The main finding of this study is that touchdown general primer PCR coupled with control of the quantity of sample DNA in a PCR supports the detection of low-copy number HPV in an excess background of human DNA sequences. A high percentage of abnormal cervical cytological samples previously recorded as HPV negative tested HPV positive by this approach. It has also been demonstrated that a wide range of HPV types and multiple infections are detectable using a touchdown protocol.
Detection of minority HPV DNA
The potential of PCR to detect minority nucleic acid species is the genius of the technique. However, the presence of background DNA may compromise PCR efficiency by giving rise to non-specific primer annealing . Assays such as the GP5+/GP6+ use one pair of primers to amplify a wide range of HPV types and necessarily each primer contains varying numbers of mismatches relative to any given HPV type . Low-stringency primer-annealing conditions were originally defined for the GP5+/GP6+ assay  however, these conditions permit GP5+/GP6+ primer annealing to human sequences .
In this study a hot start step was introduced to prevent annealing that can occur at PCR set-up between primers and single-stranded DNA sequences produced during DNA extraction [9, 16]. Four touchdown protocols were tested to assess how a touchdown annealing approach affects the amplification of HPV sequences. In the first instance, these protocols were tested on a model system consisting of small quantities of SiHa cell DNA 'hidden' in varying amounts of C-33A cell DNA.
The SiHa cell line is known to contain approximately one slightly truncated copy of the HPV-16 genome integrated in chromosome 13q21-31 . The cell line is hyper-diploid but has been shown to be disomic with respect to chromosome 13 . It is possible to approximate the number of copies of HPV-16 detected by the various protocols tested assuming for the purpose of the approximation that there are 6.6 × 109 DNA base pairs per SiHa cell. One base pair weighs 650 daltons and one dalton weighs 1.66 × 10-24g. Therefore, one diploid cell contains 7.12pg of DNA, and there will be one copy of the GP5+/GP6+ primer target per 3.56pg of SiHa cell DNA extract. It follows that the number of copies of HPV-16 in 0.1 ng of SiHa cell DNA will be ~28; in 0.5 ng SiHa DNA there will be ~140 copies; in 1 ng ~280 copies; in 25 ng ~7000 copies and in 100 ng ~28,000. HPV-16 was amplifiable from smaller quantities of SiHa cell DNA as the total quantity of DNA reduced from 2 μg to 500 ng for all protocols. Two touchdown protocols (TDP3, TDP4) were routinely able to detect 0.1 ng SiHa HPV DNA (~28 copies) in a total DNA quantity of 500 ng (approximately equal to DNA extracted from 70,000 cells [i.e. 1 copy HPV-16 detected per 2500 cells]). In contrast, the best performance of the GP5+/GP6+ protocol was the demonstration of HPV-16 from 1 ng of SiHa cell DNA (~280 HPV-16 copies) in a background of 500 ng total DNA. Unlike TDP3 and TDP4 protocols, the standard GP5+/GP6+ assay failed to demonstrate HPV when the total DNA content in a reaction was 1 μg or 2 μg. It is noticeable from Figure 1 that some of the protocols gave slightly more intense bands, ~150 bp in size, following amplification of HPV-16 sequences in a background of 500 ng DNA compared to the amplification in a background of 100 ng. This empirical finding was reproducible on repeat PCR. This observation is counter-intuitive; however, given the nature of the touchdown approach to annealing, the character of the dilutions, and the high number of PCR cycles, the PCR assays tested are essentially qualitative rather than quantitative.
The significantly improved detection of low-copy HPV-16 by touchdown protocols illustrates the important influence background DNA can have on detection sensitivity. The findings are particular noteworthy given that relative to the target HPV-16 sequence, the GP5+ primer sequence (T
m = 45°C) has just two mismatches and the GP6+ primer (T
m = 41°C) has no mismatches. The data show that under low-stringency annealing conditions and when the target is low-copy HPV, considerable non-specific primer annealing with human sequences must occur despite the near perfect homology of the primers for the HPV-16 L1 open reading frame target.
The MRT GP5+/GP6+ protocol was also found to detect low-copy HPV in a background of human sequences. This assay incorporates periods of slow temperature changes from the denaturation step to the annealing step and from the annealing step to the extension step. This approach may allow for better specific annealing of primers with HPV target during the gradual cooling to the final annealing temperature of 38°C than is possible with the standard conditions. Possibly, the express heating/cooling default settings of current generation thermal cyclers may compromise the efficiency of the GP5+/GP6+ PCR assay as developed using early generation machines [6, 7].
