GEPIA is an interactive online web tool for investigators to explore diverse functional modules based on TCGA and GTEx databases (18). Using the GEPIA database, we analyzed the gene expression profile of FBXO family members across 33 tumor samples and paired normal tissues, including pancreatic adenocarcinoma (PAAD), and performed the pathological stage plot of the FBXOs as well as evaluated the prognostic values of FBXOs in PDAC patients. “Median” was selected as the “Group Cutoff”.

Oncomine is a public database that contains substantial tumor microarray datasets (19). Log2-transformed form was utilized to represent the transcriptional levels of FBXOs. Fold change >2 and p-value <0.05 were selected as the inclusion criteria. UALCAN is a user-friendly database for the analysis of gene expression and clinical parameters from thirty-one different types of tumors, including PDAC (20). In this study, PDAC staging, grading, TP53 mutation, and methylation data were obtained from the UALCAN database.

GEO database includes a large number of high-throughput sequencing data, including PDAC datasets. We used five GEO datasets to compare the expression levels of FBXOs in PDAC tissues and normal pancreatic tissues (GSE16515 and GSE62165) or para-carcinoma tissues (GSE62452, GSE28735, and GSE15471). Besides these, five GEO datasets were used to analyze correlations between the expression levels of FBXOs and the clinicopathologic characteristics of PDAC. GSE21501 was used for the analysis of tumor size (T1–T2 vs. T3–T4) and lymphatic metastasis (N0 vs. N1); GSE62165, GSE62452, GSE19650, and GSE51971 were used for the analysis of tumor location (head vs. body/tail), differentiation degree (G1–G2 vs. G3–G4), different pancreatic precancerous lesions, and cell stemness, respectively.

The Human Protein Atlas (HPA) database is an integrated and accessible data mining platform containing a substantial distribution of information of human protein from more than twenty kinds of cancer in cellular and histopathological levels. The immunohistochemical (IHC) staining images of FBXO family members and the confocal images of their cellular localization were acquired from the HPA database.

The Kaplan–Meier plotter is an interactive and public online database for analyzing the survival of twenty-one tumor types based on substantial RNA-seq and next-generation sequencing (21). The prognostic values of FBXO family members and their prognostic subgroup analysis in gender, mutation burden, and immune infiltration were evaluated by the Kaplan–Meier plotter database. Hazard ratio (HR) and 95% confidence interval (CI) were calculated automatically according to “Auto select best cutoff”.

SurvExpress is a bioinformatics web server for gene expression and clinical data in tumors (22). The prognosis of FBXO family members and the transcriptional levels in low- and high-risk groups were assessed by using this database. “PAAD-TCGA-Pancreatic adenocarcinoma” (n = 176) was selected. The PDAC cohorts were divided into low- and high-risk groups based on the prognostic index to compare the expression levels of FBXOs. LinkedOmics is an available web portal for users to analyze multi-omics data within and across 32 cancer types (23). This database was used to evaluate the prognosis of promoter methylation of the FBXOs and top 200 co-expressed genes with individual FBXOs in PDAC patients.

TIMER is a user-friendly web server for the systematic and comprehensive analysis of immune infiltration across various tumor types via inputting function-specific parameters (24). The immune infiltration of FBXOs in PDAC tissues was estimated using TIMER database (Spearman correlation). Furthermore, TISIDB is a public portal for tumor and immune system interaction as well as integration of numerous heterogeneous data types (25). We used this database to perform a further detailed analysis of immune infiltration of FBXOs in PDAC based on high-throughput data.

The cBioPortal database is an accessible online resource to interactively explore and visualize multidimensional genomics data and clinical profiles in diverse cancer samples (26). Using cBioPortal database, we analyzed the genetic alteration, mutation, and related prognosis of FBXOs in PDAC and identified the top 200 co-expressed genes of FBXOs. The “Pancreatic adenocarcinoma (TCGA, PanCancer Atlas)” (184 samples) was selected. In terms of “Select Genomic Profiles”, the “Mutation”, “Structural Variant”, “Putative copy-number alterations from GISTIC”, “mRNA Expression”, and “Protein expression z-scores (RPPA)” were selected. Besides these, “Complete sample (168)” was selected as the option of “Select Patient/Case Set”.

