|Home | About | Journals | Submit | Contact Us | Français|
Despite clear epidemiological and genetic evidence for X-linked prostate cancer risk, all prostate cancer genes identified are autosomal. Here we report somatic inactivating mutations and deletion of the X-linked FOXP3 gene residing at Xp11.23 in human prostate cancer. Lineage-specific ablation of FoxP3 in the mouse prostate epithelial cells leads to prostate hyperplasia and prostate intraepithelial neoplasia. In both normal and malignant prostate tissues, FOXP3 is both necessary and sufficient to transcriptionally repress cMYC, the most commonly over-expressed oncogene in prostate cancer as well as among the aggregates of other cancers. FOXP3 is an X-linked prostate tumor suppressor in the male. Since the male has only one X chromosome, our data represents a paradigm of “single-genetic-hit” inactivation-mediated carcinogenesis.
The study describes two significant advances. First, we demonstrate FOXP3 as an X-linked tumor suppressor gene in the male in both human and mice. Since male has only one X-chromosome, our work represents a compelling exception to the widely accepted “two-hit” theory for inactivation of tumor suppressor genes. Second, our work demonstrates FOXP3 as a major transcriptional repressor of c-MYC oncogene in the prostate. FOXP3 inactivation is necessary and sufficient for c-MYC over-expression, which is critical for molecular pathogenesis of prostate cancer.
Genetic lesions of several autosomal tumor suppressor genes, including PTEN (Sansal and Sellers, 2004; Suzuki et al., 1998), NKX3.1 (Emmert-Buck et al., 1995; Vocke et al., 1996), and KLF6 (Bar-Shira et al., 2006; Narla et al., 2005; Narla et al., 2001) have been implicated in the molecular pathogenesis of prostate cancer patients. In addition, epidemiology studies have suggested a role for X-linked genes that control the susceptibility to prostate cancer (Monroe et al., 1995). While two loci, one in Xp11.22 (Gudmundsson et al., 2008) and one in Xq27-28 (Xu et al., 1998), have been implicated, the genes in these regions have not been identified. X-linked tumor suppressor genes are of particular interest as the majority of X-linked genes are dose compensated, making a single-hit sufficient to inactivate their functions (Spatz et al., 2004). While we and others have reported X-linked tumor suppressor genes, WTX1 (Rivera et al., 2007) and FOXP3 (residing at Xp11.23) (Zuo et al., 2007b) in female cancer patients, none have been identified for cancer in male patients.
In addition to inactivation of tumor suppressors, activation of proto-oncogenes also play a critical role in carcinogenesis. Among them, c-MYC (hereby called MYC) is known as one of the most commonly over-expressed oncogenes. MYC over-expression occurs in more than 30% of all human cancer cases studied (Grandori et al., 2000). However, the mechanism by which MYC transcription is increased in the prostate cancer remains unclear. In Burkitt’s lymphoma, the MYC locus is translocated into a constitutively active Ig locus (Dalla-Favera et al., 1982; Taub et al., 1982), which was found to lead to its transcriptional activation (Erikson et al., 1983). In lung cancer, high levels of gene amplification of the MYC locus have been documented (Wong et al., 1986), although such amplification occurred considerably less frequently than over-expression of MYC mRNA (Takahashi et al., 1989). Likewise, in breast and prostate cancer, upregulation of MYC mRNA was substantially more frequent than amplification of the MYC gene (Bieche et al., 1999) (Jenkins et al., 1997; Latil et al., 2000). Since MYC has been shown to be a target of β-catenin activation (He et al., 1998; Sansom et al., 2007), an appealing hypothesis is that MYC upregulation may be a manifestation of aberrant Wnt signaling, which occurs frequently in a variety of cancers (Fearon and Dang, 1999). However, a general significance of β-catenin-mediated MYC upregulation remains to be demonstrated (Kolligs et al., 1999) (Bommer and Fearon, 2007). Therefore, for the majority of cancer types, the genetic change involved in the aberrant MYC expression remained to be defined.
