Introduction

Human papillomavirus (HPV) infection has been recognized as an important cause of cervical precancerous lesions or cancer, and yet is necessary but not sufficient for cervical carcinogenesis process1,2. Therefore, in addition to HPV infection, HPV integration could contribute to the cervical carcinogenesis process.

HPV integration could upregulate the expression of viral oncogenes E6 and E7 and eventually promotes host genomic instability, which could be a crucial event of cervical carcinogenesis process3,4. Additionally, the level of HPV integration was positively associated with cervical intraepithelial neoplasia (CIN) grades and was proposed as a marker for cervical disease progress5,6. Therefore, comprehensively and accurately identifying the sites and level of HPV integrations from normal cervical epithelium to CIN and cervical cancer is necessary to assess HPV-induced carcinogenesis process. However, cervical HPV integrations have only been comprehensively investigated in invasive cancer and limited data were available in CINs and normal cervical epithelium7,8. Moreover, most previous assays could not identify HPV integrations sensitively, which could lead to the underestimation of HPV integration level7,9,10. Recently, the study by Hu et al. reported 3,667 HPV integration events in 26 CINs, 104 cervical cancer samples, and five cell lines based on high-throughput viral integration detection (HIVID)8. However, Nigel Dyer et al. claimed that 87% of the integration breakpoints reported by Hu et al. were likely to be experimental and computational artifacts according to their own data analysis pipeline11. Meanwhile we used the same HPV capture technology and the next generation sequencing as in Hu’s study to detect cervical HPV integrations in 39 HPV-positive primary cervical tumor samples and 2 cell lines and yet only identified 117 unique validated HPV integration breakpoints12. Moreover, we found that the Sanger sequencing validation rate based on one, two, three, or more than three discordant paired-end reads was 3.7%, 47.8%, 44.4%, and 83.3%, respectively, indicating that HIVID could be a sensitive method to detect integrated HPV and yet has a high false positive rate with fewer supporting reads. HPV integration rates in the study by Hu et al. could be overestimated in cervical cancer and even in CINs. Therefore, given that comprehensive and accurate data regarding HPV integrations in CINs and normal cervical tissues were limited, we detected HPV integrations in a series of CINs and normal cervical epithelium samples using base-resolution HPV capture technology and the next generation sequencing as previously reported12.

In this study we enrolled 166 participants with CIN or ‘normal cervical epithelium’ in order to investigate the level of cervical HPV integration in CINs and normal cervical epithelium.

Result

Distribution of HPV types

As shown in Fig. 1 and Table 1, of the 64 ‘normal cervical epithelium’, 62 CIN 1, 19 CIN 2 and 21 CIN 3 samples, we detected HPV in 7 (10.9%, 95% CI: 4.5–21.2%), 8 (12.9%, 95% CI: 5.7–23.9%), 6 (31.6%, 95% CI: 12.6–56.6%) and 15 (71.4%, 95% CI: 47.8–88.7%) samples, respectively. Of the 166 samples, we detected HPV16 (n = 23), HPV18 (n = 2), HPV33 (n = 2) and HPV58 (n = 5); four samples harbored two types of HPV with HPV45 and HPV58 in CIN 1–42, HPV 33 and HPV18 in CIN 1–62, HPV 82 and HPV52 in CIN 2–12, and HPV58 and HPV52 in CIN 3–5 (Fig. 2). HPV infection rate in younger women (age< = 50) and in older women (age > 50) were 25.2% and 12.8% (P = 0.096 by Fisher’s exact test).

Table 1 Summary of 36 HPV positive samples analyzed in this study: HPV integration status and clinical information.
Full size table
Figure 1
figure 1

Flow chart of study design, sample selection and HPV detection.

Full size image
Figure 2
figure 2

Distribution of HPV types among 36 HPV positive samples.

Full size image

Determination of potential HPV integration sites

As described in our previous study12, if a specific site had one or more discordant reads mapped on one end to the HPV reference genome and the other to human chromosome, it would be considered as a potential HPV integration locus. A total of 37, 21, 44 and 45 potential HPV integrations were identified in 7 HPV-positive ‘normal cervical epithelium’, 8 HPV-positive CIN1, 6 HPV-positive CIN2 and 15 HPV-positive CIN 3 samples, respectively (Table 1).

Validation of HPV integration sites

In order to confirm the potential HPV integration sites and to further identify the sequence between cellular and viral genome, all potential HPV integration positions were verified by targeted PCR amplification and Sanger sequencing. The validation rate of Sanger sequencing on the basis of one, two, three, or more than three different paired-end reads, was 0% (0/131), 33.3% (95% CI: 0.8–90.6%), 50% (95% CI: 1.3–98.7%), and 72.7% (95% CI: 39.0–94.0%), respectively (Table 1). After validation, only three, zero, one and six integration sites were verified in 7 HPV-positive ‘normal cervical epithelium’, 8 HPV-positive CIN1, 6 HPV-positive CIN2 and 15 HPV-positive CIN 3 samples, respectively. Of the 10 integrations, one was involved with HPV33 and nine with HPV16, respectively (Table 1).

