C-terminal deletions of Merkel cell polyomavirus large T-antigen, a highly specific surrogate marker for virally induced malignancy

Authors


Abstract

In 67–100% of cutaneous Merkel cell carcinomas (MCC) the Merkel cell polyomavirus (MCPyV) integrates into the host genome. Mutations and deletions truncating the C-terminal helicase domain of the T-antigen (TAg) protein have been detected in these MCCs, but not in healthy skin specimens. C-terminal deletions of the TAg nucleic acid sequences are characteristic for about 38% of these cases. While the association of MCPyV with MCC has been proven, it is unknown whether MCPyV may play a similar role in other tumor entities. We describe in detail the development and validation of a novel Merkel cell polyomavirus TAg C-terminus deletion assay (MCPyV ΔC-TAg). The triplex real-time PCR quantifies the N- and C-terminal part of the MCPyV TAg gene and the cellular β-globin gene. By comparing the copy numbers of the N- and C-terminal part, deletions of the MCPyV TAg C-terminus are rapidly identified. MCPyV ΔC-TAg was used to assess the physical state of MCPyV TAg in a large series of 55 MCCs, 15 cutaneous lymphomas and 47 forehead smears of healthy individuals. Neither DNA positivity nor viral load was able to discriminate MCCs from the other different types of samples. However, deleted TAg C-terminus sequences were detected only in MCPyV positive MCCs (39%). Consequently, detection of deleted C-terminal TAg sequences appears to be a highly specific surrogate marker for virally induced malignancy and should be used to support novel assumed MCPyV–tumor associations. The study further supports the notion that MCPyV does not play a role in cutaneous lymphoma pathogenesis.

Recently, the Merkel cell polyomavirus (MCPyV) was identified as a causative agent of Merkel cell carcinomas (MCC).1 MCPyV integration into the human host genome is detected in 67-100% of MCPyV positive MCC.1–5 MCPyV is, like other human polyomaviruses (HPyV), a small (40–50 nm in diameter) double-stranded DNA viruses with a circular genome that encodes several proteins, among them the large tumor antigen (TAg).5, 6 The TAg regulates the life cycle of the virus and stimulates the cell cycle of the host cell. TAg not only exhibits tumor suppressor binding domains in the N-terminus but also a C-terminally located origin binding domain and helicase/ATPase functions required for viral genome replication. MCPyV integration in MCC tumors is incompatible with the productive virus life cycle and is further characterized by premature stopcodons, mutations and deletions affecting the C-terminal helicase domain of the Tag.2, 5 These mutations result in C-terminally truncated TAg proteins.5 C-terminal deletions of the TAg nucleic acid sequence are characteristic for about 38% of these MCCs.5 In contrast, when MCPyV is detected in non-tumor sources, e.g., healthy skin or MCPyV unrelated skin tumors, no MCPyV integration and, thus, TAg deletions are detected.7

Currently, many studies are ongoing to study MCPyV prevalence in other tumor entities.6–15 However, since MCPyV DNA is commonly detected on the healthy skin, MCPyV DNA positivity alone is not specific enough to establish a causal role of MCPyV in these tumors. We are describing the prevalence of C-terminal TAg deletions in (i) DNA extracted from 47 forehead smears of healthy individuals with expected episomal MCPyV genomes, (ii) DNA obtained from 55 MCC biopsies with possibly integrated MCPyV and (iii) 15 cutaneous lymphoma biopsies with an unknown MCPyV physical state using a novel triplex real-time PCR assay, called Merkel cell polyomavirus TAg C-terminus deletion assay (MCPyV ΔC-TAg). We show that the detection of deleted TAg sequences represents a promising surrogate marker being highly specific for virally induced malignancy.

Material and Methods

Ethics statement

The study was conducted according to the principles of the Declaration of Helsinki and is covered by ethics committee approvals of the Ruhr University of Bochum and of the University of Cologne. Written informed consent was provided by study participants.

