Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1 DOI: 10.34175/jno202001004 · 31 ·

Bioinformatics Analysis of HKDC1 Expression in Non-small Cell Lung Cancer and Its Relationship to Survival

Lei Wang, Wei Yuan Li, Fang Fang Li, Chun Yan Yang, Deng Cai Mu, Shang Yong Zheng

School of Medicine, Yunnan University, Kunming, Yunnan 650091, China

Abstract: Lung cancer has become one of the most common types of cancer, and has the highest morbidity and mortality rates. We herein performed a bioinformatic analysis to explore the expression of domain containing 1 (HKDC1) to further illuminate the molecular mechanisms involved in non-small cell lung cancer (NSCLC). We also examined the relationship between HKDC1 expression and patient survival. The Oncomine and cBioPortal cancer genomics databases were used to investigate the expression of HKDC1 in NSCLC. Then the protein-protein interaction network and protein expression intensity of HKDC1 were examined with the Gene MANIA and The Human Protein Atlas databases. Finally, the prognostic impact of HKDC1 was evaluated via the GEPIA, Oncolnc and Kaplan-Meier Plotter online tools. HKDC1 was significantly up-regulated in NSCLC patients. In addition, the mutation frequency of HKDC1 in NSCLC was high in a large number of studies. Moreover, HKDC1 expression was correlated with a poor prognosis in NSCLC patients. These findings suggest that HKDC1 represents a novel therapeutic target for NSCLC. Key words: HKDC1; NSCLC; Bioinformatics; Expression; Carbohydrate

Introduction genome-scale data source in life science research. Unlike Lung cancer (LC) is a life-threatening disease and the DNA sequence analysis and genotyping, microarrays most prominent cause of cancer morbidity and mortality in analyze mRNA. with meaningful statistically both men and women [1]. Lung cancer is usually classified significant differences can be identified by comparing into two types, non-small cell lung cancer (NSCLC) and expression in adjacent normal tissues and tumor tissues. small cell lung cancer (SCLC). Non-small cell lung cancer We applied mRNA microarray analysis to 15 groups of accounts for about 80% of all lung cancers, and is mainly NSCLC tissues and adjacent non-tumor lung tissues. It composed of three subtypes: lung adenocarcinoma (LUAD), was found that hexokinase domain containing 1 (HKDC1) lung squamous cell carcinoma (LSCC) and large cell lung had relatively high expression in the tumor tissues and cancer (LLCC) [2]. Currently, the most common treatment showed a potential interaction with O-glycan synthesis strategy is chemotherapy [3]. One of the main causes of glucosaminyl (N-acetyl) transferase 3 (GCNT3) in lung cancer is human epidermal growth factor receptor enriched network diagrams. Previous studies have shown (EGFR) mutation or overexpression/gene amplification. that GCNT3 acts as a NSCLC oncogene [5]. It has also Mutations in EGFR result in sustained activation of the been reported that HKDC1 can effectively cluster LSCC EGFR signal, which drives the growth of tumor cells. patients in gene clustering models, and up-regulation of this This signaling pathway can be effectively blocked using gene may play a role in the adaptive stress response [6]. a tyrosine kinase inhibitor (TKI) that targets EGFR, such Moreover, the HKDC1 protein has enzymatic properties that as erlotinib or gefitinib. However, due to the presence are active in carbohydrate metabolism, and it is considered of intrinsic drug resistance, only a small percentage of to be a hexokinase isoform (REF). patients can benefit from the present treatments. Therefore, The present study was conducted to evaluate the it is necessary to find a more effective therapeutic target. expression of HKDC1 in NSCLC patients by using public Studies have confirmed that one of tyrosine kinase receptor gene expression datasets and bioinformatics platforms (AXL) is a potential therapeutic target, and its inhibition based on the Web. The work accomplished in this study can prevent or overcome the acquired resistance of EGFR provides additional information on the underlying molecular mutant lung cancer to EGFR-TKIs [4]. mechanism in NSCLC. We believe the present findings With the increasing application of genome sequencing, will be helpful in the clinical diagnosis of lung cancer, and DNA microarray analysis has become the most widely used provides new targets for the treatment of NSCLC patients.

