HPV has been recognized as an important driver of HNSCC [23, 56, 57]. The patient treatment varies depending on HPV positive (HPVP) vs. negative (HPVN) HNSCC, thus it is important to gain further knowledge of the genetic profile of HNSCC. Our study showed that HPVP HNSCC patients exhibit gene deregulation at the gene transcription and methylation levels different from HPVN HNSCC. When analyzed, both independently and collectively, gene expression and methylation deregulation patterns specifically point out changes in gene pathways including those involved in controlling invasion, immune response, differentiation, and cell division.
In total, the cohort of patient’s samples analyzed were 114 (HPVN = 73 and HPVP = 41). There was a disparity in the male/female self-described sample ratio, male samples account for 93, and female samples the remaining 21 (Figure 2C). A possible explanation for this disparity might be that HNSCC cases are sex biased and more prevalent in males, but a larger cohort needs to be analyzed to address this disparity. There was also an overrepresentation of white race (n = 98) in our cohort for HNSCC (Figure 2D). This lack of racial representation is unfortunately not uncommon in clinical studies. We have since identified studies that report HNSCC incidence in non-white population, and a similar analysis will be conducted in the future in order to include more equally distributed races [58–61]. There was an apparent absence of HPVP HNSCC patients in the oldest patient category (76-85 years of age - Figure 2B), and we theorize that might be due to HPVP HNSCC being rarer than HPVN patients, thus causing this age groups’ underrepresentation. Alternatively, the HPVP HNSCC patients do not survive long enough to be included in the data [62, 63]. We observed differences in anatomical sites of HNSCC dependent on HPV status (Figures 2E, 2F and 2G). Tonsil was the predominant location in HPVP patients, while the oral tongue had the most cases in HPVN patients (Figures 2F and 2G). In the U.S., regardless of HPV status, the oral tongue is the most common site for HNSCC [58].
In our analysis, genes that play a role in all HNSCC development belonged to four main functional pathways: cell survival, cellular proliferation, squamous epithelial differentiation, and invasion/metastasis. We identified differentially expresses and methylated genes in HPVP versus HPVN HNSCC. Of the 1854 DEG, 16 genes were the top hits identified in the transcriptome and methylome analyses. The functions of these genes range from cell cycle, immune response, to cell death regulation. Specifically, we found that SYCP2 and TAF7L were the two most deregulated genes in both analyses. Synaptonemal Complex Protein 2 (SYCP2) was the top hypomethylated and upregulated gene in HPVP HNSCC. This gene is testis-specific human gene and it has been associated with impaired meiosis [64]. It is known that SYCP2 aberrant expression in HPVP cancers may contribute to the genomic instability induced by high-risk HPVs and subsequent oncogenic change [65]. In 2015, a paper by Masterson et al. reported that deregulation of SYCP2 predicts early-stage human papillomavirus-positive oropharyngeal carcinoma. The same authors concluded their study proposing SYCP2 as a potential biomarker [66]. In addition to this, an independent study showed SYCP2 was hypomethylated in HPVP HNSCC, which is in concordance with what we have discovered [19]. This might imply that the previously proposed biomarker function for SYCP2 is not unlikely. Besides these reports, the elevated expression of SYCP2 in HPV-associated tumors has previously been observed in three additional gene expression analysis studies [67–69]. Similarly, the second highlighted gene that was hypomethylated and upregulated in HPVP HNSCC is TATA-Box Binding Protein Associated Factor 7 Like (TAF7L), a gene involved in spermatogenesis [70]. According to a study by Mobasheri et al., TAF7L is upregulated in breast cancer, so it is possible that is not an exclusive feature observed only in breast cancer tissue [71].
DEG analysis identified that PCNA, TNFRSF14, TRAF1, TRAF2, BIRC3, and BCL2 were significantly altered in HPVP HNSCC.
Proliferating cell nuclear antigen (PCNA) is a gene that was significantly overexpressed in HPVP versus HPVN HNSCC patient samples. It has been shown that PCNA expression levels change during cell cycle, as PCNA is associated with proliferation and cell transformation in cancer [72, 73]. PCNA is one of the crucial regulators in cell cycle as it makes complexes with cell cycle activators (cyclins and cyclin dependent kinases) and inhibitors (p21) [72]. Post-translational modifications are crucial for PCNA function, so much so, that PCNA exists in an alternative methylated form in cancers [74].
