Overexpression of HPRT1 is associated with poor prognosis in head and neck squamous cell carcinoma

Hypoxanthine phosphoribosyltransferase (HPRT1), as a salvage pathway enzyme, plays a crucial role in modulating the cell cycle and has been reported to be overexpressed in multiple cancers. Nevertheless, the relationship between the HPRT1 gene and head and neck squamous cell carcinomas (HNSCCs) has not been investigated so far. In this study, we first evaluated the expression and clinical value of HPRT1 mRNA and protein in tumor and healthy control tissues. Then, we examined mutations of the HPRT1 gene and their association with survival outcomes of patients with HNSCC. We also performed functional analyses of HPRT1 coexpressed genes and examined the association between HPRT1 expression and drug sensitivity. Both HPRT1 mRNA and protein were significantly higher in HNSCC compared with normal tissues, and up‐regulation of HPRT1 was also correlated with age, sex, pathological stage and histological grades of patients with HNSCC. Moreover, HPRT1 and its associated genes were observed to be enriched for several cancer‐related pathways, including DNA replication and cell cycle. Finally, patients exhibiting overexpression of the HPRT1 gene may be resistant to abiraterone and sensitive to several drugs, including tozasertib and teniposide. This study demonstrated that the elevated expression of HPRT1 gene is correlated with the progression of HNSCC; thus, this gene may serve as a useful indicator for the early detection, risk stratification and targeted therapy of patients with HNSCC.

Hypoxanthine phosphoribosyltransferase (HPRT1), as a salvage pathway enzyme, plays a crucial role in modulating the cell cycle and has been reported to be overexpressed in multiple cancers. Nevertheless, the relationship between the HPRT1 gene and head and neck squamous cell carcinomas (HNSCCs) has not been investigated so far. In this study, we first evaluated the expression and clinical value of HPRT1 mRNA and protein in tumor and healthy control tissues. Then, we examined mutations of the HPRT1 gene and their association with survival outcomes of patients with HNSCC. We also performed functional analyses of HPRT1 coexpressed genes and examined the association between HPRT1 expression and drug sensitivity. Both HPRT1 mRNA and protein were significantly higher in HNSCC compared with normal tissues, and up-regulation of HPRT1 was also correlated with age, sex, pathological stage and histological grades of patients with HNSCC. Moreover, HPRT1 and its associated genes were observed to be enriched for several cancer-related pathways, including DNA replication and cell cycle. Finally, patients exhibiting overexpression of the HPRT1 gene may be resistant to abiraterone and sensitive to several drugs, including tozasertib and teniposide. This study demonstrated that the elevated expression of HPRT1 gene is correlated with the progression of HNSCC; thus, this gene may serve as a useful indicator for the early detection, risk stratification and targeted therapy of patients with HNSCC.
Head and neck squamous cell carcinoma (HNSCC) has been introduced as the sixth most common malignancy globally. HNSCC originates from various subsites of the upper aerodigestive tract, such as the oral cavity, larynx, pharynx, paranasal sinuses and nasal cavity [1][2][3][4]. Despite current efforts for the risk stratification and treatment of patients with HNSCC, the overall survival (OS) rate of these patients is unsatisfactory and has not significantly improved over the past decade, which may be because of late-stage diagnosis. The lack of effective means for clinical applications urgently demands the findings of more applicable biomarkers to improve the earlier detection and targeted therapies of patients with HNSCC [5][6][7][8][9][10][11].
Hypoxanthine phosphoribosyltransferase (HPRT1) transcripts the HPRT protein, a transferase enzyme that takes part in the cell cycle through modulation of guanine and inosine production in the salvage pathway [12,13]. Although the HPRT1 is broadly used as a housekeeping gene for many expression studies, growing evidence has ascertained the differential expression of HPRT1 and its imperative role in quickly proliferating cells, such as neoplasms, due to elevated demand for nucleotides synthesis and consequently the HPRT1 during the cell cycle [14,15]. For instance, Townsend et al. [16] have shown the overexpression of HPRT1 in colorectal cancer samples compared with healthy tissues. Besides, others have demonstrated that the increased expression of HPRT1 in endometrial cancer samples was correlated with the survival outcomes of patients [17]. Moreover, Sedano et al. [18] also have discovered that up-regulation of the HPRT1 gene in tumor tissues could estimate the prognosis of patients with breast cancer. These findings provoked us to appraise the HPRT1 gene as a probable biomarker for patients with HNSCC.
In this study, we first analyzed the expression patterns of the HPRT1 gene in HNSCC using bioinformatics and laboratory investigations. We also assessed the diagnostic value and prognostic significance of the HPRT1 mRNA expression in patients with HNSCC. Besides, we conducted the mutation analysis for the HPRT1 gene in HNSCC. Finally, we explored those genes closely associated with HPRT1 in HNSCC and then performed functional enrichment analysis for these coexpressed genes to better understand the role of HPRT1 in HNSCC.

