Significant transcriptional levels of FBXOs in BC
In order to explore the prognostic and potential therapeutic values of different FBXO members in BC, the ONCOMINE databases were used to compare the mRNA expression levels of FBXOs in BC samples with normal breast samples (Figure 1). Ten FBXO genes were identified within the human BC cells. According to our findings, FBXO1, 2, 5, 6, 16, 17, 22, 28 and 45 were remarkably altered in different types of BC cells. FBXO1, 6, 16, 28, and 45 were all expressed at high levels in various pathological types of BC. FBXO2 and 17 were signiﬁcantly downregulated in different types. As for FBXO5 and 22, they showed the contrary expression pattern. The specific fold change, p-value, and the value of t-test of different significantly statistical analysis were showed in Table 1 [8, 22,32-36]. Using ONCOMINE and UALCAN databases, we compared the expression situations of FBXOs in more than 20 types of tumor and normal samples across TCGA datasets to explore the FBXOs’ regular pattern of expression (Figure S1-S2).
Next, we explored the distinction between the mRNA expression of FBXO family members and normal breast tissues in different subcategory of breast invasive carcinoma (BRCA) in GEPIA database. The overall results indicated that the expression levels of FBXO1, FBXO6, FBXO16, FBXO22 and FBXO45 in BRCA were higher than those in normal tissues, and the expression levels of FBXO17 and FBXO31 were lower in BRCA samples (Figure 1).
The correlation between mRNA expression levels of FBXOs and clinicopathological parameters of BC
We analyzed the transcriptional levels of FBXOs in different molecular subtypes of BRCA by using GEPIA database, and all data was from TCGA and GTEx datasets. Significantly increased FBXO1, FBXO6, FBXO22, and FBXO45 were observed in all BRCA subtypes compared with normal breast groups. The expression levels of FBXO17 and FBXO31 were significantly decreased in all BRCA subtypes. As for FBXO2, it was found expressed lower in HER2 and luminal B subtypes of BRCA. The mRNA of FBXO5 showed up-regulated in Basal-like, HER2 and luminal B subtypes. In luminal-types breast carcinoma, FBXO16 was inclined to over-express in luminal A and B groups and FBXO28 was a potential up-regulated biomarker of luminal B groups (Figure 2).
Based on aforesaid research, we probed into the correlation between the mRNA expression of FBXOs and clinicopathological stage of BRCA patients via UALCAN database. In all family members, there were considerable differences of transcriptional levels between normal groups and the patient groups divided by different pathological stages. Among the results, FBXO1, FBXO5, FBXO6, FBXO16, FBXO22, FBXO28 and FBXO45 were up-regulated in the pathological stage groups, FBXO2, FBXO17 and FBXO31 were negative expression factors in BRCA patients. More details of expression differences were showed in Figure 3, p < 0.05 was considered to be statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001; ****p＜0.0001).
We also used bc-GenExMiner (v4.4) online tool to assess the relationship between FBXOs expression levels and various clinical features of BC patients based on RNA-seq technology (Table 2). The clinical features include age, nodal metastasis status, ER/PR/HER2 status, basal-like statues, triple-negative statues (TNBC), P53 status, Scarff Bloom & Richardson grade status (SBR) and Nottingham Prognostic Index (NPI). The table showed clearly that both of FBXO1 and FBXO45 had significant high-expression differences in the clinical patient groups of younger age (Age＜51), lymph nodes metastasis, ER (-), PR (-), HER2 (+), basal-like subtype, triple-negative subtype, P53 gene mutation, level III of SBR and level III of NPI. The results implied that FBXO1 and FBXO45 were positively correlated with the types of highly malignant and poor-prognostic BC, which have the features of low differentiation, high invasiveness, easy to metastasize and relapse. It means that FBXO1 and FBXO45 could be potential biomarkers to identify special types of BC.
