Homozygous loss of BHD causes early embryonic lethality and kidney tumor development with activation of mTORC1 and mTORC2

Role of BHD during Early Embryogenesis.

We have analyzed mouse BHD mRNA expression levels by qRT-PCR in wild-type embryos and adult tissues (Fig. S1). We detected consistent BHD mRNA expression from E8.5 to E12.5 with 4-fold elevation at E19 and high expression in adult heart, pancreas, and prostate with moderate expression in adult brain, kidney, liver, and lung. BHD mRNA expression was further analyzed during early embryogenesis by whole mount in situ hybridization (Fig. S2). BHD mRNA was expressed consistently throughout embryogenesis. At E5.5, BHD expression was restricted to extraembryonic tissues; however, by E6.5, BHD was expressed in both embryonic and extraembryonic tissues. We saw strong expression in certain tissues including neural ectoderm, headfold, and limb buds, but the signal was relatively weak in the surrounding endoderm and heart.

Next we evaluated BHD homozygous knockout (BHD^d/d) embryos from intercrosses of BHD heterozygous knockout (BHD^d/+) mice. BHD^d/+ mice appeared normal at birth and developed normally, but no BHD^d/d mice were born in 75 neonates indicating embryonic lethality (Fig. 1A). No BHD^d/d embryos were found at E9.5. BHD^d/d embryos were found before E8.5 with lower frequency (9.3%, 15/161) than expected (25%). There were many empty deciduas suggestive of early embryonic death and resorption (21.8%, 26/119 at E5.5 and E6.5). The morphologies of all BHD^d/d embryos were abnormal with one exception at E5.5. By gross appearance BHD^d/d embryos (Fig. 1 B and D) were thinner or smaller than BHD^d/+ (Fig. 1 C and E) or BHD^+/+ embryos with occasional bleeding. There were no histological abnormalities in BHD^+/+ or BHD^d/+ embryos (Fig. 1 F and G). Most BHD^d/d embryos lacked an organized cell layer with only a small cell mass (Fig. 1I). We could occasionally see BHD^d/d embryos consisting of two types of cells, clear visceral endoderm-like cells and ectoderm cells (Fig. 1H). Immunohistochemical staining with the visceral endoderm (VE) marker DAB2 confirmed that the outer cell layer was VE (Fig. 1 J–M). Interestingly, VE of the BHD^d/d embryos showed distorted enlarged cytoplasm and a disorganized structure (Fig. 1 K and M) instead of a cuboidal epithelial layer as seen in the BHD^d/+ embryos (Fig. 1 J and L). Also the nuclei of the VE cells were disorientated in BHD^d/d embryos (Fig. 1H'), whereas nuclei were aligned along the basal membrane in an orderly fashion in BHD^+/+ embryos (Fig. 1F'). BHD^d/d embryos also displayed a disorganized ectoderm structure and did not show cavitation or a polarized epithelial cell layer (Fig. 1 H and K).

Spontaneous Kidney Tumor Development in BHD^d/+ Mice with Loss of Heterozygosity.

