Oncogene (2013) 32, 2150–2160 & 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13 www.nature.com/onc

ORIGINAL ARTICLE The p53 isoform, D133p53a, stimulates angiogenesis and tumour progression

H Bernard1,2, B Garmy-Susini1,3, N Ainaoui1, L Van Den Berghe1,4, A Peurichard1, S Javerzat5,6, A Bikfalvi5,6, DP Lane2, JC Bourdon2,7 and A-C Prats1,7

The tumour suppressor p53, involved in DNA repair, cell cycle arrest and apoptosis, also inhibits blood vessel formation, that is, angiogenesis, a process strongly contributing to tumour development. The p53 gene expresses 12 different proteins (isoforms), including TAp53 (p53 (or p53a), p53b and p53g) and D133p53 isoforms (D133p53a, D133p53b and D133p53g). The D133p53a isoform was shown to modulate p53 transcriptional activity and is overexpressed in various human tumours. However, its role in tumour progression is still unexplored. In the present study, we examined the involvement of D133p53 isoforms in tumoural angiogenesis and tumour growth in the highly angiogenic human glioblastoma U87. Our data show that conditioned media from U87 cells depleted for D133p53 isoforms block endothelial cell migration and tubulogenesis without affecting endothelial cell proliferation in vitro. The D133p53 depletion in U2OS osteosarcoma cells resulted in a similar angiogenesis blockade. Furthermore, using conditioned media from U87 cells ectopically expressing each D133p53 isoform, we determined that D133p53a and D133p53g but not D133p53b, stimulate angiogenesis. Our in vivo data using the chicken chorio-allantoic membrane and mice xenografts establish that angiogenesis and growth of glioblastoma U87 tumours are inhibited upon depletion of D133p53 isoforms. By TaqMan low-density array, we show that alteration of expression ratio of D133p53 and TAp53 isoforms differentially regulates angiogenic gene expression with D133p53 isoforms inducing pro-angiogenic gene expression and repressing anti-angiogenic gene expression.

Oncogene (2013) 32, 2150–2160; doi:10.1038/onc.2012.242; published online 25 June 2012 Keywords: angiogenesis; p53; isoform; cancer; glioblastoma

INTRODUCTION Furthermore, increased expression of D133p53a is associated with The well-known tumour suppressor p53 called ‘the guardian of the inhibition of replicative senescence in normal human fibroblasts 5 genome’ was first described as a single protein. However, we have and colon carcinoma. However, no data are available about a shown that the p53 gene expresses 12 protein isoforms.1,2 direct role of D133p53a in tumour progression. Alternative splicing of intron 9 generates p53 isoforms bearing In addition to its ability to control DNA repair, cell cycle different C-terminal domains (a, b and g; Figure 1a). The presence arrest and apoptosis, p53 is involved in inhibition of angiogenesis, of an alternative promoter in intron 4 generates D133p53 isoforms a critical mechanism in tumour progression and metastatic 6 lacking the transactivation domain (TA) and part of the DNA- dissemination. This process is induced in most solid tumours, binding domain. In addition, initiation of translation at alternative whose centres become hypoxic following an increase of the AUG codons, codon 40 or codon 160, leads to D40p53 and diffusion distance between the nutritive blood vessels and D160p53 isoforms, respectively. The p53 protein isoforms can be tumour cells. Hypoxia is one of the principal angiogenic stimuli categorized into four subclasses: TAp53, D40p53, D133p53 and leading to synthesis of angiogenic growth factors and inducing 7,8 D160p53 protein isoforms. Each subclass is composed of the three formation of new blood vessels. Such vessels, in addition to isoforms a, b and g. their ability to feed tumour cells, provide a way for cells to The p53 isoforms are conserved in Drosophila, zebrafish, mouse disseminate and form metastases. p53 inhibits angiogenesis by and human, but little is known about their functions.2,3 A gain of at least three mechanisms: (1) by regulating expression and expression of D133p53 mRNAs in human breast tumours versus activity of the central regulator of hypoxia, the hypoxia-induced normal breast tissue suggests a protumoural role of these transcription factor-1a; (2) by inhibiting production of pro- isoforms.2 Indeed, D113p53, the zebrafish ortholog of human angiogenic factors, such as vascular endothelial growth factor D133p53a, is able to regulate wild-type p53 (WTp53) pro- (VEGF) and fibroblast growth factor 2 (FGF2); and (3) by increasing 6,9–14 apoptotic function.3 This activity is conserved in human cells, as production of anti-angiogenic factors. p53 is able to induce D133p53a can inhibit p53-mediated apoptosis and G1 cell cycle tumour dormancy, and the loss of WTp53 reverses the tumour arrest without inhibiting p53-mediated G2 cell cycle arrest.4 switch from a dormant non-angiogenic state to an angiogenic

1Universite´ de Toulouse, UPS, TRADGENE, Laboratory of Translational Control and Gene Therapy of Vascular Diseases, EA4554, Institut des Maladies Me´taboliques et Cardiovasculaires, Toulouse, France; 2Department of Surgery and Molecular Oncology, University of Dundee, Dundee, UK; 3Inserm, U1037, Toulouse, France; 4Inserm, U1048, F-31432 Toulouse, France; 5Inserm, U920, Talence, France and 6Universite´ de Bordeaux 1, Angiogenesis and Cancer Microenvironment Laboratory, Talence, France. Correspondence: Dr A-C Prats, TRADGENE, Laboratory of Translational Control and Gene Therapy of Vascular Diseases, Institut des Maladies Me´taboliques et Cardiovasculaires, Universite´ Paul Sabatier, EA 4554, 1, Avenue Jean Poulhe`s, BP 84225, 31432 Toulouse Cedex 4, France. E-mail: [email protected] 7These authors contributed equally to this work. Received 6 November 2011; revised 12 April 2012; accepted 11 May 2012; published online 25 June 2012 D133p53a stimulates angiogenesis H Bernard et al 2151 invasive phenotype.15,16 Clinical data also suggest that p53 RESULTS has a crucial role in controlling tumour vascularization. However, Characterization of p53 isoforms expression in human nothing is known about the role of D133p53 isoforms (a, b and g) glioblastoma U87 in this process. The expression of p53 isoforms was analysed at the mRNA level by Among all solid tumours, glioblastoma multiformes are the nested reverse transcriptase (RT)–PCR in U87 human glioblastoma most angiogenic, by displaying the highest degree of vascular cells, which express WTp53 gene. We determined that U87 cells 17 proliferation and endothelial cell hyperplasia. Such intense express TAp53 isoforms (p53 (p53a) and p53g) and D133p53 vascularization has a critical role in the pathological features isoforms (D133p53a and D133p53b). The D40p53 isoforms of glioblastoma multiformes, including peritumoural oedema subclass as well as p53b and D133p53g were not detected at resulting from the defective blood–brain barrier in the newly the mRNA level (Figure 1b). formed vasculature. Thus, human glioblastoma U87, expressing To investigate the role of endogenous p53 isoforms in WTp53, has been chosen here to address the role of D133p53 angiogenesis, U87 cells were transfected with the small interfering isoforms in angiogenesis and tumour progression. RNA (siRNA) siD133p53, which targets specifically the D133p53 In the present study, we establish by knockdown and ectopic isoforms subclass, or with the siRNA siTAp53, which targets expression experiments that angiogenesis can be regulated by specifically the TAp53 isoforms subclass (Figure 1a). The specificity manipulating the expression ratio of D133p53a and p53. of the above siRNAs has previously been assessed and validated.4,18 Efficiency of siD133p53 and siTAp53 in U87 cells was confirmed by quantitative RT–PCR and western blotting, as P1’ ATG3 compared with the control siRNA (siNON; Figures 1c1, 1c2, 1d1 P1 Δ β/γ/α ATG1 ATG2 and 1d2). As expected, transfection of siTAp53 inhibits p53 mRNA P2 expression, but has no effect on D133p53 mRNA expression. Transfection of siD133p53 in U87 cells represses D133p53 mRNA 1562433 789 9i 10 11 expression. However, transfection of siD133p53 induces p53 expression at both protein and mRNA levels in U87 cells SiTAp53 SiΔ133p53 (Figures 1c2 and 1d2). Similar results were obtained with another siRNA specific for D133p53 isoforms, siD133p53-2 (Supplementary α Figure S1).4 Thus, induction of p53 in response to siD133p53 TA β transfection seems to be because of the inhibition of p53 mRNA Δ133 γ expression by D133p53 isoforms. Co-transfection of U87 cells with p53 Domains siTAp53 and siD133p53 leads to inhibition of D133p53 expression and prevents induction of p53, maintaining p53 expression at its TA1 TA2 PR DNA BD NLS OD basal level. Therefore, the expression ratio of endogenous D133p53 and p53 isoforms can be manipulated by transfection p53 Δ133p53 of siTAp53 and/or siD133p53.

