Abstract

Background

   Emerging studies reveals that 3‐hydroxy‐3‐methylglutaryl‐CoA synthase 1
   (HMGCS1) plays vital oncogenic roles in a broad spectrum of human
   cancers, but there is no pan‐cancer evidence on the relationship
   between HMGCS1 and various tumor types.

Aim

   To explore the potential role of HMGCS1 across various tumor types
   based on big clinical data.

Methods

   We conducted a pan‐cancer analysis across more than 30 tumor types,
   based on the most comprehensive database available, including TCGA,
   GSCA, clinical proteomic tumor analysis consortium, Kaplan–Meier
   Plotter dataset, GEPIA2, TIMER2, STRING, and GDSC dataset.

Results

   HMGCS1 was highly expressed and negatively correlated with the
   prognosis in most cancer types. The infiltration levels of cancer
   associated fibroblast and CD8+ T‐cell were closely associated with
   HMGCS1 expression. Amplification was the most common genetic alteration
   of HMGCS1 in different cancers, while the frequency of mutation was
   low. Besides, ACAT2 and MVD were closely correlated and bind to HMGCS1.
   Pathway enrichment analysis indicated that HMGCS1 was actively involved
   in steroid biosynthesis. Moreover, high HMGCS1 expression could reduce
   the sensitivity to most drugs in the GDSC dataset.

Conclusions

   Our study revealed the potential oncogenic role of HMGCS1 in cancers.

   Keywords: cancer, drug sensitivity, genetic alteration, HMGCS1,
   mevalonate pathway, prognosis

1. INTRODUCTION

   The etiology of cancer is complicated, with both genetic and epigenetic
   alterations involved. It is significant to conduct a pan‐cancer
   expression analysis of particular gene of interest and assess its
   correlation with diagnosis, treatment, prognosis, and potential
   molecular mechanisms.[30] ^1 , [31]^2

   The mevalonate (MVA) pathway is known as synthesizing sterols and
   isoprenoids, playing a key role in multiple cellular processes.[32] ^3
   , [33]^4 Recent studies highlighted the importance of MVA metabolism in
   cancer, especially the intermediates involved in this pathway.[34] ^3 ,
   [35]^5 , [36]^6 , [37]^7 For instance, cholesterol, an essential
   component of most cellular membranes, provides building blocks for
   highly proliferative cancer cells. Cholesterol also contributes to drug
   resistance and affect immune surveillance in cancer.[38] ^8 , [39]^9
   Besides, farnesyl‐diphosphate and geranylgeranyl diphosphate are
   requisites for protein isoprenylation, which are vital for
   intracellular signaling. For example, after isoprenylation, small
   GTPases such as RAS are transferred to cell membrane and exert
   oncogenic function of transforming cells.[40] ^10

   The 3‐hydroxy‐3‐methylglutaryl‐CoA reductase (HMGCR) is the
   rate‐limiting enzyme in the MVA pathway. However, other enzymes in the
   pathway have also been showed to work as controlling points.[41] ^7 ,
   [42]^11 The 3‐hydroxy‐3‐methylglutaryl‐CoA synthase 1 (HMGCS1)
   catalyzes the synthesis of HMG‐CoA, which is the substrate for
   HMGCR.[43] ^12 , [44]^13 Our previous study has proved that HMGCS1
   promoted cell growth in leukemia (data not shown) and
   BRAF^V600E‐positive human cancers.[45] ^14 Emerging studies showed that
   HMGCS1 contributed to the progression of a variety of cancers,
   including gastric, colon, and cervical cancer.[46] ^15 , [47]^16 ,
   [48]^17 Further, the high HMGCS1 expression level was correlated with
   poor prognosis in breast cancer.[49] ^18 However, there is no
   pan‐cancer evidence on the relationship between HMGCS1 and various
   tumor types based on big clinical data.

   In this study, we conducted a pan‐cancer analysis across more than 30
   tumor types, based on the most comprehensive database available. Our
   study revealed the potential oncogenic role of HMGCS1 in cancers.