Materials and Methods Corresponding author: Shang Yong Zheng, MD, PhD, School of Sample collection and human gene expression Medicine, Yunnan University, 2 Cuihu Road, Kunming, Yunnan 650091, China; Tel: +86 182 0889 0689; Fax: +86 871 65034358; microarrays Email: [email protected] The genes studied were the same as in a previous · 32 · Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1 article [5]. Fifteen pairs of lung tissue samples (including in NSCLC was conducted in the Oncomine database cancerous tissue and para-cancerous tissues) for gene chip (http://www.oncomine.org/) with the following filter analysis were collected from lung cancer patients who conditions on datasets: (HKDC1, cancer vs. normal underwent thoracic surgery at the Yunnan Cancer Hospital analysis, lung cancer, mRNA, clinical specimen). The in 2014-2015. Each patient’s condition had been confirmed mRNA expression in NSCLC and adjacent non-tumor before the resection, and none of the patients had received lung tissues (ANT) or normal tissues was downloaded treatment or had other tumors that might have metastasized from the Oncomine database. Then, raw data were to the lung. The surgeon removed the excised tissue from normalized to transform them to log2 values. The the operating room and placed it in a cell cryotube, then cBioPortal database provides a portal that simplifies immediately put the tissue cryotube into liquid nitrogen the molecular analysis of cancer tissues and cell lines for cryopreservation (the tissue was frozen within 30 min into readily understandable genetic, epigenetic, gene of resection). The sample information was recorded, and expression, and proteomic events [10]. The cBioPortal the sample was stored in an ultra-low temperature freezer for the Cancer Genomics database was used to view the at -80°C until analysis. All of the above processes were proportion of genetic alterations in NSCLC patients and approved by the patients and their families, as well as the the results of mutation analyses of the mRNA. Ethics Committee of Yunnan Cancer Hospital and the Ethics Committee of Yunnan University, and met the clinical Protein expression of HKDC1 in NSCLC norms and medical ethics regulations. The samples were The Human Protein Atlas (HPA) aims to map all human analyzed using an oligonucleotide microarray (Affymetrix proteins in cells, tissues and organs using the integration GeneChip Human Transcriptome Array 2.0, Thermo Fisher of various -omics technologies with online bioinformatics Scientific) to determine the expression profile of NSCLC- tools. The HPA currently facilitates the systematic related genes, and the original data were normalized to investigation of the transcriptome of the protein-coding obtain the expression levels of different genes, which were genes for more than 17 tumor types [11]. It also contains statistically analyzed. images of histological sections from normal and cancer tissues obtained by immunohistochemistry. Antibodies were Functional analysis for promising up-regulated genes labeled with 3,3’-diaminobenzidine (DAB) and the resulting The Metascape (http://metascape.org/gp/), an online brown staining indicates where an antibody had bound to bioinformatics resource platform, includes gene annotation, its corresponding antigen. The intensity of antibody staining enrichment summaries, and pathway and process indicates the extent of protein expression. We verified the enrichment analyses [7]. We uploaded 174 genes that were protein expression of HKDC1 in NSCLC samples using this identified as being up-regulated into the object box of platform. Metascape. The functional annotation of those genes was then elucidated by enrichment analysis on the Metascape Evaluation of the prognostic value of HKDC1 in tool. A clustering network was also constructed to identify NSCLC the valuable genes with a potential interaction with GCNT3. We evaluated the clinical prognostic significance of the HKDC1 gene using the GEPIA [12], Oncolnc [13] Construction of a protein-protein interaction (PPI) and Kaplan-Meier Plotter [14] online tools. The GEPIA network database was used to assess the disease-free survival GeneMANIA is an open online tool that can be used to (DFS) of NSCLC patients. We set a 50:50 ratio collecting construct biomolecular interaction networks [8]. This makes survival data of NSCLC patients on the Oncolnc databases. it possible to explore the protein-protein, protein-DNA and In addition, the best probe was selected in Kaplan- genetic interactions, pathways, reactions, gene and protein Meier Plotter to obtain the prognostic information of LC expression data for any given gene. We downloaded a patients. P-values < 0.05 were considered to be statistically chart that identifies the most co-expressed and pathway- significant. associated genes or proteins. Information was collected on several genes, and the most interesting were selected for Results further study. Human gene expression microarray analysis We performed human gene expression microarray Evaluation of the mRNA expression of HKDC1 via studies on 15 pairs of NSCLC and ANT. The differential Oncomine and cBioPortal data expression of genes was determined by the fold change (FC), The Oncomine is the world’s largest oncogene and the P-value calculated using a t-test. Thresholds were microarray chip database, and has an integrated data set for up-regulation and down-regulation in cases with mining platform. It has the most comprehensive cancer FC ≥ 2.0 and P-value ≤ 0.05. In total, there were 623 up- mutation spectrum, gene expression data and related regulated and 2118 down-regulated mRNAs identified in clinical information [9]. A microarray search for HKDC1 the NSCLC samples by Student’s t-test (see supplementary). Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1 · 33 ·

