Oncogene (2010) 29, 5989–6003 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 www.nature.com/onc ORIGINAL ARTICLE A novel concept in antiangiogenic and antitumoral therapy: multitarget destabilization of short-lived mRNAs by the zinc finger ZFP36L1

S Planel1,2,3, A Salomon1,2,3, P Jalinot4, J-J Feige1,2,3 and N Cherradi1,2,3

1Institut National de la Sante´ et de la Recherche Me´dicale, Grenoble, Rhoˆne-Alpes, France; 2Commissariat a` l’Energie Atomique, Institut de Recherches en Technologies et Sciences pour le Vivant, LAPV, Grenoble, France; 3Universite´ Joseph Fourier, Grenoble, France and 4LBMC/UMR5239 CNRS, ENS Lyon, Lyon, France

Angiogenesis inhibitors have shown clinical benefits in Introduction patients with advanced cancer, but further therapeutic improvement is needed. We have previously shown that the Increased vascular endothelial growth factor (VEGF) zinc finger protein 36, C3H type-like 1 (ZFP36L1) expression is associated with many angiogenesis depen- enhances vascular endothelial growth factor (VEGF) dent pathologies, such as cancer (Ferrara, 2004) and mRNA decay through its interaction with AU-rich ocular diseases (Gariano et al., 2006). Therefore, elements within VEGF 30-untranslated region. In this intensive efforts have been undertaken over the past study, we evaluated the possibility to develop an decade to develop therapeutic strategies in order to antiangiogenic and antitumoral strategy using the inhibit the pathological angiogenesis via neutralization mRNA-destabilizing activity of ZFP36L1. We engineered of VEGF or its receptors by antibodies, or blocking a cell-penetrating ZFP36L1, by fusing it to the protein VEGF receptor activation and signaling with tyrosine transduction domains (PTDs) TAT derived from HIV, or kinase inhibitors (Ellis and Hicklin, 2008). There has the polyarginine peptides R7 or R9. PTD-ZFP36L1 been success in combining antiangiogenic agents with fusion were expressed in bacterial cells and conventional chemotherapy, such as the successful use affinity-purified to homogeneity. TAT-, R7- and R9- of the monoclonal anti-VEGF antibody bevacizumab ZFP36L1 were efficiently internalized into living cells and (Avastin, Genentech/Roche Laboratories, San Francisco, decreased both endogenous VEGF mRNA half-life and CA, USA) with various chemotherapies for metastatic VEGF protein levels in vitro. Importantly, a single colorectal, renal, non-small-cell lung and metastatic injection of R9-TIS11b fusion protein into a high-VEGF breast cancer (Ellis and Hicklin, 2008). However, despite expressing tissue in vivo (in this study, the mouse adrenal the initial enthusiasm predicting the absence of resis- gland) markedly decreased VEGF expression. We further tance to these antiangiogenic treatments, such resistance evaluated the effect of R9-ZFP36L1 on tumor growth appeared to occur on time (Bergers and Hanahan, using Lewis Lung Carcinoma (LL/2) cells implanted 2008). There is therefore a real need for a variety of subcutaneously into nude mice. Intratumoral injection of antiangiogenic drugs that will target the angiogenic R9-ZFP36L1 significantly reduced tumor growth and process through distinct mechanisms. markedly decreased the expression of multiple angiogenic VEGF expression is regulated by transcriptional and and inflammatory cytokines, including VEGF, acidic post-transcriptional mechanisms. At the post-transcrip- fibroblast growth factor, tumor necrosis factor a, inter- tional level, VEGF mRNA stability is tightly controlled leukin (IL)-1a and IL-6, with a concomitant obliteration (Levy, 1998; Ciais et al., 2004; Onesto et al., 2004; of tumor vascularization. These findings indicate that R9- Cherradi et al., 2006). VEGF mRNA stability is ZFP36L1 fusion protein may represent a novel antiangio- regulated by the binding of stabilizing and destabilizing genic and antitumoral agent, and supports the emerging proteins to AU-rich elements (AREs) located in the 30- idea that modulation of mRNA stability represents a untranslated region (30-UTR) of VEGF mRNA. We promising therapeutic approach to treat cancer. have previously characterized ZFP36L1 as a VEGF Oncogene (2010) 29, 5989–6003; doi:10.1038/onc.2010.341; mRNA-destabilizing protein (Ciais et al., 2004). published online 30 August 2010 ZFP36L1 is a member of the tristetraprolin (TTP) family of CCCH tandem zinc-finger proteins, consisting Keywords: tumor angiogenesis; VEGF; ZFP36L1; of three members, which are expressed in all mammals tristetraprolin; mRNA stability; multitarget therapy (TTP also named TIS11 or ZFP36, ZFP36L1 also named TIS11b or BRF1 and ZFP36L2 also named TIS11d or BRF2) and a fourth member only present in Correspondence: Dr N Cherradi, Institut National de la Sante´et de la rodents (ZFP36L3). All four proteins bind and destabi- Recherche Me´dicale, INSERM U878/LAPV/iRTSV, CEA-Grenoble, lize ARE-containing mRNAs in vitro (Baou et al., 2009). 17 Rue des Martyrs, Grenoble Cedex 09, 38054, France. However, knockout studies have provided evidence E-mail: [email protected] Received 5 February 2010; revised 13 June 2010; accepted 1 July 2010; for their unique role in vivo (Carballo et al., 1998; published online 30 August 2010 Stumpo et al., 2004, 2009). ZFP36L1 knockout mice are R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5990 embryonic lethal because of abnormal placentation and NcoI KpnI NotI major vascular defects. Both TTP and ZFP36L1 enhance the decay of their targets by association with FLAG PTD ZFP36L1 the mRNA decay machinery (Lykke-Andersen and PreScission Wagner, 2005). Site Protein transduction domains (PTDs) have emerged PTD TAT :YGRKKRRQRRR as novel tools for delivering biologically active macro- PTD R9 : RRRRRRRRR molecules into cells (Schwarze et al., 1999). The HIV-1 PTD R7 : RRRRRRR transactivator TAT is capable of crossing the plasma TAT- R7- R9- membrane and thereby mediating the intracellular Control ZFP36L1 ZFP36L1 ZFP36L1 ZFP36L1 delivery of heterologous proteins (Chauhan et al., Plasmid 12510 12510 1 10 25 1 10 25 1 10 25 2007). Most of the recent studies indicate that PTDs (ng) 75 use various forms of endocytosis (Richard et al., 2003; 50 Snyder and Dowdy, 2004; Asoh and Ohta, 2008). They have been used to deliver a variety of cargos including 37 proteins, peptides and nucleic acids in vitro, and to kDa successfully treat several preclinical models of human 120 ** *** diseases such as cancer and cerebral ischemia (Lindsay, ** ****** *** 100 ** *** *** Control 2002; Snyder and Dowdy, 2004; Asoh and Ohta, 2008). ** *** *** ZFP36L1 The homopolymers of arginine have been shown to 80 TAT-ZFP36L1 R7-ZFP36L1 transduce various cell types with higher efficiencies than 60 R9-ZFP36L1 the 11 amino-acids of TAT (Wender et al., 2000; Futaki 40