Materials and methods
Expression analysis using the TCGA-HNSCC cohort The University of California Santa Cruz (UCSC) Xena browser (https://xenabrowser.net/) is an online web tool for analyzing and visualizing multiomic data and related clinical and phenotypic annotations [19]. We downloaded the mRNA expression levels of the HPRT1 gene and the information of several clinicopathological parameters of 520 HNSCC samples and 44 normal tissues in The Cancer Genome Atlas (TCGA) cohort from data deposited in the UCSC Xena browser. The clinicopathological features of patients of this cohort are listed in Table 1.

Expression analysis using the Gene Expression Omnibus database
The Gene Expression Omnibus (GEO) database (https:// www.ncbi.nlm.nih.gov/geo/) is a valuable resource of available gene expression data that can be integrated and explored to derive new theories and knowledge [20]. We compared the HPRT1 mRNA expression in 23 HNSCC and 23 noncancerous tissues using a previously published dataset (GSE107591) [21] obtained from the GEO database. A P value <0.05 was considered statistically significant.

Sample collection
We obtained a total of 45 tumor tissues and paired adjacent normal tissues from 45 patients with HNSCC undergoing routine surgical procedures in the Amir Alam Hospital Complex in Tehran, Iran. It is noteworthy that none of the participants received treatment before the operation. The tissue samples were frozen immediately after surgery in liquid nitrogen and were stored at À80°C until  Table 3.

Immunohistochemistry staining
The Human Protein Atlas (HPA) database (https://www. proteinatlas.org/pathology) is the largest and most comprehensive repository for proteins in tissues and cells, supplying vital data for assessment of expression profile at a single-cell resolution [22][23][24]. We evaluated the expression of the HPRT1 gene at protein levels in four HNSCC tissues and one oral mucosa using the immunohistochemistry (IHC) staining data (antibody CAB012200) collected from the HPA database.

Prognosis analysis
The Kaplan-Meier plotter (http://kmplot.com/analysis) is an open-access resource that can be exploited to estimate the impact of 54 675 genes on survival outcomes of patients in 21 cancer types [25,26]. We obtained the information about the prognostic value of the HPRT1 mRNA expression in HNSCC through this database. A log rank P value <0.05 was considered statistically significant.

Mutation analysis
The Catalogue of Somatic Mutations in Cancer (COSMIC) database (http://cancer.sanger.ac.uk) is a website for studying the influence of somatic mutations in all types of human cancers [27]. We introduced this database to figure out the distribution and substitutions of the HPRT1 gene mutations in HNSCC. The data have been imported to Excel 2019 software to illustrate the pie chart. Besides, we investigated the genetic alterations of the HPRT1 gene and their impact on the survival outcomes of patients with HNSCC using the data from the cBio Cancer Genomics Portal (http://cbioportal.org) database, a freely available platform for cancer genomic datasets [28][29][30]. A log rank P value <0.05 was considered statistically significant.

Coexpression analysis
We investigated the coexpressed genes of the HPRT1 gene in HNSCC through the UALCAN database. Genes with extremely low expression (median transcripts per million < 0.5) were removed, and only genes with a |Pearson's correlation coefficient| ≥ 0.3 were involved. We then evaluated the expression profile of the collected genes in the GEPIA2 database (http://gepia2.cancer-pku.cn/#index), and only those genes that differentially expressed (Q < 0.01, |log2 fold change| > 1) were included for further analyses.

