Activation of AKT and nuclear accumulation of wild type TP53 and MDM2 in anal squamous cell carcinoma

Abstract

Human papilloma virus (HPV) infection is considered as an important aetiological factor for anal squamous cell carcinoma (ASCC) but is not sufficient for tumour progression. This carcinoma is poorly understood at the molecular level. Using the largest cohort of cases to date we investigated the molecular mechanisms underlying ASCC development, in particular the roles of TP53, MDM2 and AKT. Viral infection in our cohort occured at high frequency (73%, 94/128) with HPV16 accounting for the majority (86%, 81/94) of infected cases. Only 4% (5/119) of ASCCs showed TP53 (exons 5–8) mutations, but a high frequency (91%, 100/110) of nuclear protein expression of TP53 was observed. There was a significant association (p < 0.001) between nuclear accumulation of TP53 and MDM2 protein although no MDM2 mutations were found, and copy number was normal. Cellular accumulation of phosphorylated-AKT was observed in 66% (82/125) of ASCCs and an association demonstrated between nuclear accumulation of MDM2 and activated AKT (p < 0.001). We observed a high frequency of copy number gain at PIK3CA (47%), and some coding sequence mutations (4%). Amplification of PIK3CA was associated with presence of phosphorylated-AKT (p= 0.008). There was no association between virus infection and TP53 nuclear accumulation (p = 0.5). However, a significant association was found between infection and MDM2 nuclear staining, and between infection and activated AKT (p = 0.04, p = 0.01, respectively). We propose that activation of AKT, possibly through the PI3K-AKT pathway, is an important component of ASCC tumorigenesis that contributes to MDM2 and TP53 accumulation in the nucleus. © 2007 Wiley-Liss, Inc.

The annual incidence of anal squamous cell carcinoma (ASCC) is 1.4–2 per 100,000 in the general, heterosexual population and accounts for approximately 300 cases per year in the UK and 3,500 in the USA.1, 2 Sexually transmitted infection with mucosal high-risk human papillomavirus (HPV) causes anal intraepithelial neoplasia, which may progress from low-grade through to high-grade and finally to invasive neoplasia.3, 4 This process has distinct parallels with cervical intraepithelial neoplasia, including, origin within squamous epithelium and infection with some viral subtypes, such as HPV type 16, which appear to confer a significant risk of malignant transformation.

ASCC is poorly understood at the molecular level and it remains to be documented whether mutational pathways in ASCC reflect those seen in cervical neoplasia, or whether there are parallels with colorectal cancer. The most common genetic alteration in human cancer is mutation of the TP53 tumour suppressor gene. To date, mutation of TP53 in ASCC has been examined in a single study that investigated only 9 tumours and found 3 coding sequence mutations.5 Whilst the frequency of TP53 mutation needs to be clarified, nuclear accumulation of TP53 protein has been shown in 37–71% of anal tumours.6, 7, 8 Whether wild type or mutant TP53 protein can account for the observed nuclear staining is unclear.

Although the genetic and functional status of TP53 in ASCC is poorly characterised, the E6 gene product of HPV is known to inactivate TP53 protein by promoting its degradation.9 Binding to the MDM2 oncoprotein, which then leads to ubiquitination and proteasomal degradation, may also inactivate TP53. Alternatively, MDM2 may bind to the TP53 transactivation domain, thereby inhibiting TP53 transactivation without subsequent degradation.10 Over-expression of MDM2, resulting in TP53 inactivation has been detected in multiple tumours, including 6% of breast carcinomas, 24% of sarcomas and 30% of bronchial carcinomas.11, 12, 13 However, the role of MDM2 in HPV infected cells remains unclear particularly as the MDM2 degradation pathway is inactive in HPV16 E7-expressing human lung fibroblasts.13

MDM2 has not been studied in anal cancers previously, but both TP53 and MDM2 have been investigated in cervical cancer. MDM2 accumulation was observed in 21–32% of cervical cancers and a positive correlation was shown between nuclear accumulation of TP53 and MDM2 proteins.14, 15 MDM2 accumulation occurs by various mechanisms including: gene amplification; mutation; over-expression of the MDM2 protein stabilizer: phosphorylated-AKT; and through a single nucleotide polymorphism (SNP, SNP309; T to G) in the MDM2 promoter. Amplification has been found in 20% of soft tissue tumours, 6% of breast carcinomas, and 6% of melanomas.11, 16, 17 MDM2 protein levels are increased by gene mutations at cysteine 438 and cysteine 464 which target MDM2 for degradation; glutamine 58 which is necessary for interaction between MDM2 and TP53; and the nuclear export signal (amino acids 197–211).10 The G allele of SNP309 confers increased affinity of the transcriptional activator Sp1 for the P2 promoter with a resultant increase in MDM2 transcription.18