2. MATERIALS AND METHODS

2.1. Gene expression analysis

   By using the “search‐expression” module of ATACdb, a comprehensive
   human chromatin accessibility database
   ([50]http://www.licpathway.net/ATACdb/), the expressions of HMGCS1 in
   primary/nonprimary cells and normal tissues were displayed from NCBI
   GEO/SRA database. In the “Expression” module of GSCA (Gene Set Cancer
   Analysis, [51]http://bioinfo.life.hust.edu.cn/GSCA/), we analyzed the
   HMGCS1 expression data from The Cancer Genome Atlas (TCGA) program,
   which was obtained from both tumor and adjacent normal tissues of
   various tumors or tumor subtypes. For some tumors without or with only
   a few normal tissues, we analyzed the difference of HMGCS1 expression
   between the tumor and the corresponding normal tissues by GEPIA2 (Gene
   Expression Profiling Interactive Analysis, version 2,
   [52]http://gepia2.cancer-pku.cn/). The “expression analysis‐box plots”
   module was used. In addition, we presented the HMGCS1 expression in
   every pathological stage (stage I–stage IV) of TCGA tumors by violin
   plots using the “pathological stage plot” module of GEPIA2.

   To compare the level of HMGCS1 protein in tumors and normal tissues, we
   utilized the UALCAN portal
   ([53]http://ualcan.path.uab.edu/analysis-prot.html) to analyze the data
   from the clinical proteomic tumor analysis consortium (CPTAC) dataset.

2.2. Survival prognosis analysis

   The correlation between HMGCS1 expression and prognosis of tumors was
   presented by Kaplan–Meier plotter ([54]http://www.kmplot.com/analysis/)
   and the GEPIA2 database. We calculated the log‐rank p value and hazard
   ratio (HR) with 95% confidence intervals.

2.3. Immune infiltration analysis

   We utilized the “immune‐gene” of TIMER2
   ([55]http://timer.cistrome.org/) to evaluate the correlation of HMGCS1
   expression with immune infiltration (IF) across all tumors in TCGA
   project. The “Gene_Corr” module of TIMER2 was employed to investigate
   the association between HMGCS1 expression and IF in different cancer
   types and in adjacent normal tissues of the TCGA project. Correlations
   between HMGCS1 expression and IF were analyzed using the
   purity‐adjusted partial Spearman's correlation test. The data was
   displayed by heatmaps and scatter plots. CIBERSORT, CIBERSORT‐ABS,
   EPIC, MCPCOUNTER, QUANTISEQ, TIDE, and XCELL algorithms were used. The
   infiltrated immune cells include cancer‐associated fibroblasts (CAF)
   and CD8+ T cells.

2.4. Genetic alteration analysis

   We utilized the cBioPortal data ([56]https://www.cbioportal.org/) to
   look at HMGCS1 alterations and clinical outcomes in the TCGA pipeline.
   We obtained the alteration frequency, mutation type, and copy number
   alteration of the HMGCS1 gene via the “cancer types summary” module of
   cBioPortal. The mutated site information of HMGCS1 was displayed either
   in the schematic diagram of the protein structure or in the 3D
   structure using the “mutations” module. The “comparison” module was
   used to acquire the mRNA and protein expression differences between the
   TCGA cancer cases with or without HMGCS1 genetic alteration. The
   heatmaps of HMGCS1 mRNA expression were generated by the TBtools. The
   Venn diagram was generated by the Xiantao web
   ([57]https://www.xiantao.love). The “Comparison” module of cBioPortal
   was employed to explore the correlation between HMGCS1 genetic
   alterations and prognoses, including overall survival (OS), progress
   free survival (PFS), and disease‐free survival (DFS). We built the
   bivariate linear regression models to study the relationship between
   HMGCS1 mutations and expression levels.

2.5. HMGCS1‐related gene enrichment analysis

   We used the STRING ([58]https://string-db.org/) database to analyze the
   protein–protein interactions (PPI) of HMGCS1. Using the STRING website,
   we obtained a list of top‐50 HMGCS1‐binding proteins by setting the
   following parameters: minimum required interaction score (“low
   confidence [0.150]”), meaning of network edges (“evidence”), max number
   of interactors to show (“no more than 50 interactors” in first shell),
   and active interaction sources (“experiments”). GEPIA2 was utilized to
   obtain the top 50 HMGCS1‐correlated targets. A pairwise gene Pearson
   correlation analysis was performed. A heatmap was generated by the
   TIMER2. To analyze the intersection between the HMGCS1‐binding and
   interacted genes, the Venn diagram viewer of the Xiantao web
   ([59]https://www.xiantao.love) was utilized. Besides, Kyoto
   encyclopedia of genes and genomes (KEGG) and Gene Ontology (GO)
   enrichment analysis were analyzed using the “KEGG/GO enrichment
   visualization” tool of XianTao.