Mapping and characterization of cellular-viral junction sequences

10 unique HPV integration breakpoints were distributed in six samples, including ‘normal cervical epithelium’ (n = 1), CIN2 (n = 1), and CIN3 (n = 4) (Fig. 3 and Table 1). Of the 10 integrations, four samples (CIN2-5, CIN3-15, CIN3-20, and CIN3-21) had one integration site and two samples (Control-31 and CIN3-2) had three integration sites, respectively (Fig. 3 and Table 1). The viral genome regions of the 10 validated integration positions in six samples were E1 (n = 5), E2 (n = 2), E2/E4 (n = 1), L1 (n = 1), and L2 (n = 1). Due to the limited number of integration events, we did not find hot spots in the human genome (Table 1).

Figure 3
figure 3

Chromosome localization of the 10 HPV integration breakpoints in six samples.

The chromosomal reference of all viral-cellular breakpoints with respect to Giemsa-stained bands was taken from the UCSC databases. Six HPV-positive samples with integration breakpoints are shown in unique color. Cell sequence breakpoints are marked with black and HPV breakpoints are marked with red.

Full size image

All integration positions were examined for the presence of fragile sites in the human genome. Of the 10 integration positions, one was located in a fragile site and three were close to a fragile site (Supplementary Table S1). Meanwhile, the human genomic sequences within 50 kb of an integration locus were investigated. Seven integration sites were located in cellular genes with six in introns and one in an exon (Supplementary Table S1).

Discussion

Our study only found three, zero, one and six verified integration sites in 7 HPV-positive ‘normal cervical epithelium’, 8 HPV-positive CIN1, 6 HPV-positive CIN2 and 15 HPV-positive CIN 3 samples, respectively. HPV integration sites were mainly located in the E1 and E2 regions of the viral genome and in cellular genes of the human genome.

We found that HPV integration rates in HPV-positive CIN 1, CIN 2, and CIN 3 samples were 0% (0/8), 16.6% (95% CI: 0.4–64.1%), and 26.7% (95% CI: 7.8–55.1%), respectively, which were similar to those reported previously7,13,14. However, the study by Hu et al.8 reported that HPV integration rates in CIN 1, CIN 2, and CIN 3 were 50% (95% CI: 18.7–81.3%), 44.4% (95% CI: 13.7–78.8%), and 71.4% (95% CI: 29.0–96.3%), respectively, which were much higher than the validated integration rates in our study and yet similar to what we considered as the “potential” HPV integration rates (37.5% [95% CI: 8.5–75.5%], 66.7% [95% CI: 22.3–95.7%], and 53.3% [95% CI: 26.6–78.7%]) based on HPV capture and the next generation sequencing technology without Sanger sequencing validation. Notably, the validation rate of Sanger sequencing on the basis of one, two, three, or more than three different paired-end reads was 0% (0/131), 33.3% (95% CI: 0.8–90.6%), 50% (95% CI: 1.3–98.7%), and 72.7% (95% CI: 39.0–94.0%) in our study, respectively. Moreover, HPV integration rates in cervical cancer was significantly higher in Hu’s study8 than in our previous study12. The overall HPV integration rates in the study by Hu et al. could be overestimated in cervical cancer and CINs, since only selected HPV integrations in Hu’s study were validated by Sanger sequencing.

It is worth noting that HPV integrations could occur in cervical tissue with normal epithelium and the integration rate in CIN 3 was significantly higher than those in CIN 1 or CIN 2. This indicated that HPV integrations could play an important role in the early stage of cervical carcinogenesis, although our results were lack of the statistical analysis and the sample size may not be enough for the analysis. In addition, we found similar characteristics of the HPV integration sites in cervical cancer and non-cancer specimens12. For example, HPV integration sites were mainly located in the E1 and E2 regions of the viral genome and in cellular genes of the human genome.

Two different types of mechanisms are presumed to explain cervical carcinogenesis process induced by HPV integration, i.e. altering viral gene expression or disrupting cellular transcripts. In order to determine the effect of these two mechanisms, it is necessary to comprehensively profile HPV integrations in host and viral genome. However, detection methods of HPV integration in most previous studies were low-throughput and lower sensitivity. In order to better understand the cervical carcinogenesis induced by HPV integration, this approach is able to discern fusion breakpoints accurately at single-base resolution for further elucidating the effect of HPV integration on viral and its flanking cellular transcripts. In addition, since HPV integration could lead to the viral persistence and moreover HPV persistent infection plays role in cervical carcinogenesis, this approach provides unbiased, genome-wide integration information to monitor the persistent or permanent infection.

