Abstract The etiology of osteosarcoma (OS) is complex and not fully understood till now. This study aimed to identify the miRNAs, circRNAs, and genes (mRNAs) that are differentially expressed in OS cell lines to investigate the mechanism of circRNA-associated competing endogenous RNAs (ceRNAs) in OS. Microarray datasets reporting mRNA ([29]GSE70414), miRNA ([30]GSE70367), and circRNA changes ([31]GSE96964) in human OS cell lines were downloaded, differentially expressed (DE) RNAs were identified, and DEmRNAs were used for the annotation of Gene Ontology (GO) biological processes (BP), and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The mechanisms of DEcircRNA-mediated ceRNAs were identified in a step-by-step process. A total of 326 DEmRNAs, 45 DEmiRNAs, and 110 DEcircRNAs were identified from 3 datasets. The DEmRNAs were associated with GO BP terms, including cholesterol biosynthetic process, angiogenesis, extracellular matrix organization and KEGG pathways, including p53 signaling pathway and biosynthesis of antibiotics. The final ceRNA network consisted of 8 DEcircRNAs, including 5 pappalysin (PAPPA) 1-derived DEcircRNAs (hsa_circ_0005456, hsa_circ_0088209, hsa_circ_0002052, hsa_circ_0088214 and has_circ_0008792, all downregulated), 3 DEmiRNAs (hsa-miR-760, hsa-miR-4665-5p and hsa-miR-4539, all upregulated), and downregulated genes (including MMP13 and HMOX1). The ceRNA regulation network of OS was built, which played important roles in the pathogenesis of OS and might be of great importance in therapy. Keywords: Osteosarcoma, human osteosarcoma cell lines, competing endogenous RNA, pappalysin 1 Introduction Osteosarcoma (OS) is a primary bone malignancy with a higher incidence in both children and older individuals.^[32]1,[33]2 The overall 5-year survival rate is approximately 65% and <25% in patients with metastasis^[34]2-[35]4 and 0% in those with pulmonary metastasis.^ [36]5 The etiology of OS is complex and not fully understood till now. It has been reported that there is vast tumor heterogeneity in patients with OS.^ [37]6 The commonly used OS-derived cell lines, including U2OS, U2OS/MTX300, HOS, MG63, 143B, ZOS and ZOSM, serve vital roles in understanding the pathogenesis and therapeutic responses of OS and are used as conventional 2D models. Cell lines reflect only a part of the subtypes of tumors and cannot cover the tumor heterogeneity and full genetic spectrum of the primary OS.^[38]7-[39]9 Most anticancer drugs in cell lines are ineffective in clinical research, which restricts drug screening, personalized therapy, and the survival of patients. However, the application of conventional 2D models has invaluable roles in improving the understanding of the pathogenesis, drug screening, and therapeutic responses in tumors. Moreover, the integration of more than 1 cell line in preclinical research might suggest the typical etiology, homogeny, and common therapeutic responses. Previous research has shown that changes in the combination of chemotherapy drugs and the method of administration cannot improve the 5-year survival rate, even if the dose is increased.^ [40]10 Therefore, the discovery of biomarkers related to the treatment of OS is of great significance. The genetic and epigenetic alterations in protein-coding mRNAs and non-coding RNAs, including microRNAs (miRNAs), long non-coding RNAs (lncRNAs) and circular RNAs (circRNAs), serve crucial roles in the pathogenesis, development, drug, and therapy response, and recurrence of tumors. Compared with genes, miRNAs, lncRNAs, and circRNAs participate in and modulate tumor etiology and development indirectly through their targets or competing endogenous RNA (ceRNA) mechanism.