2.6. Drug sensitivity analysis

   We performed an unbiased comparison of the drug screening data using
   GSCA, which integrated over 10 000 genomic data in 33 cancer types from
   TCGA and over 750 small molecule drugs from genomics of drug
   sensitivity in cancer (GDSC) and cancer therapeutics response portal.

3. RESULTS

3.1. . The expression of HMGCS1 in cancers

   First, we studied the expression of HMGCS1 in various
   primary/nonprimary cells and normal tissues based on ATACdb database.
   HMGCS1 was expressed abundantly in normal tissues. The top five normal
   tissues with highest HMGCS1 mRNA expression were vagina, liver, skin,
   salivary gland, and esophagus (Figure [60]1A). In primary cells,
   keratinocyte cells had the highest HMGCS1 expression, followed by
   neurosphere, foreskin keratinocyte, mucoepidermal cells (MEC) and
   fibroblast of lung cells. However, the expression was low in various
   blood cells, including PBMC, CD14+ monocyte, T‐cell, CD4+ alpha‐beta T
   cell, natural killer cell, B cell, and common myeloid progenitor CD34+
   cell (Figure [61]1B). Interestingly, we found that HMGCS1 was obviously
   over‐expressed in tumor cell lines, especially in hematological
   malignancies and sarcoma (Figure [62]1C).

FIGURE 1.

   FIGURE 1
   [63]Open in a new tab

   The expression of HMGCS1 in cancers and normal tissues. (A–C) Heatmap
   of HMGCS1 expressions in normal tissues (A), primary cells (B), and
   nonprimary cell lines (C). (D) Differential HMGCS1 mRNA expressions
   between normal tissues and various cancers in the GSCA database.
   *p < .05; **p < .01; ***p < .001. (E) The levels of HMGCS1 mRNA in
   tumor tissues and corresponding normal tissues. *p < .01. (F) HMGCS1
   total protein levels of normal and primary tissues of breast cancer,
   ovarian cancer, colon cancer, clear cell RCC and UCEC

   Next, we analyzed the HMGCS1 expression of normal and tumor tissues
   from GSCA approach across different cancer types in TCGA
   (Figure [64]1D). Compared to adjacent control tissues, HMGCS1 was
   over‐expressed in the tumor tissues of breast invasive carcinoma
   (BRCA), cervical squamous cell carcinoma (CSCA) and endocervical
   adenocarcinoma (CESC), colon adenocarcinoma (COAD), esophageal
   carcinoma (ESCA), liver hepatocellular carcinoma (LIHC), lung squamous
   cell carcinoma and uterine corpus endometrial carcinoma (UCEC), and
   lower in tumor tissues of cholangiocarcinoma (CHOL), glioblastoma
   multiforme, kidney renal clear cell carcinoma (KIRC), kidney renal
   papillary cell carcinoma (KIRP), lung adenocarcinoma (LUAD),
   pheochromocytoma and paraganglioma (PCPG), prostate adenocarcinoma, and
   thyroid carcinoma (THCA) (Figure [65]1D). Furthermore, we used the data
   of normal tissue from the GTEx dataset as controls to compare the
   HMGCS1 expression between the tumor and normal tissues. We found that
   HMGCS1 level was significant higher in adrenocortical carcinoma (ACC),
   diffuse large B‐cell lymphoma, pancreatic adenocarcinoma, rectum
   adenocarcinoma (READ), and thymoma (THYM) tumor tissues
   (Figure [66]1E), and lower in skin cutaneous melanoma and acute myeloid
   leukemia (AML) tumor tissues, compared with normal tissues
   (Figure [67]S1A). However, for other tumors such as LHIC (LGG), ovarian
   serous cystadenocarcinoma (OV), testicular germ cell tumors (TGCT), and
   uterine carcinosarcoma (UCS), the expression of HMGCS1 was not
   significantly different from their controls (Figure [68]S1B). Using the
   pathological stage plot module of GEPIA2, we observed significantly
   negative correlation between HMGCS1 expression and pathological stages
   of KIRC, OV, and THCA (Figure [69]S1C).