Detection of HPV recombinant plasmid DNA
To examine the effect of a touchdown protocol on a wider range of HPV types, recombinant HPV/plasmid DNA samples were diluted in C-33A DNA and tested with the TDP3 protocol. There are three primer mismatches with respect to HPV-33, one in the GP5+ primer and two in the GP6+ primer. For HPV-45 there are three mismatches that are all in the GP5+ primer sequence. HPV-31 and 56 each have one mismatch in the GP5+ primer and three in the GP6+ primer. For HPV-52 there are seven mismatches, including five in the GP5+ primer. HPV-51 has the greatest number of mismatches of any HPV type amplified by the GP5+/6+ system: six mismatches in the GP5+ primer and four in the GP6+ primer. Not surprisingly, HPV-16 was the most efficiently amplified type (Figure 3). Amplification product was detectable from 15ag DNA (approximately equivalent to 1 copy of the recombinant HPV plasmid) in a background of 100 ng human DNA (equivalent to 14,000 cells). HPV-45 and 51 were amplifiable from 150ag (10 HPV/plasmid copies); types 33, 52, and 56 from 15fg (100 copies of HPV) DNA; and, HPV-31 from 150fg (1000 copies) of recombinant DNA. Clearly, for such small quantities of DNA, slight variations in pipetting accuracy and/or DNA concentration estimate might alter the final sensitivity data. Nonetheless, the data indicates firstly that a touchdown protocol can sensitively amplify a range of HPV types, and secondly that the efficiency of amplification may not be a simple function of the number of mismatches between the primer and target sequence. With the TDP3 protocol HPV-31 (4 mismatches) was less efficiently amplified than HPV-56 (4 mismatches), HPV-52 (7 mismatches), or HPV-51 (10 mismatches) (Figure 4, Table 3). The TDP3 amplified six of the seven HPV types tested with greater sensitivity than the standard GP5+/GP6+ conditions. HPV-31 was detectable as a faint band from 15fg template DNA following the standard PCR, whereas the detection limit with the TDP3 protocol was 150fg (Figures 3 and 4, Table 3). HPV-52 was not detectable by the standard protocol at the dilution range tested but was detectable from 15fg starting template using the TDP3 protocol. These data again indicate that the detection efficiency for different HPV types by a general primer method may not follow easily predictable rules. General primer annealing involves a dynamic inter-relationship depending on HPV copy number, primer affinity for different HPV types, human sequence concentration, and, primer affinity for human sequences at a given annealing temperature. The empirical data indicate that a touchdown annealing approach helps negotiate these variables to support sensitive detection of a wide range of HPV types.
Detection of multiple HPV types
This study also examined the ability of a touchdown protocol to amplify HPV types-16 (2 mismatches), 45 (3 mismatches), 51 (10 mismatches), and, 52 (7 mismatches) combined in one PCR (Figure 5). Again, HPV-16 was detected with the highest sensitivity; however, the other three types were identifiable following dot blot hybridization demonstrating that the touchdown protocol did not favor HPV-16 amplification to the exclusion of other types. An exhaustive assessment of the effect of different concentrations of HPV types within a PCR on the detection of multiple types was beyond the resources available for this study.
Detection of HPV in cervical smear samples
An excess of background human DNA is likely to be a common situation when performing PCR on DNA extracted from routine cervical smear samples. The extent to which normal or lesion tissues are scraped from the cervix cannot be controlled. For efficient HPV PCR amplification, the study data strongly support the use of standardized quantities of DNA and not regular volumes containing arbitrary amounts of DNA. Overall, the cell line and clinical sample studies indicate that in a 50 μl reaction, between 100 ng and 200 ng sample DNA supports the most efficient PCR. Using optimized amounts of DNA, the TDP3 protocol demonstrated one or other of seventeen different HPV types in 47 of 65 (72%) abnormal cytology samples HPV negative by the regular GP5+/GP6+ protocol. Multiple infections were found in 9/47 (19%) of the samples.
The MRT GP5+/GP6+ conditions were tested on 54 of the 65 GP5+/GP6+ negative samples. Twenty-eight (64%) of 44 samples positive by the TDP3 were also positive by the MRT protocol, and 1 of 10 samples negative by the TDP3 protocol was positive by the MRT. Of an additional 25 abnormal cytological samples negative for HPV by the TDP3 assay, none tested positive with the standard GP5+/GP6+ method, but two samples were positive by the MRT protocol. Again, these data show that HPV detection can be highly dependent on assay conditions and also demonstrate that two or more methods may be required for inclusive HPV screening. Nevertheless, the above comparison of methods together with the demonstration of HPV in 94% of 799 abnormal cytological samples indicates a touchdown approach supports wide-ranging and sensitive HPV detection.