MYC over-expression in benign prostate hyperplasia and prostate cancer was documented over 20 year ago (Fleming et al., 1986). Ectopic expression of MYC causes hyperplasia of prostate tissue (Thompson et al., 1989). Further studies demonstrated that in combination, Myc and Ras could induce prostate cancer in mice (Lu et al., 1992). More recently, gene expression profiling of a mouse prostate cancer model induced by Myc transgene indicated similarity with human prostate cancer (Ellwood-Yen et al., 2003). Consistent with a role for MYC in the pathogenesis of prostate cancer, several whole genome scanning studies have strongly implicated a region 260 kb telomeric to the MYC gene in susceptibility to prostate cancer (Amundadottir et al., 2006; Gudmundsson et al., 2007; Haiman et al., 2007a; Haiman et al., 2007b; Witte, 2007; Yeager et al., 2007).
Here we investigated whether the FOXP3 gene is frequently inactivated in prostate cancer samples by deletion and somatic mutation. Moreover, we determined the significance of such inactivation by growth inhibition of normal and cancerous prostate cell lines and by identification of FOXP3 targets. The impact of prostate-specific ablation of the FoxP3 gene in the mouse was also tested.
We first evaluated the expression of FOXP3 in both normal and malignant prostate tissues by immunohistochemistry. While our previous studies have demonstrated the expression of FoxP3 in the mouse prostate using an affinity purified anti-FoxP3 peptide antibody (Chen et al., 2008), the FOXP3 expression was not reported in either normal or malignant prostate tissues in human by immunohistochemistry (IHC), even though FOXP3 expression on infiltrating regulatory T cells was clearly detectable (Fox et al., 2007; Roncador et al., 2005). As the master regulator of regulatory T cells, FoxP3 is expressed there at levels comparable to house keeping genes such as GAPDH and HPRT (Fontenot et al., 2003; Hori et al., 2003). Since FoxP3 expression in prostate tissue is approximately 100-fold lower than what was found in regulatory T cells (Chen et al., 2008), we reasoned that the lack of detectable FOXP3 in normal prostate tissue may be caused by low sensitivity of staining and/or tissue processing conditions. Therefore, we first fixed the frozen tissues in 10% formalin for 8-12 hours and screened a large panel of commercially available anti-FOXP3 antibodies for their reactivity to endogenous FOXP3 in epithelial tissues. A shown in supplemental Fig. S1, anti-FOXP3 mAb stained prostate epithelial uniformly. However, compared with infiltrating lymphocytes, the level of FOXP3 is considerably lower (supplemental Fig. S2).
As summaried in supplemental Table S1, four of commercially available mAbs gave uniform staining of FOXP3 in normal prostate. The fact that multiple anti-FOXP3 mAbs reacted to FOXP3 demonstrated that FOXP3 is expressed at significant levels in normal prostate tissue. Among them, two (hFOXY and 236A/E7) were also tested and found to react specifically with FOXP3 protein in Western blot of lysates made from immortalized mammary epithelial cell line MCF-10A. The specificity of the reactivity to human FOXP3 was further confirmed by comparing reactivity of scrambled and FOXP3 ShRNA-transduced normal epithelial cell line MCF10A by Western blot and by IHC (supplemental Fig. S3).
Using the uniform fixation and processing conditions, we evaluated the expression of FOXP3 in 85 cases of normal and 92 cases of cancer tissues. As shown in Fig. 1A, immunohistochemistry with anti-FOXP3 mAb can detected nuclear FOXP3 staining in 100% of the normal prostate tissues tested. In contrast, only 31.5% of the prostate cancer samples show nuclear FOXP3 staining (P = 8.82 × 10-21). Perhaps due to harsher fixation conditions in preparing samples for tissue microarrays, the FOXP3 protein was generally difficult to detect by IHC, unless high concentrations of antibodies were used (data not shown). Using an extended antigen retrieval (37°C overnight following heating in microwave oven), we have obtained clear, albeit somewhat weaker staining from tissue microarray samples using low concentration of anti-FOXP3 mAb. However, 13% of normal tissues were not stained, presumably due to harsher fixation conditions. As shown in Fig. 1B, a significant reduction was observed among prostate cancer tissues. Furthermore, when the samples with PIN were compared with normal tissues, we observed a statistically significant reduction of FOXP3 expression in PIN (Supplemental Fig. S4A, B). Taken together, our data demonstrated that FOXP3 down-regulation is widespread in prostate cancer and that such down regulation may have occurred at an early stage of prostate cancer.
