Astragalus Polysaccharide Regulates EGFR and ANXA1 Expression to Inhibit Tumor Growth

Background: Astragalus polysaccharide (APS) has been frequently used as an adjuvant agent responsible for its immunoregulatory activity to enhance ecacy and reduce toxicity of chemotherapy used in the management of breast cancer. However, the other synergism mechanism of APS remains unclear. Methods: The expression proles of breast cancer biological data (E-GEOD-10780), (E-GEOD-65194), (E-GEOD-10810), (E-GEOD-42568), (E-GEOD-61304), (E-GEOD-7904) and (E-GEOD-15852) were retrieved from European bioinformatics institute (EBI) database to identify differentially expressed genes (DEGs). Gene Ontology and pathway enrichment analyses were carried out. The protein–protein interaction (PPI) networks of the DEGs were constructed by STRING online database. Finally, STRING was applied to construct the PPI networks of the DEGs associated with cell proliferation, cell cycle and cell apoptosis from GO enrichment analysis. Subsequently, hub and common genes that might be the potential targets and possible mechanism behind APS in vivo direct anti-tumor activity on breast cancer were identied. Meanwhile, we evaluated the mRNA and protein expression levels of these key targets using multiple biological analyses. Results: In total, 116 down-regulated and 73 up-regulated differential expressed genes (DEGs) were examined from seven gene expression datasets. Top ten hub genes were obtained in four typical protein-protein interaction (PPI) network of DEGs involved in each specic biological process (BP, cell cycle, cell proliferation, cell apoptosis and death) that was related to inhibitory activity of APS in vitro against breast cancer cell lines. Four common DEGs (EGFR, ANXA1, KIF14 and IGF1) were further identied in the above four BP-PPI networks, among which EGFR and ANXA1 were the hub genes that were potentially linked to the progression of breast cancer. The results of biological detections indicated that APS could downregulated EGFR and up-regulated ANXA1 expression. Conclusion: In conclusion, the present study may provide potential molecular therapeutic targets and a new insight into the mechanism of APS against breast cancer.


Introduction
Breast cancer is one of most commonly diagnosed cancers in women all around the world. Despite the remarkable advancements in the detection and treatment of breast cancer, it remains the most lifethreatening disease accounted for 25% of all cancers in women [1]. The exploration of novel therapeutic agents with preferable e cacy and low toxicity is urgently needed for breast cancer treatment. Currently, several polysaccharides have been developed and attracted growing interests in the adjuvant therapy of various carcinomas since they are effective or sensitize chemotherapy contributing to improve quality of life and patient survival [2].
Astragalus polysaccharide, the main active ingredient extracted from the natural product astragalus membranaceus, has been considered as a good candidate for adjuvant treatment of cancer owing to its low toxic, immunomodulatory and anti-tumor properties [3]. APS mainly exert anti-tumor effects via the following ways: immune system enhancement, direct cytotoxicity, and synergistic effects in combination with chemotherapy. Extensive studies have been investigated the tumor-killing capability of APS in multiple tumor types [4,5]. Ye et al indicated that APS could inhibit the proliferation of breast tumor cell line by up-regulating the expressions of p53 and PTEN [6]. APS combined with vinorelbine and cisplatin was reported to increase tumor response, signi cantly improve performance status of patients in advanced non-small-cell lung cancer [7]. It has been evident from various animal studies and clinical trials that APS produces indirect cytotoxicity or promotes host immune response in combination with other chemotherapeutic drugs [8,9]. Beside this, the other underlying molecular anti-cancer mechanism of APS remains unclear.
It is inevitable to assess the in vivo anti-tumor potential of APS by enhancing immune functions of host organism with suitable animal models. However, the immune system is relatively complex and the e cacy of tests in vivo displayed through immune system will emerge slowly after at least 3 weeks [10]. Consequently, it is time-and cost-consuming and di cult to explore the speci c immune-regulation mechanism of APS against breast carcinogenesis. The introduction of bioinformatics analysis to evaluate the essential genes involved in the development of breast cancer provides a value prediction that is e cient and less loss in clinical expense prior to the following experimental veri cation [11].
Previous studies performed in our lab investigated the anti-cancer potentials of APS towards 4T1-solid tumor and breast cancer cell lines both in vivo and in vitro, mainly by enhancing immune response, coordinating chemotherapy drugs, inducing apoptosis, arresting cell cycle and inhibiting cell proliferation [8,12]. To discover the underlying anti-cancer molecular mechanism of APS, we employed bioinformatics approaches to nd out differentially expressed genes (DEGs) of normal and breast tumor tissues.
Additionally, these DEGs underwent Gene ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment and protein-protein interaction (PPI) network construction. In especial, representative DEGs were screened out involved in biological processes (BPs, cell cycle, cell proliferation and cell apoptosis) to further construct BPs-PPI networks. The common hub genes were picked out followed by multiple experimental validations to reveal the potential targets and mechanism of APS against breast cancer. We found that the inhibitory activity of APS towards 4T1-solid tumor may be associated with the regulation of EGFR and ANXA1 ( Figure.1).

