Potential Therapeutic Value of Mangiferin for Lung Adenocarcinoma (Luad): A Comprehensive Study Based on in Vitro Experiments and Bioinformatics

Background: Lung adenocarcinoma (LUAD), which is the most common lung cancer type in never-smokers and primarily occurs in women, requires effective treatment methods. Mangiferin is a polyphenol widely found in mango trees and has been reported to have chemotherapeutic and preventive potential against various types of cancer. Thus, we investigated the potential therapeutic value of mangiferin for LUAD. Methods: Cell lines A549, H2030 and H1299 were processed with mangiferin to detect and screen for differentially expressed lncRNAs and mRNAs. Close associations between extracted common lncRNAs and mRNAs were identied for lncRNA-mRNA network construction. Based on the network and an online database, target lncRNAs, target genes and meaningful lncRNA-mRNA pairs were identied, and signaling pathway analysis was performed. Results: The top 200 lncRNA-mRNA pairs were used to construct the network. We identied 12 target lncRNAs and 18 target genes. Gene nodes in the network were mostly visualized on the PI3K-Akt signaling pathway (P < 0.01). Furthermore, lncRNA-mRNA pairs that contained the genes ARHGAP29, BRIX1, CD109, CDK1, CTNNAL1, DAB2IP, DDIT4L, GPR162, ICAM5, KCNAB3 and MMP7 were considered to be meaningful lncRNA-mRNA pairs affected by mangiferin in LUAD. Conclusions: Our study provided a comprehensive understanding of the potential therapeutic value of mangiferin for LUAD.


Introduction
Lung adenocarcinoma (LUAD), which belongs to the category of non-small-cell lung cancer and arises from the distal alveolar epithelium, is the most common lung cancer type. LUAD can be detected by radiography, ne needle aspiration biopsy and pathological analysis, while the expression variations of signaling molecules in LUAD also provide new candidates for diagnosis [1,2]. Recently, treatment methods for LUAD, including surgery, chemotherapy or molecular-targeted therapy, and even immunotherapy, have been employed [3][4][5][6][7]. Interestingly, traditional Chinese medicine (TCM) treatment can alleviate the symptoms caused by LUAD and ease multidrug resistance, has a synergistic and detoxifying effect on radiotherapy and chemotherapy, and has become an important component of comprehensive treatment [8,9]. Active ingredients in TCMs for LUAD treatment have attracted considerable attention in recent years [10][11][12][13].
Mangiferin, an attractive TCM, is also known as 1,3,6,7-tetrahydroxyxanthone-C2-β-d glucoside and is one kind of polyphenol primarily derived from the Anacardiaceae and Gentianaceae families and is widely found in mango trees. Mangiferin has attracted considerable attention for its promising chemotherapeutic and preventive potential against various types of cancer, such as lung, brain, breast, cervix, ovarian, prostate cancers, glioma and leukemia, by single administration or in combination with known anticancer chemicals. In terms of antineoplastic mechanisms, mangiferin has been shown to act on multiple molecular targets, which mediate the underlying biological signaling processes that inhibit the initiation, promotion, and metastasis of cancer [14][15][16][17][18][19][20]. For lung cancer, studies have found that decreased activities of some biochemical pathways in lung cancer-bearing animals were prevented after pre-and posttreatment with mangiferin and even approached normal control animal values [21][22][23][24][25]. It was also shown that the anticancer effect of mangiferin was more pronounced when used as a chemopreventive agent, rather than as a chemotherapeutic agent, against B(a)P-induced lung carcinogenesis [26]. In addition, mangiferin was able to decrease the tumor mass by helping the cisplatin with antiproliferative effects on A549 cells [25]. According to recent ndings by Grauzdytė D et al., mangiferin ameliorates oxidative stress, accelerates the wound healing process and restores the proliferation rate in polycyclic aromatic hydrocarbon-exposed bronchial epithelium [27]. Positive effects of mangiferin stimulate further research on this promising phytochemical. Therefore, for LUAD, we are also optimistic about the potential therapeutic value of mangiferin.
On the other hand, molecular biological methods for exploring the origin, development and sequelae of cancers are attracting increasing interest; among these methods, the use of long noncoding RNA (lncRNA) carries great weight. LncRNAs, which are primarily located in the nucleus, are a class of RNA molecules that measure over 200nt and contain many types of transcripts. The expression of lncRNAs is tissue-and cell-speci c. LncRNA can bind to DNA/RNA and thus is considered to perform important regulatory functions, and this RNA and is closely related to disease development. Several important lncRNAs have been identi ed in lung cancer. For example, Castillo J et al. summarized multiple approaches for discovering and providing consensus rankings of deregulated lncRNAs in carcinoma, especially focusing on lncRNAs that have improved the e cacy of LUAD early diagnosis, clinical assessment, and prognosis analysis [28]. Therefore, taking advantage of the characteristics of signaling molecules, we attempted to evaluate the expression levels of lncRNAs and messenger RNAs (mRNAs) in LUAD cell lines before and after treatment with mangiferin. Differentially expressed lncRNAs or mRNAs were identi ed for further exploration with online databases to verify whether mangiferin has potential inhibitory effects and even therapeutic value on LUAD.
Materials And Methods