We then used microdissection to obtain benign prostate tissue and cancer tissues from the same patients and compared the FOXP3 mRNA levels. Since inflammatory T cells are a major source of FOXP3 expression, we carefully avoided areas of inflammation for dissection. After normalizing against the house-keeping gene, 14/18 cases showed 2-10 fold reduction of FOXP3 mRNA in comparison to the benign tissues (Fig. 1C). Six of the 18 samples contain clearly identifiable PIN lesions. We therefore micro-dissected the PIN lesions and compared the levels FOXP3 mRNA transcript with normal and cancerous tissues. As shown in Fig. S4C, FOXP3 transcript is down regulated in PIN lesion. Thus, reduced FOXP3 expression is widespread among prostate cancer samples, perhaps starting at early stage of carcinogenesis.
We used fluorescence in situ hybridization (FISH) to determine FOXP3 gene deletion in the prostate cancer tissue. As shown in Fig. 2 and supplemental Table S2, 23 of 165 samples (13.9%) tested showed a deletion of the FOXP3 gene. Among them, 18 of the 23 cases had a single copy of the X chromosome. However, 5 out of the 23 cases showed an increase in the number of X-chromosomes. Interestingly, in cells with X polysomy and FOXP3 deletion, the deletion was complete in all X-chromosomes. Thus, X-chromosome duplications in cancer tissues likely occurred after deletion of FOXP3.
In order to determine whether FOXP3 was somatically mutated in primary prostate cancer samples, we isolated cancerous and normal prostate tissues from the same patients and compared the DNA from exons and some exon-intron junctions. A summary of the data is shown in Fig. 3A and a representative chromatogram is shown in Fig. 3B, with other chromatograms provided in supplemental information Fig. S5. Our sequencing analyses demonstrate single base-pair changes in 5/20 samples tested (Table S3). Among them, four were missense mutations while one caused a change in intron 6. One of the missense mutation (K227R) was also reported in the breast cancer (Zuo et al., 2007b). The tumors with the intron 6 mutation showed reduced expression of FOXP3 (Fig. 3C). Among the five samples that contain FOXP3 mutation in cancer tissue, two contained identifiable PIN lesions. We therefore microdissected PIN lesion to determine if the same mutation can be found. As shown in Fig. S5, both samples had the same mutations in PIN and cancerous tissues.
In order to substantiate tumor suppressor activity of FOXP3, we transfected FOXP3 cDNA into prostate cancer cell lines PC3, LNCaP, and Du145. Our data demonstrated strong growth inhibition by FOXP3 (supplemental Fig. S6, 7). Importantly, while vector-transfected LNCaP response to hormone 5α-dihydrotestosterone (5α-DHT), FOXP3 expression abrogated its stimulation by the hormone (Supplemental Fig. S7). The growth inhibition by wild-type (WT) FOXP3 provided an important functional test for the somatic mutants uncovered from the clinical samples. As shown in Fig, 3D, only WT FOXP3, but not any of the missense mutants, abrogated growth of prostate cancer cell line Du145. Similar data were obtained with another cell line PC3 (data not shown). Therefore, the somatic mutations of the FOXP3 are functionally inactivated.