et al., 2001; Ho et al., 2001). (% of Control) In this study, we hypothesized that fusion of 20 Firefly Luc/Renilla Luc ZFP36L1 to a PTD would allow delivery of ZFP36L1 0 into cells and target not only VEGF but also a number 12510 of inflammatory cytokines and products of stress- Plasmid (ng) response whose mRNA 30-UTRs possess AREs. Figure 1 PTD-ZFP36L1 constructs decrease VEGF mRNA We confirmed this hypothesis by demonstrating that 30-UTR-mediated luciferase activity. (a) Schematic representation of cell-penetrating R9-ZFP36L1 markedly reduces the ZFP36L1 fusion-proteins and amino-acid sequences of the protein transduction domains (PTDs) studied. The PreScission site allows physiological expression of VEGF in vivo and leads to cleavage of the fusion protein by PreScission protease after a significant inhibition of tumor growth and vascular- purification to yield a fusion protein devoid of the Flag tag. ization with a concomitant downregulation of VEGF (b) Western blot analysis of the expression levels of ZFP36L1 and and several proangiogenic inflammatory cytokines. PTD-ZFP36L1 constructs in COS7 cells showing that ZFP36L1, and After proper vectorization, ZFP36L1 might thus repre- the fusion proteins TAT-, R7-and R9-ZFP36L1 are equally expressed at each dose of the transfected plasmid. Control cells refer to sent a promising second generation antiangiogenic and transfection with pTarget empty vector. The apparent molecular antitumorigenic agent. weight of ZFP36L1 and PTD-ZFP36L1 proteins (TAT-, R7- and R9- ZFP36L1) were 48 and 51 kDa, respectively. (c) Dose-dependence of the inhibitory effect of ZFP36L1 and PTD-ZFP36L1 on VEGF mRNA 30-UTR-mediated luciferase activity. COS7 cells were transfected with various amounts of ZFP36L1, TAT-ZFP36L1, R7- Results ZFP36L1 or R9-ZFP36L1 pTarget plasmids in the presence of 500 ng of pLuc-30-UTR (Firefly luciferase) and 25 ng of pRL-TK (Renilla Overexpressed PTD-ZFP36L1 fusion proteins decrease luciferase) plasmids for 24 h as described in Supplementary Materials 0 reporter gene activity through VEGF 30-UTR and Methods. The pLuc-3 -UTR construct contains the full-length 30-UTR of rat VEGF mRNA (2201 bp) (Ciais et al., 2004). Firefly A previous study using transfection of a ZFP36L1 luciferase and Renilla luciferase activities were measured as stated in expression vector and a reporter gene construct com- Supplementary Materials and Methods. Results are expressed as prising the firefly luciferase cDNA, cloned upstream of relative light units of firefly luciferase activity over relative light units VEGF mRNA 30-UTR (Luc-30-UTR), allowed us to of Renilla luciferase activity to compensate for variations in 0 transfection efficiency and are represented as a percentage of demonstrate that VEGF mRNA 3 -UTR mediates a luciferase activity in control cells transfected with empty vector. With ZFP36L1-induced decrease in reporter gene activity, 25 ng of ZFP36L1, TAT-ZFP36L1, R7-ZFP36L1 or R9-ZFP36L1 which was because of a decrease in luciferase transcript transfected plasmids, luciferase activity was 33.6±6.5 (ZFP36L1), stability (Ciais et al., 2004). In order to check that the 43.5±9.7 (TAT-ZFP36L1), 42.2±9 (R7-ZFP36L1) and 46.5±5.6% fusion of TAT, R7 or R9 to ZFP36L1 does not alter (R9-ZFP36L1) of controls (Po0.001, n ¼ 3). Transfections were performed in triplicate and values are means ± s.e.m. from four ZFP36L1 function, we first analyzed the effect of each independent experiments. Each value was compared to its respective fusion protein on luciferase activity of Luc-30-UTR control using ANOVA with Dunnett’s post-test. **, ***, significantly construct. COS7 cells were transfected with increasing different from control with Po0.01 and Po0.001, respectively. doses of expression vectors encoding ZFP36L1, TAT- ZFP36L1, R7-ZFP36L1 or R9-ZFP36L1 (Figure 1a), and a fixed concentration of Luc-30-UTR (500 ng). The activity was markedly decreased by ZFP36L1 and PTD- expression level of each construct was assessed in COS7 ZFP36L1 in a dose-dependent manner (Figure 1c). The cell lysates by western blot (Figure 1b). Luciferase effect of ZFP36L1 was not statistically different from

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5991 the effect of the fusion proteins PTD-ZFP36L1 at all of ZFP36L1 in bacteria yields a 38 kDa protein (in doses tested, indicating that ZFP36L1 and PTD- agreement with the molecular weight deduced from the ZFP36L1 were equally efficient in the inhibition of amino-acid sequence), whereas overexpression in mam- reporter gene activity. malian cells yields a 48 kDa protein (Figure 1b), indicating post-translational modifications of ZFP36L1 in mammals. PTD-ZFP36L1 fusion proteins enhance endogenous VEGF mRNA decay and impair VEGF protein production In order to investigate the effect of ZFP36L1 and PTD- ZFP36L1 proteins on endogenous VEGF mRNA PTD-ZFP36L1 fusion proteins are efficiently delivered stability, COS7 cells were transfected either with into live cells ZFP36L1, TAT-ZFP36L1 and R7-ZFP36L1, or with Internalization of purified ZFP36L1 (negative control) R9-ZFP36L1 expression vectors then used in 5,6- and PTD-ZFP36L1 proteins was examined in live cells. dichloro-1-b-d-ribofuranosylbenzimidazole (DRB) time COS7 cells were incubated with 100 nM of Alexa Fluor course experiments. Overexpression of ZFP36L1, TAT- 488 dye-labeled ZFP36L1 or PTD-ZFP36L1 for 2 h. ZFP36L1, R7-ZFP36L1 and R9-ZFP36L1 triggered Fluorescence microscopy analysis showed that very little rapid VEGF mRNA decay in COS7 cells with a half- staining was detected in COS7 cells incubated with life of 62±4, 56±3, 65±6 and 71±5 min, respectively, Alexa-labeled ZFP36L1 (Figure 4a). In contrast, we as compared with 101±10 min for control cells trans- observed a significant uptake of Alexa-labeled R9- fected with an empty plasmid (Figures 2a and b, left ZFP36L1 (green fluorescent signal), which indicates panel). ZFP36L1, TAT-ZFP36L1, R7-ZFP36L1 and that R9 was efficient in delivering ZFP36L1 to cell R9-ZFP36L1 decreased the steady state level of cytoplasm (Figure 4a). In addition, evidence for VEGF mRNA to 61±10, 62±9, 43±6 and 52±5% successful PTD-ZFP36L1 intracellular delivery was of control, respectively (Figure 2b, right panel). ELISA provided by confocal scanning microscopy (Figure 4b). analysis of culture medium from COS7 cells revealed Internalized TAT-ZFP36L1 and R9-ZFP36L1 were that ZFP36L1, TAT-ZFP36L1, R7-ZFP36L1 and distributed throughout the cytoplasm with a punctuate R9-ZFP36L1 reduced the VEGF secretion down to appearance, indicative of localization in transport 47.9±1.9, 46.6±8.4, 54.4±7.6 and 48.5±4.2% of vesicles like endosomes. Deconvolution of the confocal control, respectively (Figure 2c). These data indicated microscopy images along the z axis confirmed that R9- that ZFP36L1- and PTD-ZFP36L1-induced decreases ZFP36L1 was indeed internalized in the COS7 cell in VEGF mRNA stability are accompanied by a cytoplasm (Figure 4c). decrease in VEGF protein production. This prompted us to produce and purify recombinant fusion PTD-ZFP36L1 fusion proteins inhibit VEGF mRNA proteins. and protein expression in live cells We next evaluated the effect of TAT-, R7- or R9- ZFP36L1 and PTD-ZFP36L1 protein production ZFP36L1 purified proteins on endogenous VEGF and purification mRNA levels. COS7 cells were incubated with 100 nM We constructed negative control ZFP36L1 without PTD of protein for 24 h and VEGF mRNA was analyzed by and PTD-ZFP36L1 bacterial expression vectors. Ex- northern blot (Figure 5a). Internalization of TAT-, pression and purification of the recombinant proteins R7- and R9-ZFP36L1 in two independent experiments are described in details in Supplementary Materials and reduced VEGF mRNA steady state levels to 63.3±10.1, Methods. The major problems encountered during our 44.4±5.3 and 49.8±3.1% of control level (vehicle), initial production and purification procedures were the respectively (Figure 5b). ZFP36L1, which has no cell- accumulation of PTD-ZFP36L1 proteins into inclusion penetrating ability, displayed only a discrete effect on bodies, as well as their precipitation, thus confirming the the basal expression of VEGF mRNA (80.4±9.6% difficulty of expressing and purifying TTP family of control level). We further investigated whether members from various expression systems (Cao, 2004; the PTD-ZFP36L1-induced decreases in VEGF mRNA Cao et al., 2008; Cao and Lin, 2009). However, we was accompanied by decreases in VEGF protein succeeded in increasing the solubility of PTD-ZFP36L1 production. As shown in Figure 5c, VEGF amounts fusion proteins by lowering the expression temperature in the culture medium from R7-ZFP36L1- and and optimizing the isopropyl b-D-1-thiogalactopyrano- R9-ZFP36L1-transduced COS7 cells were significantly side treatment. In addition, an increased stability was decreased to 64.4±10.8 and 63.03±14.1% of observed in the presence of zinc chloride (100 mM). control cells (n ¼ 3, Po0.05). No significant effect of Figure 3A illustrates the different steps of TAT-ZFP36L1 TAT-ZFP36L1 on VEGF secretion was observed. This purification on a Flag-affinity chromatography column. latter result is in contrast to the one obtained in Purification of ZFP36L1, R7- and R9-ZFP36L1 was transfection experiments, which showed a marked achieved using the same procedure. SDS–PAGE analy- inhibition of VEGF secretion in TAT-ZFP36L1 trans- sis of the eluates identified homogeneous preparations fected cells (Figure 1c). The above results led us to for ZFP36L1 and PTD-ZFP36L1 proteins with expected choose R9-ZFP36L1 fusion protein to examine the molecular weights of 38 and 42 kDa, respectively effect of ZFP36L1 transduction on VEGF expression (Figure 3B). It is worth mentioning that overexpression in vivo.

