Multiple Genetic Alterations within the PI3K Pathway Are Responsible for AKT Activation in Patients with Ovarian Carcinoma

The phosphatidylinositol 3-kinase (PI3K)/AKT pathway is activated in multiple cancers including ovarian carcinoma (OC). However, the relative contribution of the single components within the PI3K pathway to AKT activation in OC is still unclear. We examined 98 tumor samples from Italian OC patients for alterations in the members of the PI3K pathway. We report that AKT is significantly hyperactive in OC compared to normal tissue (n = 93; p<0.0001) and that AKT activation is preferentially observed in the elderly (>58 years old; n = 93; p<0.05). The most frequent alteration is the overexpression of the p110α catalytic subunit of PI3K (63/93, ∼68%); less frequent alterations comprise the loss of PTEN (24/89, 27%) and the overexpression of AKT1 (18/96, 19%) or AKT2 (11/88,12.5%). Mutations in the PIK3CA or KRAS genes were detected at lower frequency (12% and 10%, respectively) whereas mutations in AKT1 or AKT2 genes were absent. Although many tumors presented a single lesion (28/93, of which 23 overexpressed PIK3CA, 1 overexpressed AKT and 4 had lost PTEN), many OC (35/93) presented multiple alterations within the PI3K pathway. Apparently, aberrant PI3K signalling was mediated by activation of the canonical downstream AKT-dependent mTOR/S6K1/4EBP1 pathway and by regulation of expression of oncogenic transcription factors that include HMGA1, JUN-B, FOS and MYC but not by AKT-independent activation of SGK3. FISH analysis indicated that gene amplification of PIK3CA, AKT1 and AKT2 (but not of PI3KR1) and the loss of PTEN are common and may account for changes in the expression of the corresponding proteins. In conclusion, our results indicate that p110α overexpression represents the most frequent alteration within the PI3K/AKT pathway in OC. However, p110α overexpression may not be sufficient to activate AKT signalling and drive ovarian tumorigenesis since many tumors overexpressing PI3K presented at least one additional alteration.


Introduction
Epithelial ovarian carcinoma (OC) is the most lethal gynecological malignancy with most patients diagnosed only at advanced stages [1,2]. OC is a complex disease that exhibits remarkable heterogeneity at the clinical, cellular and molecular level. [3][4][5][6] OC is classified by histotype and grade. Histologically, OC is classified into serous (S-OC), mucinous (M-OC), endometrioid (E-OC), clear cell (CC-OC), transitional (or Brenner cancer), squamous cell, and undifferentiated types. S-OC are the most common type of OC, accounting for about two thirds of ovarian carcinomas [7]. As to the grade, Type I lesions constitute 10-20% of OC and include low-grade S-OC, M-OC E-OC, and CC-OC. Low-grade type I cancers present in early stage (I-II), show lowmalignant potential, grow slowly, and are relatively resistant to platinum-based chemotherapy. Conversely, Type II lesions include high-grade S-OC, and undifferentiated cancers that present at late stage (III-IV), grow more aggressively, though they respond more frequently to platinum-based treatment. The cure rate of patients affected by OC remains low (approximately 30%) although the outcome of patients in the late stage has recently improved, with 5-year survival rates approaching 50% [8][9][10].
OC represent independent diseases, being characterized by genetic changes that are remarkably different in Type I and Type II OC. Type I low-grade OC are almost euploid, retain wild-type p53, and are apparently driven by activating mutations of RAS and PIK3CA, and inactivating mutations of PTEN. Conversely, type II high-grade OC almost invariably present genomic instability caused by mutation and/or silencing of BRCA1 or BRCA2 and p53 mutation [11]. Such genomic instability results in diverse subsequent events that include alterations within the PI3K/AKT pathway, which are believed to drive tumor growth and metastatic progression, [3,[12][13][14].
The PI3K/AKT pathway is activated in multiple cancers leading to oncogenic transformation [15,16]. The mechanisms responsible for activation of the PI3K/AKT pathway in human cancers are diverse and include dysregulation of growth factor receptor and integrin signalling, activating RAS mutations, activating mutations or gene amplification of the gene encoding the p110a catalytic subunit of PI3K (PI3KCA), inactivating mutations in the phosphatase and tensin homolog (PTEN) tumor suppressor gene or in the gene encoding the p85 regulatory subunit of PI3K (PI3KRA).