DAVID is a bioinformatic analytic tool for the functional enrichment of genes derived from high-throughput genomic experiments, such as functional classification and annotation, to acquire an in-depth understanding of the biological function of gene lists (27). Functional enrichment and prediction of FBXOs were performed using gene ontology (GO) and KEGG pathway analysis from DAVID. GO functional enrichment consists of biological process (BP), cellular component (CC), and molecular function (MF).

Human immortal pancreatic epithelial cell line HPDE6 and PDAC cell lines (Panc-1, AsPC-1, SW1990, T3M4, and CAPAC-1) were purchased from the National Infrastructure of Cell Line Resource (Beijing, China). HPDE6, Panc-1, and T3M4 were cultured in DMEM medium (HyClone); AsPC-1, SW1990, and CFPAC-1 were cultured in RPMI-1640, Leibovitz’s L-15, and IMDM medium (Coning), respectively. All media were supplemented with 10% fetal bovine serum (Gibco) and 1% penicillin–streptomycin (Sigma), and cells were incubated in a humidified atmosphere at 37°C and 5% CO₂.

Total RNA was isolated from exponentially divided cell lines using Trizol Reagent (RNAiso Plus, #9109, Takara) and then concentrated with trichloromethane, isopropanol, and ethanol according to standard protocols. The RNA was reverse-transcribed with PrimeScript™ RT reagent Kit (#RR037A, Takara). Real-time qPCR was administrated in three duplicates for each sample with TB Green^® Premix Ex Taq™ (#RR420A, Takara) on the QuantStudio3 apparatus (ThermoFisher, USA). GAPDH was used for standard normalization, and the data were analyzed using the 2^−ΔΔCT method. The primer sequences were summarized in Supplementary Table S1 .

Statistical analysis and graphs were performed and plotted by GraphPad Prism 6.0 (Lajolla, CA, USA). Comparisons between the two groups were conducted using a two‐tailed Student’s t-test. For comparisons of three or more groups, one-way ANOVA with post-hoc Dunnett test or Tukey’s test was utilized. Statistical significance was indicated as p-value <0.05.

To investigate the expression profiles of FBXO family members in PDAC, the GEPIA database, an online analysis tool based on TCGA and GTEx datasets, was used to compare the difference between the mRNA expression levels of FBXOs in PDAC tissues (N = 179) and normal pancreatic tissues (N = 171). A total of twenty-one aberrantly expressed genes were identified within the human PDAC tissues. Of them, FBXO1, FBXO5, FBXO7, FBXO8, FBXO11, FBXO13, FBXO18, FBXO20, FBXO22, FBXO23, FBXO28, FBXO32, FBXO34, FBXO38, FBXO42, FBXO45, and FBXO46 were upregulated, while FBXO2, FBXO12, FBXO35, and FBXO50 were downregulated in PDAC tissues ( Figure 1 ). No abnormal expression was found in other FBXO members ( Supplementary Figure S1 ). By consulting the previously published literature, there are extremely limited reports related to FBXO1, FBXO20, FBXO22, FBXO28, FBXO32, and FBXO45 in PDAC, and we therefore further explored the six genes. Furthermore, the transcriptional levels of six-FBXOs were explored in 32 other types of tumor. We found that six-FBXOs were also abnormally expressed in the other multiple tumors ( Supplementary Figure S2 ). Next, Oncomine databases were also used to investigate the expression of six-FBXOs in PDAC tissues versus normal pancreatic tissues. The results confirmed that the mRNA expression levels of FBXO20, FBXO32, and FBXO45 were strongly upregulated in patients with PDAC in different datasets ( Supplementary Figures S3A–D and Table 1 ). In Buchholz’s data, FBXO32 exhibited a higher expression in pancreatic duct intraepithelial neoplasia (PanIN) than in normal pancreatic tissues ( Supplementary Figure 3C ). Moreover, five GEO datasets were collected to verify the abnormality in the expression patterns of six-FBXOs found above via a comparison between PDAC and normal pancreatic tissues (GSE16515 and GSE62165) or adjacent non-cancerous tissues (ANCT) (GSE62452, GSE28735, and GSE15471). As shown in Figure 2 , the overall results confirmed that the expression levels of six-FBXOs were robustly elevated, although FBXO1 and FBXO22 did not show statistical differences in two of the datasets ( Figures 2A–E ).

