MOLECULAR AND CELLULAR BIOLOGY, Sept. 2011, p. 3885–3895 0270-7306/11/$12.00 doi:10.1128/MCB.05089-11 Copyright © 2011, American Society for Microbiology. All Rights Reserved.
Vol. 31, No. 18
Hypoxia-Inducible Factor 1 Is Activated by Dysregulated Cyclin E during Mammary Epithelial Morphogenesis䌤 Tanushri Sengupta,1 Gathi Abraham,1 Yanfei Xu,1 Bruce E. Clurman,3 and Alex C. Minella1,2* Department of Medicine, Division of Hematology/Oncology,1 and Robert H. Lurie Comprehensive Cancer Center,2 Northwestern University Feinberg School of Medicine, Chicago, Illinois 60611, and Divisions of Clinical Research and Human Biology, Fred Hutchinson Cancer Research Center, Seattle, Washington 981093 Received 21 January 2011/Returned for modification 18 February 2011/Accepted 1 July 2011
Increased cyclin E expression has been identified in human tumors of diverse histologies, and in studies of primary breast cancers, high cyclin E is associated with poor prognosis. We have studied dysregulated cyclin E in epithelial tissues using organotypic cultures of human mammary epithelial cells and a murine model. We unexpectedly discovered that dysregulated cyclin E impairs normal acinar morphogenesis in vitro, and this is associated with the induction of p21Cip1, p27Kip1, and cellular senescence. Cyclin E-induced morphogenesis arrest is dependent upon hypoxia-inducible factor 1␣ (HIF-1␣), which itself is induced by high cyclin E both in cultured mammary acini and in mammary epithelial tissues in a mouse model of deregulated cyclin E expression. We next determined that E2F activity directly regulates and is required for induction of HIF1A by cyclin E. Additionally, we found that cyclin E deregulation in mammary acini decreases, in an E2F-independent manner, expression of the EGLN1 prolyl hydroxylase that regulates HIF-1␣ degradation within the VHL ubiquitin ligase pathway. Together, our findings reveal a direct link between cyclin E and HIF-1 activities in mammary epithelial cells and implicate HIF-1 as a mediator of proliferation-independent phenotypes associated with high cyclin E expression in some human breast cancers. ubiquitin-mediated proteolysis, causes genomic instability that is associated with the centrosome hyperamplification, unstable DNA replication intermediates, and defective chromosome segregation during mitosis (1, 11, 42, 43). In vivo, impaired Fbw7-mediated cyclin E degradation causes increased cell proliferation and increased apoptosis as well as impaired maturation in a cell-type-specific manner (34), in addition to increased genomic instability and tumorigenesis (29, 30, 46). Cancers of diverse cell lineages express high levels of cyclin E, and in various studies, high cyclin E expression correlates with increased tumor aggression (18). This correlation likely does not simply reflect increased proliferation (22, 23); rather, inappropriate cyclin E expression may have far-reaching biological consequences to cancer cells, such as promoting the acquisition of new genetic lesions and altering gene expression programs. In this study, we first investigated the effects of deregulated cyclin E expression in human mammary epithelial cells (MCF10A) undergoing acinar morphogenesis in organotypic cultures. This experimental system provides opportunities to study phenotypes in culture that are not observable in standard monolayer growth conditions, including the formation of epithelial structures that depend on a complex interplay of progrowth, prosurvival, and cell death-promoting signals (7, 27). We found, unexpectedly, that expression of cyclin E mutated at threonine 380 (T380) [cyclin E(T380A)] to prevent Fbw7-dependent degradation (6, 47, 51) causes reduced proliferation and impairs acinar morphogenesis in a majority of mammary epithelial structures. These findings are associated with induction of both p21Cip1 and p27Kip1 and markers of cellular senescence and are dependent on expression of hypoxia-inducible factor 1␣ (HIF-1␣). Notably, induction of HIF-1␣ and p21/p27 by high cyclin E expression are features of MCF-10A
Cyclin E–cyclin-dependent kinase 2 (CDK2) is an evolutionarily conserved kinase that phosphorylates substrates involved in transcriptional control (e.g., Rb) (16), DNA replication (e.g., NPAT, Cdt1, and Cdc6) (28, 31, 52), centrosome duplication (e.g., nucleophosmin and CP110) (5, 39), and cell survival regulation (FoxO1A) (17). The essential functions of cyclin E, revealed by germ line knockout studies, include the regulation of cell cycle reentry from quiescence and endoreduplication (13). Moreover, cyclin E appears to play a critical role during tumorigenesis, as cyclin E-null fibroblasts are resistant to transformation by activated Ras (13). Cyclin E activity normally oscillates throughout the cell cycle, peaking at early S phase, and these oscillations result from E2F-dependent transcription and ubiquitin-mediated proteolysis controlled by the SCFFbw7 ubiquitin ligase (38, 50). When cyclin E is bound to CDK2, it is targeted for ubiquitination by the SCFFbw7 ubiquitin ligase, and this process requires cyclin E phosphorylation within a conserved phospho-epitope, termed the Cdc4 phosphodegron (CPD) (24, 47, 51). Phosphorylation of cyclin E at CPDs can be performed by both CDK2 and glycogen synthase kinase 3 (GSK3) (51). Thus, a number of signaling pathways that regulate GSK3 activity, including the Wnt and PI3K/AKT signaling pathways, can potentially modulate Fbw7-dependent degradation of cyclin E as well as other Fbw7 substrates (50). Cyclin E dysregulation may play a fundamental role in tumorigenesis. Excess cyclin E activity, resulting from defective * Corresponding author. Mailing address: Northwestern University Feinberg School of Medicine, 303 E. Superior Street, Lurie 5-131, Chicago, IL 60611. Phone: (312) 503-2041. Fax: (312) 503-0189. E-mail:
[email protected]. 䌤 Published ahead of print on 11 July 2011. 3885
3886
SENGUPTA ET AL.