However, there are some limitations in our study. Firstly, since HPV integration rates in ‘normal cervical epithelium’ and CIN were low and moreover only 36 HPV-positive women were involved in HPV integration analysis, comprehensively evaluating the sites and the level of HPV integrations was limited to some extent. Secondly, the cross-sectional study did not investigate the temporal relationship between HPV integration and CINs. Thirdly, in our study, CIN enriched tissue was not sampled by laser microdissection, which did not rule out contamination from normal adjacent epithelium or the underlying stroma. This would overestimate HPV integration rate in CINs to some extent although this effect was small due to significantly lower HPV integration rate in ‘normal cervical epithelium’ than in CINs. Fourthly, Since HPV DNA was detected using a highly sensitive PCR primer set (SPF1/GP6+) amplifying a 184-bp fragment of the L1 open-reading frame before performing HPV capture and sequencing, this might produce HPV false negatives from L1 breakpoints among the 166 samples. However, since proportion of breakpoints occurring in this targeted region of L1 was low12 and in most situations, HPV viral genome may be existed in both episomal and integrated forms, HPV false negative probability due to L1 breakpoints was small.

In summary, the accurate identification of HPV integrations in CINs and normal cervical tissues could provide basic evidence for HPV integration-driven cervical carcinogenesis and be served as individualized markers in cervical cancer screening in the future.

Materials and Methods

Study population and specimen collection

A total of 166 cervical biopsy specimens were collected and diagnosed with normal cervical epithelium or acute/chronic cervicitis without atypical hyperplasia (n = 64), CIN 1 (n = 62), CIN 2 (n = 19) and CIN 3 (n = 21) from Beijing Cancer Hospital, Beijing, China, between 2014 and 2015. All biopsy specimens were reviewed by two experienced pathologists who confirmed the diagnosis of CIN. Cervical biopsies were histologically diagnosed using criteria defined by the World Health Organization15. All cases have no histological evidence of epithelial malignancy of the cervix in this study. Normal cervical epithelium or acute/chronic cervicitis without atypical hyperplasia is defined as ‘normal cervical epithelium’ in this study. Punch biopsy samples were divided into two parts; one was kept for histopathological analysis, and the other one was used for HPV typing and integration analysis. Individual informed consents had been collected from all participants. This study received ethical approval from the Institutional Review Board of the Peking University School of Oncology, China. All experiments were performed in accordance with relevant guidelines and regulations.

The specimens were stored at −80 °C and genomic DNA was extracted from the frozen tissues using DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. The β-globin gene was evaluated in all specimens by PCR.

HPV typing and integration detection

HPV DNA in valid (β-globin positive) specimens was detected using a highly sensitive PCR primer set (SPF1/GP6+) amplifying a 184-bp fragment of the L1 open-reading frame16. Specimens showing the PCR amplification product were used to identify HPV genotypes and integrations. HPV probes were designed according to the full-length genome of 17 HPV types (6, 11, 16, 18, 31, 33, 35, 39, 45, 52, 56, 58, 59, 66, 68, 69, and 82) by MyGenostics (MyGenostics, Baltimore, MD, USA). Details of HPV typing and the detection of HPV integrations, as well as Sanger sequencing validation of potential HPV integration sites were described previously12. In brief, the whole-genomic libraries were hybridized with HPV probes (MyGenostics GenCap Technology), adsorbed onto the beads via biotin and streptavidin magnetic beads, and the uncaptured DNA fragments were removed by washing. Then the eluted fragments containing the targeted gene were enriched by PCR to generate libraries for sequencing. Libraries were quantified and sequenced for paired-end 125 bp using the Illumina HiSeq 2500 sequencer (Illumina Inc., San Diego, CA, USA). Illumina clean reads were mapped to human genome (GRCh37/hg19) and HPV genome of 17 types using the BWA program. The paired-end read, uniquely mapped with one end to a human chromosome and the other to the HPV reference genome, is identified as a discordant read pair. If a specific position has one or more discordant read pairs, it would be considered as a potential HPV integration site. PCR and Sanger sequencing were used to verify all the potential HPV integration breakpoints. All sequences of the fusion genes were characterized by the NCBI human mega Blast database alignment tool and the UCSC Blat database.

Statistical analyses

Fisher’s exact test was used to determine the relationship between age and HPV infection. Statistical analyses were performed using STATA version 12.0 software (STATA Corporation, College Station, TX, USA). P values less than 0.05 (two-sided) were considered to be significant.

Additional Information

How to cite this article: Liu, Y. et al. Genome-wide profiling of the human papillomavirus DNA integration in cervical intraepithelial neoplasia and normal cervical epithelium by HPV capture technology. Sci. Rep. 6, 35427; doi: 10.1038/srep35427 (2016).