Effects and mechanisms of AMIGO2 in proliferation, migration and drug resistance of bladder cancer

Background Bladder cancer is the most common malignancy in urinary system, but the therapeutic targets remain elusive. This study aims to reveal the relationship between AMIGO2 and proliferation, migration, drug-resistance and tumorigenicity of bladder cancer, and explore the potential molecular mechanisms. Methods The expression of AMIGO2 in human bladder cancer tissues is measured by qRT-PCR and immunohistochemistry (IHC). Stable AMIGO2 knockdown cell lines T24 and 5637 were established by lentivirus transfection. Cell viability assay (CCK-8 assay) was used to determine cell proliferation, ow cytometry analysis was utilized to detect cell cycle, and wound healing assay was proceeded to test migration ability of bladder cancer cells. Chemosensitivity to cisplatin was measured by CCK-8 assay. Xenograft mouse model was established for investigating the effect of AMIGO2 on tumor formation in vivo. The RNA Sequencing technology was used to explore differentially expressed genes (DEGs) between knockdown group and negative control group of T24. Bioinformatics analysis upon the results of RNA-Seq was proceeded to understand underlying mechanisms. AMIGO2 was upregulated in bladder cancer cells and tissues. Inhibited expression of AMIGO2 suppresses cell proliferation which might be cell cycle in G1 phase. AMIGO2 could reduce chemoresistance to cisplatin in bladder cancer cells. Low AMIGO2 expression inhibited tumorigenicity of T24 in nude mice. 917 DEGs were identied by RNA-Sequencing technology and bioinformatics analysis. The DEGs were mainly enriched in cell-cell adhesion, ATP-binding cassette transporters (ABC transporters), PPAR signaling pathway and some other pathways. Among ten hub genes, four of them might be associated with the prognosis of bladder cancer patients. gel and transferred to polyvinylidene diuoride (PVDF) membranes (Millipore, USA). The membranes were blocked with 5% milk, incubated overnight with the primary antibodies: Anti-AMIGO2 (Santa-Cruz, USA); Anti-GAPDH (Santa-Cruz, USA), and probed with horseradish peroxidase-conjugated secondary antibody: Anti-Mouse IgG (Santa-Cruz, USA). The blots were then detected using Pierce™ ECL Western Blotting Substrate Kit (Thermo, USA).


Flow cytometric cell cycle analysis
Cells (1×10 6 ) were harvested, re-suspended in PBS, xed in 75% ice-cold ethanol, and incubated in propidium iodide (PI, 20µg/ml; Sigma, USA) in the dark for 20min. Cell cycle analysis was performed using the BD LSRII Flow Cytometry System with FACSDiva software (BD Bioscience, Franklin Lakes, USA). The data were analyzed using the Mod t LT software.

CCK-8
Cells were seeded into 96-well plates at 2000 cells/well. Cell proliferation was detected using the Cell Counting Kit-8 (CCK-8, Dojindo Laboratories, Japan). At indicated time point, CCK-8 solution was added to each well and incubated for 3h. The absorbance was determined at 450nm using a microplate reader (Tecan in nite, Switzerland).
Wound-healing assay 3×10 5 T24 or 5637 cells were seeded into 6-well plates with complete culture medium. The con uent monolayer of both cells was scratched with a 1 ml pipette tip. The cells were then washed with PBS three times and kept in the 1640 (supplemented with 10% FBS) for 48h. The wound width was recorded at 0h and 24h under a microscope.

Xenograft mouse model
Nude mice (BALB/c-nu, female, 4 weeks) were bred and housed in AAALAC-accredited speci c pathogenfree rodent facilities. Mice were housed in sterilized, ventilated microisolator cages and supplied with autoclaved commercial chow and sterile water. The mice were randomly divided into two groups (n=8).
Tumorigenicity was determined by subcutaneously injection of T24 cells into the anks of female nude mice (1×10 6 cells per site). The tumor size was measured 2~3 times per week, up to 28d. Tumor volumes were calculated using the following formula: V=π/6 x largest diameter x smallest diameter 2 (11) . No blinding was done. All mouse experiments were conducted with standard operating procedures approved by the University Committee on the Use and Care of Animals at the second hospital of Lanzhou University (Approval No. D2019-171).