   In addition, CPTAC dataset was utilized to analyze the HMGCS1 protein
   level difference among cancers. Compared to normal tissues, HMGCS1
   protein was higher in the primary tissues of breast cancer, colon
   cancer, and UCEC, but was lower in clear cell RCC (also known as KIRC)
   (Figure [70]1F).

3.2. The correlation of HMGCS1 expression with prognosis in cancers

   We utilized the Kaplan–Meier lotter dataset and GEPIA2 to investigate
   the correlation between HMGCS1 level and OS/DFS. In Kaplan–Meier
   Plotter dataset, we found that higher level of HMGCS1 was linked to
   shorter OS in multiple tumor types, including BCA (bladder carcinoma),
   CSCA, HNSC, PDAC, SARC, UCEC, BRCA, CESC, CHOL, KICH, LUAD, MESO, and
   THCA (Figure [71]2A, C). However, highly expressed HMGCS1 was
   correlated with better prognosis in a few types of tumors including
   LGG, READ, and KIRC ([72]Figure S2A). Similarly, HMGCS1 expression was
   negatively correlated with the DFS of ACC, CESC, KIRP, PDAC, PCPG,
   SARC, STAD, CHOL, KICH, and UVM (Figure [73]2B, D), but positively
   correlated with the DFS of KIRC and LGG [74](Figure S2B). Collectively,
   HMGCS1 expression was negatively correlated with both OS and DFS in
   CSCA, PDAC, SARC, ACC, CESC, CHOL, and KICH, while positive
   associations were found in LGG and KIRC. In conclusion, high expression
   of HMGCS1 was correlated with poor prognosis in most tumor types, but
   good prognosis for a small number of tumors.

FIGURE 2.

   FIGURE 2
   [75]Open in a new tab

   The correlation between the HMGCS1 expression and the prognosis in
   cancers. (A, B) The correlation between HMGCS1 expression level and
   overall survival (OS) and disease‐free survival (DFS) in different
   tumors. (C‐D) The correlation between HMGCS1 expression level and OS
   (C) and DFS (D) in the GEPIA2 databases. HR, hazard ratio

3.3. The correlation between HMGCS1 expression and IF in cancers

   HMGCS1 expression was negatively correlated with the IF of CD8+ T‐cells
   in most cancers, such as CESE, HNSC, STAD, and THYM ([76]Figure S3A,
   B). In addition, HMGCS1 expression was positively correlated with the
   estimated infiltration value (EIV) of CAF in the TCGA tumors of BRCA,
   BRCA−basal, BRCA−LumA, SARC, TGCT, THYM, MESO, and OV (Figures [77]3A,
   B). Collectively, we found that the expression of HMGCS1 was correlated
   with the IF in most cancers.

FIGURE 3.

   FIGURE 3
   [78]Open in a new tab

   The correlation between HMGCS1 expression and the immune infiltration
   (IF) in cancers. (A) Using four algorithms, we estimated the
   correlation between HMGCS1 expressions and IF level of four cancer
   associated fibroblast cell types across TCGA cancer types.
   Statistically significant positive/negative correlation is represented
   by the solid red/blue square, respectively. (B) Some examples of
   scatter plots show the correlation between HMGCS1 expression and tumor
   purity (first column), and the infiltration level of cancer associated
   fibroblast cell estimated by EPIC (second column), MCPCOUNTER (third
   column), XCELL (fourth column) and TIDE (fifth column), in BRCA (first
   row), SARC (second row), TGCT (third row), and THYM (fourth row),
   respectively

3.4. Genetic alterations of HMGCS1 in cancers

   We analyzed the genetic alterations of HMGCS1 in various tumors based
   on the TCGA dataset. We found that gene amplification (copy gain) was
   far more common than other forms of genetic alterations in HMGCS1, such
   as mutation, deep deletion (copy loss) and fusion (Figure [79]4A).
   Sarcoma had the highest HMGCS1 genetic alteration frequency, among
   which amplification was the primary type. It was noteworthy that the
   proportions of mutation were significant in UMEC (100%) and UEC
   (Figure [80]4A). Figure [81]4B displayed the types, distribution, and
   case numbers of HMGCS1 genetic alterations. Missense mutation was the
   most common type. The R277Q site mutation in the C‐terminal was
   detected in one case of UCEC, while R277W mutation was detected in one
   case of COAD. The R277Q/W site mutation resulted in a truncated form of
   HMGCS1 protein, impairing its function. Figure [82]4C showed the 3D
   structures of HMGCS1 with or without R277 mutation.