bp αβγ αβ γ 1500 Tumour-conditioned medium from U87 glioblastoma cells or 1000 U2OS osteosarcoma cells transfected with siD133p53 impairs 80 angiogenesis in vitro 0 Anti-angiogenic activities reported for p53 prompted us to look at the effect of D133p53 isoforms in angiogenesis. U87 cells Δ133p53 mRNA expression p53 mRNA expression were transfected with siD133p53, siTAp53 or siNON, and the 100 250 80 200 Figure 1. Regulation of expression of endogenous p53 isoforms in 60 150 U87 cells. (a) Human p53 gene structure. Alternative splicing of intron 9 generates p53 isoforms bearing different C-terminal 40 100 domains (a, b and g). TAp53 isoforms include p53 (p53a), p53b % of control 20 % of control 50 and p53g, whereas D133p53 isoforms include D133p53a, D133p53b and D133p53g.2 The different siRNAs used (siTAp53 and siD133p53) 0 0 SiNON + --- + --- and the regions they target are indicated underneath the p53 gene SiTAp53 - + - + - + - + structure. The different p53 domains are shown: transactivation SiΔ133p53 --+ + --+ + domains (TA), proline rich domain (PR), DNA binding domain (DNA BD), nuclear localization signal (NLS) and oligomerization domain p53 protein expression (OD). (b) Gel electrophoresis (1% agarose) of p53 isoform mRNAs amplified by nested RT–PCR in U87 cells. Total RNA was extracted 133p53 protein p53 55 from U87 cells and nested RT–PCR were performed, as previously expression described.2 (c) D133p53 (c1) and p53 mRNAs (c2) relative expression Δ133p53 38 was determined by quantitative RT–PCR following siRNA transfec- β-tub 50 tion in U87 cells. Results shown are the average of two independent experiments performed in duplicate. Relative quantification of p53 β-tub 50 250 isoforms was determined by the 2-DDCt by normalizing to 18S rRNA 4 SiNON +- - 200 and control (siNON-transfected U87 cells). Results are presented as SiΔ133p53 --+ 150 percentage compared with control. (d1) D133p53 protein Δ133p53 - + + expression was analysed by western blotting after transfection of 100 U87 cells with siRNA (siD133p53 or siNON) and D133p53a

% of control 50 expression vector, as indicated. (d2) p53 expression was analysed 0 by western blotting after siRNA transfection (b-tubulin was used as a SiNON + --- loading control). The histogram shows densitometry analysis of SiTAp53 - + - + multiple scanned western blot images to quantify the percentage SiΔ133p53 --+ + increase over baseline of p53a (four independent experiments).

& 2013 Macmillan Publishers Limited Oncogene (2013) 2150 – 2160 D133p53a stimulates angiogenesis H Bernard et al 2152 Migration quantification 100%

80%

60%

40% SiNON SiTAp53

% of migration 20%

0% C + (FGF-2) --+ --- C ------SiNON --+ --- SiTAp53 + Si133p53 SiTAp53 --- + - + Si133p53 SiΔ133p53 ----+ +

Tubulogenesis quantification (U87 conditioned media) 16 14 12 10 SiNON SiTAp53 8 * 6 4

Branch point/field 2 0 SiTAp53 + SiNON + ---  Si133p53 Si 133p53 SiTAp53 - + - + SiΔ133p53 --+ +

HUVEC Proliferation Tubulogenesis quantification (U2OS conditioned media) 0.7 50 *p<0.005 vs siNON 0.6

0.5 40 0.4 30 SiNON * 0.3 SiTAp53 OD 540 nm 0.2 SiΔ133p53 20 SiΔ133p53 + SiTAp53 0.1 Branch points/field 10 0.0 0244872 0 Hours SiNON + --- SiTAp53 - + - + SiΔ133p53 --+ + Figure 2. Tumour conditioned medium from U87 cells transfected with siD133p53 impairs HUVEC endothelial cell migration and tube formation, but not their proliferation. Conditioned media from siRNA-transfected U87 cells were used to assess HUVEC migration, tube formation and proliferation. (a1) Micrographs of scratch wound-healing assay of HUVEC grown in conditioned media from U87 cells transfected with siRNAs as indicated. (a2) Quantification of HUVEC migration by scratch wound-healing assay. (b1) Micrographs ( Â 100 magnification) of tube formation by HUVEC grown on Matrigel containing 10 ng/ml FGF2 (HUVEC migration must be stimulated to detect a possible inhibition by siRNAs) in conditioned media from siRNA-transfected U87 cells as indicated. (b2) Quantification of the number of branch points formed during tubulogenesis. Asterisk indicates statistical significance. *Po0.05 versus control (conditioned medium from siNON-transfected U87 cells). (c) Effects of conditioned media from siRNA-transfected U87 cells supplemented with 3 ng/ml FGF2 on HUVEC proliferation. HUVECs were incubated with conditioned media from siRNA-transfected U87 cells for 24, 48 and 72 h. Cell proliferation was determined by MTT assay. (d) Conditioned media from U2OS cells transfected with siRNAs as above. Quantification of the number of branch points formed during tubulogenesis. Asterisk indicates statistical significance. *Po0.05 vs control (conditioned medium from siNON- transfected U2OS cells).

corresponding conditioned medium was added onto human conditioned medium from U87 cells transfected with siTAp53 had umbilical vein endothelial cells (HUVECs). Migration of HUVECs no effect. Interestingly, conditioned medium from U87 cells co- was then analysed in the wound-healing assay (Figures 2a1 and transfected with siTAp53 and siD133p53 had no effect on HUVEC 2a2). The ability of cultured HUVECs to form vessel-like structures migration and tube formation, as well as proliferation (Figures (tubulogenesis) on matrigel was also investigated (Figures 2b1 and 2a1–2b2). Cell proliferation was also assessed; neither siD133p53 2b2). Conditioned medium from U87 cells transfected with nor siTAp53, nor the double siRNA treatment had any effect on U87 siD133p53 inhibited HUVEC migration as well as their ability to proliferation (Supplementary Figure S2), and the different condi- form vessel-like structures (Figures 2b1 and 2b2), whereas tioned media had no effect on HUVEC proliferation (Figure 2c).

