Main

Cervical cancer is the second leading cause of cancer deaths for women worldwide.1 Today, cervical cancer screening programs mostly rely on cytology. The Pap-test introduced more than 60 years ago by Papanicolaou2 remarkably reduced the mortality by cervical cancer.3 Pap screening has high specificity for disease, but suffers from suboptimal sensitivity and observer subjectivity that is partially compensated by frequent testing.4 Therefore, supportive molecular alternative methods are highly desirable.

Human papillomavirus is a necessary cause of invasive cervical cancer.5

Therefore, HPV testing is being discussed as a primary screening method, for triaging of equivocal results, and follow-up after therapeutic intervention.6 High-risk HPV infections are not sufficient for the development of cervical cancer. Most individuals remain asymptomatic and clear HPV infections spontaneously within approximately 8–10 months.7 The immortalization of primary human keratinocytes can be sufficiently maintained by the expression of the high-risk HPV E6 and E7 oncogenes. These oncogenes are required for initiation and all subsequent stages of carcinogenic progression.8 The focus on HPV oncogene mRNA expression could be a promising strategy to identify transformation events, but requires multiplex approaches to deal with HPV type-specific transcripts. The proteins p16INK4a (CDKN2A) and p14ARF (CDKN2B) are generally upregulated in transformed cells as a consequence of high-risk HPV E6 and E7 expression.

The CDKN2A/B gene locus encodes two proteins, p14ARF and p16INK4A, generated by alternative mRNA splicing.9 The p14ARF consists of the unique first exon, 1-beta, 15 kb upstream of the exon 1-alpha of p16INK4A. Both contain the common exons 2 and 3. p14ARF and p16INK4A are involved in cell cycle regulation, but despite their functional relationship, distinct promoters independently control the mRNA transcription of p14ARF and p16INK4A, and different reading frame usage results in unique amino acid sequences.10

p14ARF is a cell cycle regulator, arresting cell growth at G1–S and G2–M11. p14ARF interaction inhibits the Mdm2 oncoprotein, thereby blocking the Mdm2–p53 complex and preventing Mdm2-induced p53 degradation.10 p14ARF expression is positively regulated by the transcription factor E2F-1.11 HPV E7 inhibits Rb, releasing active E2F-1. Therefore, E7 has the ability to induce p14ARF expression. Additionally, p53 negatively regulates p14ARF mRNA transcription. Consequently, as E6 inhibits p53, it can elevate p14ARF expression.11, 12

p16INK4A stops cell cycle progression by binding cyclin-dependent kinase (CDK) 4 and 6, and inhibiting cyclin D. By inhibiting phosphorylation of Rb, p16INK4A promotes the formation of a Rb–E2F repressive transcriptional complex, which also blocks cell cycle progression at the G1–S restriction point.10, 13 The HPV E7 protein binds to the phosphorylated Rb. This event releases the transcription factor E2F, and thereby induces uncontrolled cell replication. A positive feedback control loop leads to overexpression of p16INK4a in HPV-transformed cells.14, 15

Cervical cancer was shown to be strongly associated with p16INK4a and p14ARF expression by immunohistochemical studies.11, 16, 17, 18, 19 Immunohistochemical staining for p16INK4a became the most widely accepted surrogate biomarker in cervical cancer screening. p16INK4a has the advantage to detect any high-risk HPV infection in conjunction with uncontrolled, malignant cell replication.20, 21 Furthermore, it has been shown to improve inter-observer agreement on histological diagnosis,22 and commercial kits are widely used to support histological findings.

Apart from HPV-induced expression, p14ARF and p16INK4A also accumulate in many tissues as a function of advancing age as shown in rodents23 and in human skin.24 Their physiological function is the control of the state of cellular replicative senescence. Such upregulation in older patients may interfere with the high-risk HPV-related upregulation of p14ARF and p16INK4A in cervical intraepithelial neoplasia.

In contrast to the successful establishment of immunohistochemical staining, very little data exists about the use of CDKN2A-derived mRNA in cervical cancer diagnosis. However, despite RNA derived from exfoliated cervical cells being challenging, other mRNA-based biomarkers such as HPV oncogenes,25 Brn3a,26 or MCM5,27 were successfully tested on cervical smears and biopsies.

We analyzed the mRNA expression levels of p16INK4a and p14ARF in cervical smears from HPV-negative women with normal cytology and compared it with women with high-grade findings (HSIL). We used validated SYBR-Green I assays in a read-out described by Kanellou et al.28 Expression was related to β-actin (ACTB) that was used as a housekeeping gene,29 because it turned out to be the most stable reference in exfoliated cervical cells.30

The aim of this study was to (i) study p14ARF and p16INK4A mRNA expression with respect to their feasibility in cervical cancer diagnosis and (ii) to analyze the influence of age on the expression of p14ARF and p16INK4A. Our results support wariness in interpretation of p16INK4a and p14ARF mRNA expression results from older patients.

