Men1 ^+/- mice carrying an inactivated Men1 allele were generated as previously described 5 . These mutant mice with mixed C57BL6/129-Sv genetic background give rise to multiple endocrine lesions mimicking the human MEN1 tumour spectra as previously described 6 . Littermate Men1 ^+/- and Men1 ^+/+ male mice were used for the analyses to limit the bias possibly originating from the mixed genetic background. All animal experiments were conducted in accordance with accepted standards of animal care and were approved by the International Agency for Research on Cancer's Animal Care and Use Committee.

PCR genotyping analyses were performed to determine the presence of the wild-type and mutant Men1 alleles as previously described 6 . For LOH analysis, menin-immunostained prostate sections were used to locate menin-negative lesions. Genomic DNA was then extracted from corresponding microdissected paraffin-embedded sections as previously described 6 . The wild-type (+) and mutant (-) Men1 alleles were first amplified from 1 μg of genomic DNA by PCR, carried out as described for genotyping 6 , using the following three primers: 2f0, 5-CTTACCTCTTCTCATGTCTG; 2r0, 5-CTCAGTACATTGCACGGAGA; and tk1.1, 5-GCGTTGCGTGGGGTCAG. Southern blots were then performed by standard methods on these PCR products using a probe for each Men1 allele generated by amplification of genomic DNA from a Men1 heterozygous mutant mouse with the PCR described above. Autoradiography films were scanned with the EC3 imaging system (UVP Bioimaging Systems), and the intensity of each band was quantified with Visionworks LS software. The ratio between the intensities of the + and - Men1 alleles (+/- ratio) was then calculated to evaluate the loss of the wild-type Men1 allele.

Prostate glands and adjacent tissues were collected from wild-type and heterozygous Men1 mutant mice, fixed in 4% buffered formalin for at least 24 h at room temperature, and processed for paraffin-embedding, sectioning (3 μm) and haematoxylin and eosin staining. Histopathological examination of prostate samples was performed independently by three pathologists (B Fontanière, L Frappart and WM Tong), including a mouse model specialist (WM Tong), on a blinded basis according to the criteria recommended by the consensus report from the Bar Harbor meeting of the mouse models of human cancer consortium (MMHCC) prostate pathology committee 13 . The WHO Classification of Tumours of the Urinary System and Male Genital Organs (2004) was also used. More than 3 prostate sections were examined for each mouse and lobe origin was distinguished during the scoring process. The final diagnosis was then reached by a common decision.

Immunohistochemical (IHC) staining was performed on prostate serial sections as previously described 6 , using antibodies against menin (BL342, Bethyl Laboratories, Montgomery, TX, USA, 1:6000), androgen receptor (C-19, Santa Cruz Biotech, Santa Cruz, CA, USA, 1:4000) and CDKN1B (F-8, Santa Cruz Biotech, 1:2000). In all these analyses, a control without primary antibody was systematically included to rule out non-specific staining due to secondary antibody reactivity (data not shown). Immunofluorescence (IF) double staining was performed on paraffin-embedded prostate sections. Briefly, after endogenous peroxidase inactivation, antigen retrieval and antibody diluent treatment (DAKO, Capinteria, CA, USA), sections were incubated overnight with primary antibody against menin (BL342, 1:6000). After secondary antibody incubation (anti-rabbit, Vector Lab, 1:200), the signal was amplified using the ABC Vectastain elite kit (Vector Lab) and the TSA Cyanine 3 kit (PerkinElmer, Waltham, MA, USA, 1/100). Slides were saturated once more in DAKO antibody diluent and then incubated overnight with primary antibodies against p63 (Ab-1, Oncogene Research products, San Diego, CA, USA, 1:2000) or CDKN1B (Santa Cruz Biotech, 1:2000), followed by incubation with secondary antibody (anti-mouse, Vector Lab, 1:200). The signals were revealed by the ABC kit and TSA Cyanine 5 (PerkinElmer, 1/100). Slides were mounted in IF Vectashield DAPI-containing mounting medium (Vector Lab), and the immunofluorescence signal was visualised with a confocal microscope (LEICA TCS SP2).

To determine the effect of Men1 inactivation on prostate cancer development in mice, we followed a cohort of 47 male mutant mice (Men1 ^+/-) and 23 wild-type (Men1 ^+/+) age-matched littermate mice from 18 to 26 months of age, based on a previous study that showed no prostate cancer in younger mice 6 .

