Flavin Adenine Dinucleotide Synthetase 1 Is an Up-regulated Prognostic Marker Correlated With Immune Infiltrates in Breast Cancer


 Background: Flavin adenine dinucleotide synthetase 1 (FLAD1) is a prognostic biomarker in several cancers. This study aimed to identify its effect on survival and potential mechanism in breast cancer (BRCA).Methods: We used the OncoMine, PrognoScan, and UALCAN databases and Kaplan-Meier Plotter to explore FLAD1 expression and survival prognosis in pan-cancer and BRCA. The LinkedOmics database, Gene Ontology-Biological Process (GO-BP), and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were used for co-expression analyses. The Tumor Immune Estimation Resource, Sangerbox online tools, and the Gene Expression Profiling Interactive Analysis database analyzed the relationship between FLAD1 and cancer immune infiltrates. Results: FLAD1 expression was higher in tumor tissue than adjacent normal tissue in many cancers. Increased FLAD1 expression was correlated with poor overall, relapse-free, and disease-free survival in BRCA. The KEGG pathway and GO-BP analyses showed that FLAD1 expression was correlated to the tRNA metabolic process and mitochondrial gene expression. The DNA mismatch repair (MMR) and methylation status indicated a potential BRCA metastasis mechanism. Immune infiltration was associated with MMR, improved BRCA prognosis, and was negatively associated with FLAD1 expression. High FLAD1 expression led to the down-regulation of macrophages and monocyte infiltration, which might relate to cancer metastasis.Conclusions: FLAD1 expression is higher in BRCA than in normal tissue. High FLAD1 expression was associated with worse survival and might be associated with MMR pathway and immune infiltrations. These results suggest that FLAD1 could serve as a novel biomarker for prognosis and metastasis in BRCA.


Background
Breast cancer (BRCA) is one of the most common cancers and the primary cause of cancer-related mortality in females worldwide. Almost 268 000 women are newly diagnosed with BRCA every year in the United States, and approximately 40 000 die 1 . Although we can predict the prognosis and guide clinical treatment of patients based on traditional clinicopathological factors and some other indicators identi ed in clinical practice 2 , more precise biomarkers are needed to help improve the diagnostic accuracy and prognostic prediction.
Flavin adenine dinucleotide synthetase 1 (FLAD1) is a vital protein-coding gene. It encodes the avin adenine dinucleotide synthetase (FADS), a key enzyme that catalyzes adenylation of avin mononucleotide (FMN) to form avin adenine dinucleotide (FAD) coenzyme 3 . The FLAD1 gene is located on chromosome 1q21. 3 and is widely expressed in human tissues, including the thyroid, lymph nodes, and breasts 4 . FAD, found in the cytoplasm and mitochondria, is an accessory factor responsible for transport in the mitochondrial dehydrogenase reaction. It is closely related to DNA phosphorylation and protein metabolism and is involved in nuclear redox activity that directly reprograms the chromatin structure and Page 3/17 leads to dependent transcriptional activation or repression 5 . FAD imbalance might be an important cause for cell degeneration or tumorigenesis. Hence, FADS is important for tumor cell progression as it functions in the oxidation-reduction chain. In previous studies on other cancers, including prostate cancer 6 , nonsmall cell lung cancer 7 , and gastric cancer 8 , FLAD1 expression was signi cantly associated with poor survival and prognosis. However, no study has explored the relationship between FLAD1 expression and BRCA.
This study evaluated the relationship between FLAD1 and BRCA patients' survival and searched for potential mechanisms, using public databases, including OncoMine, PrognoScan, Kaplan-Meier Plotter, LinkedOmics, Sangerbox online tools, Tumor Immune Estimation Resource (TIMER), and Gene Expression Pro ling Interactive Analysis (GEPIA). Our study results shed light on the predictive value of FLAD1 in BRCA, suggesting a potential correlation and mechanism associating FLAD1 with BRCA prognosis and immune in ltration.

Methods
The Expression of FLAD1 in Pan-Cancer in the OncoMine Database FLAD1 expression level in pan-cancer was analyzed in the OncoMine database with a P-value of 0.01 and a fold change of 1.0 41 .

