1.02784(0.99593,1.06077) |
| Gender | 0.16581 | 2.19214(0.72238,6.65229) |
| pT_stage | 0.00044 | 14.43798(3.25786,63.98533) |
| pM_stage | 0.23939 | 3.80949(0.41039,35.36194) |
| pTNM_stage | 0.06092 | 0.10886(0.01071,1.10698) |
0.01071 Hazard Ratio
C
C-index : 0.88 (0.829-0.931) p<0.001
D
1.0
1
I
1-year
0
10
20
30
40
50
60
70
80
90
100
2-year
Points
0.8
3-year
FUT2
5-year
0
0.5
1
1.5
2
2.5
3
3.5
4
FUT4
0
0.5
1
1.5
2
2.5
3
3.5
Observed(%)
0.6
FUT6
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
pT_stage
T2
T4
0.4
TI
T3
Total Points
0
20
40
60
80
100
120
140
160
180
200
Linear Predictor
0.2
-4
-3
-2
-1
0
1
2
3
4
1-year survival Pro
0.95
0.9
0.8
0.7 0.6
2-year survival Pro
0.0
*
0.95
0.9
0.8
0.7
.60.
0.4 40
0.
0.2
30
0.1
3-year survival Pro
0.0
0.2
0.4
0.6
0.8
1.0
0.95
0.9
0.8
0.7
.60.50.
50.4
.3
).2 0.1
5-year survival Pro
Nomogram-prediced(%)
n=77 d=27 p=6, 25.6666666666667 subjects per group
resampling optimism added, B=200 Based on observed-predicted
0.95
0.9
0.8
0.7 0.60.50.40.30.2
0.1
Gray: ideal
Discover
After that, the median risk score was employed to divide patients into 2 categories: high-risk and low-risk. All patients with ACC were included in the distribution of survival status and risk plots, as shown in Fig. 7C; OS rates were considerably lower among high-risk patients in comparison with low-risk patients, according to the findings. Additionally, the gene heatmap demonstrated that the high-risk group had substantially greater levels of FUT11, FUT6, FUT4, FUT2, and FUT1 (Fig. 7C). Eventually, five FUTs, namely, FUT11, FUT6, FUT4, FUT2, and FUT1, were found to provide predictive value for OS. These 5 FUTs were employed to create a prognostic risk model. By applying the FUT signature to the entire dataset comprising 169 samples, its prognostic significance was validated. OS was found to be shortened in the high-risk patients relative to those at low risk, as determined by KM survival analyses (p=0.000354; Fig. 7D). The AUC value for the FUT signature in the training cohort was 0.826, 0.813, 0.852, and 0.78 at 2, 4, 6, and 8 years (Fig. 7E). Overall, these outcomes show that the ACC prognostic risk model, which was developed utilizing the selected five FUTs, was accurate.
We determined whether differentially expressed FUTs exhibited a strong link to worse prognosis in ACC patients by conducting univariate analyses. This study set out to examine the entire TCGA dataset (n = 169) for any correlation between FUT mRNA expression levels and OS in ACC patients. Through the use of univariate analysis on the entire TCGA dataset (169 patients), we determined the predictive value of FUTs that showed differential expression in ACC patients. Elevations of the four differentially expressed FUTs (FUT11, FUT4, FUT2, and FUT1) were substantially linked to ACC patients’OS (p<0.05, Fig. 8A). Subsequently, it was determined that FUT4, FUT2, and FUT6 affected ACC patients’ OS rates and clinical outcomes, as evidenced by multivariate analyses (p <0.05, Fig. 8B). Moreover, OS was found to be strongly linked to pathological T stage, M stage, and TNM stage in ACC patients, according to univariate analysis (p <0.05, Fig. 8A). Additionally, T stage and TNM stage were linked to ACC patients’s OS as evidenced by multivariate analysis (p <0.05, Fig. 8B). The ACC patients’ prognoses were predicted using a nomogram (Fig. 8C) that integrated the risk score with the T stage. The C-index for survival prediction was 0.88 (P <0.001). Furthermore, the nomogram model’s calibration curves for 1, 2, 3, and 5 years exhibited a high degree of concordance between the nomogram’s predicted values and the actual results (Fig. 8D). High accuracy in anticipating OS for patients with ACC was shown by the integrated prognostic nomo- gram that uses the FUT signature.