Materials and regents
The green uorescent protein (GFP)-labeled and luciferase-labeled 4T1 cells used in the study were purchased from Mingjin Energy Technology Co., Ltd. (Shanghai, China). The required medium was renewed at regular basis after 2 days. When the cells con uence was 80-90% then a passage was required. Astragalus polysaccharide was obtained from Pharmagenesis Inc. (Beijing, China). 5uorouracil (5-FU) was acquired from Sigma Aldrich Co. (St Louis, MO, USA). Fatal bovine serum (FBS), trypsin-EDTA, RPMI-1640 medium and penicillin-streptomycin were obtained from Gibco (Grand Island, NY, USA). The anti-ANXAI and Anti-EGFR antibody was obtained from Abcam Inc. (Cambridge, MA, USA). All the necessary chemicals utilized in the study, were of analytical grade.

Identi cation of DEGs
Microarray expression pro le breast cancer biological data (E-GEOD-10780), (E-GEOD-65194), (E-GEOD-10810), (E-GEOD-42568), (E-GEOD-61304), (E-GEOD-7904) and (E-GEOD-15852) were retrieved from European bioinformatics institute (EBI) database, which was comprised of 251 normal samples and 483 breast cancer samples. RMA algorithm in Affy package was performed to preprocess the raw expression data in the R statistical software consist of background correction, medianpolish summarization and quantile normalization methods. Furthermore, SAM method (signi cance analysis of the microarrays) was used for the assessment of differentially expressed genes between normal and BC samples.

Functional and pathway enrichment analysis
To describe the biological characteristics and functional annotation of DEGs, Go and KEGG pathway enrichment analysis were performed using Database for Annotation, Visualization and Integrated Discovery (DAVID, https://david.ncifcrf.gov/) [13] online tool. P < 0.05 was set as the cut-off criterion.

Construction of PPI network and hub gene selection
The online database Search Tool for the Retrieval of Interacting Genes (STRING, http://string.embl.de/) was employed to build up PPI network of DEGs, for the assessment of considerably important linkage between DEGs. Besides, another PPI network of the DEGs was also constructed in the rst six KEGG pathways with the most signi cant difference. The con dence score ≥ 0.2 was set as the cutoff criterion.
For the visualization of PPI network Cytoscape 3.5.1 software was used.
The previous ndings in our lab indicated that the activated macrophages by APS in vitro could inhibit the progress of breast cancer cells via cell cycle arrest, inhibiting cell proliferation and promoting cell apoptosis [8,12]. Specially, STRING was applied to construct the PPI networks of the DEGs associated with cell proliferation, cell cycle and cell apoptosis from GO enrichment analysis. Subsequently, hub genes were determined on the basis of degree > 10 by cytoHubba plugin, the common genes were identi ed from the PPI network constructed by the DEGs involved in the above biological processes.
Transplanted tumor and treatment protocol A sum of 60 female BALB/c mice (18-22g, 6 ~ 8 weeks old) were obtained from Dalian Medical University (Dalian, China) and were housed at 25 ± 1°C in 50-60% humidity under a 12 h dark/light cycle. Tumor was transferred to the healthy mice by injecting 1.5 × 10 6 of the GFP-4T1 cells, targeting the female BALB/c mice right unk area. Besides, the equal numbers of 4T1-luc cells were injected to evaluate tumor progress and drug e ciency using live animal imaging. When the tumor was extended to a size of 100mm 3 , all animals were randomly allocated to ve groups (n = 10/group) as follows: (I) Negative control group: intraperitoneal injection with 0.1 ml normal saline; (II) a low dose APS group: intraperitoneal injection with 100 mg/kg/d APS; (III) a high dose APS group: intraperitoneal injection with 200 mg/kg/d APS; (IV) 5-FU group: intraperitoneal injection with 20 mg/kg/d 5-FU as a positive control; (V) combined group: the mice were administered APS at 200 mg/kg/d and 5-FU at 20 mg/kg/d, once daily. Using Vernier caliper scales the volume of existing tumor was measured regularly every next day, and was calculated with the following equation: length × width 2 × 0.5. After fourteen days, live animal imaging was conducted to observe the effects of APS on the development of subcutaneous tumors in vivo using IVIS Spectrum imaging system (PerkinElmer, MA, USA). Mice were anesthetized with 3.5% chloral hydrate and real-time images were taken and further analyzed using the Living Image 3.1 software. Afterwards, the mice were killed all together, and the tumors were excised, xed in formalin (10%), and were embedded in para n for subsequent studies.

