Tumor Spectrum, Tumor Latency and Tumor Incidence of the Pten-Deficient Mice

Background Pten functionally acts as a tumor suppressor gene. Lately, tissue-specific ablation of Pten gene in mice has elucidated the role of Pten in different tumor progression models. However, a temporally controlled Pten loss in all adult tissues to examine susceptibility of various tissues to Pten-deficient tumorigenesis has not been addressed yet. Our goal was to explore the genesis of Pten-deficient malignancies in multiple tissue lineages of the adult mouse. Methods and Findings We utilized an inducible Cre/loxP system to delete Pten exon 5 in the systemic organs of ROSA26 (R26)-CreERT;Ptenfx/fx mice. On reaching 45 weeks 4OHT-induced Pten loss, we found that the R26-CreERT;Ptenfx/fx mice developed a variety of malignancies. Overall tumor mean latency was 17 weeks in the Pten-deficient mice. Interestingly, mutant females developed malignancies more quickly at 10∼11 weeks compared with a tumor latency of 21 weeks for mutant males. Lymphoma incidence (76.9% in females; 40.0% in males) was higher than the other malignancies found in the mutant mice. Mutant males developed prostate (20.0%), intestinal cancer (35.0%) and squamous cell carcinoma (10.0%), whereas the mutant females developed squamous cell carcinoma (15.4%) and endometrial cancer (46.1%) in addition to lymphomas. Furthermore, we tested the pharmacological inhibition of the PTEN downstream effectors using LY294002 on Pten-deficient prostate hyperplasia. Our data revealed that, indeed, the prostate hyperplasia resulting from the induced Pten loss was significantly suppressed by LY294002 (p = 0.007). Conclusions Through monitoring a variety of Pten-deficient tumor formation, our results revealed that the lymphoid lineages and the epithelium of the prostate, endometrium, intestine and epidermis are highly susceptible to tumorigenesis after the Pten gene is excised. Therefore, this R26-CreERT; Ptenfx/fx mouse model may provide an entry point for understanding the role of Pten in the tumorigenesis of different organs and extend the search for potential therapeutic approaches to prevent Pten-deficient malignancies.


INTRODUCTION
Tumors occur in normal tissues that are challenged by a variety of genetic and environmental risk factors. These risk factors may consequently evoke aberrant effects on cytoplasmic signaling network, which result in uncontrolled cellular proliferation, cell survival and heterotypic cross-talk with neighboring cells. Among the various types of complicated cellular signaling circuitry that have been identified, the tumor suppressor PTEN, a phosphatase and tension homolog on chromosome 10, acts as the most important negative regulator of the phosphatidylinositol 3-kinase (PI-3K) signaling pathway [1,2]. PTEN converts PIP3 to phosphatidylinositol (4,5) diphosphate (PIP2) in order to maintain a homeostasis of PIP3, which is generated by PI-3K activation [3]. Inactivation of PTEN results in an accumulation of PIP3, which recruits the serine/threonine kinase PKB/AKT to the cell membrane and this further activate downstream of the PI-3K/ AKT signaling pathway to promote cellular proliferation, antiapoptosis, cell survival and tumorigenesis [1,4].
Mutation of PTEN is often seen in human cancers [5,6,7]. Using a gene targeting strategy to ablate Pten gene function in the mouse causes embryonic lethality between days 6.5 to 9.5 of gestation [8,9,10,11]. The early lethality of Pten-deficient mice limits our understanding regarding how PTEN regulates downstream signaling and is related to tumorigenesis at adult stage. Up to this point, mice heterozygous for Pten have provided some basis for study because they develop a variety of cancers, including breast cancer, endometrial cancer, prostate tumors and lymphoma [8,9,11,12]. Loss-of-heterozygosity of Pten might contribute to the tumor development in Pten +/2 mice [9,11,12]. In addition, other genetic lesions such as Nkx3.1, Trp53 or p27KIP1 might be involved in decreasing latency and increasing invasiveness/metastasis during tumor development in Pten +/2 mice [13,14,15,16,17].