cells cultured for acinar morphogenesis, in contrast to monolayer conditions, and are observed also with mammary epithelial cells isolated from cyclin ET74A T393A knock-in mice, which are defective for Fbw7-dependent cyclin E destruction (34). We identified candidate E2F binding sites within the HIF1A promoter sequence and used reporter assays to demonstrate transactivation of a HIF1A reporter by E2F-1. We also demonstrated occupancy of the HIF1A promoter by E2F-1 and verified that endogenous E2F activity mediates HIF1A transactivation in the setting of dysregulated cyclin E. We additionally determined that dysregulated cyclin E downregulates expression of the EGLN1 prolyl hydroxylase, which promotes degradation of HIF-1␣ via the von Hippel-Lindau (VHL) pathway, in both cultured human mammary epithelial cells and mouse mammary tissues. Given the proposed role of HIF-1 in the promotion of neoplastic progression (44), our data suggest one way in which dysregulated cyclin E may drive malignant phenotypes in mammary epithelial cells, independent of promoting cell proliferation. MATERIALS AND METHODS Cell culture. MCF-10A cells, obtained from Joan Brugge (Harvard Medical School) were cultured and maintained in monolayer conditions as previously described (8). Acinar morphogenesis cultures were established by overlaying suspensions of MCF-10A cells onto Matrigel basement membrane extract (BD Biosciences) that was dispensed onto chamber slides (BD Biosciences) for microscopy studies or into 24-well plates for RNA, protein, and cell cycle analyses. 293T and Phoenix-amphotropic producer cells (G. Nolan, Stanford University) were grown in Dulbecco modified Eagle medium (DMEM) with 10% fetal calf serum. Transfection of 293T and Phoenix cells used calcium phosphate precipitation. MCF-10A cells were infected with filtered retroviral supernatants containing 4 mg/ml polybrene, and cells were maintained in antibiotic selection until seeding in basement membrane cultures. Plasmids and antibodies. All human cyclin E retroviral expression constructs (in pBabe) were previously described (35). The short hairpin RNA (shRNA) constructs for p21 and p27 knockdown were previously described (33, 49). The HIF1A shRNA construct was obtained from Open Biosystems (identification no. V2HS_132151), and the VHL shRNA construct was provided by Navdeep Chandel (Northwestern University). A vector expressing a luciferase-targeting shRNA sequence was used as a control in the knockdown studies. The pBabe–E2F-1 and E2F-1–E132 constructs were provided by Kristian Helin (University of Copenhagen). Antibodies used in immunoblots were anti-cyclin E (HE12), anti-p21 (C19), anti-p27 (C19), anti-Rb (C15), and anti-E2F-1 (KH95), all from Santa Cruz Biotechnology; anti-HIF-1␣ (H1alpha67), anti-HIF2␣ polyclonal, and antiEGLN1 polyclonal antibodies, all from Novus Biologicals; anti-GRB2 (BD Biosciences); anti-LC3B (Cell Signaling); and anti-phospho-Rb (Ser807/811, Cell Signaling). In immunoprecipitations for in vitro kinase assays performed as previously described (35), antibodies were anti-cyclin E (HE111) and anti-CDK2 (M2), both from Santa Cruz Biotechnology. For immunofluorescence, antibodies were anti-Ki67 polyclonal (Zymed), anti-cyclin E (HE111; Santa Cruz Biotechnology), anti-laminin 5 (D4B5; Chemicon), anti-GM130 polyclonal (BD Biosciences), anti-cleaved caspase 3 (Asp175; Cell Signaling), and anti-HP1␥ (MAB3450; Chemicon). For immunohistochemistry (IHC), anti-p21 and antip27 monoclonal antibodies were from BD Biosciences and anti-HIF-1␣ was the same as that for immunoblotting. Immunoblot and gene expression analysis. Crude extracts from MCF-10A cells grown in a monolayer were made by lysing cells in NP-40 buffer supplemented with protease and phosphatase inhibitors, as previously described (35). Lysates from mammary acini were made by first adding two volumes of 0.25% trypsin to basement membrane cultures (after removing overlay media). Then, Matrigel containing acini was removed by scraping the plate surface with a pipette tip, and the mixtures were gently agitated at 37° for 15 min. Cells were collected by centrifugation after washing in cold PBS and lysed in NP-40 buffer containing inhibitors. Protein concentrations were quantified using the Bradford protein assay (Bio-Rad), and normalized lysates were electrophoresed and transferred to nitrocellulose membranes. Quantification of immunoblots was performed using ImageJ software (NIH). RNA from cells grown in basement membrane cultures was purified using TRIzol (Invitrogen) and added directly to the
MOL. CELL. BIOL. cultures. cDNA was synthesized using 1 g total RNA and the Affinity Script QPCR cDNA synthesis kit (Stratagene). Quantitative real-time PCR (RT-PCR) was performed using a Light Cycler 480 PCR system (Roche). Gene expression values were normalized to GAPDH (glyceraldehyde-3-phosphate dehydrogenase) mRNA levels, and differences computed using the threshold cycle (⌬CT) method. Primer sequences are available on request. Acinar size quantification. MCF-10A cells were grown in basement membrane extract, and on day 8 or 18, each well was washed carefully and kept in phosphate-buffered saline (PBS) for the duration of image acquisition. Images were taken using a Nikon SMZ1500 microscope at ⫻11.25 magnification and calibrated for distance measurement. All images were acquired using the same microscope settings, and acinar surface areas were measured using ImageJ software, with a fixed pixel-to-distance ratio maintained for all experiments. Each comparison of acinar sizes shown is comprised of data from at least two independent experiments, and statistical significance was determined using the twotailed t test (Prism 5.0; GraphPad). Immunofluorescence and IHC. Cells were grown in chamber slides coated with Matrigel basement membrane extract. Cultures were grown for 18 days and then fixed with 2% paraformaldehyde. Washing, blocking, and primary antibody incubation steps were performed as previously described (8). Alexa fluorophoreconjugated secondary antibodies were used (Molecular Probes). Image acquisition was done using a Zeiss LSM 510 META laser-scanning confocal microscope, and analysis was performed using ImageJ software. For detection of senescenceassociated (SA) chromatin marks, acinar structures were dissociated from Matrigel using trypsin, as performed for cell cycle analyses. Single cells were then recovered by centrifugation, spotted onto slides, air dried, fixed with 2% paraformaldehyde, and stained with antibody. For senescence-associated beta-galactosidase (SA -Gal) detection, cells were similarly spotted onto slides, and then staining was performed as previously described (10). For immunohistochemistry assays, Matrigel was dispensed onto plastic coverslips precut to fit into the chamber slides prior to cell seeding, and then after culturing, 10% neutral buffered formalin (Sigma) was added to the wells for 24 h. After scraping from the plastic coverslips with a razor, the Matrigel/acini were then fixed for an additional 24 h prior to paraffin embedding. Only one section per culture well was scored to avoid duplication of data from the same structures. Cell cycle analyses. For bromodeoxyuridine (BrdU) labeling, acinar cultures were incubated with 20 mM BrdU for 16 h. Then, cells were liberated from Matrigel, collected using the same method as that used for lysate preparation, fixed, permeabilized, and stained for detection of incorporated BrdU and cell cycle distribution (BrdU flow kit; BD Biosciences). Autophagy assays. For transmission electron microscopy (TEM), Matrigel was dispensed onto plastic coverslips, which were precut and placed into chamber slides prior to cell seeding. Then after culturing for 8 to 18 days in morphogenesis cultures, one-half-strength Karnovsky’s fixative in 0.1 M cacodylate acid buffer was added to each well for in situ fixation for 24 h at 4°. Then, the Matrigel/acini were scraped from the plastic coverslips and fixed for additional time, prior to washing them in 0.1 M cacodylate acid buffer, postfixation treatment with 1% osmium tetroxide in 0.1 M cacodylate acid buffer, and ethanol dehydration. The Fred Hutchinson Cancer Research Center Electron Microscopy Facility performed processing steps to yield 1-m sections. Samples were viewed and photographed using a JEOL 1230 instrument. In autophagic flux assays, day 8 acinar cultures were treated with bafilomycin A1 (Molecular Probes) at a dose of 200 nM or vehicle, and LC3-II accumulation was quantified by densitometry and normalized to the loading control. Apoptosis assays. The ethidium bromide retention assay to identify cell death in situ was performed as previously described (8). Image acquisition utilized a Leica DM 4000B microscope, and analysis was performed using ImageJ software. Cleaved caspase 3 detection used immunofluorescence methods described above. Gene expression analysis. RNA was extracted from sorted CD71highTer119high bone marrow cells using RNeasy minikits (Qiagen), and DNase treatment followed. Microarray analysis was performed using Sentrix Mouse-6 version 2 expression BeadChips (Illumina). Microarray analyses were performed in duplicate for three littermate pairs of wild-type and cyclin ET74A T393A mice. Data were analyzed subsequently using the gene set enrichment analysis (GSEA) tool (Broad Institute). Transactivation assays. A luciferase reporter plasmid containing the human HIF1A promoter was made by amplifying nucleotides ⫺2180 to ⫹28 (relative to the transcriptional start position) from genomic DNA and subcloning into pGL2b vector (Promega). The construct, here designated the ⌬ mutant, was made by deleting nucleotides ⫺255 to ⫺56, and a second mutated promoter construct was made by introducing mutations within positions ⫺89 to ⫺74 (here designated *; CTCCCCGCGCGCCCG to GGATCCGTCGACCAG). Cell lysates of transfected 293T cells were prepared 2 days after transfection, and cell