Oncogene (2013) 2150 – 2160 & 2013 Macmillan Publishers Limited D133p53a stimulates angiogenesis H Bernard et al 2153 Δ133p53α None 0.1μg/ml Doxycycline 1μg/ml Doxycycline 150 * NS 100

50

Branchpoints/field 0 0 0.01 0.1 μg/ml Doxycycline

Δ133p53β 150 NS NS 100

50

Branchpoints/field 0 0 0.01 0.1 μg/ml Doxycycline

Δ133p53γ

150 § * 100

50

Branchpoints/field 0 0 0.01 0.1 μg/ml Doxycycline Figure 3. D133p53a and D133p53g stimulate endothelial cell tubulogenesis. U87 TetOn cells were transduced with lentivectors expressing D133p53a, D133p53b or D133p53g under the control of a tetracyclin-inducible promoter. Conditioned media from transduced and doxycycline-treated U87 cells were used to assess HUVEC tube formation as in Figure 2b, except that FGF2 was no added to Matrigel. Left panels: bright-field images (  100 magnification) of HUVEC in vitro ‘vessel’ formation grown in conditioned media from U87 cells expressing D133p53a (a), D133p53b (b), or D133p53g (c) in response to doxycycline treatment, as indicated. Right panels: quantification of the number of y branch points formed during tubulogenesis. Mean number of branching points per  100 field ±s.e.m. (*Po0.05, P ¼ 0.001).

Endothelial cell migration was then studied in a transwell D133p53a and D133p53g overexpression stimulates endothelial migration assay. In this experiment, adult bovine aortic endothelial cell tubulogenesis cells were seeded in Boyden Chambers and their migration To determine which D133p53 isoform (a, b or g) is responsible for quantified after treatment with conditioned media of siRNA- endothelial cell migration and tubulogenesis, lentivectors expres- transfected U87 cells (Supplementary Figure S3). Here, the sing each of these isoforms under the control of a tetracyclin- knockdown of D133p53 in U87 cells again resulted in a significant inducible promoter were used to transduce U87 TetOn cells decrease of endothelial cell migration, whereas the knockdown of (Supplementary Figures S4 and S5). To measure tubulogenesis, p53 generated a significant increase, suggesting that the absence HUVECs were seeded on matrigel as above and incubated in of increase of HUVEC migration and tubulogenesis following conditioned media produced by doxycycline-treated U87 cells treatment with siTAp53-conditioned media is probably because of (Figure 3). Data clearly showed that overexpression of D133p53a experimental conditions. Such an increase would probably have and D133p53g, but not of D133p53b, are able to stimulate been observed as for the adult bovine aortic endothelial cells, if endothelial cell tubulogenesis, suggesting that D133p53a and migration and tube formation would have been quantified at D133p53g isoforms have a pro-angiogenic activity. As U87 cells earlier times. express D133p53a and D133p53b, but not D133p53g (Figure 1b), Altogether, these data show that the secretion of angiogenic we concluded that only D133p53a is responsible for the secretion and/or anti-angiogenic factors by U87 cells into the conditioned of factors by U87 cells that regulate endothelial cell migration and medium is regulated by the expression ratio of p53 and tubulogenesis of HUVECs. D133p53, and suggest that the knockdown of D133p53 in U87 cells impairs the ability of these tumour cells to stimulate angiogenesis. SiD133p53 inhibits angiogenesis and tumour growth in vivo on To determine whether this effect of D133p53 can be general- the CAM ized to other tumours, HUVEC tubulogenesis was performed using The effect of D133p53a on angiogenesis and tumour growth was conditioned media of U2OS osteosarcoma cells (WTp53) treated addressed in vivo using the experimental glioma assay developed with the different siRNAs. The same data were obtained as for U87 on the chicken CAM, as previously described.19 This assay cells, showing that the pro-angiogenic effect of D133p53 is not recapitulates hallmarks of human glioblastoma multiformes and restricted to U87 glioblastoma (Figure 2d). allows following the first steps of tumoural angiogenesis. The

& 2013 Macmillan Publishers Limited Oncogene (2013) 2150 – 2160 D133p53a stimulates angiogenesis H Bernard et al 2154 Window in Injection of U87 Harvest of Incubation egg shell cells over CAM U87 tumours

D0 D3 D10 D14

SiNON SiTAp53 SiΔ133p53 SiΔ133p53 + SiTAp53

SNA-lectin

Vimentin

Angiogenic appearance Vessel quantification Tumour volume no tumour avascular vascular 30 * 100% ** 35 25 * 30 80% 20 25 60% 20 15 40% 15

% of field 10 10

20% Volume (mm3) 5 5 0% 0 0 SiNON (nM) + --- + --- + --- SiTAp53 - + - + - + - + - + - + SiΔ133p53 --+ + --+ + --+ + Figure 4. Transfection of siD133p53 inhibits angiogenesis and tumour growth in vivo on the chorio-allantoic membrane (CAM). U87 cells transfected with siNON, siTAp53, siD133p53 or siTAp53 þ siD133p53 (as indicated) were deposited on CAM of fertilized eggs, and tumours were analysed 4 days later. (a) Experimental protocol. (b) Tumour growth was assessed by biomicroscopy ( Â 3 magnification; I, II, III and IV), fluorescent immunohistochemistry ( Â 20 magnification; V, VI, VII and VIII) and haematoxylin and eosin (H&E) staining ( Â 20 magnification; IX, X, XI and XII). Tumour sections were co-stained for the mesenchymal marker vimentin (tumour cell marker, in red) and SNA-lectin (marker of endothelial cells, in green). Numerous irregular and dilated blood vessels are visible in all tumour sections (V, VI and VIII), except for tumours transfected with siD133p53 isoforms (VII). H&E staining of the corresponding tumour sections is shown and blood vessels are indicated by an arrow (IX–XII). (c)Left panel: angiogenic appearance presented as the percentage of absent (no visible tumour), vascular or avascular tumours following treatments with the different siRNAs (10–18 tumours for each siRNA treatment). Middle panel: SNA-lectin staining allowed quantification of the blood vessel network invading the tumour by counting the number of green pixels on each field with Adobe photoshop software (at least five tumours per group). Right panel: tumour volume was quantified for the tumours obtained in CAM presented in Figure 4 (11 to 16 tumours/condition).

siRNA-transfected U87 cells were deposited on the CAM of vascularization, whereas transfection of siD133p53 led to the fertilized eggs and tumours were analysed 4 days later (Figure 4a). formation of avascular tumours, showing that angiogenesis can be Transfection of siTAp53 resulted in an increase of tumour regulated by change in expression ratio of p53 and D133p53a

Oncogene (2013) 2150 – 2160 & 2013 Macmillan Publishers Limited D133p53a stimulates angiogenesis H Bernard et al 2155 (Figures 4b and 4c, left panel). Similar results were obtained with and Table 1). At day 7 after xenograft, U87 cells transfected with siD133p53-2 (Supplementary Figure S6). Angiogenesis inhibition siD133p53 generated significantly smaller tumours compared with upon siD133p53 transfection was assessed by counting chick U87 cells transfected with siNON (Figure 5a). However, tumours blood vessels in tumour sections stained with Sambucus nigra generated from U87 cells transfected with siTAp53 or cotrans- (SNA)-1 lectin conjugated to fluorescein isothiocyanate (Figure 4c, fected with siTAp53 and siD133p53 had similar sizes compared middle panel). The anti-angiogenic effect upon siD133p53 with tumours generated from U87 cells transfected with siNON. transfection was also correlated with smaller size of tumours at The mice were followed until day 44, which clearly showed that day 4 (Figure 4c, right panel). Western blot analyses confirmed tumours generated after siD133p53 treatment of U87 grew that siRNAs were still efficient in tumours 4 days after transfection; significantly more slowly than the control and were drastically tumours treated with siTAp53 or siD133p53 showed a decrease or smaller than the control at day 35, as well as day 44 (Figure 5b and an increase of p53 expression, respectively (Supplementary Figure Table 1). However, in contrast to the data obtained at day 7, the S7). Interestingly, tumours resulting from the co-transfection of size of tumours from U87 cells co-transfected with siD133p53, and siTAp53 and siD133p53 looked similar to the ones obtained after siTAp53 was similar to the size of tumours from U87 cells transfection of siNON (control). This confirms in vitro data and transfected with siD133p53. indicates that angiogenesis can be regulated by altering the Angiogenesis was quantified at day 7 by staining of the expression ratio of p53 and D133p53a. endothelial cell marker, CD31. Our data showed that tumours from siD133p53-transfected U87 cells were significantly less vascular- ized at day 7 as compared with tumours from U87 cells siD133p53 inhibits angiogenesis and tumour growth in mouse transfected with siNON or siTAp53, or co-transfected with siTAp53 xenografts and siD133p53 (Figure 5c). To study D133p53a involvement in tumour progression, siRNA- These results are consistent with the ones obtained in CAM, and transfected glioblastoma U87 cells were subcutaneously confirm that change in expression ratio of p53 and D133p53a implanted in nude mice and followed during 44 days (Figure 5 regulates angiogenesis and tumour growth in both in vivo models.