RNA-Seq data analysis
Total RNAs of shAMIGO2 and shCtrl were extracted. We used RNA sequencing technology (Novogene, China) to identify DEGs. Brie y, the RNA-seq raw fastq data were rst trimmed using Trimmomatic (V0.35). The trimmed reads were aligned to the human reference genome (NCBI GRCh38) with TopHat V2.0.12 with default parameter settings. The aligned bam les were then processed using Cu inks V2.2.1 for gene quanti cation. Reads were then mapped to ERCC transcripts and quanti ed using TopHat V2.0.12 and Cu inks V2.1.1 with default parameter settings. Genes with FPKM ≥ 1 in all samples were used for DEG analysis. We could upload the data from RNA-Seq to a public repository if necessary.

Data analysis and DEGs identi cation
The sequencing data set was normalized and analyzed using the DESeq package (1.10.1). The criteria of a false discovery ratio (FDR) <0.01 and |logFC| >2 were set as the threshold. Visualization of the volcano plot and heat map were done using R software. The most upregulated 100 genes and downregulated 100 genes were chosen for the heat map.

Biological function and pathway enrichment analysis
Through calculating the corresponding topological overlap, genes positively associated with AMIGO2 were found out and subjected to gene ontology (GO) analysis (GOSeq, Release2.12) to determine clusters of DEGs with enriched molecular functions. Kyoto encyclopedia of genes and genomes (KEGG) pathway analysis was performed via the "clusterPro ler" package in R/Bioconductor software to acquire the enriched biological process and KEGG pathway. p<0.05 and counts ≥ 4 were considered signi cant.

Module analysis of protein-protein interaction (PPI) network
Search Tool for the Retrieval of Interacting Genes (STRING) database (http://www.string-db.org/) was used to acquire PPI information for the DEGs. Cytoscape software (3.7.2) was applied to visualize the PPI network. The top DEGs with a high degree of connectivity in the PPI network were selected to discuss their function and effect on bladder cancer.

GSEA of DEGs on the whole gene expression level
Gene set enrichment analysis (GSEA) online tool (http://software.broadinstitute.org/gsea/index.jsp) 12 was applied to verify the results of GO and KEGG analysis. The cut-off criteria for GSEA were p<0.05. We created a chip expression pro le le and a sample data le, and imported them into GSEA software. After choosing gene sets database and corresponding chip platform, setting other parameters as default, we could run GSEA and acquire pathway enrichment results on the total gene expression level.

Statistical analysis
All data were reported as mean ± SEM from three independent experiments. The two-tailed Student's t-test was used to evaluate statistical differences between two groups. The survival curve was described by the Kaplan-Meier plot and was calculated using the log-rank test. *p< 0.05; **p<0.01; and ***p<0.001 were considered statistically significant.

AMIGO2 is upregulated in bladder cancer cells and tissues
The result of qRT-PCR showed that the AMIGO2 expression level was elevated in bladder cancer cell lines UMUC3, 5637 and T24 ( Figure 1A). According to TCGA database, the relative differential expression level (logFC) of AMIGO2 in bladder cancer is also higher than adjacent normal tissue (n=19) ( Figure 1B). The relative mRNA expression of AMIGO2 in bladder cancer tissues was significantly higher than that of their matched adjacent normal tissues (n=11) ( Figure 1C). We also found that AMIGO2 expression was upregulated in bladder cancer tissues compared to the matched adjacent normal tissues by immunohistochemistry (n=43) ( Figure 1D). Upon clinicopathological correlation analysis, elevated AMIGO2 is positively correlated with advanced tumor stage (p=0.022) and tumor grade (p=0.041) ( Table 1).   Figure 3C). The shAMIGO2 group exhibited a lower suppression ratio than shCtrl group in a series of dilute concentrations of cisplatin ( Figure 3D). Meanwhile, the IC50 of shAMIGO2 is higher than that of shCtrl (T24, p=0.002; 5637, p=0.004) ( Figure 3E). All these results supported that downregulation of AMIGO2 reduces the proliferation and migration and induces G1 phase cell cycle arrest, and that AMIGO2 might reduce chemoresistance to cisplatin in bladder cancer cells.