VOL. 31, 2011
HIF-1 INDUCTION BY DYSREGULATED CYCLIN E
luminescence was measured using a dual luciferase reporter assay kit (Promega) and a Moonlight 2010 luminometer (Analytic Luminescence Laboratory). Firefly luciferase activities were normalized based on Renilla reporter activity (expressed from cotransfected pRL-CMV; Promega). All assays were performed in triplicate. HIF1A reporter-expressing acini were made by transfecting MCF-10A cells with pGL4.20 (Promega) into which we subcloned the HIF1A promoter. After selecting in puromycin, we expanded the pooled, stable transfectant cell clones. Following transduction with retroviral vectors and selection, cells were seeded in basement membrane extract and allowed to grow for 8 days. Lysates were then prepared for bioluminescence measurement as described above, except, for these assays, total protein content was used to normalize samples. Chromatin immunoprecipitation (ChIP). MCF-10A cells stably transduced with cyclin E(T380A) or a control plasmid were fixed at room temperature for 10 min with 1% formaldehyde. After sonication, cross-linked protein-DNA complexes were precipitated using anti-E2F-1 antibody or normal mouse IgG and protein G Sepharose (GE Healthcare) to adsorb immune complexes. DNAprotein complexes were washed and reverse cross-linked. Immunoprecipitated DNA was amplified by RT-PCR, and product quantity was determined relative to a standard curve generated from the titration of input chromatin. Mouse mammary epithelial cell preparations. Mice were housed in barrier facilities, and euthanasia was performed using CO2 asphyxiation according to Northwestern University ACUC guidelines. Female wild-type and cyclin ET74A T393A mice (34) (129Sv background), matched for age and prior pregnancies, were sacrificed at day 14.5 postcoitum (p.c.), and no. 4 mammary glands were harvested. Mammary glands were minced and resuspended in DMEM/F12 (Invitrogen) without serum, which was supplemented with collagenase/hyaluronidase (Stem Cell Technologies). Tissues were incubated at 37° for 30 min with gentle agitation and centrifuged to separate fat from epithelial organoids. Organoids were treated with DNase (2 U/ml), then washed, and pelleted at least four times. Protein extracts and RNA were then prepared as described for MCF-10A acini.
pressing, and cyclin E(T380A)-expressing acini for cleaved caspase 3 expression, and though we found a slight increase in apoptosis within the large cyclin E(T380A) structures compared to control and wild-type cyclin E acini, the arrested structures showed significantly less activated caspase 3 staining (Fig. 1F). Thus, increased apoptosis does not account for the arrested acinar morphogenesis associated with dysregulated cyclin E expression. We also tested whether increased autophagy accounts for impaired morphogenesis of cyclin E(T380A) acini. To do this, we measured autophagic flux after treating acinar cultures with bafilomycin A1 (36). We actually noted modestly decreased accumulation of the phosphatidylethanolamine-conjugated LC3 form (LC3-II) at the early time point in cyclin E-overexpressing acini compared to controls, possibly indicative of defective autophagic degradation, but this was not statistically significant with repeated experiments. Moreover, accumulation of LC3-II at a later time point was nearly identical across the different acinar cultures (Fig. 1G), and using transmission electron microscopy, there was no evidence of increased autophagic vesicles within cyclin E(T380A) acinar cells that would be consistent with impaired clearance of autophagosomes (Fig. 1H). Thus, significantly altered autophagic flux in cyclin E(T380A) acinar cells does not account for their impaired epithelial morphogenesis in organotypic culture. In contrast, with both an immunofluorescence-based assay to identify HP1␥ nuclear staining and senescence-associated beta-galactosidase detection (10), we found that there are significantly increased numbers of cells expressing these markers of senescence in the cyclin E(T380A)-expressing acini compared to in the control acini (Fig. 1I). Under monolayer growth conditions, in comparison, cyclin E(T380A) expression did not induce senescence-associated beta-galactosidase activity. Thus, dysregulated cyclin E expression induces cell cycle arrest with some features of cellular senescence and impairs acinar morphogenesis of mammary epithelial cells in organotypic cultures. Dysregulated cyclin E induces expression of p21Cip1 and p27Kip1, as well as hypoxia-inducible factor 1␣ (HIF-1␣), in cultured mammary acini and primary mouse mammary epithelial cells. Oncogene-induced senescence is associated with induction of cyclin-dependent kinase inhibitors (2, 9, 32). We thus assayed cyclin E(T380A)-expressing and control acini for the induction of both p21Cip1 and p27Kip1. Using IHC, we readily detected p27 expression in both control and cyclin E(T380A) acini, and this was modestly increased in the latter. In contrast, p21 expression was more significantly upregulated in the cyclin E(T380A) acini, and by IHC, p21 was found to be highly expressed in the arrested acinar structures (Fig. 2A). Notably, the induction of both p21 and p27 proteins (Fig. 2B) was associated with increased gene expression, and this was specific to MCF-10A acini, as the same cyclin E(T380A)-transduced cell population showed no increase in CDK inhibitor expression under monolayer growth conditions (Fig. 2C). The induction of both p21 and p27 expression suggests activation of a response to dysregulated cyclin E in mammary epithelial cells that is distinct from one that we have previously described for primary fibroblasts, in which p53-dependent p21 induction is a characteristic feature (33, 35). Data from gene expression studies performed for ongoing studies with cyclin
RESULTS Dysregulated cyclin E expression impairs acinar morphogenesis. We transduced MCF-10A cells prior to seeding them in basement membrane extract to induce acinar morphogenesis, using retroviral constructs to express either wild-type cyclin E or the threonine 380-to-alanine cyclin E mutant [the cyclin E(T380A) mutant]. This mutant is incompetent for binding to the SCFFbw7 ubiquitin ligase, resulting in sustained expression through the cell cycle compared to that of wild-type cyclin E (29). We found that expression of cyclin E(T380A) impairs mammary acinar formation in the organotypic cultures, with many structures not reaching half the size of the empty vectoror wild-type cyclin E-expressing acini (Fig. 1A). These differences were most apparent by day 8 of growth in basement membrane extract and did not resolve over additional time in culture (Fig. 1B). A minority of cyclin E(T380A) acini were enlarged, showing increased proliferation and crowded centers compared to controls (Fig. 1A and C). Using both immunoblotting to measure cyclin E expression in extracts made from isolated acini and immunofluorescence to stain cyclin E in situ, we were able to discount poor expression of cyclin E as an explanation for the aborted cyclin E(T380A)-expressing epithelial structures (Fig. 1D). Moreover, we found that the proliferation defect associated with cyclin E(T380A) expression is specific to cell growth in organotypic conditions and is not observed with the same transduced cell populations maintained in monolayer culture (Fig. 1E). We next sought to identify the mode of morphogenesis arrest associated with expression of cyclin E(T380A). Using an ethidium bromide retention assay, we first determined that the arrested cyclin E(T380A) acini were not comprised primarily of dead cells. Next, we scored control, wild-type cyclin E-ex-
3887
3888
SENGUPTA ET AL.
MOL. CELL. BIOL.
