The retinoblastoma protein/p16INK4A pathway but not p53 is disrupted by human papillomavirus in penile squamous cell carcinoma

Stankiewicz E, Prowse D M, Ktori E, Cuzick J, Ambroisine L, Zhang X, Kudahetti S, Watkin N, Corbishley C & Berney D M (2011) Histopathology58, 433–439 The retinoblastoma protein/p16INK4Apathway but not p53 is disrupted by human papillomavirus in penile squamous cell carcinoma


Introduction
Penile carcinoma is rare in developed countries, representing 0.3-0.5% of male malignancies in Europe and the United States. 1 In the United Kingdom there are approximately 600 (rate one per 100 000) new cases each year, mainly after the sixth decade. 1,2 The vast majority (95%) are squamous cell carcinomas (SCC). 3 These may be divided into usual type (70%) followed by more aggressive basaloid (10%) and a slow-growing, low-grade group of 'verruciform' tumours (20%). Verruciform lesions include verrucous carcinoma, warty carcinoma and papillary squamous cell carcinomas. 4 Mixed tumours of usual type and one or more of other subtypes of SCC also exist.
Risk factors for penile cancer include lack of circumcision during childhood, phimosis and cigarette smoking. 5 There is some confusion as to the role of infection with human papillomavirus (HPV) in penile cancer 3 as, contrary to cervical cancer, 6 the incidence rate of HPV positivity varies from 15% to 77.5%, 7,8 depending upon detection range, population studied and tumour type. 9 HPV16 is the most prevalent infection, 7 similar to other HPV-related ano-genital cancers. 10 Basaloid and warty tumours have been shown to be associated strongly with human papillomavirus infection. 11 Therefore, penile cancer may resemble vulvar cancer, which has two different aetiologies, one related to HPV infection and one that is not. 12 The carcinogenic abilities of high-risk HPV types are well known due to extensive studies on cervical cancer. Viral oncoproteins E6 and E7 can disrupt cell-cycle checkpoints and apoptosis by interacting respectively with tumour suppressor proteins, retinoblastoma (RB) protein and p53. RB protein regulates cell-cycle progression, protecting the cell from uncontrolled proliferation, and is regulated by cyclindependent kinase phosphorylation. p16 INK4A can inhibit cyclin-dependent kinase-mediated RB protein phosphorylation, preventing cell-cycle progression. In cells infected with high-risk HPV viral E7 protein binds directly to RB protein, causing its inactivation and down-regulation, which prevents cell-cycle control by p16 INK4A . 13,14 In these circumstances accumulation of p16 INK4A can occur, which is an indication of HPV infection. 15 p53 can also inhibit cancer development and tumour growth through its ability to efficiently inhibit cell proliferation and promote apoptotic cell death. 16 In cancers harbouring high-risk HPV, expression of viral E6 protein can inactivate p53 through its down-regulation, and an inverse correlation between HPV positivity and p53 overexpression has been found in some cancer sites but not others. 17 HPV E7 protein can additionally overcome the inhibitory function of p21. p21 is a p53-responsive protein and induces cell cycle arrest in the presence of DNA damage. E7 protein binds to p21 and abrogates its inhibitory functions, therefore overcoming DNA damage-induced cell-cycle arrest, despite high levels of p21. 18 The mechanisms of oncogenesis in penile cancer are not fully understood. There are no data on RB protein expression in penile cancer and results on p53 in relation to HPV are inconclusive. Limited data are available on p16 INK4A immunodetection 19,20 and p21 expression 21 in penile tumours. Therefore, we investigated HPV infection type in one of the largest series of penile SCC to test the hypothesis that HPV type and its association with key cell-cycle proteins has differential effects on the tumour subtypes, in order to elucidate their role in tumour pathogenesis.