We selected the HKDC1 gene because it had a P-value < 0.05 research significance, we selected the top 174 up-regulated and FC = 3.31. In addition, based on a literature search, as genes (P < 0.05, FC > 3.26), and performed well as gene expression, functional enrichment, and survival (GO) (Figure 1A), and function clustering network analyses, we found that the HKDC1 gene has potential (Figure 1B) analyses using the based massive data tool, implications in lung cancer. Metascape. GCNT3 (GO: 0005975) showed involvement in carbohydrate metabolic processes in the GO analysis. Functional analysis for promising up-regulated genes Interestingly, a bioinformatics analysis found that HKDC1 To narrow down the study to genes with meaningful and GCNT3 were in the center of clusters with the same

A R-hsa-1640170: Cell cycle G0:0000819: Sister chromatid segregation R-hsa-983189: Kinesins G0:0048589: Developmental growth R-hsa-70614: synthesis and interconversion (transamination) M176 :Pid foxml pathway G0:0030155: Regulation of cell adhesion G0:0005975: Carbohydrate metabolic process GO:0071103: DNA conformation change G0:0007052: Mitotic spindle organization Hsa04950: Maturity onset diabetes of the young G0:0048729: Tissue morphogenesis G0:0010466: Negative regulation of peptidase activity R-hsa-190861: Gap junction as sembly G0:0006260: DNA replication G0:0001822: Kidney development G0:0030856: Regulation of epithelial cell differentiation G0:0033628: Regulation of cell adhesion mediated by integrin G0:0031214: Biomineral tissue development G0:0001824: Blastocyst development 0 1 2 3 4 5 6 7 8 -Log10(P) B

Carbohydrate metabolic process

Figure 1 Distribution of gene ontology terms and the function clustering network for the top 174 genes up- regulated in NSCLC. A. The enrichment summary. B. The function clustering network: Networks are coloured by cluster ID, where nodes that share the same cluster ID were typically close to each other. · 34 · Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1 enrichment. be related and form functional interaction modules involved in the regulation of HKDC1 in NSCLC. The GPR39 gene, Construction of a protein-protein interaction (PPI) which had a close relationship, had previously been reported network to have abnormal mutations [15] and different expression We input GCNT3 and HKDC1, or HKDC1 alone, into [16-18]. Therefore, we further verified the expression and GeneMANIA. There were 20 related genes identified, with mutation status of HKDC1. 22 total genes, 0 attributes, and 155 total links (Figure 2A) for the GCNT3 and HKDC1 network, In addition, when we Differences in the expression of mRNAs in NSCLC only entered HKDC1, there were 20 related genes, with 21 tissues total genes, 0 attributes, and 75 total links (Figure 2B). Thus, In a total of 20 ANT and 226 NSCLC samples included 6 genes (HK1, HK2, HK3, ENTPD7, NR5A2, GPR39) may in the databases, the mean ± SD of ANT and NSCLC were

