OS cells (MG63 and 143B) and hFOB1.19 cells were bought from the cell bank of the Chinese academy of sciences (Shanghai, China). Cells were authenticated by short tandem repeat analysis and passaged for fewer than 6 months before experiments. All cell lines were tested to be negative for mycoplasma contamination and were cultured in an atmosphere of 5% CO₂ and 90% relative humidity. MG63 and 143B cells were cultured in DMEM (Gibco, United States of America) supplement with 10% (vol/vol) fetal bovine serum (FBS, Gibco, United States of America), 100 μg/mL penicillin and streptomycin (Gibco, United States of America) at 37°C. The hFOB1.19 cells were cultured in DMEM/F12 (Gibco, United States of America) supplement with 10% (vol/vol) FBS (Gibco, United States of America), 1% (vol/vol) nonessential amino acid solution (Gibco, United States of America), 0.3 mg/mL G418 and 100 μg/mL penicillin and streptomycin (Gibco, United States of America) at 33.5°C.

Tumor specimens of OS patients and control non-tumor specimens of polydactyly patients were obtained from Shanghai Children’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. All samples were collected with the donor being informed completely and with their consent. The procedures were approved by the Institutional Ethical Review Board of the Shanghai Children’s Hospital (2021R095-E01, 2020R159-E02 and 2020R029-E03).

Lenti-GFP containing an IDO1 overexpression sequence and its negative control sequence (NC) were purchased from Shanghai HuaGene Biotech Co., Ltd. Cells were transfected with lenti-GFP-IDO1 or lenti-GFP-NC. Polyclonal cells with green fluorescent protein signals were purified for further experiments using a fluorescence-activated cell sorting flow cytometer.

Western blots were performed as described (Yang et al., 2019; Yang et al., 2022) using the following antibodies against IDO1 (ab211017, Abcam), Flag (ab205606, Abcam), CD63 (ab217345, Abcam), CD81 (ab109201, Abcam), RUNX2 (ab23981, Abcam), Calnexin (#2433, CST) and GAPDH (AF2823, Beyotime). The whole protein was extracted by RIPA Lysis Buffer (Beyotime Biotechnology, China), and the concentration was detected by a BCA protein assay kit (Beyotime Biotechnology, China). Cell lysates were kept on ice for 30 min and centrifuged at 16,000 g for 3–5 min at 4°C. Supernatants were collected and boiled in SDS loading buffer, and the same amounts of protein were separated by 10% SDS-PAGE and blotted onto polyvinylidene fluoride membranes (Millipore, United States of America). Bands were visualized using chemiluminescence.

Immunohistochemistry analysis was performed as described previously (Yang et al., 2019; Du et al., 2021). Samples were fixed in 4% paraformaldehyde (PFA), paraffin-embedded and cut into 4–6 μm slides. All slides were dehydrated in gradient ethanol. Following the antigen retrieval done in 10 mM citrate buffer pH 6.0 (Na3H6H5O7, Beyotime, China), blocking was done using PBS with 10% normal goat serum (Sigma-Aldrich, United States of America) for 40 min. Slides were incubated overnight at 4°C with primary antibodies of IDO1 (ab211017, Abcam) or Ki67 (sc-23900, Santa Cruz). Slides were then incubated with goat anti-rabbit secondary antibody (G-21234, Thermo Fisher Scientific) or goat anti-mouse secondary antibody (G-21040, Thermo Fisher Scientific), stained using 3,3-diaminobenzidine solution and counterstained with hematoxylin.

The concentrations of Trp and Kyn in cell supernatant were detected with Elisa kits following the instructions of the manufacturer (MM-85226O2 and MM-926267O2, Meimian Biotech). The optical density was measured with a microplate reader (Thermo Fisher Scientific) at a wavelength of 450 nm. Concentrations were calculated by reference to the standard curve.