Histopathologic analysis of the prostate glands from these mice was performed in accordance with the criteria defined by the MMHCC 13 . A comparison between age-matched Men1 wild-type and mutant mice revealed notable histological differences, showing a marked increase in cancer incidence in Men1 ^+/- prostates (Table 1) and a significantly decreased proportion of mutant mice with normal prostate histology. The results showed 30.4% of Men1 wild-type prostates with normal histology while only 6.4% of prostate glands from mutant mice were free of lesions (p = 0.0266, Fisher's two-tailed exact test). Normal prostate tissue presented as a single layer of secretory epithelial cells lined by a well-defined outer basal cell layer, with uniform small nuclei containing inconspicuous or small nucleoli. The epithelial hyperplasia diagnosed in this study appeared as increased numbers of epithelial cells, with or without atypia, associated with increased gland size. The intraluminal proliferation of markedly atypical epithelial cells, with tufted, cribriform or micropapillary growth pattern, was recognised histologically as mouse prostatic intraepithelial neoplasia (mPIN) (Figure 1B). Carcinomas were characterised by frequent mitotic figures, apoptotic debris and cytologic atypia, such as amphophilic cytoplasm, increased nuclear-cytoplasmic ratio, hyperchromasia, prominent and multiple nucleoli, chromatin clumping and pleomorphism (Figure 1C, D, E).

Hyperplastic and mPIN lesions, mainly presenting in the lateral and anterior lobes of prostatic glands, were observed with similar frequency in mutant and wild-type aged mice (Table 1, 43.5% in Men1 ^+/+ versus 46.8% in Men1 ^+/- mice for hyperplasia and 26.1% in Men1 ^+/+ versus 34% in Men1 ^+/- mice for mPIN), and a few of them were also found in dorsal and ventral lobes. However, the incidence of mPIN in mutant mice was underestimated because four mPIN lesions observed in mutant mice with in situ carcinomas were not counted, since only the most severe lesion was taken into account for each mouse. The real incidence of mPIN in Men1 ^+/- mice should thus be 42.5%. The occurrence of both hyperplasia and mPIN lesions has already been reported in some wild-type mouse strains 14 .

More importantly, our analysis revealed that, of 47 heterozygous Men1 mutant mice, six developed prostate cancers (12.8%). No prostate carcinoma was ever found in age-matched Men1 ^+/+ littermates (0/23). Among the prostate cancers observed in Men1 ^+/- mice, four were identified as in situ prostate carcinomas presenting as microscopic lesions (one in dorsal prostate, shown in Figure 1C, two in lateral prostate, and one in anterior prostate), one papillary adenocarcinoma in the anterior prostate (Figure 1D) and one differentiated invasive adenocarcinoma in the lateral prostate (Figure 1E, F), with the latter two visible on gross examination. Invasive tissue was found in epithelial areas lacking an intact basal cell layer, with cancerous cells invading the surrounding stroma (Figure 1F). Among the observed prostatic lesions, none displayed histopathological features typical of neuroendocrine differentiation (solid or sheet-like proliferation of closely spaced oval or spindle cells with scant cytoplasm and hyperchromatic nuclei, with areas of rosette formation) 13 .

Taken together, our data revealed the development of prostate carcinomas in a small proportion of aged heterozygous Men1 mutant mice.

Menin expression was assessed in the different prostatic lobes from Men1 wild-type mice using either IHC for menin or double IF with antibodies against menin and p63, a basal epithelial cell marker. IHC analysis revealed that menin expression, which is mainly nuclear but can also be detected in the cytoplasm, was found in all prostatic lobes from Men1 wild-type mice in both luminal and basal prostatic epithelial cells (Figure 2A) but not in interstitial cells. Double IF staining of menin and p63 confirmed this result (Figure 3A-D). To evaluate the association between the occurrence of prostatic lesions in heterozygous Men1 mice and inactivation of the Men1 gene, we examined menin expression by IHC in the corresponding lesions. We found that menin expression was readily detectable by IHC in all four tested mPIN lesions from WT mice (data not shown), but clearly reduced in two of four mPIN lesions from mutant mice (Figure 2B, F). The data suggest that, although a large proportion of mPINs observed in mutant mice occur independently of Men1 inactivation, as those seen in wild-type mice, some may still be related to menin loss. In contrast to mPIN lesions, no menin expression was detectable in the six prostate cancers found in Men1 ^+/- mice (Figure 2C-E and 2G-I), suggesting that prostate cancer in these mice was likely caused by menin inactivation, rather than the secondary effects of MEN1 pathology, such as gene expression and/or hormone disturbance in endocrine tissues. In two of the prostate cancers shown in Figure 2 (panels D, H and E, I), a small number of cells remain positive for menin expression. We reasoned that these cells were in fact normal basal epithelial cells that remained in cancer lesions, which was confirmed by further analyses (see below).