Recently, a multiplatform genomic analysis by The Cancer Genome Atlas (TCGA) Research Network identified alterations in the PI3K/AKT and RAS pathways in approximately 45% of high-grade S-OC [11]. Here, we performed an integrated analysis of OC in an Italian cohort of patients in order to characterize the molecular mechanisms that lead to the activation of the PI3K/ AKT pathway in OC.

Tissue Microarray (TMA) and Immunohistochemistry
TMAs (257.1 and 257.2) were constructed in collaboration with the Unit of Immunostaining at the Centro Nacional de Investigaciones Oncologicas (Madrid, Spain) according to established methods [27] using a Tissue Arrayer (Beecher Instruments, Gene Micro-Array Technologies, Silver Spring, MD). Two cores of ovarian carcinoma were arrayed from each case. TMA slides were deparaffinized, heated in a pressure cooker with 1 mM EDTA, pH 8.0 for 10 min, and incubated with pepsin at 37uC for 30 min. Slides were then dehydrated in increasing ethanol concentrations, and then air-dried. Probes were denatured at 96uC for 5 min, and hybridization solution was applied on each slide and incubated at 75uC for 1 min. After overnight incubation at 37uC in a humid chamber, slides were washed with 0.4 X SSC and 0.3% NP40 for 2 min at 75uC, air-dried in darkness, counterstained with DAPI, and a coverslip was applied.
The immunohistochemical score of pAKT and pSGK3 used in this work was selected on the basis of widely established criteria existing in the literature [31] by multiplying the percentage of labelled cells (ranging from 0% to 100%) by the intensity of the staining (1-weak, 2-moderate and 3-strong). Scores above 150 were considered positive (+).

Fluorescence In Situ Hybridization (FISH)
FISH analysis was performed on TMAs. BAC clones were designed according to the Ensembl database (www.ensembl.org). BAC clones covering the AKT1 gene were RP11-982M15, RP11-477I4 and RP11-556J09. Control BAC probes covering chromosome region 14q11 was RP11-324B11. BAC clones covering the AKT2 gene were RP11-36B02, RP11-688J23, RP11-725P04. Control BAC probes covering chromosome region 19p13.1 were RP11-737I1, RP11-520G3. BAC clones covering the PIK3CA gene were RP11-360P21 and RP11-245C23. Control BAC probes covering chromosome region 3p14.1 were RP11-175F9 and RP11-15B21. All BAC clones were labelled with dUTP-Sprectrum Orange (Vysis Inc., DownersGrove, IL; USA). All Control probes were labelled with dUTP-Sprectrum Green (Vysis Inc., Down-ersGrove, IL; USA). Two different investigators (R.F., S.L.) that had no previous knowledge of the genetic, clinical and IHC results evaluated FISH analysis. All FISH were scored in an average of 130 (60-210) nuclei. For evaluation of copy number of the genes encoding AKT1, AKT2 and PIK3CA, a gene-to-control ratio of 1.0 was classified as disomy; ratios between 1.0 and 2.0 were considered gene low-level gains; ratios .2.0 were considered as high polysomy and/or gene amplification [41,42]. Accordingly, tumours were divided into different classes: disomy, trisomy (3 copies of chromosomes in .40% of cells), low polysomy ($3 copies of chromosomes in .40% of cells), high polysomy ($4 copies of chromosomes in $40% of cells), and gene amplification (presence of gene clusters with a ratio of gene-to-chromosome of $2 per cell in $40% of cells or presence of small or non-enumerable clusters of the gene signal). On this basis patients were classified into two groups: FISH-negative (disomy and gains) and FISH-positive (high polysomy and/or gene amplification).