Cell viability evaluation
Live/dead staining and CCK-8 analysis were performed to assess the viability of GFP-4T1 cells sorted from in vivo tumors after 14 days of drug treatment. In brief, GFP-4T1 cells were harvested after the excisional tumor tissues were underwent cutting, digestion with collagenase, ltration with 200 mesh lter, erythrocytes removal with lysate. GFP-4T1 cells were sorted using FACS Aria III cell sorter (BD Biosciences, USA). Upon overnight adherent culture, 4T1 cells in each group were incubated with calcein-AM(2 µmol/L), propidium iodide (PI, 4 µmol/L), and Hoechst 33342 staining (5µg/L) in PBS at 37°C for 15 min. Afterwards, cells were imaged with an inverted uorescence microscope. Besides, CCK-8 analysis was conducted to evaluate cell viability as previously described [14]. The optical density value of each group was measured at a wavelength of 450 nm using an iMark microplate reader (Bio-Rad, CA, USA). as a ratio between the internal control β-actin and hub genes occurring in the same sample of mRNA, and was calculated by the comparative CT method. The 2-ΔΔCt method was used for the calculation of relative mRNA fold changes. The primers of hub genes were synthesized by GenScript Biotech Corp.

Quantitative real-time PCR analysis
(Nanjing, China), and their sequences were given in Table 1.

Immunohistochemical analysis
Immunohistochemical (IHC) analysis was conducted to assess the APS effects on the expression of key proteins in 4T1 cells and analyze anti-cancer mechanism of APS based on bioinformatic approach.
Brie y, para n-embedded tumor sections (5 µm) were depara nized, rehydrated and boiled in a citrate buffer solution for antigen retrieval. Tumor sections were incubated in 3% H 2 O 2 and blocked with goat serum. Subsequently, the samples were incubated with antibodies for EGFR and ANXA1 at 4°C overnight, followed by being incubated with goat anti-rabbit (HRP), stained with DAB chromogen and counterstained with hematoxylin. All sections were observed, using light microscope (Olympus, Tokyo, Japan). A semi-quantitative analysis of the IHC images was performed by utilizing the Image-Pro Plus 6.0 (IPP, Media Cybernetics, USA) software.
Immuno uorescence and ow cytometry assay Immuno uorescence (IF) and ow cytometry assays were conducted to evaluate the expression of EGFR and ANXA1 proteins in GFP-4T1 cells sorted from subcutaneous tumors after treatment with APS for 14 days. GFP-4T1 cells were harvested and cultured for adhesion. Cells were xed with 4% paraformaldehyde for 20 min, permeabilized at 0.1% Triton X-100 for 30 min and blocked with 10% goat serum in PBS for 60 min. Subsequently, cells were incubated with primary antibody (EGFR, ANXA1, 1:100) overnight at 4°C prior to incubation with second antibody (Cy3, 1:100) at room temperature for 1h and counterstaining with DAPI at 37°C for 15 min. Images were taken using a laser confocal microscope (FV1000, Olympus, Japan). Besides, the expressions of EGFR and ANXA1 were analyzed using ow cytometry as described above except for adding centrifugation before each procedure. All ow cytometry data were analyzed with FlowJo7.6.2 (Tree Star, San Carlos, CA, USA).