FIG. 1. Dysregulated cyclin E expression in MCF-10A cells impairs acinar morphogenesis in organotypic cultures. (A) MCF-10A cells transduced with the indicated retroviral constructs were seeded on basement membrane extract (Matrigel)-coated chamber slides. Cultures were grown for a total of 8 days, after which acinar structures were photographed (⫻11.25 magnification; insets display representative images), and acinar surface area was measured using ImageJ software. Graphs represent distribution of structures, with individual acinar measurements binned according to the indicated surface areas. Calculated P values for differences between control (vector) acinar size distribution and those of wild-type cyclin E (wt) and cyclin E(T380A) acini are 0.65 and 0.006, respectively. (B) MCF-10A cells transduced with control or cyclin E(T380A)-expressing retroviral vectors were grown in Matrigel for 18 days. (C) Representative acini were fixed after 18 days and stained for KI67 (green), GM130 (red, Golgi stain), and TOPRO (blue, nuclear counterstain). Arrowhead indicates arrested cyclin E(T380A) structure, and other cyclin E(T380A) structures shown are representative of the hyperproliferative acini. (D) (Left) Protein extracts were prepared from day 8 acinar cultures and immunoblotted for cyclin E. GRB2 expression is shown as loading control. (Right) Images are shown of representative control and arrested and hyperproliferative cyclin E(T380A) acini, fixed after 18 days of growth and stained for cyclin E (red) and TOPRO. (E) Tabulated BrdU incorporation results are shown for MCF-10A cells that were transduced with the indicated constructs and grown for 8 days in either acinar or monolayer cultures. (F) (Top) Representative (magnification, ⫻20) images are shown of ethidium bromide staining of live, day 8 acini, comprised of cells expressing the indicated constructs, for in situ cell death detection. (Bottom) Tabulated cleaved caspase 3 immunofluorescence (IF) detection results are displayed for fixed, day 18 acini. Smaller cyclin E(T380A) structures are those comprised of 12 or fewer cells per image, and larger structures are those with 13 or more. (G) Endogenous LC3 processing was measured in the indicated mammary acinar cultures, treated with either bafilomycin A1 for the indicated times or vehicle alone, and the accumulation of LC3-II was measured. Values displayed indicate average increase in normalized LC3-II signal after treatment, in two independent experiments. SD, standard deviation. (H) Representative transmission electron micrograph images of an intact, control MCF-10A acinar structure (magnification, ⫻400) or an arrested, cyclin E(T380A) structure (magnification, ⫻1,000), following growth in Matrigel. (I) (Top) HP1␥ and/or SA -Gal staining was performed for acinar cells, disaggregated from the indicated acinar cultures, or for cells grown in a monolayer. Representative results with staining for SA -Gal (magnification, ⫻11.25) and nuclear HP1␥ (magnification, ⫻100) are shown. Arrowheads indicate rare positive cells detected in monolayer cultures. (Bottom) The graph displays average fraction of cells positive for either SA -Gal or HP1␥ per field. Error bars indicate standard deviations.
ET74A T393A bone marrow cells suggested the possibility that HIF-1 activity is increased by dysregulated cyclin E (Fig. 2D), in these and in mammary epithelial cells, especially as both p21 and p27 are positively regulated by HIF-1 (12, 14, 25). Indeed,
we found activation of a panel of known HIF-1 transcriptional targets (summarized in reference 21) in cyclin ET74A T393A primary mammary epithelial cells (Fig. 3A) and in cyclin E(T380A)-expressing mammary acini, though not in MCF-10A
VOL. 31, 2011
HIF-1 INDUCTION BY DYSREGULATED CYCLIN E
3889
FIG. 2. Dysregulated cyclin E induces HIF-1␣ during acinar morphogenesis. (A) MCF-10A cells transduced with control or cyclin E(T380A)expressing vectors were cultured for acinar morphogenesis over 8 days, formalin fixed and paraffin embedded (FFPE), and stained for detection of p21 and p27 (magnification, ⫻40, representative images [left]). (Right) Cells positive for either p21 or p27 are expressed as ratios to total cell number per ⫻40 magnification field. (B) Extracts were prepared from the indicated acinar cultures after 8 days of growth and immunoblotted for p21, p27, and GRB2 proteins. Relative abundances of p21 and p27 (normalized to GRB2 signal) are shown as averages from three independent experiments (tripl.) along with calculated standard deviations. (C and E) RNA was extracted from monolayer or acinar cultures after 8 days of culture growth and analyzed by quantitative RT-PCR for the expression of CDKN1A (p21) and CDKN1B (p27) (C) or the indicated HIF-1regulated genes (E). Quantitative RT-PCRs were performed in triplicate, and the data shown represent the average from at least two different experiments. Error bars show standard deviation. (D) Shown are GSEA output data generated from comparison of gene expression signatures of CD71high Ter119high bone marrow cells from cyclin ET74A T393A mice, compared to those from wild-type mice. A signature corresponding to dysregulated expression of predicted HIF-1 targets is shown along with the corresponding P value and false-detection rate q value (FDR q) (48). (F) (Top) MCF-10A acini were FFPE processed as described above for panel A and stained for HIF-1␣ detection. Smaller structures are those comprised of 12 or fewer cells per section, and larger structures are those with 13 or more. Acini considered positive for HIF-1␣ staining contain at least one positive cell. Inset, representative image of a HIF-1␣-positive structure is shown (magnification, ⫻40). (Bottom) Extracts prepared from the indicated acinar cultures after 8 days of growth were immunoblotted for HIF-1␣ and GRB2 proteins, and relative abundance of HIF-1␣ was quantified in four independent experiments. (G and H) HIF1A or HIF2A gene expression for the indicated acini or monolayer cultures was measured as described above.
cells grown in a monolayer (Fig. 2E), similar to our observations of CDK inhibitor induction in cyclin E(T380A) acini but not in monolayer cells. The rate-limiting component of the HIF-1 heterodimeric transcription factor is HIF-1␣ (HIF-1␣), which is increased in abundance in cyclin E(T380A) mammary acini (Fig. 2F). Indeed, by IHC, we identified a marked dichotomy in HIF-1␣ staining between “small” (⬍12 cells) and larger cyclin E(T380A) structures, suggesting that HIF-1␣ induction plays a key role in the induction of senescence associated with dysregulated cyclin E expression during mammary epithelial morphogenesis (Fig. 2F). For wild-type cyclin E-expressing structures, we noted an intermediate induction of expression of both HIF-1␣ and the CDK inhibitors, by Western blotting (Fig. 2B and F). We found that, comparable to our results with the CDK inhibitors, HIF1A gene expression was induced by cyclin E(T380A) expression (and only modestly by wild-type cyclin E), findings that were not reproduced with monolayer MCF-
10A cells (Fig. 2G). Importantly, we confirmed increased protein and mRNA levels of HIF-1␣ as well as p21 and p27 in primary cyclin ET74A T393A mouse mammary epithelial tissues (Fig. 3B and C). Thus, impaired Fbw7-mediated cyclin E regulation induces expression of the HIF-1␣ protein, and this is associated with upregulation of HIF-1 target gene expression. Notably, HIF2␣, a component of the related HIF2 transcription factor, is not upregulated similarly in cyclin E(T380A) acini (Fig. 2H). Arrested acinar morphogenesis induced by dysregulated cyclin E is HIF-1␣ dependent. To test the functional consequences of HIF-1␣ induction by cyclin E(T380A) in mammary cells, we utilized a retrovirally encoded short hairpin RNA (shRNA) targeting HIF1A mRNA and verified its efficacy by measuring HIF-1␣ protein in MCF-10A cells grown in hypoxic conditions (Fig. 4A, left) and in mammary acini (Fig. 4A, right), as well as by assaying gene expression (Fig. 4B). We determined that knockdown of HIF-1␣ expression in cyclin
3890
SENGUPTA ET AL.