Skin samples

To study the physical state of MCPyV in clinical specimens, three different types of clinical specimens were selected: forehead smears from healthy individuals, cutaneous lymphoma biopsies and MCC biopsies. Samples used for the present study were chosen because of their positive MCPyV-DNA status as determined by a singleplex MCPyV quantitative PCR (qPCR), including also a small proportion of MCPyV-negative samples. Formalin-fixed paraffin-embedded (FFPE) MCC specimens were derived from previous studies8–10 or had been sent to the German National Reference Laboratory for Papilloma- and Polyomaviruses for MCPyV-DNA determination. Included in the present study were 5 MCPyV-DNA negative MCC and 50 MCPyV-DNA positive MCC. FFPE biopsies of various cutaneous lymphomas were also derived from a previous study.11 Two MCPyV-DNA negative (1 mycosis fungoides (MF), 1 folliculotropic MF) and 13 MCPyV-DNA positive lymphomas (3 MF, 4 folliculotropic MF, 1 primary cutaneous aggressive epidermotropic CD8+ T-cell lymphoma, 3 primary cutaneous anaplastic large cell lymphomas, 2 Sézary syndrome biopsies) were included in the present study.11 The skin smears were collected with cotton swabs from an area of about 10 cm2 of normal forehead skin of 47 healthy volunteers (22 males, 25 females, mean age 47.9 years (range 21–89)) participating in a study on cutaneous MCPyV prevalence. Forty-one smears were MCPyV-DNA positive and six swabs did not contain MCPyV-DNA.

DNA extraction and MCPyV DNA load determination

DNA extraction of all samples was performed using QIAamp DNA Mini kit (Qiagen, Hilden, Germany) as described.12 MCPyV-DNA load (MCPyV-DNA copies per β-globin-gene copy) had been determined in all biopsies and smears by qPCR prior to this study using primers and probes located in the N-terminal part of the TAg gene as reported (Light-Cycler 480 real-time PCR, Roche, Mannheim, Germany).16

MCPyV ΔC-TAg assay

The primers and hydrolysis probe were designed with the aid of the LightCycler Probe Design 2 program (Roche) for specific amplification of the TAg N- and C-terminus, both located in exon 2 of TAg, as well as human β-globin gene (Table 1). Primers and probes were designed to completely match all MCPyV TAg and human β-globin sequences available in the National Center for Biotechnology Information nucleotide sequence database (GenBank). All primers and probes were synthesized at TIB Molbiol, Berlin, Germany. Real-time PCR was performed with the Roche LightCycler 480. The amplification conditions were 10 min at 95°C, followed by 45 cycles of 95°C for 10 sec, 60°C for 30 sec and 70°C for 1 sec. The ramp rate was 4.4 °C/sec while heating and 2.2 °C/s while cooling. The PCR amplifications were performed in a 10 μL volume containing 5 μL of 2× LightCycler 480 Probes master mix (Roche Diagnostics, Mannheim, Germany), 0.5–0.1 μM of each primer and 0.2–0.75 μM of each probe (Table 1) and 1 μL of DNA template derived from clinical specimens or plasmid dilutions. In all runs, tubes that contained all PCR components but without template DNA were used to ensure that the reagents were free of contamination. To compensate for spectral overlap of the probes, a color compensation template was generated according to the LightCycler 480 Instrument Operator's Manual. Signals were expressed as Cp (crossing point) values that indicate the cycle at which the increase of fluorescence is highest.

Table 1. MCPyV ΔC-TAg primer and probe sequences
inline image

Standard curves

Standard curves for the N- and C-terminus of MCPyV TAg were obtained by amplification of a dilution series of 1 × 106 to 1 copy of a plasmid clone containing the MCPyV full-length TAg (kindly provided C.B. Buck, NCI, Bethesda, MD) in 50 ng/μL of human placenta DNA. TAg copy numbers were determined on the basis of the molecular weights of the plasmid. The standard curve for human β-globin was obtained from dilution series of 100–0.01 ng of human placenta (HP) DNA. Cellular genome equivalents in the sample were calculated based on the weight of 1 genome equivalent (6.6 pg/cell), i.e., 100 ng of cellular DNA corresponds to 15,151 cells. Viral load in each specimen was expressed as the number of MCPyV TAg N-terminus copies per cell.