Using Cre-loxP conditional genetics [18,19], tissue-specific inactivation of Pten results in tumor formation in the targeted tissues of the mouse. Prostate-specific ablation of loxP-flanked Pten gene (Pten fx/fx ) utilizing different Cre transgenic mice lines such as probasin-Cre, prostate-specific antigen (PSA)-Cre or mouse mammary tumor virus (MMTV)-Cre has shown the occurrence of prostate cancer progression similar to that which occurs in humans [20,21,22,23]. Likewise, conditionally ablating the Pten gene in hematopoietic lineages causes various leukemias in the mouse [24]. Such Cre activity might also be present in the diseaseinitiating stem cells from which the Pten gene is deleted, which is followed by tumor initiation, expansion and progression [24,25,26]. Thus, tissue-specific and cell type-specific Pten knockout mice have provided a fundamental basis for an understanding of the role of Pten in different tumor progression models. However, somatic inactivation of Pten in a temporally controlled manner in all adult tissues to test susceptibility of various tissues to Pten-deficient tumorigenesis has not been carried out up to the present.
To assess the tumor incidence, latency and spectrum of Ptendeficient adult mice, we utilize two mouse strains: firstly, the ROSA26 Cre-estrogen receptor (R26-CreER T ) knock-in mouse line, which expresses inducible Cre recombinase driven by the ubiquitous promoter ROSA26 [27] and secondly, the Pten fx/fx mouse carrying loxP-flanked exon 5 encoding the phosphatase domain of PTEN [28,29]. 4-hydroxytamoxifen (4-OHT) is administrated to induce the Pten gene excision in a temporally controlled manner in the crossed mutant offspring (R26-CreER T /+;Pten fx/fx ). In this report, we found that lymphoid lineages, the squamous epithelium of the epidermis, the simple epithelium of the prostate, the endometrium and the intestine are highly susceptible to tumorigenesis in such a Ptendeficient background. Application of the PI3K inhibitor LY294002 in these Pten-deficient mice resulted in the enlargement of the anterior prostate glands being significantly suppressed. Thus, we have established a Pten-deficient tumor model, which demonstrates that multiple cell lineages can be provoked to undergo aberrant cellular signaling simultaneously and that this results in major changes of normal tissues.

Temporally controlled CreER-mediated recombination by 4OHT administration
The inducible Cre activity of the R26-CreER T transgenic mice was demonstrated in neuronal tissues by Badea et al. [27]. In this report, we have examined temporally controlled Cre activity in the systemic organs of R26-CreER T ; R26R bigenic mice. After 4OHT treatment for one week, whole mount X-gal staining revealed that the 4OHT-induced b-galactosidase expression showed focal or mosaic blue patterns in the brain, liver, pancreas, kidney, intestine, uterus and bladder of R26-CreER T ;R26R bigenic males or females ( Figure 1). Non-specific X-gal staining was also observed in the hind-stomach, the gut, the prostate and the vas deferens ( Figure 1 and data not shown). These results suggested that the inducible Cre activity of R26-CreER T transgenic mice had been successfully controlled by 4OHT in a variety of organs, although the level of inducible Cre activity may have differed as indicated by the variable X-gal stained intensity among these systemic organs.
To investigate tumor susceptibility in the Pten-deficient mice at an adult stage, we generated R26-CreER/+;Pten fx/fx (referred to as R26-Pten fx/fx hereafter) and control (R26-Pten fx/+ , Pten fx/+ or Pten fx/fx ) mice and followed this with 4OHT induced Cre/loxP recombination to excise exon 5 of Pten gene (Figure 2A). After 4OHT treatment for one week, we examined the efficiency of exon 5 of Pten gene excision using the genomic PCR method in a variety of organs dissected from males and females carrying the Pten fx/+ , R26-Pten fx/+ and R26-Pten fx/fx genotypes ( Figure 2B). Our result showed that exon 5 of the Pten gene excision was detected in R26-Pten fx/+ and R26-Pten fx/fx tissues, indicating that 4OHT was able to induce Cre-mediated excision of loxP-flanked region in the tissues examined ( Figure 2B). Notably, the excision efficiency of Pten showed no overt differences between males and females carrying the R26-Pten fx/+ or R26-Pten fx/fx genotypes. However, inducible Cre activity may vary slightly across the organs of the R26-Pten fx/+ or R26-Pten fx/fx mice ( Figure 2B). We next examined the expression of PTEN protein in selected organs, specifically the anterior prostate and the intestinal tract, of 4OHT-injected R26-Pten fx/fx and control (R26-Pten fx/+ ) mice using immunofluorescence microscopy. Our results showed that PTEN could be detected in the prostate and colon epithelium, which was indicated by cytokeratin 8 (CK8) staining of the controls (yellow; CK8-positive, PTEN-positive). In contrast, PTEN expression was lost in most of the hyperplastic epithelial lesions and tumor lesions (red; CK8positive, PTEN-negative) found in the corresponding organs of the R26-Pten fx/fx mice ( Figure 2C). We also examined the expression of PTEN in various other systemic organs, namely the lung and the kidney, of R26-Pten fx/fx and the control mice to demonstrate PTEN loss ( Figure S1). Our results suggested that the majority of PTEN was lost from the systemic organs of the R26-Pten fx/fx mice ( Figure 2B & C; Figure S1).