AKT is a protein kinase with numerous roles in the promotion of cell proliferation and inhibition of apoptosis that is activated by phosphorylation at threonine 308 and serine 473; AKT stabilises MDM2 by phosphorylation.19 Genomic amplification and oncogenic activating mutations, that result in an increase in AKT activity, have been demonstrated in PIK3CA in colorectal (32%), gastric (25%), hepatocellular (36%) and breast (18–40%) cancers.20, 21 Amplification of PIK3CA has been shown in 70% (28/40) of cervical cancers and a positive correlation shown between high level PIK3CA amplification (>3 copies) and phosphorylated-AKT immunostaining.22

Using the largest cohort of ASCC samples analysed to date, we now report an investigation into the molecular mechanisms underlying the development of anal cancer.

Material and methods

Patients and samples

One hundred twenty-eight patients with ASCC (cases) were identified from the pathology archives of St. Mark's and Northwick Park Hospital, Harrow, UK and Queen's Medical Centre, Nottingham, UK between 1988 and 2004. An expert gastrointestinal histopathologist has reviewed all the cases. Peri-anal, rectal tumours and those of histological type other than squamous cell carcinoma were excluded. Sixty specimens of normal anal epithelium (controls) were identified from the St Mark's Hospital database and processed as described for ASCC. All normal controls had been previously diagnosed as haemorrhoids and were only selected if there was no evidence of inflammation, dysplasia or wart infection on examination by consultant pathologist. Paraffin embedded blocks of relevant specimens were sectioned into 10 μM slices and DNA extracted using standard protocols. Specimens containing tumour and normal tissue on the same slide were dissected under a microscope and processed separately. Ethical approval for this study was obtained.

HPV typing

Presence of HPV DNA was established in all samples by PCR with consensus Gp5/Gp6+ primers.23 This primer set amplifies 160 bp of the viral L1 gene which shares 70% homology amongst oncogenic HPV types and can detect the majority of mucosal HPV types, including all the predominant high risk types (HPV16, 18, 31, 33, 45). PCR reaction conditions are available on request. A plasmid containing the full HPV genome was used as a positive control (provided by A. Storey and K. Purdie, Centre for Cutaneous Research, Queen Mary's School of Medicine and Dentistry). PCR products were purified and sequenced using Big Dye Terminator version 3.1 according to manufacturer's protocol (Applied Biosystems, CA, USA). Only nucleotide changes verified by repeat amplification and sequencing in both directions were recorded as mutations. DNA sequences were checked against the NCBI database using nucleotide–nucleotide BLAST to determine the type of human papillomavirus.

Mutation screening

Exons 5–8 of TP53, exons 4, 8 and 11 of MDM2 and exons 9 and 20 of PIK3CA were direct sequenced. Exons were chosen based on previous published mutation data. Genotype analysis was also performed for a functional T to G SNP at nucleotide 309 in the MDM2 promoter region (rs2279744; SNP309) presumed to be germline in all cases using standard laboratory protocols. ASCCs were assessed for microsatellite instability using BAT26. PCR primers and conditions are available on request.

Immunohistochemistry

Immunohistochemistry was performed on 4 μM sections of formalin fixed, paraffin-embedded tissue samples using the avidin–biotin complex protocol. Sections were dehydrated, deparaffinised and stained with one of the following mouse monoclonal antibodies: beta-catenin (E-5, Santa Cruz, CA) 1:100 dilution overnight at room temperature; TP53 (DO-7, Dako Cytomation, USA) 1:25 dilution at 4°C overnight; MDM2 (Santa Cruz, CA) 1:200 for 1 hr at room temperature; and phosphorylated-AKT (Cell Signalling, Hitchin, UK) 1:180 at 4°C overnight. The antibody used to assess the activation state of the PI3K-AKT pathway shows specificity towards serine 473-phosphorylated AKT; phosphorylation at serine 473 is required for full activation of AKT. Positive controls were used in each experiment: breast carcinoma for phosphorylated-AKT; colorectal carcinoma for TP53 and beta-catenin; and leiomyosarcoma for MDM2. Negative controls were prepared by omitting the primary antibody and showed no evidence of staining.

Most slides used for immunohistochemistry had some normal epithelium surrounding the ASCC and served as internal controls. Cut off values for positive staining were selected based on previously published studies.14, 22 Samples were scored positive for TP53 and MDM2 if more than 10% of tumour nuclei demonstrated staining.14 Positive staining for phosphorylated-AKT was reported if there was nuclear or cytoplasmic staining in more than 30% of tumour cells.22 Scoring was performed by two independent observers and less than 5% inter-observer variation found; a third observer resolved any disagreement.