Functional enrichment analysis
We applied the Enrichr database (http://amp.pharm. mssm.edu/Enrichr) to carry out the Gene Ontology (GO)  and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis for coexpressed genes of the HPRT1 gene that differentially expressed in HNSCC. The Enrichr database is an extensive platform for curated gene sets and a search engine that gives biological awareness for further biological hypothesis [31]. The GO analysis contained biological processes (BP), molecular function and cellular component (CC). The retrieved terms were imported to Microsoft Excel 2019, and after selecting those terms with an adjusted P < 0.01, they were ranked based on the combined score. This index was produced by the Enrichr database and multiplied the log (Fisher's exact test P value) by z-score deviation from the expected rank. The bar chart for the top 10 terms in each category was illustrated using Microsoft Excel 2019.

Drug sensitivity analysis
To demonstrate whether aberrant expression of the HPRT1 gene could influence the clinical response to treatment and could be considered a potential biomarker for drug screening, we performed drug sensitivity analysis through the GSCALite database. This database has accumulated 481 and 265 small molecules from the Therapeutics Response Portal and Genomics of Drug Sensitivity in Cancer, respectively, and performs the Spearman's correlation analysis to evaluate the association of gene expression with drug sensitivity. The positive correlation suggests that the elevated expression of an interesting gene is resistant to the drug and vice versa.

Statistical analysis
To compare the relative expression of the HPRT1 gene among different groups, we performed a Student's t-test or one-way ANOVA tests as needed. Receiver operating characteristic (ROC) curves were illustrated to demonstrate the diagnostic utility of HPRT1 mRNA expression for patients with HNSCC. The area under the ROC curve (AUC) closer to 1.0 reflected that the test has the most diagnostic excellence. The maximum Youden index was used as a cutoff point. A P value <0.05 was regarded as statistically significant. GraphPad Prism Software, version 9.0.0 (GraphPad Software Inc., La Jolla, CA, USA), was used to perform statistical analysis.

Results
The mRNA expression levels of the HPRT1 gene in HNSCC We first investigated the transcript expression profiles of the HPRT1 gene. The RNA sequencing data retrieved from the UCSC Xena browser revealed that the mRNA expression levels of the HPRT1 gene were higher in HNSCC samples when compared with normal tissues (P < 0.0001; Fig. 1A). This was in accordance with our findings obtained from the GSE107591 dataset (P = 0.022; Fig. 1B). To confirm the results of our data mining, we assessed mRNA expression levels of the HPRT1 gene in 90 samples (45 from HNSCC tissues and 45 from adjacent normal tissues) using quantitative real-time PCR analysis. The results concluded that the transcript levels of the HPRT1 gene were apparently elevated in HNSCC compared with adjacent normal tissues (P < 0.0001; Fig. 1C).
The relationship between the HPRT1 mRNA expression and clinicopathological characteristics of patients with HNSCC We further assessed the association of HPRT1 mRNA expression levels with clinicopathological features of patients in TCGA-HNSCC and our validation cohorts. As presented in Fig. 2A-H, data determined that in comparison with normal control counterparts, the expression levels of the HPRT1 gene were higher in patients with HNSCC with different age, sex, pathological stage and histological grade. However, further analysis indicated the absence of remarkable variation in mRNA expression levels of the HPRT1 gene for patients with HNSCC of different ages ( Fig. 2A,E), different sexes (Fig. 2B,F), different stages (Fig. 2C,G) and different grades (Fig. 2D,H).

Prognostic value of HPRT1 mRNA expression for HNSCC
We evaluated the prognostic value of HPRT1 mRNA expression for HNSCC using the Kaplan-Meier Plotter database. The statistics suggested that higher levels of HPRT1 transcripts were correlated with the poor OS (P = 1.3eÀ5; Fig. 4A). However, there was no significant difference between relapse-free survival (RFS) in patients with HNSCC with high expression levels of HPRT1 mRNA compared with those with low expression levels (P = 0.22; Fig. 4B).