Tumour volume Tumour growth Day 7 8000 100 7000 SiNON 90 6000 80 SiTAp53 5000 70 SiΔ133p53 60 4000 50 SiΔ133p53 + 40 3000 SiTAp53 30 Volume (mm3) Volume (mm3) * 2000 20 * 10 1000 0 * SiNON + --- 0 SiTAp53 - + - + 0 5 10 15 20 25 30 35 40 45 SiΔ133p53 --+ + Days after tumour injection

Tumour angiogenesis Day 7 CD31/DAPI CD31/DAPI

30 25 20 SiNON SiTAp53 15 * CD31/DAPI CD31/DAPI /field 10 5 Vessels number 0 SiNON + --- SiTAp53 - + - + SiΔ133p53 --+ + SiTAp53 +  SiΔ 133p53 Si 133p53 Figure 5. Transfection of siD133p53 inhibits angiogenesis and tumour growth in mouse xenografts. U87 cells transfected with siNON, siTAp53, siD133p53 or siTAp53 þ siD133p53 at 50 nM were xenografted into nude mice. Tumour volumes were quantified from day 7 to day 44 (eight tumours/condition; Table 1). (a) Quantification at days 7. *Po0.05. (b) Kinetics of tumour growth unitl day 44 (c) Immunostaining to detect blood vessels in glioblastoma at day 7. Cryosections of xenografted tumour of U87 cells were immunostained using anti-CD31 antibody. Nuclei were stained with DAPI (blue), Â 200 magnification. Right panel: quantification of number of blood vessels/field. Asterisk indicates statistical significance. *Po0.05 versus the SiNON-treated group.

& 2013 Macmillan Publishers Limited Oncogene (2013) 2150 – 2160 D133p53a stimulates angiogenesis H Bernard et al 2156 Table 1. Effect of siD133p53 treatment on tumour growth in nude siTAp53 and siD133p53, whereas ANG expression was still induced mice (Figure 6c). In addition, upon transfection of siD133p53, the angiogenic marker angiopoietin-like 4 is downregulated, a SiNON SiTAp53 SiD133 SiD133 þ repression that is not observed in U87 cells co-transfected with SiTAp53 siTAp53 and siD133p53 isoforms. This indicates that these pro- angiogenic factors and markers are differentially regulated by the Day 7 Average 58 68 15 76 expression ratio of D133p53a and p53 (Figure 6d). Moreover, it s.e.m. 14 14 5 13 P-value 0.3060 0.0095 0.1904 suggests that some genes, such as ANG, are regulated by Day 14 Average 79 86 20 76 D133p53a independently of p53. s.e.m. 19 19 8 13 Altogether, our results indicate that upon transfection of U87 P-value 0.4026 0.0095 0.4471 tumour cells with siD133p53, which represses D133p53a expres- Day 20 Average 143 140 31 97 sion and induces p53, anti-angiogenic factors are induced, s.e.m. 38 30 13 11 whereas expression of secreted pro-angiogenic factors is P-value 0.4750 0.0107 0.1356 repressed. Day 27 Average 321 252 44 161 s.e.m. 83 49 13 38 P-value 0.2431 0.0060 0.0564 DISCUSSION Day 35 Average 1336 841 173 474 s.e.m. 335 258 53 159 We and others have previously shown abnormal expression of p53 P-value 0.1317 0.0051 0.0220 isoforms, including p53b, p53g or D133p53 mRNA expression in 20 Day 44 Average 6062 4171 736 1577 human tumours compared with normal corresponding tissue. s.e.m. 1206 974 239 614 p53g expression is associated with good prognosis in breast P-value 0.1218 0.0014 0.0039 cancer patients expressing mutant p53.21 Moreover, D133p53a U87 cells transfected with siNON, siTAp53, siD133p53 or siTAp53 þ inhibits replicative senescence, p53-mediated apoptosis and G1 cell cycle arrest, without preventing p53-promoted G2 cell cycle siD133p53 were xenografted into nude mice. Tumour volumes were 2–5 quantified from day 7 to day 44 (eight tumours/condition). arrest by regulating gene expression. Altogether, it suggests that p53 isoforms have a key role in carcinogenesis. However, the role of p53 isoforms in angiogenesis was unknown. The present study reveals that D133p53 is responsible for a strong Hence, an increase of D133p53a compared with p53 would angiogenesis stimulation that mainly contributes to acceleration stimulate angiogenesis, whereas an increase of p53 compared of human glioblastoma progression. D133p53 angiogenic effect is with D133p53a would inhibit angiogenesis. due to D133p53a and g, but not the b isoform. To study the role of p53 isoforms on angiogenesis, we used the glioblastoma U87 model expressing WTp53. Glioblastoma multi- p53 and D133p53a regulate angiogenesis by differentially formes are the most angiogenic by displaying the highest degree modulating expression of angiogenesis-related genes of vascular proliferation and endothelial cell hyperplasia. We have To decipher how p53 and D133p53a regulate angiogenesis, shown that U87 cells express p53 and p53g (TAp53 isoforms), as expression of 96 angiogenesis-related genes was analysed well as D133p53a and D133p53b (D133p53 isoforms) at the mRNA by TaqMan low-density array in siRNA-transfected U87 cells, as level. Only p53 protein could be detected at the protein level indicated (Figure 6). We used a Taqman human angiogenesis because of the low affinity of current p53 antibodies and the low array providing a panel of 96 angiogenesis-related genes expression level of p53 protein isoforms. However, it is important (Supplementary Table S1). In the control, expression of 14/39 to note that several studies have reported that p53 protein anti-angiogenic genes (35%) and 20/28 pro-angiogenic genes isoforms are biologically active at low expression level.3–5,18,22 (71%) was detected (Figure 6a). Results with siD133p53 and By a knockdown approach using siRNAs specific of TAp53 siTAp53 were normalized to siNON-transfected U87 cells. isoforms or of D133p53 isoforms, we have noticed that siD133p53 Regarding the anti-angiogenic factors, transfection of concomitantly represses D133p53mRNAs and induces p53 at the siD133p53 significantly induced interleukin 12A (IL12A) and matrix mRNA and protein levels in U87 cells, whereas siTAp53 represses metallopeptidase 2, whereas collagen IV (COL4A2) was signifi- p53 expression at the mRNA and protein level without altering cantly repressed (Figure 6b and Table 2). Transfection of siTAp53 expression of D133p53 mRNAs. Interestingly, co-transfection of induced also the matrix metallopeptidase 2 expression, whereas siD133p53 and siTAp53 represses D133p53 mRNAs expression no change of expression was noted for collagen IV (COL4A1, -A2 without inducing p53 expression. As the specificity of siTAp53 and or -A3) and XVIII (COL18A1) known as targets of p53.6 In contrast, siD133p53 has been previously validated, it suggests that IL12A was repressed upon TAp53 isoforms depletion. Interestingly, D133p53 isoforms repress p53 expression in U87 cells cultured only IL12A expression was still significantly induced upon under standard conditions. We and others have shown that p53 cotransfection of U87 cells with siTAp53 and siD133p53. This transactivates, in response to stress, the internal p53 promoter indicates that anti-angiogenic factors expression is differentially giving rise to D133p53 isoforms, thus providing a negative regulated by the expression ratio of D133p53a and p53. Moreover, feedback regulation of the p53 response.3,4,18,23 Altogether, it it suggests that some genes, such as IL12A, are regulated by suggests that p53 and D133p53 isoforms regulate each other’s D133p53a independently of p53. expression in U87 cells. Further experiments will be required to Regarding the pro-angiogenic factors, angiogenin (ANG), decipher the D133p53-mediated repression of p53 in U87 cells. midkine (MDK, neurite growth-promoting factor 2), EGF-like By incubating conditioned media from siRNA-transfected U87 repeats and discoidin I-like domains 3 (EDIL3) and hepatocyte cells with primary human or bovine endothelial cells, we have growth factor (HGF) were significantly downregulated upon demonstrated that the expression ratio of p53 and D133p53 transfection of siD133p53. In contrast, transfection of siTAp53 isoforms regulate endothelial cell migration and tubulogenesis. had no effect on the level of expression of these genes (Figure 6c). Furthermore, their ectopic expression has revealed that D133p53a Interestingly, the level of expression of the classical angiogenic and D133p53g, but not D133p53b, are able to stimulate growth factors VEGF, FGF1 and FGF2 was not significantly affected angiogenesis. As D133p53a and D133p53b, but not D133p53g, by any of the siRNAs. Moreover, the expression of MDK, EDIL3 and are endogenously expressed in U87 cells, we conclude that HGF genes remained unchanged in U87 cells co-transfected with stimulation of angiogenesis is only because of the endogenous

