Oncogene (2006) 25, 588–598 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE NOL7 is a nucleolar candidate tumor suppressor in cervical cancer that modulates the angiogenic phenotype

R Hasina1, AL Pontier1, MJ Fekete1, LE Martin1,XMQi2, C Brigaudeau3, R Pramanik2, EI Cline2, LJ Coignet4 and MW Lingen1

1Departments of Pathology, Medicine and Radiation and Cellular Oncology, The University of Chicago, Chicago, IL, USA; 2Hines Veterans Affairs Hospital, Maywood, IL, USA; 3Loyola University Medical Center, Maywood, IL, USA and 4Department of Cancer Genetics, Roswell Park Cancer Institute, Buffalo, NY, USA

Cervical cancer is associated with human papilloma virus high-risk HPV is not sufficient to induce transformation infection.However, this infection is insufficient to induce and progression (Chen et al., 1993; Montgomery et al., transformation and progression.Loss of heterozygosity 1995; Uejima et al., 1995). analyses suggest the presence of a tumor suppressor gene Most types of cancer develop in a multi-step fashion (TSG) on 6p21.3–p25. Here we report the in which there is a sequentialactivation of oncogenes cloning NOL7, its mapping to chromosome band 6p23, and inactivation of tumor suppressor (TSG) in the and localization of the to the nucleolus.Fluores- same clone of cells. These genetic changes generate cence in situ hybridization analysis demonstrated an allelic concomitant phenotypic changes in the tumor cells that loss of an NOL7 in cultured tumor cells and human tumor allow them to survive and expand. A number of samples.Transfection of NOL7 into cervical carcinoma potentialCC-associated TSG have been identified cells inhibited their growth in mouse xenografts, confirm- (Lazo, 1999; Wolf and Ramirez, 2001). In addition, ing its in vivo tumor suppressor activity.The induction of chromosomallossof heterozygosity (LOH) has been tumor dormancy correlated with an angiogenic switch consistently observed, including on the short arm of caused by a decreased production of vascular endothelial (6p), suggesting the presence of one or growth factor and an increase in the production of the more TSG. Detailed allelotype studies have indicated angiogenesis inhibitor thrombospondin-1.These data potentialsites of TSG at 6p23, 6p21.3, and 6p25 in CC suggest that NOL7 may function as a TSG in part by (Mitra et al., 1994; Mullokandov et al., 1996; Huettner modulating the expression of the angiogenic phenotype. et al., 1998; Kersemaekers et al., 1998; Rader et al., Oncogene (2006) 25, 588–598. doi:10.1038/sj.onc.1209070; 2000). However, the identity of the TSG located on this published online 3 October 2005 region of 6p has remained elusive. Here we report the cloning of a new gene, NOL7, and Keywords: cervical cancer; nucleolus; tumor suppressor locate it to chromosome band 6p23. In addition, we gene; angiogenesis; VEGF; TSP-1 demonstrate that there is allelic loss of NOL7 in both CC cell lines as well as fresh tumor samples. Transfec- tion of NOL7 into CC cells inhibited their growth in mouse xenografts, confirming that this nucleolar protein product functions as a TSG in human CC. Finally, we Introduction functionally demonstrate that the induction of tumor dormancy is caused in part by a switch in the angiogenic With approximately 470 000 new cases annually, cervical phenotype of the NOL7-transfected cells, which was the cancer (CC) is the most common gynecologic malig- result of a decrease in the production of the angiogenic nancy worldwide and is one of the leading causes of vascular endothelial cell growth factor (VEGF), as well death in women (Cannistra and Niloff, 1996). Infection as an increase in the production of the naturally with the human papilloma virus (HPV) is closely occurring inhibitor of angiogenesis thrombospondin-1 associated with disease and recent reports have found (TSP-1). These data suggest that this novel nucleolar that essentially all cases of CC demonstrate the presence protein may function as a TSG in part by modulating of the virus (Walboomers et al., 1999). However, the expression of the angiogenic phenotype in CC. experimentalevidence has shown that the presence of

Correspondence: Dr MW Lingen, Department of Pathology, The Results University of Chicago, MC 6101, 5841 S. Maryland Avenue, Chicago, IL 60637, USA. Cellular and chromosomal localization of NOL7 E-mail: [email protected] Received 19 April2005; revised 1 August 2005; accepted 1 August 2005; While seeking to identify genes that were differentially published online 3 October 2005 expressed in endothelial cells under different conditions, Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 589

Figure 2 Physicalmapping of NOL7 to the short arm of chromosome 6p. (a) Hybridization pattern of the Cos7 probe containing the entire NOL7 gene on a normalmetaphase spread; (b) the NOL7 gene was mapped to 6p23 using G/DAPI banding as illustrated with the chromosome 6 ideogram.