A HK3 B C6orf25 HK2 LIPH GRID2 HK1 NR5A2 GCTP1 PDZD3 GCK C1orf106 IMPG1 NR5A2 CEACAM5 GCNT7 HFE ENTPD7 HKDC1 KRAS MUC13 GCNT1 HKDC1 DAPK2 SCGN GPR39

GPR39 GCNT4 GCNT3 MEF2C PLOD2 HK3 HK2 ENTPD7 GCNT2

B3GNT3 XYLT2 C3orf36 GCK HK1 Co-expression GCLM KRAS Pathway DHRSX XYLT1 Figure 2 The PPI network of possible target genes. Each node represents a gene-encoded protein, while lines between the nodes represent protein associations. A. The protein interaction map between GCNT3 and HKDC1 (co-expression and pathways). B. The co-expression and pathways of the HKDC1 protein interaction network.

A B

75 10

60 8 45 6 30 4 15

0 2 Expreeion of HKDC1 mRNA Expreeion of HKDC1 mRNA -15 0 Normal tissue NSCLC Adjacent normal tissue NSCLC (n = 20) (n = 226) (n = 20) (n = 226)

Figure 3 Expression of HKDC1 in NSCLC patients based on Oncomine data. A. The mRNA expression of 20 ANT and 226 NSCLC samples from the Okayama Lung clinic (all LUAD). B. 58 pairs of NSCLC and ANT samples from the selamat lung study (all LUAD). The expression level of HKDC1 in NSCLC tissues was obviously higher than that of ANT. (**: P < 0.001;***: P < 0.0001). Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1 · 35 ·

1.189 ± 0.4135 and 5.054 ± 6.033 (Figure 3A), respectively. Evaluation of the prognostic value of HKDC1 in NSCLC In addition, 58 pairs of NSCLC and ANT samples from We examined the association of HKDC1 with patients the Oncomine database were included in the current survival based on data from three databases. When the data comparison. As shown in Figure 3B, the expression of sources were overlapped, they showed significant differences HKDC1 in NSCLC was obviously higher compared with in prognostic information (Figure 6). In the Oncolnc data, that in ANT (P-value < 0.0001). The cBioPortal datasets we observed that the prognostic relationship was significant revealed that there were genetic alterations in NSCLC (P = 0.00628). Overall, the data indicated that HKDC1 was patients, and a mutation analysis of HKDC1 indicated a highly expressed in NSCLC, with a high mutation rate, and high mutation rate (Figure 4). may indicate a poor prognosis of NSCLC patients.

Protein expression of HKDC1 in NSCLC tissues Discussion The protein expression of HKDC1 was relatively weak. In this paper, high expression of HKDC1 was identified As shown in Figure 5, HKDC1 was downregulated in from 15 clinical samples of lung cancer. NSCLC is a NSCLC samples compared to normal tissues in the Human common cancer, but the treatment effects are difficult to Protein Atlas databases. The intensity of antibody staining predict. Therefore, it is important to look for predictive was weak in NSCLC, and its immunohistochemical score biomarkers to provide new insights, diagnostic information was low. and targets for treatment. In the current study, we not only However, there was a relatively limited number of cases identified the high expression of HKDC1 in NSCLC, provided by the Human Protein Atlas database, and more and found that it is related to the patient survival, but samples are needed to confirm these findings. also provided some insight into its potential molecular

A B 14

5% 12

10 2% 8 5% 6

4 Alteration Frequency 1%

2 0.5% 0

-2 Lung adenoNSCLC Lung(TCGA (TCGA squLung PanCan) (TCGA 2016) squLung (TCGA) adenoLUADPanCan) (TCGA) (MSK, 2017) Truncating (VUS) HKDC1 Expreeion --- RNA Seq V2 (log2) Seq HKDC1 Expreeion --- RNA -4 Missence (VUS)

-6 Not mutated Not profiled for mutated -8

Lung adeno (TCGALung PanCan) adeno (TCGALung Pub) adeno (TCGA)Lung squ (TCGALung PanCan) squ (TCGA)