Analysis of R26-Pten fx/fx and R26-Pten fx/+ malignancies Furthermore, R26-Pten fx/fx (n = 33; 20 males and 13 females) and control (Pten fx/fx , n = 24; R26-Pten fx/+ , n = 21) mice were given 4-OHT injection (i.p.) at age of 6 weeks and then monitored for tumor development over about 60 weeks. The mice were sacrificed for pathological analysis (H&E and IHC) when they manifested sign of distress or on tumor detection. The cumulated tumor-free survival curves of the R26-Pten fx/fx and control mice are shown in Figure 3A. Our results showed that all the R26-Pten fx/fx mice died from their tumor burden by 45 weeks post 4OHT injection ( Table 1). The overall mean latency of Pten-deficient tumor formation was 17 weeks. In addition, we found 4 R26-Pten fx/+ mice (19.0%) developed various different types of tumors by 60 weeks post 4OHT treatment ( Table 1). We found that one R26-Pten fx/+ mouse suffered from metastatic lung cancer at 42 weeks, two R26-Pten fx/+ females developed mammary tumors at 52 and 55 weeks and another R26-Pten fx/+ male mouse developed hepatocellular carcinoma at 58 weeks (Table 1; Figure S2). No tumors were found in the control group Pten fx/fx mice at 60 weeks.
Interestingly, we found that the tumor latency and spectrum of the Pten-deficient mice exhibited gender differences. Approximately, 50% R26-Pten fx/fx males and females had malignant tumors at 21 weeks and between 10 and 11 weeks, respectively ( Figure 3B). Most R26-Pten fx/fx females quickly developed lymphomas (10 of 13; 76.9%) and died at about 9 weeks post-4OHT injection compared to a much lower incidence of lymphoma formation (8 of 20; 40%) by 13 weeks in the males ( Figure 3C). The Pten-deficient male mice developed intestinal cancers (7 of 20, 35%) arising from the colorectum (n = 6) and small intestine (n = 1), whereas no female mice developed this type of cancer. This might be partly due to the quick development of the lymphomas, which caused early lethality before the intestinal cancer could develop in the R26-Pten fx/fx females. In addition, the R26-Pten fx/fx mice developed squamous cell carcinoma of the epidermis (4 of 33 mice, 12.1%; 2 males, 10% and 2 females, 15.4%), endometrial cancer (6 of 13 females, 46.1%) and prostate cancer (4 of 20 males, 20%) ( Figure 3C). The mean latency of these malignancies is showed in Figure 3D. The sequential occurrence of the Pten-deficient malignancies was about 11 weeks for the lymphomas and the endometrial cancer, about 19 weeks for the prostate cancer and squamous cell carcinoma and about 30 weeks for the intestinal cancer ( Figure 3D).
