Introduction
Von-Hippel-Lindau (VHL) gene inactivation results in a cancer syndrome characterized by tumors at various sites, including vascular tumors of the retinas and central nervous system, renal cancers, adrenal pheochromocytomas, and pancreatic tumors. The VHL protein (pVHL) is an E3 ubiquitin ligase that regulates hypoxia inducible factors alpha subunits (HIF1α and HIF2α) through targeted ubiquitination and proteasomal degradation under normoxic conditions [
1‐
3]. Under hypoxic conditions pVHL levels are decreased [
4] and HIFα subunits accumulate. The activated HIF transcriptional response results in expression of hundreds of genes whose products are involved in metabolism, cellular proliferation, and angiogenesis [
5]. Although conventional deletion of VHL is embryonic lethal in the mouse due to placental vascular defects [
6], conditional VHL deletions have been shown to induce hepatic vascular lesions [
7‐
10], renal cysts [
11] and pancreatic lesions [
12].
Previously, we have shown that the androgen-responsive tumor suppressor ELL-associated factor 2 (EAF2) could block HIF-dependent angiogenesis by binding and stabilizing pVHL [
13]. EAF2
−/− mice had reduced levels of pVHL in the testes; and mouse embryonic fibroblasts derived from EAF2
−/− mice had an increased activity and level of HIF1α [
13]. Furthermore, EAF2
−/− mice displayed an increase in blood vessel formation in the liver at 3 months of age [
14]. These studies suggest that EAF2 may cooperate with VHL in the regulation of blood vessel formation and growth, and concurrent loss of function of VHL and EAF2 could contribute to a pro-angiogenic phenotype.
In order to further characterize the role of EAF2 and VHL in angiogenesis, we examined the effects of EAF2 loss and VHL heterozygosity in murine prostate and liver. Here we report that EAF2−/−VHL+/− mice had an increased incidence in hepatic vascular lesions as well as increased liver and prostate vascularity compared to wild-type, EAF2−/− and VHL+/− mice. Increased vascularity in the liver was characterized by an increase in HIF1α and VEGF immunoreactivity, particularly in hepatocytes near vascular lesions. In the prostates of EAF2−/−VHL+/− mice there was an increased incidence of PIN, stromal inflammation, fibrosis and smooth muscle proliferation compared to wild-type, EAF2−/− and VHL+/− mice.
Materials and methods
Generation of strain-specific VHL deletion mice
VHL
+/− mice [
6] were generated by backcrossing to the BALB/c, FVB/N and C57BL/6 J strains (NCI Animal Production Facility, Frederick, MD, USA) for >12 generations to generate VHL
+/− mice with a pure background. Wild-type and VHL
+/− mice were maintained identically, under approval by the Institutional Animal Care and Use Committee of the Louisiana State University Health Sciences Center, New Orleans. Genotyping was determined by PCR analysis of mouse tail genomic DNA as described previously [
6]. Mice were sacrificed at 12–15 mos of age.
Generation of EAF2 and VHL deletion mice on a C57Bl/6 background
Preparation of mice with specific deletion of EAF2 gene using HM1 embryonic stem cells has been described previously [
14,
15]. EAF2
+/− mice were subsequently backcrossed to the C57BL/6 J strain (Jackson Laboratory, Bar Harbor, ME, USA) for >12 generations to generate EAF2
−/− mice with a pure C57BL/6 J background. EAF2
−/− VHL
+/− mice were generated from heterozygous intercrosses of EAF2
+/− and VHL
+/− mice (obtained from Dr. Laura Schmidt [
8]) and subsequently from homozygous EAF2
−/−VHL
+/− intercrosses. Experimental cohorts were wild type, EAF2
−/−, VHL
+/−, and EAF2
−/−VHL
+/− male littermates. All mice were on a C57BL/6 background and were maintained identically, under approval by the Institutional Animal Care and Use Committee of the University of Pittsburgh. Genotyping was determined by PCR analysis of mouse tail genomic DNA as described previously [
15]. Mice were sacrificed at 12–15 mos and 20–24 mos of age. Organs were cleaned of excess fat and membrane with phosphate-buffered saline and mass was determined by weighing after blotting with filtration paper to remove excess liquid. Prostate mass was determined as absolute mass as well as a ratio of organ to body mass to correct for body weight differences as described [
16]. Samples were fixed in 10% formalin for at least 24 h, then embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin and eosin.
