GALNT14 expression in renal clear cell carcinoma and its effect on prognosis

Background The expression of GALNT14 in kidney renal clear cell carcinoma (KIRC) and its clinical signicance remains unknown. Methods The KIRC data expressed by GALNT14 was downloaded from The Cancer Genome Atlas (TCGA) database. The expression of GALNT14 was analyzed by R software, Perl software and online analysis database. The relationship between GALNT14 expression and clinicopathological features in KIRC was analyzed by univariate, multivariate Cox regression and some databases. Gene Expression Proing Interactive Analysis (GEPIA), Starbase v3.0, UALCAN, and Kaplan-Meier were used to analyze the relationship between GALNT14 expression and overall survival (OS) in KIRC. UALCAN detects the expression of GALNT14 methylation in KIRC. Linkedomics and Genemania were used to analyze the gene co-expression of GALNT14. Gene Set Enrichment Analysis (GSEA) was performed to search for potential regulatory pathways. Results We found that GALNT14 was overexpressed in KIRC (p=1.433e-25). Patients with high GALNT14 expression in KIRC had a better prognosis than patients with low GALNT14 expression (p=0.008). In addition, high GALNT14 expression in KIRC was signicantly associated with low T stage and positive OS (p<0.05). Univariate Cox analysis showed that GALNT14 was positively correlated with OS (p<0.001). Multivariate Cox analysis showed that GALNT14 was associated with OS (p<0.001), age (p=0.01) and histological grade (p=0.02). GALNT14 methylation is low expressed in KIRC (p<0.001). GSEA analysis showed that GALNT14 was enriched in histidine metabolism, peroxisome, and renin-angiotensin system pathways. Conclusion GALNT14 can be used as an independent prognostic for renal clear cell and a target for clinical and of KIRC.

Besides, GALNT14 has been identi ed as a destructive Mendelian mutation that led to the death of embryos, suggesting that GALNT14 plays an irreplaceable role in human development [9]. Otherwise, GALNT14 is abnormally expressed in multiple cancers. It has been reported that GALNT14 can change the catalytic e ciency of apoptosis signal and tissue invasiveness, regulate the migration of cancer cells, etc., and play a great role in malignant tumors of the breast, lung, gastrointestinal tract and ovary [10,11]. Together, we know that GALNT14 plays a unique role in the progression of malignant tumors. Given the above, most published studies have focused on the role of GALNT14 in cancers of the respiratory system and digestive system [2,4,5]. However, to date, no studies have examined the effect of GALNT14 expression in KIRC.
GALNTs have 20 members in total: GALNT1 through 14, GANLTL1 through L6, and GALNT14 is one of them [10,12]. GALNT is an inducer of mucin O-glycation, and its expression can alter many biological behaviors of tumor cells, including formation, proliferation, migration and drug resistance [10,13].
Nevertheless, other members of the GALNTs also play a role in cancer [14]. For example, Muhammad et al.
[8] proposed that the abnormal glycation caused by the mutation of GALNT1 was associated with the occurrence of bladder cancer, and GALNT3 was highly expressed in RCC and prostate cancer in patients with high malignant tumors. In addition, GALNT4 is a positive prognostic factor in patients with DFS KIRC, and high GALNT10 expression is associated with TNM stage progression in KIRC [8,15]. In summary, we speculate that GALNTs are involved in the occurrence, development, and metastasis of cancers in various systems. However, no comprehensive bioinformatics analysis has been reported to investigate the expression and clinical signi cance of GALNT14 in KIRC. Here, we demonstrate for the rst time that high GALNT14 expression in KIRC is associated with lower T stage and higher OS. Our study not only provides a new independent prognostic factor for KIRC, but also reveals the pathway by which GALNT14 regulates the occurrence and development of KIRC.

The cancer genome atlas (TCGA) database
We downloaded a total of 611 primary KIRC tumor tissue samples with detailed GALNT14 expression data and 537 clinical data from KIRC tumor patients from TCGA (https://portal.gdc.cancer.gov/).

Clinical correlation analysis--Kolmogorov-Smirnov test
First, we examined the correlation between GALNT14 and the clinical characteristics of patients with KIRC in Perl software. GEPIA and UALCAN were used to investigate the expression characteristics of GALNT14.
Logistic regression analysis was used to investigate the clinical signi cance of GALNT14 in KIRC. Cox analysis was performed using the "Survival" and "SurvMiner" software packages of R software.