Calculations

Absolute quantification of copy numbers was achieved by linear regression analysis comparing the Cp value of the unknown sample against the standard curve with known copy numbers. The ratio of the copy numbers of the TAg N-terminus to the C-terminus (N-to-C-ratio) was used to assess the physical state of the MCPyV genome. The relation between viral load measurements performed with the singleplex qPCR and MCPyV ΔC-TAg was evaluated with the Pearson correlation coefficient. Differences in median viral load of the MCPyV TAg N-terminus across cutaneous smears, lymphoma and MCCs were assessed by the Kruskal–Wallis test for MCPyV ΔC-TAg.

Results

Detection limit and reproducibility of MCPyV ΔC-Tag

The detection limit of each amplification was evaluated in both, three single- and one triplex real-time PCR format. For both formats, as little as ten MCPyV TAg gene copies could be detected by the C- and N-terminal TAg amplification, and 0.01 ng of HP DNA (∼1.5 cell equivalents) by the β-globin PCR. There was a linear relationship between the Cp values plotted against the log of the copy number over the entire range of dilutions (exemplarily shown for MCPyV TAg N- and C-terminus amplification, Fig. 1). All three amplifications displayed similar PCR efficiencies between 1.9 and 2.1. The intra- and interday coefficient of variation was below 1% (data not shown).

Figure 1.

Standard curves for MCPyV TAg N- and C-terminus. Serial ten-fold dilutions from 1 × 106 to 1 copies per PCR of MCPyV plasmid DNA diluted in 50 ng/μL human placenta DNA were analyzed by the MCPyV ΔC-TAg triplex PCR. The detection limit was 10 copies per PCR. PCR efficiencies and R2 values are given.

Specificity of MCPyV ΔC-Tag

The specificity of each of the three real-time PCRs was confirmed by analyzing dilution series of MCPyV TAg and human placenta DNA in the multiplex format. No cross-amplification was observed between primers and probes targeting the MCPyV TAg and HP DNA and vice versa (data not shown). After the color-compensation, no cross-talk was observed between different probes, even when 1 × 106 template copy numbers were analyzed.

MCPyV DNA prevalence, viral load and physical state in clinical specimens by MCPyV ΔC-Tag

DNA extracted from 47 forehead smears of healthy individuals, 55 MCC and 15 lymphoma biopsies were submitted to MCPyV ΔC-TAg.

DNA of the MCPyV TAg N-terminus was detected in 36 of the 41 (88%) previously MCPyV-positive smears of the forehead from healthy individuals, in 7 of 13 (54%) previously MCPyV-positive lymphoma biopsies and in 46 of 50 (92%) previously MCPyV-positive MCC biopsies. In all previously qPCR positive but MCPyV ΔC-TAg negative samples, qPCR had revealed very low MCPyV copy numbers with less than two median MCPyV copies per PCR. With the exception of one forehead smear, all samples that had been MCPyV-negative in qPCR were also negative in the MCPyV ΔC-TAg assay.

Viral load estimates (MCPyV copy numbers per cell) from both methods (qPCR and MCPyV ΔC-TAg assay) showed a good correlation (R2; 0.73) (data not shown). The median viral load of about 0.4 MCPyV copies per cell did not significantly differ between forehead smears, lymphoma and MCC biopsies as measured by N-terminal MCPyV ΔC-TAg PCR (p = 0.69, Fig. 2).

Figure 2.

MCPyV load in MCPyV DNA-positive clinical specimens by MCPyV ΔC-TAg. The MCPyV TAg N-terminus loads (copies/cell) are shown in forehead smears of healthy individuals (n = 36), and lymphoma (n = 7) and MCC (n = 46) biopsies. Each sample was measured in duplicates, the mean value is indicated. The horizontal lines represent the median and the interquartile range.