FOXP3 is a transcriptional regulator that functions by interacting with DNA in the nuclei (Zuo et al., 2007b). As the first step to understand the mechanism by which the mutations in FOXP3 affect its function, we tagged the FOXP3 protein with the green fluorescence protein (GFP) at the N-terminus and visualized its intracellular localization by confocal microscopy. As shown in Fig 3E, three out of four somatic mutants disrupted its translocation into nuclei. To substantiate these obserations, we isolated cytoplasm, nucleoplasm, and chromatin from PC3 transcted with vector control, WT and somatic FOXP3 mutant cDNA and determined distribution of FOXP3 by Western blot. As shown in Fig. 3F, while WT FOXP3 and the V79A mutant resides in both the nucleoplasm and the chromatin, the overwhelming majority of the proteins encoded by other three missense mutants are excluded from the nucleus. Since these three mutations had a more severe impact on the growth inhibition by FOXP3, preventing nuclear localization of FOXP3 appear to be the major mechanism to inactivate the tumor suppressor function. To confirm that disruption of nuclear localization is sufficient to abrogate growth inhibition by FOXP3, we used site-directed mutagenesis to inactivate the known nuclear localization sequence of FOXP3. As shown in supplemental Fig. S8, mutation in nuclear localization sequence was sufficient to abrogate growth inhibition by FOXP3.
To test the cell-intrinsic effect of FoxP3 deletion, we crossed the mice with a floxed FoxP3 locus (diagrammed in Fig. 4A) (Fontenot et al., 2005) to a transgenic line that expresses Cre gene under the probasin promoter (PB-Cre4) (Wu et al., 2001). Previous studies have demonstrated that this promoter causes prostate-specific deletion of Floxed genes starting in newborn mice (Wu et al., 2001). Using microdissected tissue samples of 8-12 week-old mice, we observed more than 80% deletion of the FoxP3 locus among the micro-dissected prostate epithelial tissue (Fig. 4B). The FoxP3 mRNA was reduced by more than 16-fold (Fig. 4C). The less profound reduction in DNA levels likely reflected the fact that our micro-dissected samples also contained non-epithelial cells that do not express FoxP3 (Chen et al., 2008). The reduction of FoxP3 protein is confirmed by Western blot using the lysates of total prostate (Fig. 4D) and immunohistochemistry staining (Fig. 4E). Consistent with the kinetics and levels of the PB-Cre4 transgene expression (Wu et al., 2001), the deletion is more complete in the ventral and lateral prostate lobes than in the anterior and dorsal lobes.
We took several approaches to determine the impact of prostate-specific deletion in the FoxP3 locus. First, we used magnetic resonance imaging (MRI) to monitor the prostate size in the live mice. As shown in Fig. 5A, 12-15 week-old mice with prostate-specific deletion of the FoxP3 locus had significant enlargement of the prostate. In comparison to WT, a 5-fold increase in the percentage of Ki67+ proliferating epithelial cells was observed in the mutant mice (Fig. 5B). Histological examination of the prostate revealed signs of prostate hyperplasia as early as 14-16 weeks in five out of six mutant mice. At 23-26-week old, 4/5 mutant mice but none of the six age-matched WT mice exhibited extensive hyperplasia (Fig. 5C and Fig. 6A). Early PIN was detectable at 23-26 weeks in a small fraction of ventral and dorsal prostate lobes with FoxP3 deletion, characterized by increased layers of epithelial cells and nuclear atypia (Fig. 5C and data not shown). By 43-60 weeks, all cKO mice examined had hyperplasia. Moreover, all but 1/9 cKO mice exhibited early PIN, including multiple layers of epithelial cells (Fig. 6B, C). The epithelial cells in this region had significantly enlarged nuclei, in comparison to either the single-layered epithelial cells in the same glandular structure (data not shown), or to those in the control mice (Fig. 6B, middle and lower panels). In most cases, the epithelia formed both papillary and tufting (Fig. 6C) patterns. Under high power, the luminal epithelial cells in these areas appeared transformed, as demonstrated by enlargement of nuclei, and more active nucleoli (Fig. 6C). In most mice, multiple PINs were found in the anterior, ventral and lateral prostate lobes, although the lesions are all focal in nature. All WT prostates have normal morphology through the course of the study. Therefore, targeted mutation of the FoxP3 gene in prostate tissue is sufficient to initiate the process of prostate cancer development.