Immunohistochemistry
Immunohistochemical stains were performed on five-micron sections of paraffin blocks. Briefly, the sections of all groups were deparaffinized and rehydrated through a graded series of ethanol. Heat induced epitope retrieval was performed using 10 mmol/L of citrate buffer (pH 6), followed by rinsing in TBS buffer for 5 min. The primary antibodies (working dilution 1:400) were rat monoclonal anti-CD31 (MEC 13.3, 550274, BD Biosciences, San Jose, CA, USA), Ki-67 (TEC-3, M7249, Dako, Carpinteria, CA, USA), VHL (Ig32, 556347, BD Biosciences), HIF1α (54, 610959, BD Biosciences), VEGF (P-20, sc-1836, Santa Cruz Biotechnology, Santa Cruz, CA, USA). Slides were then counterstained in hematoxylin and cover-slipped. Sections were imaged using a Zeiss Axioplan2 microscope and Axiovision Rel. 4.5 imaging software. Composite images were constructed with Photoshop CS (Adobe Systems, San Jose, CA). Extent of immunostaining was determined according to the presence or absence of specific staining when compared to positive and negative controls (Supplemental Figure 1). CD31-positive vessel density and Ki-67-positive cell density were determined by analysis of sections from at least 4 independent mice from each genotype. Assessment of microvessel density was determined based on CD31-positive blood vessel count as previously described [
14]; proliferative index was determined based on cell count as described [
10]. Briefly, microvessel density was determined from at least 20 fields imaged at 10× magnification for prostate and 40× magnification for liver with no overlap and identified by evaluating histological sections, and CD31-positive vessels were counted to determine the average vessel numbers per field for each section. Proliferative index was determined from at least 20 fields imaged at 40× magnification with no overlap, Ki-67-positive cells were counted to determine the average number of proliferating cells for each section. All tissues were examined by an animal pathologist (L. Rigatti).
Cell culture and transfection
Primary mouse embryonic fibroblasts (MEF) were generated from embryonic day 12.5 (E12.5) to day 13.5 (E13.5) mixed C57BL/6 J/129 background embryos [
13]. MEFs were maintained in DMEM with 10% fetal bovine serum (FBS), 0.1 mmol/L nonessential amino acids (Invitrogen), 100 μmol/L 2-mercaptoethanol (Sigma), and penicillin/streptomycin. MEFs were transfected using Lipofectamine 2000 (Invitrogen) in OPTI-MEM (Invitrogen). GFP-positive cells were sorted using flow cytometry. GFP-positive MEFs were then cultured in DMEM with 250 μm CoCl
2 to simulate hypoxia for 4 h, and cell lysates were collected for Western blot analysis. To control for variation, transfection experiments were repeated using several different clones.
Plasmids
The full length human EAF2 cDNA was subcloned into pEGFP-C1 vector (Clontech, Mountain View, CA, USA) to generate pEGFP-EAF2. In-frame cloning of the pEGFP-EAF2 fusion was verified by sequencing.
Western blot
Proteins were lysed in RIPA lysis buffer (50 mM Tris HCl pH7.4, 1% NP-40, 0.25% sodium-deoxycholate, 150 mM NaCl. 1 mM EDTA pH 8.0, 1 mM NaF) supplemented with 1 mM PMSF, 1 mM NaV3O4, and 1× protease inhibitor cocktail (P8340, Sigma–Aldrich, St Louis, MO, USA). Frozen livers were homogenized, and cells were placed on ice for 30 min and then centrifuged at 13,000g for 10 min at 4°C. Protein concentration was determined by BCA Protein Assay (Thermo Scientific, Rockford, IL, USA). Proteins (50 μg) from livers or whole cell MEF lysates were denatured and separated on a 10% SDS-polyacrylamide gel and transferred onto a nitrocellulose membrane. Blotted proteins were probed with antibodies as follows: mouse monoclonal anti-HIF1α antibody (1:2,000, 610958, BD Transduction Laboratories), rabbit polyclonal HIF1α (1:500, NB-100-449, Novus Biologicals, Littleton, CO, USA), rabbit polyclonal GFP (1:5,000, TP401, Torrey Pines Biolabs, Houston, TX, USA), rabbit polyclonal GAPDH (1:1,000, FL-335, sc-25778, Santa Cruz Biotechnology) and goat polyclonal β-actin (1:1,000, C-11, sc-1615, Santa Cruz Biotechnology), followed by horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology). Signals were visualized by enhanced chemiluminescence (ECL Western Blotting Detection Reagents, GE Healthcare) and were exposed to X-ray film (Fuji film, Valhalla, NY, USA). Membranes were stripped between antibody probes using a stripping solution (β-ME, 10% SDS, 0.375 M Tris pH 6.8).
Computational analysis of gene expression datasets from human prostate tissue specimens
Previously published datasets were queried for differential expression of CD31, EAF2 and VHL pathway genes VHL, HIF1A, HIF1AN (HIF1A inhibitor), VEGFA, VEGFB, and VEGFC in human prostate tissue specimens. Cell-type specific transcriptome profiles from normal prostate CD26
+ luminal epithelial and CD49
+ stromal cells [
17] were compared to CD26
+ cancer [
18] and CD90
+ cancer-associated fibroblasts [
19]. The data were reported as robust multi-array average (RMA) [
20] normalized Affymetrix signal intensities implemented in an in-house analysis pipeline SBEAMS [
21], or as a composite value: X = log
2(Cancer normalized intensity;Normal normalized intensity).