Statistical analysis
All the results of this study were presented as mean ± SD. A one-way analysis of variance (ANOVA) was used for the comparison of the means among the different groups. Moreover, a signi cant difference of P < 0.05 was also noted.

Page 7/27
The identi cation of DEGs We downloaded seven gene expression datasets from EBI database including 483 breast cancer samples and 251 normal samples. Sums of 201 genes were screened to be differentially expressed genes. Furthermore, 189 genes were identi ed as the uniformly regulated genes after selecting the intersections of differential gene sets, among which 73 genes were upregulated and 116 genes were downregulated ( Table 2). Table 2 Common up-regulated and down-regulated DEGs in the seven breast cancer data GO functional annotation and KEGG pathway enrichment analysis of DEGs To explore in detail the function of these DEGs in BC, GO terms including biological process (BP), molecular function (MF), and cell component (CC) were evaluated by utilizing DAVID. It has been revealed from the results that DEGs were speci cally enriched in certain biological processes, including cell division, cell cycle, chromosome segregation and cell proliferation (Fig. 2). As to CC, DEGs were primarily enriched in cytoplasm, membrane, cytoskeleton and proteinaceous extracellular matrix. For MF, DEGs were profoundly enriched in protein binding, glycosaminoglycan binding, receptor binding and heparin binding. All changes of these functional enrichments are closely associated with the adhesion, growth, metastasis, invasion and proliferation of tumor cells. On this basis, the occurrence and development of tumors result from the comprehensive regulation of multiple pathways with network structure rather than the action of single or several pathways.
In order to evaluate the 201 DEGs for the enrichment of KEGG pathway, DAVID was used, and the results were shown in Fig. 3a and Table 3. KEGG pathways results indicated that the DEGs were predominantly enriched in cell cycle (12 genes), AMPK signaling pathway (9 genes) and pathways in cancer (14 genes).
In addition, the subnetworks of the top six pathways including cell cycle, AMPK signaling pathway, pathways in cancer, ECM-receptor interaction, proteoglycans in cancer and focal adhesion were constructed by STRING software (Fig. 3b), which intuitively presented the direct connection between the different genes enriched in each pathway. Here, the PPI network of 189 DEGs (simultaneous up-regulation or down-regulation) was constructed with 146 nodes and 733 edges by STRING database (Fig. 4a). A total of 56 genes in the BPs associated with cell cycle, cell proliferation, cell apoptosis and death were constructed to the BP-PPI network that containing 56 nodes and 267 edges (Fig. 4b). In addition, four PPI networks of the involved DEGs in each representative BP were built as shown in Fig. 4c I -IV. The top ten genes with high degree score were presented in these 4 BPs, among which 7 genes were remarkably identi ed (degree > 45) such as CCNB1, DLGAP5, CDC7, CSE1L, GINS1, AURKA and ECT2, suggesting their dominant roles in the regulation of the proliferation or apoptosis of breast cancer cells.