The lymphomas mainly arose from the thymus and mesenteric lymph nodes. A representative thymic lymphoma is shown grossly in Figure 4A. H&E staining revealed a mature appearance for the small lymphocytic cells and a loss of cortical-medullary architec-  The 4OHT inducer was used to temporally active the CreER T recombinase activity, which resulted in the excision of the DNA fragment containing the Pten exon 5. (B). PCR genotyping of the Pten floxed allele and Cre-mediated exon 5 excision; After one week of 4OHT injection, genomic DNA was isolated individually from multiple organs including the lung, the liver, the intestine, the kidney, the spleen, the thymus, the tail, the anterior prostate and the uterus of mice carrying different genotypes (C, Pten fx/+ ; 1, R26-Pten fx/+ male; 2, R26-Pten fx/+ female; 3, R26-Pten fx/fx male; 4, R26-Pten fx/fx female). PCR genotyping is shown that allows the identification of the Pten floxed allele (,1100-bp), the wild-type (,1000-bp) and the exon 5 excised alleles (D5; ,400-bp) from the 4OHT-treated mice. (C). Immunostaining of PTEN in the prostate and the colon. Representative merged images of the immunofluorescence analysis reveal the expression level of PTEN in the anterior prostate lobe (at 7 weeks post-4OHT treatment) and the colon (at 37 weeks post-4OHT treatment) of the R26-Pten fx/+ and R26-Pten fx/fx mice. Antibody against PTEN (green) and TROMA-I antibody against cytokeratin 8 (CK8; red) were used to study the expression of PTEN in the simple epithelial subsets. Arrows indicated the coexpression of PTEN and CK8 (yellow) in anterior prostate gland and colon epithelium of R26-Pten fx/+ mice (left panels), whereas the hyperplastic epithelial subsets (middle panels; red) of the R26-Pten fx/fx mice showing PTEN loss (arrowheads). In addition, the immunofluorescence images in the right panels reveal PTEN loss in the tumor lesions of prostate and colon of R26-Pten fx/fx males (at 13 and 39 weeks, respectively). Scale Bar, 200 mm. doi:10.1371/journal.pone.0001237.g002 ture in this thymic tumor ( Figure 4A). To verify the potential origins of this lymphoma, T-cell and B-cell origins were respectively determined by antibodies against CD3 and B220 in the liver with the infiltrated lymphoid tumor cells in the portal area identified using IHC, which showed that this tumor was a CD3positive T-cell lymphoma ( Figure 4B). Through evaluation of CD3 and B220 expression in the lymphomas formed in the mutant mice, we found that all lymphomas were derived from CD3-positive T-cells with either a mature appearance of small lymphocytes or large lymphoblasts (data not shown). Occasionally, the more detail surface markers that were expressed on these lymphomas were analyzed by flow cytometry in order to assist our verification of these lymphomas. As an example, a CD4-rich T-cell lymphomas was found to have developed in a R26-Pten fx/fx mouse at six weeks post-treatment ( Figure 4C). We also found that one mouse developed both CD4-and CD8-rich lymphomas arising from the thymus and mesenteric lymph node, respectively, suggesting that 4OHT-induced Pten loss in different lymphoid origins may develop simultaneously (data not shown).
In addition to lymphomas, prostate cancer, endometrial cancer, intestinal cancer and squamous cell carcinoma of the epidermis were found in R26-Pten fx/fx mice using H&E and IHC analyses (Figures 3 & 5). The focal invasions of the prostate, endometrial and colorectal carcinomas were observed and characterized by H&E or IHC using antibodies against E-cadherin ( Figure 5). Precancer lesions (hyperplasia/dysplasia and adenoma) of corresponding malignancies were also observed in the R26-Pten fx/fx mice ( Figure 6). We found that R26-Pten fx/fx males had developed small or large intestinal polyps at late onset (.35 weeks; Figure 6A & B). In addition, most mutant females developed endometrial hyperplasia (7 of 13) at between 7 and 9 weeks post-4-OHT injection ( Figure 6C). The early alternations associated with prostate hyperplasia (BPH) or prostate intraepithelial neoplasm (PIN) were observed at 4,6 weeks post 4-OHT injection ( Figure 6D).
Moreover, multiple tumors were also observed in R26-Pten fx/fx mice (9 of 33, 27.3%; 5 females and 4 males). Among these animals, all females developed T-cell lymphoma and endometrial cancer at 7,23 weeks post-4OHT injection. One of these females also developed squamous cell carcinoma additionally. Furthermore, two males developed prostate cancer together with lymphoma (at 13 weeks) and colorectal cancer (at 27 weeks), respectively. Finally, two males developed squamous cell carcinomas and colorectal cancer at 28 and 31 weeks post-4OHT injection.