Cell culturing and processing
We obtained the wild-type epidermal growth factor receptor (EGFR) and KRAS-mutated cell lines A549, H2030 and H1299 [29][30][31] from the Institute of Biochemistry and Cell Biology of the Chinese Academy of Sciences (Shanghai, China) and performed high-throughput screening before and after mangiferin treatment [32][33][34]. We stored the cell lines in the Dulbecco-modi ed Eagle's medium. All medias were added with 10% FBS, 100 mg/ml streptomycin, 100 U/ml penicillin and 0.03% L-glutamine. All cells were cultured under 37 °C in a humid incubator with 5% CO 2 . The responses of these cell lines to mangiferin were tested by detecting lncRNA and mRNA expression [25].
2. Statistical analysis and network construction.
The lncRNA and mRNA expression rates in A549, H2030 and H1299 cells before and after mangiferin treatment were compared using paired sample t-tests. To acquire the differentially expressed lncRNAs and mRNAs in mangiferin-processed LUAD cells, we intersected the lists of upregulated or downregulated lncRNAs and mRNAs from three cell lines. Next, we measured the relevance between the extracted common lncRNAs and mRNAs to obtain the closely associated lncRNA-mRNA pairs, which were used to construct the underlying lncRNA-mRNA regulatory network by Cytoscape v3.6.1 [35].
3. Exploration of potential mechanisms in the lncRNA-mRNA network Based on the network that contains the top 200 related lncRNA-mRNA pairs, we further subjected the gene nodes to protein-protein interaction (PPI) analysis via STRING v11.0 (https://string-db.org/), which was last updated on January 19, 2019 and provides online interaction evidence regarding proteins of various organisms. In addition, to explore the potential biological pathways that in uenced mangiferin in LUAD, gene nodes from the network were also used for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis by the online Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.8 (https://david.ncifcrf.gov/, last updated in 2019).

Identi cation of signi cant lncRNA-mRNA pairs
Gene Expression Pro ling Interactive Analysis 2 (GEPIA2, http://gepia2.cancer-pku.cn/) is an updated online tool to analyze the RNA sequencing expression data from totally over 18,000 tumor and normal specimens, of the TCGA and GTEx projects origin (http://xena.ucsc.edu) [36]. We fully utilized GEPIA2 to download differentially expressed genes (DEGs) of LUAD by |Log2FC| Cutoff ≥ 1 and q-value Cutoff ≤ 0.01. Furthermore, two gene groups from the lncRNA-mRNA network and GEPIA2 were intersected to obtain genes that were mangiferin-affected, as well as dysregulated in LUAD, of which the expression levels and survival analysis with hazard ratio (HR) in LUAD were also acquired from GEPIA2. Among the intersections, the dysregulated genes, which indicated poor or good prognosis, but were found to be inhibited or promoted by mangiferin toward the positive direction for LUAD treatment, were identi ed as signi cant target genes. In addition, according to the intersected group, those lncRNA-mRNA pairs that consisted of those target genes were considered to be signi cant pairs in mangiferin-processed LUAD cell lines.

Differentially expressed lncRNAs and mRNAs
We collected differentially expressed lncRNAs and mRNAs in the A549, H2030 and H1299 cell lines. Next, those targets were intersected by Venn gram (Fig. 1), and we acquired 44 upregulated and 52 downregulated intersecting lncRNAs in three cell lines, while 229 upregulated mRNAs and 423 downregulated mRNAs were also obtained, among which the top 10 differentially expressed lncRNAs and mRNAs are shown in Table 1. Furthermore, correlations between common differentially expressed targets, which combined the upregulated and downregulated groups by lncRNAs or mRNAs, were calculated by R language. Consequently, the top 200 closely related lncRNA-mRNA pairs were extracted to built the lncRNA-mRNA network (Fig. 2). With degrees not less than 10 inclusive, we also identi ed NONHSAT024774.2 and NONHSAT077537.2 from the network. Thus, in addition to the top 10 differentially expressed lncRNAs, we obtained 12 target lncRNAs (Table 1). Intuitively displayed by the lncRNA-mRNA correlation network, all 115 mRNA nodes were further involved in PPI network construction. The PPI network showed that there was protein interactions among 58 items (Fig. 3), in which the genes CDK1, CXCL8, CCL2, HMGB1 and DDX58 were considered to be preliminary mangiferin-affected hub genes in this network with degrees not less than 5 inclusive. By adding the top 10 differentially expressed mRNAs and the genes DLX2, EFNB3 and GEMIN2 directly screened from the lncRNA-mRNA network, we nally obtained 18 target genes in total (Table 1).

