Differential expression of human papillomavirus 16-, 18-, 52-, and 58-derived transcripts in cervical intraepithelial neoplasia

Background Human papillomavirus (HPV) infection is a primary cause of cervical cancer. Although epidemiologic study revealed that carcinogenic risk differs according to HPV genotypes, the expression patterns of HPV-derived transcripts and their dependence on HPV genotypes have not yet been fully elucidated. Methods In this study, 382 patients with abnormal cervical cytology were enrolled to assess the associations between HPV-derived transcripts and cervical intraepithelial neoplasia (CIN) grades and/or HPV genotypes. Specifically, four HPV-derived transcripts, namely, oncogenes E6 and E6*, E1^E4, and viral capsid protein L1 in four major HPV genotypes—HPV 16, 18, 52, and 58—were investigated. Results The detection rate of E6/E6* increased with CIN progression, whereas there was no significant change in the detection rate of E1^E4 or L1 among CIN grades. In addition, we found that L1 gene expression was HPV type-dependent. Almost all HPV 52-positive specimens, approximately 50% of HPV 58-positive specimens, around 33% of HPV 16-positive specimens, and only one HPV18-positive specimen expressed L1. Conclusions We demonstrated that HPV-derived transcripts are HPV genotype-dependent. Especially, expression patterns of L1 gene expression might reflect HPV genotype-dependent patterns of carcinogenesis.

genome into the human genome, is critical in cervical cancer progression [4]. High-risk HPV E6 and E7 are likely to transform CIN lesion to cancer. Especially, HPV 16 and 18 are the most carcinogenic. The prevalence of HPV 16 and 18 in cervical cancer and cervical intraepithelial neoplasia (CIN) are quite different from other high-risk HPV. About 50 and 15% of cervical cancer are positive for HPV 16 and 18, whereas about 40% and 3-7% of high-grade CIN (CIN2/3) are positive, respectively [5,6]. Other data show that the rate of progression of HPV 16-or 18-infected cervical epithelium to CIN3 or more is around 15% at 10 years post-infection, which is much higher compared to other HR-HPV [7]. Further, HPV 18 is likely to integrate the viral genome into the host genome compared to HPV 16 [8]. Furthermore, there are HPV type-dependent features among cancer histological types. Most HPV 16-positive cancers are squamous cell carcinomas, whereas around 50% of HPV 18-positive cancers are adenocarcinomas [5].
HPV generates numerous viral transcripts via differential RNA splicing. For example, at least 13 transcripts are derived from eight HPV genes in HPV 16-infected W12E cells [9]. There are six genes (E6, E7, E1, E2, E4, and E5) located in the early region of the HPV genome, and two genes (L1 and L2) in the late region. Expression of these genes is altered during epithelial differentiation and/or CIN progression. E6 and E7 are oncogenes encoding proteins that suppress p53 and pRb activation, respectively [10]. E6* is a splicing isoform of E6, which is the main E6 isoform in cervical cancer, and might facilitate E7 expression [11]. Although the roles of E1^E4 are not explicitly defined, E1^E4 is considered to be associated with viral replication [12]. The L1 and L2 proteins are components of the viral capsid and are associated with HPV infection [13].
The expression patterns of HPV-derived transcripts vary depending on CIN grade. For example, the expression of E6 and E7 is higher in high-grade squamous intraepithelial lesions (high-grade SILs) than in lowgrade SILs [14]. In contrast, expression of the L1 protein is lower in high-grade SILs [15,16]. The expression patterns of HPV-derived transcripts also differ among HPV genotypes. For example, among E6 isoforms, HPV 18 cancers exhibit significantly higher ratios of the nonspliced isoform of E6 oncoprotein than HPV 16 cancers [17]. Furthermore, Griffin et al. demonstrated that CIN3 with HPV 18 exhibited no E4 protein expression [18].
In this study, we analyzed each HPV-derived transcript to gain a better understanding of HPV genotypedependent carcinogenesis. To represent the viral life cycle, we focused on the expression levels of HPVderived transcripts E6/E6*, E1^E4, and L1. E6/E6* are oncogenes regulated by the early promoter, the E1^E4 splicing site relates to both early and late gene expression and can contribute to viral replication, and L1 expression is observed in the late phase of viral differentiation, which is regulated by the late promoter [19]. In addition to HPV 16 and 18, we focused on HPV 52 and 58, which are highly prevalent in East Asia [20].

Patients and sample collection
All experimental procedures were approved by the institutional review boards of The University of Tokyo (approval number: G10082), Keio University (approval number: 2015-388), Chiba University (approval number: 560), Akita University (approval number: 2174), the National Institute of Infectious Diseases (approval number: 659), and Nihon University (approval number: 234-0), and signed informed consent for the use of tissues was obtained from each participant.
In total, 382 patients with cervical cytological abnormality who were admitted to the University of Tokyo, Chiba University, or Keio University between February 2016 and December 2017 were enrolled. Cervical tissues were obtained from biopsy under colposcopic examination. Samples were stored at − 80°C until analysis.