Quantitative PCR

Genomic DNA samples were purified with QIAamp DNA Micro kit (Qiagen, Hilden, Germany; n = 96) and assessed for amplification of MDM2 and PIK3CA. ALB and GCK were used as reference genes, respectively. As a representative of normal anal epithelium, control DNA was extracted and purified from paraffin embedded haemorrhoid tissue (n = 15) in the same manner as tumour DNA. Primers and probes were used as described previously.22, 24 All samples and blank controls with no template, were tested in triplicate.

Samples were analysed on an ABI 7900 with an automatic threshold and a Ct value was obtained for each reaction. All samples were run in triplicate and a mean value was calculated from the 3 Ct values for both the test primer (MDM2/PIK3CA) and for the control primer (ALB/GCK). Data was analysed by the ΔΔCt method where ΔCt = mean Ct test (PI3K/MDM2) − mean Ct control (GCK/ALB). At least 6 normal anal epithelium samples were run in each QPCR and a mean ΔCt(normal) was obtained. Finally, the ΔΔCt value was obtained for each individual tumour sample using the following formula: ΔΔCt = ΔCt(normal) − ΔCt(tumour).

Statistics

Data were analysed in Stata 8.2 (Stata statistical software release 8.2; Stata Corporation College Station, TX, USA) as 2 × 2 tables with Fisher's exact test; 2-tailed tests were used and a p-value of <0.05 considered significant.

Results

HPV detection and typing shows high frequency of HPV infection

Tumour samples underwent PCR with consensus Gp5/6+ primers and 73% (94/128) showed a positive PCR product (Fig. 1). The most prevalent type was HPV16, which was isolated in most samples (86%, 81/94). HPV 18 was present in 3% of samples (3/94), HPV 6 in 2% (2/94), and HPV 31, 58 and 59 in 1% (each, 1/94). It was not possible to differentiate between HPV 16, 31 and 33 in 4% (4/94) of cases due to sequence homology of the L1 amplicon. A series of normal controls (n = 60) were also tested for HPV infection and no HPV types detected in any of these samples.

Figure 1. Detection of HPV using PCR with consensus Gp5/6+ primers. Representative examples: lane 1, positive (POS) control, plasmid DNA containing HPV genome; Lane 2 negative (NEG) control, haemorrhoid DNA; lanes 3-8, ASCC samples; lane M, DNA size ladder showing 300, 200 and 100 bp.

TP53 and MDM2 are mutated at a low frequency, but nuclear accumulation of both proteins occurs in most anal squamous cell carcinoma

Mutation of TP53 was found in 4% of ASCC (5/119), all in exon 5 (summarised in Table I). There were no mutations in exons 6, 7 or 8 of TP53 or in any region of MDM2 examined including the nuclear export signal (exon 11), cysteines 438 and 464 (exon 8) and glutamine 58 (exon 4). Genotyping of ASCC (n = 94) for the MDM2 SNP309 (rs:2279744) revealed: T/T, 43% (40/94); T/G, 48% (45/94); G/G, 9% (9/94), a distribution very similar to that observed in control populations (T/T, 43%; T/G, 43%; G/G, 13%, χ² = 0.8; p = 1.0; 25). There was no evidence for microsatellite instability in ASCC (data not shown).

Table I. Summary of Somatic TP53 and PIK3CA Mutations Observed in Anal Squamous Cell Carcinoma

Gene	Tumour id	Mutation type	Amino acid change
TP53	D5	Deletion + insertion	170delTEVVRRCPHER (or170del11AA), 170insS
	D44	Deletion inframe	177delPHHERS
	D25	Substitution	K132N
	C23	Substitution (splice site mutation)	ntG876A-1bp3′of Exon5
	C66	Substitution	P151A
PIK3CA	D39	Substitution	H1047R
	D66	Substitution	H1047R
	C69	Substitution	H1047L
	C71	Substitution	H1047R
	C6	Substitution	E545K

Immunohistochemistry revealed nuclear TP53 or MDM2 accumulation in a majority of tumours (91%, 100/110; and 72%, 79/110, respectively; Figs. 2a–2d). All five samples with TP53 mutations demonstrated TP53 nuclear staining. Normal anal epithelial controls (n = 30) showed only basal staining of TP53 and parabasal staining of MDM2. A positive association between the presence of nuclear staining of TP53 and nuclear accumulation of MDM2 was demonstrated in this set of tumours (p < 0.001, chi goodness of fit test; Table II). There was no association between HPV infection and TP53 nuclear accumulation (p = 0.5, chi goodness of fit test; Table III) and marginal significance for detection of HPV and MDM2 nuclear staining (p = 0.04, chi goodness of fit test, Table III). There was no beta-catenin nuclear staining observed in ASCC (Figs. 2e–2f) suggesting that perturbation of Wnt signalling was not a factor in ASCC tumorigenesis.