Exosomes were isolated from cell culture supernatant with the exoEasy Maxi Kit (Qiagen, No. 76046, United States) according to the manufacturer’s instructions. Total RNA was isolated from exosomes following the protocol of the exoRNeasy SerumKit (Qiagen, No. 77064, United States). The RNA yield and exosome size range were analyzed on an Agilent 2,100 Bioanalyzer. Extracted RNA was used to prepare the miRNA sequencing library with QIAseq miRNA Library Kit (Qiagen, No. 331505, United States). The sequencing was performed on an Illumina sequencer at the Cloudseq Biotechnology Co., Ltd. (Shanghai, China).

Raw data were generated after sequencing, image analysis, base calling, and quality filtering on the Illumina sequencer. Firstly, Q30 was used to perform quality control. The adaptor sequences were trimmed and the adaptor-trimmed reads ( ≥ 15 nt) were left by cutadapt software (v1.9.3). Then, trimmed reads from all samples were pooled, and miRDeep2 software (v2.0.0.5) was used to predict novel miRNAs. The trimmed reads were aligned to the merged pre-miRNA databases (known pre-miRNA from miRBase v22 plus the newly predicted pre-miRNAs) using Novoalign software (v3.02.12) with at most one mismatch. The numbers of mature miRNA mapped tags were defined as the raw expression levels of that miRNA. The read counts were normalized by the edgeR approach. Differentially expressed miRNAs between two samples were filtered through Fold change. Differentially expressed miRNA between two groups was filtered by Fold Change and p-value. miRNA targets were performed based on Miranda and Targetscan, miRNA-targets networks were plotted by Cytoscape software (v2.8.0), and the GO analysis was performed based on the top 10 differentially expressed miRNA target genes.

Total RNA was isolated using Trizol (Invitrogen, CA, United States) following the manufacturer’s guidelines. Reverse transcription of mRNA was performed as previously described (Yang et al., 2019). Total RNAs of miRNA were reverse-transcribed to cDNA by using an All-in-one™ miRNA First-Strand cDNA Synthesis Kit (GeneCopoeia, Rockville, MD, United States) and Oligo (dT) priming method (Prime Script TMRT Reagent Kit; TaKaRa, Shiga, Japan). QPCR (Applied Biosystems 7,500, Foster City, CA, United States) was performed using power SYBR^® Green PCR Master Mix (Applied Biosystems). The expression levels of miRNAs and mRNAs were normalized to the levels of miR-103a-3p and β-Actin, respectively. qPCR analysis was performed in quadruplicate for each sample with specific primers: IDO1 Forward: 5′-ATG​CAA​GAA​CGG​GAC​ACT-3′, Reverse: 5′-GCC​TTT​CCA​GCC​AGA​CAA-3’; β-Actin Forward: 5′-GGG​AAA​TCG​TGC​GTG​AC-3′, Reverse: 5′-GGA​AGG​AAG​GCT​GGA​AGA​G-3’; hsa-miR-23a-3p Reverse primer: 5′-GTC​GTA​TCC​AGT​GCA​GGG​TCC​GAG​GTA​TTC​GCA​CTG​GAT​ACG​AC GG AAAT-3′, Forward: 5′-GCG​ATC​ACA​TTG​CCA​GGG-3′, Reverse: 5′- AGTGCAGGGTCCGAGGT ATT-3’; hsa-miR-103a-3p Reverse primer: 5′- GTC​GTA​TCC​AGT​GCA​GGG​TCC​GAG GTATTCG CAC​TGG​ATA​CGA​CTC​ATA​G-3′, Forward: 5′-GCG​AGC​AGC​ATT​GTA​CAG​GG-3′, Reverse: 5′-AGT​GCA​GGG​TCC​GAG​GTA​T T-3’.

All graphing and statistical analyses were performed using GraphPad Prism (version 9). All microscopy images were randomly taken from different areas. The positive count of each immunostaining photo ranged from 0 to 300 and was calculated as the percentage of weakly stained cells plus the percentage of moderately stained cells multiplied by two plus the percentage of strongly stained cells multiplied by three. Details of sample size n), statistical test, and p-value applied for each experiment were indicated in the figure legends. Results are presented as median ±SD or SEM. The statistical significance of differences between the two groups was assessed by Student’s t-test. One-way analysis of variance (ANOVA) was used to compare several treatment groups with one control group. p values <0.05 were considered statistically significant.