BHD^d/+ mice spontaneously developed cysts, complex cysts and solid tumors in their kidneys after the age of 10 months (Fig. 2 A and B). Two of 23 BHD^+/+ mice displayed small, isolated simple cysts in their kidneys. The median age of kidney-lesion-free survival for BHD^d/+ mice was 23 months (n = 65), compared with an undefined kidney-lesion-free survival for BHD^+/+ littermate controls (n = 28, P < 0.0001) (Fig. 2C). No solid kidney tumors developed in BHD^+/+ mice (BHD^d/+ median tumor-free survival = 25 months, n = 65; BHD^+/+ median tumor-free survival = undefined, n = 28; P = 0.0026) (Fig. 2D). The number of kidney lesions in BHD^d/+ mice between the ages of 20 and 25 months is shown in Fig. 2E (n = 35; no. cysts per animal, mean = 3.43, SD = 3.13; no. mixed lesions per animal, mean = 0.51, SD = 0.85; no. solid tumors per animal, mean = 0.51, SD = 0.82). Histological examination showed that kidney cysts that develop in BHD^d/+ mice were lined by hyperplastic cells with enlarged cytoplasm and nuclei (Fig. 3A). The cysts found in BHD^+/+ mice were lined by flat cyst cells characteristic of simple cysts, which were distinct from the hyperplastic cysts that developed in BHD^d/+ mice. Complex cysts, defined as cysts with some structure or solid portions in the lumen, were observed. These complex cysts showed papillary protrusions into the lumen (Fig. 3B) as well as regions containing solid tumor (Fig. 3C). Interestingly the solid tumors displayed histologies similar to human kidney tumors that develop in BHD patients, including clear cell (Fig. 3D), oncocytic hybrid consisting of a mixture of chromophobe renal carcinoma and oncocytic cells (Fig. 3E), and oncocytoma (Fig. 3F). Southern blot analysis of tumors that formed in BHD^d/+ mice showed a severely reduced wild-type BHD allele signal compared to adjacent normal kidney tissue, supporting LOH at the BHD locus (Fig. 3G). Western blot analysis showed only very weak FLCN expression in the tumors compared with adjacent BHD^d/+ kidney or wild-type kidney tissue (Fig. 3H), thus confirming the inactivation of both BHD alleles. Loss of endogenous FLCN expression was also confirmed by immunofluorescent staining using the Duolink in situ proximity ligation assay (PLA), which enables visualization of endogenous FLCN with two different antibodies (Fig. 3 I and J). Importantly all five kidney tumors that were analyzed from BHD^d/+ mice showed very low FLCN expression compared to normal mouse kidneys. These results strongly support the premise that kidney tumors, which developed in the BHD^d/+ mouse model, were produced as a consequence of loss of FLCN function.

Activation of the PI3K-AKT-mTOR Signaling Pathway in Kidney Tumors from BHD^d/+ Mice and Human BHD Patients.

Total AKT protein levels, AKT1, and AKT2 as well as total AKT, were dramatically elevated in kidney tumors compared with normal kidney tissue (Fig. 4A). Levels of AKT phosphorylation were also dramatically elevated at both the PDK1 phosphorylation site (Thr308) and the mTORC2 phosphorylation site (Ser473) (15) (Fig. 4 A and B). AKT mRNA levels were measured by qRT-PCR in the same samples (Fig. 4C). Contrary to the Western blot results, mRNA levels of AKT1 and AKT2 were not significantly different between tumors and normal kidneys. Increased phosphorylation of downstream effectors of AKT signaling would support AKT activation in kidney tumors that developed in BHD^d/+ mice. Indeed, p-GSK3, p-FOXO1, and p-FOXO3a were elevated in tumors compared with normal kidneys (Fig. 4 D–F). We also saw Cyclin D1 elevation, which might be a consequence of GSK3 phosphorylation (Fig. 4D). mTOR phosphorylation on serine 2448, which reflects mTORC1 activation (16), was higher in tumors than in normal kidneys. Furthermore, mTOR phosphorylation on serine 2481, indicating mTORC2 activation (17), was also higher in tumors compared with normal kidney tissue (Fig. 5A). Interestingly Rictor was more highly expressed in tumors from BHD^d/+ mice, which may lead to mTORC2 activation and result in higher AKT phosphorylation on serine 473. Levels of phospho(p)-S6 kinase (Thr421/Ser424) and phospho S6 ribosomal protein (Ser240/244), readouts of mTORC1 activation, were higher in tumors than in normal kidney tissue, although total protein levels of S6K and S6R were also elevated (Fig. 5A). Consistent with the Western blot results, we were able to see stronger p-mTOR (Ser2448) and p-S6R (Ser240/244) staining in cells lining the cysts and in tumors from BHD^d/+ mice when compared to adjacent normal kidney (Fig. 5 B–G).