Figure 1 Subcellular localization of NOL7. Cos7 cells were transiently transfected with GFP-tagged full-length NOL7. (a) DAPI staining for nuclei (blue); (b) GFP-NOL7; (c) anti-B23 antibody staining (red); (d) merged pictures. The yellow areas demonstrate colocalization of NOL7 and B23 confirming the nucleolar localization of NOL7. we recently cloned NOL7, a gene encoding an mRNA of 902 base pairs and a protein with a predicted molecular Figure 3 Northern blot analysis of NOL7 expression in human cervical carcinoma cell lines. Three out of eight tumor cell lines weight of 27 kDa that bore no significant homologies to demonstrate a marked reduction in NOL7 mRNA. Molecular any known family of genes (accession number weight is indicated on the right in kilobases. NM_016167). Northern blot analysis of commercially available human tissue blots demonstrated that NOL7 was expressed in a number of different tissue types throughout the body and was particularly prevalent in 1998; Kersemaekers et al., 1998; Rader et al., 2000). To the adrenal gland, thyroid gland, heart, skeletal muscle, determine if NOL7 might be one of these suppressor and kidney (data not shown). genes, we initially determined levels of its mRNA In order to determine the cellular localization of the expression in CC cell lines. Northern blot analysis of NOL7 protein, we transiently transfected Cos7 cells with eight independent human CC cell lines showed a GFP-tagged full-length NOL7. The chimeric protein decrease in NOL7 mRNA in three of the eight cell lines localized to the nucleus of the cells, particularly to the (Figure 3). This proportion of 38% is consistent with nucleolus (Figure 1). The nucleolar localization of previous findings that approximately 40% of CCs NOL7 was confirmed by immunocytochemistry using demonstrate a LOH for 6p23 (Rader et al., 2000). an antibody that was specific for B23, another indepen- Seeking to define the deletion pattern of the NOL7 dent nucleolar marker (Figure 1). In addition, three gene in tumor cell lines as well as in fresh tumor samples, independent BAC clones containing sequences from the we made use of a smaller 8-kb genomic probe containing NOL7 gene were hybridized to normal metaphase cells the entire NOL7 gene, in combination with a chromo- in order to determine the chromosomallocation of some 6-centromere-specific probe (Figure 4, Table 1). In NOL7. All three clones hybridized to the same all, 600 interphase cells from three different lymphocyte chromosome locus, which diamidino-phenylindole preparations showing two distinct signals for the control (DAPI) banding identified as chromosome 6, band p23 probe were scored as normalcontrolsto define the (Figure 2a and b). This mapping has been confirmed in cutoff values for NOL7 deletion. The threshold for subsequent experiments using a chromosome 6-centro- detection of the NOL7 gene deletion was set to 22.4%, mere-specific probe in combination with the BAC defined by the mean plus three standard deviations of clones. the percentage of nuclei showing only one NOL7 gene signaland two signalsfor the controlprobe. For the Ca Ski cell line, the Cep6 control probe exhibited five copies Allelic loss of NOL7 in CC cell lines and human tumor of the centromeric region of chromosome 6. Two of samples these five signals hybridized on identical short deriva- LOH on the short arm of chromosome 6 (6p) is tives of chromosome 6. Rearrangements involving the frequently observed in CC, suggesting the presence of entire short arm of chromosome 6 could explain the one or more TSG on this chromosomalarm (Mitra absence of NOL7 signal. A deletion of one or two NOL7 et al., 1994; Mullokandov et al., 1996; Huettner et al., signals on the three remaining copies of chromosome 6

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 590

Figure 4 Demonstration of allelic loss of NOL7 in cervical carcinoma cell lines and human tumors. FISH experiments were performed on: (a) ME-180 CC cell line hybridized with the RPCI1-55N8 BAC clone; (b, c) ME-180 and SiHa CC cell lines hybridized with an 8-kb NOL7 gene-specific probe; and (d–f) three consecutive CC samples hybridized with the 8-kb NOL7 genomic probe. NOL7 probes were labeled with bio-11-dUTP, and detected with FITC. NOL7 probes were cohybridized with Spectrum Orange directly labeled probe for the centromere of chromosome 6 (Cep6). Experiments demonstrate loss of at least one NOL7 allele in both the tumor cell lines as well as the tumor samples.

Table 1 Results of FISH analysis on Ca Ski, ME-180 and SiHa cervical cancer-derived cell lines and three cervical cancer patientsa Cep 6 signal 1 2 3 4 5

NOL7 signal 1 1 2 2 1 2 1 2 3

Ca Ski 54 36 10 ME-180 92 8 SiHa 81 19 Patient 1 34 28 38 Patient 2 11 43 21 3 9 13 Patient 3 8 44 37 3 8

aFor each patient sample, at least 200 interphase nuclei were analysed and the percentages reported.

was observed in 90% of interphase nuclei (Table 1). The cells (two Cep6 signals) having loss of one NOL7 signal BAC clone RPCI1-55N8 did not demonstrate a signal (Figure 4). In all, 25 and 11% of nuclei exhibited a pattern deletion in the ME-180 cell line (Figure 4). hyperdiploid pattern (three and four Cep6 signals), However, when hybridized with an 8-kb NOL7-specific where the NOL7 gene was always deleted with loss of probe, both the ME-180 and SiHa cell lines demon- one, two or three signals. Overall, for patients 2 and 3, strated abnormalsignalpatterns in 92 and 81% of the 79 and 63% of interphase nuclei demonstrated a loss of interphase nuclei, respectively. NOL7 signals, respectively. To determine whether the The same 8-kb NOL7 genomic probe was hybridized high number of nuclei with one Cep6 signal (patient 1) to interphase nuclei from three consecutive CC-derived was due to an aneuploid pattern or chromosome 6 tissue samples. Their overall ploidy was determined abnormalities, we hybridized all patient samples with a using the Cep6 centromeric probe. Patient 1 showed a centromeric chromosome 12 probe (data not shown). high percentage (34%) of nuclei with one Cep6 control This probe was selected because chromosome 12 was signalassociated with onlyone NOL7 signal(Figure 4). not reported as frequently implicated in chromosomal When two controlsignalswere counted, the NOL7 gene aberrations of CC (Mitra et al., 1994). Hybridization was deleted in 28% of nuclei. Overall, 62% of cells results confirmed that patient 1 had a diploid pattern, exhibited an abnormalNOL7 profile.Patients 2 and 3 with 93% of cells exhibiting two Cep12 probe signals. A showed a similar pattern, with 43 and 44% of diploid diploid profile associated with a clonal evolution toward

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 591 a tetraploid pattern was found in patients 2 and 3. These results indicate that the high frequency of Cep6 probe loss was probably due to a rearrangement involving the entire chromosome 6 (monosomy) or due to a chromo- some 6 short arm aberration, including the pericentro- meric Cep6 locus (6p11–6q11).

Suppression of in vivo tumor growth by NOL7 In order to test the hypothesis that NOL7 is able to function as a tumor suppressor in cervicalmucosa, the full-length NOL7 cDNA was cloned into the pcDNA3.1/V5-His-TOPO. vector and stably transfected into SiHa cells. As a negative control, the LacZ gene was also cloned into the pcDNA3.1/V5-His-TOPO vector and stably transfected into SiHa cells. Individual clones were selected, ring-cloned and expanded from individual100 mm dishes. Expression of the transfected NOL7 was confirmed via Western blot analysis using an antibody for the V5 tag (Figure 5a). In vitro proliferation assays were performed in order to determine if the introduction of NOL7 into CC cells would alter their proliferation profile. The NOL7- containing SiHa clones proliferated at a rate that was similar to both the parental SiHa cell line as well as the LacZ-containing SiHa clones (Figure 5b), suggesting that the introduction of NOL7 into the tumor cells had no effect on their in vitro proliferation. However, the subcutaneous injection of NOL7-containing SiHa clones into nude mice garnered significantly different results. Tumorigenicity was significantly reduced in the NOL7 clones when compared to the parental cell line or the LacZ-containing SiHa cells (Figure 5c).

