Identifying potential anti‐metastasis drugs for prostate cancer through integrative bioinformatics analysis and compound library screening

Metastasis poses the greatest threat to the lives of individuals with prostate cancer. Therefore, it is imperative to identify the underlying mechanism driving metastasis. Doing so would facilitate the detection of new diagnostic biomarkers and the advancement of treatment options for patients.


| INTRODUCTION
Prostate cancer (PCa) is the second most common cancer in men and the fifth leading cause of cancer-related death in men. 1 According to the SEER database, although the 5-year overall survival rate for prostate cancer is very high (98.2%),among the 5% of men with distant (metastatic) prostate cancer at diagnosis, their 5-year survival rate is still only slightly higher, at 30%. 2 The leading cause of prostate cancer death is metastasis.Worse, some cases exhibit rapid resistance to standard androgen deprivation therapy (ADT) combined with/without microtubule-targeted taxane chemotherapy and this leads to fatal disease. 3The commonly used drugs are α-receptor blockers and 5α-reductase inhibitors. 4However, the present clinical predictors (such as serum prostate-specific antigen, biopsy Gleason score) of biological outcomes for PCa still remain imprecise and imperfect. 5cently a chemical compound library has been widely used as a tool to identify active compounds in the cancer research field. 6However, the efficiency of compound library-based drug screening relies on large capacity, making the screening method expensive, timeconsuming and cumbersome, known as high-throughput screening.It is necessary to optimize the strategy.The development of highthroughput methods (such as large-scale microarrays and wholegenome sequencing) and bioinformatics methods has provided a wealth of information to guide drug screening.One such method is the weighted gene co-expression network analysis (WGCNA), which is able to analyze both differentially expressed genes, as well as their interconnectivity, enabling the discovery of synergistically expressed modules. 7The discovery of hub genes and subsequent screening for candidate biomarkers or therapeutic targets has been applied in various cancers.
Therefore, in the present study, we performed a systematic integration analysis of PCa metastasis.For the first time, a metastasisspecific gene co-expression module has been identified in PCa.These modules were then subjected to Gene Ontology (GO) (http:// geneontology.org)and Kyoto Encyclopedia of Genes and Genomes (KEGG) (https://www.genome.jp/kegg)pathway enrichment analysis.
The key gene regulation of each module was then identified.The compound library was then used to identify potential drugs that might be beneficial in treating prostate cancer.

| MATERIALS AND METHODS
The overall flowchart of this work is summarized in Figure 1.We downloaded PCa samples from the GSE6919 database (https://www.ncbi.nlm.nih.gov/gene/?term=GSE6919).Hub genes were analyzed by WGCNA.The data were then calculated according to diseasedrug-gene correlation analysis.Finally, the effect of Cu 2+ on PCa treatment was verified.

| Raw data
Systematic analyses were performed based on the data gathered from GEO and NCBI-gene.GSE6919 gene expression profiles were extracted from the GEO (https://www.ncbi.nlm.nih.gov/geo)database. 8Data from GSE6919 including 61 metastatic prostate tumors from the same batch and 24 primary prostate tumor samples were normalized using a quantile method for extended analyses (see Supporting information, Tables S1 and S2).Collected clinical pathological data included sex, age, stage, grade, survival status and survival duration.PCa-related genes were selected by comprehensively searching through the online databases NCBI-gene. 9These datasets were subsequently merged.

| WGCNA analysis
In the WGCNA analysis method, we employed a strategy to screen differentially expressed genes.Specifically, we selected differentially expressed genes with p < 0.05 and a fold change greater than 2 (i.e., jlog 2 FCj > 1) from the candidate gene list.This step helped to retain genes relevant to biological processes at the same time as reducing network noise.Next, we used the "WGCNA" package in R (R Foundation, Vienna, Austria) to perform gene co-expression network analysis.Prior to building the co-expression network, we preprocessed the data by removing low expression genes and outlier samples.We constructed a gene correlation matrix based on Pearson's correlation coefficient.We then used hierarchical clustering to select highly similar gene modules and applied a dynamic tree cutting algorithm to partition the modules.We chose a robust modeling approach by adjusting the deepSplit value to 2 and setting the minimum gene module size to 30.To further determine the biological significance of the gene modules, we performed GO enrichment analysis and KEGG pathway analysis.For each module, we calculated the enrichment p-value and adjusted p-value and selected pathways and functions relevant to the studied biological processes.We also calculated Pearson's correlation coefficient between each module and tumor.After completing the analysis, we used a network visualization tool to display the data and better illustrate the relationships between modules.To merge highly similar modules, we set the merge cut-off value to 0.25 and repeated the dynamic tree cutting. 10

| Experimental preparation
Hub gene expression in WGCNA was quantified using a quantitative  S3).