PCR, RT-PCR and Mutation Analysis
Total RNA and genomic DNA were prepared as described [43,44]. Q-RT-PCR and Q-PCR were performed using the Power SYBR Green PCR Master Mix in an ABI Prism 7300 thermocycler (Applied Biosystems, Foster City, CA, USA). cDNAs were synthesized from 1 mg of total RNA using QuantiTect Reverse Trascription (Qiagen, The Netherlands, Venlo). Normalization was performed to GAPDH mRNA content. The relative amounts of mRNA or DNA were calculated by the comparative cycle threshold (CT) method by Livak and Schmittgen [45]. Mutation analysis for PIK3CA using LightCycler was performed with DNA Master/Hybridization probes kit (Roche Molecular Biochemicals, Mannheim, Germany). Direct sequencing was performed using the BigDye v3.03 cycle sequencing kit (Applied Biosystems) in a capillary automatic sequencer (ABI PRISM 3100 Genetic Analyzer; Applied Biosystems). Protocols and primers for Q-PCR, Q-RT-PCR and sequencing KRAS (exons 2 and 3) and PIK3CA (exons 9 and 20) are reported in Supporting Information.

Statistics
The association between the phosphorylation status of Akt and the expressed proteins or the clinicopathologic variables was

AKT Activation in OC
As a read-out of PI3K/AKT signalling in OC from Italian patients we determined the phosphorylation status of aminoacid S473 of AKT1 (pAKT). pAKT was evaluated on TMAs containing duplicated core biopsies of 98 OC. As controls 50 matched normal samples (of which 3 tubes) were used. Patients' clinico-pathological characteristics are described in Materials and Methods and summarized in Table S1.
Immunostaining analysis of TMAs failed to find AKT activation in 50 control tissues ( Figure S1A). In contrast, AKT activation was observed in 73 out of 93 OC analysed and was significantly higher in cancer samples than in normal controls (Table 1; p,0.0001). pAKT staining was observed in 51/66 Serous Ovarian Carcinomas (S-OC) and in 14/16 Endometrioid Ovarian carcinomas (E-OC) ( Table 1). See Figure 1 for representative stainings.
We then correlated pAKT staining with the clinico-pathological parameters of patients (n = 98, median age 58 years old, range 21-86 years) and found a significant association between pAKT staining and the age of patients: Akt activation was more represented in patients $58 years old (p,0.05). On the contrary, we observed no significant association between pAKT staining and grade or FIGO stage of the disease (Table 2). See also Tables S2 and S3 for analysis of the correlation between pAKT and clinicopathological parameters of patients with S-OC and E-OC, respectively, and Table S4 for a patient-by-patient list of pAKT status.

Mechanisms of AKT Activation in OC: AKT1 and AKT2
To investigate the molecular mechanisms leading to AKT activation in Italian patients affected by OC we performed a comprehensive analysis of the expression and/or the genetic status of AKT1 and AKT2 and their closest regulators (KRAS, PI3K and PTEN). Of the 98 cases on TMAs 257.1 and 257.2, 96 could be properly analysed for AKT1, 88 for AKT2, 89 for PTEN, 93 for p110a (PIK3CA) and 94 for p85a (PIK3R1) ( Table S5). The evaluation criteria for the staining of each protein are reported in Materials and Methods. Accordingly, samples were classified as follows: negative (2), moderate (+) or high (++). In the case of PTEN staining samples were classified as positive (+), reduced (+/ 2) or negative (2).
We then analysed expression of AKT2 in OC and found that it was moderately expressed in 41/88 specimens and overtly overexpressed in 11/88 specimens (12.5%) (Table S5 and Figure 3). See also Figure S2B for representative staining of AKT2 expression in S-OC. In the case of AKT2 all cases overexpressing AKT2 showed pAKT staining ( Table 3). As with AKT1, AKT2 overexpression was similarly distributed among S-OC and E-OC [7 out of 61 S-OC (,12%); 3 out of 16 E-OC (,19%)] (Table S6).
As to the relationship between the expression of AKT1 and AKT2, we found that ,14% OC showed an aberrant expression of only AKT1, ,6% OC overexpressed AKT2 and ,7% OC showed increased staining of both proteins (6/88 samples). On the other hand, 22 samples showed no sign of AKT1 or AKT2 overexpression. Notably, 32 out of 33 samples with aberrant staining for at least one of the proteins were positive for pAKT, while 12/22 (55%) negative specimens showed AKT activation.