Tumor necrosis factor receptor superfamily member 14 (TNFRSF14) is known to be a herpesvirus entry mediator by being a part of signal transduction pathways that activate inflammatory and inhibitory T-cell immune response [75]. It is not surprising to see that it was upregulated in HPVP HNSCC, although it is interesting that a herpesvirus related gene has been upregulated upon HPV infection in this cancer type. TNFRSF14 is known to interact with TNF Receptor Associated Factor 2 (TRAF2), that is also upregulated in HPVP HNSCC. This protein directly interacts with the TNF receptors, and forms a complex with another TRAF family member, TRAF1 that is also upregulated in HPVP HNSCC. This is all necessary for TNFα mediated activation of MAPK8/JNK and NF-kβ, which are known to be involved in cell survival. The protein complex formed by TRAF2 and TRAF1 interacts with the inhibitor-of-apoptosis proteins (IAPs), and functions as a mediator of the anti-apoptotic and pro-survival signals from TNF receptors. One of those IAPs that is upregulated in HPVP HNSCC is BIRC3 - apoptosis inhibitor [76–78]. According to The Human Protein Atlas (THPA), TRAF2 has the highest expression in HNSCC, followed by cervical cancer among all sampled cancer types (17 cancer types) [79]. BIRC3 shows similar observations, implying this pattern may be specific for HPV related HNSCC [79]. Another role of TRAF1 is a negative regulation of Toll-like receptor (TLR) and Nod-like receptor (NLR) signaling. TRAF1 can also, independently from TRAF2, contribute to NF-kβ activation; conversely, during TLR and NLR signaling, TRAF1 can also negatively regulate NF-kβ activation. According to THPA, TRAF1 has been found to be overexpressed in HNSCC. Additionally, TRAF1 can contribute to chronic viral infection and limit inflammation, contributing to survival of Epstein-Barr virus dependent cancers [76, 79]. TRAF family genes (TRAF1 and TRAF2 specifically) have been found to be differentially expressed in a couple of HPV related studies, including one in our lab [80, 81]. An interesting question follows: does TRAF1 have a similar role in HPV dependent cancers as well? To investigate this, more research is needed.
In addition to BIRC3 - apoptosis inhibitor being upregulated in HPVP HNSCC, BCL-2, an anti-apoptotic gene has been observed to be upregulated in HPVP HNSCC as well. An existing model explains the observed picture in our data. Similarly to oncogene addiction, some tumor cells may become dependent on BCL-2 for survival [82]. As tumor environment may induce higher stress signal production that is pro-apoptotic nature, a proportion of cancer cells manage to overexpress BCL-2 and survive producing this anti-apoptotic signal. In this way, BCL-2 helps cancer progression by promoting the survival of altered cells [83, 84]. BCL-2 is also known to be overexpressed in non-hematologic tumors as ovarian, neuroblastoma, colorectal and HNSCC [85–88].
Starburst analysis combined DEG and DMR results and highlighted genes that were the most hypomethylated and upregulated and the most hypermethylated and downregulated. We performed Starburst with FDR cut=1 and a more stringent parameter FDR cut=3 and maintained top highlighted gene profile (Supplementary Table 5, and Figures 5C and 5D), specifically SYCP2 and TAF7L. Taken together, some of the DEG identified as top hits may be used as potential biomarkers for early identification of HPVP HNSCC, including SYCP2, TAFL7 and ZFR2. The analysis of DEG of tonsil HPVP HNSCC and oral tongue HPVN HNSCC (predominant anatomical locations of samples), identified unique genes that were downregulated in HPVP tonsil HNSCC (Supplementary Table 6). One of these, RBM24 is shown to mediate repression of p53/TP53 mRNA translation and INHBA, a member of transforming growth factor-beta (TGF-β) superfamily of proteins. According to THPA, the highest expression of RBM24 is observed in HNSCC, followed by cervical cancer, although we have not seen its use as a diagnostic tool [79, 89]. This implies that these genes when downregulated might specifically indicate HPVP HNSCC site specific (tonsil) cancer development.