NOL7 represses the angiogenic phenotype by modulating the expression of VEGF and TSP-1 The observation that the introduction of NOL7 into CC cells altered their in vivo, but not in vitro, growth characteristics suggested that the expression of this gene might contribute to the regulation of the angiogenic phenotype. To determine if the induction of tumor Figure 5 Tumor formation of CC cells transfected with NOL7 in nude mice. (a) Stable independent clones of the SiHa cell line dormancy in the NOL7-containing cells was related to a expressing the full-length NOL7 cDNA were generated and verified switch in the angiogenic phenotype, we first evaluated via Western blot using an antibody to the V5 sequence tag present the rates of proliferation and apoptosis as well as the in the vector; (b) in vitro proliferation of SiHa parental, LacZ, and microvesseldensity of the tumor xenografts. The NOL7 NOL7 clones; (c) in vivo tumor growth of SiHa parental, LacZ, and NOL7 transfected cells inoculated in nu/nu mice and monitored tumors as well as the parental and LacZ tumors were over a period of 49 days (n ¼ 12 animals per group). The in vitro found to contain both proliferating cells (Figure 6) as proliferation of the three cell types was similar, while the in vivo well as cells undergoing apoptosis (Figure 6). However, tumor growth of the NOL7-transfected cells was significantly while the proliferation indices of the three cell types were decreased when compared to the SiHa- and LacZ-containing cells. similar, the apoptotic index of the NOL7-containing cells was much greater (Table 2). This suggested that the inability of the NOL7-containing tumor cells to grow beyond a certain threshold in size was not due to a lack was observed in the parentaland LacZ clones(Figure 6, of tumor cell proliferation. Rather, it was likely due to a Table 2). The combined findings of an imbalance in the balance in the rate of proliferating cells and cells rates of proliferation/apoptosis as well as altered undergoing apoptosis. To revealmicrovessels,tumor microvesseldensities suggest that one of the mechanisms sections were stained for CD31, a marker that is by which NOL7 acts as a tumor suppressor may be by expressed on endothelial cells. Rare CD31-positive causing a switch in the angiogenic phenotype. staining vessels were found scattered within the dormant Whether or not angiogenesis occurs depends on the NOL7 tumors (Figure 6, Table 2). Conversely, a robust relative levels of inducers and inhibitors of neovascular- angiogenic response, as determined by CD31 positivity, ization present within the microenvironment surround-

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 592

Figure 6 Histological analysis of SiHa parental and NOL7 tumor xenografts. SiHa parental and NOL7-transfected cells were injected s.c. in nude mice. After 7 weeks tumors were resected and prepared for histology. The histological sections were stained with H&E ( Â 25) or stained for proliferating cells (anti-PCNA, Â 400), apoptotic cells (cleaved caspase-3, Â 400), and microvesseldensity (anti- CD31, Â 400). The SiHa parental and NOL7 cells had a similar rate of proliferation, while the NOL7 transfectants had a higher degree of apoptosis and lower microvessel densities.

Table 2 Quantitation of proliferation, apoptosis, and angiogenesis of parental, LacZ, and NOL7-transfected tumor cells in mouse tumor xenograftsa Parental LacZ NOL7

% Proliferation 30.175.2 28.675.5 (Po0.7)* 29.174.8 (Po0.7)* % Apoptosis 1.571.3 1.871.2 (Po0.7)* 7.771.5 (Po0.0001)** Vesselcount 87.6 711.2 86.8713.3 (Po0.7)* 6.272.5 (Po0.000001)**

aValues represent mean7standard error. *Statistically not significant as compared to parental cells. **Statistically significant as compared to parental cells. n ¼ 12 tumors/group.

ing the endothelial cells. To better define the mechanism angiogenesis in vitro using the endothelial cell migration of the angiogenic switch observed in in vivo tumor assay. CM from the parental cell line as well as the LacZ xenografts, serum-free conditioned media (CM) from SiHa clones strongly stimulated endothelial cell migra- the parental as well as NOL7 and control LacZ clones tion (Figure 7). However, CM from the NOL7- were collected and assayed for their ability to stimulate transfected SiHa clones had a significantly reduced

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 593

Figure 7 In vitro angiogenic activity of parentaland NOL7- transfected SiHa cells. Serum-free CM from parental SiHa, LacZ, and NOL7-transfected clones were functionally tested for their ability to stimulate endothelial cell migration. CM alone from the SiHa parental and the LacZ clones (open bars) strongly stimulated endothelial cell migration in a fashion similar to VEGF alone, while the NOL7 clones demonstrated a significant reduction in the stimulation of migration. In addition, CM from the NOL7 clones were able to inhibit the induction of endothelial cell migration by VEGF (filled bars), suggesting that these cells were secreting a protein that had angioinhibitory activity.