Common gene selection of DEGs in speci c BP-PPI network
The PPI network of the DEGs strongly connected with the 4 representative biological processes was further constructed and 4 common genes were identi ed including EGFR, ANXA1, KIF14 and IGF1 after the division of DEGs in each BP as shown in Fig. 5 (blue square). Of particular note, EGFR (degree = 16) and ANXA1 (degree = 15) were also the hub genes in the 4 BP-PPI networks as presented in Table 4, indicating their particular importance in the enhancement of breast cancer cell apoptosis and progression. Thus, key targets EGFR and ANXA1 were identi ed for experimental veri cation to study the possible mechanism of anti-breast cancer of APS. Effect of APS on the expression of EGFR and ANXA1 The results of live animal imaging demonstrated that tumor volumes markedly decreased with the intervention of APS for 14 days as compared to that in negative control group (Fig. 6a). Besides, APS inhibited tumor growth evidenced by the tumor appearance and signi cantly depressed tumor weight ( Fig. 6b-c). In addition, GFP-4T1 cells were sorted from subcutaneous tumors in BALB/c mice after treatment with APS for 14 days. Upon incubation overnight, live/dead staining indicated that the number of living GFP-4T1 cells reduced, while the number of dead cells increased in APS groups (Fig. 6d). CCK-8 analysis showed APS (200mg/kg) signi cantly decreased cell viability as compared to that in untreated group (P < 0.001). Further, it's more cytotoxic of APS towards 4T1 cells when combined with 5-FU (P < 0.001, Fig. 6e). All these results demonstrated the growth-inhibiting effect of APS on 4T1-solid tumor.
In this study, we further examined whether the inhibition of tumor growth in vivo by APS was associated with the regulation towards EGFR and ANXA1. The mRNA and protein expression of the ANXA1 and EGFR were measured by RT-PCR and IHC. Upon treatment with APS, the positive expression of ANXA1 protein increased, while the level of EGFR protein decreased in 4T1 solid tumors, especially in the high dose APS group. The trends of increased and decreased expression of two proteins in breast cancer tissues were more remarkable in the combined group (Fig. 7a). The results of RT-PCR and semi-quantitative evaluation of IHC revealed that APS treatment profoundly up-regulated the expressions of ANXA1 mRNA and protein comparatively to that in negative control group (P < 0.05, Fig. 7b-c). On the contrary, APS administration markedly down-regulated the expression of EGFR mRNA and protein compared with that in negative control group (P < 0.05, Fig. 7b-c).
Post analysis for the sorted GFP-4T1 cells from subcutaneous tumors in BALB/c mice, the results of ow cytometry analysis revealed that decreased percentage of GFP-4T1 cells from solid tumors was identi ed in APS groups compared with that in untreated group (Fig. 8a), which also suggested the anti-tumor activity of APS. Post overnight culture, IF and ow cytometry assays were further performed to study the expression of EGFR and ANXA1 proteins from the sorted GFP-4T1 cells. Note that morphologic changes between non-labeled and GFP-labeled 4T1 cells. GFP-4T1 cells showed a contracted morphology and were prone to aggregate growth, while non-labeled 4T1 cells were atter (Fig. 8b). Decreased expression of EGFR in 4T1 cells was observed after exposure to APS, particularly in combination with 5-FU (Fig. 8c,   e), while the expression of ANXA1 was increased after undergoing APS and combined treatment (Fig. 8d,  e). Meanwhile, DAPI staining showed APS induced cell apoptosis with nuclear chromatin condensation and nuclear fragmentation (white arrow, Fig. 8d). APS (200mg/kg) signi cantly down-regulated EGFR expression and up-regulated ANXA1 expression compared with that in negative control group (P < 0.01, Fig. 8f). All these ndings indicated that the regulation of EGFR and ANXA1 in 4T1 solid tumor may be the underlying anti-cancer molecular mechanism of APS in addition to or following the improvement of the immune function.