MYC is over-expressed in 80% of the prostate cancer samples starting as early as benign hyperplasia (Fleming et al., 1986). However, the mechanism by which MYC transcription is increased remains unclear. We tested if MYC upregulation correlates with down-regulation of the FOXP3 transcripts. We measured the levels of the mRNA transcripts from microdissected cancerous and benigh tissues from 18 patients by real-time PCR. We normalized the transcript levels in cancer tissue against the normal epithelial from the same patients in order to avoid differential RNA degradation under different sample procurement conditions. We observed an increased MYC expression in 15/18 cases. Importantly, a significant correlation was observed between FOXP3 down-regulation and MYC over-expression among malignant tumor samples (Fig. 7A). When the levels of normal and cancer tissues were compared separately, a negative correlation between FOXP3 and MYC levels were found in cancer but not normal samples (Supplemental data Fig. S9). To test the relevance of this observation in human prostate cells, we tested the effect of the FOXP3 shRNA on MYC expression in early passage primary human prostate epithelial cells. Normal prostate cell culture grew slowly and expressed low levels of MYC. ShRNA silencing increased the growth rate of the culture (Supplemental Fig. S10). As shown in Fig. 7B, FOXP3 shRNA caused a major reduction in the expression of FOXP3 mRNA and protein. Correspondingly, the level of MYC transcripts and protein were significantly elevated by FOXP3 shRNA. To test whether the correlation could be causative and independent of cancer development in vivo, we microdissected normal WT and FoxP3-deleted prostate tissues and compare the Myc transcript levels. As shown in Fig 7C, prostate deletion of the FOXP3 locus caused more than a 4-fold increase in Myc mRNA. Moreover, the increased transcript levels were also reflected in elevation of the Myc protein in the nuclei (Fig. 7D). These data demonstrated that FoxP3 is a necessary repressor for the Myc locus.
To test whether ectopic expression of FOXP3 is sufficient to repress MYC, we transfected two prostate cancer cell lines with FOXP3. As shown in Fig. 7E, FOXP3 transfection almost completely abrogated the expression of MYC in both cell lines. In order to determine whether the growth inhibiton was mediated by repression of MYC, we co-transfected FOXP3 with MYC cDNA into Du145 cells. The cells were transfected with either pcDNA6-blasticidin vector or MYC cDNA (comprising of the entire coding region but no untranslated regions) and either the pEF1-G418 vector or FOXP3 cDNA. As shown in Fig. 7F, ectopic expression of MYC overcome FOXP3-mediated tumor suppression. These data demonstrated that MYC repression explains the growth inhibition of FOXP3, at least for established prostate cancer cell line.
To understand the mechanism by which FOXP3 represses MYC, we used ChIP to identify the site of FOXP3 binding in the MYC promoter. As shown in Fig. 8A, quantitative PCR analysis indicated that despite the abundance of forkhead binding sites, a strong binding of FOXP3 centered around -0.2 kb 5’ of the first transcription starting site (TSS-P1). To test the significance of this site for the repression, we carried out a deletional analysis to map the region that conveys susceptibility to FOXP3 repression. As shown in Fig. 8B, little repression by FOXP3 can be observed when the reporter was truncated before the forkead binding site at the -0.2 kb region (F1-F2). Strong inhibition was observed when the binding motif was included (F3-F5). Additional sequences did not increase the efficiency of repression. Sequence alignment revealed a conserved forkhead-binding site surrounding the promoter region with the highest ChIP signal (Fig. 8C). When the site was either deleted or mutated, the repression was completely abrogated (Fig. 8C). These data demonstrated that FOXP3 represses MYC promoter activity by interacting with the forkhead motif at the -0.2 kb 5’ of the MYC TSS.
To test whether somatic mutations of FOXP3 affect MYC repression, we transfected WT and mutant FOXP3 cDNA into the Du145 prostate cancer cell line in conjunction with the MYC promoter. Despite similar levels of FOXP3 protein, somatic mutations substantially reduced MYC repression (Fig. 8D). Since three of the four mutants failed to localize into the nuclei (Fig. 8E, F), we tested the remaining mutant for its ability to bind to the MYC promoter. As shown in Fig. 8E, the V79A mutation significantly reduced the binding of FOXP3 to the MYC promoter. Taken together, the data presented in this section demonstrates that FOXP3 represses MYC expression by binding the forkhead binding motif in the promoter. Somatic mutations uncovered in human prostate cancer abrogated the MYC repression by either preventing FOXP3’s nuclear localization or its binding to a cis-element in the MYC promoter.