Then, we evaluated the clinical significance of six-FBXOs in PDAC by using three GEO datasets (GSE62165, GSE21501, and GSE62452), GEPIA, and UALCAN databases. The data from GSE62165 showed that none of the six-FBXOs was associated with tumor location (N = 118) ( Supplementary Figure S4A ), while GSE21501 displayed that FBXO28 and FBXO32 were tightly linked to tumor size (N = 98); FBXO32 was also related to lymphatic metastasis (N = 101) ( Figures 3A, B ). GSE62452 demonstrated that the higher expression levels of FBXO1, FBXO20, FBXO32, and FBXO45 were correlated with poorer tumor differentiation (N = 68), apart from FBXO22 and FBXO38 ( Figure 3C ). Besides this, the GEPIA database manifested that, among six-FBXOs, FBXO32 was associated with tumor staging (N = 179); FBXO1, FBXO20, and FBXO45 had a related trend with tumor staging, whereas no correlation with tumor staging was observed for FBXO22 and FBXO28 ( Figure 3D ). It was worth mentioning that FBXO family members FBXO8, FBXO13, and FBXO34 were linked to tumor staging, and FBXO50 had an associated trend with tumor staging ( Supplementary Figure S4B ).

The results from the UALCAN database further confirmed the GSE21501 data that FBXO32 was closely relevant in the lymphatic metastasis of PDAC as well as showed that FBOX20 expression was relatively lower in PDAC patients with diabetes, compared with PDAC patients without diabetes ( Supplementary Figure S4C ). Furthermore, the UALCAN database displayed a more detailed analysis regarding tumor differentiation and staging based on TCGA samples. The data from the UALCAN database verified that the expression levels of six-FBXOs were strongly relevant in tumor differentiation, and patients who were in worse differentiation had the higher expression of six-FBXOs on the whole ( Supplementary Figure S5A ). Further validating the results of the GEPIA database, FBXO32 was related to the tumor stage of PDAC patients. It also indicated that the mRNA expression levels of FBXO20 and FBXO45 were closely related to the tumor stage of PDAC patients ( Supplementary Figure S5B ), which meant that patients who were in more advanced tumor staging tended to express higher transcripts of FBXO20, FBXO32, and FBXO45. TP53 mutation plays a crucial role in the initiation, formation, and maintenance of PDAC (28). Therefore, we also explored the effect of P53 mutation on the expression of six-FBXOs. The data showed that FBXO1, FBXO20, and FBXO45 exhibited higher expression levels in the TP53-mutant group than in the TP53-nonmutant group, while this kind of difference was not observed for FBXO22, FBXO28, and FBXO32 ( Supplementary Figure S5C ).

Pancreatic carcinogenesis is a gradual and multi-step process from precancerous lesions (8). We compared the expression of six-FBXOs in different precancerous lesions, including intraductal papillary-mucinous neoplasm (IPMN), intraductal papillary-mucinous carcinoma (IPMC), and intraductal papillary-mucinous adenoma (IPMA) by using the GSE19650 dataset. The results indicated that FBXO20 was highly expressed in these three kinds of precancerous lesions (IPMA, IPMC, and IPMN) in comparison to normal pancreatic tissues, and FBXO20 had the highest expression level among these three pathological types, while FBXO22 was downregulated in IPMA compared with corresponding normal tissues ( Figure 3E ). FBXO32 and FBXO45 were significantly enhanced in IPMC, and FBXO32 exhibited the highest expression level among the other pathological types ( Figure 3E ). Taken together, these results demonstrated that FBXO20, FBXO22, FBXO32, and FBXO45 may act as markers of precancerous lesions and have the value of early diagnosis of pancreatic cancer.

Promoter methylation is a critical epigenetic modification to regulate gene expression in human cancers (29). Therefore, we next investigated the promoter methylation levels of six-FBXOs. The results illustrated that FBXO1, FBXO20, and FBXO32 had lower methylation levels in primary tumors than in normal pancreatic tissues, while FBXO45 had a higher methylation level in PDAC ( Supplementary Figure S6A ). In six-FBXOs, there were considerable differences in the promoter methylation level among divergent differentiated degrees or clinical stages. The methylation levels of FBXO1, FBXO20, FBXO32, and FBXO45 were associated with tumor differentiation ( Supplementary Figure S6B ). The methylation levels of FBXO1, FBXO20, FBXO22, FBXO32, and FBXO45 were correlated with clinical stages ( Supplementary Figure S7A ). In particular, FBXO1 exhibited a robust negative correlation between promoter methylation levels and differentiated degree or clinical stages. In addition, FBXO1, FBXO20, and FBXO45 had a lower methylation degree in the TP53-mutant group than in the TP53-nonmutant group, while FBXO22, FBXO28, and FBXO32 had no significant correlation between methylation level and TP53-mutant ( Supplementary Figure S7B ).