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5992 Control ZFP36L1 TAT-ZFP36L1 R7-ZFP36L1 R9-ZFP36L1 VEGF mRNA

18S

0 30 60 120180 0 30 60 120 180 0 30 60 120 180 0 30 60 120 180 0 30 60 120 180 Time of DRB treatment (min)

100 120 Control ZFP36L1 100 TAT-ZFP36L1 50 R7-ZFP36L1 80 R9-ZFP36L1 ** 60 * **

Control VEGF/18S (%) 40 ZFP36L1 TAT-ZFP36L1 VEGF mRNA remaining (%) 20 R7-ZFP36L1 R9-ZFP36L1 10 0 040 80 120 160 200 Time 0 prior to addition of DRB Time of DRB treatment (min)

140 Control 120 ZFP36L1 TAT-ZFP36L1 R7-ZFP36L1 100 R9-ZFP36L1

80

60

40

20 VEGF protein (% of control)

0 12510 Plasmid (ng) Figure 2 PTD-ZFP36L1 proteins decrease endogenous VEGF mRNA half-life and VEGF secretion by COS7 cells. (a) COS7 cells were transfected in 12-well plates (1.5 Â 106 cells/well) with 5 ng of either pTarget empty plasmid (control cells) or ZFP36L1 or PTD-ZFP36L1 plasmids. At 48 h after transfection, the inhibitor 5,6-dichloro-1-b-d-ribofuranosylbenzimidazole (10 mg/ml) was added and total RNA was extracted at the time points indicated and analyzed by northern blot. The membrane was hybridized to a radiolabeled VEGF 30-UTR probe and rehybridized to 18S RNA probe for loading control. Membranes were exposed either overnight or for 6 h for VEGF mRNA or 18S RNA detection, respectively. The two arrows indicate the two major species of VEGF mRNA detected in COS7 cells. Shown is a representative northern blot of three independent experiments. (b) Left panel: measurement of VEGF mRNA half-life in COS7 cells transfected as described in (a). VEGF mRNA values were normalized to 18S RNA values and plotted as a percentage of the initial value against time. The data shown represent the mean ± s.e.m. of three independent experiments. Each value was compared with its respective control (pTarget empty plasmid) using ANOVA with Dunnett’s post-test. For t ¼ 30 min, ZFP36L1, TAT- ZFP36L1, R7-ZFP36L1 and R9-ZFP36L1 were not significantly different from control. For t ¼ 60 min, ZFP36L1 and TAT-ZFP36L1 were significantly different from control with Po0.01, and R7-ZFP36L1 and R9-ZFP36L1 were significantly different from control with Po0.05. For t ¼ 120 and 180 min, ZFP36L1, TAT-ZFP36L1, R7- ZFP36L1 and R9-ZFP36L1 were significantly different from control with Po0.01). Right panel: Quantification of VEGF mRNA steady state levels at time 0 before addition of 5,6-dichloro-1-b-d-ribofuranosylbenzimidazole. Each point represents the mean± s.e.m. of three independent experiments performed in triplicate (*, **, significantly different from control with Po0.05 and Po0.01, respectively). (c) Dose-dependence of the inhibitory effect of ZFP36L1 and PTD-ZFP36L1 on VEGF protein levels in COS7 cells transfected with increasing amounts of ZFP36L1 and PTD-ZFP36L1. At 24 h after transfection, VEGF protein was measured in the culture medium by ELISA. VEGF protein values were normalized to total protein values and represented as a percentage of VEGF protein level in control cells. Each point is the mean ± s.d. of triplicate samples from an experiment representative of three similar experiments.

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5993 MW 1 2 3 4 5

75 50 37 TAT-ZFP36L1 25 20 15

250 150 100 75 50 37

25

R7- R9- TAT- R7- R9- TAT- ZFP36L1 ZFP36L1 ZFP36L1 ZFP36L1 ZFP36L1 ZFP36L1ZFP36L1 ZFP36L1 Figure 3 SDS–PAGE analysis of the purification steps of ZFP36L1 and PTD-ZFP36L1 recombinant proteins from Escherichia coli. (A) TAT-ZFP36L1 was induced by isopropyl b-D-1-thiogalactopyranoside in E. coli, extracted and purified with an anti-Flag affinity column. As purified Flag-PTD-ZFP36L1 proteins were not digested to a significant extent by the protease and the Flag tag did not alter ZFP36L1 activity, we used the full-length Flag-PTD-ZFP36L1 or Flag-ZFP36L1 fusion proteins in our studies. Eluted fusion protein was analyzed by SDS–PAGE and detected with Commassie brilliant blue staining. MW: protein molecular weight standards, lane 1: non-induced E. coli culture extract, lane 2: induced E. coli culture extract, Lane 3: filtrate of the anti-Flag affinity column, Lane 4: TBS buffer wash, lane 5: eluate. The arrow indicates the position of the purified protein. (B) Purified ZFP36L1, TAT-ZFP36L1, R7-ZFP36L1 and R9-ZFP36L1 fusion proteins (700–850 ng) were analyzed by SDS–PAGE and visualized by silver-staining (panels a and b). In order to confirm the identity of the purified material, fusion proteins (250–700 ng) were further analyzed by western blot (c) using anti-ZFP36L1 antibodies as described in Supplementary Materials and Methods.