GO and KEGG analysis of gene nodes in the lncRNA-mRNA network
The biological meanings behind 115 mRNA nodes in the lncRNA-mRNA network were explained by GO and KEGG analysis via DAVID v6.8. According to Table 2, we identi ed enriched GO terms, particularly extracellular matrix organization, cell adhesion and cellular response to interleukin-1 terms of biological process (BP), cell surface of cellular component (CC) (P < 0.01) and NAD binding and single-stranded RNA binding of molecular function (MF) (P < 0.05). In addition, genes were visualized on KEGG pathway maps, indicating that the most centralized PI3K-Akt signaling pathway (P < 0.01) in the network was in uenced by mangiferin and that the involved lncRNAs may participate in LUAD cell lines.

Differentially expressed gene analysis of LUAD on GEPIA2
On the strength of GEPIA2, which combines the advantages of large datasets, a comprehensive exploration of recognized dysregulated genes in LUAD was performed to characterize the cancer. Based on GEPIA2, we generated 1111 higher expressed genes and 3134 lower expressed genes in LUAD compared to paired normal samples, with lter criteria of |Log2FC| Cutoff ≥ 1 and q-value Cutoff ≤ 0.01 being applied (Fig. 4).

Signi cant lncRNA-mRNA pairs affected by mangiferin in LUAD
We organized DEGs of LUAD from GEPIA2 and gene nodes from the lncRNA-mRNA network as gene groups A and B, respectively. The intersecting genes of group A and group B included 23 genes (Fig. 5) whose expression levels were downloaded (Table 3). We collected changes in the expression levels of these proteins in our experimental cell lines treated with mangiferin. In addition, overall survival analysis of these targets was performed by employing Kaplan-Meier curves, which showed that aberrant expression of 12 genes, speci cally ADAMTSL4, ARHGAP29, BRIX1, CD109, CDK1, COL6A2, CTNNAL1, DDIT4L, EFHD1, MMP7, ITPKA and FGF11, in LUAD may indicate an optimistic survival rate (HR > 1) (Fig. 6). Therefore, according to Table 3, we hypothesized that the corresponding lncRNA-mRNA pairs,  which contain the genes ARHGAP29, BRIX1, CD109, CDK1, CTNNAL1, DAB2IP, DDIT4L, GPR162, ICAM5, KCNAB3 and MMP7, are signi cant lncRNA-mRNA pairs affected by mangiferin in LUAD (Table 4).