Mutation Amplification Deep Deletion

Figure 4 Summary of the mutation rate and RNA expression of HKDC1. A. The mutation frequency of HKDC1 was different in 300 cancer studies. B. The HKDC1 RNA sequence expression based on the TCGA provisional and pan-can atlas is shown. The mutation rate was especially high in lung adenocarcinoma. · 36 · Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1 mechanism of action. mucin synthesis. We found that 6 genes (HK1, HK2, HK3, We performed a GO analysis and pathway and process ENTPD7, NR5A2, GPR39) coexisted in the co-expression enrichment analyses on the top 174 up-regulated genes, and pathway networks of HKDC1 and GCNT3. Among and found that GCNT3 and HKDC1 occupied a central these 6 genes, we found that abnormal mutations of HK1 position in carbohydrate metabolism. We suspected that directly led to the development of disease [19]. The loss of both genes may play a regulatory role in glycolysis and NR5A2 affected the expression of acinar-specific partner

Figure 5 Protein expression of HKDC1 determined by immunohistochemistry. A. Low staining of HKDC1 in normal lung tissues (antibody: HPA011956; magnification of 3×10). B. Low staining of HKDC1 in LUSC (antibody: HPA011956; magnification of 3 ×100). C. Low staining of HKDC1 in normal lung tissues (antibody: HPA011956; magnification of 3 ×10). D. Low staining of HKDC1 in LUSC (antibody: HPA011956; magnification of 3 ×100).

A B C (GEPIA) Disease Free Survival OncoLnc (Kaplan-Meier Plotter) 1.0 100 Low High 1.0 Low HKDC1 YPM N = 244 N = 244 HR = 1.18 (1.04-1.34) High HKDC1 YPM Logrank P = 0.01 Logrank P-valua = 0.00628 0.8 Logrank P = 0.012 80 0.8 HR (high) = 1.3 Number at risk low 964 435 ˆ 95 24 3 P (HR) = 480 60 0.6 n (low) = 480 0.6 high 962 393 105 33 4

40 0.4 Probebility 0.4 % Surviving Percent survival Expression 0.2 20 0.2 low 0.0 0 0.0 high 0 50 100 150 200 250 0 1000 2000 3000 4000 5000 6000 0 50 100 150 200 Time (months) Time (days) Time (months) Figure 6 Kaplan-Meier survival curves of different NSCLC subtypes based on the HKDC1 expression levels. A. The DFS survival curve of 480 LUSC and LUAD patients (HR = 1.3, P = 0.012) in the GEPIA databases. B. survival curve of 244 LUSC patients (P = 0.00628) in the Oncolnc database. C. The survival curve of 962 LC patients (HR = 1.18, P = 0.022) based on the Kaplan-Meier Plotter online tool. Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1 · 37 ·