| qRT-PCR
Total RNA was extracted from cells using Trizol reagent (15596026; Thermo Fisher Scientific, Waltham, MA, USA).The sample RNA was reverse-transcribed to cDNA in accordance with the instructions of the reverse transcription kit (cw2569; Kangwei Century Company, Beijing, China).The reaction volume was 20 μL.A quantitative PCR was performed using a fluorescence quantitative RCP instrument (QuantStudio1; Thermo Fisher Scientific).The reaction conditions were denaturation at 95 C for 10 min, denaturation at 94 C for 15 s, and annealing at 60 C for 30 s, for 40 cycles.β-actin was used as the internal reference primer.Primer sequences are listed in the Supporting information (Table S1).With 2 μg of cDNA as a template, the relative quantitative method (2 À44Ct method) was used to calculate the relative transcription level of the target gene: 44Ct = 4 experimental group -4 Control group, 4Ct = Ct (target gene) -Ct (β-actin).

| GO term and KEGG pathway enrichment analysis
To better understand the biological function and find critical pathways of each module, we performed GO term and KEGG pathway analysis for each module respectively.Statistically, significant enrichment was indicated when p < 0.05.GO terms are organized in a directed acyclic graph, where an edge between terms represents a parent-child relationship.KEGG is a collection of hand-drawn pathway maps representing molecular interaction and reaction networks.These pathways cover a wide range of biochemical processes that can be grouped into seven broad categories: metabolism, genetic and environmental information processing, cellular processes, bodily systems, human diseases, and drug development.We visualized the results using R packet analysis.In addition, the target proteins involved in the cellular components (CC), molecular functions (MF), biological processes (BP) and pathways were also described. 11

| Identifying pivot elements
We defined elements as the pivot elements which significantly regulated the module as those that fulfilled the following criteria 12 : i. ≥ 2 relationships between the module and the element; ii. the number significantly enriched targets per module (hypergeometric test, p < 0.05).
The interaction between pivot transcription factors (TFs) and targets were retrieved from the TRRUST, version 2 database. 13Pivot long non-coding RNA (lncRNA) target interactions were extracted from the RAID, version 2.0 database. 14I G U R E 1 Overall flowchart of the work.

| Drug repositioning analysis and confirmation of drug screen using the compound library
Comprehensive drug target information was obtained from the Drug-Bank database 15 (http://www.drugbank.ca).The chemical screening was conducted using Cherry Pick Library (Selleck Chemicals, Houston, TX, USA).Cupric oxide was purchased from Sigma-Aldrich (St Louis, MO, USA).Compounds were arrayed in 96-well plates in triplicate and diluted with medium to a final concentration of 10 μM.Each plate contains dimethylsulfoxide (DMSO) or pure water control wells.Each well in a 96-well plate contains 1000 cells.Twenty-four hours after plating, the compounds or DMSO/water control were added to the cells.Forty-eight hours after treatment, cell viability was determined using the Cell Counting Kit-8 (CCK-8) (DOJINDO, Tokyo, Japan).Each cell line was subjected to duplicate testing with each selected compounds.Each cell line was tested in duplicate with each selected compound.Results were normalized using the normal prostate epithelial cell line RWPE-1.The composite Z-score for each composite/cell line pair was calculated using 16 :

| CCK-8 assay
Cell proliferation was measured using a CCK-8 kit (AWC0114a; Abiowell Biotechnology Ltd).Cells in the logarithmic growth phase were inoculated into 96-well plates at 5 Â 10 3 cells/well and cultured for 3 days.After the corresponding incubation time, the medium was absorbed and replaced with CCK8 working solution (100 μl of medium and 10 μl of CCK8 initial solution) and incubated for 4 h in a 5% CO 2 incubator at 37 C. Absorbance at 450 nm was measured using a microplate reader.