Temporally controlled prostate tumorigenesis
The advantage of the use of inducible CreER T /loxP technology in this report is that it allowed the monitoring of tumor development at specific time points after 4-OHT injection. To investigate the progression of Pten-deficient malignancies, prostate cancer progression was chosen as the initial model because a good understanding of the role of Pten in this disease has been gained from other prostate-specific Pten knockout mouse model studies [20,21,23,25]. At 4 weeks, 6 weeks and 30 weeks after 4-OHT injection, we found gradually enlargement of the anterior prostate in R26-Pten fx/fx mice compared to the controls ( Figure 7A). Histologically, these prostate lesions were studied at the above time points (4, 6 and 30 weeks) and this revealed the prostate cancer progression through BPH/PIN to invasive cancer ( Figure 7A). We then determined activation of AKT using antibody against phosphorylated AKT (p-AKT-Ser473) specifically. We found that AKT phosphorylation appeared in all prostate lesions along with hyperplasia and neoplasia in the R26-Pten fx/fx mice, unlike the controls ( Figure 7B). To identify the potential cell type of the Ptendeficient prostate cancer, we performed double immunofluorescent staining to verify the presence of cytokeratin 5 (CK5 + )-expressing basal cells and cytokeratin 8 (CK8 + )-expressing lumen epithelial cells from the normal, BPH, PIN and prostate cancer samples in the R26-Pten fx/fx mice compared with the controls. Flattened and discontinuous CK5 + -expressing cells were observed underneath the CK8 +expressing lumen epithelium in the control prostate, whereas stratification of the CK8 + lumen epithelium together with flatten CK5 + basal cells were noticed in the BPH/PIN lesions ( Figure 7C). As the disease progressed to invasive prostate cancer in the R26-Pten fx/fx males, focal expansion of the CK5 + cells and coexpression of the CK5 + /CK8 + cells were significantly increased ( Figure 7C). However, the significance of the increased CK5/CK8 double positive cells in the Pten-deficient prostate cancer remains unclear.

The PI3K inhibitor, LY294002, suppresses prostate hyperplasia
We next evaluated the effect of the potential cancer targeting agent LY294002, which is a potent inhibitor of PI3K-AKT signaling pathway in Pten-deficient prostate tumor progression. After LY294002 treatment for 4 weeks, we found that the apparent increased anterior prostate weight in R26-Pten fx/fx mice (n = 4) was significantly suppressed (p = 0.007) in comparison to the controls (LY294002-treated Pten fx/fx mice, n = 3; vehicle-treated R26-Pten fx/fx mice, n = 3; Figure 8A). Histological H&E staining revealed that the increased cellularity and nuclear atypia of the anterior prostate were diminished in LY294002 treated R26-Pten fx/fx mice compared with the untreated mice ( Figure 8B). We next examined whether LY294002 treatment blocked the PI3K-AKT signaling pathway. Using IHC, we found that only focal epithelial lesions showed pAKT(Ser473)-positive staining in LY294002 treated R26-Pten fx/fx prostates, suggesting that LY294002 treatment strongly suppresses AKT phosphorylation at Ser473 in R26-Pten fx/fx mice compared to the untreated animals ( Figure 8B). Therefore, our results demonstrated that inhibition of the PI3K-AKT pathway was sufficient to suppress Pten-deficient prostate hyperplasia.