Mechanisms of AKT Activation in OC: FISH Analysis
To determine the molecular mechanisms underling overexpression of AKT1, AKT2, PIK3CA and PIK3R1 we performed FISH and/or Q-PCR analysis in OC. See Materials and Methods for classification of tumours by FISH. We found that 16/55 OC (29%) presented copy number gain of the AKT1 gene at chromosome 14, of which 9 were high polysomy (.4 copies) and 7 focal amplification ( Figure 5A). See Table S4 for a detailed list of genetic alterations detected in single OC patients. In the case of the gene encoding AKT2, we observed 10/36 OC (,28%) with copy number gain at chromosome 19, of which 5 had high polysomy and 5 had focal amplification. See Figure 5B for a representative example. FISH analysis with chromosome 3q26.32 probes revealed the presence of an increase in the PIK3CA gene copy number in 25/84 cases (,30%), all of which characterised by high polysomy, with 16 cases showing also focal amplification ( Figure 5C). Of 25 cases showing copy number gains, 19 were S-OC, 3 was E-OC, 1 was CC-OC and 2 were M-OC.
On the other hand, the analysis of copy number variation indicated that one important mechanism that dysregulates AKT signalling in OC was the acquisition of an increased gene copy number -either amplification or high degree of polysomy. In fact, among the 63 tumors that overepressed PIK3CA, 11 presented amplification and 6 high polisomy (27%); among the 18 tumors that overepressed AKT1, 1 presented amplification and 4 high polisomy (,28%) and among the 11 tumors that overepressed AKT2, 1 presented amplification and 1 high polisomy (18%).
For each gene under analysis most samples presenting high gene copy number showed high or moderate expression of the corresponding protein (8/16, 50%; 8/10, 80%; 20/25, 80%; for AKT1, AKT2 and PIK3CA, respectively). Moreover, most of FISH-positive OC resulted in the activation of AKT signalling. As summarized in Table 4, 20/24 cases that were FISH-positive for PIK3CA were positive for pAKT staining, 13/16 cases that were FISH-positive for AKT1 were positive for pAKT staining, 8/10 cases that were FISH-positive for AKT2 were positive for pAKT staining.
Given the high frequency of PIK3R1 overexpression in OC, we investigated the mechanisms underlying the observed overexpression of PIK3R1 protein. We used Q-PCR for analysis of PIK3R1 gene copy number and Q-RT-PCR for analysis of PIK3R1 mRNA in 30 representative primary OC (16 S-OC, 7 E-OC, and 7 others), that have been previously analyzed by immunostaining. The analysis of the genetic status of PIK3R1 by Q-PCR ( Figure 6A) failed to reveal an increase in the copy number of the PIK3R1 gene in the 30 tumor analysed. Conversely, when analysed by Q-RT-PCR we found that 24 out of 30 OC (,80%) showed increased mRNA levels compared with normal tissues (n = 10). Thirteen out of 16 were S-OC (81%), while five out of 7 were E-OC (71.4%). See Figure 6B. These results indicated that the increase in the levels of p85a observed in OC may be ascribed to increased transcription rate from the PIK3R1 gene promoter or to stabilization of the p85a protein by p110a but not to gene amplification.

Mechanisms of AKT Activation in OC: Mutation Analysis
We also investigated the occurrence of mutations in the PI3KCA or KRAS genes in OC. Because the vast majority of PIK3CA gene mutations in cancer were reported in exons 9 and 20, we focused our mutation analysis on these exons [46]. Mutation detection was performed on the LightCycler (Roche) and confirmed by direct DNA sequencing (see Figure 7A). We observed the missense mutation (GAG1633-.ACG that lead to the amino acid change E545A) in the PIK3CA gene in 5 cases out of 33 analysed (12%) of which 3 S-OC and 2 were E-OC. In agreement with previous findings PI3KCA mutations were observed in E-OC [23]. In the case of PIK3CA, mutations and gene amplification were apparently mutually exclusive, though, due to the limitedness of the sample under analysis, this correlation did not reach statistical significance (p = 0.05). Interestingly, 4 tumors with mutations in PIK3CA were positive for pAKT, except for one S-OC specimen. See Figure 7B.