| Transwell assay
Cell invasion and migration were detected by an invasion assay and a transwell assay.The upper well of the chambers was seeded with 2 Â 10 4 cells in serum-free medium.The lower chambers contained 500 μl of full medium.Compound/cell mixtures were inserted into the upper compartment and left to incubate at 37 C for 48 h.Subsequently, cells in the upper chamber were removed.The cells in the bottom compartment were fixed with 4% paraformaldehyde (Sangon Biotech, Shanghai, China) for 2 h, stained with crystal violet solution (Sigma-Aldrich) for 1 h, washed three times, air dried and photographed.Three fields were randomly selected in order to count the cells at 20Â magnification.

| Flow cytometry for analysis of apoptosis
The cells were centrifuged at 1500 rpm (centrifugal force: 910.8g) for 5 min and washed once with phosphate-buffered saline.The cells were suspended in 500 μl of binding buffer.Five microliters of Annexin V-FITC (KGA108; KeyGEN, China) was added and mixed.

| Statistical analysis
All measurement data are expressed as the mean ± SD.Each test was repeated independently three times.All data were analyzed using SPSS, version 26.0 (IBM Corp., Armonk, NY, USA).An unpaired Student's t-test was used to compare the data of two groups that did not have a one-to-one correspondence.One-way analysis of variance and Tukey's post-hoc tests were used to compare the data among the three groups.p < 0.05 was considered statistically significant.

| Construction of the WGCNA and identification of key modules
Seven hundred and seventy-six genes in GSE6919 (see Supporting information, Table S4) and 1432 PCa-related genes from the NCBIgene database (see Supporting information, Table S5) were selected.
After merging these datasets, 1684 genes were chosen for further analyses (see Supporting information, Table S6).A uniform threshold power of β = 4 was chosen using unscaled topological criteria (unscaled R 2 = 0.902).In total, six modules containing 62-628 genes were identified (see Supporting information, Table S7).The relevance between each module and the PCa metastasis were tested using the eigengene dendrogram (Figure 2A) and heatmap (Figure 2B).As shown in Table 1, five modules show significant relevance with the PCa metastasis in which blue, yellow and brown show negative correlation, whereas turquoise and green show positive correlation.
Eventually, the blue and turquoise modules were found to have stronger associations with PCa metastasis in contrast to other modules, which was extracted for further analysis (Figure 2C

| GO and KEGG enrichment analyses of hub genes
Analyses of GO terms and the pathway analysis of the KEGG database were performed.Here, we report the results in the blue and turquoise module (Figure 3A,D) and the results of other modules are shown in the Supporting information (Figure S1).GO analysis results revealed that the GO terms vessel morphogenesis, angiogenesis and epithelial cell proliferation were the primary functions of the turquoise module (see Supporting information, Table S8).GO terms urogenital which also lead to cancer metastasis (see Supporting information, Table S9).
The turquoise module was mainly enriched in genes encoding cancer, PCa proteoglycans and many cancer-related signaling pathways, including the PI3K-Akt, MAPK, cAMP, Wnt and Rap1, FoxO and TNF pathways (see Supporting information, Table S10).Blue modules were mainly enriched in cancer proteoglycans, T cell receptor signaling pathway and several cancer-associated genes, including PCa (see Supporting information, Table S11).
Figure 3E,F depicts the highest scoring hub genes in the blue and turquoise modules.The top five in the blue module are CELF1, CIRBP, OS9, BPTF and CBLB, whereas those of the turquoise module are TPM2, MYH11, ACTG2, AHNAK2 and FLNA (Table 2).

| Hub gene expression confirmation
Expression of these central genes was first demonstrated in prostate cell lines.We found that the expression of four of these five genes increased from normal prostate epithelial cells to PCa in situ and metastatic PCa, whereas OS9 showed the opposite trend.Four of these five genes were also continuously altered in clinical samples (Figure 4).The lowest gene expression values were used for direct data comparison.

| TF regulatory network
To better understand the regulatory network of the above five modules, the fulcrum TF is found and the TF regulatory relationship of the characteristic genes is predicted.A TF regulatory network was constructed based on 9396 TF mRNA interactions in the Trrust v2 database (https://www.grnpedia.org/trrust/).By setting p < 0.01, we obtained a regulatory network of 127 central TFs and 392 interacting TFs (see Supporting information, Table S12).For each module, we present the five TFs in the TF regulatory network (Table 3).

| lncRNA-mRNA interaction network
To further analyze the cellular functions and processes associated with the signature genes, we designed a lncRNA-mRNA interaction  S13).Five lncRNAs are shown in each module of the lncRNA-mRNA interaction network (Table 4).