Next, we provided the immunohistochemistry (IHC) outcomes from HPA database to verify the difference of protein expression of FBXO family members from HPA database. We found that FBXO1, FBXO5, FBXO6, FBXO16, FBXO45 proteins were more highly expressed in the BC tissues than those in the normal tissues. The expression differences were not obvious of FBXO2 and FBXO31. The IHC results of FBXO17, FBXO22, FBXO28 need to be further updated in HPA database (Figure 4).
The genetic alteration and mutation information of FBXO family members
We analyzed the FBXO genes’ alterations and mutation situation in the cBioPortal online tool for breast invasive carcinoma (TCGA, Firehose Legacy). As showed in Figure 5a, target genes were altered in 723 of 1093 patient cases with the percent of 66.15%. The highest frequency of alterations was found in FBXO28 (395 of 1093 samples, 36.14%), with mRNA up-regulation of 22.6% (247 cases), genetic amplification of 5.76% (63 cases), mRNA down-regulation of 0.46% (5 cases) and other multiple alterations of 7.32% (80 cases) (Figure 5a and b). The second gene was FBXO1, and it altered in 11.89% of 1093 patient cases. The main genetic alterations involved mRNA up-regulation (70 cases, 6.4%), genetic amplification (46 cases, 4.21%), mutation (5 cases, 0.46%), deep deletion (1 case, 0.09%) and other multiple alterations (8 cases, 0.73%) (Figure 5a and b). Other gene alterations included FBXO2 (38 of 1093 samples, 3.48%), FBXO5 (93 of 1093 samples, 8.51%), FBXO6 (52 of 1093 samples, 4.76%), FBXO16 (95 of 1093 samples, 8.87%), FBXO17 (101 of 1093 samples, 9.24%), FBXO22 (103 of 1093 samples, 9.42%), FBXO31 (99 of 1093 samples, 9.06%) and FBXO45 (39 of 1093 samples, 3.57%). The specific percentage of each gene alteration is shown in Figure 5b. The largest proportion of alterations was high mRNA expression, especially in FBXO2, FBXO5, FBXO6, FBXO17 and FBXO22. Interestingly, there was no overexpression of mRNA was detected in FBXO45, but it had high frequency of genetic amplification of 3.48% (38 cases). Furthermore, we extracted the gene mutation information of the FBXOs from cBioPortal website tool. The overall somatic mutation frequency was very low. The frequency of FBXO1 and FBXO17 was 0.5%, the frequency of FBXO28 and FBXO31 was 0.3%, the rest members’ mutation frequency was no more than 0.2%. Figure 5c displayed the specific mutation site in FBXOs DNA sequences. The green dots indicate missense mutations and the black ones mean truncating sites. These results illustrated that the ten FBXOs members had excellent genetic stability as potential BC universal biomarkers.