Molecular pathogenesis of prostate cancer development includes both upregulation of oncogenes such as MYC and inactivation of tumor suppressor genes such as autosomal genes PTEN and NKX3.1. Although X-linked genes have been implicated by genetic epidemiology and linkage analysis, no X-linked tumor suppressors have been identified for prostate cancer. Our study described herein fills a major gap by identifying the FOXP3 gene as an X-linked tumor suppressor gene in males.
We and Rivera et al. have recently reported the involvement of X-linked tumor suppressor genes in female cancer patients, where the suppressor genes are functionally silenced by a programmed epigenetic event that occurred in all cells (X-inactivation) and another lesion such as gene deletion and/or mutation (Rivera et al., 2007; Zuo et al., 2007b). Here we described evidence that in the male, the FOXP3 locus is silenced by deletion and somatic mutation.
Immunohistochemical analysis demonstrated that the nuclear FOXP3 protein is absent in 68.5% of prostate cancer cases tested. Thus, in addition to deletion and mutation identified herein, additional mechanisms, such as epigenetic silencing or mutations outside the areas analyzed may also contribute to lack of FOXP3 expression. In contrast, 100% of benign prostate samples exhibited clear epithelial expression of nuclear FOXP3. The overwhelming difference strongly suggest a relationship between FOXP3 down-regulation and cancer development. This hypothesis has been confirmed by the impact of prostate-specific deletion of the FoxP3 locus. These data, together with strong growth inhibition and MYC repression by FOXP3, demonstrate that the FoxP3 gene plays a critical role in suppressing pathological transformation of the prostate. However, it should be pointed out that, similar to other tumor suppressors including Trp53 (Chen et al., 2005), NKX3.1 (Abdulkadir et al., 2002) (Kim et al., 2002), and Pten (Chen et al., 2005), we have not observed full spectrum of prostate cancer 60 weeks of observation. Since FoxP3 regulates Myc, and since the effect of transgenic Myc expression can be either hyperplasia (Zhang et al., 2000) or carcinoma with gene profiles similar to human prostate cancer (Ellwood-Yen et al., 2003), it is unclear whether extended observation time or additional factors are needed in order to achieve the full-spectrum of prostate cancer. The lack of full carcinogenesis over the observation period makes our model valuable for dissectiing contribution of other factors, such as inflammation and aging in the pathogenesis of prostate cancer. Nevertheless, the development of prostate hyperplasia and intra-epithelial neoplasia, a widely accepted precancerious lesion (Tomlins et al., 2006), demonstrates clearly that inactivation of FOXP3 in prostate cancer patients contribute to prostate cancer development. It is of interest to note that statistically significant down-regulation of FOXP3 occurred in the PIN lesion of human samples. Moreover, somatic mutations of FOXP3 were found in both PIN and cancerous lesions. Since deletion of FoxP3 in the mouse caused early PIN but so far no cancerous lesions, it is likely that FOXP3 inactivation is an early event in prostate carcinogenesis. Inactivation of FOXP3 will likely work in concert with additional genetic hits to cause prostate cancer.
FOXP3 exhibits strong growth inhibition of prostate cancer cell lines which is attenuated by somatic mutations in prostate cancer samples. Therefore, our genetic studies in mice and humans demonstrated a critical role for FOXP3 as a tumor suppressor gene for prostate cancer. The presence of a X-linked tumor suppressor gene as demonstrated in this study revealed a major exception for the generally accepted two-hit theory of tumor suppressor genes in cancer development (Knudson, 1971).