Oncogene (2013) 2150 – 2160 & 2013 Macmillan Publishers Limited D133p53a stimulates angiogenesis H Bernard et al 2157

Anti-angiogenic Factors SiTAp53 - + + SiΔ133p53 +-+ Anti-angiogenic genes MMP2 ** ** IL12A ** ** CXCL2 Pro-angiogenic (14/39 amplified) TIMP2 genes 35% HSPG2 TIMP3 (20/28 amplified) ITGA4 71% FBLN5 ADAMTS1 COL15A1 THBS1 FN1 COL4A2 * * Pro -angiogenic Factors SiTAp53 - + + 4 212 4 212 4 212 4 - + SiΔ133p53 +- +

ANG * * Angiogenic markers and targets MDK ** EDIL3 SiTAp53 - + + HGF * * Δ VEGF Si 133p53 +-+ CST3 EPHB2 EDG1 TGFA NRP1 IL8 TEK FGF2 PDGFRB FGF1 ITGB3

GRN Targets NRP2 VEGFB KDR FST PDGFRA TGFB1 CTGF ITGAV VEGFC ** ANGPTL2 PTN * * PGK1 TNF Markers ANGPTL4 * 621244 212 4 4 212 4 6 4 212 4 212 212 - + - + Figure 6. p53 and D133p53a regulate angiogenesis by differentially modulating expression of angiogenesis-related genes. Fold-change expression of anti- and pro-angiogenic genes upon transfection of U87 cells with siNON, siTAp53, siD133p53 or siTAp53 þ siD133p53 as analysed by TaqMan low-density array. (a) Expression of anti- and pro-angiogenic factors in U87 cells represented as the number of genes amplified in basal conditions (SiNON). Fold-change mRNA expression of 44 genes divided into 4 categories—pro-angiogenic (b), anti- angiogenic (c), markers and target genes (d)–according to Applied Biosystem classification form (Supplementary Table 1). We considered as relevant repression ( À ) or induction ( þ ) variations superior to two folds. Experiments were repeated twice. Details are presented in Table 2.

D133p53a in these cells. Using in vitro and in vivo models, we have regulated, leading to inhibition of angiogenesis. Reciprocally, then established that the p53 and D133p53a regulate angiogen- upon co-transfection of U87 tumour cells with siD133p53 and esis by modulating expression of angiogenic-related genes. siTAp53, which represses (not abolishes) D133p53a expression Hence, an expression ratio of p53 and D133p53a in favour of and maintains p53 expression at basal level, one can expect that D133p53a would stimulate angiogenesis whereas an expression anti-angiogenic factors are repressed, whereas expression of ratio of p53 and D133p53a in favour of p53 would inhibit secreted pro-angiogenic factors is induced, leading to stimulation angiogenesis. of endothelial cell migration and blood vessel formation A previous report showed that D113p53, the zebrafish ortholog (tubulogenesis). Our data suggest that D133p53a upregulation of human D133p53a, antagonizes p53-induced apoptosis via affects the angiogenic balance in favour of angiogenesis, and Bcl2L activation, and differentially modulates p53 target gene provides an explanation to inhibition of HUVEC tubulogenesis expression.3 Moreover, D133p53a prevents replicative senescence in vitro and tumours angiogenesis in vivo observed in our study. by inhibiting transactivation of miR-34a by p53.5 We have recently Therefore, angiogenesis through regulation of angiogenesis- established that D133p53a inhibits p53-mediated apoptosis and related gene expression could be controlled by modulating the G1 cell cycle arrest, without inhibiting p53-mediated G2 cell cycle expression ratio of p53 and D133p53a using specific siRNAs. arrest. This suggests that D133p53a does not exclusively act by Moreover, the fact that some pro and anti-angiogenic gene inhibiting p53. This has been confirmed by the fact that p21, Bcl-2 factors (such as IL12A and ANG) are still regulated upon co- and HDM2 genes are differentially regulated by D133p53a in transfection of siTAp53 and siD133p53, suggests that such genes response to DNA damage. Indeed, D133p53a represses p53 are regulated by D133p53a independently of p53. Furthermore, it transcriptional activity on p21 promoter, whereas it increases p53- is important to note that expression of the classical angiogenic dependent induction of HDM2 expression and induces Bcl-2 VEGF or FGF genes is not significantly altered upon transfection by expression independently of p53.4 Therefore, we concluded that siD133p53, suggesting a regulation of a distinct signalling D133p53a modulates apoptosis and cell cycle progression in pathway involving other angiogenic factors, including ANG, HGF response to stress by regulating gene expression in a p53- and angiopoietin-like 4. dependent and independent manner. The present study demonstrates for the first time the direct role of Our present study of angiogenesis-related gene expression by D133p53a isoform in tumour progression. Such a feature may be the TaqMan low-density array has figured out that upon critical in tumours exhibiting WTp53, which represent about 72% of siD133p53 transfection, which represses D133p53a and induces grade I and 35% of grade II glioblastoma, and for which p53, several anti- and pro-angiogenic factors are differentially angiogenesis has been recognized as a key event in their

& 2013 Macmillan Publishers Limited Oncogene (2013) 2150 – 2160 D133p53a stimulates angiogenesis H Bernard et al 2158 Table 2. Angiogenesis Taqman low density array of D133p53 and p53 knockdown