Survival Analysis in PrognoScan, Kaplan-Meier Plotter, and UALCAN Databases
The relationship between FLAD1 expression and survival was analyzed in PrognoScan 42 and Kaplan-Meier Plotter 43 . FLAD1 expression was searched in the microarray datasets of PrognoScan to determine its relationship with prognosis, including overall survival (OS), disease-free survival (DFS), and relapse-free survival (RFS). We set the signi cance threshold at a Cox P-value < 0.05. Kaplan-Meier Plotter is a powerful online to assess the effect of 54 000 genes on survival in 21 cancer types. We used it to analyze the relationship between FLAD1 expression and OS and RFS. Hazard ratios (HRs), 95% con dence intervals 44 , and log-rank P-values were calculated. Different clinical characteristics and prognosis and their relationship with FLAD1 in BRCA were analyzed by the UALCAN database 45 .

Co-expression Analysis of FLAD1 in LinkedOmics Database
We analyzed the co-expression of FLAD1 by the LinkedOmics database, which contains a large dataset from The Cancer Genome Atlas project 46 , with more than a billion data points 47 . We used the Spearman correlation coe cient and present the results in volcano plots and heat maps. We used the Function Module of the gene set enrichment analysis (GSEA) for kinase enrichment, miRNA, and transcription factor-target analyses in the Gene Ontology-Biological Process (GO-BP) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. We used a false discovery rate (FDR) < 0.05 as the rank criterion and performed 500 simulations.

Correlations Between FLAD1 Expression and DNA Mismatch Repair and Methylation Status
The Sangerbox online tools, updated synchronously with dozens of databases worldwide every day, were used to analyze the relationship between FLAD1 expression and DNA mismatch repair (MMR) and DNA methyltransferases (DNMTS) status.
Correlation Between FLAD1 Expression and Immune Cells in the TIMER and GEPIA Databases TIMER and the Sangerbox online tools were used to determine the relationship between FLAD1 expression and immune in ltration 48 . TIMER is a comprehensive resource for systematic analysis of immune in ltration in pan-cancer. It contains 10 897 samples across 33 cancer types from TCGA. We analyzed the relationship between FLAD1 expression and the abundance of all six immune in ltrating cell types, including B cells, CD4 + T cells, CD8 + T cells, neutrophils, macrophages, and dendritic cells (DCs). Tumor purity was also determined 49 .
GEPIA is an online database with 9 736 tumors and 8 587 normal tissue samples from TCGA and the Genotype-Tissue Expression (GTEx) projects 52 . We used GEPIA to study the correlation between FLAD1 expression and immune in ltration gene markers and o con rm the TIMER results.

Statistical Analysis
The results generated in OncoMine are presented with fold changes, gene ranks, and their P-values as determined by t-tests. The log-rank test was used to calculate the HR and log-rank P-value in the Kaplan-Meier Plotter to compare survival curves. PrognoScan was used to calculate the HR and Cox P-value with a univariate Cox regression model. The gene expression correlation was evaluated by Spearman's rank correlation coe cient. Differences with a P < 0.05 were considered statistically signi cant.

Results
FLAD1 mRNA Expression Level in Pan-Cancer FLAD1 expression is higher in bladder, brain, breast, cervical, colorectal, gastric, kidney, liver, lung, and ovarian cancers and leukemia, lymphoma, and myeloma (Fig. 1A). We found no cancer in which FLAD1 expression in tumor tissue was lower than in the normal tissue. Detailed information on FLAD1 mRNA expression in pan-cancer is summarized in Table S1. FLAD1 mRNA expression level was in the top 1% in BRCA.
The TIMER database was used to analyze single gene expression in pan-cancer in relation to FLAD1. The differential expression of FLAD1 between cancer and normal tissues is shown in Fig. 1B. BRCA had higher FLAD1 expression than the other cancer types.