| Drug repositioning and confirmation of drug screen using the compound library
We also look forward to drugs that target modular signature genes important for prostate cancer therapy.Fifty-two targets in the blue module, 49 targets in the brown module, 10 targets in the green mod- ule and 10 targets in the yellow module were obtained with p < 0.01.
Twenty-two drugs and 93 blue-green target drug forms were obtained.
Drugs containing more than three compounds are shown in Table 5 (a complete list is provided in the Supporting information, Table S14).
To confirm our drug screening strategy, we selected the top 36 drugs with the connections >3 as screen library reported by CCK8 (see Supporting information, Table S15).By setting ZscoreA and ZscoreB, we identified 17 compounds that significantly target the PCa cells (ZscoreA and ZscoreB > 0.5 in at least two cancer cell lines, 47.2% of the library) (Figure 5A,C) and four compounds that selectively target the PCa cells but not influence normal cells (ZscoreA and ZscoreB > 0.5 in the cancer cell and < 0.5 in RWPE-1,11.1% of the library) (Figure 5B,D).Of particular note is copper, which shows the most connections and selective influence for PCa cell lines.Copper refers to cupric oxide in DrugBank, showing an excellent score in our screening.Another cupric oxide-cuprous oxide is known to be a potent agent which has been well studied in cancer therapy including PCa. 17 By contrast, the anti-tumor properties of cupric oxide have yet to be reported.Thus, we selected cupric oxide for further studies and evaluated its effects across a range of doses.

| Cupric oxide shows selectively cytotoxicity to PCa cells and upregulates CIRBP
Cell viability post-exposure to cupric oxide were again assessed with the CCK-8 kit.The results demonstrate that cupric oxide was consistently selectively toxic against metastatic PCa cell lines PC-3, LNCaP, DU145 and C4-2B in a time and dose-dependent manner (Figure 5E).For cupric oxide treatment, the half maximal inhibitory concentration (IC 50 ) for the PC-3 cells was 28.25, 25.20 and 9.624 μg/ml after 24, 48 and 72 h; for the LNCaP cells, 11.39, 8.583 and 5.780 μg/ml after 24, 48 and 72 h treatment; for the DU145 cells, 9.533, 4.750 and 3.373 μg/ml after 24, 48 and 72 h treatment; and, for C4-2B cells, 10.15, 8.270 and 4.225 μg/ml after 24, 48 and 72 h treatment, respectively.Selective cytotoxicity to tumor cells is important, and one of the greatest challenges facing tumor therapy is the inability of anticancer drugs to effectively distinguish between cancer cells and normal cells.We also examined the cytotoxicity of these compounds to the normal prostate epithelial cell line RWPE-1, and the IC 50 for RWPE-1 cells was 66.42, 62.94 and 35.73 μg/ml after 24, 48 and 72 h of cupric oxide treatment, which indicated that our candidate drugs were selectively cytotoxic to tumor cells.
The effects of drug candidates on tumor cell invasion and migration were assessed in vitro by invasion and transwell assays.
We selected the appropriate concentration according to the cell viability curve, and cupric oxide at a concentration of 2.5 μg/ml significantly inhibited the invasion and migration of PCa cells Furthermore, cancer cell lines that were incubated with different cupric oxide concentrations for 48 h were subjected to propidium iodide-based apoptosis and annexin V assays.Cupric oxide was found to significantly trigger malignant cell apoptosis in a dose-dependent manner (Figure 5H).He et al. 18 have reported that PDA-PEG/Cu 2+ showed a cancercell-selective-killing effect through reverse cancer cell upregulated oncogene CIRBP, which is also a hub gene in our network.To confirm the relationship between cupric oxide treatment and CIRBP expression, we also detected CIRBP by qRT-PCR in prostate cell lines treated with cupric oxide for 48 h.Interestingly, CIRBP expression was increased in treated PCa cells but remained stable in normal cell lines, suggesting that cupric oxide can selectively kill PCa cells by regulating CIRBP (Figure 5I).