SiD133p53 SiTAp53 SiD133p53 þ SiTAp53

Moy P-values Moy P-values Moy P-values

Pro-angiogenic factors TNF 3.43 0.08 1.37 0.20 4.70 0.16 ANGPT2 2.15 0.23 6.94 0.25 À 3.12 0.13 PTN 1.48 0.28 2.06 0.05 1.43 0.37 VEGFC 1.28 0.21 3.50 0.04 1.20 0.27 CTGF 1.23 0.10 À 1.07 0.28 À 1.35 0.03 TGFB1 1.21 0.37 1.55 0.06 À 1.25 0.20 FST 1.16 0.30 À 1.28 0.17 À 1.60 0.15 VEGFB 1.15 0.06 1.33 0.16 1.11 0.38 GRN 1.15 0.36 À 1.04 0.08 À 1.09 0.34 FGF1 1.03 0.48 1.00 0.50 1.50 0.08 FGF2 À 1.03 0.48 À 1.53 0.00 À 0.90 0.36 IL8 À 1.03 0.48 À 1.02 0.43 1.02 0.48 TGFA À 1.04 0.43 À 1.12 0.04 À 1.30 0.21 EPHB2 À 1.07 0.45 À 1.04 0.39 À 1.17 0.14 CST3 À 1.31 0.38 À 1.62 0.10 À 1.38 0.36 VEGF À 1.80 0.09 À 1.12 0.39 À 1.86 0.04 HGF À 2.01 0.01 1.06 0.41 À 1.75 0.22 EDIL3 À 2.58 0.10 À 1.25 0.18 À 1.84 0.20 MDK À 2.77 0.01 À 1.01 0.49 1.24 0.01 ANG À 6.91 0.02 1.10 0.44 À 5.08 0.07

Anti-angiogenic factors VASH1 3.70 0.21 4.84 0.24 1.84 0.13 MMP2 3.51 0.02 3.21 0.05 2.02 0.20 IL12A 2.64 0.04 À 2.55 0.05 2.63 0.17 CXCL2 1.78 0.20 2.35 0.15 2.34 0.20 TIMP2 À 1.05 0.27 1.27 0.13 À 1.65 0.04 HSPG2 À 1.12 0.40 1.20 0.26 À 1.72 0.19 TIMP3 À 1.21 0.34 1.11 0.12 À 1.45 0.16 ITGA4 À 1.27 0.37 À 1.12 0.23 À 1.22 0.37 FBLN5 À 1.32 0.17 2.07 0.28 1.05 0.49 ADAMTS1 À 1.38 0.13 À 1.48 0.05 À 1.32 0.30 COL15A1 À 1.46 0.28 1.16 0.29 À 1.74 0.14 THBS1 À 1.53 0.03 À 1.44 0.02 À 1.24 0.25 FN1 À 3.08 0.09 À 1.13 0.21 À 2.94 0.09 COL4A2 À 4.70 0.05 À 1.61 0.10 À 1.93 0.06

Angiogenesis targets EDG 15.23 0.15 1.29 0.38 3.62 0.25 NRP1 1.19 0.37 À 1.04 0.32 1.14 0.36 TEK À 1.01 0.33 À 1.35 0.04 À 2.32 0.06 PDGFRB À 1.22 0.01 1.15 0.17 À 1.04 0.34 ITGB3 À 1.36 0.18 À 1.10 0.26 À 1.48 0.09 NRP2 À 1.51 0.06 À 1.41 0.08 À 1.42 0.08 KDR À 2.04 0.15 À 2.15 0.12 À 1.74 0.07 PDGFRA À 2.39 0.08 À 1.32 0.17 À 1.55 0.31

Angiogenesis markers ITGAV 1.30 0.06 1.13 0.16 1.12 0.19 ANGPTL2 1.07 0.09 À 1.20 0.31 1.32 0.14 PGK1 À 1.44 0.28 1.49 0.09 À 2.65 0.09 ANGPTL4 À 7.93 0.02 1.22 0.20 À 3.15 0.07 Abbreviations: Pro-angiogenic factors. TNF: tumor necrosis factor, ANGPT2: angiopoı¨etin-2, PTN: pleı¨otrophin, VEGFC: vascular endothelial growth factor C, CTGF: connective tissue growth factor, TGFB1: transforming growth factor b 1, FST: follistatin, VEGFB: vascular endothelial growth factor B, GRN: granulin, FGF1: fibroblast growth factor 1, FGF2: fibroblast growth factor 2, IL8: interleukine 8, TGFA: transforming growth factor a, EPHB2: ephrin type-B receptor 2, CST3: cyctatin C, VEGF: vascular endothelial growth factor, HGF: hepatocyte growth factor, EDIL3: EGF-like repeats and discoidin I-like domains 3, MDK: midkine, ANG: angiogenin. Anti-angiogenic factors. VASH1: vasohibin 1, MMP2: matrix metallopeptidase 2, IL12A: interleukin12 subunit a, CXCL2: chemokine (C-X-C motif) ligand 2, TIMP2: TIMP metallopeptidase inhibitor 2, HSPG2: heparan sulfate proteoglycan 2, TIMP3: TIMP metallopeptidase inhibitor 3, ITGA4: integrin a-4, FBLN5: fibulin 5, ADAMTS1: ADAM metallopeptidase with thrombospondin type-1 motif 1, COL15A1: collagen type XV a 1, THBS1: thrombospondin 1, FN1: fibronectin 1. Angiogenesis targets. EDG: endothelial differentiation gene, NRP1: neuropilin 1, TEK: TEK tyrosine kinase endothelial, PDGFRB: platelet-derived growth factor receptor beta polypeptide, ITGB3: integrin b 3, NRP2: neuropilin 2, KDR: kinase insert domain receptor, PDGFRA: platelet-derived growth factor receptor alpha polypeptide. Angiogenesis markers. ITGAV: integrin a V, ANGPTL2: angiopoietin-like 2, PGK1: phophoglycerate kinase 1, ANGPTL4: angiopoietin-like 4.

