Activated Hedgehog-Gli1 signal pathway contributed to chondrosarcoma
To explore the potential activated signal pathways in chondrosarcoma, we processed the GSE30835 data from GEO datasets, which included two growth plates, four normal cartilages, and four chondrosarcomas, but excluded the Ollier disease17. After quality control, we performed a PCA analysis and found that the normal cartilage and chondrosarcoma have different developmental trajectories from the growth plate (Fig. 1A). Finally, we found 326 up-regulated genes in normal cartilage and 116 up-regulated genes in chondrosarcoma respectively (log2FC>0.5, p.value< 0.01, Fig.1B). Of these, the Hh signal pathway is in off sate in normal cartilage, while “Regulation of IGF Transport And Uptake By IGFBPs” signal pathway is activated in chondrosarcoma (Figure 1C, D). In the previous study, IGFBP3 is regulated by GLI signaling in the progression to malignant chondrosarcoma18. We also found that GLI1 could act directly on IGFBP3 to regulate the downstream signal in one chip-seq analysis (GSE100936) (Fig. E). GLI1 is the final and most crucial transcript factor in the hedgehog signal pathway19. Then, we collected 10 normal cartilages, 26 low-grade (Ⅰ-Ⅱ), 33 high-grade (Ⅲ) , as well as 25 dedifferentiated chondrosarcomas to detect the expression of GLI1, which has no significant difference between normal cartilage and low-grade chondrosarcoma, but obviously increased in high-grade and dedifferentiated chondrosarcoma in our IHC staining (Fig. F, G). According to the results, we deeply deem that Hedgehog-GLI1 pathway is on state and further activates the downstream signals in chondrosarcoma.
Hedgehog-GLI1 signal mediated the RNAP III signal pathway and tRNA synthesis to regulate cell cycle and death receptor binding in chondrosarcoma
To further determine the mechanism of the Hedgehog-GLI1 signal in malignant chondrosarcoma, we reanalyzed a gene expression profiling of 17 fresh frozen chondrosarcoma biopsies20. After normalizing the data, the expression of GLI1 was dichotomized using the median as a cutoff to define ‘‘high’’ or ‘‘low’’ expression categories for each sample (Fig. 2A). Furthermore, we compared the transcriptional profiles of GLI1high to GLI1low samples. A total of 542 differentially expressed transcripts (p< 0.05; |log2FC | > 0.25) were identified, and 117 of them were elevated in GLI1high chondrosarcomas. We analyzed the BioPlanet signal pathways using the GLI1high expressed transcripts to gain insight into the functional relevance. RNA polymerase III (RNAP III) transcription-relevant pathways were enriched, which are attributed to the Hedgehog-GLI1 pathway activation (Fig. 2B). In addition, we processed chondrocyte-specific transcription factors that could directly convert human amnion cells into chondrosarcoma to demonstrate the RNAP III transcription signal pathway in chondrosarcoma progression. Enrichment plot GSEA indicated that RNAP III transcription signal pathway enriched in chondrosarcoma (P<0.01, NES=2.48) (Fig. 2C). The heatmap representation of this subset of genes was shown in Fig. 2D.
The abundant tRNAs in eukaryote cells that are synthesized by RNAP III have key functional roles21. Therefore, we analyzed a modulated expression of endogenous tRNA fragments (tRFs) in six normal cartilage and fourteen chondrosarcomas. We detected 478 up-regulated and 321 down-regulated tRFs in chondrosarcoma (|log2FC|>1.0, p< 0.01, Fig.2E). Then we used the Genomic tRNA Database (GtRNAdb) to predict the tRNA gene. Five tRNAs were detected from 21 enriched tRFs in GtRNAdb of human species. Another dataset called tsRFuntion (tsRFun) was applied to predict the functions of tRFs target genes by GO enrichment analysis. The enrichment analysis highlighted “cell cycle” and “death receptor binding” were the most significant biological processes (BP) and molecular functions (MF) respectively (Fig. 2F, G). Taking all the results together, the Hedgehog-GLI1 signal mediated the RNAP III signal pathway and tRNA synthesis to regulate the cell cycle and death receptor binding in chondrosarcoma.
GLI1 inhibitor (GANT61) suppressed chondrosarcoma by inhibiting the RNAP III signal pathway and tRNA-Gly-CCC synthesis in vivo
As the Hedgehog-GLI1 signal pathway is activated in chondrosarcoma, we established a vivo subcutaneous xenograft model by SW1353 cells. After five days of xenograft, mice were treated with GANT61 intravenous injection per three days for one month (Fig. 3A). Tumor volumes were calculated in the GANT61 group were decreased by 53±8.9%, as well as had a lower proliferation rate than the control group (Fig. 3B, C, P<0.01). The tumor weights were also decreased by 51.2±11.0% (Fig. 3D). The significant increase in regions of necrosis elicited by GANT61 was assessed by microscopic examination of H&E-staining is shown in Fig.3E, in which necrosis rate already accounted for approximately 70% when compared to 5% of the control group (Fig.3F). To further test the functions of GANT61 on Hh signal pathway, we assessed the key members IHH, PTCH1, SMO, GLI1, GLI2 in the hedgehog pathway by using qRT-PCR. We found that the PTCH1 and GLI1 expression was significantly decreased, while the GLI2 expression was increased, which means GANT61 works by reducing the GLI1 rather than GLI2.