Moreover, a total of 31 tumours were successfully analyzed for the presence of mutations in the gene encoding KRAS. Overall, mutations were found in 3 samples (,10%). Three different mutations were detected in the OC analysed: G12V (GGT-.GTT); G12R (GGT-.CGT); G13D (GGC-.GAC). The G12V mutation was found in one S-OC patient, while the G12R and G13V mutations were detected in Mu-OC. The specific nucleotide changes and the corresponding amino acid substitutions are shown in Figure 7C. The overall KRAS mutation frequency in this cohort of OC (,10%) was in agreement with that described in previous works (10-15%) [47,48]. KRAS mutations were observed prevalently in the Mu-OC subgroup [49,50] and in a low grade S-OC (S53) [51]. KRAS mutation was reflected into AKT activation in 2 tumors out of three ( Figure 7D). See Table S4 for details.

Mechanisms of AKT Activation in OC: PTEN
Patients accrued for this study were also characterised for PTEN expression. Complete loss of PTEN protein was observed in 24 of 89 (,27%) OC; reduction of PTEN expression was observed in 7  Tables S4 and S6) [52]. However, AKT activation was significantly associated with the loss and/or reduction of PTEN expression only in S-OC (n = 63; p = 0.019) (Table S7).
We used quantitative RT-PCR analysis of normal ovarian tissue samples (n = 10) and 31 representative primary OC (16 S-OC, 6 E-OC, and 9 others), that have been analyzed by immunostaining, to investigate whether the loss of PTEN protein observed in the TMA studies occurred through mechanisms that dysregulate its mRNA transcription ( Figure 8C). By matching the quantitative RT-PCR analysis of PTEN mRNA with the protein analysis performed on TMAs, we observed poor correspondence between loss of PTEN mRNA and loss of the corresponding protein. Using a cut-off of 0.7 (arbitrary unit) as lower limit for normal PTEN mRNA level, we found that 23 OC showed reduced mRNA levels; however, only 7 out of them also showed consistently loss of PTEN protein. On the other hand, 2 out of 8 samples, that showed mRNA expression comparable with normal tissue, had lost protein expression suggesting also the existence of a post-translational mechanism.
We also investigated the genetic status of PTEN by Q-PCR on the same group of samples analysed by Q-RT-PCR ( Figure 8D). The average value of PTEN gene in normal tissues was similar to the PBL's value that had been set arbitrarily as 2. We used a cutoff of 1.3 (arbitrary unit) as lower limit for PTEN biallelic status and 0.3 as lower limit for monoallelic status. We found that 3 S-OC, 2 Mu-OC and one E-OC and M-OC showed monoalellic loss of PTEN gene whereas one E-OC showed biallelic loss of PTEN gene, for a total of 8 samples out of 31 analysed (26%). Importantly, most of the samples that showed monoallelic or biallelic deletion of PTEN gene (6/8) presented drastically reduced or absent mRNA and PTEN protein. Moreover 7/8 samples with monoallelic or biallelic deletion of PTEN gene showed AKT activation. Altogether these results indicate that multiple mechanisms are responsible for PTEN loss during OC carcinogenesis.
It is of note that from the analysis of TMAs we found that aberrant expression of more than a single gene within the PI3K pathway (PTEN loss, overexpression of AKT1, AKT2 or p110a, respectively) was observed only in tumours showing AKT activation. In particular, PTEN loss was mutually exclusive with increased AKT2 expression but not with increased expression of AKT1 or PIK3CA and that .50% of tumors overexpressing PIK3CA presented a second molecular alteration (see Table 5 and Tables S8-S10).

Analysis of Pathways Activated Downstream PI3K in OC: mTOR and SGK3
AKT is the most important effector of PI3K signalling, and is a key regulator of a variety of proteins involved in cell proliferation, metabolism, invasion, migration, and apoptosis that include mTOR (mammalian target of rapamycin), GSK3, and forkhead transcription factors [53]. In particular, mTOR is a critical component of the PI3K/AKT pathway that activates protein synthesis and cell proliferation [54].