^[41]11-[42]14 For instance, Song and Li^ [43]14 reported that circ_0001564 regulated the proliferation and apoptosis of OS cells and cell lines via sponging miR-29c-3p. Zhang et al showed that circRNA ubiquitin-associated protein 2 promoted OS progression via sponging miR-143. The characterization of circRNA-associated ceRNA in OS has been widely studied. However, the consistency and integrated analysis of circRNA-associated ceRNAs in OS cell lines have not been well reported until now. We performed this present study using microarray dataset of OS cell lines. The miRNAs, circRNAs, and genes (mRNAs) that are differentially expressed in >5 OS cell lines were identified and retained in the analysis of circRNA-associated ceRNA in OS. This study might provide novel insights into the pathogenesis or therapeutic treatments in OS concerning circRNA-associated ceRNA and maybe offer new insight for the development of OS drugs. Materials and Methods Microarray data Microarray datasets reporting the mRNA ([44]GSE70414), miRNA ([45]GSE70367) and circRNA changes ([46]GSE96964) in human OS cell lines were obtained from the National Center for Biotechnology Information Gene Expression Omnibus database ([47]https://www.ncbi.nlm.nih.gov/). [48]GSE70414 (Affymetrix Human Genome U133 Plus 2.0 Array) and [49]GSE70367 datasets ([50]GPL16384 (miRNA-3) Affymetrix Multispecies miRNA-3 Array) show the mRNA and miRNA changes in 5 OS cell lines including HOS, MG63, HY, Saos and Hu09 relative to human mesenchymal stem cells (hMSCs). [51]GSE96964 dataset ([52]GPL19978 Agilent-069978 Arraystar Human CircRNA microarray V1) presents the circRNA profiles in 7 human OS cell lines including U2OS, U2OS/MTX300, HOS, MG63, 143B, ZOS and ZOSM compared with human osteoblasts hFOB1.19. Data processing and differentially expressed RNA (DERNA) identification The GEL files derived from all 3 datasets were downloaded and processed using the Affy package ([53]GSE70414 and [54]GSE70367, version 1.52.0, [55]http://bioconductor.org/help/search/index.html?q=affy)^ [56]15 or Limma package ([57]GSE96964; version 3.10.3, [58]http://www.bioconductor.org/packages/2.9/bioc/html/limma.html).^ [59]16 Microarray datasets were preprocessed for background correction and quantile normalization. The expression of related RNAs was calculated. The average value (expression value) was calculated, while multiple probes were mapped to the same RNA (mRNA/miRNA/circRNA) and used as the expression level. Differential expression analysis of RNAs was performed using Bayesian approach methods in the Limma package.^ [60]17 DERNAs (DEmRNAs, DEmiRNAs and DEcircRNAs) in OS cell lines and controls were identified following the thresholds of the P-value <.05 and |log[2]FC (fold change)| ⩾ 1. Hierarchical clustering analyzes of DERNAs were performed using R package pheatmap (version 1.0.8, [61]https://cran.r-project.org/package=pheatmap) and presented as clustering heatmaps. Enrichment analysis The online tool Database for Annotation, Visualization and Integrated Discovery (version 6.8; [62]https://david.ncifcrf.gov/)^ [63]18 was employed to identify the Gene Ontology (GO) biological processes (BP) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways^[64]19,[65]20 that were significantly associated with DEmRNAs (P-value ⩽.05 and number of enriched genes count ⩾2) in OS cell lines and controls. DEmiRNA-mRNA regulatory network construction Due to the large number of DEmiRNAs, the difference threshold was further screened, and DEmiRNAs with |log2FC| ⩾ 2 were included in the subsequent prediction analysis. Before the construction of the miRNA-mRNA regulatory network, we predicted the targets of DEmiRNAs in miRWalk, miRanda, miRDB, miRMap, Pictar2, RNA22 and TargetScan using miRWalk2.0 ([66]http://zmf.umm.uni-heidelberg.de/apps/zmf/mirwalk2/).^ [67]21 The overlapping genes between the predictions of targets at least in 5 databases and DEmRNAs were identified as potential targets of DEmiRNAs. The DEmRNAs in the miRNA-mRNA regulatory network were used in the KEGG pathway enrichment analysis using clusterProfiler (version 2.4.3, [68]http://bioconductor.org/packages/3.2/bioc/html/clusterProfiler.html ).^ [69]22 The miRNA-mRNA regulatory network was visualized using Cytoscape (version 3.2.0, [70]http://www.cytoscape.org/).