Next, we probed into the protein levels of six-FBXO expression between PDAC tissues and normal pancreatic tissues from the HPA database. The IHC staining of six-FBXOs revealed that the protein expression levels of FBXO1, FBXO20, and FBXO45 were markedly increased compared with the corresponding normal pancreatic tissues ( Figure 4A ), which corroborated the aforesaid result. The IHC data of FBXO22, FBXO28, and FBXO32 were pending cancer tissue analysis in the HPA database. Besides this, GSE45757 was used to evaluate the six-FBXO expression in human normal pancreatic ductal epithelial cells (hn-PDEC) and PDAC cell lines, including Panc-1, BxPC-3, Capan-2, Mia PaCa-2, SW1990, SU86.86, CFPAC-1, HPAF_II, and Hs766T. The data demonstrated that FBXO1, FBXO20, FBXO22, and FBXO28 were strongly upregulated in multiple PDAC cell lines relative to hn-PDEC ( Figures 4B–E ). The expression levels of FBXO32 and FBXO45 in cell lines were not available in the GSE45757 dataset. The cellular localization of gene expression often determines its corresponding significant function, so we explored the subcellular localization of six-FBXOs by using confocal images of the HPA database. The results warranted that FBXO1 was detected in the centrosome, FBXO20 was detected in actin filaments and cytosol, FBXO28 was detected in the nucleoplasm and focal adhesion sites, FBXO32 was detected in the nucleoplasm and cytosol, and FBXO45 was detected in the cytosol ( Figure 4F ). The confocal result of FBXO22 was absent in the HPA database. More importantly, we also explored the cell stemness of six-FBXOs in PDAC by using the GSE51971 dataset, whose data were classified into triple-positive group and triple-negative group according to whether these expressed the three important cancer stem cell markers, CD44, CD133, and EpCAM. The results showed that the expression levels of FBXO22, FBXO28, and FBXO32 were potently enhanced in the triple-negative group ( Figure 4G ). Overall, these results suggested that six-FBXOs are likely to play a vital role in PDAC initiation and progression.

In order to investigate the prognostic values, we explore the prognostic data of six-FBXOs from three primary public databases: GEPAI, Kaplan–Meier plotter, and SurvExpress. The data from GEPAI displayed that the high levels of six-FBXOs were tightly linked to a decreased probability of OS and DFS in PDAC (89 vs. 89) ( Supplementary Figure S8 and Table 2 ). The Kaplan–Meier plotter further substantiated the aforesaid prognostic results ( Figure 5 and Table 2 ). Specifically, FBXO20 had the highest HR in OS and DFS at 2.73 and 9.6, respectively, which meant that the FBXO20 high-expression group had 2.73 times and 9.6 times the risk of PDAC-related death and relapse, respectively, versus the FBXO20 low-expression group. The median OS time of the low- and high-expression groups was 18.93 and 10.27 months, respectively. More detailed prognostic information was shown in Figure 5 and Table 2 . Additionally, the SurvExpress online web tool also confirmed that six-FBXOs were the negative prognosis factors of PDAC patients ( Supplementary Figures S9 , S10A ). The respective six-FBXO expression levels of PDAC patients in the SurvExpress datasets were divided into low- and high-risk groups (88 vs. 88) according to the prognostic index. The data from SurvExpress revealed that six-FBXOs had higher expression levels in the high-risk group than in the low-risk group ( Supplementary Figure S10B ). The LinkedOmics databases demonstrated that the hypermethylation group of FBXO1, FBXO20, and FBXO32 indicated a better OS than the hypomethylation group, while no significant correlation with OS was observed for FBXO22, FBXO28, and FBXO45 ( Supplementary Figure S11 ). Collectively, these results in this section outlined the important prognostic values of six-FBXOs, which may act as effective prognostic markers for PDAC patients.