Materials and methods
The study was conducted upon approval from East London and the City Research Ethics Committee. We reviewed retrospectively the Cellular Pathology Department Registry of St George's Hospital to identify patients treated for penile SCC between 2001 and 2007. We retrieved 148 penile SCCs. Ninety-seven samples were usual-type SCCs, 17 basaloid, 15 pure verrucous carcinomas, seven mixed verrucous ⁄ usual type, seven mixed verrucous ⁄ warty, two warty and three warty ⁄ usual types. Twenty-one cases were obtained from excision biopsies ⁄ circumcisions, 82 from glansectomies and 45 from partial ⁄ total penectomies. All cases were re-reviewed by an expert uropathologist (CC), including subtyping, grading and staging by standard methodologies. 4,22 polymerase chain reaction ( pcr) One hundred and two wax blocks from penile SCC cases were available for DNA extraction with a QIAamp DNA Mini kit (51304; Qiagen, Crawley, UK). Beta-globin PCR was performed using primers B1 and B19 to confirm the adequacy of the extracted DNA. Validated samples were tested for the presence of HPV DNA by a broad-spectrum HPV PCR method using short PCR fragment (SPF10) primers, which amplify a 65-base pairs (bp) fragment of the L1 open reading frame and HPV genotypes identified by the INNO-LiPA line probe assay (Innogenetics NV, Ghent, Belgium). 19 immunohistochemistry Tissue microarray blocks were prepared using a manual microarrayer. Three · 1 mm tissue cores were taken from each tumour. Four lm sections were cut and immunostained using standard heat-induced antigen retrieval methods and the ABC kit (PK-6200; Vector Laboratories, Peterborough, UK), according to the manufacturer's instructions. 23 Primary antibody dilutions were: 1:50 for RB (NCL-RB-358; Novocastra, Newcastle, UK), 1:100 for p16 INK4A (MS-1064-PO; Neomarkers, Fremont, CA, USA), 1:1000 for p53 (M7001; Dako, Glostrup, Denmark) and p21 (M7202; Dako, Glostrup, Denmark). The antibody used against RB (clone 13A10) binds to the N-terminal region of the protein and detects RB regardless of phosphorylation status. Positive controls included CIN III for p16 INK4A , placenta for p21, anaplastic thyroid cancer for p53 and tonsil for RB. The staining pattern of RB, p53 and p21 was nuclear. p16 INK4A showed both nuclear and cytoplasmic immunoreactivity.
Sections were scored semiquantitatively by a consultant genitourinary pathologist (DB). For nuclear positivity each core was given an estimated visual score between 0 and 100%, representing the percentage of positively stained neoplastic nuclei. The intensity of staining was also measured as: 1 (weak), 2 (medium) and 3 (strong). The final score was deduced by multiplying the percentage of staining by intensity to give an expression score from 0 to 300. p53 expression was always strong; therefore, nuclear score alone was applied. Cytoplasmic expression of p16 INK4A was determined by intensity of staining alone. The core with the highest score was selected for analysis. Statistical analysis was performed using StatsDirect software, version 2.60.6000. The correlations between antibodies were evaluated using Spearman's rank correlation test and the tumour type or HPV infection was evaluated by v 2 test or Fisher's exact probability test. Comparisons between antibody expression in different histological subtypes of SCCs were restricted to usual type, verrucous and basaloid only. The warty group of tumours was too heterogeneous and included only two pure warty samples. Similar to previous papers, the cut-off points selected for antibody positivity were: >0 for p16 INK4A , 24 ‡5 for p53 and >5 for p21. 25,26 As RB is normally highly expressed in tissue we chose the median value of 240 and above (cut-off ‡ 240) as indicative of high expression of RB in penile SCC. All analyses were two-sided; P < 0.05 was considered to be significant.

Results
One hundred and forty-eight tumours were analysed, which comprised 97 usual-type SCCs, 17 basaloid, 15 verrucous, seven mixed verrucous ⁄ usual type and 12 mixed warty and other SCC subtypes. The histopathological features of the tumours are listed in Table 1. Of these, 102 cases were also suitable for HPV analysis.
Differences in HPV infection were observed between the histological subtypes of PSCC. For the usual type, HPV DNA was detected in 38 of 64 (59%) tumours, with high-risk HPV16 present in 33 of 38 (87%) cases. Mixed warty subtypes were positive for HPV DNA in six of 11 (55%) cases, with HPV16 present in three of six (50%) cases. Basaloid tumours showed higher positivity, 10 of 13 (77%) for HPV DNA, with HPV type 16 present in 100% of these cases. In contrast, HPV was detected in only three of 13 (23%) verrucous tumours, and HPV16 was not found (0%). immunohistochemistry The positive expression of proteins and mean values are listed in Table 3. High RB protein expression ( Figure 1A) was detected in 85 of 147 (58%) of penile SCC, and significant differences were observed between histological groups (P < 0.0001). A high percentage of verrucous cases (87%), an intermediate number of usual type (60%) and few basaloid cancers (12%) expressed high RB levels. The mean RB expression was twofold lower in basaloid than in verrucous and usual subtypes. p16 INK4A demonstrated both nuclear and cytoplasmic immunoreactivity ( Figure 1B). Overall, 47% of PSCCs were positive for cytoplasmic p16 INK4A expression, with significant differences between histological groups: 13% of verrucous, 52% usual and 94% of basaloid cases were p16 INK4A -positive. Basaloid samples had very high mean expression of nuclear and cytoplasmic p16 INK4A , while mean expression of p16 INK4A in verrucous samples was very low and    Figure 1C) showed no significant difference between histological subtypes of SCC, and was present in 79% of cases overall. Intensity of p21 staining was weak ( Figure 1D) and present in 62%, with no difference in the expression between different SCC subtypes. No relationship was detected between p53 and p21 expression. There was a positive correlation between p21 and p16 INK4A expression (Spearman's q = 0.658793, P < 0.0001) and negative correlation with RB (Spearman's q = )0.499952, P < 0.0001).
There was also a strong positive correlation between HPV infection and p21 (P = 0.0002) and p16 INK4A (P < 0.0001) immunodetection and negative correla-tion with RB expression (P < 0.0001) in penile SCC. p53 did not show any correlation with HPV infection (P = 0.5682).