To validate whether the anti-tumor effects of GANT61 were reflected at gene expression levels, we performed microarray-based transcriptomic profiling of chondrosarcoma subcutaneous tumor tissues. It was found that massive genes in vivo xenograft models were differentially expressed (|log2FC|≥1; p<0.01) following treatment with GANT61. There were 1772 genes with up-regulated and 2479 genes with down-regulated mRNA expression levels (supplement table 2). Following the analysis of the microarray data, the hub genes of the RNAP III signal pathway were downregulated (Fig. 3H). To test whether the down-regulated RNAP III signal pathway affects the tRNA synthesis or not, the five upregulated candidate tRNAs in Fig.2 E were further checked by qRT-PCR, Interestingly, only tRNA-Gly-CCC was significantly declined in our study (Fig. 3I). SNAPC1 in the RNAP III signal pathway is the bridge genes between GANT61 and tRNA-Gly-CCC, and the target sequence of tRNA-Gly-CCC was described in the Fig. 3J. All these findings imply that GANT61 tends to inhibit chondrosarcoma tumor formation by suppressing the RNAP III signal pathway regulated tRNA-Gly-CCC expression in vivo.
GANT61 blocks the chondrosarcoma cells in G2/M cell cycle phase
Just as described in Fig.2F, the cell cycle signal pathway is activated in chondrosarcoma through the endogenous tRNA fragments enrichment analysis. We also acquired a consistent result in transcriptomic expression levels, which was that cell cycle pathway is significantly elevated in chondrosarcoma when compared to the amnion cells, fetal and adult chondrocytes by GSEA analysis (P<0.01, NES=1.88) (Fig.4A, B). We carried out a pathway enrichment analysis by using the down-regulated genes, and found that the cell cycle signal pathway was successfully suppressed by the GANT61 in our microarray data (Fig. 4C). We further confirmed that GANT61 treatment caused marked accumulation of a G2/M population in the isolated tumor cells of the animal models with flow cytometry (Fig.4D). G2/M arrest population was 30.41±4.67% in those animals exposed to GANT61, while approximately 2.5±0.53% of the control group (Fig. 4E). The further mechanism was that GANT61 inhibited the molecular changes of CDK1 and Cyclin 2A that associated with G2/M checkpoint (Fig.4F).
GANT61 induced chondrosarcoma cell death through apoptosis and autophagy
Given that the “Death receptor binding” function was observed in the endogenous tRNA fragments enrichment analysis of chondrosarcoma (Fig. 2G). Then we put our efforts into which cell death manners can be induced by GANT61. Firstly, apoptosis rate distribution was determined by flow cytometry and increased by 16.89 ± 2.475% in GANT61 treated group (Fig. 4D, F). And we found this is a caspase3 mediated apoptosis (Fig. 4G). Subsequently, we applied GSEA analysis to observe the autophagy level in chondrosarcoma. The interesting fact is that the autophagy in chondrosarcoma is at a lower level than in amnion cells, fetal and adult chondrocytes (p<0.01, NES=-2.03) (Fig. 5A, B). In our transcriptomic expression data, the enrichment degrees of the cellular components also showed that differentially encoded product proteins were mainly distributed on vacuole, pre-autophagysomal structure membrane, autophagic vacuole by using the up-regulated genes in GANT61 group (Fig. 5C). Meanwhile, GSEA analysis showed differentially up-regulated genes were related to the activation of autophagosome (p<0.01, NES=1.15) and Macroautophagy (P<0.01, NES=1.14) (Fig. 5D, E). The KEGG signaling pathways analysis results also expounded that their encoding proteins were most involved in the autophagy signaling pathway (Fig. 6F).
To verify whether GANT61 is involved in autophagy, TEM was used to detect the ultrastructures of autophagy. We found some representative features in the group of GANT61 treatment, such as autophagic vacuole and autophagosome (Fig. 5F). Endogenous LC3-II visualized by immunofluorescence is recognized as an increased autophagosomal formation in living cells. We observed that GANT61 significantly increased endogenous LC3-II puncta in the isolated tumor cells of the xenograft models (Fig. 5G). Detecting LC3-II and p62 combined with bafilomycin A1 (Baf A1, inhibitor of lysosomal degradation) by immunoblotting has become a reliable method for monitoring autophagy and autophagy-related processes. Our results show that GANT61 increased the LC3-II localization in the cytoplasm and resulted in the process of transferring LC3-I to LC3-II with or without Baf A1 (Fig. 5H). A link between autophagy and cell death has been demonstrated using pharmacological (e.g., Rapamycin (Rapa) is a potent inducer of autophagy, chloroquine (CQ) is a lysosomotropic agent that has been suggested to inhibit autophagy)22. The autophagy inhibitor CQ (10 µM) significantly enhanced the viability of cells, but the autophagy inducer Rapa (100 nM) further depresses the viability in response to GANT61 (Fig. 6A). Overall, the enrichment analysis of the up-regulated genes, along with the TEM, immunofluorescence technique and immunoblotting results indicated that autophagy is silent in chondrosarcoma, but GANT61 can arouse the autophagy to against this malignant sarcoma.
GANT61 activated ULK1 and MAPK Signaling pathway to regulate autophagy
To declare the underlying mechanism of autophagy, GSEA analysis was also dedicated to performing leading edge analysis of the differential genes which was the most critical one. Genes in the subset list showed that ULK1 (unc-51 like autophagy activating kinase 1) had the highest impact on the biological process of autophagy (Fig. 6B). We uncovered that MAPK signal pathway is involved in the regulation of autophagy in our previous study. Subsequently, we applied GSEA analysis to observe the MAPK signal level in chondrosarcoma and found that the MAPK signal is not in an activated state in chondrosarcoma (p<0.01, NES=-1.61) (Fig. 6C, D). However, the autophagy and MAPK signal pathway were activated simultaneously when exposed to GANT61 in the KEGG signal pathway analysis (Fig. 6E). Also, there is a potential correlation between autophagy and MAPK signaling pathway (Fig. 6F).