Prognostic values of FBXOs’ mRNA expression levels in BC patients
In order to evaluate the clinical significance of FBXOs, we used publicly Kaplan-Meier Plotter tools to explore the correlation between FBXO family members’ transcriptional level and the survival of patients with overall BC and different molecular subtypes of BC patients further. The main parameters of survival analysis include relapse free survival (RFS), overall survival (OS), distant metastasis free survival (DMFS) and post progression survival (PPS). Survival curves according to Kaplan‐Meier showed in Figure 6, suggesting that high mRNA levels of FBXO1, 5, 31 and 45 were signiﬁcantly associated with worse prognosis in BC patients. By contrast, high transcription levels of FBXO2, 6, 16, 17 symbolized a better prognosis of BC patients (p＜0.05). Moreover, we found that increased expression of FBXO1 mRNA revealed a significant correlation with worse RFS, OS and DMFS in overall BC patients (Figure 6), as well as in luminal A subtype (Figure S3). The high mRNA levels of FBXO2 was signiﬁcantly associated with better RFS, OS and DMFS in overall BC patients (Figure 6). In luminal B and HER2 subtypes, FBXO2 symbolized a better prognosis similarly (Figure S3). The increased transcriptional levels of FBXO5 were related to poor RFS, OS, DMFS and PPS in overall BC and luminal A subtype patients (Figure 6). In luminal B and triple-negative subtypes, high expression of FBXO5 was related to poor RFS (Figure S3). FBXO6 is a marker for good prognosis of BC patients, high mRNA level of FBXO6 meaning better RFS, OS, DMFS in overall patient groups (Figure 6). Increased expression of FBXO6 was related to better RFS in HER2 subtypes and better RFS, OS in TNBC (Figure S3). As for FBXO16 and FBXO17, they are favorable prognosis markers in BC (Figure 6). High transcriptional levels of FBXO16 was associated with better RFS and OS in luminal A and better RFS in luminal B of BC groups (Figure S3). Increased mRNA levels of FBXO17 revealed a significant correlation with better PPS in HER2 BC patients (Figure S3). The overexpression of FBXO22 was only related to worse OS in HER2 BC (Figure S3) and overexpression of FBXO28 was only related to worse RFS in luminal B and TNBC types of BC (Figure S3). High transcription level of FBXO31 was interrelated with poor RFS and OS in overall BC and poor RFS is luminal A, luminal B and HER2 subtypes patients (Figure S3). FBXO45 is a poor prognosis marker in BC. We found that it was related to worse RFS and OS in overall BC when overexpressed. In luminal subtypes of BC, high transcriptional level of it also suggested poor RFS and OS (Figure S3). In a conclusion, FBXO2, FBXO6, FBXO16 and FBXO17 were potential favorable prognostic factors for BC. FBXO1, FBXO5, FBXO22, FBXO28, FBXO31 and FBXO45 may be the independent poor prognostic factors in BC.
Functional enrichment analysis of FBXO1 and co-expressed genes in BC
All our preliminary results throw light on the importance of FBXO1. As a novel biomarker in human BC, FBXO1 may play a crucial role in the process of tumorigenesis and development and may be the potential target of precision therapy for patients with BC. We further performed the IHC staining in clinical different molecular subtypes of BC tissues to verify the expression situation of FBXO1 protein. Our IHC results showed that significantly increased FBXO1 was observed highly expressed in all clinical subtypes of BC tissues than in the normal tissues (Figure 7a). The additional clinical information of samples used in IHC assay was showed in Table S1. Next, we analyzed the specific mutations of FBXO1 in BC by employing the COSMIC database. The largest proportion of mutations were missense substitution (15%) and synonymous substitution (15%). The largest proportion of nucleotide changes was C > T (41.67%), the rest included 8.33% of A > G, C > A, C > G, G > A, G > C, G > T and T > C (Figure 7b). Then We screened the top 150 co-expressed genes that were most related to FBXO1 from the cBioPortal and COXPRESdb online tools. The top 20 genes from both databases were displayed in Figure 7c and d. We obtained a cohort of 108 crossed genes shown by Venn diagram in Figure 7e.
GO enrichment analysis indicated that the biological processes (BP) including mitotic nuclear division, chromosome segregation, nuclear division and organelle fission were mostly signiﬁcantly regulated by the FBXO1 and co-expressed genes alterations in breast adenocarcinoma. Mostly signiﬁcant cell component (CC) included chromosomal region, spindle, centromeric region and condensed chromosome. Besides, as molecular function (MF), microtubule binding, tubulin binding and ATPase activity were mostly signiﬁcantly affected by targeted genes in Figure 7g. KEGG analysis demonstrated the pathways were mostly correlated with the functions of FBXO1 and co-expressed genes shown in bubble chart (Figure 7f). Cell cycle (hsa04110) was considered as the most relevant pathway which FBXO1 and co-expressed genes participated in Table 3. Furthermore, using DAVID database, we marked the key points regulated by FBXO1 and co-expressed genes alteration refer to Figure S4. Collectively, through experiment and database analysis, FBXO1 protein was truly increased in BC tissues. It may be an excellent therapeutic target for clinical BC patients because the stability of FBXO1 gene is of a high degree and the mutations are very rare. Moreover, GO and KEGG analysis suggested that FBXO1 and 108 co-expressed genes may play essential roles in regulating the tumorigenesis and proliferation in BC.