Patients and methods

Patients

Patient numbers and characteristics are summarized in Table 1. Pap smears were taken under colposcopic control at the Clinic for Gynecology, Campus Benjamin Franklin, Charité-Universitätsmedizin Berlin, Germany. Two smears were taken, the first for cytology and RNA extraction, and the second for HPV typing. To analyze the effect of age, patients with normal results (cytology and HPV test negative) or with high-grade dysplasia were divided into groups of upper and lower mean-age and analyzed for p16INK4a/p14ARF mRNA expression. Cytological diagnosis was made by cytopathologists blinded to the real-time PCR results. The study was approved by the IRB (EA4 217-20) and patients consented to the scientific use of scrape material.

Table 1 Patient groups
Full size table

HPV Typing

Genomic DNA from cervical smears was extracted with the Genomic DNA Mini Kit (Qiagen, Hilden, Germany) according to the standard protocol and eluted in 160 μl TE buffer. A β-globin PCR was made as described and analyzed on a 2% agarose gel, to monitor DNA quality.31 The generic GP5+/bio6+ primer set was used to amplify a sequence from the L1 gene of a large number of different HPV types.31 After PCR amplification, HPV sequences were genotyped by MPG Luminex suspension array technology (Multimetrix, Heidelberg, Germany).32

RNA Extraction and cDNA Synthesis

Cervical smears were taken and the cyto-brushes were immediately immersed and stored in RNAlater (Qiagen), and transferred to −20°C within 24 h. The brush head was placed into a shredder column (Macherey & Nagel, Düren, Germany), filled with 350 μl lysis buffer RA1 containing 1% mercaptoethanol, and centrifuged at 15 000 r.p.m. for 1 min. Residual material in the RNAlater solution was centrifuged at 12 000 r.p.m. after adding one volume of PBS. The pellet was combined with the flow-through from the shredder column, reloaded to the column, and centrifuged again. From the lysed material, RNA was extracted with the Macherey & Nagel RNA II Mini Isolation Kit (Macherey & Nagel) according to manufacturer instructions, with DNA digest using recombinant DNAse. RNA was eluted with 36 μl RNAse free water. RNA concentration was measured by Nanodrop (PaqLab, Erlangen, Germany). RNA integrity was analyzed for representative samples using Agilent Bioanalyzer 2100 (Agilent Santa Clara, USA). The extracted RNA was stored at −80°C. The cDNA synthesis was done with the Quantitect cDNA Kit (Qiagen) according to manufacturer instructions, using 0.25 μl/reaction RNAse inhibitor (Sigma Aldrich, St Louis MO, USA), and up to 2 μg RNA per reaction. The cDNA was diluted 1:2 with RNAse-free water and used as template in subsequent real-time reverse transcriptase PCR assays.

Real-Time PCR

All real-time PCR experiments were performed on a Chromo4 real-time detector (Bio-Rad, Hercules, USA). All quantitative PCRs were prepared in a dedicated facility in a laminar flow benchhood, using aerosol-resistant pipette tips.

Specific p16INK4a and p14ARF expression was quantitated using a real-time reverse transcriptase PCR assay with SYBR-Green I. The specific primer pairs for p16INK4a and p14ARF, and the temperature were described by Kanellou et al.28 For normalization, we used ACTB as an internal control. ACTB primers were described by Kreuzer et al.29 All primer sequences are listed in Table 2. The temperature protocol was 15 min at 95°C, followed by 40 cycles of 30 s at 95°C, 1 min at 60°C and 30 s at 72°C. Melting curves were made between 50 and 95°C with a temperature increment of 0.2°C. cDNA (1 μl) was amplified in 1 × Qiagen multiplex mastermix containing 1 × SYBR-Green I, and 300 nM of each primer. Each sample was run in duplicate and each run contained two ‘no template controls’ for each gene.

Table 2 Primer sequences
Full size table

Expression analysis and statistical evaluation was made using the pair-wise fixed re-allocation randomization test by the Qiagen REST 2009 Software, V2.0.13.33 Additionally, relative copy numbers were calculated for each patient.

For the purpose of quantification, and to show efficiency and sensitivity, plasmid standard curves were made by cloning the p14ARF, p16INK4A, and ACTB amplicons into pCR2.1, using the TOPO TA cloning system (Invitrogen, Groningen, The Netherlands), according to the manufacturer's instructions. Plasmids were verified by sequencing, and serial dilutions of 107–102 plasmid copies containing the PCR inserts were made. Standard curves were generated by amplified plasmid DNA serial dilutions.

Results

The efficiency of the real-time PCR assays was tested by evaluating the results of serial template dilutions of plasmid DNA containing the respective amplicons. The efficiency of all assays was high, and all assays had a dynamic range between at least 102–107 (Figure 1a). Highly specific melting curves are shown for representative samples (Figure 1b). RNA integrity number (RIN) was analyzed in each of five cervical smears and cervical cell lines (CasKi, HeLa, SiHa and the recently developed HPV16-positive cell lines, Goe and Marq). As expected, RNA isolates from smears were highly degraded (RIN(mean)=4.6, range=2.4–6.6). RNA from cervical cancer cell lines showed less degradation (RIN(mean)=8.5, range=5.1–9.1), although it was exposed to the same storage and extraction procedure. The cervical cancer cell lines were used as positive controls to estimate the expression level of samples with good RNA quality and 100% transformed cells.