Survival analysis
We combined Perl software, survival analysis and Kaplan-Meier prognostic analysis, and used R software to calculate p-value by log-rank test. In addition, Kaplan-Meier overall survival curves and log-rank were used to examine the OS differences between GALNT14 expression groups. Meanwhile, GEPIA, Starbase v3.0, UALCAN and Kaplan-Meier Plotter (http://kmplot.com/analysis/) were used to detect GALNT14 expression and prognosis (OS or DFS) in KIRC, KIRP and KICH.

Methylation analysis
We used UALCAN to detect the expression of GALNT14 methylation in KIRC and its relationship with clinical features.

Analysis of gene co-expression
In order to explore the genes associated with GALNT14 in different molecular functions and biological pathways, we used Linkedomics (http://www.linkedomics.org/) and Genemania (http://genemania.org) to analyze gene co-expression. The proteins interacting with GALNT14 were obtained from the STRING database (https://www.string-db.org/).

Gene Set Enrichment Analysis
KEGG ( http://www.genome.jp/kegg/) was used to analyze the enrichment of GALNT14-related Genes in KIRC. Each GSEA analysis performed 1000 genome sequences.

Characteristics of the study population
Clinical data of 537 patients were downloaded from TCGA in September 2020. According to the age classi cation standards of the United Nations WHO, the age groups are divided into young people from 18 to 65 years old, middle-aged people from 66 to 79 years old and old people from 80 to 99 years old (Table 1). First, we studied the expression of GALNT14 in KIRC by differential expression analysis. GALNT14 expression in KIRC and normal tissues were compared in Fig. 1A, and the results indicated that GALNT14 expression was elevated in KIRC(p = 1.433e-25). Similarly, paired differential analysis showed that GALNT14 was highly expressed in KIRC tissues (p = 1.752e-18) (Fig. 1B). Additionally, the results of GEPIA, Starbase v3.0 and UALCAN all showed that GALNT14 was increasingly expressed in KIRC (Fig. 1C, F, G). Accordingly, all the results illustrated high expression of GALNT14 in tumor samples. Moreover, we found high GALNT14 expression in human renal carcinoma cell line A498 and ACHN through CCLE database. To determine whether the high expression of GALNT14 in KIRC is speci c, we also analyzed the expression of KIRP and KICH by GEPIA, and found that GALNT14 was low and high in KICH and KIRP respectively, and the results were statistically signi cant. (Fig. 1D, E).

Related clinicopathologic variables
To investigate whether GALNT14 is involved in the clinical progression of KIRC, we downloaded data from a total of 537 clinical samples of KIRC with GALNT14 expression from TCGA. As shown in Fig. 2, differences in GALNT14 expression were observed according to clinical stage(p = 0.009), T classi cation(p = 4.104e-05), M classi cation(p = 0.005) and distant metastasis(p = 3.988e-05). Univariate logistic regression analysis showed that GALNT14 was associated with clinicopathologic features with a favorable prognosis ( Table 2). Subgroup analysis indicated that high GALNT14 expression in KIRC was signi cantly associated with low T stage and high OS (p < 0.05). Likewise, UALCAN results demonstrated that GALNT14 was upregulated in stage, grade, and lymph node metastasis ( Fig. 3A-E). Namely, the higher level of the grade, the less the expression of GALNT14 (p < 0.05) ( Table 3). Besides, the expression of GALNT14 in KIRC manifested a signi cant gender difference(p = 2.21E-02) (Table 3). Similarly, to determine whether GALNT14 is involved in the clinical process of KICH and KIRP, the results of GEPIA analysis of the clinical stage of KICH and KIRP showed no difference in GALNT14 expression in KICH, while there was a difference in GALNT14 expression in KIRP (Fig. 3F, G).   Fig. 4E indicated that KIRC patients with high GALNT14 expression had a higher OS (p = 0.008). Besides, the GEPIA results showed that GALNT14 was not statistically signi cant in OS in KICH, but higher GALNT14 expression was associated with higher OS in KIRP (p = 0.012), which was consistent with the result obtained by Kaplan-Meier Plotter(p = 0.023) ( Fig. 4F-H). Finally, the forest diagram showed the correlation between GALNT14 and OS (Fig. 4I). In all, GALNT14 was found to be signi cantly associated with patient OS in KIRC.