Discussion
The rate of HPV infection in penile cancer varies widely, depending upon the population studied and sensitivity and specificity of the method used. Our results suggest that in a developed country, unlike cervical cancer, penile cancer has at least two aetiologies: one HPV-related and one unrelated (similar to vulvar cancer), as we detected HPV DNA in 56% (57 of 102) of PSCC cases. This is consistent with our previous report of HPV prevalence in PSCC of 54%, 19 and a recent review that found that 48% of 1266 cases from 30 studies of invasive penile cancer were HPV-positive. 27 We confirm the existence of differences in HPV infection between histological subtypes. 11 HPV prevalence in usual-type SCC varies between 11% and 71%, 7,8,28 and the 59% (38 ⁄ 64) reported by us falls well within this range. As reported previously by our group, verrucous tumours were mainly HPV-negative, confirming the lack of HPV involvement in this neoplasm, 29 while basaloid carcinomas showed a strong correlation with HPV infection. 11,28 HPV16 is the most prevalent type in our study and was detected in 81% (46 of 57) of positive samples and in more than half of these as a single infection, suggesting that this HPV genotype is more likely to contribute to the carcinogenic process. However, 17 of 57 (30%) patients had multiple HPV-type infections, which is similar to other penile studies. 30,31 The significance of multiple HPV infections in cancer development is not clear, and reports on increased risk for carcinoma in women with multiple HPV infections, compared to those infected with a single HPV type, are contradictory. 32,33 Comparing these results with protein expression reveals that basaloid tumours have an aetiology related to high-risk HPV infection, which manifests itself in high p16 INK4A and decreased RB expression, as has been shown in cervical cancer. 34 The aetiology of usual-type SCC can be attributed to HPV infection in only approximately half the tumours, showing corresponding loss of RB and gain of p16 INK4A protein expression. This is analogous to the involvement reported for HPV in carcinogenesis of anal SCC 35 and tonsillar carcinoma. 36 Detection of p53 protein by immunohistochemistry in penile cancer varies between 41.5% and 89%, 21,37,38 and there is a lack of reports comparing different histological types. We detected p53 immunostaining in 79% (103 of 143) of penile SCCs, and the value was very similar regardless of histology (Table 3). There was no correlation between high-risk HPV infection and p53 immunostaining, which is in agreement with previous reports on penile SCC. 37,38 We used the DO-7 clone antibody, which mainly detects mutated p53 but is also able to recognize wild-type (wt) p53. Therefore, it may detect wt p53 when it is highly expressed in cells. Highrisk HPV infection also causes oncogenic stress to the cell, which induces normal cellular responses such as increased p53 levels in order to induce cell cycle arrest. Keratinocytes or fibroblasts expressing high-risk HPV E7 protein are well known to overexpress wt p53; however, there is evidence that in E7 expressing cells wt p53 is transcriptionally inactive. 39 Additional studies on p53 mutations are necessary to confirm its mutational status in these tumours. p21 was expressed in 62% (88 of 143) of penile SCC, with no significant difference between tumour subtypes. Lam and Chan 21 showed lower p21 expression in penile cancer but in a much smaller cohort. Interestingly, p21 did not correlate with p53 and some cases expressed high levels of p21 despite low or absent p53 protein, suggesting p53-independent activation of p21. Surprisingly, similar to tonsilar SCC, 36 we found a positive correlation of p21 with HPV infection (P = 0.0002). Additionally, p21 correlated positively with p16 INK4A (P < 0.0001) and negatively with RB protein expression (P < 0.0001). Funk et al. 18 reported that high-risk HPV16 E7 protein can bind directly to p21 and abrogate DNA damage-induced cell-cycle arrest, despite high levels of p21. He suggested that it is possible that the release of E2F from RB and inactivation of p16 INK4A and p21 are all necessary for the ability of E7 to bypass cell-cycle arrest signals. Conversely, there is emerging evidence that p21 in certain cancers may itself act as an oncogene and actually promote proliferation. 40 To our knowledge, this is the largest study to examine penile cancer pathogenesis by comparing HPV type with proteins affected commonly by HPV infection. We have demonstratedthat HPV infection in penile SCC disrupts the RB ⁄ p16 INK4A pathway through down-regulation of RB and elimination of cell-cycle control from p16 INK4A , manifesting itself in accumulation of p16 INK4A , which fails to block cell-cycle progression. p21 was widely expressed, consistent with abnormal cell-cycle regulation. However, p21 seems to be regulated independently from p53 and may be involved in the oncogenic process. We confirm that penile tumours seem to have two different aetiologies: one related to HPV and one unrelated. These data suggest that use of the bivalent HPV16 ⁄ 18 prophylactic vaccine in men could reduce the occurrence of penile SCC by about 45%.