FLAD1 Prognostic Value in Pan-Cancer
We used PrognoScan and Kaplan-Meier Plotter for a double validation exploring the relationship between FLAD1 expression and prognosis in pan-cancer. High FLAD1 expression in uenced survival in seven cancers (lung, breast, skin, colorectal, head and neck, bladder, and blood) in the PrognoScan database. Increased FLAD1 expression had a protective role in colorectal and head and neck cancer, while it was negatively associated with survival prognosis in the other cancer types (breast, lung, skin, brain, and bladder). However, the Kaplan-Meier Plotter showed that prognosis was not associated with FLAD1 expression in bladder, colorectal, lung, and head and neck cancer. Data on skin and blood cancers was absent from this database. Detailed information on these cancer cohorts and FLAD1 expression in PrognoScan and Kaplan-Meier Plotter is presented in Table S2 and Figures S2 and S3. Using FLAD1 median expression level as a standard, analysis of the BRCA cohorts in PrognoScan and Kaplan-Meier Plotter showed that increased FLAD1 expression was associated with poor OS, RFS, and disease metastasis-free survival (DMFS), which presented in Fig. 2.

Increased FLAD1 expression is Correlated with poor BRCA Prognosis
After nding that high FLAD1 expression was associated with poor BRCA prognosis, we performed a subgroup analysis of FLAD1 expression and survival using the UALCAN database. The data were strati ed by clinical characteristics, including sample type, sex, age, race, major molecular classi cation, nodal metastasis status, and clinical stage ( Fig. 2A-G). Subgroup analysis showed that sex did not in uence FLAD1 expression in tumor tissue (Fig. 2C). However, FLAD1 expression was higher in the 21-40-year-old group than in the other age groups analyzed (Fig. 2B). The FLAD1 expression in African-Americans was higher than in Caucasians (Fig. 2D). The luminal BRCA subtype had lower FLAD1 expression than other subtypes (Fig. 2E), and BRCA patients with lymph node metastases had signi cantly higher FLAD1 expression than normal tissues, suggesting that FLAD1 might affect BRCA patients by increasing lymph node metastasis (Fig. 2F).
We also performed a survival analysis based on different pathological characteristics in the Kaplan-Meier Plotter (Table S3). The results showed that triple-negative BRCA (P = 0.0348) had poorer RFS, and luminal A-type BRCA had poor RFS and OS (P = 0.0015 and P = 0.0044, respectively) when FLAD1 expression was high. In BRCA patients negative for lymph node metastasis, high FLAD1 expression was also associated with a poor OS, indicating that FLAD1 expression level detection could improve the prognostic accuracy in lymph node-negative patients.

FLAD1 Co-expression Networks in BRCA
We analyzed the FLAD1 co-expression model in the BRCA cohort in the LinkedOmics database to explore FLAD1 biological signi cance. FLAD1 was signi cantly correlated with 9 733 genes (black dots in Fig. 4A), of which 4,883 were negatively correlated 9 and 4 850 were positively correlated (red). Heat maps of the top 50 genes positively and negatively correlated with FLAD1 are presented in Figs. 4B and 4C, respectively. Information on the co-expressed genes is detailed in Table S4.
The KEGG pathway analysis results showed that the co-expressed genes were enriched in the proteasome, RNA degradation, aminoacyl-tRNA biosynthesis, RNA transport, and more (Fig. 4D). GO-BP analysis showed that enrichment was mainly in the tRNA metabolic process, mitochondrial gene expression, telomere organization, deoxyribonucleotide metabolic process, and others (Fig. 4E). These results indicated that FLAD1 might have an extensive effect on the genome replication, transcription, and repair systems.

FLAD1 Expression is Associated with MMR in BRCA
We analyzed DNA methylation and repair system status to explore the relationship between FLAD1 expression and DNA genome activity. DNMTs are essential for genome integrity in humans. Promoter hypermethylation inhibits gene expression, while methylation of the gene body contributes to its expression 10,11 . Meanwhile, gene body DNA methylation in cancer tissues occurs more frequently than in the promoters 12,13 . The levels of DNMT1 (R = 0.21, P < 0.001), DNMT2 (R = 0.08, P = 0.013), DNMT3A (R = 0.32, P < 0.001), and DNMT3B (R = 0.33, P < 0.001) were all positively correlated with FLAD1 expression in BRCA (Fig. 5B).
Expression of the four MMR genes, MLH1, MSH2, MSH6, and PMS2, and epithelial cell adhesion molecule (EpCAM), a transmembrane glycoprotein, is crucial for cell migration and metastasis. The relationship between FLAD1 expression and the mutation status of the MMR genes was analyzed by the Sangerbox online tools. The results showed that higher mutation rates of MLH1 (P < 0.05), MSH2 (P < 0.0001), MSH6 (P < 0.0001), and EpCAM (P < 0.0001) were signi cantly correlated with FLAD1 expression in BRCA (Fig. 5A).