Cox regression analyses
Moreover, combining with the results of the clinicopathological characteristics of survival status, the univariate Cox analysis revealed that GALNT14-high correlated signi cantly with a positive OS in KIRC (p < 0.001). Other clinicopathologic variables associated with better OS included clinical stage, lymph nodes, and distant metastasis (Table 4). Similarly, multivariate Cox analysis showed that GALNT14 was associated with OS (p < 0.001), along with histological grade(p = 0.02) and age(p = 0.01). To summarize, Cox regression analyses showed that GALNT14 was an independent prognostic factor as compared with other confounders.

The prognostic values of GALNT14 for DFS in KIRC patients
Equally, GEPIA and Kaplan-Meier Plotter were used to analyze the relationship between the expression of GALNT14 and patient DFS in KIRC. Data from GEPIA showed that high expression of GALNT14 predicted poor DFS of patients with KIRC (Fig. 5A). Kaplan-Meier Plotter results also showed that high GALNT14 expression was associated with low DFS in patients with KIRC (p = 0.0091) (Fig. 5B). Thereby, we can know that the up-regulation of GALNT14 was signi cantly associated with poor DFS in KIRC patients. In addition, the GEPIA results showed no statistically signi cant difference in the expression of GALNT14 in DFS in KICH patients (Fig. 5C), while high expression of GALNT14(p = 0.033) predicted higher DFS of patients with KIRP (Fig. 5D). But there was no statistical difference in the expression of GALNT14 in DFS of KIRP by Kaplan-Meier Plotter (Fig. 5E).
3.6 Methylation of GALNT14 in UALCAN Correspondingly, we used UALCAN to determine the differential expression of GALNT14 methylation and its relationship to clinical features in KIRC. The results demonstrated that GALNT14 methylation levels in KIRC were lower than those in normal tissues (p < 0.001) ( Table 5). Moreover, speci c patterns of DNA methylation differed in indicators including tumor grade and clinical stage, as shown in Fig. 3H-M and Table 5.
3.7 GALNT14 functions as a protection factor in KIRC Given that GALNT14 was signi cantly up-regulated in KIRC and closely correlated with patient prognosis including both OS and DFS, we speculated that GALNT14 might act as a protective factor in KIRC. Furthermore, LinkedOmics and GeneMANIA were used to conduct gene co-expression analysis on GALNT14. In this way, we haven't ignored the high correlation between genes[16]. The result of gene co-expression network analysis of GALNT14 in KIRC is shown in (Fig. 5F). Meanwhile, in the STRING database, we obtained 10 proteins that interact with GALNT14 (Fig. 5G). The overall closely co-expressed genes of GALNT14 in KIRC are shown as a volcano plot (Fig. 5H). The top-50 positively and negatively co-expressed genes of GALNT14 in KIRC are shown as heat maps (Fig. 5I, J), respectively.   Table 6. The results indicated that GALNT14 was highly differentially expressed in histidine metabolism, peroxidase and renin-angiotensin system pathways (Fig. 6). Moreover, to make the results of gene enrichment analysis more obvious, we combined the high-expression phenotypic pathway of GALNT14 with the low-expression phenotypic pathway of GALNT14 (Table 6) obtained through GSEA (p < 0.05) to conduct GALNT14, and the results are shown in Fig. 6J.