^ [71]23 CeRNA regulatory network construction The DEmiRNAs in the aforementioned miRNA-mRNA regulatory network were listed, and all DEcircRNA IDs were converted to standard names. The FASTA files of these DEmiRNAs and DEcircRNAs were extracted and processed using miRanda (version 3.3a, [72]https://omictools.com/miranda-tool)^ [73]24 to identify the circRNA-miRNA interaction pairs. Interaction pairs with interaction score ⩾150 and energy <−30 were retained and used in the construction of the ceRNA regulatory network using Cytoscape. The overlapped DEmiRNAs between the circRNA-miRNA and miRNA-mRNA regulatory networks were identified, and the associated DEcircRNAs and DEmRNAs were used in the construction of the ceRNA regulatory network using Cytoscape. The circRNA-associated genes and locations were predicted using CircInteractome ([74]https://circinteractome.nia.nih.gov/index.html). Drug-gene interaction prediction DEmRNAs in the circRNA-miRNA-mRNA regulatory network were identified and used in the selection of potential therapeutic drugs. Drug-gene associations were identified in Drug-Gene Interaction Database 3.0 (DGIdb 3.0, [75]http://www.dgidb.org/) (default parameters),^ [76]25 and the document-proven relationship pairs were screened to build the drug-gene associations. Finally, drug-gene associations were visualized using Cytoscape. Results Summary of DERNAs After normalization, a total of 326 DEmRNAs (76 upregulated and 250 downregulated), 45 DEmiRNAs (24 upregulated and 21 downregulated), and 110 DEcircRNAs (8 upregulated and 102 downregulated) were differentially expressed in the OS cell lines in [77]GSE70414, [78]GSE70367 and [79]GSE96964 datasets, respectively ([80]Figure 1A). The clustering heatmap of these DERNAs in each dataset is shown in [81]Figure 1B. Enrichment analysis for the DEmRNAs in [82]GSE70367 showed that there were 8 GO BP terms associated with upregulated DEmRNAs and downregulated DEmRNAs were enriched in 57 GO BP terms (GO:0006695 cholesterol biosynthetic process, GO:0001525 angiogenesis, and GO:0030198 extracellular matrix organization) and 4 KEGG pathways (hsa04115:p53 signaling pathway and hsa01130: biosynthesis of antibiotics) ([83]Figure 2). [84]Figure 2 shows the top 8 or 10 GO BP terms and all KEGG pathways. Figure 1. [85]Figure 1. [86]Open in a new tab Summary of the differentially expressed RNAs (DERNAs) in human OS cell lines. (A) Volcano plot of the DEmRNAs, DEmiRNAs, and DEcircRNAs in OS cell lines versus controls (hMSCs or hFOB1.19). Blue and red dots indicate DERNAs and non-significantly DERNAs, respectively. X- and Y-axis represent log2FC and–log10P-value of expression changes, respectively. (B) Clustering heatmaps of DERNAs. Green and red indicate down- and upregulation, respectively. Figure 2. [87]Figure 2. [88]Open in a new tab The pathways and biological processes (BPs) associated with differentially expressed mRNAs (DEmRNAs) in [89]GSE70414. Bubble size indicates gene number. The higher the gene number, the larger the bubble diameter. Red and green indicate high and low values of −log10(P-value), respectively. Construction of the DE miRNA-mRNA regulatory network Before the construction of DE miRNA-mRNA regulatory network, we selected 20 DEmiRNAs (9 upregulated and 11 downregulated) with |log2FC| ⩾ 2 between OS cell lines and hMSCs in [90]GSE70367 due to a large number of candidate DEmiRNAs (n = 45). Then, a total of 136 target mRNAs of these 20 DEmiRNAs, covered by DEmRNAs in [91]GSE70414, were predicted from 7 databases. Finally, the overlapping genes between target genes of 20 DEmiRNAs and DEmRNAs were acquired. The corresponding DE miRNA-mRNA regulatory network consisted of 294 miRNA-mRNA pairs (lines) ([92]Figure 3). The top 20 DEmRNAs and all DEmiRNAs in this network are presented in [93]Table 1. About 19 DEmiRNAs (8 upregulated and 11 downregulated) and 136 DEmRNAs (31 upregulated and 105 downregulated) ([94]Table S1). Figure 3. [95]Figure 3. [96]Open in a new tab Differentially expressed (DE) miRNA-mRNA regulatory network. Green arrows and red triangles represent down- and upregulated DEmiRNAs, respectively. Pink circles and green prisms represent up- and down-regulated DEmRNAs, respectively. Table 1. The differentially expressed (DE) miRNAs and top 19 DEmRNAs in the DE miRNA-mRNA regulatory network. DEmiRNAs