Therefore we investigated whether the activation of the PI3K/ AKT pathway observed in OC in this study was associated with activation of mTOR or of two well characterized downstream targets such 4EBP1 and p70 S6 kinase (S6K1). To this aim, we studied the activation status of mTOR, 4EBP1, S6K1 and its substrate S6 in immunostaining on TMAs, by use of phosphospecific antibodies (anti-phospho-mTOR, Ser2448; anti-phospho-S6K1, Thr389; anti-phospho-4EBP1, Thr37/46; anti-phospho-S6, Ser235/236), and correlated them with AKT activation and/ or PIK3CA overexpression.
We found that the protein encoding mTOR was expressed in all normal and cancer-derived samples (data not shown) whereas mTOR phosphorylation was detected at high level in 46/92 OC (50%), of which 33 were S-OC e 9 were E-OC. Similarly, S6K1 protein was expressed in all normal and cancer-derived samples (data not shown) whereas S6K1 phosphorylation was detected at high level in 65/91 OC (71.5%), of which 47 were S-OC e 11 were E-OC.
Finally, 4EBP1 was phosphorylated in 52 out of 92 OC (56%) whereas S6 was significantly phosphorylated in 36 out of 91 OC (40%). As with mTOR and S6K1, both S6 and 4EBP1 proteins were constitutively expressed in tumors (data not shown).
To determine whether the mTOR/S6K1/4EBP1 pathway contributed to aberrant PI3K signalling observed in OC we correlated the phosphorylation of mTOR, S6K1, 4EBP1, and S6 with PIK3CA overexpression and/or AKT activation. We found that several OC that were positive for pmTOR or pS6K1 were also positive for either pAKT (34/44 and 48/62, respectively) or p110a (33/44 and 44/62, respectively) staining, though no significant correlation was observed among these proteins (Table 6). Conversely, phosphorylation of 4EBP1 was correlated with p110a overexpression, AKT activation (pAKT) and mTOR activation (pmTOR) ( Table 6) and, similarly, phosphorylation of S6 presented a significant correlation with AKT activation  (pAKT) and mTOR activation (pmTOR) ( Table 6). Immunostaining data were confirmed by Western blot analysis of a representative group of freshly-frozen OC (Figure 9) and on a selected subset of OC cell lines ( Figure S4). These results suggest that the mTOR/S6K1/4EBP1 pathway contribute to aberrant PI3K signalling and AKT activation in OC.
On the other hand, recent studies uncovered an AKTindependent signalling pathway in PIK3CA mutant cancer cell with low AKT activation that involved PDK1-dependent activation of SGK3 [55]. Therefore, to understand whether SKG3 could mediate PI3K signalling in tumors that presented low AKT activation in the setting of high PIK3CA expression, we analyzed SGK3 activation using phosphorylation specific antibody (Thr320). We found that that 33/92 OC (36%) were positive for phospho-SGK3 staining though phosphorylation of SGK3 did not correlate with p110a overexpression or AKT activation (see Table 6 and Figure 9C). On the other hand, activated pSGK3 was detected only 3 out of 9 OC that were positive for p110a and negative for pAKT, showing a frequency of SGK3 activation similar to OC positive for p110a and positive for AKT. Altogether, these results suggest that, at least in the OC analysed here, aberrant PI3K activity is mediated by the canonical AKTdependent mTOR/S6K1/4EBP1 pathway but not by the novel AKT-independent activation of SGK3.