Figure 2. Representative images showing TP53, MDM2 and beta-catenin immunohistochemistry. (a) nuclear protein expression of TP53 in ASCC (×200); (b) strong nuclear TP53 antibody reaction in ASCC (×100; arrow) compared to the adjacent squamous mucosa (very weak/negative stain; arrowhead); (c) distinct nuclear accumulation of MDM2 in ASCC (×200); (d) MDM2 positive, leiomyosarcoma, positive control (×200); (e) cytoplasmic staining of beta-catenin in ASCC (×200); (f) beta-catenin nuclear positive, colorectal adenocarcinoma as positive control (×100).

Table II. Nuclear Accumulation of MDM2 with TP53, MDM2 with AKT and AKT with TP53 in Anal Squamous Cell Carcinoma

	MDM2+	MDM2−	p
Numbers of tumours showing nuclear accumulation are shown; p-value was calculated using chi goodness of fit test.
TP53+	79	21
TP53−	2	8	<0.001
AKT+	61	21
AKT−	21	22	<0.001
	AKT+	AKT−
TP53+	71	31
TP53−	5	4	0.5

Table III. HPV Infection Status of Anal Squamous Cell Carcinoma and Nuclear Accumulation of MDM2, TP53 and Presence of AKT Protein in Tumour Cells

	HPV+	HPV−	p
Number of tumour samples are given. A total of 128 samples were available for analysis: 18 failed for TP53; 10 for MDM2 and 11 for AKT. Statistical p values are calculated using Fisher's exact test.
TP53+	79	29
TP53−	1	1	0.5
MDM2+	65	16
MDM2−	23	14	0.04
AKT+	64	14
AKT−	23	16	0.01

Copy number of PIK3CA was increased and phosphorylated-AKT staining was positive in a significant proportion of ASCC

We next analysed ASCC samples for mutation of the PIK3CA gene, including copy number changes. Five coding sequence mutations were found (4%, 5/127), all substitutions: one in exon 9, E545K; and 4 in exon 20: H1047R (n = 3) and H1047L (n = 1). Genomic copy number of PIK3CA was increased (>2n) in a substantial proportion (47%, 33/70) of ASCC (Table IV; Fig. 3). MDM2 amplification was not observed in any ASCC sample (data not shown).

Table IV. PIK3CA Copy Number Changes in Anal Squamous Cell Carcinoma

PIK3CA copy number	Number of samples (%)
A total of 70 samples were successfully analysed using quantitative PCR. Numbers of tumours are given for each category of copy number with percentages in parenthesis.
≤2	37 (53%)
>2 to ≤4	30 (43%)
>4 to ≤8	2 (3%)
>8	1 (1%)

Figure 3. Copy number of PIK3CA determined by quantitative real time PCR. Panel (a) shows an ASCC with 9-fold increase in PIK3CA copy number (line shifted to the left) compared to normal control (line to right of ASCC) and Panel (b) shows example of no amplification of ASCC with overlapping control.

Immunohistochemistry using a phosphorylated-AKT antibody provided a positive result in 66% of ASCC samples (82/125): 21% showed predominant nuclear stain (17/82), 6% had cytoplasmic staining (5/82) and a composite nuclear and cytoplasmic staining was found in 73% (60/82) (Fig. 4). There was a positive association between phosphorylated-AKT cellular accumulation and MDM2 nuclear staining (p > 0.0001, Fisher's exact test; Table II), between HPV infection and phosphorylated-AKT staining (p = 0.01, Fisher's exact test; Table III) and amplification of PIK3CA was associated with activated AKT in ASCC (p = 0.008, Fisher's exact test; Table V).

Figure 4. Immunohistochemistry showing activated phosphorylated-AKT (serine 473). Panel (a) Anti phosphorylated-AKT in ASCC; (b) Anti phosphorylated-AKT in breast carcinoma, positive control; Anti phosphorylated AKT; (c) negative control omitting primary antibody in ASCC.