Injection of R9-ZFP36L1 into adrenocortical tissue RT–PCR (Figure 6b). R9-ZFP36L1 injection down- induces a marked reduction in VEGF protein expression regulates the levels of all VEGF mRNA isoforms The adrenal cortex is a steroid hormone-producing detected in mouse adrenal cortex, within 24 h. This tissue and one of the most highly vascularized tissues in effect was even more pronounced at 48 h after injection. the organism. To evaluate the bioactivity of R9- ZFP36L1 on VEGF expression in vivo, the fusion proteins were injected in the adrenal glands of severe Intratumoral injection of R9-ZFP36L1 inhibits LL/2 combined immunodeficiency mice. Immunohistochem- tumor growth ical analysis of VEGF revealed a strong and uniform We next evaluated the effect of R9-ZFP36L1 on the staining in the adrenal cortex of control mice (Figure 6a, growth of luciferase-expressing Lewis Lung Carcinoma M1 and M2 mice). At 24 h after R9-ZFP36L1 injection, (LL/2) cells’ allografts in mice. As the engineered a marked decrease in VEGF staining was observed ZFP36L1 was of human origin, these experiments were throughout the entire zona fasciculata-reticularis of the conducted in nude mice (nu/nu). Tumors were allowed adrenal cortex (Figure 6a, M3 and M4 mice). Some to develop subcutaneously (up to a size of 100 mm3), VEGF immunoreactivity was still detected in the zona after which R9-ZFP36L1 (40 ng) or vehicle was injected glomerulosa, the outer part of the adrenal cortex. While into the tumors every other day. As shown in Figure 7a, the capillaries in the zona fasciculata-reticularis paral- at day 20 post-implantation, R9-ZFP36L1 significantly leled the columns of steroidogenic cells in control inhibited LL2 tumor growth by 42% as measured by animals, the vascular network in R9-ZFP36L1-treated caliper (mean 1104±116 mm3 for controls, versus animals appeared discontinuous. These histological 634±62 mm3 for R9-ZFP36L1-treated group; n ¼ 5 for modifications were reminiscent of those obtained during each group, Po0.001). Injection of non-penetrating dexamethasone-induced adrenal cortex regression, a ZFP36L1 (negative control) had no effect on tumor condition when VEGF protein levels are also dramati- growth (Supplementary Figure 1). The inhibitory effect cally reduced (Thomas et al., 2004). In order to check R9-ZFP36L1 was confirmed and even more pronounced whether R9-ZFP36L1-mediated decrease in VEGF by measuring luciferase activity of live cells at day 20 protein expression in the adrenal cortex was correlated using bioluminescence imaging (Figure 7b). R9- with VEGF mRNA levels, the expression of the ZFP36L1 induced a 52% reduction in luciferase activity different VEGF mRNA isoforms was analyzed by in R9-ZFP36L1-treated tumors (mean activity

Oncogene 5994 Oncogene 9ZP61rpessmlil uo nigncfactors angiogenic tumor multiple represses R9-ZFP36L1 R9-ZFP36L1 TAT-ZFP36L1 R9-ZFP36L1 ZFP36L1 Alexa 594-WGA Alexa 594-WGA

R9-ZFP36L1 x-z Planel S Alexa 488-ZFP36L1 Alexa 488-ZFP36L1 tal et y-z Merge Merge R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5995 3.43 Â 108±7.9 Â 107 photons/s for controls versus IL-1a, leptin, acidic fibroblast growth factor, tumor 1.66 Â 108±8 Â 107 photons/s for R9-ZFP36L1-treated necrosis factor a, IL-6 and TGFa were decreased by tumors, n ¼ 4, Po0.05). Differences in luciferase activity 40–90% in R9-ZFP36L1-treated tumors as compared correlated with tumor weights (Figure 7c, mean weight with the untreated ones (Figure 9b). Consistent with the 1116±109 mg for controls versus 628±32 mg for immunohistochemical analysis (Figure 8C), VEGF R9-ZFP36L1-treated tumors, n ¼ 5, Po0.01). More- protein levels were decreased by 70% in R9-ZFP36L1- over, R9-ZFP36L1-treated tumors looked much paler treated tumors. The most abundant antiangiogenic than the reddish control ones. factors interferon-g and TIMP-1 were decreased by 58 and 28%, respectively. R9-ZFP36L1-induced tumor growth inhibition is associated with reduced microvessel density and decreased VEGF expression Discussion To determine whether R9-ZFP36L1 had an effect on LL/2 tumor vasculature, immunohistochemical analysis In this study, we hypothesized that ZFP36L1, was performed on resected control and treated tumors a regulator of mRNA turnover might represent a novel (Figure 8A). New blood vessel formation was dramati- antiangiogenic agent that may suppress not only VEGF cally reduced in R9-ZFP36L1-treated tumors (c and d) (Ciais et al., 2004; Cherradi et al., 2006), but also other as compared with controls (a and b). In addition, proangiogenic cytokines that support tumor growth. vascularization in R9-ZFP36L1-treated tumors was The development of multityrosine kinase inhibitors has often limited to the periphery of the tumor mass (c). emphasized the superior efficacy of simultaneously Vessels in control tumors displayed an important targeting several signaling pathways as compared with variability in diameter compared with the narrow vessels monospecific inhibitors (Petrelli and Giordano, 2008). observed in R9-ZFP36L1-treated tumors (b and d). We show that R9-mediated intracellular delivery of Quantitation of microvessel density in tumor sections ZFP36L1 inhibits tumor growth and is a potent revealed that R9-ZFP36L1 decreased microvessel den- suppressor of tumor vasculature in correlation with a sity by 76% (Figure 8B, mean 91±4.2/200 Â field for marked decrease in the levels of several tumor angio- control versus 22±1.34/200 Â field for R9-ZFP36L1- genic factors. Our results provide proof of the concept treated tumors, n ¼ 3, Po0.0001). As VEGF is a potent that ZFP36L1-mediated targeting of multiple ARE- inducer of microvessel growth in tumors, we also bearing mRNAs is an effective antiangiogenic and determined the effect of R9-ZFP36L1 on VEGF antitumoral strategy. expression (Figure 8C). VEGF protein levels (brown Transfection and transduction studies revealed that staining) were significantly decreased in R9-ZFP36L1 PTD-ZFP36L1-mediated decrease of VEGF mRNA treated tumors (g and h) as compared with controls levels was associated to a 50% similar decrease in (e and f). VEGF protein secretion, indicating a direct coupling between VEGF mRNA stability and translation. R9-ZFP36L1 targets multiple angiogenic and Although the half-life of ZFP36L1 has been reported inflammatory cytokines to be rather short (3 h) (Benjamin et al., 2006), PTD- As ZFP36L1 could target not only VEGF but also a ZFP36L1-induced decrease in VEGF protein levels was number of short-lived cytokines in vitro (Baou et al., detected up to 24 h after their addition to COS7 cells. 2009), we further characterized the antiangiogenic effect This observation is in agreement with previous studies of R9-ZFP36L1 by analyzing the relative abundance of showing that some TAT-fusion proteins are still active several angiogenic factors in control and R9-ZFP36L1- at least up to 24 h (Cai et al., 2006). It is worth treated tumors (Figure 9). Using mouse angiogenesis mentioning that we used low concentrations of purified antibody arrays, we observed that leptin, interferon-g, PTD-ZFP36L1 (100 nM), which is in contrast to acidic fibroblast growth factor, tumor necrosis factor a previous studies using micromolar concentrations of and interleukin (IL)-6 were the most abundant cytokines PTD fusion proteins to transduce different cell types detected in LL/2 tumors (Figure 9a). Overall expression (Richard et al., 2003; Nakase et al., 2004; Duchardt levels of proangiogenic factors were significantly lower et al., 2007). We demonstrated that R7- and in R9-ZFP36L1-treated tumors than in the untreated R9-ZFP36L1 induced a 40% decrease in VEGF protein ones. Densitometry analysis of the arrays revealed that levels compared with a 20% decrease obtained the expression levels of the proangiogenic factors EGF, with TAT-ZFP36L1, thus confirming that arginine-rich

Figure 4 PTD-ZFP36L1 and not ZFP36L1 penetrate efficiently in live COS7 cells. COS7 cells were incubated for 2 h at 37 1C in the presence of 100 nM of green Alexa 488-labeled- ZFP36L1 or -PTD-ZFP36L1, and subsequently incubated with red Alexa 594 wheat germ agglutinin for 10 min to label the plasma membrane. (a) Internalization of ZFP36L1 (non-penetrating negative control) or R9-ZFP36L1 was visualized by fluorescence microscopy. (b) Laser confocal microscopy analysis of TAT-ZFP36L1 and R9-ZFP36L1 intracellular localization. (c) Deconvolution fluorescence microscopy of R9-ZFP36L1 in COS7 cells confirming an intracellular localization of the fusion protein. The analysis was performed along the x–y axes (central), the x–z axes (bottom) and the y–z axes (right). The white lines indicate the axes along which the deconvolution was performed. Scale bars, 20 mm. Note the fluorescent internal signal clearly visible along the z axis (green).

