HMGCS2 is differentially expressed between tumor tissues and normal tissues
The UCSC database contains normalized pan-cancer datasets from TCGA, TARGET and GTEx, from which we obtained 19,131 normal tissues and 60,499 tumor samples. We calculated differences in ENSG00000134240 (HMGCS2) expression between normal and tumor samples in each tumor type and observed that HMGCS2 was significantly downregulated in 15 tumors in the TCGA database (Fig. 1A), including LUAD, COAD, COADREAD, BRCA, ESCA, KIRP, KIPAN, HNSC, KIRC, LUSC, LIHC, THCA, READ, KICH and CHOL (P < 0.05). After combining the TARGET and GTEx databases, we additionally observed that HMGCS2 was up-regulated in ESCA, STES, PRAD, OV, PAAD, UCS (Fig. 1B). In addition, we queried the HPA database for the expression of HMGCS2 in different cell lines, and the results also showed that it is lowly expressed in most cells, except RT4 (human bladder transitional cell papilloma cells) and OE19 (human esophageal cancer cells) (Fig. 1C). Further, the cellular localization of HMGCS2 was mainly located in mitochondria (Fig. 1D), which was also confirmed by immunofluorescence staining with different antibodies. Overall, HMGCS2 was significantly downregulated in most tumors relative to normal tissues, and its expression levels were associated with possible mitochondrial function.
Pan-cancer analysis of the prognostic value of HMGCS2
To further study the significance of HMGCS2 expression for clinical patients, we analyzed the relationship between HMGCS2 expression in pan-cancer patients with overall survival (OS), primary tumor status (T), lymph node metastasis (N), distant metastasis (M), tumor grade (G) and tumor stage (Stage). Univariate Cox analysis showed that HMGCS2 expression was associated with overall survival in STES, GBMLGG, KIRC, KIPAN, LIHC and LGG (P < 0.05) (Fig. 2A). In terms of primary tumor status (T), HMGCS2 was associated with BRCA, ESCA, STES, KIPAN, KIRC, LIHC and CHOL (P < 0.05) (Fig. 2B). Lymph node metastasis (N) was associated with BRCA, STES, KIPAN, STAD, HNSC, KIRC and THCA (P < 0.05) (Fig. 2C). Distant metastases (M) aspects were COADREAD, KIPAN and KIRC (Fig. 2D). In terms of tumor grade (G), it was associated with STES, KIPAN, HNSC, KIRC, and LIHC (Fig. 2E). Tumors in which HMGCS2 expression correlated with tumor stage were BRCA, KIRC, LIHC, THCA, SKCM, BLCA, and KICH (Fig. 2F). Overall, HMGCS2 expression appeared to be strongly associated with both KIRC and LIHC tumors. Therefore, we further analyzed the prognostic role of HMGCS2 in KIRC (Fig. 2G, J) and LIHC (Fig. 2H, K) by K-M survival analysis and ROC curves. The results showed that HMGCS2 had good survival prediction ability in both tumors. In addition, we additionally analyzed whether HMGCS2 expression was associated with progression-free interval (PFI) using univariate Cox. The results showed that HMGCS2 expression correlated with PFI in KIRC, but not LIHC (Supplementary Fig. 1A). K-M survival analysis and ROC curves also showed that HMGCS2 had some working power in the prediction of PFI (Supplementary Fig. 1B, C).
Correlation of HMGCS2 expression with genomic heterogeneity and tumor stemness in various cancer types
Cancer stem cells (CSCs) are unique cell populations within tumors with self-renewal and tumorigenic capabilities. CSCs play a key role in cancer progression, metastasis and recurrence, as well as tumor resistance to cytotoxic therapy (22). We calculated the EREG.EXPss tumor stemness score by mRNA signature for each tumor obtained from a previous study (23), integrating the stemness index and gene expression data of the samples. We observed significant correlations in 12 tumors, of which there were significant positive correlations in 9 tumors, such as: LUAD, ESCA, STES, SARC, KIRP, KIPAN, STAD, PRAD, KIRC, in 3 tumors There was a significant negative correlation between the two groups, such as: BRCA, LIHC, TGCT (P < 0.05) (Fig. 3A).