FIG. 3. HIF-1 induction occurs in primary mammary epithelial cells of cyclin ET74A T393A knock-in mice. (A) Expression of HIF-1 target genes was measured as described in the legend to Fig. 2C in freshly isolated mammary epithelial cells from four cyclin ET74A T393A knock-in mice versus wild-type mammary epithelial cells (obtained from age-matched, day 15 p.c., no. 4 mammary glands). (B) HIF-1␣, p21, p27, and GRB2 (loading control) protein levels were assayed by immunoblotting protein extracts prepared from wild-type and knock-in mammary epithelial cells. (C) Transcript levels of HIF1A, CDKN1A, and CDKN1B were measured as described above for panel A.
E(T380A)-expressing acini is sufficient to abrogate the induction of p21 and p27 protein and mRNA levels (Fig. 4A, right, and 4B). We next found that HIF1A knockdown prevents the emergence of senescence-associated HP1␥ foci in cyclin E(T380A)-expressing mammary acini (Fig. 4C). A major regulator of HIF-1␣ expression is the ubiquitin ligase that contains the VHL protein (pVHL) as its substrate binding module. We used shRNA against VHL, and after verifying its efficacy (Fig. 4D), we found that the formation of MCF-10A acini is indeed sensitive to increased expression of HIF-1␣, as VHL knockdown is associated with impaired acinar morphogenesis (Fig. 4E). We then demonstrated that reduced expression of HIF-1␣ using shRNA restored acinar morphogenesis in the setting of cyclin E(T380A) expression (Fig. 4F). Therefore, induction of HIF-1␣ expression by dysregulated cyclin E directly impairs normal mammary acinar formation in culture, and this is associated with induction of both p21 and p27, which, we also determined, using shRNA-mediated knockdown, cooperatively mediate the cyclin E(T380A)-associated morphogenesis arrest (Fig. 4G and H). Increased expression of HIF1A associated with dysregulated cyclin E is E2F dependent. We expected cyclin E(T380A) expression to generate increased E2F transcriptional activity, as cyclin E-CDK2 promotes E2F activation by negatively regulating Rb (16). We verified that a number of known E2F targets are upregulated in cyclin E(T380A)-expressing acini compared to control and wild-type cyclin E acini and that this correlated with Rb phosphorylation (Fig. 5A). To investigate the possi-
MOL. CELL. BIOL.
bility that increased HIF1A expression in cyclin E(T380A) acini is E2F dependent, we first utilized the ECR Browser application (40) to scan the HIF1A promoter for predicted E2F binding sites and found a number of noncanonical sites located within a highly conserved, GC-rich region less than one kilobase upstream of the transcriptional start site (Fig. 5B). To test the validity of these predicted sites, we cotransfected 293T cells with a luciferase reporter construct containing the HIF1A promoter region and either wild-type E2F-1 or an E2F-1 mutant that is defective for DNA binding (E2F-1–E132) (37). The E2F-1–E132 mutant allows us to distinguish direct activation from indirect activation, possibly through E2F association with another transcription factor (41). We found that transfected wild-type E2F-1 significantly activates the HIF1A reporter, in contrast to the E2F-1–E132 mutant (Fig. 5C, top). These results were reproduced in separate assays done with MCF-10A acini, comprised of cells that stably express the HIF1A reporter (Fig. 5C, bottom). Next, to better define the E2F binding site(s) within the HIF1A promoter, we designed two mutant HIF1A promoter constructs: one in which we deleted the 200base pair GC-rich region containing all predicted, conserved E2F-binding sites (here designated the ⌬ mutant) and a second in which we introduced nucleotide substitutions within a 14nucleotide stretch (Fig. 5D, *) located inside the GC-rich region and containing overlapping, predicted E2F-binding sites. We tested these mutants in 293T cells and found that the deletion mutation nearly completely disabled the ability of the HIF1A reporter to be transactivated by ectopically expressed E2F-1 (Fig. 5D). The construct containing mutations within the 14-nucleotide region showed 50% of the activation with E2F-1 expression, as seen with the wild-type reporter (Fig. 5D). These results are consistent with previously published observations that E2F transcription factors exhibit considerable flexibility in their binding motifs (41) and the concept that E2F likely binds multiple sites within the GC-rich region located just upstream of the HIF1A transcriptional start site. To demonstrate the importance of this putative E2F binding site in mediating HIF1A induction by cyclin E further, we compared the activation of the HIF1A deletion mutant reporter construct (Fig. 5E, ⌬), stably expressed in mammary acini, with that of the wild-type HIF1A reporter. We found, as expected, that the wild-type reporter is activated by cyclin E(T380A) expression, whereas the mutant reporter showed no evidence of activation by cyclin E (Fig. 5E). We next evaluated E2F binding to the HIF1A promoter in MCF-10A cells by performing chromatin immunoprecipitation (ChIP) analysis. We found a significant increase in E2F occupancy at the HIF1A promoter in cyclin E(T380A)-expressing cells (Fig. 5F), consistent with the notion that increased cyclin E activity stimulates recruitment of E2F to the HIF1A promoter. Indeed, we verified that cyclin E kinase activity is higher in cyclin E(T380A)-expressing cells, although net CDK2 activity is lower (Fig. 5G), likely due to the skewed distribution of p21 and p27 throughout the pool of cyclin/CDK2 complexes resulting from cyclin E-CDK2-catalyzed degradation of bound p21 and/or p27 (3, 45). To further demonstrate the importance of endogenous E2F activity in mediating HIF-1␣ induction by dysregulated cyclin E, we utilized the HIF1A reporter-expressing MCF-10A cell line to show that expression of the E2F-1– E132 mutant, which acts to inhibit endogenous E2F activity by
VOL. 31, 2011
HIF-1 INDUCTION BY DYSREGULATED CYCLIN E
3891
FIG. 4. Impaired morphogenesis of cyclin E(T380A) acini is dependent on HIF-1␣. (A) MCF-10A cells first transduced with control or cyclin E(T380A)-expressing constructs were again transduced with vectors expressing a short hairpin targeting HIF1A (sh-HIF1A) or luciferase (sh-ctrl). (Left) Protein lysates were prepared from monolayer MCF-10A cells transduced with the indicated vectors and then subjected to low oxygen for 16 h. HIF-1␣ immunoblot to determine gene knockdown efficiency is shown. (Right) Acini from cells transduced with the indicated constructs were grown for 8 days, and then protein extracts were prepared and immunoblotted for cyclin E, HIF-1␣, p21, p27, and GRB2. (B) Gene expression of HIF1A, PGK1, CDKN1A, and CDKN1B was measured from the indicated cultures by quantitative RT-PCR. (C) Individual cells were isolated from the indicated acini and stained for HP1␥ as described in the legend to Fig. 1E. (D) Gene expression of VHL and PGK1 was measured from the indicated cultures by quantitative RT-PCR. (E and F) MCF-10A cells transduced with the indicated constructs were grown in mammary acinar cultures, and epithelial structure sizes were measured after 8 days of growth as described in the legend to Fig. 1A. Calculated P values for differences in size distributions between both control and sh-VHL structures and control and sh-HIF1A were ⬍0.001. (G) Knockdown efficiency of CDKN1A and CDKN1B was assayed with the indicated cultures by quantitative RT-PCR. (H) MCF-10A cells transduced with the indicated constructs were grown in mammary acinar cultures, and epithelial structure sizes were measured after 8 days of growth as described in the legend to Fig. 1A. Calculated P value for difference in size distribution of control and p21/p27 knockdown acini, ⬍0.001.