Then, we analyzed the relationship between the expression of six-FBXOs and the immune infiltration levels in PDAC using the TIMER online tool. The data revealed that the expression levels of FBXO1, FBXO22, FBXO28, and FBXO45 were associated with B cell infiltration; FBXO20, FBXO22, FBXO28, FBXO32, and FBXO45 were correlated with CD8⁺ T cell infiltration; FBXO20 and FBXO45 were inversely related to CD4⁺ T cell infiltration; FBXO22, FBXO28, FBXO32, and FBXO45 were linked to macrophage and neutrophil cell infiltration; FBXO1, FBXO22, FBXO28, FBXO32, and FBXO45 were relevant in dendritic cell infiltration ( Supplementary Figure S12 ). However, it was contradictory that these negative prognostic genes were positively associated with immune infiltration, especially CD8⁺ T cells, which often predicted a good clinical outcome. We speculated that this was related to whether immune cells were activated and contained different subgroups. Therefore, the TISIDB databases were utilized to perform a more detailed immune infiltration analysis of six-FBXOs in PDAC among subgroups with different functions. The results from TISIDB showed that, on the whole, six-FBXOs had a negative association with various immune infiltration, especially activated B cell, activated CD8⁺ T cell, and macrophage infiltration ( Figures 6A, B ), indicating the expression of six-FBXOs might be associated with a suppressive tumor immune microenvironment in PDAC.

To determine whether these six-FBXOs affected the prognosis of immune subgroup in PDAC patients, the Kaplan–Meier plotter was used to evaluate it between the immune cell-enriched group and the immune cell-decreased group. The data from the Kaplan–Meier plotter illustrated that FBXO1, FBXO20, FBXO22, FBXO28, and FBXO32 were the negative prognosis factors in the B cell-decreased group and the macrophage-decreased group, while FBXO45 was the negative prognosis factor in the B cell-enriched group and the macrophage-enriched group. FBXO20 and FBXO45 were the negative prognosis factors in the CD8⁺ T cell-decreased group, while FBXO1, FBXO22, FBXO28, and FBXO32 were the negative prognosis factors in the CD8⁺ T cell-enriched group. All six-FBXOs were the negative prognosis factors in the regulatory T cell-enriched group ( Supplementary Figure S13 ). On the whole, the six-FBXOs tended to be the negative prognosis factors in the immunosuppressed state of PDAC. Furthermore, we also detected the prognostic analysis of subgroups in different gender or mutation burden. The results showed that FBXO1 and FBXO32 were the negative prognosis factors in the female group, whereas FBXO20, FBXO22, FBXO28, and FBXO45 were the negative prognosis factors in the male group. FBXO1, FBXO20, FBXO22, FBXO28, and FBXO32 were the negative prognosis factors in the low-mutation-burden group, and FBXO45 was the negative prognosis factor in the high-nutation-burden group ( Supplementary Figure S13 ). For the subgroup analysis of DFS, the data are shown in Supplementary Figure S14 . The HR, 95% CI, and p-value for the OS and DFS of six-FBXOs in different subgroups are summarized in Supplementary Table S2 .

We investigated the genetic alterations of the six-FBXO and the mutation in the cBioPortal database for pancreatic adenocarcinoma (TCGA and PanCancer Atlas). The results indicated that the genetic alteration of six-FBXOs in PDAC patients accounted for more than 40% ( Figure 7A ). Genetic alterations account for about 10% of each gene on average ( Figures 7B, C ), including mutation, amplification, mRNA high, mRNA low, and multiple alterations. For the progression of the genetic alterations of six-FBXOs, the data manifested that the altered group had a worse OS (p = 0.0331) and progression-free survival (p = 0.0428) than the unaltered group ( Figures 7D, E and Supplementary Table S3 ). There is also such a trend for disease-specific survival in PDAC (p = 0.0592), but no statistical difference for DFS of two groups (p = 0.2020) ( Figures 7F, G and Supplementary Table S3 ). Besides these, the altered group always had the higher typical PDAC gene mutations than the unaltered group, such as KRAS, TP53, CDKN2A, etc. ( Figure 7H ). Next, we analyzed the difference in sample-level enrichments between the altered group and the unaltered group. Using p <0.05 and log ratio >2 as the screening conditions, 1,194 genes were obtained in the altered group. The top 20 differently enriched genes are listed in Supplementary Table S4 , such as ADCY8, ASAP1, EFR3A, KCNQ3, and PCAT1, which had been reported to play roles in tumor onset and progression (30–32).

We screened the top 200 positively co-expressed genes using the LinkedOmics database and the cBioPortal database, respectively, and obtained the intersected genes ( Figure 8A ). These shared gene sets were used to predict the functions and pathways of six-FBXOs by performing the GO and KEGG enrichment analysis. The GO enrichment analysis contains three points: BP, CC, and MF.