Analysis of Pathways Activated Downstream PI3K in OC: Transcription Factors
We have recently demonstrated that aberrant PI3K signalling in Non-Small Cell Lung Cancer cells induces the mRNA expression of oncogenic transcription factors (HMGA1, JUN-B, FOS and MYC). Thus, to further extend the analysis of the pathways acting downstream PI3K and/or AKT, we investigated whether overexpression of PIK3CA affected the expression of these transcription factors also in OC. To this aim, we performed Q-RT-PCR analysis of HMGA1, JUN-B, FOS and MYC in a representative number of samples that were classified into two groups: low expressors/non-mutated p110a (n = 7) or high expressors/mutated p110a (n = 9). As shown in Figure 10A, we found that the group of PIK3CA overexpressors showed an We further confirmed this observation by RT-PCR analysis of selected OC cell lines (OVCA429, TOV112D) that had been treated with PI3K or mTOR inhibitors (LY294002 and RAD001, respectively). Of note that both inhibitors reduced the levels of all 4 transcription factors in OVCA429 but not in TOV112D ( Figure 10B), suggesting that mTOR was implicated in the PI3K-dependent control of the expression of transcription factors.

Discussion
We report a comprehensive analysis of the contribution of the different members of PI3K/AKT pathway to AKT deregulation and to the development of ovarian cancer in a cohort of Italian OC patients. We report that in the cohort of OC patients studied here, the PI3K/AKT pathway is activated in ,79% of the cases (73/93) and was detected more frequently in the elderly (.58 years old; n = 93, p,0.05). This value is in agreement with the average degree of PI3K/AKT activation as reviewed in Bast [12] but apparently differs from data reported in previous studies on cohorts of Norwegian (51.6%) [56] or Middle Eastern OC patients (52.1%) [42], thus highlighting possible ethnical and/or geographical differences.
Immunoistochemical studies have suggested that AKT activation is common in high-grade, late stage S-OC [20,18] and is associated with resistance to cytotoxic therapies [17]. However, at difference with some of these previous studies, in Italian OC patient AKT S473 phosphorylation was not associated to advanced disease or higher grade. The difference in AKT activation between FIGO stages I-II and FIGO stages III-IV was not statistically significant, possibly due to the relative paucity of FIGO stages I-II in our analysis. Moreover, no gross difference in AKT activation was observed with regard to histological   subtypes (77% in S-OC; 87% in E-OC) and grade (71% in lowgrade; ,78% in high-grade). Altogether these results suggest that deregulation of the PI3K/AKT signalling occurs early during the tumorigenic process in the ovary and represents a common event for the different molecular routes through which OCs develop.
A second finding of this study was that several OCs simultaneously over-express the catalytic (PIK3CA) and the regulatory (PIK3R1) subunits of PI3K (,57%) or present activating mutations in the gene encoding PIK3CA (,12%), implying that the deregulated expression and/or activity of this enzyme represents a crucial oncogenic event during cancer development in the ovary. Accordingly, aberrant activation of p110a in OCs -either due to overexpression or mutations -was reflected into AKT activation and, as a whole, represented the major determinant of AKT activation in this cohort of patients. These results extend to the Italian cohort of patients the observation that PI3K is an important oncogene for the development and progression of OC [21,42,57]. Less frequent alterations comprised PTEN loss (,27%), overexpression of AKT1 (,19%) or of AKT2 (,14%) and mutations in the KRAS genes (10%). Conversely, OCs analysed in this study were negative for mutations in AKT1 or AKT2, confirming previous reports [58]. As to PTEN inactivation, somatic inactivating mutations of PTEN are rather rare in OCs, having been described preferentially in CC-OCs and in low-grade E-OCs [59,60]. Accordingly, we did not detect PTEN mutations, possibly because of the low number of CC-OCs analysed (data not shown). Conversely, we observed a large reduction in PTEN protein expression in approximately 27% of cases including high-grade S-OCs, E-OCs and Mu-OCs. A third issue to discuss of our results is that the type and/or the position of the alterations identified within the PI3K/ AKT pathway apparently have different strength and effects [61][62][63]. In fact, whereas mutations of PIK3CA are mutually exclusive with other genetic alterations, the majority of OCs under analysis here (n = 35) presented alterations in more than one gene, suggesting that the different genetic