Figure 1
figure 1

Sensitivity and specificity of real-time PCR assays. (a) Dynamic range and efficiencies of P14ARF and P16INK4a. (b) Melting curve analysis of SYBR-Green I assays. E=efficiencies, C(p)=crossing point, NTC=no template control.

Full size image

The p14ARF and p16INK4A expression levels relative to ACTB were significantly upregulated in HSIL patients. A significant age-related bias was observed. The results are summarized in Table 3. The data show the N-fold upregulation that was calculated in 45 high-risk HPV-positive tested women with HSIL diagnosis compared with 48 normal women with no detectable HPV infection. Both groups were divided at their median age into four subgroups (Table 1). To analyze the sensitivity and specificity of these tests, we calculated the individual p14ARF and p16INK4a copy numbers in relation to ACTB. The respective cut-offs were defined as the median value for normal patients, plus one s.d. For p16INK4a, we calculated a sensitivity to detect HSIL positive patients of 79% (95% CI=65–88%) and a specificity of 83% (95% CI=67–92%). For p14ARF, we calculated a sensitivity to detect HSIL-positive patients of 74% (95% CI=60–84%) and a specificity of 83% (95% CI=67–93%). A total of 7 out of 48 normal women had an elevated p16INK4a value, and 6 out of 48 normal women had an elevated p14ARF value. Of the 45 HSIL patients, 11 had p16INK4a and 15 had p14ARF mRNA levels below the defined cut-off.

Table 3 Relative p14ARF and p16INK4A expression in relation to dysplasia and age
Full size table

Discussion

Despite the successful reduction of cervical cancer deaths by Pap cytology screening, a reliable biomarker-based screening method that verifies high-grade disease is highly desirable to support cytological findings.

The most popular biomarker for the detection of cervical intraepithelial neoplasia is the immunohistochemical staining for p16INK4a. Rare data exist about p16INK4a mRNA quantification. Our aim was to explore the transcripts p14ARF/p16INK4a with respect to their usability in cancer screening, because higher levels of these mRNAs may indicate HPV type independently the upregulation of E6 and E7 oncogene expression that in turn indicates malignant transformation. However, p14ARF/p16INK4a expression was also described to be a biomarker of age in rodents and in human skin.23, 24 The influence of age in p16-based cervical cancer screening has not been described yet, and may help to improve data interpretation.

We compared normal HPV-negative patients (n=48) with high-risk HPV-positive patients referred as HSIL (n=45) with regard to their p14ARF and p16INK4a expression. We used published SYBR-Green I assays and related the expression to the housekeeping gene ACTB. As expected, we found that relative p16INK4a and p14ARF transcript frequencies increased significantly in samples of patients with HSIL diagnoses. We also found a significant correlation between the patient age and the increased expression of p16INK4a and p14ARF that interfered with HPV-induced upregulation.

Overall, p16INK4a expression was higher than the p14ARF expression in HSIL patients. In women with equal diagnosis, both transcripts were significantly upregulated in older patients compared with younger patients (see Table 3). This age-dependent upregulation can strongly influence the HPV oncogene-related upregulation of p14ARF and p16INK4a. For p16INK4a, the comparison between younger normal women with older HSIL patients would result in an upregulation of 17.12-fold. In turn, older normal women compared with younger HSIL would result in 2.34-fold higher expression only. Therefore, age induces a bias concerning p14ARF and p16INK4a expression.

Patients in the HSIL group were younger compared with the normal group, because of the higher HPV prevalence in young women. Although the age difference of normal women was wider compared with patients with HSIL diagnosis, the age-related upregulation was in the same range.

The quantification of mRNA using real-time PCR has the advantage to be independent from subjective visual sample assessment, as well as being high-throughput compatible. However, the use of mRNA from exfoliated cervical cells is challenging, because cervical scrapes consist of cervical mucus, cellular debris and mRNA is mostly degraded. This critical fact affords small PCR amplicons, good storage solutions and a rapid sample processing. We tested the RNA integrity number in a subset of patients and found mRNA from cervical smears to be highly degraded. Indeed, some samples may be not suitable for mRNA-based analysis. A total of 11 and 15 out of 45 HSIL patients did not reach sufficient p16INK4a and p14ARF levels, respectively. This may be due to insufficient RNA quality. To investigate whether degradation occurred in situ or during storage and extraction, we validated our sampling and extraction procedure by simultaneous treatment of cervical cell lines. The high RNA integrity numbers in the cell lines verified our appropriate extraction and storage procedure. We suggest that beside RNA quality, it may be advantageous to extract the relevant material to reduce the ratio of healthy to dysplastic tissue. Our samples were taken under colposcopic control. This may have enhanced the quality of sampling. Tumor cell enrichment through techniques such as laser capture microdissection was not investigated in this study, as the design aimed to analyze the feasibility of CDKN2A/B mRNA analysis in a standard clinical setting. Our results warrant further analysis of mRNA expression of p16INK4a and p14ARF from cervical smears.