4 Discussion
Among the glycosyltransferases (GT), fucosyltransferases (FUTs) stand out for their folded structure, substrate selectivity, the way they interact with substrates (donor and acceptor), and catalytic activity. Interestingly, fucosylated epitopes are hallmarks of cancer cells, and their expression is changed when FUTs are dysregulated [7, 8]. At present, there remains a dearth of research concerning their molecular biology and prospects as drugs for the treatment of cancer. Although the FUTs family has been investigated in some studies as targets for therapy and prognostic markers, its precise appli- cability in pan-cancer remains uncertain. A limited number of studies have examined the potential therapeutic utility and prognostic value of the FUT family; however, its applicability to pan-cancer remains inadequately explored. Due to the intricate interrelationships among FUTs, inflammatory processes, and cancer, this study sought to comprehensively analyze the genetic, immune, and clinical characteristics associated with FUTs in 33 distinct types of cancer.
According to our analysis, tumor samples exhibited a substantially higher CNV rate of FUTs in comparison with normal samples. A direct relationship was identified between CNV and FUT expression; this finding suggests that CNV may influ- ence FUT expression, which in turn may have implications for carcinogenesis and patient prognosis. Subsequently, an examination of epigenetic modifications revealed that aberrant hypermethylation may inhibit FUTs expression. Accord- ingly, SNVs manifest in FUTs with considerable frequency, according to our investigation. Furthermore, a substantial link was observed between SNVs and the FUT expression, implying that SNVs might influence the FUT expression. This might foster carcinogenesis and survival. As a result, we postulate that genetic and epigenetic alterations in FUTs could induce their dysregulation, thereby promoting tumorigenesis.
Then, using the TCGA dataset, we determined the survival implications, associated pathways, and expression levels of FUTs in a variety of types of cancer. Our analysis uncovered that FUTs are present in many different types of tumors, with these genes being significantly upregulated in the majority of tumors. A review of the relevant pathways has demonstrated that these FUTs have the potential to regulate cancer-related pathways. They participate in various signaling pathways, including those associated with apoptosis, EMT, RAS/MAPK, RTK, TSC/mTOR, PI3K/AKT, response to DNA damage, ER, AR, and cell cycle [17-25]. According to the findings, FUTs facilitate the formation of a network of cancer-related pathways, thereby impeding the progression of the disease and improving patient survival. A correlation analysis between FUTs and survival was undertaken to provide further insight into the function of FUTs
Discover
in clinical risk stratification. Survival study revealed a link between FUT overexpression-particularly POFUT2 over- expression-and PFS, DSS, DFS, and OS. Particularly, FUT overexpression was linked to a dismal prognosis in most tumors. This led us to investigate immune infiltration in subsets of TCGA tumors classified by POFUT2 expression.
Dunn et al. in 2002 [26] proposed the idea of cancer immunoediting in 2002. It comprises elimination, equilibrium, and escape as its three constituent processes, which rely on different types of immune cells found in the TME. We found a robust positive correlation between POFUT2 expression and the number of infiltrating immune cells after performing a comprehensive analysis across various types of cancer. Fucosyltransferase 1 and 2 play pivotal roles in breast cancer cells [27, 28]. This work stands out as the first to establish a clear link between immune responses to cancer and FUTs, including POFUT2, to our knowledge. Nevertheless, according to our findings, POFUT2 expression was moderately positively linked to that of gamma delta T cell, CD8 +T cell, CD4 + memory activated T cells, activated NK cell, and memory B cell, and POFUT2 expression was negatively linked to the abundance of CD4 + memory rest- ing T cells and resting NK cell. Insights like these may help clarify why POFUT2 is effective in warding against certain cancers.
Also, we hypothesized that the four candidate POFUT2-targeted drugs identified in this study with viable docking modes would be effective therapies for patients who develop chemoresistance to standard therapies. There are still many limitations in the treatment of cancer, and it is necessary to seek newer and more effective treatment methods, especially the exploration of new targets [29-31]. Nonetheless, additional research must be done to clarify the precise mechanisms that govern the impact of drugs on POFUT2 expression and the advancement of cancer. Furthermore, apoptosis, EMT, and hormone AR are positively linked to the FUT score in ACC. Additional evidence from survival studies supports the use of FUTs for predicting the efficacy of immunotherapies in malignancies, including ACC.
Five genes were utilized in the development of a risk signature by employing LASSO analysis. The reliability of the risk model comprising the five selected FUTs as an independent maker for prognosticating the clinical outcomes in ACC patients was validated. FUT11, FUT6, FUT4, FUT2, and FUT1 were identified to be risk factors in the risk model.