Discussion
First, differential expression analysis showed that GALNT14 was highly expressed in KIRC tissues, and this was veri ed in the CCLE database. The results of the clinicopathological analysis showed that GALNT14 expression differed signi cantly in clinical stage, T stage, distant metastasis, and survival status. In addition, the UALCAN results showed that GALNT14 methylation was under-expressed in KIRC.
The OS outcome of the Kaplan-Meier curve also showed that high GALNT14 expression was associated with better outcomes in patients with KIRC. In addition, univariate and multivariate Cox analyses suggested that GALNT14 may be a biomarker for the prognosis of KIRC. Subsequently, Genegen network analysis demonstrated that GALNT14 was associated with multiple MUC genes. Finally, the results of GSEA revealed that GALNT14 overexpression regulates KIRC mainly through histidine metabolism, peroxidase, renin-angiotensin system and other pathways. In conclusion, GALNT14 may serve as a potential marker for prognostic and therapeutic targets of KIRC.
Our work is the rst to suggest that GALNT14 is a KIRC protective factor. The results of differential expression analysis and Cox regression analysis showed that GALNT14 high expression was signi cantly different in clinical-grade, T stage and distant metastasis. Further, the overexpression of GALNT14 is associated with a low T grade. In other words, GALNT14 upregulation inhibited the size and extent of the primary KIRC tumor. In the stage, Hillman et al. [2] proposed that GALNT14 is expressed in several cancers, including ovarian cancer, and its upregulation promotes clinical progression in highgrade serous cancer. Beyond that, most patients with KIRC are diagnosed with metastases, particularly to the lung, bone, and brain [17,18]. We speculated that the high expression of GALNT14 in KIRC may affect the biological functions of tumor cell proliferation, invasion and metastasis.
In addition, we found that GALNT14 expression was signi cantly associated with survival in patients with  [9,19] noted that the GALNT14 "TT + TG" genotype resulted in a longer OS in patients. Besides, Kwon et al. [20] indicated that patients with GALNT14 overexpression had poor DFS. What's more, Lin et al. [10] suggested that in patients with positive anti-HCV antibodies, the GALNT14 TT genotype was associated with longer OS. Therefore, we suggested that GALNT14 overexpression may affect the progression of cancers. Unfortunately, the prognostic value of GALNT14 expression in lung cancer and HCC is known [20,21], but its prognostic value in other cancers, including renal cancer, remains unknown.
Our Cox analysis suggests that GALNT14 may serve as an independent good prognostic biomarker for KIRC. Indeed, GALNT14 is not the only protective factor for KIRC. Zhang et al. [22] indicated that ACE2 is one of the protective factors of KIRC, and ACE2 down-regulation is associated with poor OS and tumor progression in KIRC. Correspondingly, there are other genes that may act as protective factors for KIRC. For example, Qiu et al. [23] proposed that the high expression of PHYH improves the survival of patients with KIRC mainly by affecting peroxisomes and may serve as a potential prognostic molecular marker. In brief, there is not only one protective factor in each cancer. Absolutely, protective factors are not just present in KIRC or in a particular type of cancer, but in a wide range of cancers. For instance, Jiang et al. [24] found that overexpression of PTGD was positively correlated with early TNM, suggesting that this gene may be a protective prognostic factor for breast cancer. Moreover, the survival rate of cervical cancer patients with high KIR3DL1 transcription levels in tumor tissues is signi cantly higher than that of cervical cancer patients with low/moderate transcription levels, suggesting that KIR3DL1 is a protective factor of cervical cancer [25]. Collectively, protective factors exist not only in cancers of the urinary system, but also in cancers of other systems. Thus, we propose that the most intuitive manifestation of cancer affecting protective factors is to inhibit the metastasis and spread of cancer cells, thus improving the OS of cancer patients.
According to univariate and multivariate analysis, GALNT14 was predicted to be a novel biomarker in KIRC samples. GALNT14-mediated protein glycosylation has been reported to be associated with tumor development and malignant transformation [20,26]. In addition, Wang et al. [27] pointed out that cancerspeci c glycation changes are mainly based on the abnormal expression and activity of related glycosyltransferases, which provides theoretical support for our belief that the glycation of KIRC is mainly based on the abnormal expression and activity of GALNT14. We know that GALNT14 encodes the glycosyltransferase GalNA-T14, which is involved in O-chain glycosylation after protein translation [26]. Oglycosylation is a type of glycosylation that plays an important role in cell growth, differentiation, transformation, adhesion and tumor immune monitoring [27][28][29]. In detail, DeDarya et al. [13] proposed that abnormal expression of GALNTs would lead to abnormal glycation of cancer cells, which would interfere with cell adhesion, migration and proliferation and promote tumor development. Besides, in melanoma cells, non-small cell lung cancer, and pancreatic cancer, high GALNT14 expression enhances O-glycation of DR4 and DR5, thereby reducing tumor cell sensitivity to TRAIL [30]. Hence, GALNT14 overexpression interacts with glycosylation to in uence cancer progression in various systems of the body. In conclusion, we suggest that GALNT14 affects a variety of tumor cell functions, including proliferation, apoptosis, invasion, and metastasis, by altering glycation, and further affects the genesis and development of KIRC.