ability to stimulate endothelial cell migration. In addition to having a decreased capacity to induce endothelial cell migration, the CM from the NOL7 clones contained high levels of one or more molecules that were capable of inhibiting angiogenesis. When the NOL7 CM was mixed 1:1 with media from either the parental or the LacZ clones, it was able to block angiogenesis, as measured by the in vitro endothelial cell migration assay (data not shown). These results suggest that a switch from an angiogenic to an angioinhibitory phenotype is one of the potentialmechanisms by which the introduction of NOL7 into the SiHa CC cell line resulted in a change in the angiogenic status. We subsequently performed genomic profiling on the SiHa parentaland NOL7 clonesin order to obtain global signatures of angiogenesis-related genes whose Figure 8 Introduction of NOL7 into cervical carcinoma cell lines expression might be altered by introduction of NOL7. alters the angiogenic phenotype by modulating the expression of The expression data demonstrated that transfection of VEGF and TSP-1. (a) Western blot analysis of serum-free CM NOL7 into the parental cells resulted in an upregulation from SiHa and NOL7-transfected cells for expression of VEGF and TSP-1. Significant upregulation of TSP-1 and downregulated of TSP-1 and a concomitant downregulation of VEGF expression of VEGF is observed in both of the NOL7-transfected (please see supplemental material). In order to validate clones. Serum-free CM from U87 and HT1080 cells were used as these findings, we performed Western blot analyses for positive controls for VEGF and TSP-1, respectively. (b) SiHa VEGF and TSP-1 on CM from the parentalSiHa and parental and NOL7-transfected cells were injected s.c. into nude the NOL7-containing cells. The CM from the parental mice. After 7 weeks tumors were resected and prepared for histology. The histological sections were incubated with antibodies SiHa tumor cells contained high levels of VEGF and for VEGF ( Â 400) and TSP-1 ( Â 400), demonstrating a significant extremely low levels of TSP-1 (Figure 8a), while the in vivo upregulation of TSP-1 and downregulation of VEGF in NOL7 transfectants demonstrated a decreased expres- NOL7-transfected cells in vivo;(c) serum-free CM from parental sion of VEGF and an increased expression of TSP-1. SiHa cells as well as NOL7-transfected clones were functionally tested for their ability to stimulate endothelial cell migration in the Similarly, we performed immunohistochemistry for presence of neutralizing antibody to either VEGF or TSP-1. These VEGF and TSP-1 on the SiHa parentaland the results functionally validate the role of TSP-1 and VEGF in CC. NOL7-containing tumor xenografts. The SiHa parental Furthermore, they provide further support for the modulation of tumors demonstrated high levels of VEGF expression these two by NOL7 in the context of CC.

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 594 and low levels of TSP-1 in vivo (Figure 8b). Conversely, Traditionally, the nucleolus has been considered to be the NOL7 clones expressed markedly reduced amounts a subnuclear zone whose sole function was related to of VEGF and considerably higher levels of TSP-1 than ribosomaltranscription and subunit assembly.Recent parentals in vivo (Figure 8c). findings have suggested that the nucleolus may play an In order to determine if the altered levels of VEGF important additionalrolein a number of physiologic and TSP-1 were biologically important for the angio- and pathologic processes such as regulation of cell cycle, genic switch observed in the NOL7 clones, we apoptosis, aging, and cancer biology (Olson et al., 2000; performed additional in vitro endothelial cell migration Visintin and Amon, 2000). For example, NOL8, another assays using serum-free CM with neutralizing antibodies novel nucleolar protein, was recently reported to act as to either VEGF or TSP-1 (Figure 8c). As anticipated, an oncogene in diffuse-type gastric carcinomas (Jina- the angiogenic activity of the parentaland LacZ clones wath et al., 2004). Conversely, the p14ARF TSG protein was due primarily to the secretion of VEGF, because was initially characterized as such for its ability to neutralizing antibodies to VEGF reduced the ability of inactivate mdm2, the negative regulator of the p53 the CM from these cells to induce the migration of transcription factor, by relocalizing it to the nucleolus endothelial cells. The combination of CM and anti- (Weber et al., 1999). In addition, it has recently been VEGF resulted in a 60% reduction in the ability of CM shown that p14 can also act as a tumor suppressor by from either the SiHa parentalor the LacZ clonesto inhibiting ribosomalRNA processing by retarding induce endothelial cell migration. However, anti-VEGF processing of 47/45S and 32S precursors (Sugimoto did not alter the migration profile of the NOL7 clones. et al., 2003). To our knowledge, this is the only other This observation is in keeping with our in vivo findings known TSG whose protein product localizes to the that the NOL7-containing clones express low levels of nucleolus. VEGF, while the parental and LacZ clones express high Whether or not angiogenesis occurs in a particular levels of this cytokine. Conversely, the antiangiogenic tissue depends upon the balance between the relative activity of the NOL7 clones was due primarily to the amounts of molecules that induce and molecules that secretion of TSP-1. The addition of TSP-1-neutralizing inhibit angiogenesis (Bouck et al., 1995). In normal antibody to the NOL7 CM relieved the inhibitory tissues, blood vessels are usually quiescent and cells activity present in the CM from the NOL7 clones, while secrete low levels of inducers and high levels of addition of anti-TSP-1 had no effect on the SiHa inhibitors. As normal cells progress towards malig- parentalor LacZ clones(Figure 8c). nancy, they must develop the ability to induce angio- genesis. In order to achieve this switch, tumor cells increase the amount of inducers and decrease the amount of inhibitors they secrete. There is considerable Discussion interest in determining how cells, progressing from normalto tumorigenic, switch from being antiangio- CC continues to be a major worldwide health problem. genic to angiogenic. In some animalmodels, a distinct HPV is recognized as a major contributor to the switch to the angiogenic phenotype is seen (Hanahan development of CC. It is now well established that and Folkman, 1996). In other cases, the cells developing high-risk HPVs containing the E6 and E7 viralgenes can into tumors sequentially become more angiogenic in a disrupt the cell cycle by altering the function of the p53 stepwise fashion (Lingen et al., 1997; Volpert et al., and Rb TSG (Lazo, 1999; Wolf and Ramirez, 2001). 1997). The exact mechanisms regarding how this occurs HPV E6 induces rapid degradation of p53 via ubiquitin- in most neoplasms have for the most part remained dependent proteolysis. Similarly, HPV E7 complexes elusive. However, a growing body of evidence suggests with the hypophosphorylated form of Rb, causing its that the activation of oncogenes as well as the proteolysis via the proteosome pathway. However, there inactivation of TSG play an important role in this are still many aspects regarding the molecular mechan- phenotypic switch. For example, expression of VEGF isms of CC that remain unanswered, such as the identity can be stimulated by various oncogenic stimuli such as of other genetic alterations required for CC progression. ras, v-src, c-fos, and c-jun (Mukhopadhyay et al., 1995; A number of TSG have been postulated to play a role. Rak et al., 1995; Kraemer et al., 1999; Marconcini et al., Here we have provided functional in vitro and in vivo 1999). Similarly, inactivation of the p53, VHL, and evidence that NOL7 is the candidate TSG located at RB2/p130 TSG results in increased VEGF production 6p23. Northern blot analysis demonstrated a decreased (Mukhopadhyay et al., 1995; Gnarra et al., 1996; expression of NOL7 mRNA in approximately 40% of Claudio et al., 2001). Conversely, activation of onco- CC cell lines. Furthermore, FISH analysis revealed an genes such as c-jun, v-myc, v-src, and ras, or the allelic loss of NOL7 in cultured tumor cells and human inactivation of the p53, Rb, PTEN, and nm23 TSG, tumor samples. Transfection of NOL7 into CC cells results in a decreased expression of TSP-1 (Dameron inhibited their growth in mouse xenografts, confirming et al., 1994; Slack and Bornstein, 1994; Zabrenetzky its in vivo tumor suppressor activity. Finally, the et al., 1994; Gnarra et al., 1996; Sheibani and Frazier, induction of tumor dormancy correlated with an 1996; Tikhonenko et al., 1996; Dejong et al., 1999; Wen angiogenic switch caused by decreased production of et al., 2001; Watnick et al., 2003). Data presented here VEGF and an increase in the production of the show that NOL7, like p53, can modulate both sides of angiogenesis inhibitor TSP-1. the angiogenic balance by increasing the expression of