Rbpjl, so that the acinar cells couldn’t fully differentiate [20]. are still limited. In particular, more clinical samples need These findings suggest that these pathways are associated to be examined, and the findings needs to be confirmed in with the development of lung cancer. independent studies. Next, we found that HKDC1 was highly expressed in NSCLC in public databases, and also had a higher mutation Abbreviation rate, especially in LUAD. With regard to protein expression, HKDC1, Hexokinase domain containing 1; HKDC1 was only weakly expressed, which we suspect was NSCLC, Non-small cell lung cancer; due to its unstable nature and rapid degradation. Finally, LC, Lung cancer; we evaluated the survival prognosis of HKDC1 using SCLC, Small cell lung cancer; three different online network tools and found that high LUAD, Lung adenocarcinoma; expression of HKDC1 was associated with a poor prognosis LSCC, Lung squamous cell carcinoma; in patients. LLCC, Large cell lung cancer; HKDC1, encoding Hexokinase Domain-Containing EGFR, Epidermal growth factor receptor; Protein 1, is located on chromatin position chr10: TKI, Tyrosine kinase inhibitor; 55,977,981-56,025,237. It is 47257 bp in size according AXL AXL, receptor tyrosine kinase; to the USSC Genome browser. It encodes three protein GCNT3, O-glycan synthesis enzyme glucosaminyl subtypes, with subtype I being the most commonly encoded (N-acetyl) transferase 3; protein. Studies have shown that the HKDC1 is highly HK1, Hexokinase 1; enriched in metabolism, lipid metabolism and HK2, Hexokinase 2; protein metabolism pathways [21]. In addition, in LSCC, HK3, Hexokinase 3; high-mutation genes were identified by a TCGA database, LIPH, Lipase H; and 14 of them were found to provide the best clustering NR5A2, Nuclear receptor subfamily 5 group a member model. The HKDC1 gene clustering model could effectively 2; cluster patient samples. Moreover, in some primary tumors, C1orf106, 1 open reading frame 106; HKDC1 affected the brain and muscle ARNT-like-1 gene CEACAM5, Carcinoembryonic antigen related cell (BMAL1), leading to disturbances in its metabolism, adhesion molecule 5; increasing glycolytic activity, and changing the response MUC13, Mucin 13, cell surface associated; to treatment [22]. These findings indicate that HKDC1 is GPR39, G protein-coupled receptor 39; highly involved in tumors by participating in the hexose ENTPD7, Ectonucleoside triphosphate diphosphohy- metabolism pathway. drolase 7; Meanwhile, in the PPI network map, we found that B3GNT3, UDP-GlcNac:betaGal beta-1,3-N-acetyl- only the myocyte-specific enhancer factor 2C (MEF2C) glucosaminyl transferase 3; gene is associated with HKDC1. It had previously been KRAS, KRAS proto-oncogene,GTPase; reported that mutation of the MEF2C gene was related to DHRSX, Dehydrogenase/reductase X-linked; the phenotype of Rett syndrome patients [23]. Mutations of XYLT1, Xylosyltransferase 1; the MEF2C gene were also associated with congenital heart XYLT2, Xylosyltransferase 2; defects (CHD) [24]. GCNT1,2,4,7 Glucosaminyl (N-acetyl) transferase It has been reported that the missense mutation of 1,2,4,7; HKDC1 affects highly-conserved amino acids, and the GCK, ; mutant HKDC1 protein partially loses activity [25]. C6orf25, Chromosome 6 open reading frame 25; Our study found that HKDC1 had a higher frequency of GRID2, Glutamate ionotropic receptor delta type mutations in NSCLC than ANT, supporting this concept. subunit 2; Although studies had reported that HKDC1 could be used GCTP1, Growth hormone regulated TBC protein 1; as a therapeutic target [26], we confirmed that HKDC1 was PDZD3, PDZ domain containing 3; associated with the prognosis of NSCLC patients based on a IMPG1, Interphotoreceptor matrix proteoglycan 1; multi-modal survival prognosis evaluation using the online HFE, Hemochromatosis; bioinformatics analysis of ‘massive data’. DAPK2, Death associated protein kinase 2; Our findings indicated that the high expression of SCGN, Secretagogin, EF-hand calcium binding protein; HKDC1 may be associated with decreased survival. MEF2C, Myocyte enhancer factor 2C; It is expected that the use of targeted therapy using a PLOD2, Procollagen-lysine,2-oxoglutarate 5- combination of multiple genes with high tumor expression dioxygenase 2; would be more effective against malignant tumors than C3orf36, Chromosome 3 open reading frame 36; conventional cytotoxic chemotherapy. GCLM, Glutamate-cysteine ligase modifier subunit; However, despite the extensive data available in the HR, Hazard ratio; databases, the specific data regarding the impact of HKDC1 ANT, Adjacent non-tumor lung tissues; · 38 · Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1