Moreover, by UALCAN analysis, we found that GALNT14 methylation was down-expression in KIRC, and different DNA methylation patterns differed in pathological grade, clinical stage and other clinical indicators. To our knowledge, DNA methylation has been identi ed as a potential target for diagnostic, prognostic and predictive biomarkers in cancer [31,32]. Nevertheless, the PubMed database of 14 biomarkers used to measure gene methylation does not include GALNT14 [31]. Thus, the correlation between GALNT14 methylation and the clinicopathological characteristics of KIRC explored in this study is the rst to date, which will provide a reference for subsequent related studies. Altogether, methylation markers play an important role in the diagnosis and prediction of survival of solid tumors [33][34][35], which is why we explored the relationship between GALNT14 methylation and KIRC in this study. To our surprise, Chen et al. [3] proposed that the hypomethylation of oncogenes plays an indispensable role in the carcinogenesis process. In addition, Li et al. [33] proposed that the abnormal methylation contributing to cancer is common in all types of cancer. Interestingly, abnormal enhancer hypermethylation has been shown to be a predictor of poor survival in patients with KIRC [31,36], which could fully explain the role of GALNT14 as a protective factor in KIRC. Similarly, studies have shown that the increase of hypermethylation is associated with a higher stage and grade of KIRC [3,37], which is consistent with our nding. Therefore, we estimate that GALNT14 methylation may affect its function and thus contribute to the early diagnosis and prognosis of KIRC.
Finally, we used GSEA analysis to identify the pathway by which GALNT14 affects KIRC. The results showed that GALNT14 was highly expressed in histidine metabolism, peroxisome, renin-angiotensin system and other pathways. Our results are similar to those of Wang et al.
[38], who identi ed three important pathways associated with KIRC, namely, arginine and proline metabolism, aldosteroneregulated sodium reabsorption, and oxidative phosphorylation. This suggests that GALNT14 plays an important role in the metabolic pathway of the KIRC enrichment pathway, which is consistent with the physiological function of the kidney. Previous studies have suggested a strong link between cancer and metabolism, with one even pointing to kidney cancer as a disorder of cell metabolism [16]. Recent research has shown that the characteristic that distinguishes KIRC from other types of tumors is its speci c metabolic changes [39]. In terms of amino acid metabolism, it has been reported that the deletion of the 3P chromosome in clear cell renal cell carcinoma leads to the inactivation of FHIT gene [40], so we speculate that this may be the reason why the histidine metabolic pathway is in the rst place. Besides, Weiss et al. [41] proposed that tryptophan is catabolized by the renin pathway and that the renin metabolites have immunosuppressive effects. In addition, Liu et al. [5] have shown that tryptophan depletion leads to apoptosis through an accumulation of KP metabolites. Therefore, we speculate that GALNT14 may reduce tryptophan metabolism to maintain the immune function of patients, thereby limiting the proliferation of KIRC primary tumor. Then, in terms of fatty acid metabolism, Hillman et al. [2] showed that KIRC is associated with abnormal fatty acid metabolism. Interestingly, reduced fatty acids in KIRC were associated with increased tumor invasion and poor prognosis [41]. This suggests that the high expression of GALNT14 in KIRC may inhibit the expression of fatty acid synthase, leading to the higher OS in patients with GALNT14 upregulation. Surprisingly, there are also reports of dysregulation of carbon metabolism in KIRC[6]. This is basically consistent with our ndings in KIRC enrichment analysis that GALNT14 overexpression affects glucose metabolism. KIRC has been reported to enhance uptake of glucose transporters, leading to glucose accumulation and further promoting tumor progression [39], so inhibition of gluconeogenesis will promote the development of RCC. In a word, we suggest that GALNT14 overexpression in KIRC reduces the absorptive activity of glucose transporter and accelerates glucose decomposition. Hence, we hypothesized that GALNT14 regulates the occurrence of KIRC through the above metabolism-related pathways. However, the prognosis of high GALNT14 expression is limited to KIRC and IV stages, and further experiments are needed.

Conclusion
Consequently, GALNT14 can be served as an independent prognostic factor for KIRC and a potential target for clinical diagnosis and treatment of KIRC.