Inhibition of AMIGO2 reduces tumorigenicity of bladder cancer cells in vivo
The growth of tumors derived from the shAMIGO2 group was prominently suppressed compared with the shCtrl group from 20 days after tumor inoculation (n=5) (Fig 4A-B).
Also, we detected the expression of AMIGO2 in the xenograft tumors by qRT-PCR. The result showed that the expression level is lower in shAMIGO2 group than that of shCtrl group ( Fig   4C). These results indicated that inhibition of AMIGO2 reduces bladder cancer cell growth and tumorigenicity in vivo, which was consistent with the in vitro results.
Identification of differentially expressed genes (DEGs) and molecular function and pathway enrichment analysis As the volcano plots illustrated, after data integration, gene expression profiles from RNA Sequencing identified 917 differentially expressed genes. 627 genes were upregulated and 290 were downregulated in shAMIGO2 compared with the shCtrl in T24 cell ( Figure 5A), grounded on the cut-off criteria (|logFC|>2, Padj<0.01). DEGs were selected for integrated analysis. We gained the most upregulated 100 genes and 100 downregulated ones to portray the heat map ( Figure 5B), with red represents high expression level and blue stands for low.
In order to investigate the molecular function and biology pathways of the DEGs, GO and KEGG analysis were performed. GO (Gene Ontology) includes molecular function, biological process and cellular component ( Figure 5C-E). Enriched KEGG pathways of the DEGs are shown in Figure 5F, including ABC transporters (ATP-binding cassette transporters), oxytocin signaling pathway, PPAR signaling pathway, etc.
Protein-Protein Interaction (PPI) network construction, hub genes survival analysis and

GSEA
The PPI network was constructed based on the SRTING database. A total of 174 proteins were obtained from the DEGs, including 136 nodes and 234 edges ( Figure 6A). In the network, nodes with top 10 highest degrees were ZAP70, AKR1C1, MAP2K6, SCN2A, genes. The information of the 10 hub genes is shown in Table 2, including full gene names and primary functions. Total 10 hub genes were obtained from PPI network. The Kaplan-Meier survival analysis ( Figures 6B-E) shows that four of the 10 hub genes associated with survival were ZAP70, AGMAT, AKR1C1 and AKR1C3. We further explored and verified KEGG pathway on the whole gene expression level by GSEA. The results, as shown in Figure   6F-G, displayed two pathways including oxidative phosphorylation and spliceosome.

