1. IntroductionHead and neck squamous cell carcinoma (HNSCC) contributed to the death of 450,000 people worldwide in 2018, which makes it the seventh most severe cancer. HNSCC is mainly associated with tobacco and alcohol abuse. However, human papillomavirus (HPV) infection, mostly with HPV-16, also appears as a crucial etiologic factor in HNSCC development [1,2]. The characteristics of HNSCC are a poor response to treatment and high mortality, where only about 50–60% of patients reach the 5-year survival rate. Thus, there is an urgent need to develop novel, more effective, personalized therapies and specific prognostic biomarkers which are based on genes with protein-coding and non-coding abilities [3,4,5,6]. However, knowledge of the exact molecular mechanisms driving HNSCC is still limited.Zinc finger proteins (ZNFs) constitute the most numerous family of sequence-specific DNA-binding proteins encoded by 2% of human genes. They bind to their target DNA sequences through the zinc finger domain [7] and exert various functions, including transcriptional regulation, signal transduction, or protection against DNA double-strand breaks [8,9,10]. They may interact with DNA sequences, RNAs, proteins, and post-translational modifications [8,9,10]. ZNFs are divided into several subgroups based on their structural conformation, and C2H2 ZNFs are the most common [7,11]. Besides various zinc finger motifs, the C2H2 class contains additional domains involved in gene expression or cellular localization—such as the KRAB (Krüppel-associated box), SCAN, or BTB/POZ domain [12,13]. Due to the considerable complexity within the ZNF family, little is known about the exact molecular function of most of its members. Of note, many KRAB-ZNFs were shown to play an essential role in carcinogenesis, acting as oncogenes, suppressors, or both, depending on the cancer type [14]. The association with tumor biology was already described for several ZNFs in various cancers, including melanoma [15], colorectal [16,17], renal [18], gastric [19], and esophageal cancers [20], or lung adenocarcinomas [13]. Numerous KRAB-ZNFs show altered expression in various tumors, e.g., HNSCC, as was demonstrated in the transcriptomic profiling based on the TCGA datasets [21]. Nevertheless, the contribution to biological processes and the potential diagnostic utility of specific ZNFs in HNSCC remain undefined. Moreover, there isstill no data on ZNFs’ involvement in head and neck cancers with HPV origin.For this study, based on the preselection with the UALCAN database [22], we chose six ZNF genes: ZFP28, ZNF132, ZNF418, ZNF426, ZNF540, and ZNF880. In our previous analysis, these factors were shown to be downregulated in multiple tumor types, including HNSCC [21]. Moreover, ZNF132 was reported to be epigenetically inactivated in laryngeal squamous cell carcinoma due to promoter hypermethylation [23,24]. In the same tumor type, ZNF418 promoter methylation was demonstrated as a potent diagnostic factor distinguishing between high- and low-risk groups of patients [25]. To the best of our knowledge, no other report has been published to date describing the involvement of ZFP28, ZNF132, ZNF418, ZNF426, ZNF540, and ZNF880 in HNSCC. Here, we hypothesize that these genes may be implicated in HNSCC biology and related HPV phenotypes. To test this hypothesis, we used the TCGA data and performed bioinformatics analyses of mRNA expression. We aimed to explore the correlation of ZNF expression with clinico-pathological parameters, their engagement in various cancer-associated processes, and their potential role as biomarkers in HNSCC. 4. Discussion

The zinc finger proteins (ZNFs) are one of the most abundant proteins encoded in the human genome. However, due to the vast complexity of this large family of transcriptional factors, the exact roles of ZNFs are still unexplored. In this study, we analyzed ZFP28, ZNF132, ZNF418, ZNF426, ZNF540, and ZNF880 in HNSCC, focusing on their biological role, association with various clinico-pathological parameters, and potential utility as biomarkers. The analysis was carried out using the TCGA data, followed by the validation with an alternative dataset from GEO.