DISCUSSION
In this study, we demonstrated that the tumor spectrum, tumor incidence and tumor latency during temporally controlled Pten loss in the mouse. This is the first in vivo analysis of Pten-deficient tumorigenesis in the systemic organs of the adult mouse. It is likely that lymphoid lineages and the epithelium of the prostate, endometrium, intestine and epidermis are highly susceptible to tumorigenesis after the Pten gene is excised by 4OHT-induced Cre/loxP recombination in the mouse's adult tissues. Although we have observed that Cre-mediated Pten excision did slightly vary among the systemic organs of R26-Pten fx/+ or R26-Pten fx/fx mice, the variation seemed to show no correlation with the occurrence of Pten-deficient malignancies. Neoplastic cells may initiate from a single cell or a small number of cells because they are highly susceptible to tumorigenesis after PTEN loss and therefore slight variations in the Cre-mediated Pten excision will not affect the presence of such foci. Clinical studies have shown that somatic alterations of PTEN are found in proximately 5% of lymphomas [30], 29% of prostate cancer [31] and 34% of endometrial cancer [32]. Moreover, germline PTEN mutation shows strong linkage to Cowden syndrome (CS) and Bannayan-Zonana syndrome (BZS). These are inherited tumor predisposition syndromes in which hamartomas may develop in the central nervous system, gastrointestinal tract, skin, breast and thyroid [33]. Affected individuals of CD families also have a higher risk of malignancy in corresponding tissues [33,34,35]. Thus, our findings for the R26-Pten fx/fx tumor malignancies resemble to some degree of the tumor spectrum observed in human PTEN genetic studies. Interestingly, the tumor latency and tumor spectrum of R26-Pten fx/fx mice exhibit gender-bias. Pten-deficient females are predisposed to lymphoma with an early onset of about 7 weeks, which may preclude the formation of other malignancies; the result of this may be the overall gender differences in the tumor spectrum. Particularly, intestinal cancer was not observed in Ptendeficient females, which may partly be due to the longer period time required to develop intestinal cancer, namely 25 weeks to 35 weeks post 4OHT treatment in mutant male mice.
Furthermore, we observed that only four mice heterozygous for the Pten floxed allele, among 21 R26-Pten fx/+ mice analyzed, developed a variety of malignancies (up to 60 weeks).The incidence (19.0%, 4 of 21), latency (42 to 58 weeks) and spectrum (lung, breast and liver cancers) of the malignant tumors in our R26-Pten fx/+ mice is quite different from that of a previous reported Pten +/2 mouse model [9]. This report described approximately 18% lymphomas, 22% uterine cancer and 49% breast cancer occurring at ages ranging from 26 to 65 weeks. This discrepancy may be explained by the focal or mosaic excised activity of Cre/loxP-mediated recombination in the R26-Pten fx/+ mice. In addition, it is possible that the genetic background of the two mouse models or some other undefined mechanism may affect the tumor latency and spectrum between the different mouse models of Pten heterozygosity [8,11,12,36]. In a previous study, Freeman et al. reported that tumorigenesis in Pten heterozygotes was strongly affected by the genetic background of the mouse [36]. Mutant mice with a 129;C57BL6 background tend to have a higher incidence of lymphoid and endometrial malignancies. In contrast, Pten heterozygotes with a 129;BALB/c background develop prostate cancer more frequently. These findings are significant because they suggest that there are different modifier(s) present in different mouse strains that affect tumorigenesis in the mice after Pten loss. However, it remains unclear currently whether such background differences or putative modifier(s) are able to efficiently contribute to tumorigenesis in bi-allelic Pten loss. This might be further examined by generating R26-Pten fx/fx mice with different genetic backgrounds or different mutant strains in the future.
Although the multiple tumors that occurred simultaneously in the Pten-deficient mice may have complicated the monitoring of tumor formation compared to other mouse models that have used Pten ablation in a tissue-specific manner, our study is able to provide knowledge that will help an understanding of Pten tumor suppressor functional requirements in multiple tissue lineages and cells. In addition, the latency of the different Pten-deficient cancers provides a valuable time frame for a more detailed dissection in the future of the downstream effectors of Pten loss using pharmacological or genetic approaches. In this study, we tested the pharmacological inhibition of the Pten downstream signaling cascade using the PI3K inhibitor, LY294002, which has been demonstrated to suppress tumor outgrowth in xenografted ovarian tumors and pancreatic cancer cells in the nude mice [37,38]. Recently, Shukla et al showed that LY294002 treatment suppresses the invasive properties of several human prostate cancer cell lines [39]. In our study, we found that prostate hyperplasia after induced Pten loss at an adult stage is suppressed by LY294002. To our knowledge, this is the first evidence to show tumor inhibitory activity of LY294002 in spontaneous tumor formation using a genetically engineered mouse model. Moreover, recent reports have shown that mTOR inhibition effectively suppresses neuronal hypertrophy, endometrial hyperplasia and leukemia initiation after tissue-specific Pten loss or in the Pten +/2 mice [24,40,41]. Likewise, mTOR inhibition could also be tested in our temporal controlled Pten-deficient tumor model and this will allow the dissection of mTOR pathway requirements when Pten is lost in systemic organs.