For FBXO1, BP terms indicated cell division, mitotic nuclear division, DNA replication, DNA repair, cell cycle, and cell proliferation. CC terms were implicated in nucleoplasm and condensed chromosome kinetochore, midbody. MF terms were involved in protein binding, ATP binding, and protein kinase binding. The KEGG pathways showed that FBXO1 was related to cell cycle, DNA replication, and p53 signaling pathway ( Figure 8B ). For FBXO20, BP terms were implicated in cell–cell adhesion, hemidesmosome assembly, actin cytoskeleton organization, and negative regulation of extrinsic apoptotic signaling pathway. CC terms were involved in cell–cell adherens junction, extracellular exosome, plasma membrane, and focal adhesion. MF terms contained cadherin binding involved in cell–cell adhesion, protein binding, and actin binding. The KEGG pathways indicated that FBXO20 was related to tight junction, pathways in cancer, and Hippo signaling pathway ( Figure 8C ). For FBXO22, BP terms were implicated in protein folding, vesicle-mediated transport, and ubiquitin-dependent protein catabolic process. CC terms were involved in membrane, cytosol, and cytoplasm. MF terms contained poly(A) RNA binding, cadherin binding involved in cell–cell adhesion, and unfolded protein binding. The KEGG pathways indicated that FBXO22 was related to protein processing in the endoplasmic reticulum, antigen processing, and presentation ( Figure 8D ). For FBXO28, BP terms were implicated in Golgi to plasma membrane protein transport, regulation of protein export from nucleus, protein K11-linked deubiquitination, and EGFR signaling pathway. CC terms were involved in nucleoplasm, cytoplasm, and nucleus. MF terms contained protein complex binding, K63-linked polyubiquitin binding, and ubiquitin-protein transferase activity. The KEGG pathways indicated that FBXO28 was related to the regulation of actin cytoskeleton ( Figure 8E ). For FBXO32, BP terms were implicated in extracellular matrix organization, collagen catabolic process, cell differentiation, and angiogenesis. CC terms were involved in extracellular matrix, proteinaceous extracellular matrix, and extracellular space. MF terms contained extracellular matrix structural constituent and integrin binding. The KEGG pathways indicated that FBXO32 was related to ECM–receptor interaction, PI3K-Akt signaling pathway, proteoglycans in cancer, and regulation of actin cytoskeleton ( Figure 8F ). For FBXO45, BP terms were implicated in protein import into the nucleus, DNA repair, and activation of protein kinase activity. CC terms were involved in nucleoplasm, cytoplasm, and nuclear pore. MF terms contained poly(A) RNA binding, protein binding, and mRNA binding. The KEGG pathways indicated that FBXO45 was related to RNA transport, ErbB signaling pathway, insulin signaling pathway, and T cell receptor signaling pathway ( Figure 8G ).

To further validate the findings of the bioinformatics analysis, we used qPCR to confirm the transcriptional levels of six-FBXOs in five pancreatic cancer cell lines, respectively, that is, Panc-1, AsPC-1, SW1990, T3M4, and CFPAC-1, compared with human immortal pancreatic epithelial cell line HPDE6. The results revealed that, overall, the expression levels of six-FBXOs are markedly elevated in multiple PDAC cell lines ( Figure 9 ), in concordance with the above-mentioned data, and it substantiated our bioinformatics analysis to some extent, although further in vitro and in vivo experiments were needed to support it.

FBXO1 (CCNF) had been reported to be downregulated in hepatocellular carcinoma and was related to poor differentiation and adverse clinical outcome (34). In breast cancer, the overexpression of FBXO1 could suppress tumor progression, indicating that it has the role of a tumor suppressor (35). Intriguingly, FBXO1 mRNA was highly expressed in primary breast cancer tissues, but its protein level was strikingly reduced (35), hinting that FBXO1 was subjected to post-transcriptional modifications and protein degradation. Of note is the fact that FBXO1 was recently reported to be involved in the modification of the ubiquitination-proteasome system (36). FBXO1 was negatively regulated by the E3 ligase (FZR1) and the co-regulator of E3 ligase (FBXL8), which were known to play an oncogenic role in breast cancer, and FBXO1 could inhibit the expression of ribonucleotide reductase M2 (RRM2), a pro-tumorigenic protein (35). In contrast, FBXO1 was overexpressed in ovarian cancer tissues and facilitated the cell growth and invasion of ovarian cancer (37). Herein we found that FBXO1 was upregulated in PDAC tissues and the majority of PDAC cell lines as well as associated with the unfortunate prognosis of PDAC patients based on different public databases. These findings have been validated by a recently published study regarding the protein expression of FBXO1 in PDAC via IHC (38). Furthermore, our data showed that FBXO1 expression was linked to tumor differentiation, pathological grading, promoter methylation, P53 mutation, and immune infiltration in PDAC.