The expression levels of the HPRT1 protein in HNSCC
To check the expression levels of the HPRT1 gene at protein levels, we then acquired the relevant expression data using the IHC staining images from the HPA database. The findings showed negative HPRT1 staining in normal oral mucosa of one patient, and moderate and strong HPRT1 staining in tumor tissues of one and three patients with HNSCC, respectively ( Fig. 5A-E). Regarding the distribution, the expression of the HPRT1 protein is mainly noticed in the cytoplasm and cell membrane (Table 4).

Mutation analysis of the HPRT1 gene
The mutations of the HPRT1 gene in HNSCC samples were examined by the COSMIC database. The pie chart depicted that among four mutations in the HPRT1 gene, three were missense (75.00%) and one was synonymous (25.00%) substitutions (Fig. 6A). There were 50.00% A>G, 25.00% G>A and 25.00% G>C mutations in the HPRT1 gene coding region (Fig. 6B). We also explored the types of genetic alterations in the HPRT1 gene and their frequencies in 496 samples from the TCGA-HNSCC cohort using the cBioPortal database. As shown in Fig. 6C, the HPRT1 gene was altered in 7 (1%) cases of HNSCC, and the amplification was responsible for most changes. Moreover, using the "Survival" tab, we investigated the relationship between genetic alterations of the HPRT1 gene and survival times in patients with HNSCC. The Kaplan-Meier plot and log rank test uncovered that genetic alterations of the HPRT1 gene were not associated with the disease-specific survival (P = 0.666; Fig. 6D), disease-free survival (P = 0.194; Fig. 6E), progression-free survival (P = 0.300; Fig. 6F) and OS (P = 0.130; Fig. 6G) of patients with HNSCC.

Demonstration of the dysregulated coexpressed genes of the HPRT1 gene
To explore the potential roles of the HPRT1 gene in HNSCC, we first collected the coexpressed genes of the HPRT1 in HNSCC via the UALCAN database.
Our data mining unveiled that 1966 genes were correlated with HPRT1 in HNSCC. Figure 7A,B shows the top 20 genes with positive and negative correlation with HPRT1 in HNSCC, respectively. When we assessed the expression value of the acquired genes in the GEPIA2 database, we found that only 268 genes were the top five up-regulated (Fig. 7C), and SPINK5, SLURP1, HOPX, CRCT1 and CNFN were the top five down-regulated coexpressed genes of the HPRT1 gene in HNSCC (Fig. 7D).

GO and KEGG pathway analysis
Subsequently, we applied the Enrichr database to conduct the GO and KEGG pathway analysis for dysregulated coexpressed genes of the HPRT1 gene in HNSCC. The GO analysis unearthed that these genes     Table 7). The KEGG pathway analysis also indicated that these genes were enriched in DNA replication, cell cycle, homologous recombination, Fanconi anemia pathway, p53 signaling pathway, progesterone-mediated oocyte maturation, mismatch repair, oocyte meiosis, one carbon pool by folate and cellular senescence pathways (Table 8).

Drug sensitivity analysis
We investigated the correlation between the expression profile of the HPRT1 gene with drug sensitivity through the GSCALite database. The results have suggested that patients with cancer with high expression of the HPRT1 gene probably were sensitive to tozasertib, teniposide, manumycin A, clofarabine, GSK-J4, COL-3, CD-437, BRD-K01737880 and 3-Cl-AHPC and resistant to abiraterone (Fig. 8).