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 595 TSP-1 and decreasing the expression of VEGF. How- lines were cultured in Eagle’s Minimal Essential Medium with ever, since p53 is inactivated as a result of HPV infection 10% FBS. ME-180 and HEC-1-A cell lines were cultured in in CC, and is functionally inactivated via degradation in McCoy’s 5A medium with 10% FBS. The C-4 I cell line was the SiHa cell line used in these experiments (Yaginuma cultured in Waymouth’s 752/1 medium with 10% FBS. The and Westphal, 1991; Srivastava et al., 1992), the HeLa and Cos7 cell lines were cultured in DMEM with 10% FBS. All cells were cultured at 371C in a 5% CO2–95% air modulations of both VEGF and TSP-1 by NOL7 environment in humidified incubators. All media and reagents appear to be p53-independent events in CC. It is also were purchased from Invitrogen (Carlsbad, CA, USA). interesting to note that the oncogenic stimulus of Serum-free CM was generated by rinsing the cells with PBS hepatocyte growth factor (HGF) modulates VEGF once, RPMI three times, incubating them in RPMI for 4 h, and TSP-1 in the opposite fashion (Zhang et al., 2003) refeeding them with RPMI and collecting the CM 48 h later. in other tumor types. Therefore, it is possible that The CM were concentrated and dialysed using Centriprep-3 overexpression of HGF in conjunction with the loss of concentrators (Amicon Inc., Beverly, MA, USA). Protein NOL7 in CC may represent a coordinated TSG/ concentrations were determined using the Coomassie Protein oncogene switch that contributes to the induction of Assay Reagent 23200 (Pierce BiomedicalCompany, St Louis, the angiogenic phenotype in CC. However, to our MO, USA). knowledge, the expression of HGF has not been thoroughly investigated in the setting of CC. Similarly, mRNA Isolation and Northern blots detailed analyses of the potential role for other TSG/ mRNA was isolated with Oligotex Direct mRNA Mini Kit (Qiagen, Valencia, CA, USA). In all, 2 mg of denatured RNA oncogenes in the modulation of VEGF and TSP-1 in the was loaded onto a formaldehyde agarose gel and blotted onto context of CC have not been performed to date. Further a GeneScreen Plus membrane (NEN Life Science/Perkin- investigations into these possibilities are warranted. Elmer, Boston, MA, USA). The membrane was hybridized At this time, the status of the second NOL7 allele is with the [alpha-32P]dATP-labeled (Amersham, Piscataway, NJ, undetermined. From preliminary RTQ–PCR experi- USA) NOL7 probe at 681C for 1 h. ments, we do know that the parental cell lines of Ca Ski, ME-180, and SiHa continue to produce NOL7 Chromosomal localization mRNA. However, NOL7 expression in the parental cells The human RPCI1 BAC library in the vector pCYPAC2, with is approximately 60% less than a number of different an average insert size of 110 kb, was obtained from the UK normal cell types containing two NOL7 alleles (data not HGMP Resource Centre (Cambridge, UK) and screened to shown). Therefore, it is possible that the loss of function obtain NOL7-specific genomic probes. Three gene-specific is due to a mutation of the remaining allele or the result positive individualclones,223-E5, 80-J10, and 113-O18, were of haplo-insufficiency. We are presently investigating identified on the filters. The clones were obtained from the this issue, as well as determining when during the HGMP Resource Centre, DNA was extracted with the QiaFilter Maxi DNA extraction kit (Qiagen, Valencia, CA, carcinogenic process the allele is lost. We also do not USA) and their gene specificity was confirmed by Southern know what the incidence of NOL7 inactivation in other blot. FISH was performed according to standard protocols as malignancies may be. A deletion of 6p23 has been described previously (Coignet et al., 1999). Briefly, the probes observed in a number of different types of myeloid and were labeled by nick translation with digoxigenin-11-deoxyur- lymphoid malignancies (Nagai et al., 1999; Chen et al., idine triphosphate (Roche, Indianapolis, IN, USA). Competi- 2000; Nakase et al., 2000; Takeshita et al., 2004). tion hybridization by preannealing with a 100–200-fold excess However, to our knowledge, data regarding a detailed of Cot-1 DNA (Invitrogen, Carlsbad, CA, USA) was cytogenic analysis of the short arm of chromosome 6 in performed before overnight hybridization to normal cells other malignancies are limited. Finally, it is presently complement chromosome with previously denatured DNA. unknown if alterations in NOL7 are the consequence of Probes were visualized by incubation with sheep antidigox- igenin–fluorescein isothiocyanate (FITC) (Roche, Indianapo- biological selection for specific chromosomal locations lis, IN, USA). Cells and were counterstained containing integrated HPV DNA or the result of other with DAPI and embedded in Vectashield (Vector Labora- independent genetic changes also essential for the tories, Burlingame, CA, USA). development of malignancy. Further investigation into these and other questions will provide additional Generation of probes for FISH information about this novel nucleolar candidate TSG. LANL Human Single Chromosome Library on chromosome However, our findings support the hypothesis of an 6, LA06NC01 was obtained from UK HGMP Resource increasingly complex role of the nucleolus in a wide Centre (Cambridge, UK). Four positive clones LA0634c7, range of physiologic and pathologic conditions, includ- LA0655n13, LA0658k18, and LA0668j14 were identified. ing cancer. Clone LA0634c7 was used to generate an NOL7-specific genomic DNA probe by PCR. Briefly, cosmid DNA extracted from clone LA0634c7 was used as template to generate an 8-kb PCR product by the Platinum Pfx DNA Polymerase kit Methods (Invitrogen, Carlsbad, CA, USA) according to the manufac- turer’s instructions. The forward and reverse primers were ggt Cells gct gtt gct gat gtg ggt atg and tct ttg atg tgt atg gtt ggt t. This 8- The human cervical carcinoma cell lines SiHa, Ca Ski, C-33A, kb product was subsequently used as a probe for FISH on CC AN3 CA, ME-180, HEC-1-A, C-4 I, and HeLa cell lines were cell lines and human patient samples. Spectrum Orange- obtained from the ATCC. The SiHa and Ca Ski cell lines were directly labeled centromere probes specific for chromosome 6 cultured in RPMI and 10% FBS. The C-33A and AN3 CA cell (CEP 6-D6Z1) and chromosome 12 (CEP 12) (Vysis, Downers