Description Degree DEmRNAs Description Degree hsa-miR-424-5p Down 34 ZMAT3 Down 6 hsa-miR-495-3p Down 33 EPG5 Down 6 hsa-miR-182-5p Up 29 TENM4 Down 5 hsa-miR-34a-5p Down 27 SOX11 Down 5 hsa-miR-543 Down 27 PTGS1 Down 5 hsa-miR-760 Up 23 PLXNA2 Down 5 hsa-miR-1270 Down 19 GALNT10 Down 5 hsa-miR-199b-5p Down 16 ABHD2 Down 5 hsa-miR-329-3p Down 15 SLC7A11 Up 5 hsa-miR-485-3p Down 13 CEP85 Up 5 hsa-miR-346 Up 11 ADAMTSL1 Down 4 hsa-miR-758-3p Down 10 APH1B Down 4 hsa-miR-323a-3p Down 9 CBX7 Up 4 hsa-miR-486-5p Up 9 CDC42EP3 Down 4 hsa-miR-154-5p Down 7 ENPP1 Down 4 hsa-miR-1292-5p Up 6 ERCC6 Down 4 hsa-miR-4539 Up 3 GALNT1 Down 4 hsa-miR-4665-5p Up 2 IGFBP3 Down 4 hsa-miR-941 Up 1 KLF6 Down 4 SCN8A Down 4 [97]Open in a new tab Based on the DEmRNAs in the DE miRNA-mRNA regulatory network, the KEGG pathways associated with these DEmiRNAs in the network were identified. We found that hsa-miR-1292-5p was associated with the “Notch signaling pathway”; hsa-miR-1270, hsa-miR-543, and hsa-miR-758-3p were associated with “mucin-type O-glycan biosynthesis”; hsa-miR-4539 was related to “C-type lectin receptor signaling pathway” and “viral carcinogenesis”; hsa-miR-4665-5p was enriched in the “IL-17 signaling pathway” and “relaxin signaling pathway” ([98]Figure 4). Figure 4. [99]Figure 4. [100]Open in a new tab Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways associated with differentially expressed miRNAs (DEmiRNAs) in the miRNA-mRNA regulatory network. The horizontal axis indicates the miRNA number, the vertical axis indicates the entry name, the dot size represents the miRNA number, and the color represents significance. The redder the color, the greater the importance. Construction of the ceRNA network The interaction between all DEcircRNAs in [101]GSE96964 and the aforementioned 19 DEmiRNAs in the DEmiRNA-mRNA regulatory network was predicted using miRanda. About 9 DEcircRNA-miRNA pairs were identified ([102]Table 2), including 3 upregulated DEmiRNAs, 1 upregulated DEcircRNA, and 7 downregulated DEcircRNAs. All ceRNA-DERNA pairs are shown in [103]Table S2. Then, the DEcircRNAs and DEmRNAs regulated by 3 upregulated DEmiRNAs (hsa-miR-4665-5p, hsa-miR-760, and hsa-miR-4539) were identified, and the subsequent ceRNA network was comprised of 9 circRNA-miRNA and 28 miRNA-mRNA pairs ([104]Figure 5). ceRNA pairs of hsa_circ_0088737/0005456/0088209/has-miR-4665-5p/collagenase-3 (MMP13), hsa_circ_0064136/has-miR-4539/early growth response 2 (EGR2)/Cep85, and hsa_circ_0088209/0002052/0088214/0008792/has-miR-760/heme oxygenase-1 (HMOX1)/transforming growth factor (TGF)-β-stimulated clone 22 domain 3 (TSC22D3) were identified. We found the hsa_circ_0088209, hsa_circ_0088214, hsa_circ_0008792, hsa_circ_0002052, and hsa_circ_0005456 are located in chr9, and its associated gene is pappalysin 1 (PAPPA) ([105]Table 2), suggesting that these circRNAs were derived from the same gene PAPPA. hsa_circ_0032462, hsa_circ_0088737, and hsa_circ_0064136 are associated with signal-induced proliferation-associated 1-like 1 (SIPA1L1), CDKN1A interacting zinc finger protein 1 (CIZ1), and SET domain containing 5 (SETD5), respectively. Table 2. The predicted interaction pairs of differentially expressed (DE) circRNAs and miRNAs. miRNA Up/down circRNA Up/down Gene symbol Location Score Energy hsa-miR-760 Up hsa_circ_0088209 Down PAPPA chr9:118949432-118997916 449 −73.23 hsa-miR-4665-5p Up hsa_circ_0088209 Down PAPPA chr9:118949432-118997916 433 −77.48 hsa-miR-760 Up hsa_circ_0088214 Down PAPPA chr9:118969734-118989831 301 −50.44 hsa-miR-760 Up hsa_circ_0008792 Down PAPPA chr9:118969734-119033695 301 −50.44 hsa-miR-760 Up hsa_circ_0002052 Down PAPPA chr9:118969734-118997916 301 −50.44 hsa-miR-4665-5p Up hsa_circ_0005456 Down PAPPA chr9:118949432-118950495 289 −54.12 hsa-miR-4665-5p Up hsa_circ_0032462 Up SIPA1L1 chr14:72054287-72090953 284 −45.66 hsa-miR-4665-5p Up hsa_circ_0088737 Down CIZ1 chr9:130931330-130943078 282 −47.18 hsa-miR-4539 Up hsa_circ_0064136 Down SETD5 chr3:9482139-9506356 163 −32.53 [106]Open in a new tab Figure 5. [107]Figure 5. [108]Open in a new tab The ceRNA