Statistical analysis
Comparison between groups was calculated using the two-tailed Fisher’s exact test method of summing small P values and the 1-way ANOVA and Bonferonni’s Multiple Comparison Test as appropriate. The level of significance was set at P = 0.05. GraphPad Prism version 4 was used for graphics (GraphPad Software, San Diego, CA, USA). Values are expressed as means ± SEM.
Discussion
It has long been recognized that tumor angiogenesis is critical to tumor growth, survival and metastatic potential [
29]. Microvessel density was been shown to be highly correlated with disease progression in both prostate [
30] and hepatocellular carcinoma [
31]. Loss of tumor suppressors VHL and EAF2 have individually been shown to induce pro-angiogenic phenotypes in murine models. VHL
+/− mice developed hepatic vascular lesions [
7] and EAF2
−/− mice developed PIN lesions [
15]. Furthermore, EAF2 has been shown to bind and stabilize pVHL, thereby inhibiting HIF-driven angiogenesis; and EAF2
−/− mice have reduced levels of pVHL in several major organs [
13]. In the current study, combined deficiency of EAF2 and VHL led to a highly penetrant pro-angiogenic phenotype in the murine liver and prostate.
VHL deficiency consistently increased the incidence of hepatic vascular lesions across three mouse strains at ages >12 mos, and there seemed to be no significant discrepancies in the hepatic phenotype among different strains (Fig.
1). EAF2 deficiency increased the angiogenic effects of VHL heterozygosity in the murine liver. While not statistically significant (
P = 0.08), EAF2
−/−VHL
+/− animals had a marked increase in the incidence of hepatic vascular lesions compared to VHL
+/− animals (100% vs. 55.6%) (Fig.
2). No lesions occurred in EAF2
−/− or wild-type mice. VHL or EAF2 deficiency alone had no effect on cellular proliferation in the liver (Fig.
4). Hepatic lesions in EAF2
−/−VHL
+/− mice displayed a statistically significant increase in proliferation compared to livers of EAF2
−/−, VHL
+/− and wild-type mice as measured by Ki-67 positive cells. The combined effects of EAF2 and VHL loss on proliferation were not seen in the prostate, however. Liver blood vessel formation determined by CD31 immunostaining was also significantly increased in EAF2
−/−VHL
+/− compared to the wild-type liver (Fig.
5). To a lesser degree, the livers of EAF2
−/− and VHL
+/− alone mice also displayed increased microvessel density which was consistent with previous reports [
10,
14].
EAF2 deficiency has previously been shown to induce PIN in mice [
15] and down-regulation of EAF2 expression has been reported in prostate cancer [
15,
18]. VHL heterozygosity increased the incidence of PIN lesions in EAF2
−/− mice, from 60% in EAF2
−/− mice to 100% in EAF2
−/−VHL
+/− mice (Fig.
3). PIN lesions in EAF2
−/−VHL
+/− mice were accompanied by stromal inflammation, fibrosis and fibroplasia and smooth muscle proliferation and a statistically significant increase in microvessel density when compared to EAF2
−/−, VHL
+/− and wild-type mice.
Decreased pVHL expression has been reported in the affected tissues of VHL-deficient mice [
8,
32,
33], due at least in part to the absence of one VHL allele. Rankin et al. [
11] showed that renal tubule cysts, but not normal renal cortical cells exhibited biallelic VHL gene inactivation in VHL-conditionally deficient mice. We found lost or reduced pVHL expression in EAF2
−/−VHL
+/− in areas of PIN in the prostate and in angiectasis of the liver, suggesting that VHL inactivation was associated with the development of these lesions as well (Fig.
8). Loss or reduced pVHL expression could be the result of allelic loss. However, it could also occur through transcriptional or post-transcriptional mechanisms. It was not clear from our analyses whether pVHL loss was associated with a higher grade of PIN in EAF2
−/−VHL
+/− mice as compared to PIN seen in EAF2
−/− mice. Although VHL expression in the prostate epithelium has been reported previously [
27,
28], the physiological role of the VHL gene in prostatic tissue remains to be elucidated.
Here we demonstrate that combined deficiency of VHL and EAF2 in mice increases angiogenesis in murine prostate and liver. Furthermore, this angiogenic phenotype was consistent with an increased incidence of pVHL-negative PIN and hepatic vascular lesions. The EAF2−/−VHL+/− mouse model demonstrates that these two tumor suppressors cooperate in the regulation of angiogenesis and that loss of these genes contributes to the development of neoplasia in the liver and prostate.
Acknowledgments
We are grateful to Ricardo Vencio and Julio Garcia for gene array data analyses and to Jianhua Xu, Aiyuan Zhang, Katie Leschak, Dawn Everard, Marie Acquafondata and Marianne Notaro for technical support. We also thank Dr. Laura Schmidt from NCI for providing VHL+/− mice. This work was funded in part by grants T32 DK007774, R37 DK51193, R01CA120386, CA78335 and CA125930 (JG).