Among 9 cases of OS tissues, 66.7% (6/9) showed moderately or strongly positive immunohistochemical staining signal of IDO1, and 33.3% (3/9) weakly positive (Figure 1A). The IDO1 positive count of each immunostaining photo was calculated, and the relationship between IDO1 positive count and clinical stage of the patient was analysed. It was found that in II and III/IV grade OS patients the expression of IDO1 was significantly higher than that in I grade (Figure 1B). , We then performed Ki67 immunostaining on tumor tissue and obtained the Ki67 positive count of patients, since a higher Ki67 immunostaining is associated with a higher grade of malignancies and poorer prognosis of patients (Mardanpour et al., 2016; Kim et al., 2021). Ki67, like IDO1, is differently-expressed in OS tissues (Figure 1C), and the Ki67 index was positively correlated with the clinical grade of OS patients (Figures 1D,E). Based on the above results, it was deduced that the expression of IDO1 were involved in the malignancy of OS.

To investigate the correlation between IDO1 expression and prognostic related laboratory test indexes, patients were divided into IDO1-high or IDO1-low group according to the average positive count of IDO1 in their tumor tissues. Additionally, laboratory test indexes at diagnosis of these OS patients were collected (Table 1), including serum alkaline phosphatase (ALP) (Hao et al., 2017; Bao et al., 2019), lactate dehydrogenase (LDH) (Fu et al., 2018), white blood cell (WBC) count (Viola et al., 2021) and C-reactive protein (CRP) (Jettoo et al., 2019), as they are prognostic factors for OS reported in multiple studies. We found that between IDO1-high and IDO1-low groups, serum ALP and LDH concentrations showed significant difference, while WBC and CRP level did not (Figures 2A–D). Moreover, serum ALP and LDH concentrations but not WBC and CRP were significantly positively correlated with IDO1 expression (Figures 2E–H). These results indicated that IDO1 expression was positively correlated with the poor prognosis of patients with OS.

Dysregulation of exosomal miRNAs has been reported to be associated with the malignancy of OS (Vickers et al., 2011; Redis et al., 2012). To investigate whether IDO1 participated in OS malignant progression through miRNA, we selected non-metastatic OS cell line (MG63), metastatic OS cell line (143B) and non-tumorigenic immortalized osteoblastic hFOB1.19 cell line as the research object (Ren et al., 2015; Jerez et al., 2019). IDO1-overexpressing, stably transfected cell lines of these cells (MG63 OE, 143B OE and hFOB1.19 OE) were constructed and validated as IDO1 expression were significantly higher in these cells than that of their corresponding control cells (MG63 CON, 143B CON and hFOB1.19 CON) (Figure 3A). IDO1 activity was represented by the concentration of Kyn and the ratio of Kyn to Trp in cell supernatant. We found that the concentration of Trp in the supernatant of MG63 OE, 143B OE and hFOB1.19 OE cells was not significantly different from that of corresponding control cells (Figure 3B). However, the concentration of Kyn and the ratio of Kyn/Trp of these OE cells were significantly higher (Figures 3C,D).

Exosomes were isolated from conditioned culture media of these cells. The diameter sizes and concentrations of isolated exosomes were analyzed by Zetaview nanoparticle tracking analyzer. The peak diameter of the most of exosomes is 125 nm (Figure 3E), which is consistent with the report in the literature (Jerez et al., 2019). These exosomes also expressed characteristic exosomal markers (Figure 3F), indicating successful extraction. Next-generation sequencing was conducted to identify miRNAs enriched in exosomes in our panel of cell lines. A total of 1,244 miRNAs were sequenced and identified (Figures 3G,H), in which fold changes>2.0 and p-value < 0.05 were used as the screening threshold for differently expressed miRNAs. The expression profile of miRNAs was distinguishable in hierarchical cluster analysis (Figure 3I). Taken together, the detection of cell-specific miRNAs in these cell lines indicates considerable heterogeneity in the presence of miRNAs in exosomes, not only consistent with the distinct biological phenotypes of each cell type but also affected by the expression of IDO1.