regulatory network of circRNA-miRNA-mRNA. Red triangles represent upregulated differentially expressed miRNAs (DEmiRNAs). Pink circles and green prisms represent up- and down-regulated DEmRNAs, respectively. Orange hexagon and blue squares indicate up- and down-regulated DEcircRNAs, respectively. Prediction of drug-gene association The 26 DEmRNAs in the ceRNA network were then employed in the drug prediction in DGIdb 3.0. About 6 downregulated DEmRNAs (MMP13, HMOX1, and EGR2) were targeted by 26 chemicals ([109]Figure 6), including 3-methylpyridine, sorafenib, aspirin, biliverdin, sunitinib, and tretinoin. MMP13 might be a target of 3-methylpyridine, marimastat, and other chemicals. HMOX1 was targeted by sorafenib, aspirin, biliverdin, sunitinib, formic acid, and zinc chloride. Both EGR2 and TSC22D3 were targeted by tretinoin and dexamethasone, respectively. Figure 6. [110]Figure 6. [111]Open in a new tab Drug-gene interactions. Green and gray squares represent downregulated mRNAs and chemicals, respectively. Discussion ceRNA has been a novel mechanism for exploring the interactions of RNA. For example, mRNA, miRNA and circRNA were known as the RNA interaction molecules, which have important biological significance.^ [112]26 Previous research has reported that the ceRNA was associated with the molecular mechanism of cancers, such as breast cancer,^ [113]27 glioblastoma,^ [114]28 and lung cancer.^ [115]29 Our study performed an integrated analysis of circRNA-associated ceRNAs in >5 OS cell lines. We confirmed that they have similar characteristics with miRNAs, circRNAs, and genes in OS cell lines, including HOS, MG63, HY, Saos, Hu09, U2OS and 143B. In our research, regarding the parameter settings, we set P-value <.05 and |log2FC| ⩾ 1 to screen for DERNAs, but we have obtained a large number of DEmiRNAs, so we further set |log2FC| ⩾ 2 to screen for DEmiRNAs; and in the DEmiRNA-mRNA regulatory network construction section, we predicted the targets of DEmiRNAs at least in 5 databases, compared with some studies,^ [116]30 we used relatively more databases, which can provide us with more accurate results. After rounds of selection, we identified that hsa_circ_0088737, hsa_circ_0005456, hsa_circ_0088209, hsa_circ_0002052, hsa_circ_0088214, and has_circ_0008792 played vital roles in OS pathogenesis by regulating miR-4665-5p, miR-4539, and miR-760. Hsa_circ_0088209 is located in chr9 (118949432-118997916), and its associated gene symbol is PAPPA. The interactions of hsa_circ_0088209 (downregulated), and 4 other PAPPA-derived DEcircRNAs (hsa_circ_0088214, hsa_circ_0008792, hsa_circ_0002052, and hsa_circ_0005456) with hsa-miR-4665-5p and/or hsa-miR-760 (both upregulated) were identified in OS cell lines. Our present study suggested that both hsa-miR-4665-5p and hsa-miR-760 were upregulated in OS cell lines compared with hMSCs, while all PAPPA-derived DEcircRNAs were downregulated in OS cell lines. These revealed the potential circRNA-mediated ceRNA via sponging upregulated hsa-miR-4665-5p and hsa-miR-760 in OS cell lines. Previous research has proved that hsa-miR-760 is upregulated in ovarian cancer, and its overexpression promotes the proliferation and aggressiveness of ovarian cancer cells and is associated with poor prognosis.^ [117]31 Research has found that hsa-miR-760 is upregulated in hepatocellular carcinoma, and it could serve as a potential prognostic miRNAs of hepatocellular carcinoma.^ [118]26 However, it has been proved that hsa-miR-760 is downregulated in non-small cell lung cancer, and its expression suppresses the proliferation and metastasis of cells and enhances sensitivity to TNF-related apoptosis-inducing ligand.^[119]32,[120]33 These results suggested heterogeneity in tumor pathogenesis. Our present study demonstrated that the upregulation of both hsa-miR-4665-5p and hsa-miR-760 in OS cell lines compared with hMSCs might suggest the promotion of hsa-miR-4665-5p and hsa-miR-760 overexpression in OS progression. Our study demonstrated that both hsa-miR-4665-5p and hsa-miR-760 (both upregulated) were targeted by PAPPA-derived DEcircRNAs, including hsa_circ_0088214, hsa_circ_0008792, hsa_circ_0002052, and hsa_circ_0005456 in OS cell lines. PAPPA, also known as pregnancy-associated plasma protein-A, is an oncogenic metalloproteinase that promotes the growth and invasion of human cancers, including lung cancer, ovarian cancer, and Ewing sarcoma.