Discussion
Recently, APS has attracted growing interests in biological elds especially in terms of cancer therapy.
Most of studies have clari ed that APS inhibit tumor growth through immunopotentiation activity and generally be considered as a powerful adjunctive strategy in combination with chemotherapy for various tumors [15,16]. There are also parts of studies suggest the direct anti-tumor activity of APS, such as the inhibition of cell proliferation, the induction of cell apoptosis and cell cycle arrest [17,18]. In the present study, for the rst time, we employed bioinformatics approach to speci cally target the common hub genes involved in the in vitro anti-tumor biological processes, and further underwent experimental veri cation. The aim of this study was to explore other possible anti-tumor mechanisms of APS and to nd the potential therapeutic targets for breast cancer.
KEGG pathways analysis demonstrated that the DEGs were mainly enriched in cell cycle, AMPK signaling pathway and pathways in cancer, which are potentially associated in the progression of breast tumor and used as the regulating target for cancer treatment [19,20]. For instance, downregulation of AMPK activity or decreased expression appear to promote breast tumorigenesis, while AMPK activation was found to impede tumor growth. Pan et al showed that cetyl trimethyl ammonium bromide could enhance DOX chemosensitivity mainly through activation of AMPK signaling cascades in breast cancer [21]. Another report found that targeting AMPK alpha signaling was capable of improving radiosensitivity of triple negative breast cancer [22]. All these potentially enriched pathways will contribute to identify important molecules involved in the growth of breast cancer.
It is more meaningful to study the mechanism of tumor progression by classifying the genes involved in the same function or pathway using GO functional annotation. In this study, PPI networks of all DEGs and corresponding DEGs in four BPs including cell cycle, cell proliferation and cell apoptosis were specially constructed and the hub nodes were selected in all PPI networks. Such treatment focused on the biological process network pathway of differential genes associated with breast cancer rather than being con ned to individual genes to study the expression changes of key genes or proteins throughout the entire pathway. Here, the hub nodes were purposefully analyzed in PPI networks constructed by DEGs in 4 typical BPs (cell cycle, cell proliferation, cell apoptosis and death) because they were involved in the anticancer mechanism of APS mediated macrophages in our previous study [12,23]. Moreover, the regulatory genes of cell cycle are frequently linked to the incidence and growth of tumor, which are mutated under the action of carcinogenic factors, causing the disorder of cell cycle followed by the malignant proliferation of cells and tumor formation [24,25]. It has been extensively acknowledged that cell apoptosis plays a potential role in the inhibition of tumor growth, or it may develop resistance to tumor progression [26,27]. Further, tumor cell death induced by most of anti-cancer strategies is directly related to the activation of apoptosis. On this basis, the main focus of this study was to investigate the hub genes in the four BP-PPI networks to explore other possible anti-breast cancer mechanisms using bioinformatics based on previous ndings in our lab.
In this study, a sum of 56 genes in the four BP-PPI networks, among which 7 genes (CCNB1, DLGAP5, CDC7, CSE1L, GINS1, AURKA and ECT2) were particularly signi cant with the degree over 45. For these genes, CCNB1 is essential for controlling the cell cycle at the G2/M transition, which has been displayed in multiple cancer types playing important roles in the promotion and transformation of tumor [28].
DLGAP5 and CDC7 are potential cell cycle regulators that play roles in carcinogenesis of cancer cells. Elevated expressions of DLGAP5 were noticed in breast cancer specimens and were associated with poor prognosis [29]. Moreover, a cellular apoptosis susceptibility protein (CSE1L), is extensively expressed in certain types of cancers. Mounting evidences from several pathological reports have been indicated that highly expressed CSE1L are most frequently associated with the tumor proliferation [30]. Tai et al indicated that CSE1L was of paramount importance in facilitating tumor cells apoptosis imparted by chemotherapeutic agents and could also regulate the metastasis and invasion of breast cancer [31].
Accumulating evidence has indicated that the aberrant expression of GINS1 was involved in cancer pathogenesis [32]. Breast cancer patients with elevated expression of GINS1, have been predicted with signi cantly poor prognosis, however the knockdown of GINS1 could inhibit the malignant features of breast cancer cells [33]. AURKA is involved in regulating many early mitotic events that also plays an important role in tumorigenesis and tumor progression. Extensive studies have demonstrated that AURKA was overexpressed in certain types of malignancies, such as bladder cancer, esophageal squamous cell carcinoma and breast cancer [34,35]. It is evident from previous studies that ECT2 can improve cell growth, invasion and tumorigenicity [36]. All these ndings demonstrate the vital roles of these genes in the progression of breast cancer, and will be veri ed in our further study.
Previous study in our lab demonstrated that the in vivo the inhibitory activity of APS towards 4T1-solid tumor was partially associated with the enhancement of immune responses [8]. Here, the other in vivo underlying molecular mechanism of APS against breast cancer was further studied using bioinformatics method targeting in vitro anti-tumor strategies. In the present study, common hub genes EGFR and ANXA1 were identi ed in biological processes involving in cell cycle, cell proliferation and cell apoptosis targeting to the inhibitory mechanism of APS mediated macrophages against breast cancer cells.
Correspondingly, IHC and PCR were carried out to study the effect of APS on the proteins and genes expression of EGFR and ANXA1, respectively. Currently, accumulating evidences have demonstrated that EGFR and ANXA1 are aberrant expression in multiple tumors [37,38]. EGFR, a transmembrane glycoprotein, belongs to the extracellular protein ligand of the epidermal growth factor family that is involved in a myriad of biological processes and carcinogenic events. It can be used as a tumor marker for serological detection. Signi cant difference of EGFR expression was identi ed in the serum of healthy people and gastric cancer patients [39]. EGFR can be detected in the serum of patients with early-stage gastric cancer, which thus can be used as one of the screening indicators for high-risk groups. Upregulated EGFR protein and transcript levels were related to poor prognosis in various cancers, including colorectal cancer, endometrial cancer and non-small cell lung cancer [40]. Besides, EGFR is also a hot target and predictive indicator of early recurrence and death of breast cancer. The degree of EGFRs high expression determines the amount of the down regulation, which occurs in the natural functions of family members (ErbB4/Her4, ErbB3/Her3, ErbB1/EGFR1 and ErbB2/Her2) and their use as indicators for clinical outcome in breast cancer and other tumor types [41]. The study reported that the high levels of surface tyrosine kinase receptors of EGFR family were frequently found in breast cancers [42]. The overexpression of EGFR was identi ed in approximately 40% of triple-negative breast cancer (TNBC) patients that provide a potential strategy for targeting TNBC therapy [43]. The result of EGFR with high degree indicated its signi cant role in the progression of breast cancer and re ected the feasibility of PPI analysis.
ANXA1 refers to the rst protein found in the Annexins family that plays an extremely vital role in tumor proliferation and differentiation, in ammatory response, cell signal transduction, calcium ion signaling channels, and interactions between cytoskeletal proteins [44]. Currently, numerous reports have demonstrated differential expression of ANXA1 depending on cancer type. Several studies have reported higher levels of ANXA1 in esophageal cancer, skin squamous cell carcinoma and colorectal cancer, while others have reported decreased expression of ANXA1 in head and neck cancer, prostate cancer, oral squamous cell carcinoma and cervical cancer [45]. Signi cant down-regulation of ANXA1 protein was identi ed in esophageal squamous cell carcinoma compared with that in paracancerous normal tissue [46]. Notably, there are contradictory reports about the role of ANXA1 plays in breast cancer. Mussunoor et al indicated that the over expression of ANXA1 was found in triple negative breast cancer and metastatic cancer, which plays a prominent role in tumor progression and development [47], while another report showed that ANXA1 was downregulated during late stage of progression and metastasis in breast cancer [48]. Using high-throughput analysis and also evidenced by IHC and RT-PCR that the event of decreased expression of ANXA1 occurs in the human breast cancer [49]. Besides, previous report showed that the expression of ANXA1 remarkably decreased in breast tumor including ductal carcinoma in situ and invasive carcinoma. Whereas, overexpression of ANXA1 was related to unfavorable prognostic factors in invasive cancer [50]. All these ndings suggested that ANXA1 played dualistic roles and was involved in variable mechanisms associated with the advancement and proliferation in breast cancer.