The strength of the correlation between the risk score and the pathological T stage, pathological M stage, and path- ological TNM stage of the tumor was confirmed through clinical relevance analysis. The subgroup survival analysis illustrated that the risk signature composed of five genes accurately predicts the prognosis for various ACC subgroups classified according to their pathological T stage. Additionally, the nomogram models’ calibration curves for 1-, 2-, 3-, and 5-year OS exhibited a high degree of agreement between the predicted values and the actual results. Notably, the ROC curves AUC values representing five-gene signatures at 2, 4, 6, and 8 years exceeded 0.75. Overall, these data confirm that the risk model consisting of five genes has reliable prognostic significance for ACC.
There are a few limitations to our study. First, the temporal complexity associated with dynamic analysis of FUT expres- sion in paired tumor samples necessitates further research for validation. TCGA patient transcriptomic data were utilized to ascertain the correlation between FUTs and cancer; nonetheless, experimental verification of these results is necessary. Even so, our results provide a new understanding of the involvement of FUTs in cancer. Subsequently, variations in FUTs were examined across expression, epigenetic, genetic, and pathway scales. The response to therapy and the prognosis might differ from patient to patient due to these variances. As a result, a comprehensive examination of each type of cancer is required, and therapies must be tailored accordingly.
5 Conclusion
This study provides a comprehensive analysis of the genetic, immune, and clinical features related to FUTs in 33 cancer types. Our findings reveal that FUTs, especially POFUT2, are overexpressed in various tumor types and are significantly associated with cancer-related pathways. The expression of FUTs, especially POFUT2, is linked to poor clinical outcomes, suggesting their potential as prognostic markers. Additionally, POFUT2 expression is positively correlated with cancer immunity, indicating its role in tumor microenvironment remodeling. The study also identifies potential POFUT2-tar- geted drugs that may be effective in treating chemoresistant cancers. Furthermore, FUTs are capable of anticipating the effectiveness of immunotherapies in ACC. A risk signature with five FUTs was developed and shown to be an accurate predictor of clinical outcomes in patients with ACC. Our results offer a novel understanding of the involvement of FUTs in cancer and their potential as therapeutic targets. Future studies should examine the role of FUTs in more tumor types and clarify the mechanisms of their effects on cancer progression.
Discover
| https://doi.org/10.1007/s12672-024-01447-6
Author contributions Z.J., Pan.L. and B.Y. wrote the manuscript; Z.J., B.Y. and Pan.L. participated in the critical revision of the manuscript for important intellectual content. Ping.L provided guidance on the coordination of authors and reviewed the manuscript. All authors reviewed and approved the final manuscript.
Funding Not applicable.
Data availability Data is provided within the manuscript or supplementary information files.
Declarations
Competing interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativeco mmons.org/licenses/by-nc-nd/4.0/.
References
1. Mattiuzzi C, Lippi G. Current cancer epidemiology. J Epidemiol Global Health. 2019;9(4):217-22.
2. Chan TA, Yarchoan M, Jaffee E, et al. Development of tumor mutation burden as an immunotherapy biomarker: Utility for the oncology clinic. Ann Oncol. 2019;30(1):44-56.
3. Chang L, Chang M, Chang HM, et al. Microsatellite instability: a predictive biomarker for cancer immunotherapy. Appl Immunohistochem Mol Morphol. 2018;26(2):e15-21.
4. Cui X, Zhang X, Liu M, et al. A pan-cancer analysis of the oncogenic role of staphylococcal nuclease domain-containing protein 1 (SND1) in human tumors. Genomics. 2020;112(6):3958-67.
5. Hu J, Xu J, Feng X, et al. Differential expression of the TLR4 gene in pan-cancer and its related mechanism. Front Cell Dev Biol. 2021;9: 700661.
6. Miao Y, Wang J, Li Q, et al. Prognostic value and immunological role of PDCD1 gene in pan-cancer. Int Immunopharmacol. 2020;89(Pt B): 107080.
7. Keeley TS, Yang S, Lau E. The diverse contributions of fucose linkages in cancer. Cancers. 2019;11:9.
8. Blanas A, Sahasrabudhe NM, Rodríguez E, et al. Fucosylated antigens in cancer: an alliance toward tumor progression, metastasis, and resistance to chemotherapy. Front Oncol. 2018;8:39.