FBXO20 (LMO7) was downregulated in lung adenocarcinoma and relevant in tumor size, nodal involvement, and pathological stage as well as with a poor prognosis (15). The depletion of FBXO20 resulted in an increase of susceptibility for spontaneous lung cancer in the murine model (39). In contrast, FBXO20 could facilitate the migration ability of breast cancer cells in a cell-specific manner via modulating Rho-MRTF-SRF signaling (14). Liu et al. verified that the expression levels of FBXO20 mRNA and protein were increased in mouse and human pancreatic cancer tissues (16). The results of Liu were consistent with our findings in multiple public databases to some extent. FBXO20 acted an oncogenic role to promote metastasis and progression. In terms of proliferation, FBXO20 exhibited dual roles, that is, cell cycle acceleration and apoptosis inhibition (16). In addition, our data showed that a higher expression of FBXO20 tended to result in poorer pathological grading and worse tumor differentiation and predict an adverse prognosis in PDAC patients. FBXO20 was significantly elevated in different pancreatic precancerous lesions, including IPMA, IPMC, and IPMN. Among them, IPMA had the highest expression levels of FBXO20. These data indicated that FBXO20 possessed the potential to be a sensitive indicator of precancerous lesions, which is important for the early diagnosis of PDAC. Besides this, FBXO20 was tightly correlated with diabetic status, promoter methylation, P53 mutation, and immune infiltration in PDAC. The functional analysis and KEGG pathways revealed that FBXO20 might be involved in apoptosis-related signaling pathway and Hippo signaling pathway, which deserve to be further investigated.

Accumulating evidence suggested that FBXO22 exerts its oncogenic functions through mediating the ubiquitination and degradation of substrates in many human malignancies (40). However, the roles it plays in different tumors is inconsistent in physiological and pathological processes. In renal cell carcinoma, FBXO22 restrained cancer metastasis and progression by suppressing VEGF-induced angiogenesis and MMP-9-induced invasion and migration (41). On the contrary, FBXO22 facilitated tumorigenesis and progression via regulating the ubiquitination and degradation of p21 in hepatocellular carcinoma (42), of LKB1 in lung cancer, and of nuclear PTEN in colorectal cancer (43). Interestingly, FBXO22 exhibits a paradoxical dual role of pro-tumorigenic and anti-metastatic function in breast cancer progression (44). The functional role of FBXO22 in PDAC has not yet been reported. Herein we explored that FBXO22 was upregulated in PDAC and associated with cell stemness, poor pathological grading, and worse clinical outcomes. Notably, FBXO22 expression was closely correlated with the infiltration of a variety of immune cells, including B cells, CD8⁺ T cells, CD4⁺ T cells, NK cells, macrophages, neutrophils, and dendritic cells, implicating that it might be involved in the reprogramming of the tumor immune microenvironment in PDAC. Similarly, our functional analysis suggested that FBXO22 was related to the ubiquitin-dependent protein catabolic process, protein processing in the endoplasmic reticulum, antigen processing, and presentation, which allows our data and the published literatures to be mutually corroborated.

SCF (Skp1/Cul1/F-box) ubiquitin ligase serves as a key modulator of cell homeostasis via regulating downstream a variety of critical proteins for ubiquitylation (45). The IHC analysis displayed that FBXO28 and its phosphorylation are strong and predicted a poor prognosis in breast cancer (46). The CDK1/2-driven activation of the E3 ubiquitin ligase SCF^FBXO28 accelerated MYC-dependent transcription by non-proteolytic ubiquitylation and promoted transformation and tumorigenesis (46). Fagerholm et al. showed that FBXO28 had a correlation with survival and treatment outcome using interaction analysis of cis-eQTL variants in breast cancer (47). Herein we found that FBXO28 had a high level in the majority of PDAC cell lines and tissues, which often predicted an adverse clinical survival for PDAC patients. Besides this, FBXO28 expression was related to cell stemness, tumor size, and pathological grading, but no significant correlation with promoter methylation. Notably, FBXO28 expression was strongly linked to the infiltration of multiple immune cells, such as B cells, CD8⁺ T cells, macrophages, neutrophils, and dendritic cells. The functional analysis revealed that FBXO28 may be involved in protein transport, protein deubiquitination, EGFR signaling pathway, and regulation of actin cytoskeleton.