FIGURE 4.

   FIGURE 4
   [83]Open in a new tab

   Genetic alterations of HMGCS1 in cancers. (A) The frequencies of
   various HMGCS1 genetic alterations in cancers (from cBioportal
   database). (B) Distribution of mutations on HMGCS1 protein sequence.
   (C) The 3D structures of HMGCS1 protein. Left panel: wildtype; right
   panel: R277 mutation. (D, E) The mRNA (D) and protein (E) expression
   levels of HMGCS1 in gene unaltered and altered groups of different
   cancers. (F) Number of genes with altered mRNA level in HMGCS1
   unaltered and altered groups. (G) Gene expression levels in HMGCS1
   unaltered and altered group (left panel). The top 20 differentially
   expressed genes between unaltered and altered group (right panel). (H)
   Overall, disease free, disease‐specific and progress free survival in
   cancers with or without HMGCS1 alterations, presented by Kaplan–Meier
   plots

   In addition, we aimed to study the relationship between HMGCS1
   mutations and expression levels. We found that the mRNA levels were
   different between gene‐altered and gene‐unaltered group
   (Figure [84]4D). However, there is no significant difference of protein
   level between HMGCS1 altered and unaltered group (Figure [85]4E). The
   mRNA expression of total 618 genes, consisting of 167 up‐regulated
   genes in the gene‐altered group and 451 up‐regulated genes in the
   gene‐unaltered group, were associated with HMGCS1 mutation
   (Figure [86]4F). Left panel of Figure [87]4G displayed the difference
   of mRNA expression between altered and unaltered group. The top 20
   differentially expressed genes were also shown (Figure [88]4G, right
   panel). Furthermore, we compared the OS, DFS, disease‐specific
   survival, and PFS between patients with and without genetic
   alterations. Kaplan–Meier analysis revealed that there was no
   significant difference UCEC (Figure [89]4H). Together, these data
   suggested that HMGCS1 mutation was correlated with alterations of
   transcriptome and proteome in tumors. However, R277 mutation had no
   significant effect on the prognosis of UCEC.

3.5. Enrichment analysis of HMGCS1‐related partners

   Next, we tried to find the HMGCS1‐binding proteins as well as the
   HMGCS1 expression‐correlated genes, and further conducted pathway
   enrichment analyses. First, we searched the PPI networks using STRING
   online tools, and screened out fifty HMGCS1‐binding proteins with the
   closest relationship (Figure [90]5A). In addition, GEPIA2 was utilized
   to obtain the top 50 HMGCS1 expression correlated genes. The top five
   genes were HMGCR, LDL1 (low‐density lipoprotein 1), ACAT2 (acetyl‐CoA
   acetyltransferase 2), NSDHL (NAD[P] dependent steroid
   dehydrogenase‐like), and LDLR1 (low‐density lipoprotein receptor 1)
   (Figure [91]5B), all of which were positively correlated with HMGCS1 in
   most cancer types (Figure [92]5C). An intersection analysis of top 50
   members from STRING and GEPIA2 showed two common members, namely ACAT2
   and MVD (Figure [93]5D). Further, we utilized these 98 members to
   conduct KEGG and GO enrichment analyses to establish the functions of
   HMGCS1. GO analysis showed that the top three enriched terms in the
   biological process ontology were steroid metabolic process, steroid
   biosynthetic process, and sterol metabolic process. For the cellular
   component ontology, the top three enriched terms included mitochondrial
   matrix, mitochondrial inner membrane, and intrinsic component of
   endoplasmic reticulum membrane. For the molecular function ontology,
   the top three terms were oxidoreductase activity, with incorporation or
   reduction of molecular oxygen; iron ion binding; and monooxygenase
   activity (Figure [94]6E). KEGG pathway analysis disclosed involvement
   of HMGCS1 in steroid biosynthesis, Cushing syndrome, cortisol synthesis
   and secretion, and so forth (Figure [95]6F).

FIGURE 5.