Discussion
The HPRT1 aberrant expression has been documented in several cancers [15]. Nevertheless, the relationship between the HPRT1 gene and HNSCC is unknown. To our knowledge, our report is the first to investigate the expression of the HPRT1 gene at transcriptomic and proteomic levels, as well as its genetic alterations, diagnostic and prognostic merit, and probable functions in the head and neck squamous cancer. We hope that the findings of the present research advance the current knowledge, enhance therapeutic targets and strengthen the accuracy of diagnosis and prognosis for patients with HNSCC. This study initially employed the data from the TCGA-HNSCC cohort and GSE107591 dataset to demonstrate the HPRT1 mRNA expression in HNSCC. The results showed that the HPRT1 mRNA expression was significantly up-regulated in cancer tissues compared with healthy control samples, which was consistent with the findings of our quantitative real-time PCR analysis for 45 paired HNSCC and normal tissues. We also assessed the expression levels of the HPRT1 gene in patients with HNSCC with different sex, age, pathological stage and histological grades in TCGA-HNSCC and our validation cohorts. The sex-based and age-based differences have contributed to the clinical presentations and prognosis of patients with HNSCC, respectively [32][33][34][35]. The sex-based analysis for both cohorts indicated the significant overexpression of HPRT1 gene level in tumor samples of both sexes compared with the normal tissues. However, there were no statistically significant differences between transcript levels of the HPRT1 gene in cancerous tissues of males and females. Besides, we observed no significant changes in the HPRT1 mRNA expression levels in the tumor tissues of patients with HNSCC of different ages in both cohorts. These may reflect that the patients' sex and age probably did not influence the expression levels of HPTR1 mRNA in cancer tissues. We also discovered that the HPRT1 mRNA expression levels were remarkably elevated in advanced pathological stages and histological grades of HNSCC tissues in comparison with healthy control subjects in both cohorts. These observations suggested that the elevated expression levels of the HPRT1 gene may take part in the pathogenesis of HNSCC. The IHC staining images from the HPA database have revealed the medium and high expression levels of the HPRT1 protein in HNSCC tissues. Altogether, these findings unearthed that expression of the HPRT1 gene was significantly increased in HNSCC tissues.
One of the most difficult challenges about patients with HNSCC is the time and procedure of screening and diagnosis. It has been declared that the delayed diagnosis and treatment of patients with HNSCC aggravate the prognosis outcomes and increase the undesired morbidity and mortality of patients with cancer. Hence findings of new biomarkers would boost the clinical outcomes of patients with HNSCC [36][37][38][39]. In the present research, we carried out the ROC curves analysis to appraise the diagnostic utility of the HPRT1 mRNA expression levels for HNSCC. The data from TCGA (AUC = 0.901), GSE107591 (AUC = 0.701) and our validation cohort (AUC = 0.806) concluded that expression levels of the HPRT1 gene could be considered as a helpful marker for the diagnosis of patients with HNSCC. Moreover, the prognosis data obtained from the Kaplan-Meier Plotter disclosed that the higher expression levels of the HPRT1 gene in cancer tissues were significantly correlated with inferior survival time of patients with HNSCC, in line with the findings of previous studies that suggested the dysregulation of the HPRT1 expression was associated with poor survival outcomes of breast and endometrial cancers [17,18]. In summary, these results may indicate the great value of the HPRT1 tissue expression for diagnosis and risk stratification of patients with HNSCC.
It is widely accepted that the accumulation of genetic alterations plays a causal role in the tumorigenesis of HNSCC [40]. The further assessments using the COSMIC and cBioPortal databases unveiled that genetic alterations of the HPRT1 gene were rare in patients with HNSCC (7/496), and the missense mutations were responsible for the most types of genetic aberration in the HPRT1 gene coding strand. Besides, the genetic alterations of the HPRT1 gene were not associated with survival outcomes of patients with HNSCC.