First of all, our results based on the single-gene approach show that the expression of all analyzed ZNFs was lower in HNSCC samples compared to healthy controls, which is a favorable feature for diagnostic biomarkers. This approach confirmed the outcomes of our previous pan-cancer transcriptomic analysis utilizing the TCGA data [21]. Surprisingly, we found that the cross-correlation between the expression of ZNFs was only marginal. Of note, our ROC curve test showed a very good capacity for each ZNF to discriminate between tumor and normal samples, further pinpointing their potential applicability as biomarkers.Although none of the existing studies explored the above transcripts in more detail in HNSCC, our data align with other published observations. For example, ZNF132 was shown to be epigenetically silenced via promoter hypermethylation in HNSCC [23,24], esophageal squamous cell carcinoma (ESCC) [20], and lung adenocarcinoma (LUAD) [29]. Phenotypically, the cells with reduced ZNF132 expression had decreased mobility in LUAD [29] and growth, migration, invasion, and tumorigenicity in ESCC [20]. These observations suggest the tumor suppressor function of ZNF132. High promoter methylation was also reported in the case of ZNF418 in laryngeal squamous cell carcinoma [25] and ZNF540 in clear cell renal cell carcinoma [18]. Moreover, the study by Hui et al. indicated that ZNF418 was significantly downregulated in gastric carcinoma patients [19]. In contrast, ZNF880 and ZFP28 were found upregulated in colorectal cancer [16] and melanoma [15], respectively. Interestingly, promoter hypermethylation frequently leads to the epigenetic inactivation of ZNF genes with the TSG features in various cancer types [14]. Thus, it is likely that at least some of the above-analyzed ZNFs may also become downregulated via the CpG methylation mechanism in HNSCC, and this possibility warrants further studies.We further investigated whether the mRNA expression of selected ZNFs may differ depending on various clinico-pathological parameters. We observed particular similarities between ZFP28, ZNF540, and ZNF132 signatures. First, we found that the expression level of most ZNFs (apart from ZNF880) depended on tumor location. The pharynx was the site of the highest expression for ZFP28, ZNF540, and ZNF132, and the lowest for ZNF426. Of note, ZNF132 and ZNF540 expression differed in all three anatomical sites: oral cavity, larynx, and pharynx. Secondly, ZFP28, ZNF540, and ZNF132 were downregulated in tumors with a higher T stage, in older patients, and in the cohort that underwent lymph node dissection from the neck. In contrast, these factors demonstrated an increased expression in high-grade tumors (G3 + G4). For further explanation of whether these ZNFs were associated with more aggressive forms of HNSCC, we analyzed their expression profile in different molecular subtypes. We observed that all examined ZNFs were upregulated in the mesenchymal and less aggressive, atypical [30] tumors.According to our knowledge, there are no comprehensive studies on the expression of ZNFs in the context of HNSCC and related risk factors. Here, we demonstrate that a higher ZFP28 expression was associated with alcohol consumption, whereas smoking was related to higher ZFP28, ZNF880, and ZNF418 levels and lower ZNF132 expression. Moreover, we observed that ZNF132 and ZNF540 were upregulated, and ZNF426 was downregulated in the HPV(+) group compared to the HPV(−) group. Although ZNF132 expression and promoter methylation were analyzed previously in HPV(+) HNSCC cases, no association with HPV was demonstrated [24]. Such a discrepancy between our study and [24] may reflect ethnic differences between populations analyzed or may be due to the lower number of patients included in [24].Since their biological roles remain largely uncharacterized, we further sought to determine ZNFs’ involvement in tumor-associated pathways. Using REACTOME and GSEA tools, we confirmed the correlation of ZNFs with various signaling pathways engaged in tumorigenesis and immune responses. In summary, those pathways include MAPK, NF-κB, TNF, JNK, and RAS signaling. For example, ZNF418 and ZNF540 expression was linked to KRAS signaling. In addition, ZNF540 expression positively correlated with defective base excision repair associated with OGG1 and with the expression of the TNF receptor superfamily (TNFSF) that mediates non-canonical NF-κB signaling, which is essential for immune response and cell growth regulation [31]. Our data also indicate