Oncogene (2013) 2150 – 2160 & 2013 Macmillan Publishers Limited D133p53a stimulates angiogenesis H Bernard et al 2159 progression to malignancy.17,24 Our data may find an important (5  104 cells) were plated in 100 ml of media endothelial growth medium-2 relevance in anti-angiogenic therapeutics of cancer, presently based with 0.5% serum and 300 ml of siRNA tumour-cells-transfected supernatant on angiogenesis inhibitors targeting the VEGF signalling pathway. containing 20 ng/ml FGF2 final. Cultures were photographed and the mean Such treatments have been proven to be efficacious in patient number of branch points (s.e.m.) per microscopic field was determined. survival, associated with a chemotherapy. However, although Studies were performed in triplicate. Cell proliferation was determined by using the MTT [3-(4,5– antitumoural effects and survival benefit are often evident, relapse dimethylthiaxol–2–yl)-2,5-diphenyltetrazolium bromide] assay. HUVECs to progressive tumour growth has been recently reported, reflecting 3 25,26 (2  10 per well) were incubated with endothelial basal medium 2 multiple mechanisms of adaptation to anti-angiogenic therapies. medium with 2% FBS in 96-well plates. HUVECs were serum-starved In this context, D133p53 may have a critical role by stimulating overnight and incubated in 50 ml of media endothelial growth medium-2 angiogenesis through pathways involving several angiogenic factors (EBM-2) with 0.5% serum and 150 ml of siRNA tumour cells transfected distinct from VEGF. Thus, targeting D133p53 may have a strong supernatant containing 3 ng/ml FGF2 final. The medium then was impact in cancer therapeutics. aspirated and MTT was added to each well (0.25 mg/ml). Cells then were incubated for a further 4 h at 37 1C. The medium then was aspirated and the cells were lysed with DMSO and absorbance at 540 nm was measured. MATERIALS AND METHODS Cells and embryos Tumour in CAM and xenografts in mouse HUVECs (Promocell, Heidelberg, Germany) were cultivated in endothelial Three million siRNA-transfected U87 cells were deposited on CAM as growth medium-2 containing 2% fetal bovine serum (Cambrex/Lonza, 19 St Beauzire, France). Adult bovine aortic endothelial cells (primary cells) described. A total of 18 digital photos were taken on day 4, using a were cultivated as previously described.28 Human glioblastoma U87-MG stereomicroscope (Nikon SMZ800, Champigny s/ Marne, France). (No ATCC: HTB-14, provided by LGCpromochem, Molsheim, France), For xenografts, one million cells were injected subcutaneously into nude osteosarcoma U2OS (No ATCC HTB-96) and fertilized chicken eggs (EARL background (n ¼ 8). Animals were killed 10 days later. Six-week-old nude Morizeau, Dangers, France) were handled as described.19 mice were purchased from Janvier (St Berthevin, France). Mice were injected with 106 siRNA-transfected U87 cells subcutaneously into the mid- back region. The tumour size was measured in three dimensions with SiRNA knockdown, cell transfection and western blotting calipers twice weekly starting at Day 3. Mice were observed for any change The siRNAs were purchased from Eurogentec (Liege, Belgium) and Sigma- in behaviour, appearance or weight. At the end of the experiment at Aldrich (Saint-Quentin Fallavier, France). Sequences are provided in Day 10, the mice were killed and xenograft specimens were harvested for Supplementary Table 2. SiNON served as a negative control. SiTAp53 further analyses. targets the full transactivation domain (TA) of p53, and SiD133p53 1 and 2 Tumour volumes were measured according to the international formule target the 50-untranslated region of D133p53. U87 cells were transfected V ¼ 1/2 (length  width2).29 Length and width were estimated using a with siRNAs at 50 nM using InterferIN (Polyplus transfection, New York, NY, stereomicroscope and Adobe photoshop software. USA) or with pD133p53, using FuGENE reagent (Roche, Meylan, France). Western blots were performed as previously described.2 Primary 2 Histology and immunochemistry antibodies were CM-1 and Sapu for p53 and D133p53, respectively, 30 2A10 (Abcam, Cambridge, UK) for mdm2, TUB2.1 (Sigma-Aldrich) for Tumours from CAM xenografts were fixed and cryosectioned as decribed. b-Tubulin and AC-15 (Sigma-Aldrich) for b-Actin. Horseradish peroxidase- Sections of 10 mm were stained with haematoxylin and eosin. For conjugated goat antibodies (Jackson ImmunoResearch, Baltimore, MD, immunohistology studies, the following primary antibodies were used: USA), were used as secondary antibodies in immunoblots. Quantitative Ab-2 (clone V9, NeoMarkers Ab, Montluc¸on, France) for vimentin, Anti Ki-67 analysis of the immunoblot data was performed using the ImageJ 1.40g (AnaSpec Inc., Fremont, CA, USA) for Ki-67, fluorescein-coupled SNA-1 (AbCys, software (Slimware Utilities, D’Iberville, MS, USA). Paris, France) for SNA-lectin. Cell nuclei were visualized by DAPI (4’,6- diamidino-2-phenylindole) dye. Corresponding secondary antibodies were from Molecular Probes (Invitrogen, St Aubin, France). SNA-lectin staining Lentivector production and U87 cell transduction allowed quantification of the capillary network invading the tumour by The cDNAs coding D133p53a, D133P53b or D133P53g, respectively, were counting the number of green pixels on each field with Adobe photoshop subcloned into the lentivector pTRIP-DU3-TRE-MCS, derived from the self- software. The whole surface of tumours has been counted because of inactivating lentivector pTRIP-DU3-EF1a-EGFP.27 The EF1a promoter was capillary staining heterogeneity (at least five tumours per group). replaced by the tetracyclin-inducible pTRE promoter coming from the For mice tumour xenografts, five 5 mM cryosections of tissues were vector pTRE-tight (Clontech, Issy-Les-Moulineaux, France) and a multiple prepared using a Leica cryostat CM3050 (Leica Microsystems, Nanterre, cloning site was inserted in place of the green fluorescent protein (GFP) France). Slides were fixed in ice-cold acetone for 2 min, permeabilized in 0.1% gene (Supplementary Figure S4). The resulting lentivector plasmids were Triton X-100 in phosphate buffered saline (PBS), blocked in 5% bovine serum pTRIP-TRE-D133p53a,-D133P53b, and -D133P53g. In parallel, the trans- albumin in PBS for 1 h at room temperature and then incubated with 5 mg/ml activator TetOn rTTA-advanced from Clontech was subcloned into the primary antibodies for 1 h at room temperature. Blood vessels were detected lentivector pTRIP-DU3-CMV-MCS in which we had inserted the CMV with 5 mg/ml anti-CD31 (BD Pharmingen, Le Pont de Claix, France) antibody. promoter. U87 cells were transduced with the resulting lentivector pTRIP- After extensive washing, slides were incubated with 1–2 mg/ml cross- rTTA-TetOn to generate U87 TetOn cells. Lentivector particles were absorbed goat anti-rat DyLight 549 (Tebu-bio, Le Perray en Yvelines, France) produced by the Inserm U1037 vector facility (Toulouse, France) from secondary antibody for 1 h at room temperature. Slides were counterstained HEK293 cell co-transfection by each plasmid pTRIP with plasmids pLvPack with DAPI (Tebu-bio). Coverslips were mounted with Dako Cytomation and pLvVSVg (Sigma-Aldrich) coding HIV1 capsid proteins and vasicular fluorescent mounting medium (Dako, Trappes, France). For quantification, stomatitis virus G envelope protein, respectively, as previously described. number of blood vessels in 5–10 microscopic fields per cryosection (per animal) were quantified and the mean number of vessels ±s.e.m. for the entire treatment group determined. Endothelial cell migration, tube formation and proliferation HUVECs migration in the ‘wounding assay’ was performed during 8 h after treatment with conditioned media from siRNA-transfected U87 or U2OS RNA extraction, nested RT–PCR, real time quantitative RT–PCR and cells, as described previously.28 Conditioned media were harvested 72 h TaqMan low-density array after cell transfection and administrated to HUVEC 48 h after starvation. Total RNA was isolated from U87 cells using Nucleospin RNA II kit Adult bovine aortic endothelial cells migration in the transwell migration (Macherey Nagel, Hoerdt, France), assessed and nested RT–PCR was assay was performed in Boyden Chambers (Millipore, Molsheim, France) performed as decribed previously.2 according to the manufacturer’s instruction. Quantification was performed The quantitative PCR was performed on StepOne þ (Applied Biosystems, 6 h after treatment with conditioned media. Villebon s/ Yvette, France) using 10–100 ng of cDNA and Power SYBER To assess the ability of cultured HUVECs to form vessel-like structures in Green or TaqMan Universal PCR master mix (Applied Biosystems) for culture (tube formation), Matrigel (BD Biosciences, Le Pont de Claix, France) detection of FLp53, D133p53 and 18S. was added to the wells of an 24-well plate in a volume of 300 ml and TaqMan low-density array (Applied Biosystems) was performed follow- allowed to solidify at 37 1C for 30 min. After the Matrigel solidified, HUVECs ing manufacturer’s instructions with 50 ng of cDNA. The results were