Variables
Clinical data, such as age, gravidity, smoking history, usage of steroids or immunosuppressants, and time from first detection of abnormal cytology, were obtained by a medical interview. Histological results were classified into three CIN grades: CIN1, CIN2, and CIN3. Diagnosis was confirmed by a pathologist at Akita University.

HPV genotyping
DNA was extracted from cervical specimens using the Tissue Genomic DNA Extraction Mini Kit (Favorgen Biotech Corp., Ping-Tung, Taiwan) at The University of Tokyo. HPV genotyping was performed at the National Institute of Infectious Diseases using the PGMY-CHUV assay method as described previously [21]. Briefly, standard PCR was conducted using the PGMY09/11 L1 consensus primer set and human leukocyte antigen-DQ (HLADQ) primer sets. Subsequently, reverse blotting hybridization was performed. Heat-denatured PCR amplicons were hybridized to probes specific for 31 HPV genotypes and HLA-DQ references [22].

Primer design and standard plasmid
PCR primers were designed using Primer-Blast (NCBI) in reference to PaVE, the papilloma virus genome database. The following criteria were considered when designing the primer pairs: (1) each primer should be 19-23 bp in length, and (2) the amplicon should be between 70 and 260 bp in length. The primer design is shown in S1 Fig. Plasmid standards (Eurofin Scientific, Luxembourg City, Luxembourg) were used to derive standard curves for absolute quantification.

RNA extraction and quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from cervical specimens using an miRNeasy Mini Kit (Qiagen, Hilden, Germany) after DNase treatment using the RNase-Free DNase Set (Qiagen, Hilden, Germany) at The University of Tokyo. Extracted RNA was reverse-transcribed using the SuperScript III First-Strand Synthesis System for RT-PCR (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's instructions. To assess mRNA expression levels, qRT-PCR was performed using a Light

Statistical analysis
Categorized clinical features such as gravidity, smoking history, usage of steroids or immunosuppressants according to HPV categories and CIN grades were evaluated using the Analysis of Variance. Other clinical features such as age and time from first detection of abnormal cytology and the expression levels of each transcriptome according to HPV types and CIN grades were analyzed using a Steel-Dwass test. The relationship between each HPV infection and positive ratio of each transcriptome was evaluated using the Cochran-Armitage trend test. Statistical analyses were performed with the JMP Pro software (13.0.0). p < 0.05 was considered significant. If the viral gene copy number was greater than 10 copies/L, the sample was considered positive for gene expression.

HPV prevalence of four major genotypes
For the 382 patients with cervical cytological abnormality enrolled in the study, CIN grades and infected HPV types are summarized in Fig. 1 (Fig. 1). The ages of patients with each HPV genotype were significantly different, whereas no significant differences were found in gravidity, smoking history, usage of steroids or immunosuppressants, or time  (Tables 2 and 3).

Differential E6/E6* expression in HPV-positive specimens
Expression levels of oncogene E6 and its isoform E6* were evaluated in each sample. Although there was no difference in the expression levels of E6 or E6* among CIN grades ( Fig. 2a and b), the detection rate of E6 and/ or E6* increased with severity of the CIN grade (Cochran-Armitage test, p < 0.01, Fig. 2c). In terms of HPV genotype-dependent analysis, E6 expression was lowest and E6* expression was highest in HPV 16 positive-specimens among the four HPV genotypes. In HPV 18-positive specimens, although no significant differences compared to other HPV genotypes were observed, possibly due to the small sample size, E6 and E6* expression patterns were similar to HPV 16-positive specimens. Conversely, in HPV 52-positive specimens, E6 expression was higher than E6*, while comparable expression levels of both E6 and E6* were observed in HPV 58-positive specimens (Fig. 2b).

Differential E1^E4 expression in HPV-positive specimens
Subsequently, expression of the E1^E4 splicing site was evaluated. Positivity of this site is related to both early and late gene expression [19]. First, we compared E1^E4 expression levels across CIN grades and found its highest expression in CIN3 (Fig. 3a). The detection rate of E1^E4 tended to increase with CIN progression in HPV 16-, 52-, and 58-positive specimens (Cochran-Armitage test, p = 0.10, p = 0.13, and p = 0.04, respectively) (Fig. 3c). HPV genotype-dependent analysis revealed that E1^E4 expression levels were highest in HPV 16-positive specimens and lowest in HPV 18-positive specimens (Fig. 3b). Further, the detection rate of E1^E4 was higher in HPV 16-and 58-positive specimens than HPV 52-positive specimens (Fig. 3c).

HPV type-dependent expression of major capsid protein L1
Further, expression levels of the L1 gene, which encodes a major capsid protein, was assessed. There was no difference in L1 expression among CIN grades (Fig. 4a). Comparison of L1 gene expression among HPV genotypes revealed the highest L1 expression in HPV 52-positive specimens, followed by HPV 58-positive specimens, and there was almost no L1 expression in HPV 18-positive specimens (Fig. 4b). Additionally, L1 expression was HPV type-dependent, in which nearly 100% of HPV 52-positive specimens, around 50% of HPV 58-positive specimens, approximately 33% of HPV 16-positive specimens, and almost 0% of HPV 18-positive specimens expressed L1 (Fig. 4c).
Since L1 expression is a hall mark of viral production and viral production is typically accompanied by epithelial differentiation, we performed immunohistochemistry analysis of KRT10 and KRT13 to investigate epithelial differentiation. However, there was no difference in KRT10 and KRT13 expression among samples with each HPV genotype (S2 Fig).