   FIGURE 5
   [96]Open in a new tab

   Enrichment analysis of HMGCS1‐related partners. (A) Protein–protein
   interaction network analysis based on STRING database. (B) The
   expression correlations between HMGCS1 and HMGCR, LDL1, ACAT2, NSDHL,
   and LDLR1, respectively. (C) The correlations in Figure [97]5B are
   further displayed with more details in different cancer types. (D) The
   intersection of HMGCS1‐binding and correlated genes. (E‐F) Functional
   enrichment analyses of HMGCS1 by GO (E) and KEGG (F). The top 10 terms
   in each analysis were listed. Color of dot represent adjusted p‐value.
   Size of dot represent number of genes of each term

FIGURE 6.

   FIGURE 6
   [98]Open in a new tab

   Analysis of the correlation between drug sensitivity and HMGCS1
   expression using GDSC database. Figure summarized the top 30 drugs
   whose effects showed most significant associations with HMGCS1
   expression in pan‐cancer. Only three drugs (GSK690693, KIN001‐102, and
   navitoclax) presented with negative correlation

3.6. The prediction of the correlation between HMGCS1 expression and drug
sensitivity

   For drug sensitivity analysis, we conducted drug response predictions
   using the GDSC database. We assessed the correlation between HMGCS1
   expression and IC50 values of molecules, and found that high HMGCS1
   expression could reduce the drug sensitivity of most drugs.
   Interestingly, high expression of HMGCS1 enhanced the sensitivity to
   three drugs or small molecules, which included GSK690693, KIN001‐102,
   and navitoclax (Figure [99]6). These results indicated that HMGCS1 was
   a potential therapeutic target in cancer.

4. DISCUSSION AND CONCLUSION

   Mounting evidence demonstrated a close link between HMGCS1 and the
   initiation and progression of various tumors.[100] ^14 , [101]^15 ,
   [102]^16 , [103]^17 To elucidate the general role of HMGCS1 in
   different tumors, we conducted a pan‐cancer analysis of HMGCS1 in 33
   different cancers based on the data from the most comprehensive
   database available, such as TCGA and CPTAC databases. We analyzed the
   gene expressions and genetic alterations of HMGCS1 in various caner
   types, and explored the impact of HMGCS1 on disease survival, IF and
   drug sensitivity.

   HMGCS1 was highly expressed in multiple tumor types, such as tumors
   originated from alimentary tract, breast, and uterus (Figure [104]1D).
   Correlations of increased HMGCS1 expression with a shortened survival
   were seen in most types of cancer (Figure [105]2). In BRCA, both HMGCS1
   mRNA and protein level were significantly elevated (Figure [106]1D, F).
   High HMGCS1 expression was closely correlated with poor OS as well as
   PFS (Figure [107]2). In addition, there was a positive correlation
   between the HMGCS1 expression and the EIV of CAF (Figure [108]3).
   Consistently, Walsh et al identified HMGCS1 as a specific marker of
   cancer stem cell, whose expression strikingly indicated aggressive
   feature of breast cancer.[109] ^6 Like BRCA, HMGCS1 expression was
   elevated and negatively associated with OS and PFS in tumors of uterus,
   including CESC and UCEC (Figure [110]1D and [111]2). In CESC, there was
   a significantly negative correlation between the HMGCS1 expression and
   IF of CD8+ T‐cells ([112]Figure S3). Further study is warranted to
   understand more about the function of HMGCS1 in CESC and UCEC. Taken
   together, these data highlighted the oncogenic role of HMGCS1 in
   cancer.