Finding dysregulated coexpressed genes of the HPRT1 gene in HNSCC may lead to understanding how HPRT1 and its correlated genes help the progression of HNSCC. In the present research, we uncovered that HPRT1 and its coexpressed genes that are differentially expressed in HNSCC probably were enriched in the DNA replication, cell cycle, homologous recombination, Fanconi anemia, p53 signaling and mismatch repair pathways. When we checked the included coexpressed genes and their expression value in DNA replication (RFC5, PRIM2, FEN1, POLA2, RNASEH2A, RFC4, MCM7, RPA3, POLE2, MCM3, MCM6), cell cycle (MCM7, PLK1, BUB1B, TTK, CDC6, CDC25C, CDC20, CCNA2, CCNB2, CCNB1, CDC45, CDK4, CHEK2, CHEK1, CDK2, CDK1, MCM3, MCM6, BUB1, MAD2L1), homologous recombination (BLM, RAD51, EME1, RPA3, XRCC3, RAD54L, BRCA1, RAD54B), Fanconi anemia (FANCI, BLM, RAD51, EME1, UBE2T, RPA3, FANCA, BRCA1), p53 signaling (CCNB2, CCNB1, RRM2, CDK4, CHEK2, CHEK1, CDK2, CDK1, GTSE1) and mismatch repair (RFC5, RFC4, EXO1, RPA3) pathways, we found that all of them were significantly up-regulated in HNSCC. Previous studies have documented the homologous recombination and Fanconi anemia pathways deficiency in HNSCC [41,42]. Besides, further investigations showed several genes in DNA replication, and cell-cycle pathways were duplicated in the p53 signaling (CCNB2, CCNB1, CDK4, CHEK2, CHEK1, CDK2, CDK1) and mismatch repair (RFC5, RFC4, RPA3) pathways. Moreover, several pieces of evidence have reported that upregulation of DNA replication and cell-cycle pathways genes in HNSCC [43][44][45]. Others have proposed that overexpression of the RRM2, GTSE1 and EXO1 genes in HNSCC have active roles in tumor progression and inhibition of the apoptosis pathway [46][47][48][49]. Some researchers have suggested that the HPRT1 gene, as a crucial component of the purine salvage pathway, plays a role in mediating the proliferation, autophagy and apoptosis-related processes in cancer cells [50][51][52]. IGF2BP2 and SPINK5 were the top up-regulated and down-regulated coexpressed genes of the HPRT1 gene in HNSCC. Deng et al. [53] have determined that IGF2BP2 is a prognostic gene and has vital roles in the progression of HNSCC. Others also have recognized SPINK5 as a novel tumor suppressor that suppresses migration and invasion of HNSCC cells [54,55]. These findings imply that HPRT1 and its associated gene likely mediate the progression of HNSCC through activation of the DNA replication and cell cycle and inhibition of apoptosis. Finally, the data obtained from the GSCALite database indicated that patients with increased expression of the HPRT1 gene may not respond to treatment with abiraterone very well and hopefully are responsive to therapies that consist of tozasertib, teniposide, manumycin A, clofarabine, GSK-J4, COL-3, CD-437, BRD-K01737880 and 3-Cl-AHPC. Our study had some limitations: 1 As we stated in the introduction, HNSCCs are heterogeneous and can develop at different locations, such as the oral cavity, pharynx and larynx [4]. When we aimed to classify patients with HNSCC of TCGA, GSE107591 and our validation cohorts into oral or laryngeal squamous cell carcinoma, the number of patients in each group was few, and we could not perform meaningful statistical analyses. Therefore, we decided to consider all of them as HNSCC. 2 Although the expression analysis for the TCGA cohort demonstrated the absence of significant differences in the expression levels of the HPRT1 gene for patients with HNSCC who were human papillomavirus (HPV) positive in comparison with those with HPV negative (data not shown), we could not find any data about the HPV history of patients with HNSCC in GSE107591 and our validation cohorts. 3 The samples in the HPA database were few. 4 Although we determined that the expression levels of the HPRT1 gene were increased in HNSCC, the underlying mechanisms for this phenomenon remained unclear. 5 In this report, we uncovered that the HPRT1 gene and its dysregulated coexpressed genes likely mediate HNSCC progression through the cell cycle and apoptosis-related pathways. However, the exact function of the HPRT1 gene in these pathways remained ambiguous. 6 Except for mRNA expression analysis, other findings were not confirmed with reliable laboratory analysis, and we just focused on bioinformatics investigations. Relying on bioinformatics assessments may induce deviations.
Therefore, additional studies with appropriate sample size are required to explore: (a) the expression pattern of the HPRT1 gene in HNSCCs, such as oral squamous cell carcinoma (OSCC) -and laryngeal squamous cell carcinoma (LSCC); (b) the association between the expression levels of the HPRT1 gene and HPV status of patients with HNSCC; (c) the potential molecular etiologies for up-regulation of the HPRT1 gene in HNSCC; and (d) the exact underlying mechanisms in which the HPRT1 gene promotes the progression of HNSCC.

Conclusions
Overall, the findings of this research indicated that overexpression of HPRT1 is significantly correlated with the progression of HNSCC. Besides, the observed up-regulation of the HPRT1 expression in tumor tissues may be a valuable biomarker for diagnosis, prognosis and targeted treatment of patients with HNSCC. However, further investigations and clinical trials are mandatory to elucidate the involvement of HPRT1 in HNSCC.
cohort used for this research are available from the corresponding author upon request.