Since the FOXP3 resides near a region of a putative prostate cancer susceptibility locus (Gudmundsson et al., 2008), an interesting issue is whether identification of FOXP3 may help to explain reported X-linked genetic susceptibility to prostate cancer (Monroe et al., 1995). In our analysis of more than 100 probands with familiar prostate cancer, we have so far identified no germline mutation in the coding region of FOXP3 (data not shown). Given the early onset of lethal autoimmune diseases associated with inactivating mutation of FOXP3 (Bennett et al., 2001; Chatila et al., 2000; Wildin et al., 2001), it is unlikely that inactivating germline mutation of the FOXP3 gene can be found in prostate cancer patients. Additional analyses are needed to determine whether the FOXP3 polymorphisms affect susceptibility to prostate cancer.
MYC is arguably the most up-regulated oncogene in human cancer, as over-expression has been observed in as many as 30% of all human cancer cases. Our results show freqenct over-expression of MYC mRNA in prostate cancers in comparison to adjacent normal epithelia. Although it is clear that a mutation of the APC gene resulted in up-regulation of MYC and MYC is essential for the development of colon cancer associated with APC mutation (He et al., 1998; Sansom et al., 2007), APC mutation is rare in prostate cancers thus cannot explain a MYC over-expression in prostate cancer. Likewise, although MYC over-expression has been observed in all stages of prostate cancer, a low level of gene amplification is observed only in late stages of prostate cancer (Jenkins et al., 1997; Latil et al., 2000). A similar disparity can be observed in other tumor types, including breast and lung cancer (Takahashi et al., 1989) (Bieche et al., 1999).
Our data presented here provide strong evidence that FOXP3-mediated repression of MYC is necessary to control MYC levels in normal prostate epithelial cells and explains much of the widespread over-expression of MYC in prostate cancer. Since transgenic expression of Myc resulted in development of prostate cancer with similar genetic profiles to human prostate cancer (Ellwood-Yen et al., 2003), disruption of the MYC regulator FOXP3, as documented here, likely plays a critical role for molecular pathogenesis of human prostate cancer. Nevertheless, using expression array analysis, we have uncovered a large array of genes that are either up- or down-regulated by ectopic expression of FOXP3. Notably, the genes involved in cancer are the most affected group (Figure S11). Therefore, much like what was observed in breast cancer, FOXP3 will likely suppress development of prostate cancer by targeting multiple genes, including both tumor suppressor and oncogenes (Liu et al., 2009; Zuo et al., 2007a; Zuo et al., 2007b).
Three sources of prostate tissue samples were used for this study. Frozen tissues were obtained from a Prostate Cancer Tissue Bank of the Ohio State University with no patient identification information. Tissue microarray samples came from the University of Michigan, Biomax (US Biomax, Inc., Rockville, MD) and and Target Biotech (Target Biotech, Inc., Thurmont, MD). All human studies have been approved by the Institutional Review Board of the Ohio State University and University of Michigan. Mouse prostate samples were fixed in 10% formalin. The sections were stained with H&E and diagnosed by two pathologists based on the criteria set forth by Park et al (Park et al., 2002)
Transgenic mice PB-Cre4 expressing the Cre cDNA under the control of the probasin promoter have been described (Wu et al., 2001) and were obtained from the National Cancer Institute Mouse Model Deposit. Mice with Floxed FoxP3 locus (Fontenot et al., 2005) were provided by Dr. Alexander Rudensky. Male PB-Cre4+/- mice were crossed to female FoxP3flox/flox or flox/+ mice. The male F1 mice of given genotypes were used in the study. The pathological evaluation was carried out according the Bar Harbor meeting guideline (Shappell et al., 2004). All animal studies have been approved by the University of Michigan Animal User and Carer committee.