that both ZNF540 and ZNF418 are associated with IL-2 signaling (and IL-15 and IL-21 in the case of ZNF418). Moreover, we found that both factors were related to altered immunological profiles in HNSCC patients. Thus, it may be hypothesized that the decreased expression of ZNF540 and ZNF418 may affect tumor formation not only through various oncogene-related pathways but also via interfering with the immune response.Furthermore, our study, for the first time, demonstrates that ZNF418 and ZNF540 expressions could be used as potential biomarkers in HNSCC. Notably, the patients with increased levels of ZNF418 showed significantly longer OS, while those with higher ZNF540 expression had prolonged disease-free intervals. Other reports may indirectly support our findings. For example, high ZNF418 expression correlated with improved OS in gastric carcinoma [18], whereas its promoter hypermethylation showed a good discriminatory potential between high- and low-risk patient cohorts [25]. Additionally, Arai et al. determined that ZNF540 is frequently methylated in clear renal cell carcinoma patients with worse survival [18]. Of note, our data reveal a link between ZNF540 expression and NF-κB, MAPK, and JNK pathways, which contribute to the epithelial-to-mesenchymal transition (EMT) [32,33]. Hypothetically, ZNF540 may interfere with the EMT events, thus lowering the chance of local and distant metastases and improving the prognosis of HNSCC patients. Nevertheless, these hypotheses require further clinical and wet-lab investigation.Importantly, our detailed analysis of the potential clinical usage of ZNFs revealed that ZNF540 expression might serve as a prognostic marker in the context of HPV infection. Based on the TCGA and GEO data, we showed that the ZNF540 level is higher in HPV(+) patients than in HPV(−) patients. Moreover, we indicated that a higher expression of ZNF540 is observed in patients with an active HPV infection. So far, only one study revealed that ZNF540 is upregulated in HPV(+) active vs. HPV(+) inactive patients, as well as HPV(+) active and HPV(+) inactive in comparison to HPV(−) HNSCC patients [34]. However, no in vitro studies describe the biological role of ZNF540. As mentioned previously, the highest ZNF540 expression was observed in the HNSCC molecular subtype characterized as “atypical”. It was indicated that the atypical subtype was a less aggressive type of HNSCC and was associated with a strong immune signature [35]. Based on our immunological profile results, we observed that patients with higher levels of ZNF540 displayed higher levels of CD8, follicular helper T cells, memory and naive B cells, and lower levels of M2 macrophages. Cillo et al. described the immune landscape of viral- and carcinogen-driven HNSCC and indicated that a higher level of follicular helper T cells was associated with longer progression-free survival [35]. We also observed in the TCGA data that patients with higher levels of ZNF540 displayed longer disease-free intervals. While the TCGA dataset comprised all patients, HPV(+) and HPV(−), further analysis based on the GEO dataset clearly showed that the HPV(+) patients with higher levels of ZNF540 had significantly longer overall survival.

In conclusion, we showed that ZFP28, ZNF132, ZNF418, ZNF426, ZNF540, and ZNF880 had reduced expression in HNSCC compared to healthy tissues. Moreover, their expression levels were associated with various clinical parameters, risk factors, and signaling pathways crucial for tumorigenesis and immune responses. We revealed that the high expression of ZFP540 and ZFP418 correlated with a favorable prognosis in HNSCC. Specifically, high ZFP540 levels were associated with improved survival of HPV(+) patients. Altogether, our findings emphasize the potential applicability of ZNF418 and ZNF540 as prognostic biomarkers in HNSCC. These promising data open new avenues for additional research to dissect the mechanisms responsible for ZNF downregulation. The limitation of our study is that it is based on the TCGA and GEO data, where we had no control over the quality of samples and their sequencing. However, in both data sets, different methodologies were implicated, and the results are similar, which confirms that ZNF540 is closely associated with HPV infection. More importantly, however, the molecular mechanisms contributing to the ZNF540 involvement in HNSCC biology are unknown and need to be clarified in the in vitro cell line models and in vivo based on large patient samples with known HPV status.