Detection of HPV in FFPE samples
This study also examined the effect of touchdown PCR and sample DNA quantity on the amplification of HPV from FFPE tissues, where DNA quality is likely to be poor. VAIN lesions were chosen as an example of tissues from which HPV virions are actively shed and therefore where HPV DNA is in a relative abundance . Amplification of HPV from VAIN lesions was readily accomplished by both the GP5+/GP6+ and TDP3 protocols although the TDP3 assay was able to amplify HPV from smaller quantities of template DNA than the GP5+/GP6+ assay (Figure 8). The data again illustrates the effect of DNA content on amplification efficiency. Assay of 114 VAIN lesions by TDP3 has demonstrated 21 different HPV types in 111 HPV positive samples, with double or multiple infections in 17 (15%) of the samples, again suggesting that a touchdown approach sensitively detects a wide range of HPV types.
Breast carcinomas were chosen for study as an example of a tumor where there are widely differing estimates of HPV prevalence. Most early studies of breast tumors found no evidence of HPV following PCR assay [5, 19–21]. However, other investigations (by Southern blot hybridization, and by PCR) have reported high-risk HPV types (e.g. HPV-16, 18, or 33) in 24/50 (48%), 5/17 (30%), 19/41 (46%), 14/32 (44%), and 6/17 (35%) breast carcinomas [4, 22–25]. de Villiers et al. PCR tested paired breast carcinoma and nipple tissues from 29 patients for anogenital and skin HPV types. HPV was detected in 25/29 (86%) carcinomas and in 20/29 (69%) nipple tissues . In the present study, application of the GP5+/GP6+ assay using an arbitrary DNA quantity in the PCR indicated 5/26 (19%) invasive ductal carcinomas were HPV positive. Subsequent assays using the TDP3 protocol and 200 ng of sample indicated 15/26 (58%) tumors were HPV positive. These data add to the evidence that HPV sequences are present in breast tumor tissues. However, that HPV was detectable in most tumors only after use of a highly optimized PCR protocol may indicate that HPV is in a latent form and/or is confined to a subset of tumor cells or other cells associated with a tumor mass. Perhaps there may be parallels with Epstein Barr Virus (EBV) detection in breast tumors. EBV has been reported in up to 51% of breast carcinomas by sensitive PCR assay , but it has since been suggested the EBV detected is amplified from EBV positive tumor infiltrating lymphocytes and not from tumor cells per se . Further studies are required to determine whether HPV detected in breast tumors represents an incidental 'passenger' or has causal significance. Additionally, given the disparate estimates of HPV prevalence in breast tumors, future studies might incorporate more rigorous control measures such as cutting sections of HPV negative tissues for DNA extraction and PCR in between successive breast tumor specimens to test for cross-contamination with HPV positive samples.
Carry-over contamination control
In this study dTTP was replaced with dUTP in all protocols in order to identify a sensitive PCR test that also includes a control measure against carry-over contamination. However, uracil N-glycosylase (UNG) is an enzyme that has a degree of thermostability and may also become reactivated after completion of a PCR protocol [29, 30]. HK™-UNG (Epicentre) is designed to be heat-labile and free of such shortcomings, but it has been reported that even heat-labile forms of UNG may retain some residual activity and spoil PCR efficiency . Further, dUTP is less efficiently incorporated into PCR amplicons than dTTP . UNG/dUTP usage might therefore result in suboptimal PCR assay performance and even greater sensitivity for HPV might be possible for all the protocols tested using regular dNTPs.
The impetus for this study was to determine whether abnormal cytological samples that tested HPV negative by a commonly used PCR assay were in fact HPV positive. Using a touchdown protocol, HPV has been demonstrated in a high percentage of samples previously recorded as HPV negative consistent with HPV representing a necessary cause of most (if not all) abnormal cervical cytological conditions. However, an assay with high analytical sensitivity for HPV might not be appropriate as a clinical test. The HPV status of cytological samples has been proposed as a marker to identify patients with underlying high-grade cervical intraepithelial neoplasia (CIN)  and current clinically approved HPV tests detect the bulk of high-grade CIN and invasive cervical carcinoma. Increased sensitivity for HPV might lead to reduced test specificity for high-grade CIN.