Finally we evaluated these protein levels in human BHD tumors and normal kidneys. Western blotting and immunofluorescent staining showed higher levels of p-AKT(Ser473) in kidney tumors compared to normal kidney controls (Fig. 6 A, E, and F). We quantified p-AKT(Ser473), p-S6K(Thr421/Ser424), and p-S6R(Ser240/244) using a Meso Scale Discovery multiplex microtiter plate assay, a quantitative assay system using a combination of electrochemiluminescence detection and patterned arrays (Fig. 6 B–D). The protein levels of p-AKT (Ser473) were more than 30 times higher in human BHD tumors than in normal kidneys. Levels of p-S6K (Thr421/Ser424) protein and p-S6R protein were also higher in the BHD kidney tumors (Fig. 6 C and D). We compared the genotype and phenotype of the BHD patients whose kidney tumors were evaluated in this study (Fig. 6G). The kidney tumors were surgical specimens from three different female patients with different types of germline BHD mutations: missense, splicing defect and frameshift. All of the patients with different BHD mutations showed the classic triad of BHD phenotypic features- fibrofolliculomas, pulmonary cysts and renal tumors (Fig. S3). Importantly, the molecular phenotype was similar among the tumors derived from these three different types of BHD alterations, underscoring the importance of PI3K-AKT-mTOR activation for kidney tumorigenesis in BHD patients.

Development of BHD Knockout Mouse Model.

The BHD heterozygous knockout mice were generated as previously described (13). Details of the targeting strategy are described in the SI Methods (Fig. S5). All mice which were used in these experiments were housed in the National Cancer Institute (NCI)-Frederick animal facility according to the NCI-Frederick Animal Care and Use Committee guidelines.

PCR-Based BHD Genotyping.

Mouse genomic DNA was isolated from tails (weaned neonates), yolk sacs (E8.5 or later), and whole embryos (E7.5 or earlier). Primers and details are in the SI Methods.

Quantitative Real Time-PCR.

The qRT-PCR for BHD, AKT1, and AKT2 was performed as described in the SI Methods.

Whole Mount in Situ Hybridization.

The embryos were collected from wild-type C57BL/6 mice intercrossed at different stages of gestation and processed for whole mount in situ hybridization as previously described (25). Negative staining was confirmed using a sense probe with wild-type embryos. Probe information is in the SI Methods.

Histological and Immunohistochemical Analysis.

Embryo sections were stained with hematoxylin and eosin (H&E) or with DAB2 antibody (BD Bioscience) for immunohistochemical evaluation performed as previously described (26). The slides were read by at least three persons, including two pathologists (M.J.M. and D.C.H.).

Western Blotting and Antibodies.

Immunoblotting was performed as described in the SI Methods.

Tissue Genotyping by Southern Blotting.

Nonradioactive Southern blotting was performed with DIG OMNI System for PCR Probes according to the manufacturer's protocol (Roche). Probe information is in the SI Methods.

Endogenous FLCN Detection by Duolink System.

Duolink in situ PLA was performed per manufacturer's instruction (OLink Biosciences). For further information, see the SI Methods.

Human Sample Preparation and MSD Analysis.

Renal tumors were obtained from BHD patients surgically treated at the Urologic Oncology Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD with patient permission under a National Institutes of Health Institutional Review Board (IRB)-approved protocol #97-C-0147. All patients signed informed consent. MSD (Meso Scale Discovery) 96-well multispot AKT signaling pathway (phospho-AKT [Ser-473]/total GSK3β/phospho-S6K [Thr-421/Ser-424]), and phospho-S6R (Ser-240/244) assays were carried out according to the manufacturer's protocol. For further information, see the SI Methods.

Immunofluorescence Imaging of the AKT-mTOR Pathway.

Immunostaining for p-AKT, p-mTOR, and p-S6R was performed on frozen sections as described in the SI Methods.