| DISCUSSION
The qRT-PCR results showed that four out of 10 hub genes were consistent with PCa progression in prostate cell lines and clinical specimens.Wu et al. 19 reported overexpression of ACTG2 in hepatocellular carcinoma cells (HCC), which may be a promising therapeutic target in HCC metastasis.CELF1 has been reported to reduce apoptosis in oral tumor cells. 20Wu et al. 21also reported that CELF1 may have significant roles in the progression of lung cancer.Cifdaloz et al. 22 confirmed CELF1 as a marker for melanoma as well as a prognostic factor.OS9 codes for a protein that is highly expressed in osteosarcomas.CIRBP is a cold-shock protein that can be regulated by many factors and is considered as a general stress-response protein. 23 has been reported that CIRBP is overexpressed in bladder cancer tissues and cell lines, promoting proliferation and migration. 24Sakurai et al. 25 considered that CIRBP promotes the development of liver hepatocytes.Both OS9 and CIRBP are interact with hypoxia-inducible factor 1 (HIF-1), a master regulator of angiogenesis. 26These results suggested that, although most of the hub genes have not previously been reported to participate in PCa progression, they were documented to be present in a broad range of other tumors, suggesting their suitability as targets for prognosis and treatment of PCa metastasis.
Interactions between TFs and their targets reveal complex mechanisms of PCa development and progression.It was found that four central TFs could target more than 10 genes, including WT1, AR, HDAC1 and YY1, among which HDAC1 could target both blue-green and yellow modules.HDAC1 is a multifunctional TF that mediates the expression and activity of proteins associated with the progression and initiation of malignancies, such as p63 and p53. 27Moreover, Burdelski et al. 28 reported that raised HDAC1 expression was associated with high Gleason grade, increased cell proliferation, early prostatespecific antigen (PSA) recurrence, elevated preoperative PSA-level, positive nodal status and advanced tumor stage.AR plays an important role in the progression of PCa.As a hormone-dependent cancer, ADT is standard pretreatment for advanced prostate cancer, but patients eventually relapse and progress to castration-resistant prostate cancer.This treatment resistance mainly lies in AR. 29 Seligson et al. 30 reported that YY1 possibly contributes towards PCa growth and may be a useful diagnostic and prognostic marker.WT1 is a zinc finger transcription factor involved in encoding growth factors and receptors such as AR, insulin-like growth factor II, IGF1 receptor and epidermal growth factor receptor. 31There were also other studies that showed WT1 may function as an oncogene for PCa. 32 recent years, in addition to protein-coding genes, non-coding genes have also attracted much attention because of their important biological functions and possible clinical relevance.We found three lncRNAs with more than 10 connections to key modules.ANCR, a more closely related lncRNA, regulates both modules.The ANCR gene is located on human chromosome 4 and mediates repression of progenitor cell differentiation. 33Previous studies have reported that ANCR can influence the progression of breast cancer. 34FENDRR is another lncRNA that shows high connections to the brown module.
There are studies showing that FENDRR plays an important role in the growth and progression of gastric cancer. 35The present study also predicted RNU1-1 as a pivot lncRNA that regulates the brown module.To date, studies of RNU1-1 related cancer are limited and the function of this small nuclear RNA remains unclear, comprising an area worth further exploration.We aim to investigate the mechanism of action of RNU1-1 at the clinical level in the future.
We introduced a valuable approach for drug discovery that combined bioinformatics methods with screening by the compound library, among which docetaxel, etoposide and other seven drugs have been widely used in PCa or cancer therapy.These pivot drugs also included many compounds belonging to the estrogen family.In addition to the drugs that have a certain effect on PCa progression, we also found drugs showing tentative promise in cancer.Copper (cupric oxide) was found to have the most drug connections.Furthermore, dabigatran etexilate, known as reversible direct thrombin inhibitor, salicylic acid, which is the most important medications needed in a basic health system, and naringenin, the predominant flavanone in grapefruit, which shows anti-inflammatory and antibacterial bioactivity, were also identified as the most promising drugs for metastatic PCa.These novel compounds warranting further research on their potential anticancer mechanisms.

| CONCLUSIONS
In conclusion, the present study has established gene co-expression networks and possible regulatory networks of TFs and lncRNAs based on the microarray data of metastatic PCa.We have also uncovered some pivot anti-metastasis drugs using the compound library for future studies into the biological mechanisms of molecular-targeted therapy for PCa metastasis.The study introduces a valuable approach for drug discovery that combines bioinformatics methods with screening by a compound library.