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 596 Grove, IL, USA) were used in the FISH experiments as control permeabilized with 0.2% Triton X-100 in PBS for 10 min, probes. followed by a solution of 2% bovine serum albumin (BSA)/2% normalgoat serum in PBS for 10 min. The goat primary anti- B23 (C-19) antibody (Santacruz, Santa Cruz, CA, USA; Fluorescence hybridization in situ Catalog #: SC-6013, 1:250 dilution) was incubated on the slide Metaphase preparations and single-cell suspensions obtained in a moisture chamber for 1 h. Bovine anti-goat IgG-R from fresh tumors were fixed in acetic acid/methanol(1:3) secondary antibody (Santacruz, Santa Cruz, CA, USA; solution. The different probes were labeled by nick translation Catalog #: SC-2349; 1:250 dilution) was then applied for (BioNick, Invitrogen, Carlsbad, CA, USA) with biotin-11- 30 min. Fluorescence images were obtained using an Axiovert deoxyuridine triphosphate (Invitrogen, Carlsbad, CA, USA). 200 M inverted microscope (CarlZeiss, Oberkochen, Ger- Biotinylated probes were visualized by subsequent incubations many), with a Hamamatsu Orca ER camera (Opelco, Dulles, with (i) streptavidin-FITC (Vector Laboratories, Burlingame, VA, USA) and Openlab capturing software (Improvision, CA, USA), (ii) biotinylated goat anti-streptavidin (Vector Lexington, MA, USA). Laboratories, Burlingame, CA, USA), and (iii) streptavidin- FITC (Vector Laboratories, Burlingame, CA, USA). Cells and chromosomes were counterstained with DAPI and embedded NOL7 transfection into CC tumor cells in Vectashield (Vector Laboratories, Burlingame, CA, USA). NOL7-expressing CC tumor cell lines were generated by Digital images were captured using a cooled-coupled device cloning the full-length cDNA into the HindIII/XbaI sites of the (CCD) camera (Photometrics, Tucson, AZ, USA) mounted on expression vector pcDNA3.1/V5-His-TOPO. (Invitrogen, an Axioplan fluorescence microscope (Zeiss, Oberkochen, Carlsbad, CA, USA). The final product was amplified by Germany) equipped with selective filters for fluorescein, PCR, and the construct was sequenced to confirm orientation. rhodamine, and DAPI, and controlled by a G3 Macintosh At 24 h before transfection, the cells were seeded in 100 mm computer with SmartCapture (DigitalScientific, Cambridge, plates so that they would reach approximately 80% confluency UK) and MacProbe softwares (PSI/Applied Imaging, San by the next morning. Transfections were performed using Jose, CA, USA). In all cases, a minimum of 20 metaphases and calcium phosphate according to the manufacturer’s protocol, 200 interphase nuclei were analysed. using 20 mg DNA and selecting for 2–3 weeks using neomycin FISH was performed as described previously (Coignet et al., (Invitrogen, Carlsbad, CA, USA; 300 mg/ml). Individual clones 1999). Cell lines were cultured and harvested for chromosome were picked via ring cloning and transferred into six-well preparation. Fresh tumor tissue, acquired with Institutional plates, grown and expanded as required. Review Board approvalunder the exemption status, was lacerated and digested overnight with Collagenase (1 mg/ml, Sigma, St Louis, MO, USA) and Hyaluronidase (100 mg/ml, Western blot of NOL7 Sigma, St Louis, MO, USA). Metaphase preparations and Western blots were conducted according to the protocol single-cell suspensions from fresh tumors were fixed in acetic provided with the antibodies. Briefly, whole-cell lysates were acid/methanol (1:3) solution. The probes were labeled by nick separated on a 12% SDS–PAGE geland blottedonto translation with biotin-11-deoxyuridine triphosphate (Bio- Immunoblot-PVDF membrane (Bio-Rad, Hercules, CA, Nick, Invitrogen, Carlsbad, CA, USA). Competitive hybridi- USA). The blot was blocked and then incubated with 1:3000 zation was performed by preannealing the denatured DNA Anti-V5 antibody (Invitrogen, Carlsbad, CA, USA), followed with a 100–200-fold excess of Cot-1 DNA (Invitrogen, by 1:5000 anti-mouse antibody (Santa Cruz Biotechnology, Carlsbad, CA, USA) before overnight hybridization. Biotiny- Santa Cruz, CA, USA). After washing with TBS-Tween, the lated probes were visualized by subsequent incubations with (i) blot was incubated in ECL reagents (Amersham, Piscataway, streptavidin–FITC, (ii) biotinylated goat antistreptavidin, and NJ, USA), followed by exposure to film. (iii) streptavidin–FITC (all from Vector Laboratories, Burlin- game, CA, USA). Cells and chromosomes were counterstained with DAPI and embedded in Vectashield (Vector Labora- Western blot for VEGF and TSP-1 tories, Burlingame, CA, USA). In all cases, a minimum of 20 Serum-free CM from the SiHa- and NOL7-containing clones metaphases and 200 interphase nuclei were analysed. were separated on 12% (VEGF) and 5% (TSP-1) SDS–PAGE gels and blotted onto immunoblot-PVDF membrane (Bio- Rad, Hercules, CA, USA). The blot was blocked and then Cellular localization incubated with 1:200 anti-VEGF antibody (Santa Cruz, Santa A GFP–NOL7 fusion protein was generated using the GFP Cruz, CA, USA; Catalog #: SC-152) or 1:200 anti-TSP-1 Fusion TOPO TA Cloning kit (Invitrogen, Carlsbad, CA, (Santa Cruz, Santa Cruz, CA, USA; Catalog #: SC-12312) USA). Briefly, the full-length NOL7 cDNA was ligated into followed by 1:2500 anti-rabbit (for VEGF Western) or anti- the plasmid vector pcDNA3.1/NT-GFP-TOPO, and the goat (for TSP-1 Western) antibodies (Santa Cruz, Santa Cruz, cloning reaction was transformed into chemically competent CA, USA). After washing with TBS-Tween, the blot was cells provided in the kit. A total of 10 colonies were selected, incubated in ECL reagents (Amersham, Piscataway, NJ, USA) the DNA was purified with Qiagen Plasmid Mini kit (Qiagen, followed by exposure to film. Valencia, CA, USA), and sequenced for verification of insert orientation. Cos7 cells were plated in duplicate wells of LabTek chamber glass slides at a density of 105 cells/ml and In vivo tumor studies cultured overnight. The cells were transiently transfected using Female 3–4-week-old nu/nu mice were used. Mice (Charles the CaPO4 Transfection Kit (Invitrogen, Carlsbad, CA, USA), River Laboratories, Wilmington, Maine, USA) were accli- following the manufacturer’s protocol. Briefly, cells were mated and caged in groups of four or less. Cervical carcinoma treated with CaPO4 solution containing 20 mg/mlGFP– cells were injected into the subcutaneous tissue at a final NOL7 fusion protein for 4 h. After 24 h, positive NOL7 concentration of 107 cells in 0.1 ml of PBS (n ¼ 12 mice per transfectants were identified by their green fluorescence. Nuclei group). Tumor growth was monitored over a period of 30 days were counterstained with DAPI. In addition, cells were fixed by measurement with a caliper, and tumor volumes were with 4% paraformaldehyde for 10 min, rinsed with PBS, calculated.