HPA, The Human Protein Atlas; ways, and networks in a collection of 18,000 cancer gene expression DAB 3, 3’-diaminobenzidine; profiles. Neoplasia 2007;9(2):166-80. DFS, Disease-free survival; 10. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, FC, Fold change; Sun Y, Jacobsen A, Sinha R, Larsson E, Cerami E, Sander C, Schultz N. GO, Gene ontology; Integrative analysis of complex cancer genomics and clinical profiles BMAL1, Brain and muscle ARNT-like-1; using the cBioPortal. Sci Signal 2013;6(269):l1. GEPIA, Gene expression profiling interactive analysis; 11. Uhlen M, Zhang C, Lee S, Sjostedt E, Fagerberg L, Bidkhori G, Benfeitas R, Arif M, Liu Z, Edfors F, Sanli K, von Feilitzen K, Oksvold Acknowledgements P, Lundberg E, Hober S, Nilsson P, Mattsson J, Schwenk JM, Brunn- The authors would like to thank the Metascape, Gene strom H, Glimelius B, Sjoblom T, Edqvist PH, Djureinovic D, Micke P, MANIA, Oncomine, cBioPortal, GEPIA, Oncolnc, Kaplan- Lindskog C, Mardinoglu A, Ponten F. A pathology atlas of the human Meier Plotter, and The Human Protein Atlas databases for cancer transcriptome. Science. 2017 Aug 18;357(6352). pii: eaan2507. the valuable data. doi: 10.1126/science.aan2507. 12. Tang Z, Li C, Kang B, Gao G, Li C, Zhang Z. GEPIA: a web server Conflict of Interests for cancer and normal gene expression profiling and interactive analy- The authors declare that there are no conflicts of interest. ses. Nucleic Acids Res 2017; 45(W1):W98-102. 13. Wang J, Luo XX, Tang YL, Xu JX, Zeng ZG. The prognostic values Funding of insulin-like growth factor binding protein in breast cancer. Medicine This work was supported by the National Natural (Baltimore) 2019;98(19):e15561. Science Foundation of China (No. 81860494). 14. Hou GX, Liu P, Yang J, Wen S. Mining expression and prognosis of topoisomerase isoforms in non-small-cell lung cancer by using Oncom- ine and Kaplan-Meier plotter. PLoS One 2017;12(3):e0174515. References 15. Kovacs Z, Schacht T, Herrmann AK, Albrecht P, Lefkimmiatis 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J K, Methner A. Protein kinase inhibitor beta enhances the constitutive Clin 2019;69(1):7-34. activity of G-protein-coupled zinc receptor GPR39. Biochem J 2014; 2. Inamura K. Lung cancer: Understanding its molecular pathology and 462(1):125-32. the 2015 who classification. Front Oncol 2017; 7:193. 16. Moran BM, Miskelly MG, Abdel-Wahab YHA, Flatt PR, McKillop 3. Gridelli C, Rossi A, Carbone DP, Guarize J, Karachaliou N, Mok T, AM. Zinc-induced activation of GPR39 regulates glucose homeostasis Petrella F, Spaggiari L, Rosell R. Non-small-cell lung cancer. Nat Rev through glucose-dependent insulinotropic polypeptide secretion from Dis Primers 2015; 1:15009. enteroendocrine K-cells. Biol Chem. 2019 Jun 20. pii: /j/bchm.ahead- 4. Zhang Z, Lee JC, Lin L, Olivas V, Au V, LaFramboise T, Abdel-Rah- of-print/hsz-2018-0393/hsz-2018-0393.xml. doi: 10.1515/hsz-2018- man M, Wang X, Levine AD, Rho JK, Choi YJ, Choi CM, Kim SW, 0393. Jang SJ, Park YS, Kim WS, Lee DH, Lee JS, Miller VA, Arcila M, 17. Chai X, Zhang W, Chang B, Feng X, Song J, Li L, Yu C, Zhao J, Si Ladanyi M, Moonsamy P, Sawyers C, Boggon TJ, Ma PC, Costa C, H. GPR39 agonist TC-G 1008 promotes osteoblast differentiation and Taron M, Rosell R, Halmos B, Bivona TG. Activation of the AXL ki- mineralization in MC3T3-E1 cells. Artif Cells Nanomed Biotechnol nase causes resistance to EGFR -targeted therapy in lung cancer. Nat 2019;47(1):3569-76. Genet 2012; 44(8):852-60. 18. Yang Q, Li Y, Zhang X, Chen D. Zac1/GPR39 phosphorylating 5. Li Q, Ran P, Zhang X, Guo X, Yuan Y, Dong T, Zhu B, Zheng S, CaMK-II contributes to the distinct roles of Pax3 and Pax7 in myogenic Xiao C. Downregulation of N-acetylglucosaminyltransferase GCNT3 progression. Biochim Biophys Acta Mol Basis Dis 2018;1864(2):407- by miR-302b-3p decreases Non-Small Cell Lung Cancer (NSCLC) 19. Cell Proliferation, Migration and Invasion. Cell Physiol Biochem 2018; 19. Koralkova P, Mojzikova R, van Oirschot B, Macartney C, Timr P, 50(3):987-1004. Vives Corrons JL, Striezencova Laluhova Z, Lejhancova K, Divoky V, 6. Li J, Wang J, Chen Y, Yang L, Chen S. A prognostic 4-gene expres- van Wijk R. Molecular characterization of six new cases of red blood sion signature for squamous cell lung carcinoma. J Cell Physiol 2017; cell hexokinase deficiency yields four novel mutations in HK1. Blood 232(12):3702-13. Cells Mol Dis 2016;59(71-76). 7. Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk 20. Hale MA, Swift GH, Hoang CQ, Deering TG, Masui T, Lee YK, O, Benner C, Chanda SK. Metascape provides a biologist-oriented Xue J, MacDonald RJ. The nuclear hormone receptor family member resource for the analysis of systems-level datasets. Nat Commun 2019; NR5A2 controls aspects of multipotent progenitor cell formation and 10(1):1523. acinar differentiation during pancreatic organogenesis. Development 8. Franz M, Rodriguez H, Lopes C, Zuberi K, Montojo J, Bader 2014;141(16):3123-33. GD, Morris Q. GeneMANIA update 2018. Nucleic Acids Res 2018; 21. Tian X, Zhang H, Zhao Y, Mehmood K, Wu X, Chang Z, Luo M, 46(W1):W60-4. Liu X, Ijaz M, Javed MT, Zhou D. Transcriptome analysis reveals 9. Rhodes DR, Kalyana-Sundaram S, Mahavisno V, Varambally R, Yu the molecular mechanism of hepatic metabolism disorder caused J, Briggs BB, Barrette TR, Anstet MJ, Kincead-Beal C, Kulkarni P, by chromium poisoning in chickens. Environ Sci Pollut Res Int Varambally S, Ghosh D, Chinnaiyan AM. Oncomine 3.0: genes, path- 2018;25(16):15411-21. Journal of Nutritional Oncology, February 15, 2020, Volume 5, Number 1 · 39 ·