FLAD1 Expression is Associated with Immune In ltration in BRCA
The expression status of MMR and DNMTs was reported to be correlated with tumor-in ltrating lymphocytes 3,14,15 . TIL is an independent predictor that can guide survival and improve treatment in various cancers 16,17 . Our results showed that FLAD1 expression was negatively correlated with immune cell in ltration in many cancer types (Fig. 6A). Furthermore, high FLAD1 expression was positively associated with BRCA tumor in ltration purity (R = 0.131, P < 0.001), and negatively associated with in ltration by CD8 + T cells (R = -0.264, P < 0.001), neutrophils (R = -0.128, P < 0.001), macrophages (R = -0.255, P < 0.001), and DCs (R = -0.136, P < 0.001; Fig. 6B). No association was found between FLAD1 expression and the in ltrating levels of B cells and CD4 + T cells. These results suggest that FLAD1 might affect survival by in uencing the immune in ltration into BRCA.

Immune Markers Correlation with FLAD1 Expression in BRCA
The results, adjusted by tumor purity in the TIMER database, indicated a strong correlation between FLAD1 expression and immune markers present on macrophages, monocytes, and NK cells in BRCA ( Table 2). The NOS2 and TLR4 of M1 macrophages, MRC1 and PPARG of M2 macrophages, PARK2 and CSF1R of monocytes, CD7 and KIR2DL4 of NK cells, and CD1C, HLA-DPA1, and ICAM1 of DCs were speci cally correlated with FLAD1 expression in BRCA (P < 0.001; Figure S3). Therefore, we used the GEPIA database to analyze these markers and their correlation with FLAD1 expression in BRCA tumor tissues. The results indicated that FLAD1 expression was mainly associated with macrophage in ltration into tumor tissue (Table 2). Furthermore, PPARG, a protective factor, was negatively correlated with FLAD1 expression. Treg gene marker FOXP3 and Exhausted T cell marker LAG3 were signi cantly and positively correlated with FLAD1 expression.