Discussion
In this research, we analyzed the potential therapeutic value of mangiferin for LUAD. Cell lines A549, H2030 and H1299 were processed with mangiferin, and bioinformatic analysis was employed, which helped us to construct the top 200 lncRNA-mRNA pairs network. We also obtained 12 lncRNAs and 18 mRNAs from differential expression calculations, degrees of connection in the lncRNA-mRNA network and PPI network, which were considered to be underlying targets of mangiferin. Additionally, mangiferin was mostly thought to affect the PI3K-Akt signaling pathway in LUAD. Furthermore, lncRNA-mRNA pairs that contained the genes ARHGAP29, BRIX1, CD109, CDK1, CTNNAL1, DAB2IP, DDIT4L, GPR162, ICAM5, KCNAB3 and MMP7 were determined to be meaningful lncRNA-mRNA pairs affected by mangiferin in LUAD. Generally, our study helped to elucidate the potential therapeutic value of mangiferin for LUAD.
In recent years, traditional Chinese medicine (TCM) or products have exhibited the effects of tumor prevention and treatment. Studies employing molecular biology and biochemical pharmacology to investigate the underlying mechanism governing these effects have opened new approaches for clinical antineoplastic drug development. In our study, mangiferin is a widely sourced treatment option that has shown therapeutic value in such cancer types as liver cancer, ovarian cancer, gastric cancer, and lung cancer [15,17,37,38], while potential targets, such as NFκB, PPARү, MMP-7, MMP-9 and EMT [14], were predicted. Similarly, we adopted the experimental method of a self-control study treated with mangiferin to explore the medicinal effect of mangiferin in LUAD cell lines.
Regarding how to evaluate the potential value of mangiferin for treating cancer, the trend of antitumor chemotherapy prefers targeted medicine development, which provides a useful reference for further research on this subject. A large amount of research has been devoted to the molecular mechanism of tumorigenesis and development, which encourage and contribute to the development of molecularly targeted treatment. Given several examples, Wang et al. have considered W934, a novel PI3K/Akt pathway inhibitor, to be a potential therapeutic drug candidate to treat the non-small-cell lung cancer (NSCLC) [39]. Research from Yang et al. showed that antipsychotic chlorpromazine has the potential to be a repurposed drug for breast cancer treatment [40]. Among the potentially affected molecular targets, the lncRNA-mRNA interaction network has been heavily researched, as systematic analysis indicated that an lncRNA-mRNA co-expression relevant to platinum resistance in advanced serous ovarian cancer [41].
Research of miRNA-lncRNA-mRNA interconnections also support signi cant academic, practical and clinical basis to gastric cancer [42]. Moreover, Xiao et al. found that LINC0092 and chromosome 2 open reading frame 71 were correlated with better prognosis of breast cancer by analyzing the resultant ER subtype-related miRNA-lncRNA-mRNA network in breast and gastric cancer [43]. Drawing on previous research, our research attempted to elucidate the mechanism of mangiferin in LUAD by analyzing the differentially expressed lncRNAs and mRNAs and constructing the network.
Thus, we considered three committed steps to achieve the goal. First, the lncRNAs and mRNAs that are differentially expressed after mangiferin treatment should be screened out. Second, a close association between identi ed common lncRNAs and mRNAs should be determined for lncRNA-mRNA network construction. Third, identi cation of target lncRNAs, target genes and meaningful lncRNA-mRNA pairs should be performed. Additionally, we investigated the potential mechanism by which the nodes in lncRNA-mRNA networks may be involved in the effects of mangiferin on LUAD.
Target lncRNAs and genes should be characterized as the probable targets affected by mangiferin in the network. Therefore, differentially expressed lncRNAs that have abundant connections with the mRNA group were considered. Similarly, to identify target genes, we added the protein interaction condition, as regulatory relationships among proteins have also been veri ed by large databases [44]. However, when targeting these nodes, it is not clear whether mangiferin has an anticancer effect. Furthermore, for signi cant lncRNA-mRNA pair identi cation, which was predicted to be bene cial for LUAD prognosis after mangiferin treatment, DEGs of LUAD downloaded in GEPIA2 and displayed in Venn diagrams help establish links between laboratory results and databases. To elaborate, mangiferin should promote the dysregulated genes that suggest positive prognosis and inhibit those suggesting poor prognosis by upregulating or downregulating the expression levels. Consequently, the eligible genes and corresponding lncRNAs became meaningful targets for mangiferin to produce anticancer effects in LUAD.