To assess the physical state of MCPyV, the N-to-C-ratio was analyzed when more than 10 TAg N-terminus copies per PCR were measured (Fig. 3). In episomal state as expected for the forehead smears, the copy numbers for the N- and C-terminus of TAg followed a linear correlation (R2; 0.998). The ratios of the respective copy numbers were close to 1 (mean N-to-C-ratio = 1.026, St.D. 0.31). Based on these findings, specimens were defined to contain C-terminally deleted TAg sequences and, thus, most likely integrated MCPyV, if the N-to-C-ratio exceeded the cutoff of 2.266 (mean value of the forehead smears plus fourfold standard deviation). Ratios that could not be calculated since the C-terminus was not amplified, indicated the presence of deleted TAg sequences in all present MCPyV genomes and, thus, only truncated MCPyV integrates. Ratios around 1 (range 0.441–2.266) indicated the presence of only full-length TAg and, thus, episomal or integrated full-length MCPyV genomes. Based on this cutoff, all MCPyV positive forehead smears and lymphoma biopsies contained episomal or, less likely but cannot be excluded, integrated full-length MCPyV genomes, while 14 of 36 (39%) MCPyV-positive MCC contained also C-terminally deleted TAg sequences (Fig. 3). Among those, the TAg C-terminus was completely absent in four cases (29%).

Figure 3.

Correlation of MCPyV TAg N-terminus load and the TAg N-to-C-ratio in MCPyV DNA-positive clinical specimens. The MCPyV TAg N-terminus copy numbers per PCR (x-axis) and the TAg N-to-C-ratio (y-axis) are shown in forehead smears of healthy individuals, and lymphoma and MCC biopsies. Each sample was measured in duplicates, the mean value and mean ratio is indicated. The horizontal dotted line represents the calculated N-to-C-ratio cutoff of 2.266 indicating the presence of truncated MCPyV integrates in the host genome when more than 10 TAg N-terminus copies per PCR were measured (vertical dotted line).

Discussion

Multiple mutations and deletions frequently affecting the helicase domain of the TAg gene of MCPyV have been described in 67 to 100% of MCCs.1–5 Three of eight (38%) MCC-derived MCPyV described by Shuda et al. showed deletions in the nucleic acid sequence of the entire helicase domain.5 The remaining MCPyV showed multiple nucleotide mutations in this region leading to premature stop codons. In this study, we have established a novel quantitative real-time PCR method, MCPyV ΔC-TAg, for the analysis of the physical state and viral load of MCPyV in clinical specimens. MCPyV ΔC-TAg primers/probe sets target the N- and C-terminal part of the TAg, respectively. While the amplification of the TAg N-terminus is not affected by the alterations following viral integration, at least 38% of MCC-derived MCPyV should fail to be amplified by the TAg C-terminus PCR. Our calculated N-to-C-ratio could be used as an indicator of the viral physical state: is it close to 1, only episomal or integrated full-length DNA is detected, is it above cutoff or cannot be computed due to missing C-terminal signals, MCPyV integrates are truncated suggesting carcinogenic activity of the virus. Of note, the proposed cutoff was highly stringent to minimize false-positive results in case of low N-terminal TAg copy numbers. The cutoff could be lowered to the mean value of the forehead smears plus threefold standard deviation to increase the assay's sensitivity for detecting truncated integrates (Fig. 3).