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 597 Endothelial cell migration assay 1:100 for 1 h at room temperature, and the rabbit EnVision þ The endothelial cell migration assay was performed as kit (DAKO, Carpinteria, CA, USA) was used for detection. previously described (Lingen, 2002). Briefly, human dermal For determination of cell proliferation, sections were micro- microvascular endothelial cells (Cell Systems, Kirkland, WA, waved in citrate buffer and incubated in a 1:1 dilution of USA) were starved overnight in media containing 0.1% BSA, PCNA antibody (Zymed, San Francisco, CA, USA; Catalog #: harvested, resuspended into DME with 0.1% BSA, plated on 93–1143) for 1 h. Apoptotic cells were detected using an the bottom side of a modified Boyden chamber (Nucleopore antibody to cleaved caspase-3 (Biocare Medical, Walnut Corporation, MD, USA), and allowed to attach in the inverted Creek, CA, USA; Catalog number: CP229A). Pretreatment chamber for 2 h at 371C. The chamber was then re-inverted, for caspase was achieved by using a decloaker with ET buffer. test substances (1 mg totalprotein per test compound in a HRP-conjugated secondary antibody to rabbit (Envision þ , volume of 50 ml) added to the wells of the upper chamber, and DAKO, Carpenteria, CA, USA; Catalog #: K4010) was used cells were allowed to migrate for 4 h at 371C. Membranes were with incubation for 30 min each. For thrombospondin, a 1:100 recovered, fixed, stained, and the number of cells that had dilution of anti-TSP-1 (Santa Cruz Biotechnology, Santa Cruz, migrated to the upper chamber per 10 high power fields CA, USA; Catalog #: sc-12312) was used to incubate sections counted. Each substance was tested in quadruplicate in each overnight at 41C, followed by incubation with a secondary experiment and all experiments were repeated at least twice. antibody (Dako Labeled Streptavidin-Biotin kit, Carpinteria, VEGF (R&D Systems, MN, Minnesota, USA) was used as a CA, USA; Catalog #: x0590) and streptavidin horseradish positive controlat a concentration of 200 pg/ml. peroxidase. All sections were counterstained with hematoxylin. For functionalvalidationmigration studies of VEGF and TSP-1, neutralizing antibodies were used at the following concentrations: anti-VEGF (R&D Systems, Minneapolis, MN, Quantification of immunohistochemistry USA; Catalog #: AF-293-NA; 20 mg/ml) and anti-TSP-1 (Lab To quantitate proliferation, apoptosis and CD31 staining, the Vision, Fremont, CA, USA; Catalog #: MS-418-P1ABX; sections were first scanned at low power (  40) to avoid 20 mg/ml). Optimal concentrations for growth factors and counting areas that contained necrosis. The proliferating and antibodies were determined previously by dose–response apoptotic cells were counted at high power (  400) and experiments (data not shown). All antibodies were tested expressed as a percentage of the total number of cells. alone and found to be neutral in that they neither stimulated Determination of microvesseldensity using the CD31-stained nor reduced basal levels of migration, indicating that the doses slides was performed as described previously (Lingen et al., used were neither stimulatory nor toxic (data not shown). 1998). Briefly, the region containing the most intense area of tumor neovascularization was chosen for counting in each of Histology the tumors. Individualmicrovesselswere counted using an For CD31 detection, the primary antibody (PECAM-1, BD  200 field (  20 objective lens and  10 ocular lens). Any Pharmigen, San Jose, CA, USA; Catalog #: 550274) was brown-staining endothelial cells that were clearly separate in applied to deparaffinized sections at a 1:100 dilution in PBS for appearance were counted as individual vessels. Results were 6 h at 41C. The biotinylated secondary antibody (Dako expressed as the highest number of microvessels observed at Labeled Streptavidin-Biotin kit, Carpinteria, CA, USA;  200 in the ‘hot spot’ region of each individualtumor. Catalog #: x0590) was applied for 30 min, washed and the enzyme conjugate was added for 30 min. For VEGF detection, deparaffinized sections were microwaved in citrate buffer, Acknowledgements rabbit primary antibody (Santa Cruz Biotechnology, Santa This work was supported in part by the NIH grants: DE12322 Cruz, CA, USA; Catalog #: sc-152) was applied at a dilution of and DE00470 (MWL).