sequestering critical coregulators (37), abrogates reporter activation by cyclin E(T380A) expression in mammary acini (Fig. 5H). Importantly, our ChIP studies were performed in monolayer MCF-10A cells, though we observed only increased HIF1A gene expression in MCF-10A mammary acini (Fig. 2G). Our results in aggregate, therefore, indicate that E2F binding to the HIF1A promoter is necessary but not sufficient for HIF-1␣ induction by dysregulated cyclin E and suggest that another regulator of HIF1A expression cooperates with E2F in specific contexts (e.g., in vivo and in organotypic culture conditions) to drive activation of gene expression in response to high cyclin E levels. Dysregulated cyclin E decreases expression of EGLN1 in mammary acini. The VHL ubiquitin ligase pathway is a major regulator of HIF-1␣ expression. We thus queried whether the major components of this pathway might be affected by high cyclin E expression in mammary acini. We discovered that dysregulated cyclin E expression in primary cyclin ET74A T393A knock-in mouse mammary cells elicited a significant decrease in the steady-state abundance of the EGLN1 prolyl hydroxy-
lase (Fig. 6A, left), a key positive regulator of HIF-1␣ degradation by the VHL ubiquitin ligase under conditions in which oxygen is present (19, 20). The decreased EGLN1 protein abundance is accompanied by decreased mRNA levels (Fig. 6A, right), suggesting that dysregulated cyclin E may result in repression of EGLN1 gene expression. We obtained comparable results by studying MCF-10A acini comprised of cells transduced with cyclin E(T380A)-expressing retrovirus in comparison to wild-type cyclin E-expressing or control acini (Fig. 6B). Because E2F transcription factors can function as repressors as well as activators (5a), we studied the effect of expressing the dominant-negative E2F-1–E132 mutant in cyclin E(T380A) mammary acini on HIF1A and EGLN1 transcript levels. In contrast to HIF1A message levels, which were sensitive to E2F-1–E132 expression, EGLN1 transcript levels were not altered significantly by E2F-1–E132 expression (Fig. 6C). Thus, we conclude that EGLN1 expression is likely downregulated via an E2F-independent mechanism in response to dysregulated cyclin E expression. To demonstrate the functional importance of EGLN1 down-
FIG. 5. HIF1A is a direct transcriptional target of E2F. (A) Expression of the indicated E2F target genes was analyzed using RNA prepared from day 8 acini. Insets show abundance of total and phosphorylated Rb protein in extracts made from the indicated acinar cultures. (B) (Top) Schematic representation of the 5⬘ end of the HIF1A gene. The region spanning nucleotides ⫺2180 to ⫹28 (relative to the transcription start site) was cloned for use in all reporter assays. The gray box represents nucleotides ⫺255 to ⫺56, which contain six predicted E2F binding motifs indicated by the black bars. The dark box (*) represents positions ⫺89 to ⫺74, containing overlapping, predicted E2F binding motifs that were mutated in experiments described in the legend to panel D. (Bottom) Nucleotide sequence of the E2F-binding site-containing region within HIF1A, with positions ⫺89 to ⫺74 indicated in bold typeface. (C) (Top) 293T cells transiently transfected with expression plasmids for the HIF1A promoter reporter, Renilla luciferase, and either E2F-1 or the E2F-1–E132 mutant were assayed for the activation of the HIF1A reporter. (Bottom) MCF-10A cells stably expressing HIF1A promoter reporter and the indicated E2F-expressing constructs were plated for acinar morphogenesis. Following 8 days, luciferase activity was measured. Error bars indicate standard deviations from three separate experiments. Insets show expression of wild-type E2F-1 and the E132 mutant. (D) 293T cells were transfected with expression plasmids for wild-type E2F-1, Renilla luciferase, and the wild-type HIF1A promoter reporter, a deletion mutant that removes the ⫺255 to ⫺56 region (⌬), or another mutant reporter containing nucleotide substitutions within positions ⫺89 to ⫺74 (*). Reporter activities are displayed relative to the activation of the wild-type reporter by E2F-1. (E) MCF-10A cells stably expressing the indicated HIF1A promoter reporters and cyclin E(T380A)-expressing or empty (⫺) vector were plated for acinar morphogenesis. Following 8 days, luciferase activity was measured, and change in reporter activation with cyclin E(T380A) expression was normalized to the control’s activity level within each reporter set. (F) Quantitative ChIP analysis of endogenous E2F-1 occupancy at the HIF1A promoter versus a 10-kb-upstream region was performed with MCF-10A cells transduced with the indicated constructs. Normal mouse IgG was used for the control immunoprecipitations. Average values from two different experiments are shown. (G) Cyclin E kinase and endogenous CDK2 activities using purified histone H1 as substrate were measured for protein extracts prepared from the indicated MCF-10A acini. Activity levels were quantified from autoradiograms using ImageJ and are expressed relative to those measured in control acini. (H) MCF-10A cells stably expressing the HIF1A promoter reporter were transduced with the indicated vectors, then grown in morphogenesis cultures for 8 days, and assayed for transactivation of the reporter as indicated. Insets show expression of cyclin E and the E2F-1–E132 mutant. 3892
VOL. 31, 2011
HIF-1 INDUCTION BY DYSREGULATED CYCLIN E
3893
FIG. 6. Dysregulated cyclin E downregulates EGLN1 expression in mammary acini. (A) Epithelial cells were isolated from day 15 p.c. wild-type (wt) and cyclin ET74A T393A mammary glands. Protein extracts were prepared and immunoblotted for cyclin E, EGLN1, and GRB2 (left), with quantitation shown for three independent experiments, and RNA was purified for EGLN1 transcript analysis. (B) (Left) Protein extracts were prepared from day 8 acini, and cyclin E, EGLN1, and GRB2 expression was measured by immunoblotting. (Right) Analysis of EGLN1 expression is shown from monolayer cells or day 8 MCF-10A acini transduced with the indicated constructs. (C) RNA was prepared from day 8 MCF-10A acini comprised of cells transduced with the indicated vectors, and the expression of HIF1A and EGLN1 was measured as described above. (D) MCF-10A cells were transduced with the indicated retroviral constructs. Protein extracts from day 8 acini were immunoblotted for EGLN1, HIF-1␣, and GRB2. (E) RNA prepared from MCF-10A acini transduced similarly to those described in the legend to panel D was analyzed for the expression of CDKN1B, VEGFA, and SCL2A1.
regulation by cyclin E, we transduced control and cyclin E(T380A)-expressing MCF-10A cells to enforce EGLN1 expression. We found that HIF-1␣ protein levels and induction of HIF-1 target genes are reduced by overexpressed EGLN1 in cyclin E(T380A) acini (Fig. 6D and E). Thus, EGLN1 downregulation likely accounts for part of the net activation of HIF-1 associated with dysregulated cyclin E activity. These data suggest that at least two cooperative mechanisms promote HIF-1␣ accumulation and increased HIF-1 target gene expression in response to dysregulated cyclin E in mammary epithelial cells (Fig. 7).
expression, at least in mammary epithelial cells. In contrast to its minimal effect in monolayer MCF-10A cells, dysregulated cyclin E elicits a hypoproliferative response in mammary acini, characterized by upregulation of CDK inhibitors p21 and p27 and induction of cellular senescence. Our findings of divergent phenotypes associated with dysregulated cyclin E activity in monolayer versus organotypic growth conditions recall earlier observations that growth of mammary epithelial cells in the extracellular matrix induces global changes in gene regulation and epigenetic state, compared to similar cells grown in a
DISCUSSION Cyclin E dysregulation elicits various effects upon cell proliferation, differentiation, and survival in a manner that is highly dependent on tissue type and cell lineage (34). We now have learned that not only cell type but also growth conditions (proliferation within acini versus in a monolayer) play a critical role in determining phenotypes associated with high cyclin E
FIG. 7. Schematic of the proposed cooperative mechanisms by which dysregulated cyclin E regulates HIF-1␣ expression in mammary epithelial cells during morphogenesis.