& 2013 Macmillan Publishers Limited Oncogene (2013) 2150 – 2160 D133p53a stimulates angiogenesis H Bernard et al 2160 treated with the StepOne Software v2.0 (Applied Biosystems). Primers 12 Pal S, Datta K, Mukhopadhyay D. Central role of p53 on regulation of vascular sequences are provided in Supplementary Table 2. permeability factor/vascular endothelial growth factor (VPF/VEGF) expression in mammary carcinoma. Cancer Res 2001; 61: 6952–6957. Statistical analysis 13 Ravi R, Mookerjee B, Bhujwalla ZM, Sutter CH, Artemov D, Zeng Q et al. Regulation All statistical determination was performed with GraphPad Prism, version of tumor angiogenesis by p53-induced degradation of hypoxia-inducible factor 4.0 (GraphPad, San Diego, CA, USA). All data are presented as the 1alpha. Genes Dev 2000; 14: 34–44. mean±s.e.m. One-way and two-way analysis of variance was used to 14 Ueba T, Nosaka T, Takahashi JA, Shibata F, Florkiewicz RZ, Vogelstein B et al. evaluate the significance of differences between groups. All statistical Transcriptional regulation of basic fibroblast growth factor gene by p53 in human analyses were performed with a two-tailed Student’s t-test. A P-value of glioblastoma and hepatocellular carcinoma cells. Proc Natl Acad Sci USA 1994; 91: less than 0.05 was considered statistically significant. 9009–9013. 15 Giuriato S, Ryeom S, Fan AC, Bachireddy P, Lynch RC, Rioth MJ et al. Sustained regression of tumors upon MYC inactivation requires p53 or thrombospondin-1 to CONFLICT OF INTEREST reverse the angiogenic switch. Proc Natl Acad Sci USA 2006; 103: 16266–16271. 16 Naumov GN, Akslen LA, Folkman J. Role of angiogenesis in human tumor dor- The authors declare no conflict of interest. mancy: animal models of the angiogenic switch. Cell Cycle 2006; 5: 1779–1787. 17 Wong ML, Prawira A, Kaye AH, Hovens CM. Tumour angiogenesis: its mechanism ACKNOWLEDGEMENTS and therapeutic implications in malignant gliomas. J Clin Neurosci 2009; 16: 1119–1130. We thank A Delluc-Clavie`res and F Pujol for technical assistance, Y Barreira, S 18 Marcel V, Vijayakumar V, Fernandez-Cuesta L, Hafsi H, Sagne C, Hautefeuille A Legonidec (animal production and phenotyping facilities of ANEXPLO platform, et al. p53 regulates the transcription of its Delta133p53 isoform through specific Inserm US006), F Gross (Vector facility), J-J Maoret (GeT TQ plateau of the Genotoul response elements contained within the TP53 P2 internal promoter. Oncogene Genome-Transcriptome platform), M Pucelle (CAM facility) and C Touriol (ABAE and 2010; 29: 2691–2700. HUVEC cells). This work was supported by grants from Association pour la Recherche 19 Hagedorn M, Javerzat S, Gilges D, Meyre A, de Lafarge B, Eichmann A et al. 0 sur le Cancer, Cance´ropole GSO, INCA, Fondation de l Avenir, Association Franc¸aise Accessing key steps of human tumor progression in vivo by using an avian contre les Myopathies (AFM). HB had a fellowship from the Ligue Nationale Contre Le embryo model. Proc Natl Acad Sci USA 2005; 102: 1643–1648. Cancer, then from the ARC. BGS had a postdoc fellowship from the Fondation pour la 20 Khoury MP, Bourdon JC. p53 Isoforms: an intracellular microprocessor? Genes Recherche Me´dicale, NA had a thesis fellowship from AFM. DPL and JCB were Cancer 2011; 2: 453–465. supported by Cancer Research UK (grant number: C8/A6613). 21 Bourdon JC, Khoury MP, Diot A, Baker L, Fernandes K, Aoubala M et al. p53 mutant breast cancer patients expressing p53gamma have as good a prognosis as wild-type p53 breast cancer patients. Breast Cancer Res 2011; 13:R7. REFERENCES 22 Davidson WR, Kari C, Ren Q, Daroczi B, Dicker AP, Rodeck U. Differential regulation 1 Bourdon JC. p53 and its isoforms in cancer. Br J Cancer 2007; 97: 277–282. of p53 function by the N-terminal DeltaNp53 and Delta113p53 isoforms in zeb- 2 Bourdon JC, Fernandes K, Murray-Zmijewski F, Liu G, Diot A, Xirodimas DP et al. rafish embryos. BMC Dev Biol 2010; 10: 102. p53 isoforms can regulate p53 transcriptional activity. Genes Dev 2005; 19: 23 Nakagawa T, Takahashi M, Ozaki T, Watanabe Ki K, Todo S, Mizuguchi H et al. 2122–2137. Autoinhibitory regulation of p73 by Delta Np73 to modulate cell survival and 3 Chen J, Ng SM, Chang C, Zhang Z, Bourdon JC, Lane DP et al. p53 isoform death through a p73-specific target element within the Delta Np73 promoter. Mol delta113p53 is a p53 target gene that antagonizes p53 apoptotic activity via BclxL Cell Biol 2002; 22: 2575–2585. activation in zebrafish. Genes Dev 2009; 23: 278–290. 24 Ohgaki H, Kleihues P. Genetic pathways to primary and secondary glioblastoma. 4 Aoubala M, Murray-Zmijewski F, Khoury MP, Fernandes K, Perrier S, Bernard H et Am J Pathol 2007; 170: 1445–1453. al. p53 directly transactivates Delta133p53alpha, regulating cell fate outcome in 25 Ebos JM, Lee CR, Cruz-Munoz W, Bjarnason GA, Christensen JG, Kerbel RS. response to DNA damage. Cell Death Differ 2011; 18: 248–258. Accelerated metastasis after short-term treatment with a potent inhibitor of 5 Fujita K, Mondal AM, Horikawa I, Nguyen GH, Kumamoto K, Sohn JJ et al. p53 tumor angiogenesis. Cancer Cell 2009; 15: 232–239. isoforms Delta133p53 and p53beta are endogenous regulators of replicative 26 Paez-Ribes M, Allen E, Hudock J, Takeda T, Okuyama H, Vinals F et al. Anti- cellular senescence. Nat Cell Biol 2009; 11: 1135–1142. angiogenic therapy elicits malignant progression of tumors to increased local 6 Teodoro JG, Evans SK, Green MR. Inhibition of tumor angiogenesis by p53: a new invasion and distant metastasis. Cancer Cell 2009; 15: 220–231. role for the guardian of the genome. J Mol Med 2007; 85: 1175–1186. 27 Sirven A, Ravet E, Charneau P, Zennou V, Coulombel L, Guetard D et al. Enhanced 7 Fraisl P, Mazzone M, Schmidt T, Carmeliet P. Regulation of angiogenesis by transgene expression in cord blood CD34( þ )-derived hematopoietic cells, oxygen and metabolism. Dev Cell 2009; 16: 167–179. including developing T cells and NOD/SCID mouse repopulating cells, following 8 Pouyssegur J, Dayan F, Mazure NM. Hypoxia signalling in cancer and approaches transduction with modified trip lentiviral vectors. Mol Ther 2001; 3: 438–448. to enforce tumour regression. Nature 2006; 441: 437–443. 28 Bossard C, Van den Berghe L, Laurell H, Castano C, Cerutti M, Prats AC et al. 9 Dameron KM, Volpert OV, Tainsky MA, Bouck N. Control of angiogenesis Antiangiogenic properties of fibstatin, an extracellular FGF-2-binding polypeptide. in fibroblasts by p53 regulation of thrombospondin-1. Science 1994; 265: Cancer Res 2004; 64: 7507–7512. 1582–1584. 29 Jensen MM, Jorgensen JT, Binderup T, Kjaer A. Tumor volume in subcutaneous 10 Folkman J. Tumor suppression by p53 is mediated in part by the antiangiogenic mouse xenografts measured by microCT is more accurate and reproducible than activity of endostatin and tumstatin. Sci STKE 2006; 2006: pe35. determined by 18F-FDG-microPET or external caliper. BMC Med Imag 2008; 8:16. 11 Galy B, Creancier L, Zanibellato C, Prats AC, Prats H. Tumour suppressor p53 30 Saidi A, Javerzat S, Bellahcene A, De Vos J, Bello L, Castronovo V et al. Experi- inhibits human fibroblast growth factor 2 expression by a post-transcriptional mental anti-angiogenesis causes upregulation of genes associated with poor mechanism. Oncogene 2001; 20: 1669–1677. survival in glioblastoma. Int J Cancer 2008; 122: 2187–2198.

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

Oncogene (2013) 2150 – 2160 & 2013 Macmillan Publishers Limited