In summary, we utilized an inducible Cre/loxP system to ablate Pten in a temporal-specific manner in the adult mouse. Through the monitoring of tumor spectrum, tumor latency and tumor incidence in the Pten-deficient mice, our findings provide an entry point for an understanding the role of Pten in tumorigenesis across different tissues. Furthermore, these 4OHT-inducible R26CreER T ; Pten fx/fx mice will be a useful mouse model that can be used to identify and study potential therapeutic approaches that will prevent Pten-deficient malignancies.

Mice
Pten fx/fx [28,29], ROSA26 Cre reporter (R26R) [42] and R26-CreER T [27] mice were obtained from the Jackson Laboratory (Jackson Laboratory, Bar Harbor, ME, USA). R26-CreER/+;R26R/+ bigenic mice were generated in order to examine the effect of inducible Cre activity in the systemic organs. R26-CreER T / +;Pten fx/fx and the control (R26-CreER T /+;Pten fx/+ or Pten fx/fx ) mice were generated in a mixed background (BALB/c;129;C57BL/6) by crossing Pten fx/fx and R26-CreER T mice; this was followed by crossing R26-CreER T /+;Pten fx/+ and Pten fx/fx mice. Up to five mice were housed in microisolator cages and provided with autoclaved food and water in a specific pathogen free colony. The experimental procedures using animals were in accordance with the guidelines of Institutional Animal Care and Use Committee. PCR genotyping was performed according to the standard protocols.
Pharmacologically administration 4-hydroxytamoxifen (4-OHT; Sigma-Aldrich, St. Louis, MO, USA) was dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich) at a concentration of 10 mg/ml. 4-OHT solution was emulsified in sunflower seed oil (Sigma-Aldrich) by vortexing, which was followed by mixing on a rotator for 4,6 hours. The mice at the age of 6 weeks were injected intraperitoneally with a dose of 4-OHT (4,5g per gram body weight) every other day three times. Similarly, LY294002 (Calbiochem, La Jolla, CA, USA) was prepared using the same procedure as 4-OHT. For the LY294002 administration, mice were injected intraperitoneally with LY294002 (4,5g per gram body weight) one week after the 4OHT treatment.

Whole mount X-gal staining
The systemic organs including brain, lung, liver, spleen, pancreas, lung, kidney, stomach, intestine and reproductive system were dissected. Organs were fixed transiently with 10% neutral-buffered formalin for 30 minutes and rinsed with rinse buffer (2 mM MgCl 2 , 0.1% sodium deoxycholate, 0.2% NP-40, 16 PBS, pH7.3) for 30 minutes. Organs were incubated with X-gal staining solution (5 mM potassium hexacyanoferrate, 5 mM potassium hexacynoferrate trihydrate, 1 mg/ml X-gal in rinse buffer) for 1 hour at 37uC or room temperature with gentle agitation until the appearance of blue coloration and then organs in X-gal staining solution were transferred to 4uC refrigerator overnight. Finally, stained organs were post-fixed with neutral-buffered formalin overnight and stored in 70% ethanol at 4uC or photography examined by stereomicroscopy (MZ6, Leica, Germany).

Analyses of tumor-free survival and tumor malignancy
The tumor-free survival rate was analyzed using the Kalpan-Meier method by the SPSS program (SPSS, Inc., Chicago, IL, USA). Detail analysis of tumor formation has been described previously [43,44]. Briefly, mice were inspected for illness or tumor formation twice every week. Normal tissues or tumor samples were isolated and fixed in 10% neutral-buffered formalin at 4uC overnight. Subsequently, the samples were embedded in paraffin and further processed for histopathological studies as described below. Lung, kidney, liver and enlarged lymph nodes were also isolated to histologically to be examined for metastasis. If tumors were not identified by gross examination, then a necropsy was performed. Diagnosis of tumor malignancy was carried out in consultation with pathologists. In general, tissue sections stained by hematoxylin and eosin (H&E; see below) were examined by light microscopy (BX51, Olympus, Japan). Tumor malignancy was characterized by invasive behavior or metastasis. Expansive tumor lesions, but with a relatively normal cytological appearance and not showing an invasive behavior, were considered as benign or pre-cancerous.