The promoter hypermethylation of FBXO32 was responsible for its downregulation in ovarian cancer cells. A profound methylation frequency of FBXO32 was detected in advanced-stage ovarian cancer, which predicted a shorter progression-free survival (48). The re-expression of FBXO32 dampened cell growth in platinum-resistant ovarian cancer as a result of being re-sensitized to cisplatin and increased apoptosis (48). A similar situation of aberrant methylation of FBXO32 could be found in esophageal squamous cell carcinoma (49). Furthermore, FBXO32 inhibited breast cancer oncogenesis and progression via interacting with KLF4 for its ubiquitination and degradation, a critical factor for cell fate decisions (50). Consistent with our findings, we also found aberrant promoter methylation of FBXO32, but the discordant result was that FBXO32 was hypomethylated and highly expressed in PanIN and PDAC, in comparison to normal tissues, as well as closely relevant in cell stemness, tumor size, lymphatic metastasis, tumor differentiation, tumor staging, and the prognosis of PDAC patients. Among pancreatic precancerous lesions, FBXO32 had the highest expression in IPMC. Furthermore, FBXO32 expression is also related to immune infiltration, e.g., CD8⁺ T cells, NK cells, macrophages, neutrophils, and dendritic cells. The functional analysis demonstrated that FBXO32 may be involved in cell differentiation, angiogenesis, regulation of actin cytoskeleton, and PI3K-Akt signaling pathway, which remains to be further explored.

FBXO45 propelled ubiquitin-mediated proteolysis of the tumor-suppressor PAR4 to control cancer cell survival (51). Specifically, E3 ubiquitin ligase FBXO45 could interact with PAR4, a PRKC apoptosis WT1 regulator, via a short consensus sequence motif in the cytoplasm to influence its ubiquitylation and proteasomal degradation, which, in turn, governs cell apoptosis (51). Furthermore, FBXO45 modulated the process of EMT via mediating the ubiquitination and degradation of substantial EMT-related transcription factors, such as Twist1, Snai1/2, and Zeb1/2, in tumor cells (52). In gastric cancer, the expression levels of FBXO45 in tumor tissues were increased compared with those in normal tissues. Unexpectedly, patients with high FBXO45 expression had longer survival than those with low expression (53). Herein we found that FBXO45 expression was markedly upregulated in PDAC tissues in both mRNA and protein levels and was inversely relevant in the OS and DFS of PDAC patients based on multiple online databases. FBXO45 expression was related to the degree of tumor differentiation, tumor staging, and P53 mutation in PDAC. The expression level of FBXO45 was higher in IPMC than in normal pancreatic tissues. Additionally, the expression levels of FBXO45 were related to immune infiltration, i.e., B cells, CD8⁺ T cells, CD4⁺ T cells, NK cells, macrophages, neutrophils, and dendritic cells. The functional analysis showed that FBXO45 was involved in ErbB signaling pathway and T cell receptor signaling pathway.

Among six-FBXOs, FBXO32 had attracted our attention since it exhibited a strong correlation with all the clinicopathological parameters investigated as detailed above. In terms of prognosis, FBXO32 expression could well predict the clinical outcomes of PDAC patients in both OS and DFS. For diagnosis, FBXO32 was tightly associated with clinicopathologic characteristics, cell stemness, and immune infiltration. Besides this, we noticed that FBXO32 was not only significantly upregulated in PDAC tissues but also highly expressed in some precancerous tissues, such as IPMC and PanIN, which provided a strong support for the early diagnosis of PDAC. Regarding treatment, we speculated that FBXO32 was likely to play a role in PDAC initiation and progression as well as reprogramming of the tumor immune microenvironment through regulating cell differentiation, angiogenesis, actin cytoskeleton, and some important signaling pathways, such as PI3K/Akt signaling pathway. By developing therapeutic agents targeting FBXO32, it is possible to control and block the development and malignant transformation of PDAC. Therefore, FBXO32 might be a promising prognostic/diagnostic/therapeutic target for PDAC.