9. Yan B, Liao P, Shi L, et al. Pan-cancer analyses of senescence-related genes in extracellular matrix characterization in cancer. Disc Oncol. 2023;14(1):208.
10. Akbani R, Ng PKS, Werner HMJ, et al. A pan-cancer proteomic perspective on The Cancer Genome Atlas. Nat Commun. 2014;5:3887.
11. Ye Y, Xiang Y, Ozguc FM, et al. The genomic landscape and pharmacogenomic interactions of clock genes in cancer chronotherapy. Cell Syst. 2018;6(3):314-328.e2.
12. Li T, Fu J, Zeng Z, et al. TIMER2.0 for analysis of tumor-infiltrating immune cells. Nucleic Acids Res. 2020;48(W1):W509-14.
13. Tang Z, Kang B, Li C, et al. GEPIA2: an enhanced web server for large-scale expression profiling and interactive analysis. Nucleic Acids Res. 2019;47(W1):W556-60.
14. Newman AM, Liu CL, Green MR, et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods. 2015;12(5):453-7.
15. Subramanian A, Narayan R, Corsello SM, et al. A next generation connectivity map: L1000 platform and the first 1,000,000 profiles. Cell. 2017;171(6):1437-1452.e17.
16. Clark AM, Labute P. 2D depiction of protein-ligand complexes. J Chem Inf Model. 2007;47(5):1933-44.
17. Cerhan JR, Ansell SM, Fredericksen ZS, et al. Genetic variation in 1253 immune and inflammation genes and risk of non-Hodgkin lym- phoma. Blood. 2007;110(13):4455-63.
18. Dickinson K, Case AJ, Kupzyk K, et al. Exploring biologic correlates of cancer-related fatigue in men with prostate cancer: cell damage pathways and oxidative stress. Biol Res Nurs. 2020;22(4):514-9.
19. Harada K, Hiramoto-Yamaki N, Negishi M, et al. Ephexin4 and EphA2 mediate resistance to anoikis through RhoG and phosphatidylinositol 3-kinase. Exp Cell Res. 2011;317(12):1701-13.
20. Kou X, Jiang X, Liu H, et al. Simvastatin functions as a heat shock protein 90 inhibitor against triple-negative breast cancer. Cancer Sci. 2018;109(10):3272-84.
21. McGranahan N, Favero F, de Bruin EC, et al. Clonal status of actionable driver events and the timing of mutational processes in cancer evolution. Sci Transl Med. 2015;7:283.
22. Pich C, Sarrabayrouse G, Teiti I, et al. Melanoma-expressed CD70 is involved in invasion and metastasis. Br J Cancer. 2016;114(1):63-70.
23. Rulina AV, Mittler F, Obeid P, et al. Distinct outcomes of CRL-Nedd8 pathway inhibition reveal cancer cell plasticity. Cell Death Dis. 2016;7(12): e2505.
Discover
24. Tzeng H-E, Tang C-H, Wu S-H, et al. CCN6-mediated MMP-9 activation enhances metastatic potential of human chondrosarcoma. Cell Death Dis. 2018;9(10):955.
25. Zhao S, Mi Y, Zheng B, et al. Highly-metastatic colorectal cancer cell released miR-181a-5p-rich extracellular vesicles promote liver metas- tasis by activating hepatic stellate cells and remodelling the tumour microenvironment. J Extracell Vesic. 2022;11(1): e12186.
26. Dunn GP, Bruce AT, Ikeda H, et al. Cancer immunoediting: From immunosurveillance to tumor escape. Nat Immunol. 2002;3(11):991-8.
27. Lai T-Y, Chen I-J, Lin R-J, et al. Fucosyltransferase 1 and 2 play pivotal roles in breast cancer cells. Cell death discovery. 2019;5:74.
28. Soejima M, Koda Y. Survey and characterization of nonfunctional alleles of FUT2 in a database. Sci Rep. 2021;11(1):3186.
29. Hong L, Wang X, Cui W, et al. Construction of a ferroptosis scoring system and identification of LINC01572 as a novel ferroptosis suppres- sor in lung adenocarcinoma. Front Pharmacol. 2022;13:1098136.
30. George J, Maas L, Abedpour N, et al. Evolutionary trajectories of small cell lung cancer under therapy. Nature. 2024;627(8005):880-9.
31. Mao W, Cai Y, Chen D, et al. Statin shapes inflamed tumor microenvironment and enhances immune checkpoint blockade in non-small cell lung cancer. JCI Insight. 2022;7:18.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Discover