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5996 noteworthy that the R7- and R9-mediated decrease in TAT- R7- 1 R9- VEGF expression is of a high significance as VEGF Vehicle ZFP36L1 ZFP36L1 ZFP36L ZFP36L1 expression levels in mammalian cells are tightly regu- lated. Indeed, a 50% reduction of VEGF expression VEGF during embryogenesis, as observed in VEGF þ /À mRNA heterozygous mice results in lethality (Carmeliet et al., 1996; Ferrara et al., 1996). We assessed the ability of R9-ZFP36L1 to inhibit VEGF expression in vivo by injecting R9-ZFP36L1 into mouse adrenal glands. In the adrenal cortex, the rapid release of corticosteroids into the blood flow is 18S facilitated by a dense vascular network whose main- tenance is regulated through strong VEGF expression at the adult stage (Thomas et al., 2004). A single injection of R9-ZFP36L1 in the adrenal gland induced a massive 120 Vehicle decrease in VEGF protein levels throughout the adrenal ZFP36L1 100 cortex after 24 h, indicating an efficient internalization TAT-ZFP36L1 of R9-ZFP36L1 and probably a bystander effect. R7-ZFP36L1 Whether this observation results from adrenocortical 80 R9-ZFP36L1 cell-to-cell transfer of R9-ZFP36L1 remains to be 60 investigated, as we could not determine the coverage of R9-ZFP36L1 in the injected adrenal glands, which do 40 express substantial amounts of endogenous ZFP36L1 (Cherradi et al., 2006). It should be however noted that 20

VEGF/18S (% of Control) protein transduction domains are known to facilitate the 0 passage through the plasma membrane in both direc- Purified protein tions (Fittipaldi and Giacca, 2005). Besides targeting VEGF mRNA, ZFP36L1 could also Vehicle mediate decay of other ARE-containing mRNAs, such ZFP36L1 120 as those encoding inflammatory cytokines, COX-2, the TAT-ZFP36L1 proto-oncogene c-fos, cyclins and the antiapoptotic 100 R7-ZFP36L1 R9-ZFP36L1 factor bcl-2, all of which have been reported to be abnormally stabilized in several human tumors 80 * * (Benjamin and Moroni, 2007). TTP, the most studied 60 member of ZFP36L1 family has been reported to act as a potent tumor suppressor when overexpressed in a v-H- 40 ras-expressing IL-3-dependent mast cells (Stoecklin et al., 2003). More recently, Essafi-Benkhadir et al 20 (2007) reported that TTP overexpression inhibits VEGF protein (% of Control) 0 RasVal12-dependent tumor vascularization by induc- Purified protein ing VEGF mRNA degradation. Whereas transfection Figure 5 Transduction of PTD-ZFP36L1 proteins into COS7 cells experiments yield highly significant information con- downregulates VEGF mRNA and protein levels. (a) Representa- cerning the emerging link between mRNA stability and tive northern blot of total RNA from COS7 cells incubated in the cancer, inherent limitations of transgenic technologies absence (vehicle) or in the presence of 100 nM of ZFP36L1, TAT- impede their potential use in human disease therapies. ZFP36L1, R7-ZFP36L1 or R9-ZFP36L1 for 24 h. In this experi- Therefore, additional methods of administrating intra- ment, the ratios of VEGF mRNA to 18S RNA were 0.031, 0.024, 0.020, 0.014 and 0.015 for vehicle, ZFP36L1, TAT-ZFP36L1, R7- cellularly acting proteins are needed to allow in vivo ZFP36L1 and R9-ZFP36L1, respectively. (b) Quantification of studies. We found that R9-ZFP36L1 exerts a negative VEGF mRNA signal intensities in two independent experiments. effect on tumor progression. Although control tumors VEGF mRNA values were normalized to 18S RNA values and displayed a very dense and tortuous vascular network plotted as a percentage of VEGF mRNA level in control cells (vehicle). Each point is the mean value ± s.d. of two separate with variability in vessel diameter, a hallmark of experiments performed in triplicate. (c) Culture media of COS7 excessive VEGF stimulation, R9-ZFP36L1-treated tu- cells from the experiments described in (a) were analyzed by ELISA mors displayed a dramatic loss in tumor vasculature, for VEGF protein content. VEGF protein values were normalized and a marked reduction in vessel diameter and VEGF to total protein values and represented a percentage of VEGF levels. These results are in line with those reported by protein level in control cells. The data shown represent the mean± s.e.m. of three independent experiments performed in duplicate others and showing that inhibition of VEGF signaling (*, significantly different from control (vehicle) with Po0.05). partially normalizes tumor vessels (Heath and Bicknell, 2009). peptides possess significantly enhanced protein trans- Profiling protein expression with antibody array duction potential compared with TAT in vitro and analysis of LL/2 tumor extracts revealed that R9- in vivo (Futaki et al., 2001; Ho et al., 2001). It is ZFP36L1 targets multiple growth factors (EGF, acidic

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5997 ZFP36L1

M1 M2

ZG

ZF

R9-ZFP36L1

M3 M4

ZF

ZG

ZFP36L1R9-ZFP36L1ZFP36L1R9-ZFP36L1 VEGF188 VEGF VEGF164 VEGF120

HPRT

24h 48h Figure 6 Injection of PTD-ZFP36L1 in mouse adrenal gland decreases VEGF mRNA and protein levels. (a) ZFP36L1 (negative control) or R9-ZFP36L1 fusion proteins (20 ng) were injected in mouse adrenal gland. Sections of adrenal glands were immunostained for VEGF (brown staining) and counterstained with haematoxylin and eosin (blue staining). Adrenal glands from two control mice (M1, M2) and two R9-ZFP36L1-treated mice (M3, M4) are shown. Scale bar, 100 mm. The brackets delineate the two major zones of the adrenal cortex (ZG, zona glomerulosa; ZF, zona fasciculata). Results from one experiment representative of three independent experiments are shown. (b) RT–PCR analysis of VEGF mRNA in control and R9-ZFP36L1-treated adrenal glands. Hypoxanthine- guanine phosphoribosyltransferase was used as internal standard. Results from one experiment representative of two independent experiments are shown.

fibroblast growth factor, VEGF and TGFa) and rc.kfshrc.edu.sa/ared), according to their respective inflammatory cytokines (tumor necrosis factor a, inter- ARE clusters. In addition, while VEGF is the only feron-g, IL-1a and IL-6), all of which are involved in the target known for ZFP36L1 in vivo (Bell et al., 2006), angiogenic process. Although we cannot exclude the these results were not surprising to us as ZFP36L1, TTP possibility that downregulation of these molecules by and ZFP36L2 appear interchangeable in ARE-directed R9-ZFP36L1 may be indirect, it is worth mentioning mRNA decay assays. Furthermore, our data are that most of the factors detected with the antibody array strengthened by recent studies showing that TTP are listed in the ARE-mRNA database (ARED; http:// mediates decay of IL-1a and IL-6 (Tudor et al., 2009)

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5998

) 1400

3 Control *** 1200 R9-ZFP36L1

1000 **

800

600

400

200

Average tumor volume (mm 0 6 8 10 12 14 16 18 20 Days post-implantation

* 4.5

4.0

3.5

Control 3.0 (Photons/sec) -8 2.5

2.0

1.5

1.0

5.0 Average total flux x 10

R9-ZFP36L1 0.0 Control R9-ZFP36L1

** 1250

1000

Control 750

500

250 R9-ZFP36L1

Average tumor weight (mg) 0 Control R9-ZFP36L1 Figure 7 R9-ZFP36L1 inhibits LL/2 tumor growth. Luciferase-expressing LL/2 cells were implanted subcutaneously into nude mice and R9-ZFP36L1 (40 ng), or vehicle were injected into the tumors. Tumor growth was subsequently monitored based on (a) external measurement using caliper (n ¼ 5 for each group, **Po0.01 and ***Po0.001, student’s t test), or (b) determination of emitted bioluminescence at day 20. In (b), the right panel represents quantification of the average total flux (photons/sec) in control and R9- ZFP36L1-treated mice (n ¼ 4 for each group, *Po0.05). (c). The difference in tumor volume, as well as in luciferase expression between control and R9-ZFP36L1-treated mice was confirmed by tumor weight. Control weight at the end of the experiment (day 20) was 1116±109 versus 628±32 mg for R9-ZFP36L1-treated tumors (n ¼ 5, **Po0.01).