^[121]34-[122]37 PAPPA is lowly expressed in normal tissues and highly expressed in bone cells for bone formation during fetal development. PAPPA cleaves insulin-like growth factor (IGF) binding proteins (IGFBPs), promotes the separation of IGFs from IGF-IGFBP complex, and thus activates the IGF pathway.^ [123]36 The oncogenic roles of PAPPA are mediated by the activation of the NF-κB,^ [124]38 TGF-β,^ [125]39 PI3K/AKT, and ERK signaling pathways.^ [126]40 Our present study showed that the HMOX1, which is upregulated in various human solid tumors, was a target of hsa-miR-760 and several anticancer drugs, including sorafenib and sunitinib. These DEcircRNAs in OS might suggest that PAPPA was involved in the pathogenesis of OS and potential circRNA-associated ceRNA via sponging upregulated hsa-miR-4665-5p and hsa-miR-760 in OS cell lines. MMP13 is a predicted target of hsa-miR-4665-5p and is associated with bone formation via regulation by the transcription factor gene Cbfa1 and involvement of TGF-β. The upregulation of hsa-miR-4665-5p was associated with melatonin-mediated anticancer properties in breast and prostate cancers.^[127]41,[128]42 Hirahata et al^ [129]43 reported that a higher frequency of pulmonary metastasis was observed in cases with MMP13-positive OS than MMP13-negative OS. The expression of MMP13 in human cancers promotes cancer cell invasion and metastasis and is correlated with poor outcomes.^[130]43-[131]46 These studies showed that the expression of MMP13 was crucial in cancer invasion and metastasis, while the inhibition or deletion of MMP13 suppressed the development and metastasis of tumors.^[132]44-[133]46 Our present study suggested that the expression of MMP13 was downregulated in all OS cell lines compared with hMSCs. MMP13 was only targeted by upregulated hsa-miR-4665-5p in the ceRNA network. The upregulated hsa-miR-4665-5p and downregulated MMP13 in OS might suggest different tumorigeneses or antitumor properties of hsa-miR-4665-5p/MMP13 axis. Moreover, the MMP13-marimastat interaction might show potential therapeutic value of marimastat use in OS treatment. Conclusions In conclusion, we found that 5 downregulated PAPPA-derived DEcircRNAs, including hsa_circ_0088209, hsa_circ_0088214, hsa_circ_0008792, hsa_circ_0002052, and hsa_circ_0005456, may play crucial roles in the pathogenesis and treatment of OS. Moreover, the ceRNA regulation network of OS was built, which provided novel insights into the pathogenesis of OS and might be of great importance in the therapy of OS. Supplemental Material sj-docx-1-evb-10.1177_11769343211041379 – Supplemental material for Identification of Conserved Pappalysin 1-Derived Circular RNA-Mediated Competing Endogenous RNA in Osteosarcoma [134]Click here for additional data file.^ (28.9KB, docx) Supplemental material, sj-docx-1-evb-10.1177_11769343211041379 for Identification of Conserved Pappalysin 1-Derived Circular RNA-Mediated Competing Endogenous RNA in Osteosarcoma by Guang-Fu Ming, Bo-Hua Gao and Peng Chen in Evolutionary Bioinformatics Footnotes Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. Declaration of conflicting interests: The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Author Contributions: GM and PC were responsible for the conception and design of the research, and drafting the manuscript. BG performed the data acquisition, data analysis and interpretation. GM, PC, and BG participated in the design of the study and performed the statistical analysis. All authors read and approved the final manuscript. Authors contributed equally to this work. Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. ORCID iD: Peng Chen Inline graphic [135]https://orcid.org/0000-0002-0466-1303 Supplemental Material: Supplemental material for this article is available online. References