3894
SENGUPTA ET AL.
monolayer (26). In the context of our studies, we believe that coregulatory signals, so far undefined, that derive specifically from cell growth within the organizing epithelial structure participate with E2F in the activation of HIF-1␣, since E2F-chromatin binding activity is necessary (Fig. 5H and 6C) but not sufficient for induction of HIF1A gene expression. We found that the induction of p21 and p27 in cyclin E(T380A)-expressing acini is dependent on HIF-1␣, which has been found by others to upregulate expression of both these CDK inhibitors (19, 20, 25). HIF-1␣ induction is identified (by IHC) only in hypomorphic cyclin E(T380A) acinar structures, consistent with its role in directly impairing normal morphogenesis of mammary acini. That cyclin E(T380A)-induced morphogenesis failure is dependent on high-level induction of the CDK inhibitors (Fig. 2B and 4H) is underscored by our observation that wild-type cyclin E expression, which induces lower levels of p21 and p27, has no significant effect on acinar formation (Fig. 1A). Analogous to our prior observations in studies of cyclin E-associated genomic instability in primary fibroblasts and hematopoietic cell maturation defects in vivo (33, 34), the significant functional consequences for gene expression and morphogenesis in the MCF-10A acinar cultures with expression of Fbw7-resistant cyclin E compared to wild-type cyclin E are belied by the modest differences in steady-state abundance of the proteins (Fig. 1D). This is likely because disabling Fbw7-dependent cyclin E turnover causes sustained cyclin E-CDK2 activity throughout the cell cycle that is not easily appreciated by measuring protein abundance in asynchronous cells (15, 34), with a profound impact on downstream targets such as E2F transcriptional activity. We used several methods to establish that HIF-1␣ induction by dysregulated cyclin E in mammary acini is E2F dependent. First, we showed that a HIF1A reporter could be activated directly by E2F-1. Next, we identified E2F-1 occupancy within the HIF1A promoter, which was increased in cells expressing cyclin E(T380A). Finally, we used a dominant-negative E2F mutant to show that endogenous E2F activity regulates both activation of the HIF1A promoter reporter and increased HIF1A gene expression in cyclin E(T380A) mammary acini. Thus, we have demonstrated that HIF1A is a bona fide target of E2F activity and that deregulated E2F activity, in this case associated with increased cyclin E expression, results in increased HIF1A gene expression. We also learned that high cyclin E reduces expression of EGLN1, a critical component of the VHL pathway that controls HIF-1␣ stability. In mammary epithelial cells isolated from cyclin ET74A T393A mice, compared to wild-type controls, we see increased HIF1A gene expression and HIF-1␣ protein abundance, along with increased HIF-1 target gene expression and downregulation of EGLN1 expression. Therefore, many of the consequences of dysregulated cyclin E upon HIF-1 in mammary epithelial cells do not appear to be specific only to MCF-10As. One important difference in the effect of HIF-1 activation on organotypic cultures versus acinar morphogenesis in vivo is that in the former context, HIF-1 activation is associated with increased cellular senescence, whereas with the mammary tissues of the cyclin ET74A T393A knock-in mice, we observed increased apoptosis in association with increased proliferation (34). Indeed, HIF-1 activation has been associated with increased apoptosis in some cell types (4), and we are now evaluating the role of
MOL. CELL. BIOL.
HIF-1␣ in the increased apoptosis phenotype found in cyclin E knock-in murine mammary epithelial tissues. Our results prompt several questions that we are investigating in ongoing work. For one, a subset of cyclin E(T380A) acini are large and hyperproliferative and have low HIF-1␣ expression, suggesting that some cells expressing high cyclin E levels evade HIF-1-dependent senescence. It is possible that cells within acini that appear to “escape” cyclin E-associated morphogenesis arrest do not engage a critical coregulatory mechanism that cooperates with E2F to activate HIF1A expression. Alternatively, the mechanism through which high cyclin E downregulates EGLN1 expression may be less robustly engaged in the hyperproliferative cyclin E(T380A) acini. We are currently working on defining the key regulatory elements within the EGLN1 gene that mediate repression associated with dysregulated cyclin E expression, as misregulation of EGLN1 activity may be an important mechanism for oncogenic activation of HIF-1. Determining the significance of our findings to mammary tumor biology is another focus of our ongoing studies. We speculate that HIF-1␣ induction by dysregulated cyclin E in mammary epithelial cells in vivo may have far-reaching effects on processes such as cell transformation and metastasis, through the role of HIF-1 in regulating numerous genes involved in cell metabolism and angiogenesis. We are now developing animal models to understand the degree to which dysregulated cyclin E and E2F activities co-opt hypoxia response mechanisms during mammary tumorigenesis. ACKNOWLEDGMENTS We thank Joan Brugge and her laboratory (Harvard Medical School) for providing instruction in organotypic mammary epithelial cell cultures, with support from the National Cancer Institute. Navdeep Chandel (Northwestern University) provided reagents and expert advice. This work was supported by NIH grants K22CA130984 (A.C.M.) and R01CA102742 (B.E.C.), as well as funding from the Schweppe and Zell Family Foundations (to A.C.M.). REFERENCES 1. Bartkova, J., et al. 2005. DNA damage response as a candidate anti-cancer barrier in early human tumorigenesis. Nature 434:864–870. 2. Bartkova, J., et al. 2006. Oncogene-induced senescence is part of the tumorigenesis barrier imposed by DNA damage checkpoints. Nature 444:633–637. 3. Bornstein, G., et al. 2003. Role of the SCFSkp2 ubiquitin ligase in the degradation of p21Cip1 in S phase. J. Biol. Chem. 278:25752–25757. 4. Carmeliet, P., et al. 1998. Role of HIF-1alpha in hypoxia-mediated apoptosis, cell proliferation and tumour angiogenesis. Nature 394:485–490. 5. Chen, Z., V. B. Indjeian, M. McManus, L. Wang, and B. D. Dynlacht. 2002. CP110, a cell cycle-dependent CDK substrate, regulates centrosome duplication in human cells. Dev. Cell 3:339–350. 5a.Chong, J. L., et al. 2009. E2F1-3 switch from activities in progenitor cells to repressors in differentiating cells. Nature 462:930–934. 6. Clurman, B. E., R. J. Sheaff, K. Thress, M. Groudine, and J. M. Roberts. 1996. Turnover of cyclin E by the ubiquitin-proteasome pathway is regulated by cdk2 binding and cyclin phosphorylation. Genes Dev. 10:1979–1990. 7. Debnath, J., and J. S. Brugge. 2005. Modelling glandular epithelial cancers in three-dimensional cultures. Nat. Rev. Cancer 5:675–688. 8. Debnath, J., S. K. Muthuswamy, and J. S. Brugge. 2003. Morphogenesis and oncogenesis of MCF-10A mammary epithelial acini grown in three-dimensional basement membrane cultures. Methods 30:256–268. 9. Di Micco, R., et al. 2006. Oncogene-induced senescence is a DNA damage response triggered by DNA hyper-replication. Nature 444:638–642. 10. Dimri, G. P., et al. 1995. A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proc. Natl. Acad. Sci. U. S. A. 92:9363– 9367. 11. Ekholm-Reed, S., et al. 2004. Deregulation of cyclin E in human cells interferes with prereplication complex assembly. J. Cell Biol. 165:789–800. 12. Gardner, L. B., et al. 2001. Hypoxia inhibits G1/S transition through regulation of p27 expression. J. Biol. Chem. 276:7919–7926.