Discussion
We performed an in-depth analysis of HPV-derived transcript levels according to HPV genotype and CIN grade. The detection rate of E6/E6* increased with CIN progression, which is consistent with a previous study [23], whereas there was no significant change in the detection rate of E1^E4 or L1 among CIN grades. Furthermore, type-dependent analysis revealed that expression patterns of HPV-derived transcripts were HPV genotype-dependent.
Interestingly, L1 expression level was lowest in HPV 18-positive specimens among the four genotype groups.  Clinical features were summarized according to cervical lesion grade. Statistical analysis was performed using a Steel-Dwass test (*1) and the Analysis of Variance (*2) CIN cervical intraepithelial neoplasia HPV 18 is one of the most carcinogenic genotypes among HR-HPV [24], and frequently observed in young aged cervical cancer [1]. In addition, around 40% of cervical adenocarcinoma is caused by HPV 18 infection [5]. L1 protein, a major component of the viral capsid, is a hallmark of viral production accompanied with cellular differentiation. Therefore, low level or lack of L1 expression in HPV 18-positive specimens may be associated with the loss of cellular differentiation and nonproliferative HPV infection, suggesting that stratified epithelium differentiation is not necessary for the HPV 18 genome replication and maintenance of HPV 18related carcinogenesis. Our results added a new insight on HPV 18-related carcinogenesis from the aspect of HPV-derived transcripts. Other than loss of cellular differentiation, expression of the HPV L1 capsid protein disappears when HPV DNA is integrated into the host genome. Viral genome integration occurs earlier in HPV 18-positive cervical cells than in HPV 16-positive cells [8]. Loss of cell differentiation and viral genome Conversely, L1 expression was highest in HPV 52, even in high-grade SIL (CIN2 and CIN3). Usually, L1 gene expression decreases as the CIN grades progress due to the lack of cellular differentiation [15,16]. In contrast to HPV 18, the high expression of the L1 gene in HPV 52-positive specimens, even in high-grade SILs, may indicate that proliferative HPV infection accompanied with cellular differentiation may be maintained in HPV 52-positive lesions. In this study, we could not identify the human differentiation markers reflecting HPV 52-positive lesions; further studies are needed to identify human gene expression profiles that can distinguish the expression patterns of HPV-derived transcriptomes. Furthermore, the high level of L1 gene expression in HPV 52-positive CIN3 suggests that viral genome integration occurs in the late stage of CIN progression in these samples. Combined with the previously reported epidemiological findings, i.e. frequent Fig. 3 Expression of HPV E1^E4 gene. a Copy number of the E1^E4 gene in specimens with different CIN grades. Statistical analysis was performed using a Steel-Dwass test. ** indicates p < 0.01. b Copy number of the E1^E4 gene in specimens with HPV16, 18, 52, and 58 infection. Statistical analysis was performed using a Steel-Dwass test. * indicates normalized copy number. ** indicates p < 0.01. c Detection rate of E1^E4 gene in each genotype stratified by CIN grade. Statistical analysis was performed using a Cochran-Armitage test observation of HPV 52 in CIN lesions compared to cancer lesions [5], high L1 expression represents the long-term persistence of HPV 52-related CINs.
This study has several limitations. First, regarding the analysis of E1^E4 gene expression, we only evaluated the expression level of the E1^E4 splicing site. Therefore, it is difficult to precisely demonstrate the significance of E1^E4 expression on the biology of viral replication or cancer development. Second, the sample size of the HPV 18-positive specimen was small. Therefore, further study using a large sample size is warranted to confirm our results. Third, this study is a cross-sectional study and does not investigate prognosis of CIN patients. As such, a prospective cohort study is needed to investigate whether expression of these HPV-derived transcripts can be biomarkers of CIN progression or regression.

Conclusions
In this study, we investigated the expression of three HPV-derived transcripts downstream of early, early/late, Fig. 4 Expression of the HPV L1 gene. a Copy number of the L1 gene in specimens with different CIN grades. Statistical analysis was performed using a Steel-Dwass test. ** indicates p < 0.01. b Copy number of L1 gene in specimens with HPV16, 18, 52, and 58 infection. Statistical analysis was performed using a Steel-Dwass test. ** indicates p < 0.01. c Detection rate of L1 gene in each genotype stratified by CIN grade. Statistical analysis was performed using a Cochran-Armitage test and late promoters in CIN lesions. Their expression patterns differed among HPV genotypes. In particular, L1 gene expression levels were lowest in HPV 18, while highest in HPV 52, suggesting HPV type dependence of HPV-derived carcinogenesis and viral maintenance in the cervical epithelium.