Table V. Amplification of PIK3CA was Associated with Presence of Phosphorylated-AKT in Anal Squamous Cell Carcinoma

	PIK3CA (>2 copies)	PIK3CA (≤2 copies)	p
Almost all the samples used for copy number analysis were successfully assessed using immunohistochemistry (93%, 65/70). Statistical value calculated using Fisher's exact test.
AKT+	29	21
AKT−	3	12	0.008

Discussion

At present, the molecular genetics and the tumorigenic pathways leading to ASCC are poorly characterised. To address this, we have analysed the largest cohort of anal cancer specimens to date. In accordance with previous reports, we have confirmed that a large proportion of ASCC samples (73%) are HPV-positive, reinforcing the view that HPV has a causal role in ASCC. HPV 16 was detected in 86% of positive cases, which is higher than for cervical cancer (50%). HPV oncoproteins abrogate TP53 function in vitro and it is thought that TP53 mutation may not be necessary for tumorigenesis in the context of viral infection. Prior to the present study, only a relatively small number of ASCC tumours had been assessed for TP53 mutation.5 We found that only 4% of our much larger sample set showed TP53 mutation in the mutation cluster region (exons 5–8), and to find a significant numbers of mutations outside the cluster region would appear unlikely. We found a higher frequency of nuclear TP53 accumulation (91%) than reported previously (37–71%)6, 8 and we predict the accumulated protein in most cases would be wild type.

The molecular mechanism behind our finding that wild type TP53 nuclear staining occurs in a significant proportion of ASCC remains to be elucidated. This may be due to the nuclear accumulation of MDM2, as a significant positive correlation was demonstrated between nuclear accumulation of TP53 and MDM2 in our panel of ASCCs. MDM2 protein is wild type, there is no increase in normal copy number and the distribution of the genotype of SNP309 was as expected; genetic aberration does not appear to account for the MDM2 overexpression observed in this study. The serine/threonine kinase AKT phosphorylates MDM2 at serine 166 and 186 and causes MDM2 shift to the nucleus where it is considered to play a role in TP53 degradation.26, 27, 28 This represents a paradox as TP53 nuclear staining is observed suggesting that this particular role for MDM2 is disrupted. It is possible that TP53 remains sequestered to MDM2 or viral proteins, although in the latter case we did not see a significant association between TP53 nuclear accumulation and viral infection (Table III).

Accumulation of AKT was observed in 66% (82/125) of ASCC and a positive correlation was demonstrated between nuclear MDM2 and phosphorylated-AKT accumulation (Table II). This finding indicates that the two proteins are present at the same time in the nucleus in this set of samples and it is likely that the presence of phosphorylated-AKT is responsible for MDM2 nuclear accumulation in ASCC. The copy number changes to PIK3CA in a large proportion of samples may be critical in this respect. PI3K effectively recruits AKT to the plasma membrane by catalyzing the synthesis of the phospholipids PtdIns-3,4,5-P3 which then directly interact with the AKT pleckstrin homology domain. Full activation of AKT requires phosphorylation of two conserved residues, threonine 308 and serine 473. PIK3CA is amplified and/or mutated in tumours of multiple origins, which implies a significant role for this gene in tumorigenesis. We observed a high frequency of PIK3CA copy number changes, as well as a few coding sequence mutations in PIK3CA. Furthermore, amplification of PIK3CA was associated with presence of phosphorylated-AKT in ASCC. Overall, our data indicates that the PI3K-AKT pathway may be important in ASCC.

AKT was activated in cervical precancers caused by HPV and strong phosphorylated AKT immunostaining was observed in 10 high-grade squamous cervical intraepithelial lesions.29 In contrast to our observations on invasive ASCC where staining was predominately nuclear, phosphorylated-AKT staining in cervical precancers was cytoplasmic and membranous. However, both nuclear and cytoplasmic staining has been observed previously in carcinoma of the cervix and nuclear staining has also been observed in lung and prostate cancer.30, 31 Overall, frequency of nuclear staining was higher in ASCC, which may reflect differences in mutational pathways to cancers at this site and is of interest for future research.

In summary, we have shown that activation of AKT is likely to occur through PIK3CA amplification and that this activation is important for ASCC tumorigenesis. We provide a possible pathway for ASCC development by confirming that HPV infection does indeed occur at high frequency. The presence of virus in our samples did not correlate with TP53 nuclear accumulation but did with both MDM2 and phosphorylated-AKT, although the statistical significance for MDM2 was only marginal (Table III). The HPV oncoprotein E7 is thought to activate AKT through inhibition of pRB leading to MDM2 accumulation.29, 32 However, this is only one possible mechanism of AKT activation in ASCC; the alternate mechanism via PIK3CA amplification is also well documented in multiple tumours including cervical cancer, and now shown here. Functional work will be required to further assess the role of AKT, MDM2 and TP53 in ASCC and determine the most relevant target for therapy.