According to the results, 12 lncRNAs were presented as targets. Based on NONCODE v5 (http://www.noncode.org) [45], exosome expression pro le analysis indicated that NONHSAT094064. In addition, we extracted 18 target genes of mangiferin in LUAD, among which NGFR, RP11-203J24.9, RP11-1072A3.3, FGF11, DLX2 and EFNB3 were upregulated, while ANXA3, TSPAN8, RP11-473I1.9, SEMA3A, FRK, TOMM6, GEMIN2, CDK1, CXCL8, CCL2, HMGB1 and DDX58 were downregulated, by mangiferin treatment. NGFR was suggested to be involved in the switching of KRAS + LUAD to squamous cell carcinoma when highly expressed [46]. Studies have shown that suppression of EFNB3 decreases NSCLC progression, but no evidence has been found in LUAD [47,48]. ANXA3 knockdown was found to inhibit the growth, migration, invasion, and metastasis of LUAD via in vitro and in vivo experiments [49,50]. Moreover, research has indicated TSPAN8 to be a diagnostic biomarker in lung cancer [51]. Zhou et al. found that A549 cells secrete SEMA3A to inhibit the maturation and functions of dendritic cells, which might be associated with the unidenti ed mechanism of immune evasion by tumor cells [52]. Interestingly, FRK was considered to be a underlying treatment target for drug discovery, as it has a carcinogenic effect in lung cancer cells via inducing metabolic reprogramming and nally promoting epithelial-mesenchymal transition and metastasis [53]. TOMM6 is an upregulated mRNA-related lncRNA UCA1 that might affect cisplatin resistance in LUAD [54]. CDK1 was found to be an adverse prognostic and diagnostic biomarker for LUAD [55,56]. Liu et al. found that CXCL8 played an adverse role which accelerating cancer progression and bad outcome of LUAD, while human Dachshund homologue 1 antagonized CXCL8 to enhance the survival of LUAD patients [57]. Moreover, high levels of CCL2 predict unfavorable survival in lung adenocarcinoma [58] , and HMGB1-regulated autophagy was proven to be a signi cant contributor to docetaxel resistance in LUAD cells [59]. In summary, most of these genes play roles in LUAD, at least in lung cancer. However, few or no publications have reported the associations of these genes with mangiferin. Only HMGB1, whose protein expression rate was decreased by mangiferin, effectively prevented alcohol hepatitis [60]. Therefore, further validation is needed.
In the signi cant lncRNA-mRNA pairs that contained ARHGAP29, BRIX1, CD109, CDK1, CTNNAL1, DAB2IP, DDIT4L, GPR162, ICAM5, KCNAB3 and MMP7, Shukla et al. presented the rst RNA-seq prognostic signature for LUAD, including CD109 [61]. The genovariation of DAB2IP/AIP1 is related to increasing the risk of lung cancer in Chinese males [62]. Similarly, another study revealed that an elevated circulating tumor cell count and overexpression of MMP7 correlate with metastasis and clinical progression in LUAD patients [63]. However, most of the signi cant genes predicted in our study, which are related to neither LUAD nor mangiferin, still lack literature validation, and the regulatory mechanism governing the lncRNA-mRNA pairs warrants further exploration.
Potential pathway analysis under the lncRNA-mRNA network was also performed. KEGG analysis demonstrated that mangiferin was mostly centralized in the PI3K-Akt signaling pathway (P < 0.01) when LUAD cells were treated. The PI3K-Akt signaling pathway, involving the key proteins phosphatidylinositol 3-kinase (PI3K) and Akt/Protein Kinase B, is an intracellular signal transduction road that promotes metabolism, hyperplasia, cell survival, growth and angiogenesis through responsing to extracellular signals. Dysregulation of the PI3K/Akt pathway is implicated in a number of human cancers [64][65][66][67].
Studies of the PI3K-Akt signaling pathway may also help to develop targeted medicine for LUAD. For example, Zhang et al. suggested 20(S)-protopanaxadiol (PPD) as a promising chemopreventive agent that downregulated the PI3K/Akt signaling pathway in A549 cells [68]. It was also indicated that allicin may inhibit invasion of LUAD by reducing the activity of the PI3K/AKT signaling pathway [69], and baicalein may increase cisplatin sensitivity of A549 cells via the PI3K/Akt/NF-κB pathway [70]. On the other hand, it was reported that mangiferin relieves lipopolysaccharide-induced injury by activating the PI3K/AKT pathway [71]. mangiferin also inhibits the MMP-9 gene in phorbol myristate acetate-stimulated human astroglioma, in which the PI3K/AKT pathway is involved [72]. Therefore, we demonstrated that mangiferin may in uence the prognosis of LUAD patients by regulating the PI3K/AKT signaling pathway.
Conclusions this study employed expression pro le analysis in cell lines and composited data from a large database and demonstrated that mangiferin could be a promising drug for LUAD therapy by affecting lncRNA-mRNA pairs and the PI3K/AKT signaling pathway. However, several aspects warrant further research: whether mangiferin could play a therapeutic role requires further analysis of the parameters from clinical specimens, as aberrant expression of partial genes cannot represent a whole situation. In addition, the scarcity of interactions among genes from hub genes in the network and signi cant lncRNA-mRNA pairs increased the di culty of mangiferin-targeted node veri cation, for which more samples are needed for validation. Finally, the association between lncRNAs and mRNAs warrants further experimental con rmation. Despite these limitations, the lncRNA-mRNA interaction network, as well as the predicted hub nodes and signi cant targeted lncRNA-mRNA pairs, also provided a comprehensive understanding of the potential therapeutic value of mangiferin for LUAD. In future research, we plan to further verify our results by employing larger sample sizes and animal model experiments and by employing more accurate bioinformatic methods, such as molecular docking, to evaluate the overall clinical value of mangiferin for LUAD.  Figure 1 Venn gram for extracting common lncRNAs and mRNAs among three cell lines.

Figure 5
Venn gram of DEGs from GEPIA2 and gene nodes from lncRNA-mRNA network.

Figure 6
Kaplan-Meier curves of intersected genes.