96 differently expressed miRNAs (DE miRNAs) were identified in 3 cell groups by comparing MG63 OE, 143B OE and hFOB1.19 OE with their control cells. We found that hsa-miR-23a-3p was only significantly upregulated in both MG63 and 143B OS cells but not non-tumorigenic immortalized osteoblastic hFOB1.19 cells (Figure 4A). To confirm this result, we further screened for top DE miRNAs in the above three groups of cells. Nine top significantly upregulated and downregulated miRNAs were detected from MG63 cells (Table 2), 7 from 143B cells (Table 3), and 5 from hFoB1.19 cells (Table 4). We found that compared with non-tumorigenic immortalized osteoblastic hFOB1.19 cells, MG63 and 143B OS cells did not share the same downregulated exosomal miRNAs, and only has-miR-23a-3p was significantly upregulated in both OS cells (Figures 4B,C).

We also verified the above results in clinical samples. The mRNA and protein expressions of IDO1 in OS tumor tissues were significantly higher than those in normal bone tissues excised from control multi fingered patients (Figure 4D). It was found that the expression of hsa-miR-23a-3p was higher in tumor tissues than that in control tissues (Figure 4E). MiRNAs in cancer cells act as post-transcriptional regulators of their mRNA targets to control the expression of genes involved in tumor progression (Nugent, 2014). To understand the biological functions of has-miR-23a-3p, we performed biological processes and cell components analysis by GO enrichment analysis using the protein interaction network database (string-db.com) (Szklarczyk et al., 2019) (Table 5). As the data suggests, has-miR-23a-3p targets include genes in fatty acid metabolic pathway and muscle system process which have been report to be a typical sign during cancer development (Hoy et al., 2021) (Supplementary Material). Therefore, upregulated hsa-miR-23a-3p in IDO1 overexpressed OS cell lines may signal other cells through exosomes to shape the tumor microenvironment towards tumor survival. In summary, we believe that hsa-miR-23a-3p may be the key factor for OS cells with high IDO1 level to maintain its malignancy.

Since upregulation of hsa-miR-23a-3p is shared between the two OE cell lines we examined and thus believed to have similar role in these cells, we hypothesized that other differently expressed miRNAs which are not shared between 143B OE and MG63 OE cells may contribute to their unique properties different from each other. We then investigated the function of these miRNAs using the same method and databases as mentioned above (Figure 5A; Supplementary Material). We found that hsa-miR-21-5p, hsa-miR-9-5p, hsa-miR-4488 and hsa-miR-186-5p were upregulated in 143B OE cells but not in MG63 OE cells. Many target genes of above-mentioned miRNAs showed no significant GO enrichment, while targets of hsa-miR-21-5p yielded interesting results (Supplementary Material). Among the target genes, Signal transducer and activator of transcription 3 (STAT3), transcription factor forkhead box O1 (FOXO1), Toll-like receptor 4 (TLR4) and transcription factor forkhead box O3 (FOXO3), which are involved in immune regulation and tumor cell growth and metabolism, are the most closely connected (Figure 5B). GO analysis of targets hsa-miR-21-5p has shown enrichment in “positive regulation of cytokines production involved in inflammatory response” as well as “interleukin-1 beta production” and could involve in immunosuppressive action (Supplementary Material).

Similarly, the target genes of hsa-miR-4492, hsa-miR-144-3p, and hsa-miR-148a-5p, upregulated miRNAs in MG63 OE cells but not in 143B OE cells were also analyzed (Figure 5A; Supplementary Material). We discovered that target genes of hsa-miR-4492 were mostly grouped towards immune related functions, such as basophil differentiation, dendritic cell differentiation, and positive regulation of T-helper 2 cell cytokine production (Supplementary Material), suggesting its immune suppression role in tumor microenvironment, since it could downregulate said genes in host cells. These results indicate that both cell lines have the ability of employing different strategy to evade host immune response with their own unique miRNAs, and to assure tumor survival.