Hematoxylin and eosin (H&E) staining
Tissue sections on slides were deparaffinized and rehydrated and this was followed by staining with Harris Hematoxylin (DakoCytomation, Denmark) for 3,5 minutes. Sections were washed with ddH 2 O and then transferred to Tris-HCl (pH 8.0) for 5 minutes. Subsequently, the sections were stained with Eosin Y (Shandon, Tokyo, Japan) for 30 seconds, dehydrated and then mounted.

Immunohistochemical (IHC) and immunofluorescence studies (IF)
Tissue sections were deparaffinized, rehydrated and placed in boiling antigen retrieval buffer (DakoCytomation). Endogenous peroxidase activity was quenched by 3% hydrogen peroxide for 5 minutes and this was followed by incubation with blocking reagent (1% Cold Water Fish Skin Gelatin, 5% Bovine Serum Albumin in 16 Phosphate Buffered Saline) for 1 hour at room temperature. For IHC, the tissues sections were subsequently incubated with primary antibodies against CD3 (ready to use; Abcam Ltd, Cambridge, UK), B220 (1:50 dilution; Southern Biotech, Birmingham, AL, USA;), E-cadherin (1:50 dilution; DakoCytomation) and p-AKT(Ser473) (1:100 dilution; Cell Signaling Technology, Beverly, MA, USA) for 1 hour at room temperature The tissue sections was then incubated with biotinylated-linked secondary antibody (anti-mouse/rabbit; Dako-Cytomation) for 30 minutes and Streptavidin-HRP (DakoCytomation) was added to amplify the signals. Staining was performed using a NovaRed substrate kit (Vector Laboratories, Burlingame, CA). The slides were counterstained for a short time with Hematoxylin, which was followed by dehydration and then mounting. For IF, tissues sections were incubated with primary antibodies against PTEN (1:12500 dilution; Cascade BioScience, Winchester, MA), cytokeratin 5 (CK5, 1:400 dilution; Abcam) and/or cytokeratin 8 (CK8 or TROMA-1, 1:400, Developmental Study Hybridoma Bank, University of Iowa, Iowa City, IA, USA). Secondary antibodies (AlexaFluor 568 conjugated anti-rat IgG against antibody to CK8 and AlexaFluor 488 conjugated antirabbit IgG against antibody to CK5; Molecular Probes, Invitrogen) were next incubated on the slides for 1 hour and this was followed by DAPI staining for 2 minutes. A tyramide signaling amplification (TSA) system (kit#22 with HRP-streptavidin and AlexaFluor 488, Invitrogen) was used to enhance the PTEN signal. Finally, the slides were further incubated with DAPI for 2 minutes and mounted in fluorescent mounting medium (DakoCytomation).

Flow cytometry analysis
The control or enlarged lymphoid organs (thymus and lymph nodes) were dissected and collected in DMEM medium. These lymphoid tissues were grounded and filtrated through cell strainers. The cells present were counted and then centrifuged for 5 minutes at 4uC. Then, the cell suspensions were incubated with 2.4G2 monoclonal antibody (a gift from Dr Jeffery J.Y. Yen) for 30 minutes on ice followed by labeling with FITC-conjugated antibodies against CD4 (eBioscience, San Diego, CA, USA) and PE-conjugated antibody against CD8 (eBioscience) for 15 minutes. After washing the cell pellets twice with 1xPBS, the cells were resuspended in 1xPBS containing Propidium Iodine (1g/ml, eBioscience). The percentage of live cells (PI exclusive) expressing the CD4/CD8 surface markers were analyzed by flow cytometry (FACScalibur, Becton Dickinson, USA) using Cell Quest software. Figure S1 Immunofluorescence analysis of PTEN expression. After one week of 4OHT injection, PTEN expression was determined using antibody against PTEN (green) and counterstained with DAPI (blue) in the lung and the kidney of the R26-Pten fx/+ and R26-Pten fx/fx mice.