Tumor mutational burden (TMB) reflects the amount of cancer mutations. Mutations are processed into neoantigens and presented to T cells by major histocompatibility complex (MHC) proteins, so TMB is a potential Immunotherapy Biomarker (24). MSI status can predict cancer response/resistance to certain chemotherapies (25). We observed that HMGCS2 was significantly negatively correlated with TMB in 7 tumors, such as: LUAD, COAD, COADREAD, KIPAN, KIRC, THCA and PAAD. TGCT was significantly positively correlated with MSI, while COAD, COADREAD and BRCA were significantly negatively correlated with MSI (Fig. 3B, C) (P < 0.05). Ploidy, Homologous recombination deficiency (HRD) and Loss of Heterozygosity (LOH) are important factors in tumor development and drug resistance (26–28). Therefore, we further explored the relationship between HMGCS2 and Ploidy, HRD, and LOH in various tumors. The results showed that HMGCS2 positively correlated with Ploidy in LAML, UVM, BLCA, and negatively correlated with BRCA, KIPAN, PRAD, KIRC, LIHC. In terms of HRD, HMGCS2 was positively correlated with UVM, while significantly negatively correlated in 9 tumors, such as: GBM, LUAD, BRCA, ESCA, KIPAN, PRAD, KIRC, PAAD, SKCM. Finally, we observed that HMGCS2 was significantly associated with LOH in 9 tumors, positively correlated with KIPAN, KICH, and negatively correlated with LUAD, BRCA, ESCA, PRAD, KIRC, SKCM, BLCA. In conclusion, HMGCS2 expression is strongly associated with tumor genomic heterogeneity and tumor stemness, especially in KIRC.
Correlation between HMGCS2 Expression and the TME in Different Types of Cancers
The role of the tumor microenvironment (TME) in tumorigenesis and development has been demonstrated, and therapeutic strategies targeting the tumor microenvironment (TME) have become a promising approach for cancer therapy (29, 30). The stromal score is a method to evaluate the stromal cell content in tumor tissue, which often reflects the malignancy of the tumor (31). HMGCS2 expression was observed to be significantly correlated with stromal score in 13 cancer types. Among them, 4 were significantly positively correlated, such as LGG, BRCA, KIPAN, TGCT, and 9 were significantly negatively correlated, such as LUAD, ESCA, COAD, COADREAD, KIRC, THYM, BLCA, OV, PAAD (Fig. 4A). In addition, the immune score, a method to assess the tumor immune microenvironment, has been shown to outperform the AJCC/UICC TNM classification in colorectal cancer (32, 33). Furthermore, HMGCS2 expression was found to be negatively correlated with immune score in LUAD, COAD, COADREAD, KIRC, LIHC, BLCA, THCA, OV, TGCT, while positively correlated in LGG and MESO (Fig. 4B). The stromal score and immune score were used to represent the presence of stromal and immune cells. The two scores are added to obtain the estimate score, which can be used to estimate tumor purity. Finally, we observed that the expression of this gene was significantly correlated with immune infiltration in 12 cancer types, of which 3 were significantly positively correlated, such as LGG, BRCA, KIPAN; 9 were significantly negatively correlated, such as LUAD, COAD, COADREAD, KIRC, LIHC, BLCA, THCA, OV and PAAD (Fig. 4C). Notably, KIRC and LIHC were cancer types strongly associated with HMGCS2 in multiple aspects in our results, including gene expression, prognosis prediction, and immune infiltration (Fig. 1, 2, 3, 4).