VOL. 31, 2011
HIF-1 INDUCTION BY DYSREGULATED CYCLIN E
3895
13. Geng, Y., et al. 2003. Cyclin E ablation in the mouse. Cell 114:431–443. 14. Goda, N., et al. 2003. Hypoxia-inducible factor 1alpha is essential for cell cycle arrest during hypoxia. Mol. Cell. Biol. 23:359–369. 15. Grim, J. E., et al. 2008. Isoform- and cell cycle-dependent substrate degradation by the Fbw7 ubiquitin ligase. J. Cell Biol. 181:913–920. 16. Hinds, P. W., et al. 1992. Regulation of retinoblastoma protein functions by ectopic expression of human cyclins. Cell 70:993–1006. 17. Huang, H., K. M. Regan, Z. Lou, J. Chen, and D. J. Tindall. 2006. CDK2dependent phosphorylation of FOXO1 as an apoptotic response to DNA damage. Science 314:294–297. 18. Hwang, H. C., and B. E. Clurman. 2005. Cyclin E in normal and neoplastic cell cycles. Oncogene 24:2776–2786. 19. Ivan, M., et al. 2001. HIFalpha targeted for VHL-mediated destruction by proline hydroxylation: implications for O2 sensing. Science 292:464–468. 20. Jaakkola, P., et al. 2001. Targeting of HIF-alpha to the von Hippel-Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science 292: 468–472. 21. Jiang, Y., et al. 2009. An algorithm for identifying novel targets of transcription factor families: application to hypoxia-inducible factor 1 targets. Cancer Inform. 7:75–89. 22. Keyomarsi, K., S. L. Tucker, and I. Bedrosian. 2003. Cyclin E is a more powerful predictor of breast cancer outcome than proliferation. Nat. Med. 9:152. 23. Keyomarsi, K., et al. 2002. Cyclin E and survival in patients with breast cancer. N. Engl. J. Med. 347:1566–1575. 24. Koepp, D. M., et al. 2001. Phosphorylation-dependent ubiquitination of cyclin E by the SCFFbw7 ubiquitin ligase. Science 294:173–177. 25. Koshiji, M., et al. 2004. HIF-1alpha induces cell cycle arrest by functionally counteracting Myc. EMBO J. 23:1949–1956. 26. Le Beyec, J., et al. 2007. Cell shape regulates global histone acetylation in human mammary epithelial cells. Exp. Cell Res. 313:3066–3075. 27. Lee, G. Y., P. A. Kenny, E. H. Lee, and M. J. Bissell. 2007. Three-dimensional culture models of normal and malignant breast epithelial cells. Nat. Methods 4:359–365. 28. Liu, E., X. Li, F. Yan, Q. Zhao, and X. Wu. 2004. Cyclin-dependent kinases phosphorylate human Cdt1 and induce its degradation. J. Biol. Chem. 279: 17283–17288. 29. Loeb, K. R., et al. 2005. A mouse model for cyclin E-dependent genetic instability and tumorigenesis. Cancer Cell 8:35–47. 30. Ma, Y., et al. 2007. Transgenic cyclin E triggers dysplasia and multiple pulmonary adenocarcinomas. Proc. Natl. Acad. Sci. U. S. A. 104:4089–4094. 31. Mailand, N., and J. F. Diffley. 2005. CDKs promote DNA replication origin licensing in human cells by protecting Cdc6 from APC/C-dependent proteolysis. Cell 122:915–926. 32. Majumder, P. K., et al. 2008. A prostatic intraepithelial neoplasia-dependent p27 Kip1 checkpoint induces senescence and inhibits cell proliferation and cancer progression. Cancer Cell 14:146–155.
33. Minella, A. C., J. E. Grim, M. Welcker, and B. E. Clurman. 2007. p53 and SCFFbw7 cooperatively restrain cyclin E-associated genome instability. Oncogene 26:6948–6953. 34. Minella, A. C., et al. 2008. Cyclin E phosphorylation regulates cell proliferation in hematopoietic and epithelial lineages in vivo. Genes Dev. 22:1677– 1689. 35. Minella, A. C., et al. 2002. p53 and p21 form an inducible barrier that protects cells against cyclin E-cdk2 deregulation. Curr. Biol. 12:1817–1827. 36. Mizushima, N., T. Yoshimori, and B. Levine. 2010. Methods in mammalian autophagy research. Cell 140:313–326. 37. Moroni, M. C., et al. 2001. Apaf-1 is a transcriptional target for E2F and p53. Nat. Cell Biol. 3:552–558. 38. Moroy, T., and C. Geisen. 2004. Cyclin E. Int. J. Biochem. Cell Biol. 36: 1424–1439. 39. Okuda, M., et al. 2000. Nucleophosmin/B23 is a target of CDK2/cyclin E in centrosome duplication. Cell 103:127–140. 40. Ovcharenko, I., M. A. Nobrega, G. G. Loots, and L. Stubbs. 2004. ECR Browser: a tool for visualizing and accessing data from comparisons of multiple vertebrate genomes. Nucleic Acids Res. 32:W280–W286. 41. Rabinovich, A., V. X. Jin, R. Rabinovich, X. Xu, and P. J. Farnham. 2008. E2F in vivo binding specificity: comparison of consensus versus nonconsensus binding sites. Genome Res. 18:1763–1777. 42. Rajagopalan, H., et al. 2004. Inactivation of hCDC4 can cause chromosomal instability. Nature 428:77–81. 43. Saavedra, H. I., et al. 2003. Inactivation of E2F3 results in centrosome amplification. Cancer Cell 3:333–346. 44. Semenza, G. L. 2010. Defining the role of hypoxia-inducible factor 1 in cancer biology and therapeutics. Oncogene 29:625–634. 45. Sheaff, R. J., M. Groudine, M. Gordon, J. M. Roberts, and B. E. Clurman. 1997. Cyclin E-CDK2 is a regulator of p27Kip1. Genes Dev. 11:1464–1478. 46. Smith, A. P., et al. 2006. Deregulated cyclin E promotes p53 loss of heterozygosity and tumorigenesis in the mouse mammary gland. Oncogene 25:7245– 7259. 47. Strohmaier, H., et al. 2001. Human F-box protein hCdc4 targets cyclin E for proteolysis and is mutated in a breast cancer cell line. Nature 413:316–322. 48. Subramanian, A., et al. 2005. Gene set enrichment analysis: a knowledgebased approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U. S. A. 102:15545–15550. 49. Thomas, D. M., et al. 2004. Terminal osteoblast differentiation, mediated by runx2 and p27KIP1, is disrupted in osteosarcoma. J. Cell Biol. 167:925–934. 50. Welcker, M., and B. E. Clurman. 2008. FBW7 ubiquitin ligase: a tumour suppressor at the crossroads of cell division, growth and differentiation. Nat. Rev. Cancer 8:83–94. 51. Welcker, M., et al. 2003. Multisite phosphorylation by Cdk2 and GSK3 controls cyclin E degradation. Mol. Cell 12:381–392. 52. Zhao, J., et al. 2000. NPAT links cyclin E-Cdk2 to the regulation of replication-dependent histone gene transcription. Genes Dev. 14:2283–2297.