and of interferon-g (Ogilvie et al., 2009) and IL-12 indicate that tumor necrosis factor a not only promotes (Jalonen et al., 2006). the production the angiogenic factors IL-8, VEGF and Our observation that R9-ZFP36L1 is a potent repressor bFGF (Dirkx et al., 2006), but also induces an endothelial of tumor inflammatory cytokines such as IL-1a, tumor phenotype of monocytes recruited to the tumor site (Li necrosis factor a and IL-6 suggests that R9-ZFP36L1 may et al., 2009). Finally, combined targeting of IL-6 and also potentially prevent the inflammatory responses that VEGF has been reported to inhibit glioma growth and are associated with the development of cancer. Indeed, IL- invasiveness (Saidi et al., 2009). In our study, both genes 1a has been reported to contribute to tumor angiogenesis were markedly downregulated by R9-ZFP36L1. There- and invasiveness in different experimental tumor models fore, R9-ZFP36L1 may also potentially control invasive (Voronov et al., 2003). On the other hand, recent data behavior of aggressive tumors.

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 5999 CD31

*** 100 Control

75

50 MVD / Field 25

0 R9-ZFP36L1 Control R9-ZFP36L1

VEGF

n Control R9-ZFP36L1

Figure 8 R9-ZFP36L1 inhibits tumor vascularization and VEGF expression. (A) Histological staining of the vasculature (CD31, brown immunostaining) in a representative area of the tumors resected after 20 days from control (a and b) and R9-ZFP36L1-treated mice (c and d). Panels b and d are higher magnifications of areas from a and c (n: tumor necrosis, outlined in black). (B) Quantification of microvessel density (MDV) in tumor sections from control and R9-ZFP36L1-treated mice (n=3 for each group, ***Po0.0001). (C) Intracellular VEGF immunostaining (brown) in tumor sections from control (e and f) and R9-ZFP36L1-treated mice (g and h). Panels f and h are higher magnifications of areas from e and g. The inset (i) in f shows that no staining was detected when omitting primary anti-VEGF antibody (negative control). Scale bars, 10 mm (a, c, e and g) and 5 mm (b, d, f and h).

Interestingly, although R9-ZFP36L1 treatment also time that ZFP36L1 is effective in repressing a broad affected some of the antiangiogenic factors expressed in spectrum of target mRNAs bearing structurally distinct LL/2 tumors, such as interferon-g, IL-12, IP-10 and AREs in tumor cells. Elevated expression of ZFP36L1 TIMP-1, the effect of R9-ZFP36L1 on angiogenic has also been reported to enhance cisplatin sensitivity in factors overcame its effect on antiangiogenic factors as head and neck squamous cell carcinoma by reducing the demonstrated by the inhibition of tumor growth and levels of the inhibitor of apoptosis cIAP2 mRNA (Lee tumor vascularization. It is worth mentioning that, et al., 2005). One could speculate that, in addition to although TIMP-1 is classically considered to inhibit angiogenic genes, R9-ZFP36L1 could also inhibit tumor progression, high expression of TIMP-1 has been inducers of resistance to chemotherapy. reported recently to induce MDA-MB-231 tumor growth in severe combined immunodeficiency mice (Bigelow et al., 2008), and to be associated with adverse Materials and methods prognosis in colorectal and breast human cancers (Kuvaja et al., 2005; Offenberg et al., 2008). Cloning of PTD-ZFP36L1 fusion proteins In conclusion, our study demonstrates the efficiency Plasmids containing either the Flag-TAT-, the Flag-R7- or the of a cell-penetrating ZFP36L1 in targeting tumor Flag-R9-ZFP36L1 sequences were constructed as described in angiogenesis and tumor growth. We show for the first details in Supplementary Materials and Methods.

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 6000 Control R9-ZFP36L1

Positive controls Positive controls

α γ EGF IL-1 Leptin IFN- EGF IL-1α Leptin IFN-γ

β FGFa IL-1 VEGF IL-12 FGFa IL-1β VEGF IL-12

FGFb IL-4 TNFα IP-10 FGFb IL-4 TNFα IP-10

G-CSF IL-6 TGFα TIMP-1 G-CSF IL-6 TGFα TIMP-1

Neg Ct Neg Ct TGFβ TIMP-2 Neg Ct Neg Ct TGFβ TIMP-2

120 Control R9-ZFP36L1 100

80

60

Relative intensity 40

20

0 α γ β α α β EGF IFN IL-4 IL-6 IL-1 Leptin FGFa IL-1 VEGF IL-12 FGFb TNF IP-10 TGF TGF G-CSF TIMP-1 TIMP-2 Pos Ctrl Figure 9 R9-ZFP36L1 downregulates the expression of several tumor angiogenic cytokines. (a) Mouse angiogenesis antibody array analysis of control and R9-ZFP36L1-treated tumor extracts. EGF, ; IL-1a, interleukin-1a; IFNg, interferon g; FGFa, acidic fibroblast growth factor; IL-1b, interleukin-1b; VEGF, vascular growth factor; IL-12, interleukin-12, FGFb, basic fibroblast growth factor; IL-4, interleukin-4; TNFa, tumor necrosis factor a; IP10, interferon inducible protein-10; G-CSF, granulocyte-colony stimulating factor; IL-6, interleukin-6; TGFa, transforming growth factor a; TIMP-1, tissue inhibitor of metalloproteinases-1; Neg Control, negative control; TGFb, transforming growth factor b; TIMP-2, tissue inhibitor of metalloproteinases-2. (b) Quantification of angiogenic proteins in control and R9-ZFP36L1-treated tumors. The arrays presented in (a) were scanned and the relative intensities were analyzed using the ImageJ software. Bars represent the mean ± s.d. of duplicate spots on the array. Experiments were performed twice on two control and R9-ZFP36L1-treated tumors.

Transfections and luciferase activity assay ZFP36L1 proteins (24 h) were treated with 5,6-dichloro-1-b-d- Transfection of COS7 cells with ZFP36L1, TAT-ZFP36L1, ribofuranosyl-benzimidazole (DRB, 10 mg/ml) for increasing R7-ZFP36L1 or R9-ZFP36L1 pTarget plasmids and pLuc-30- periods of time. Total RNA was extracted and subjected to UTR (firefly luciferase cloned upstream of VEGF 30-UTR) northern blot to determine the VEGF mRNA half-life as plasmid, as well as luciferase activity measurement were described in Supplementary Materials and Methods. performed as described in Supplementary Materials and Methods. Enzyme-linked immunosorbent assay (ELISA) SDS–PAGE and western blot analysis VEGF content of culture medium from COS7 cells was Details are provided in Supplementary Materials and Methods. determined as reported in Supplementary Materials and Methods. Northern hybridization and determination of VEGF mRNA half-life Production of recombinant PTD-ZFP36L1 fusion proteins COS7 cells transfected with ZFP36L1, TAT-ZFP36L1, Conditions for PTD-ZFP36L1 protein expression and R7-ZFP36L1 or R9-ZFP36L1 plasmids (48 h), or incubated purification are described in Supplementary Materials and with purified ZFP36L1, TAT-ZFP36L1, R7-ZFP36L1 or R9- Methods.