In contrast to MCPyV, the structurally related human papillomavirus type 16 (HPV16) has already been shown to not only cause cancer of the uterine cervix, but also to be the causal agent in a substantial fraction of other anogenital carcinomas, including vulvar, penile, vaginal and anal cancers, but also in oropharyngeal, in particular tonsillar carcinomas.17 Based on this knowledge, it is not unlikely that MCPyV may play a similar role among the polyomaviruses family. In particular, cancers showing an increased incidence after long-term immunosuppression may be caused by these viruses.17 Accelerated by the recent discovery of the tumorgenic MCPyV in MCC, large-scale epidemiological studies analyzing the association of MCPyV with other skin tumor types, such as basal cell carcinomas (BCC) and squamous cell carcinomas (SCC) or tumors of different entities, including chronic lymphocytic leukemia (CCL), are currently ongoing.9, 11, 13–15, 18–22 However, there is still a lot of controversy about the causal association of MCPyV with these malignancies because MCPyV load is usually lower than in MCC and MCPyV large TAg is not expressed in these tumors.23 Hitherto, MCPyV DNA detection is based on conventional or real-time PCR assays.14, 18, 24, 25 Both systems fail to ultimately proof virus–tumor associations, mostly because MCPyV DNA is commonly detected on the healthy skin.7, 8, 26 Consequently, there is a high demand for highly specific surrogate marker assays, such as MCPyV ΔC-TAg capable of further characterizing virally induced malignancy. MCPyV ΔC-TAg was designed to detect C-terminally deleted TAg and, thus, integrated MCPyV with high specificity, but with reduced sensitivity leading to 60% false negative results because mutations resulting in premature stop codons are not detectable. In excellent agreement with Shuda and coworkers, about 39% of the MCPyV DNA-positive MCCs showed C-terminally deleted TAg sequences in the present study, but none of the MCPyV DNA-positive forehead smears from healthy individuals, resulting in 100% specificity. None of the cutaneous lymphomas analyzed here had an N-to-C-ratio above the cutoff, supporting the notion that MCPyV does not play a role in lymphoma pathogenesis.27, 28 Given the report by Pantulu et al., where a TAg deletion was observed in 8.6% of CLL,29 novel MCPyV-tumor associations should be regarded as profound if a substantial fraction of tumors is MCPyV ΔC-TAg positive and additional confirmatory analyses are performed, e.g., expression analyses of TAg on the RNA and protein level.

Compared to conventional low-throughput cloning and sequencing that was used to identify TAg mutations and deletions,5, 29 or next generation sequencing that was used to identify viral integration sites,30 MCPyV ΔC-TAg is more cost-efficient and exhibits a superior high-throughput ability (90 samples can be analyzed in 2 hr). In addition, sequencing may miss TAg deletions if present only in a minority of MCPyV genomes. Moreover, molecular analysis of MCPyV in other tumor entities requires large collections of biological specimens. These are predominantly available as FFPE tissues, which are suitable for immunohistochemistry (IHC) analysis, but due to the degradation and crosslinking present a technical challenge for nucleic acid analysis, especially when long amplimers are generated. MCPyV ΔC-TAg has an advantage in the analysis of these samples as only short products are amplified. In case of extensive crosslinking, MCPyV ΔC-TAg will still allow to assess the physical state, assuming that cross-linking will equally affect the N- and C-terminal TAg PCR, because both amplimers are of similar length.

Accordingly, MCPyV ΔC-TAg would be beneficial to demonstrate or refute potential novel MCPyV-tumor associations in medium- and large-scale epidemiological studies investigating different tumor entities. Although a certain fraction of integrated full-length MCPyV genomes will be identified as false episomal, the assay's sensitivity will be sufficient to support novel MCPyV-tumor associations.

Of interest, MCPyV load was not able to discriminate between the three different types of samples analyzed in our study and also varied substantially between different MCCs. In good agreement with a previous report,2 the distribution of viral loads in MCC appeared to cluster in a high and low MCPyV load group (high, >0.2 MCPyV copies per cell; low, <0.03 MCPyV copies per cell). All samples from the low MCPyV load group, showed very low TAg N-terminus copy numbers (<7 copies detected per PCR), and were also borderline positive by the previously performed singleplex qPCR. We cannot rule out that some of the low-prevalent MCPyV genomes contain C-terminally deleted TAg, since the calculation of the N-to-C-ratio was not reliable if less than 10 copies were detected per PCR. To date, it is unknown whether the low MCPyV load-positive MCC represent merely commensal MCPyV colonization as frequently found on the skin or whether the integrated MCPyV DNA was lost in the majority of cells after establishment of the tumor.

In conclusion, MCPyV ΔC-TAg allows quantifying MCPyV load and detects C-terminally deleted TAg sequences and appears to be highly suitable for the conduction of large-scale epidemiological studies aiming at deciphering new MCPyV–tumor associations. The study further supports the notion that MCPyV does not play a role in cutaneous lymphoma pathogenesis.

Acknowledgements

The MCPyV plasmid DNA (pW9) was kindly provided by C.B. Buck (NCI, Bethesda, USA).

Ancillary