References

Bouck N, Stellmach V, Hsu H. (1995). Adv Cancer Res 69: Jinawath N, Furukawa Y, Nakamura Y. (2004). Cancer Sci 135–174. 95: 430–435. Cannistra SA, Niloff JM. (1996). N Engl J Med 334: Kersemaekers AM, Kenter GG, Hermans J, Fleuren GJ, van 1030–1038. de Vijver MJ. (1998). Int J Cancer 79: 411–417. Chen T, Pecoraro G, Defendi V. (1993). Cancer Res 53: Kraemer M, Tournaire R, Dejong V, Montreau N, Briane D, 1167–1171. Derbin C et al. (1999). Cell Growth Differ 10: 193–200. Chen Z, Issa B, Brothman LJ, Hendricksen M, Button D, Lazo PA. (1999). Br J Cancer 80: 2008–2018. Brothman AR. (2000). Cancer Genet Cytogenet 121: 22–25. Lingen MW. (2002). Wound Healing: Methods and Protocols. Claudio PP, Stiegler P, Howard CM, Bellan C, Minimo C, Totowa, NJ: Humana Press, Inc, pp. 337–347. Tosi GM et al. (2001). Cancer Res 61: 462–468. Lingen MW, DiPietro LA, Solt DB, Bouck NP, Polverini PJ. Coignet LJ, Lima CS, Min T, StreubelB, Swansbury J, (1997). Carcinogenesis 18: 329–338. Telford N et al. (1999). Genes Chrom Cancer 25: 222–229. Lingen MW, Polverini PJ, Bouck NP. (1998). Cancer Res 58: Dameron KM, Volpert OV, Tainsky MA, Bouck N. (1994). 5551–5558. Science 2651: 582–584. Marconcini L, Marchio S, Morbidelli L, Cartocci E, Albini A, Dejong V, Degeorges A, Filleur S, Ait-Si-Ali S, Mettouchi A, Ziche M et al. (1999). Proc Natl Acad Sci USA 96: Bornstein P et al. (1999). Oncogene 18: 3143–3151. 9671–9676. Gnarra JR, Zhou S, Merrill MJ, Wagner JR, Krumm A, Mitra AB, Murty VV, Li RG, Pratap M, Luthra UK, Papavassiliou E et al. (1996). Proc Natl Acad Sci USA 93: Chaganti RS. (1994). Cancer Res 54: 4481–4487. 10589–10594. Montgomery K, Tedford KL, McDougall JK. (1995). Genes Hanahan D, Folkman J. (1996). Cell 86: 353–364. Chromosomes Cancer 14: 97–105. Huettner PC, Gerhard DS, Li L, Gersell DJ, Dunnigan K, Mukhopadhyay D, Tsiokas L, Sukhatme VP. (1995). Cancer Kamarasova T et al. (1998). Hum Pathol 29: 364–370. Res 55: 6161–6165.

Oncogene Cervical cancer tumor suppressor gene inhibits angiogenesis R Hasina et al 598 Mullokandov MR, Kholodilov NG, Atkin NB, Burk RD, Tikhonenko AT, Black DJ, Linial ML. (1996). J Biol Chem Johnson AB, Klinger HP. (1996). Cancer Res 56: 271: 30741–30747. 197–205. Uejima H, Mitsuya K, Kugoh H, Horikawa I, Oshimura M. Nagai H, Kinoshita T, Suzuki H, Hatano S, Murate T, Saito (1995). Genes Chromosomes Cancer 14: 120–127. H. (1999). Genes Chromosomes Cancer 25: 277–283. Visintin R, Amon A. (2000). Curr Opin Cell Biol 12: 372–377. Nakase K, Wakita Y, Minamikawa K, Yamaguchi T, Shiku Volpert OV, Dameron KM, Bouck N. (1997). Oncogene 14: H. (2000). Leuk Res 24: 79–81. 1495–1502. Olson MO, Dundr M, Szebeni A. (2000). Trends Cell Biol 10: Walboomers JM, Jacobs MV, Manos MM, Bosch FX, 189–196. Kummer JA, Shah KV et al. (1999). J Pathol 189: 12. Rader JS, Li Y, Huettner PC, Xu Z, Gerhard DS. (2000). Watnick RS, Cheng YN, Rangarajan A, Ince TA, Weinberg Genes Chromosomes Cancer 27: 373–379. RA. (2003). Cancer Cell 3: 219–231. Rak J, Mitsuhashi Y, Bayko L, Filmus J, Shirasawa S, Weber JD, Taylor LJ, Roussel MF, Sherr CJ, Bar-Sagi D. Sasazuki T et al. (1995). Cancer Res 55: 4575–4580. (1999). Nat Cell Biol 1: 20–29. Sheibani N, Frazier WA. (1996). Cancer Lett 107: 45–52. Wen S, Stolarov J, Myers MP, Su JD, Wigler MH, Tonks NK Slack JL, Bornstein P. (1994). Cell Growth Differ 5: 1373–1380. et al. (2001). Proc Natl Acad Sci USA 98: 4622–4627. Srivastava S, Tong YA, Devadas K, Zou ZQ, Chen Y, Pirollo Wolf JK, Ramirez PT. (2001). Cancer Invest 19: 621–629. KF et al. (1992). Carcinogenesis 13: 1273–1275. Yaginuma Y, WestphalH. (1991). Cancer Res 51: 6506–6509. Sugimoto M, Kuo M-L, RousselMF, Sherr CJ. (2003). Mol Zabrenetzky V, Harris CC, Steeg PS, Roberts DD. (1994). Int Cell 11: 415–424. J Cancer 59: 191–195. Takeshita A, Naito K, Shinjo K, Sahara N, Matsui H, Ohnishi Zhang YW, Su Y, Volpert OV, Vande Woude GF. (2003). K et al. (2004). Cancer Genet Cytogenet 152: 56–60. Proc Natl Acad Sci USA 100: 12718–12723.

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc)

Oncogene