Knockdown of FBXO1 suppresses the proliferation and migration of breast cancer cells
In order to verify the results of above bioinformatics analysis, we further analyzed the FBXO1 protein levels in breast cancer and normal breast cell lines by Western blotting. FBXO1 was highly expressed in various breast cancer cell lines (MCF7, MDA-MB-231, MDA-MB-468, SK-BR3, T47D, HCC1954 and BT474), the expression levels were significantly higher than that in normal breast cell line (MCF-10A) (Figure 8a). To examine the effect of FBXO1 in breast cancer cell lines, MCF7 and MDA-MB-231 were successfully transfected with si‐FBXO1 to knockdown expression of FBXO1 and verified by Real-time qPCR, Western-blot analysis and FAM-fluorescence detection (Figure 8b-d). First of all, the CCK-8 assay was used to measure the proliferation of siRNA‐transfected cells. The MCF7 and MDA-MB-231 cell lines, treated with si‐FBXO1 #1 and #2, revealed the lower proliferative ability compared with the negative control groups (Figure 8e). Besides, it turned out that colony formation in MCF7 and MDA-MB-231 cells was signiﬁcantly reduced after FBXO1 depletion (Figure 8f). Subsequently, we found that FBXO1 knockdown caused an apparent suppression of cell migration in MCF7 and MDA-MB-231cell lines (Figure 8g-h). In conclusion, these results demonstrated that the knockdown of FBXO1 protein inhibited the proliferation and migration of breast cancer cells.
Screening and functional analysis of 10 hub genes in Protein-Protein Interaction (PPI) network of FBXO1
Combined using the STRING database and Cytoscape software, we constructed a PPI network of the co-expressed 108 genes of FBXO1 and obtained the core gene modules. The top 10 genes included CDC20, PLK1, CCNB1, CCNA2, CDK1, KIF2C, KIF23, BUB1, BUB1B and MAD2L1, which were identified as potential hub genes according to the degree score generated by MCODE plug-in of Cytoscape (marked in yellow) (Figure 9a). Meanwhile, according to the degree-rank score generated by CytoHubba plug-in, we got the similar top 10 hub nodes as Figure 9a (Figure 9b). Drawing support from STRING database, we further verified the strong correlation between FBXO1 and top 10 hub genes obtained from MCODE plug-in (Figure 9c). In addition, BINGO plug-in showed the most significant biological process influenced by the hub genes, including cell cycle M phase, organelle fission and nuclear division, which suggesting that they probably play crucial roles in the tumor cell mitosis process (Figure 9d). Hierarchical clustering of the 10 hub genes and FBXO1 was performed by UCSC Xena browser, indicating the consistent expression profile among these genes in overall and different subtypes of BC (Figure 9e). The strong positive correlationship of transcriptional levels among FBXO1 and 10 hub genes in BC patients were also proved by heatmap from the bc-GenExMiner platform (Figure 9f) and scatter diagram from the GEPIA dataset (Figure 10a). To find more in-depth clinical significance of targeted genes, we investigated the Kaplan-Meier RFS survival curves of 10 hub genes in BC. The results displayed that high expression of total 10 hub genes predicted unfavorable prognosis in patients with BC (Figure 10b). In conclusion, FBXO1 and CDC20, PLK1, CCNB1, CCNA2, CDK1, KIF2C, KIF23, BUB1, BUB1B, MAD2L1 may be tightly functional partners in regulating breast tumor cell cycle process and mediating poor prognosis of BC together.