To further study the related mechanisms leading to menin inactivation, we tested the presence of LOH in mPIN from Men1 ^+/- mice displaying reduced menin expression and in menin-negative carcinoma lesions detected by IHC analysis. To this end, the tested tissues were scraped from the corresponding paraffin-embedded slides, and genomic DNA was extracted from each lesion. Genotyping analysis demonstrated that partial LOH was indeed found in three of five analysed precancerous and cancerous lesions (Figure 2J). The quantification of amplified products corresponding to Men1⁺ and Men1 ^- alleles from the tail of a Men1 ^+/- mouse was used as a control, and an identical intensity of both alleles was achieved, thus giving a +/- ratio equal to one. The intensity of the wild-type allele (+), but not the mutant allele (-), was reduced in two mPINs (PIN1 and 2, both showing reduction of menin expression by IHC analysis) and in an adenocarcinoma lesion (ADC) from Men1 ^+/- mice, as demonstrated by the lower +/- ratios for these lesions when compared with the control (Figure 2J). We assume that the residual wild-type allele, amplified from the tested lesions, was derived mainly from the remaining normal basal cells within the lesions, which were shown to retain menin expression. The result suggests that LOH of the remaining wild-type Men1 allele may be one of the mechanisms leading to menin inactivation in the prostate cancers developed in this model, although we cannot exclude that other mechanisms may also be involved.

Although a loss of menin expression and a partial LOH were found in some mPINs from Men1 mutant mice, these occurred at similar frequencies in wild-type and mutant mice. The present study therefore focused on the analysis of the prostate cancers arising in Men1 mutant mice. To further characterise the observed cancer lesions, expression of the basal epithelial marker p63 was analysed using double IF staining, since the diagnosis of cancerous lesions can often be based on the disruption of the basal cell layer indicated by the absence or greatly reduced number of p63-positive basal cells 15 . Menin-negative cancerous cells were negatively stained with basal cell marker p63 (Figure 3E-H). In addition, normal p63-positive basal cells, which retained menin expression, were greatly reduced in number and scattered in the cancerous lesions, showing a disruption of the basal cell layer (Figure 3H, white arrowhead). The data were thus consistent with our histopathologic findings and indicated that menin inactivation did not occur in basal cells, confirming what we had speculated on the basis of the results obtained by menin IHC.

Prostatic cells express the androgen receptor (AR) and respond to the androgen pathway that regulates prostate growth, apoptosis and differentiation. Nearly 70% of clinical prostate cancer cases are AR positive. It is generally considered that AR expression in prostate cancer cells reflects not only the differentiation status of the cells, but also their sensitivity to anti-AR treatment. We therefore evaluated AR expression using IHC in different prostate tissue samples. Normal prostate tissue from Men1 ^+/+ mice showed strong AR nuclear staining in both luminal and basal epithelial cells, as well as in many stromal cells (Figure 3I, L), as previously described in the literature 16 17 18 . AR was clearly expressed in an in situ carcinoma (Figure 3J, M) and an adenocarcinoma (Figure 3K, N) and showed a heterogeneous expression pattern in these lesions in comparison with the normal control prostate (Figure 3I, L), with some cells showing absent or very low AR staining. Quantitation of epithelial cell nuclei displaying either no AR expression or very low staining revealed a significant increase in the number of cells with altered AR expression in prostate carcinomas from Men1 mutant mice compared with normal Men1 ^+/+ prostate glands (Figure 3O).

The data demonstrated the disruption of the basal cell layer in these lesions and suggested a possible association between menin inactivation and the deregulation of AR expression.

CDKN1B, an inhibitor of cyclin-dependent kinases, is a transcriptional target of the menin protein 11 12 . The involvement of its downregulation in the tumorigenesis of prostate cells has also been well documented in different mouse models and in prostate cancer patients 19 20 . We therefore analysed the expression of CDKN1B using IHC in serial sections of prostate carcinomas from Men1 ^+/- mice. Strong, predominantly nuclear CDKN1B expression was detected in virtually all luminal and basal Men1 ^+/+ prostate cells (Figure 4A, D), consistent with previous reports 21 22 . CDKN1B nuclear expression was clearly reduced in an in situ carcinoma (Figure 4B, E) and almost absent in an adenocarcinoma (Figure 4C, F) from Men1 ^+/- mice. Furthermore, double IF staining with menin and CDKN1B antibodies showed that the reduced expression of CDKN1B in prostate carcinomas correlated well with the loss of menin expression (Figure 4K-N), indicating that CDKN1B expression is indeed down-regulated in menin-negative prostatic cells.