Human Prostate Epithelial Cell (HPEC) was purchased from Lonza Group Ltd (Switzerland) and cultured with medium from the same vendor. Early passage HPEC were infected with retrovirus expressing either control shRNA or FOXP3 shRNA vector. Two FOXP3-short hairpin RNA (shRNA) constructs are FOXP3-993-shRNA and FOXP3-1355-shRNA (GenBank accession number, NM_014009). Oligonucleotides encoding small interfering RNA (siRNA) directed against FOXP3 are 5’- GCTTCATCTGTGGCATCATCC -3’ for FOXP3-993-shRNA (993 to 1013 nucleotides from TSS) and 5’- GAGTCTGCACAAGTGCTTTGT -3’ for FOXP3-1355-shRNA (1355 to 1375 from TSS). The selected shRNA oligonucleotides were cloned into pSIREN-RetroQ vectors (Clontech, Mountain View, CA) to generate siRNA according to manufacturer’s protocol. Prostate cancer cell lines Du145 and PC3 were obtained from the American Type Culture Collection (ATCC, Rockville, Maryland). Antibodies specific for the following targets were used for the study: cMyc (Santa Cruz Biotechnology, Santa Cruz, CA; Cat #: sc-40), FOXP3 (Abcam, ab20034 for IHC and ab450 for Western blot, Cambridge, MA), hFOXY (eBioscience, Cat#: 14-5779-82 for Western blot in HPEC), anti-V5 (Invitrogen, SKU# R960-25), Ki67 (Dako Cytomation, code No. M7249, Carpenteria, CA), β-actin (Sigma, Cat#: A5441, St. Louis, MO), and anti-IgG (Santa Cruz Biotechnology, Santa Cruz, CA).
Since FOXP3 is expressed at lower levels in epithelial cells than in the regulatory T cells, the conditions typically used for detecting Treg in tumor samples do not give reproducible staining in the epithelial cells. In order to facilitate replication of the current studies, we have screened commercially available anti-FOXP3 antibodies and established staining conditions that give strong staining of FOXP3 in human and mouse epithelial cells.
The normal and cancer prostate frozen samples were partially thawed at room temperature and then immersed in 10% formalin for 8-12 hours and embedded in paraffin. Antigens were retrieved by microwave in 1x target retrieval buffer (Dako) for 12 min. For TMA samples, antigens were retrieved at 37° C overnight after the microwave antigen retrieval treatment. ABC detection system was used for immunostaining according to the manufacturer’s protocol (Vectastain Elite ABC, Burlingame, CA). The incubation time for primary antibody FOXP3 (Abcam 236A/E7, 1:100), FoxP3 (BioLegend Poly6238, 1:100), c-Myc (Santa Crruz Biotechnology 9E10, 1:200), c-Myc (Abcam ab39688, 1:100), and Ki67 (Dako TEC-3, 1:100) was overnight at 4°C. After incubation with primary antibody, staining was followed by ABC detection system using biotinylated anti-mouse/rabbit/rat IgG. AEC was used as chromogen. The slides were counterstained with hematoxylin and mounted in xylene mounting medium for examination.
The distribution of samples for each group was evaluated using a One-Sample Kolmogorov-Smirnov test. In the samples with normal distributions, we compared the means of the dependent variable using a Paired-Samples t test and means of the independent variable using an Independent-Samples t test between two groups. In the samples with non-normal distribution, we compared the means of the independent variable between two groups using a Mann-Whitney U test. Chi-square test was used to compare the relationship between the expression of FOXP3 and MYC among patients. The relationship between the levels of gene expression was estimated using a linear regression model. All data were entered into an access database and analyzed using the Excel 2000 and SPSS (version 10.0; SPSS, Inc.) software.
The raw data for supplemental Fig. S11 have been submitted to ArrayExpress (Accession No. E-MTAB-108).
We thank Dr. Alexander Rudensky for the FoxP3flox/flox mice. Dr. Hong Wu, Jianti Huang from UCLA, and Dr. George V. Thomas from Institute of Cancer Research and Royal Marsden Hospital, UK for pathological evaluation of mouse prostate. Drs. Eric Fearon, Steve Gruber, Michael Sabel and Yuan Zhu for valuable discussion and/or critical reading of the manuscript and Ms Darla Kroft for editorial assistance. This study is supported by grants from NIH, Department of Defense, Cancer Research Institute in New York, American Cancer Society and by a gift made the University of Michigan Cancer Center.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.