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 6001 Transduction of PTD-ZFP36L1 proteins in living cells and Immunohistochemistry confocal laser microscopy In total, 5 mm sections of paraffin-embedded tumors from each Uptake and intracellular localization of Alexa 488-labeled group were deparaffinized in xylene and then rehydrated. PTD-ZFP36L1 proteins were assessed as described in Supple- VEGF and CD31 staining of tumor sections were conducted as mentary Materials and Methods. described in Supplementary Materials and Methods. Micro- vascular density of CD31-stained tumors was counted in seven Transduction of PTD-ZFP36L1 proteins in vivo random high-power fields (  200) per tumor section from Animal experiments were approved by the institutional guide- three R9-ZFP36L1-treated and three control animals and lines and followed the recommendations of the European expressed as a number of microvessels per  200 field. Community for the Use of Experimental Animals. Female severe combined immunodeficiency mice (6 months old) were Angiogenesis antibody array purchased from Taconic (Germantown, NY, USA) and Tissue fragments from control or R9-ZFP36L1-treated tumors maintained in the Animal Resources Center of our department. were lysed in lysis buffer (Promega, Charbonnie` res, France) In vivo delivery of R9-ZFP36L1 to mice adrenal gland was done containing a protease inhibitor cocktail (Sigma, St Louis, MO, by injecting 5 ml of 100 nM solution (20 ng) in phosphate-buffered USA). In all, 1 mg of total tumor lysate was hybridized to each saline. Control mice were injected with 5 ml of the negative membrane of an antibody-sandwich mouse angiogenesis array control ZFP36L1 without PTD (20 ng). Mice were sacrificed (Panomics, Fremont, CA, USA) according to the manufac- after 24–48 h (n ¼ 3 in each group). Adrenal glands were turer’s guidelines. In two separate experiments, the antibody collected carefully and either fixed overnight in 4% paraformal- arrays were hybridized and imaged together. The arrays were dehyde and embedded in paraffin, or processed for total RNA scanned and analyzed with the ImageJ software (NIH). isolation. Tissue sections (5 mm) were immunostained for VEGF. Background was subtracted by calculating the pixel intensity for a pixel ring outside the spots and subtracting this baseline Reverse transcription–PCR value from pixel intensity values within the spots. Interarray Details on RT–PCR analysis of VEGF or hypoxanthine- normalization was performed by using positive control spots guanine phosphoribosyltransferase (HPRT) gene expression (eight per array) on each array. are provided in Supplementary Materials and Methods. Statistical analysis Tumor growth experiments Statistical analysis was carried out using GraphPad Prism In total, 1  106 luciferase-transduced LL/2 Lewis Lung software (version 4, San Diego, CA, USA). Data were Carcinoma cells harvested in 50 ml of medium were mixed analyzed using one-way ANOVA or student’s t-test, when with 50 ml of Matrigel (Becton Dickinson, Biosciences, Le Pont appropriate. Results are expressed as means ± s.e.m. A value de Claix, France), then subcutaneously injected into the hind of Po0.05 was considered as statistically significant. flank of 10 female nu/nu mice (8 weeks old, 20–25 g, Charles River, France). When tumor size reached B100 mm3, mice were divided into two groups. Tumors of the R9-ZFP36L1- Conflict of interest treated group were injected every other day with 10 ml of R9- ZFP36L1 diluted in phosphate-buffered saline at a concentra- The authors declare no conflict of interest. tion of 100 nM (B40 ng). Tumors of the control group were injected with 10 ml of phosphate-buffered saline containing the same concentration of Tris–glycine as the R9-ZFP36L1 sample. Tumor size was monitored every other day by caliper Acknowledgements measurement of tumor length and width . Tumor volume was calculated according to the formula V ¼ 0.5ab2, where a and b This work was supported by the Institut National de la Sante´ were the largest and the smallest diameters, respectively. et de la Recherche Me´dicale (INSERM, U878), the Commis- Studies were ended when tumors in the control group reached sariat a` l’Energie Atomique (iRTSV/LAPV), the Institut du an average of 1000 mm3 (day 20–21). To perform tumor Cancer (INCa, Program ‘Emergence des Cance´ropoles 2004’), imaging, mice were injected with D-luciferin intraperitoneally the Cance´ropole Rhoˆne-Alpes (CLARA) and the Groupement (150 mg/kg body weight) before being anesthetized with des Entreprises Franc¸aises pour la Lutte contre le Cancer isofluorane. Luciferase expression of the tumors was measured (GEFLUC)-Comite´Dauphine´-Savoie. We also thank the using bioluminescence technology (IVIS Lumina II, Xenogen, Association pour la Recherche sur le Cancer and the Caliper Life Sciences, Villepinte, France) at 15 min after Fondation pour la Recherche Me´dicale for their financial luciferin administration. Measurement of total flux (photons/ support to SP. We thank Dr Didier Grunwald (LTS, iRTSV, sec) of the emitted light reflects the relative number of viable CEA-Grenoble) for helping us with the laser confocal cells in the tumor. Data were analyzed using Xenogen Living microscopy platform of our institute. We also thank Dr Image software (version 3.0). Mice were killed further by Maryline Herbet and Dr Michael Thomas (INSERM U878, cervical dislocation and tumors were excised and weighed before iRTSV, CEA-Grenoble) for their assistance in the ‘in vivo’ their fixation either in 4% paraformaldehyde or in formalin-free experiments and Dr Jean-Luc Coll (INSERM U883, UJF, fixative (Accustain, Sigma) to perform immunohistochemistry Grenoble, France) for his generous gift of the LL/2-luciferase of VEGF or CD31, respectively. cell line.

References

Asoh S, Ohta S. (2008). PTD-mediated delivery of anti-cell death Baou M, Jewell A, Murphy JJ. (2009). TIS11 family proteins and their proteins/peptides and therapeutic enzymes. Adv Drug Deliv Rev 60: roles in posttranscriptional gene regulation. J Biomed Biotechnol 499–516. 2009: 634520.