Discussion
Bladder cancer remains to be the most common malignancy in urinary system. Based on the investigations of different genes in malignancies, data are emerging to elucidate gene functions in bladder cancer as well. As is reported before, dysregulation of different genes contributes to certain types of tumor progression. In the present study, we found that AMIGO2 is upregulated in bladder cancer cells and tissues, and it could promote the proliferation, migration and tumorigenicity. AMIGO2 could also reduce chemoresistance to cisplatin in bladder cancer cells. In addition, DEGs, molecular function and pathway enrichment analysis, a PPI network and 10 hub genes were identi ed using RNA-Seq technology and bioinformatics analysis, which demonstrated the mechanisms of how AMIGO2 regulates BCa cells.
The uncontrolled cell proliferation of cancer is mainly attributed to the cell cycle deregulation 14 . DNA damage targets two cell cycle checkpoints: G1/S and G2/M. DNA damage induces program that blocks cells at one of these checkpoints until the damage is repaired or the cells move towards apoptosis 15 .
Arresting cells in G0/G1 phase offers a chance for certain cells to either undergo repairing or tend to apoptosis process. The ow cytometry assay in our study showed that suppression of AMIGO2 could inhibit its proliferative effects through blockage of cell cycle progression and arrest BCa cells in G1 phase.
In many cases, an arrest can lead to senescence or apoptosis 16 . Although few studies have described the relationship between AMIGO2 and cell cycle before, there are some pathways that regulate cell cycle indirectly enriched by our KEGG analysis, but the particular pathways and molecules are still under research.
Cisplatin is regarded as cytotoxic drugs, it kills cancer cells by damaging DNA, inhibiting DNA synthesis and mitosis, and inducting apoptotic cell death 17 . Reduced cellular intake, increased e ux and increased DNA repair are thought to be resistance mechanisms to cisplatin in cancer cells 18 . According to our research, AMIGO2 could decrease chemoresistance to cisplatin in BCa cells. Meanwhile, ABC transporters is one of the most essential pathways identi ed by KEGG pathway enrichment analysis in this study. ABC drug transporters could increase the excretion of their substrate anti-cancer agents, cause decreased intracellular concentration of anti-cancer drugs, and develop the multidrug resistance (MDR) phenotype 19 .
Previous studies showed that ATP-binding cassette subfamily B member 1 (ABCB1) is also known as multidrug resistance protein 1 or P-glycoprotein; ABCC1 is considered as multidrug resistance associated protein 1, and ABCG2 is regarded as a breast cancer resistance protein 20  worked as one of the cell adhesion molecules in human ischemic cardiomyopathy 23 . Cell adhesion participants in stimulating signals that regulate cell cycle, migration, and cell survival 24 . Cell adhesiveness is generally reduced in different kinds of human cancers. AMIGO2, as reported before, is also involved in cell adhesion and/or cell migration 6,25,26 . Changes of these molecules might be leading to the reduction of adhesion between cells and promoting the migration of tumor cells. Meanwhile, the wound healing assay in this study also proved that AMIGO2 could promote the migration of BCa cells.
Finally, we explored the interaction of DEGs. As a result, a large and complex interactome network was established, suggesting intricate links among those DEGs. Moreover, 10 hub genes were identified in total, 4 of them were related to survival, namely ZAP70, AGMAT, AKR1C1 and AKR1C3. Studies have shown that ZAP70 is involved in the development of chronic lymphocytic leukemia 27 . AGMAT could promote the lung adenocarcinoma tumorigenesis by activating the NO-MAPKs-PI3K/ Akt pathway 28 . Interestingly, as previous studies stated, AKR1C1 could mediate the invasive potential and drug resistance of metastatic bladder cancer cells, and AKR1C1 was highly expressed in metastatic lesions of human bladder cancer patients 29 . AKR1C3, comes from the same family as AKR1C1, has not been proved to have similar functions as AKR1C1 in bladder cancer, but it is often overexpressed in prostate cancer tissues and prostate cancer cell lines 30 . AKR1C3 catalyzes the formation of prostaglandin (PG) F2α and 11β-PGF2α from PGH2 and PGD2, respectively 31 . The PGF2α and 11β-PGF2α can inactivate peroxisome proliferatoractivated receptor gamma (PPAR-γ) and has anti-proliferative effects 32 . Coincidentally, PPAR is one of the pathways enriched by KEGG analysis in our study, but the exact relationships and functions remain elusive.
To the best of our knowledge, this is the rst study determining the very "new" gene AMIGO2, which promotes the proliferation, migration and tumorigenicity in BCa. In addition, AMIGO2 could reduce chemoresistance to cisplatin. Some DEGs were identi ed to be related to cell-cell adhesion. Certain pathways were recognized to be involved in tumor development or drug resistance. Whereas, the molecules and pathways identi ed by bioinformatics analysis have not been examined by our research yet. Therefore, exploring molecular functions of DEGs and verifying pathways that are regulated by AMIGO2 would be our next-step work.

Competing interests
The authors declare that they have no competing interests. There was no additional external funding received for this study. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.