Conclusion
In summary, we identi ed a possible speci c molecular mechanism of APS against 4T1 solid tumor using an integrated bioinformatics and experimental analysis, which indicated that EGFR and ANXA1 may be the potential therapeutic targets in breast tumor. In this study, a total of 201 DEGs were screened out, among which four common genes (EGFR, ANXA1, KIF14 and IGF1) participated the biological processes (cell cycle, cell proliferation and cell apoptosis) that were associated with the inhibitory effect of APS in vitro towards with breast cancer cell lines. Further, EGFR and ANXA1 were identi ed as hub genes that played important roles in the development of breast cancer. The down-regulated and upregulated expression of EGFR and ANXA1 separately may be responsible for the anti-cancer mechanism of APS in vivo veri ed by multiple biological analyses. More breast cancer data pro les coupled to various algorithm theories and methods are needed to improve the accuracy of key genes, and provide further evidence to identify the targets of breast cancer and anti-cancer mechanisms of APS against breast cancer.

Data Availability Statement
All relevant data are within the manuscript. Figure 1 Schematic diagram of the identi cation of potential targets and anti-cancer molecular mechanism of APS.

Figures
Page 21/27   PPI networks constructed by DEGs from four BP terms. Green color indicated down-regulated genes, red color indicated up-regulated genes, and blue color indicated non-co-regulated genes. Scale bar = 200 μm. e The viability of sorted GFP-4T1 cells were quantitatively assessed by CCK-8 (n=4). **P < 0.01, ***P < 0.001, compared with negative control group.