   Our data showed that HMGCS1 had increased mRNA expression in most
   alimentary tract related cancers, including COAD, ESCA, and READ, with
   a few exceptions like STAD (Figure [113]1D, E). For colon cancer, the
   HMGCS1 protein level was also elevated (Figure [114]1F). It is well
   known that RAS and RAF mutations are the most common genetic alteration
   in colon cancer (approximately 39%–53%), which are closely related to
   poor PFS, OS and nonresponse.[115] ^19 Previously, we reported that
   HMGCS1 was a synthetical partner of BRAF^V600E‐expressing colon and
   melanoma cancer cells.[116] ^14 Sheng et al recently reported that
   HMGCS1 was overexpressed in colon cancer cells and tissues, promoting
   proliferation, migration, and invasion. Simultaneous targeting HMGCS1
   could enhance the sensitivity of MEK inhibitor.[117] ^20 Thus,
   targeting HMGCS1 could overcome RAF/RAS mutation mediated drug
   resistance, which might be a promising therapeutic strategy for
   patients with colon cancer. Notably, I‐Han et al. reported higher
   HMGCS1 expression in STAD tumor cells (n = 261) and metastatic lymph
   nodes (n = 26) compared to adjacent corresponding control tissues
   (n = 261) using q‐PCR method.[118] ^15 However, in our study, the
   expression difference between control tissues (n = 32) and tumor
   tissues (n = 375) was not significant in STAD (p = 0.345), according to
   the TCGA data analysis (Figure [119]1D). The reason for the
   inconsistency may be the relatively small size of control sample in
   TCGA dataset. Interestingly, I‐Han et al. also discovered the
   nonmetabolic function of HMGCS1. Cytosolic HMGCS1 could translocate to
   nucleus and further bind and activated the promoters of downstream
   targets, contributing to the progression of gastric cancer.[120] ^15 In
   conclusion, HMGCS1 is a promising target in alimentary tract related
   cancers.

   It is also noteworthy that HMGCS1 may exert oncogenic function in
   tumors without overexpression. For instance, HMGCS1 was not
   overexpressed in prostate cancer cells but was in stroma. Through an
   autocrine/paracrine mechanism, the stromal cells with increased HMGCS1
   expression could promote the growth of prostate cancer cells.[121] ^21
   Another example is AML. We have found that HMGCS1 was not significantly
   overexpressed in newly diagnosed AML patients. However, HMGCS1
   expression was raised in refractory and relapsed AML, which played a
   vital role in reducing chemotherapy sensitivity (data not shown). More
   studies are required to understand the role of HMGCS1 in some tumors
   without elevated expression.

   HMGCS1 was negatively correlated with both OS and DFS in LGG and KIRC.
   Currently, there is no report concerning about how HMGCS1 affects LGG.
   In KIRC, differential expression of HMGCS1 was reported by Huang et
   al..[122] ^22 By analyzing data from TCGA, they identified HMGCS1 as
   one of good prognosis biomarkers.[123] ^22 However, further research is
   needed to elucidate its role in KIRC.

   Immune cells niched in the tumor microenvironment are closely linked to
   tumorigenesis.[124] ^23 CAF, differentiated from stroma cells, are
   potent tumor‐promoting cells. Their proliferation was positively
   correlated with HMGCS1 expression in multiple cancer types. However,
   the proliferations of tumor‐antagonizing cells, such as CD8+ T‐cells,
   were overall negatively correlated with HMGCS1 level (Figure [125]3 and
   [126]S2). These results indicated the role of HMGCS1 in modulating
   tumor‐infiltrating immune cells, which has not been reported yet and
   require further exploration.

   As shown in Figure [127]4, amplification was the most frequent genetic
   alteration in tumors, which was consistent with previous data, which
   showed that HMGCS1 was overexpressed and closely related with prognosis
   in multiple tumor types including COAD and UCEC. Mutations were seen in
   several cancer types, though at a quite low frequency. We found R277Q
   mutation in one case of UCEC, and R277W mutation in one case of COAD.
   By further analysis in UCEC, we found that this missense mutation did
   not show prominent impact on OS, DFS, and PFS. By consulting the
   uniport website and literature,[128] ^24 , [129]^25 we identified the
   active sites of HMGCS1, which included Glu95, Cys129, and His264, and
   the binding sites for Coenzyme A, which included Asn167, Ser221,
   Lys269, and Lys273. Currently, no mutation on these sites was found in
   tumors. The impact of R277Q mutation on prognosis was minimal, as the
   mutation seemed not to decrease the activity of HMGCS1. However, HMGCS1
   mutation could cause obvious alterations in the expressions of multiple
   genes. It will be interesting to study the impact of these expression
   alterations on UCEC cancer cells.