22. Fuhr L, El-Athman R, Scrima R, Cela O, Carbone A, Knoop H, Li Y, of-function mutation associated with congenital double outlet right Hoffmann K, Laukkanen MO, Corcione F, Steuer R, Meyer TF, Maz- ventricle. Pediatr Cardiol 2018; 39(4):794-804. zoccoli G, Capitanio N, Relogio A. The circadian clock regulates meta- 25. Zhang L, Sun Z, Zhao P, Huang L, Xu M, Yang Y, Chen X, Lu F, bolic phenotype rewiring via HKDC1 and modulates tumor progression Zhang X, Wang H, Zhang S, Liu W, Jiang Z, Ma S, Chen R, Zhao C, and drug response in Colorectal cancer. EBioMedicine 2018;33:105-21. Yang Z, Sui R, Zhu X. Whole exome sequencing revealed HKDC1 as a 23. Wang JP, Zhang QP, Chen Y, Yu SJ, Wu XR, Bao XH, Wen YX. candidate gene associated with autosomal-recessive retinitis pigmento- Novel MEF2C point mutations in Chinese patients with Rett (-like) sa. Hum Mol Genet 2018;27(23):4157-68. syndrome or non-syndromic intellectual disability: insights into geno- 26. Li GH, Huang JF. Inferring therapeutic targets from heterogeneous type-phenotype correlation. BMC Med Genet 2018;19:191. data: HKDC1 is a novel potential therapeutic target for cancer. Bioin- 24. Lu CX, Wang W, Wang Q, Liu XY, Yang YQ. A novel MEF2C loss- formatics 2014;30(6):748-52.

Received: September 4, 2019 Revised: September 19, 2019 Accepted: November 6, 2019