HMGCS2 is associated with immune cell infiltration in pan-cancer
To further explore the correlation of HMGCS2 with immune cell infiltration in pan-cancer, we comprehensively analyzed the correlation of HMGCS2 with immune cell infiltration in pan-cancer in different databases. Results from the Timer database showed that immune cell infiltration in 25 cancers was associated with HMGCS2 expression, of which LIHC, TGCT and BLCA were closely related cancers (Fig. 5A). EPIC and quanTIseq database results show that HMGCS2 is associated with immune cell infiltration in most cancers. It was closely related to TGCT, KIRC in EPIC, and LIHC, KIRC and BLCA in quanTIseq (Fig. 5B, C). Studies have shown that HMGCS2 has a tumor suppressor effect in HCC (3, 7). Therefore, we will further explore the role of HMGCS2 in KIRC. Notably, HMGCS2 appears to alter the immune status of KIRC via macrophages (red box).
60 immune checkpoint genes and 150 immune pathway genes were used to further analyze the immune correlates of HMGCS2 (15, 20). The results showed that BLCA was the most associated with HMGCS2 among many cancers, followed by TGCT. Notably, in KIRC, 33 out of 60 immune checkpoint genes were associated with HMGCS2 expression (Supplementary Fig. 2A), and 75 out of 150 marker genes of immune pathways were associated with HMGCS2 (Supplementary Fig. 2B). Drug discovery and programmed death in cancer are closely related to m6A modification (34, 35). Therefore, we extracted 44 gene expression profiles of three types of RNA modifications (m1A, m5C, m6A) and performed correlation analysis with HMGCS2. The results showed that 34 of the 44 genes were associated with HMGCS2 expression in KIRC, the most relevant cancer type (Supplementary Fig. 2C).
HMGCS2 expression is closely related to renal proximal tubules
To further understand the role of HMGCS2 in kidney development, we analyzed two single-cell sequencing datasets from adult kidneys. All cells were divided into four major populations, including immune, endothelium, fetal nephron and stroma (Fig. 6A). Then we displayed HMGCS2 as a marker gene, and the results showed that HMGCS2 was specifically expressed in pelvic epithethelium-distal UB, proximal UB and proximal tubule (Fig. 6B). All cells in another dataset were divided into thirteen cell populations, including kidney capillary endothelial cell, glomerular visceral epithelial cell, epithelial cell of promximal tubule, etc. (Fig. 6C). In addition, the expression of HMGCS2 was shown to be correlated with kidney connecting tubule epithelial cell and renal principal cell (Fig. 6D).
HMGCS2 promotes the expression of tumor suppressor-related proteins
We overexpressed exogenous GFP-HMGCS2 in the human renal clear cell carcinoma cell line OSRC2, and the results of fluorescence and western blotting confirmed the successful overexpression of GFP-HMGCS2 (Fig. 7A). We then examined the main members of the cyclin family and showed that overexpression of HMGCS2 suppressed the expression of CDK1 but no other members of the family. We also additionally examined markers associated with apoptosis (Fig. 7B). The results showed that in HMGCS2-overexpressing OSRC2 cells, the expressions of p65, caspase3 and p53 were increased, while the expressions of Bcl2 and Erk were decreased (Fig. 7C).
Overexpression of HMGCS2 inhibits proliferation but not apoptosis of OSRC2 cells in vitro
To further explore the effect of HMGCS2 on the proliferation of OSRC2 cell lines, we performed Cell Viability Assay and Colony formation assays to characterize the function of HMGCS2. The experimental results of CCK8 showed that the proliferation of OSRC2 was inhibited at 24h, 48h, and 72h after HMGCS2 overexpression (Fig. 8A). Consistently, cloning agar experiments also showed that overexpression of HMGCS2 suppressed colony number formation (Fig. 8B). In addition, the overexpression of HMGCS2 promoted the apoptosis of OSRC2 cells (Fig. 8C). These results suggest that HMGCS2 may function as a tumor suppressor by inhibiting cell cycle progression and promoting cell apoptosis.