Discussion
In this study, we analyzed the differential expression of FLAD1 between tumor and normal tissues in pancancer. The results showed that FLAD1 was up-regulated in bladder, breast, cervical, colorectal, liver, and lung cancers (Fig. 1). Due to differences in data collection and analysis methods, FLAD1 expression differed in various cancer types between the databases. However, consistent FLAD1 expression trends were found in these databases for bladder, breast, colorectal, liver, and lung. Increased FLAD1 expression was associated with poor survival in most cancer types (e.g., bladder, breast, and lung), while it correlated with superior survival in colorectal and head and neck. Increased FLAD1 expression is correlated with worse BRCA survival. The subgroup survival analysis showed that FLAD1 expression was higher in young (21-40-year-old) and African-American patients, indicating poor prognosis (Fig. 3A-G). Therefore, FLAD1 could be an independent negative prognostic factor associated with poor survival in BRCA ( Fig. 2A-F).
The co-expression analysis revealed a potential mechanism of how FLAD1 affected the prognosis in BRCA (Fig. 4). We found that FLAD1 correlated with the SRF transcription factor family network and miRNAs, including miR422A and miR31 ( Table 1). The SRF family can attenuate BRCA cell apoptosis and support the survival of tumor epithelial cells 18,19,20 . Lower miR442A and higher miR31 promote tumor expansion and tumorigenesis and act as unfavorable factors affecting patients' survival 15,21 . Furthermore, the KEGG pathway analysis indicated that the FLAD1 mechanism in BRCA was associated with nucleic acid and protein synthesis.
MMR is an intracellular mismatch repair mechanism. Loss of function of key genes, including MLH1, MSH2, MSH6, PSM2, and EpCAM, could lead to DNA replication errors that cannot be repaired, producing more somatic mutations. Genomic instability, including DNA repair defects 22 , is one of the most common characteristics of tumor cells. MMR is involved in the process of tumor cell apoptosis due to DNA damage. The expression level of MLH1 and MSH2 was negatively correlated with FLAD1 due to increased mutation status, and could ultimately promote tumor cell metastasis and progression into advanced BRCA stages 23 . Higher expression of miR422A could affect MMR by changing the expression of MLH1, thus inhibiting the regulatory feedback mechanism between MMR and miRNA expression. Reduced MLH1 expression might cause tumor progression by decreasing apoptosis 24,25,26 . MSH6 and EpCAM are associated with poor prognosis and metastasis in BRCA 27,28 . The MMR results revealed that high FLAD1 expression might promote metastasis and poor prognosis by increasing the mutation rate of proteasomes and key MMR genes and participating in the transcription regulation mechanism.
DNA methylation is associated with carcinogenesis 29 . Ding et al. reported that MSH2-MSH6 is involved in recruiting DNMT1 to oxidative damage sites where it acts to decrease transcription and interfere with the repair process 30 . Down-regulated DNMT1 promotes BRCA tumorigenesis and is correlated with poor prognosis 31,32 . Inhibition of DNMT1 methylation can enhance estrogen receptor and BRCA type 1 (BRCA1) activity, improve chemotherapy effect, reverse cellular transformation, and block tumor growth without increasing invasiveness 33,34 .
Previous studies have found that TIL affects the prognosis 17 and that MMR is negatively correlated with TIL 14 . Our TIL analysis results showed that FLAD1 expression is correlated with immune in ltration in many cancers, including BRCA. Furthermore, FLAD1 expression was signi cantly correlated with the macrophage and neutrophil in ltration level in BRCA (Fig. 6). The relationships between FLAD1 expression and some immune markers differed from the overall trend, suggesting a speci c interaction might exist (Table 2). Increased FLAD1 expression was positively associated with tumor purity in the immune in ltration analysis. FLAD1 expression correlated with macrophages after adjusting for tumor purity, indicating that FLAD1 plays a vital role in the recruitment and regulation of immune in ltration in BRCA and is associated with carcinogenesis and metastasis 35 . The macrophage cell marker PPARG was found to have a protective value in cancer and was negatively correlated with FLAD1 expression. Therefore, its down-regulation contributes to cancer formation and provides an appropriate microenvironment for malignant tumor cell proliferation and metastasis, resulting in poor prognosis 36 . PARK2 negatively regulates the EGFR/AKT/mTOR signaling pathway in monocyte in cancer. It is also negatively correlated with FLAD1 expression, leading to lymph node metastasis 37 . FOXP3 increased in Treg along with FLAD1 expression, negatively in uencing the OS and DFS. It also promoted proliferation and metastasis by generating adenosine to down-regulate the immune function 38,39 . LAG3 expressed on Exhausted T cell can downregulate the function of T cells and inhibition of LAG3 can improve the response of anti-tumor T cell 40 . Therefore, FLAD1 expression could be considered an unfavorable factor for BRCA patients' survival by promoting tumor cell metastasis. However, additional downstream mechanism validation tests are still needed.
Our study had several limitations, even though we integrated information from multiple databases into our analysis. First, we only used online databases for bioinformatic analysis of the relationship between FLAD1 expression and survival, lacking In vivo and in vitro experiments. Second, although we have found that increased FLAD1 expression was associated with poor survival and immune in ltration in BRCA patients, we have no proof that FLAD1 affects the patients' survival by immune in ltration. However, we currently conduct experiments on the effect of FLAD1 overexpression on BRCA prognosis and explore its downstream mechanisms.

Conclusion
According to our analysis results, FLAD1 is overexpressed in BRCA and is associated with macrophage in ltration. Increased FLAD1 expression was negatively correlated with prognosis, possibly by promoting tumor cell progression and metastasis. Therefore, we think that FLAD1 could be a biomarker for BRCA diagnosis and precise clinical treatment. All authors read and approved the nal manuscript.  *P < 0.01; **P < 0.001; *P < 0.0001.