alterations of the PI3K/ AKT pathway in OCs are not functionally redundant. In fact, activating mutations in PIK3CA are mutually exclusive with amplification or other forms of copy number gains (except for patient S56), suggesting that they are sufficient not only to activate AKT but also to drive ovarian tumorigenesis independently, as demonstrated in other cohorts [42]. On the other hand, more than half OC that overexpressed PIK3CA presented two or more alterations (14 OCs simultaneously overexpressed PIK3CA and had lost PTEN, 6 OCs overexpressed both PIK3CA and AKT1, and 5 OCs overexpressed both PIK3CA and AKT2). Also KRAS mutations were detected in a contest of PTEN loss (Mu6, Mu7). At difference with the OCs that presented a single alteration, all OCs with at least 2 alterations showed strong AKT activation. These findings are reminiscent of breast or endometrial cancer, in which PIK3CA mutations are frequently detected in the setting of low PTEN expression or mutations [64,65], and suggest that p110a over-expression alone may not be sufficient to activate AKT Table 5. Altered expression of genes within PI3K pathway (AKT2, AKT2, PTEN, PIK3CA) and AKT activation (pAKT) in OC.    signalling and hence requires other alterations to be fully oncogenic in OC. The functional effects of mutant or amplified PIK3CA in OC cells have defined through adoptive expression of mutant p110a or by silencing endogenous activated PIK3CA alleles [66][67][68]. Collectively, the data available indicate that PI3K signalling is required both in vitro and in vivo. However, our results in the human indicate that both PI3K and PTEN act in concert with other oncogenic hits to promote malignant transformation of ovarian epithelial cells and recent data in the mouse have demonstrated that the overexpression of activated PIK3CA or the loss of PTEN in the mouse ovarian surface epithelium do not lead to tumor formation [69].
As to the mechanisms whereby expression and/or activity of the genes under analysis are dysregulated in OC, we found that, in many OC, the abnormal expression/activity of the genes within the PI3K/AKT pathway could be ascribed to underlying genetic alterations. PIK3CA presents focal amplification at 3q26.3 or high polysomy of the chromosome 3 in ,30% of primary OCs while AKT1 and AKT2 present increased copy number in ,30 and ,28% of OCs. Expectedly, most samples presenting high gene copy number of PIK3CA, AKT1 and AKT2 showed high or moderate expression of the corresponding proteins (80%; 50%; 80% for PIK3CA, AKT1 and AKT2, respectively), with most FISH-positive OC resulting in the activation of AKT signalling.
As to PTEN, the majority of tumors showed loss of PTEN protein and consistently reduced mRNA levels, suggesting that the major mechanism of PTEN inactivation occurs at the transcriptional level that in some cases (about 30%) may be ascribed to monoalellic or biallelic loss at the PTEN locus on chromosome 10. However, the finding of tumors with high mRNA and low protein suggests the existence of a post-translational mechanism of PTEN inactivation in OCs as demonstrated in lung cancer. At least two potential mechanisms can account for the observed discrepancy between PTEN mRNA and protein levels. The first is an increased turnover of the PTEN protein due to overxpression of the ubiquitin ligase NEDD4-1 as recently demonstrated to occur in NSCLC [70]. A second mechanism that may account for reduced levels of PTEN protein in the presence of normal mRNA level is the potential effects of oncogenic miRNAs. Accordingly, Yang and coworkers have reported that miR-214 blocks PTEN translation leading to AKT activation drug resistance [71].
A final observation deriving from our results is that the aberrant PI3K activity observed in OC, is apparently mediated by activation of the downstream AKT-dependent mTOR/S6K1/ 4EBP1 canonical pathway and by regulation of expression of oncogenic transcription factors that include HMGA1, JUN-B, FOS and MYC but not by AKT-independent activation of SGK3.
In conclusion, the results reported in this manuscript indicate that the different genetic alterations of the PI3K/AKT pathway in OC are not functionally redundant and that the type or the position of the alteration within the PI3K/AKT pathway may influence mechanisms and effects of pathway deregulation. In particular, PI3KCA over-expression occurs, through gene copy number gains, at a much higher frequency in ovarian cancer than do activating mutations, apparently representing the major determinant of AKT activation in OC. However, p110a overexpression alone is not apparently sufficient to activate AKT signalling in OCs and hence may require other alterations to be fully oncogenic.