   As we can see in Figure [130]5, HMGCS1 was closely correlated and
   interacted with ACAT2 and MVD. All of three are involved in the MVA
   pathway and regulated by sterol regulatory element binding protein 2
   (SREBP2).[131] ^26 It has been reported that, after statin treatment,
   cancer cells could sustain the MVA metabolism by upregulating the
   expression of other genes in the MVA pathway via the feedback
   regulation of SREBP2.[132] ^27 As MVA pathway is complicatedly
   intertwined, combined inhibition of multiple targets may enhance the
   cytotoxicity effects. Furthermore, enrichment analysis indicated a
   prominent role of HMGCS1 in steroid metabolism (Figure [133]5F). It is
   widely known that MVA‐derived steroids are the raw material for the
   synthesis of many gonadal steroid hormones, contributing to the
   development and progression of hormone dependent tumors such as
   prostate and breast cancer.[134] ^28 , [135]^29 As mentioned before,
   several reports have demonstrated the oncogenic role of HMGCS1 in
   prostate and breast cancer. Besides, steroid biosynthesis has been
   shown to play a central role in T cells to achieve immune evasion.[136]
   ^30 In conclusion, HMGCS1 is a promising target in cancer treatment.

   High HMGCS1 expression could reduce the sensitivity of many drugs,
   indicating its potential role in drug resistance. However, HMGCS1
   enhanced the sensitivity of GSK690693, KIN001‐102, and navitoclax
   (Figure [137]6). GSK690693 and KIN001‐102 are Akt inhibitors, while
   navitoclax is a BCL‐2 inhibitor. This finding sheds lights on the
   potential clinical application of these drugs to overcome HMGCS1
   mediated drug resistance. Hymeglusin is the specific HMGCS1
   inhibitor,[138] ^31 which functions through binding to Cys‐129.[139]
   ^25 We have proved its functions of suppressing cell growth and
   promoting apoptosis (data not shown). Further studies are required to
   know more about its functions and sides effects.

   There were several limitations about our study. The evaluation of
   HMGCS1 alone may not recapitulate the whole characteristics of the MVA
   pathway. Therefore, it is important to further explore other vital
   genes and metabolites in this pathway to comprehensively uncover the
   function of MVA pathway in pan‐cancer. Additionally, a negative
   correlation between HMGCS1 expression and survival outcomes was
   presented after data analysis in this study. To make the conclusion
   more convincing, more basic and clinical research were needed.
   Furthermore, according to our and others previous reports, HMGCS1 also
   had impacts on cell proliferation, apoptosis, and immunity. The
   detailed mechanisms needed to be studied further.

   Taken together, our pan‐cancer analysis provides a broad perspective of
   the oncogenic roles of HMGCS1 across different tumors and points out
   new directions for the treatment of cancer.

CONFLICT OF INTEREST

   The authors declare that they have no conflict of interest.

AUTHOR CONTRIBUTIONS

   C.Z.; Conceptualization (lead); data curation (lead); writing –
   original draft (equal). Z.W.; Conceptualization (equal); data curation
   (equal); funding acquisition (equal); writing – original draft (equal).
   Y.C.; Data curation (supporting). L.Z.; Conceptualization (supporting);
   funding acquisition (lead); supervision (lead); writing – review and
   editing (lead).

ETHICS STATEMENT

   This study is exempt from institutional review board approval.

Supporting information

   Figure S1 Expression levels of HMGCS1 in different tumors and
   pathological stages. (A‐B) The HMGCS1 expression in AML, SKCM, LGG, OV,
   TGCT, and UCS. *P < 0.05. T: tumor tissues; N: normal tissues; AML:
   acute myeloid leukemia; SKCM: skin cutaneous melanoma; LGG: liver
   hepatocellular carcinoma; OV: ovarian serous cystadenocarcinoma; TGCT:
   testicular germ cell tumors; UCS: uterine carcinosarcoma. (C) The
   HMGCS1 mRNA levels in different cancer stages of KIRC, OV and THCA,
   which were analyzed by the Pathological Stage Plot module of GEPIA2.
   Log2 (TPM + 1) was applied for log‐scale. KIRC: kidney renal clear cell
   carcinoma; THCA: thyroid carcinoma.

   Figure S2. The correlation between HMGCS1 expression level and OS (A),
   and DFS (B) in different tumors. OS: overall survival; DFS:
   disease‐free survival.

   Figure S3. An illustration of immune component outputs upon a query to
   the immune components. (A) Correlation analysis between HMGCS1
   expression and immune infiltration of CD8+ T‐cells across all types of
   cancer in TCGA. (B) Some representative scatter plots showed the
   potential correlation between HMGCS1 expression level and the
   infiltration level of CD8+ T‐cells.
   [140]Click here for additional data file.^ (956.6KB, docx)

ACKNOWLEDGMENTS