AUTHOR CONTRIBUTIONS
KD designed the project.JFY and FH performed the experiments.
ZWL, JFY, YL and KD analyzed the data and interpreted the data.KD wrote the first draft.FH, ZWL, JP, YL and KD revised the manuscript critically for important intellectual content.All authors approved the final version of the manuscript submitted for publication.
real-time PCR (qRT-PCR) performed on prostate cell lines including cells derived from in situ PCa: 22Rv1, and cells derived from metastatic site: RWPE-1 (AW-CNH065), DU145 (AW-CCH043), LNCaP (AW-CCH042) and PC-3 (AW-CCH111) were purchased from Abio-well Biotechnology Ltd (Changsha, China).C4-2B (BNCC341733) were purchased from BeNa Culture Collection.Cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (A11-102; Ausbian, Shanghai, China) and 1% streptomycin and penicillin at 37 C in a 5% CO 2 humidified atmosphere.These cells were treated with various doses of 0, 3, 6 and 9 μg/ml cupric oxide.Seventy-two PCa samples from Xiangya Hospital of Central South University were collected and categorized into low (45 samples, Gleason score = 6, 7) and high (27 samples, Gleason score = 8, 9, 10) grade PCa.This was approved by the Medical Ethics Committee of Xiangya Hospital of Central South University (ethics number: 202211268).The samples were then subjected to qRT-PCR analysis.The primers used for the qRT-PCR are shown in the Supporting information (Table ,D).A F I G U R E 2 Results of WGCNA.(A) The eigengene dendrogram.(B) The heatmap for the whole WGCNA.(C) The heatmap for the blue module.(D) The heatmap for the turquoise module.(E) The scatterplot for the blue module.(F) The scatterplot for the turquoise module.Coexpression modules were colored as the name assigned by WGCNA.T A B L E 1 The correlation between modules and the PCa metastasis.Module Correlation coefficient p-value Blue À0.91846 3.48 Â 10 À35 Turquoise 0.853001 3.70 Â 10 À25 Yellow À0.7319 1.76 Â 10 À15 Brown À0.61887 2.74 Â 10 À10 Green 0.594725 1.95 Â 10 À9 F I G U R E 3 Results of GO and KEGG analysis.(A) GO analysis result of the blue module.(B) KEGG analysis result of the blue module.(C) GO analysis result of the turquoise module.(D) KEGG analysis result of the turquoise module.(E) PPI network for hub genes in the blue module.(F) PPI network for hub genes in the turquoise module.Each node represents a hub gene.The highest scoring hub genes can be found in the center of the network.scatterplot of Gene Significance versus Module Membership in these two modules was plotted (Figure 2E,F).The heatmap and scatterplot of other modules are shown in the Supporting information, (Figure S1).

3 F I G U R E 4
system development indicates the genes in the turquoise module play important roles in urogenital system cancer.Moreover, GO terms ossification and osteoblast differentiation should be concerned because prostate cancer shows tendency to bone metastasis.Blue module genes were enriched in the cell migration and locomotion, T A B L E 2 Hub genes in blue and turquoise module.Hub genes expression in cell lines and clinic samples.(A) Gene expression profiles in different cell lines.(B) OS9 expression profiles in clinical samples.(C) ACTG2 expression profiles in clinical samples.(D) CIRBP expression profiles in clinical samples.(E) CELF1 expression profiles in clinical samples.In clinical samples, the target gene expression was normalized to β-actin (ΔCt) and compared with the maximum ΔCt.Data are presented as ΔΔCt.Results are presented as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

F I G U R E 5
Confirmation of drug screen and cytotoxicity of cupric oxide.(A) Z scores for each compound/cell line pair, the X and Y vectors represent two normalized Z-scores replicates.(B) Plots with ZscoreA and ZscoreB > 0.5 are extracted, represent the positive compound/cell line pair.(C) Z-score for all the compound/cell line pair.(D) Z-score for 17 compounds that significantly target the PCa cells.(E) The cell viability curve after cupric oxide treatment.(F) Invasiveness of PCa cells that underwent cupric oxide treatment was determined in the invasion assay.(G) Migration of PCa cells that underwent cupric oxide treatment was determined in the transwell assay.ImageJ (NIH, Bethesda, MD, USA) was used to analyze photos.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.(H) Cupric oxide selectively trigger PCa cell apoptosis in a dosedependent manner.(I) The expression of CIRBP was upregulated after treatment in PCa cells, whereas it remained stable in normal cell lines.An unpaired t-test was used to analyze comparisons between two groups, and comparisons among multiple groups were analyzed by one-way analysis of variance, followed by Tukey's post-hoc test (N = 3).
Pivot Nc RNA-module interactions.
T A B L E 3 Pivot TF-module interactions.tion networks were constructed with 627 lncRNA interactions, with p < 0.01 (see Supporting information, Table T A B L E 5 Pivot drug-module interactions.