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 6002 Bell SE, Sanchez MJ, Spasic-Boskovic O, Santalucia T, Gambardella permeable peptides having potential as carriers for intracellular L, Burton GJ et al. (2006). The RNA binding protein Zfp36l1 is protein delivery. J Biol Chem 276: 5836–5840. required for normal vascularisation and post-transcriptionally Gariano RF, Hu D, Helms J. (2006). Expression of angiogenesis- regulates VEGF expression. Dev Dyn 235: 3144–3155. related genes during retinal development. Gene Expr Patterns 6: Benjamin D, Moroni C. (2007). mRNA stability and cancer: an 187–192. emerging link? Expert Opin Biol Ther 7: 1515–1529. Heath VL, Bicknell R. (2009). Anticancer strategies involving the Benjamin D, Schmidlin M, Min L, Gross B, Moroni C. (2006). BRF1 vasculature. Nat Rev Clin Oncol 6: 395–404. protein turnover and mRNA decay activity are regulated by Ho A, Schwarze SR, Mermelstein SJ, Waksman G, Dowdy SF. (2001). protein kinase B at the same phosphorylation sites. Mol Cell Biol Synthetic protein transduction domains: enhanced transduction 26: 9497–9507. potential in vitro and in vivo. Cancer Res 61: 474–477. Bergers G, Hanahan D. (2008). Modes of resistance to anti-angiogenic Jalonen U, Nieminen R, Vuolteenaho K, Kankaanranta H, Moilanen therapy. Nat Rev 8: 592–603. E. (2006). Down-regulation of tristetraprolin expression results in Bigelow RL, Williams BJ, Carroll JL, Daves LK, Cardelli JA. (2008). enhanced IL-12 and MIP-2 production and reduced MIP-3alpha TIMP-1 overexpression promotes tumorigenesis of MDA-MB-231 synthesis in activated macrophages. Mediators Inflamm 2006: 40691. breast cancer cells and alters expression of a subset of cancer Kuvaja P, Talvensaari-Mattila A, Paakko P, Turpeenniemi-Hujanen promoting genes in vivo distinct from those observed in vitro. Breast T. (2005). The absence of immunoreactivity for tissue inhibitor of Cancer Res Treat 117: 31–44. metalloproteinase-1 (TIMP-1), but not for TIMP-2, protein is Cai SR, Xu G, Becker-Hapak M, Ma M, Dowdy SF, McLeod HL. associated with a favorable prognosis in aggressive breast carcino- (2006). The kinetics and tissue distribution of protein transduction ma. Oncology 68: 196–203. in mice. Eur J Pharm Sci 27: 311–319. Lee SK, Kim SB, Kim JS, Moon CH, Han MS, Lee BJ et al. (2005). Cao H. (2004). Expression, purification, and biochemical characteriza- Butyrate response factor 1 enhances cisplatin sensitivity in human tion of the antiinflammatory tristetraprolin: a zinc-dependent head and neck squamous cell carcinoma cell lines. Int J Cancer 117: mRNA binding protein affected by posttranslational modifications. 32–40. Biochemistry 43: 13724–13738. Levy AP. (1998). Hypoxic regulation of VEGF mRNA stability by Cao H, Lin R. (2009). Quantitative evaluation of His-tag purification RNA-binding proteins. Trends Cardiovasc Med 8: 246–250. and immunoprecipitation of tristetraprolin and its mutant proteins Li B, Vincent A, Cates J, Brantley-Sieders DM, Polk DB, Young PP. from transfected human cells. Biotechnol Prog 25: 461–467. (2009). Low levels of tumor necrosis factor alpha increase tumor Cao H, Lin R, Ghosh S, Anderson RA, Urban Jr JF. (2008). growth by inducing an endothelial phenotype of monocytes Production and characterization of ZFP36L1 antiserum against recruited to the tumor site. Cancer Res 69: 338–348. recombinant protein from Escherichia coli. Biotechnol Prog 24: Lindsay MA. (2002). Peptide-mediated cell delivery: application in 326–333. protein target validation. Curr Opin Pharmacol 2: 587–594. Carballo E, Lai WS, Blackshear PJ. (1998). Feedback inhibition of Lykke-Andersen J, Wagner E. (2005). Recruitment and activation of macrophage tumor necrosis factor-alpha production by tristetra- mRNA decay enzymes by two ARE-mediated decay activation prolin. Science 281: 1001–1005. domains in the proteins TTP and BRF-1. Genes Dev 19: 351–361. Carmeliet P, Ferreira V, Breier G, Pollefeyt S, Kieckens L, Nakase I, Niwa M, Takeuchi T, Sonomura K, Kawabata N, Koike Y Gertsenstein M et al. (1996). Abnormal blood vessel development et al. (2004). Cellular uptake of arginine-rich peptides: roles and lethality in embryos lacking a single VEGF allele. Nature 380: for macropinocytosis and actin rearrangement. Mol Ther 10: 435–439. 1011–1022. Chauhan A, Tikoo A, Kapur AK, Singh M. (2007). The taming of the Offenberg H, Brunner N, Mansilla F, Orntoft Torben F, Birkenkamp- cell penetrating domain of the HIV Tat: myths and realities. Demtroder K. (2008). TIMP-1 expression in human colorectal J Control Release 117: 148–162. cancer is associated with TGF-B1, LOXL2, INHBA1, TNF-AIP6 Cherradi N, Lejczak C, Desroches-Castan A, Feige JJ. (2006). and TIMP-2 transcript profiles. Mol Oncol 2: 233–240. Antagonistic functions of tetradecanoyl phorbol acetate-inducible- Ogilvie RL, Sternjohn JR, Rattenbacher B, Vlasova IA, Williams DA, sequence 11b and HuR in the hormonal regulation of vascular Hau HH et al. (2009). Tristetraprolin mediates interferon-gamma endothelial growth factor messenger ribonucleic acid stability by mRNA decay. J Biol Chem 284: 11216–11223. adrenocorticotropin. Mol Endocrinol 20: 916–930. Onesto C, Berra E, Grepin R, Pages G. (2004). Poly(A)-binding Ciais D, Cherradi N, Bailly S, Grenier E, Berra E, Pouyssegur J et al. protein-interacting protein 2, a strong regulator of vascular (2004). Destabilization of vascular endothelial growth factor endothelial growth factor mRNA. J Biol Chem 279: 34217–34226. mRNA by the zinc-finger protein TIS11b. Oncogene 23: 8673–8680. Petrelli A, Giordano S. (2008). From single- to multi-target drugs in Dirkx AE, Oude Egbrink MG, Wagstaff J, Griffioen AW. (2006). cancer therapy: when aspecificity becomes an advantage. Curr Med Monocyte/macrophage infiltration in tumors: modulators of Chem 15: 422–432. angiogenesis. J Leukoc Biol 80: 1183–1196. Richard JP, Melikov K, Vives E, Ramos C, Verbeure B, Gait MJ et al. Duchardt F, Fotin-Mleczek M, Schwarz H, Fischer R, Brock R. (2003). Cell-penetrating peptides. A reevaluation of the mechanism (2007). A comprehensive model for the cellular uptake of cationic of cellular uptake. J Biol Chem 278: 585–590. cell-penetrating peptides. Traffic 8: 848–866. Saidi A, Hagedorn M, Allain N, Verpelli C, Sala C, Bello L et al. Ellis LM, Hicklin DJ. (2008). VEGF-targeted therapy: mechanisms of (2009). Combined targeting of interleukin-6 and vascular endothe- anti-tumour activity. Nat Rev 8: 579–591. lial growth factor potently inhibits glioma growth and invasiveness. Essafi-Benkhadir K, Onesto C, Stebe E, Moroni C, Pages G. (2007). Int J Cancer 125: 1054–1064. Tristetraprolin inhibits Ras-dependent tumor vascularization by Schwarze SR, Ho A, Vocero-Akbani A, Dowdy SF. (1999). In vivo inducing vascular endothelial growth factor mRNA degradation. protein transduction: delivery of a biologically active protein into Mol Biol Cell 18: 4648–4658. the mouse. Science 285: 1569–1572. Ferrara N. (2004). Vascular endothelial growth factor: basic science Snyder EL, Dowdy SF. (2004). Cell penetrating peptides in drug and clinical progress. Endocr Rev 25: 581–611. delivery. Pharm Res 21: 389–393. Ferrara N, Carver-Moore K, Chen H, Dowd M, Lu L, O’Shea KS Stoecklin G, Gross B, Ming XF, Moroni C. (2003). A novel et al. (1996). Heterozygous embryonic lethality induced by targeted mechanism of tumor suppression by destabilizing AU-rich growth inactivation of the VEGF gene. Nature 380: 439–442. factor mRNA. Oncogene 22: 3554–3561. Fittipaldi A, Giacca M. (2005). Transcellular protein transduction Stumpo DJ, Broxmeyer HE, Ward T, Cooper S, Hangoc G, Chung YJ using the Tat protein of HIV-1. Adv Drug Deliv Rev 57: 597–608. et al. (2009). Targeted disruption of Zfp36l2, encoding a CCCH Futaki S, Suzuki T, Ohashi W, Yagami T, Tanaka S, Ueda K et al. tandem zinc finger RNA-binding protein, results in defective (2001). Arginine-rich peptides. An abundant source of membrane- hematopoiesis. Blood 114: 2401–2410.

Oncogene R9-ZFP36L1 represses multiple tumor angiogenic factors S Planel et al 6003 Stumpo DJ, Byrd NA, Phillips RS, Ghosh S, Maronpot RR, directed decay of interleukin-10 and pro-inflammatory Castranio T et al. (2004). Chorioallantoic fusion defects and mediator mRNAs in murine macrophages. FEBS Lett 583: embryonic lethality resulting from disruption of Zfp36L1, a gene 1933–1938. encoding a CCCH tandem zinc finger protein of the Tristetraprolin Voronov E, Shouval DS, Krelin Y, Cagnano E, Benharroch D, family. Mol Cell Biol 24: 6445–6455. Iwakura Y et al. (2003). IL-1 is required for tumor invasiveness and Thomas M, Keramidas M, Monchaux E, Feige JJ. (2004). Dual hormonal angiogenesis. Proc Natl Acad Sci USA 100: 2645–2650. regulation of endocrine tissue mass and vasculature by adrenocortico- Wender PA, Mitchell DJ, Pattabiraman K, Pelkey ET, Steinman L, tropin in the adrenal cortex. Endocrinology 145: 4320–4329. Rothbard JB. (2000). The design, synthesis, and evaluation of Tudor C, Marchese FP, Hitti E, Aubareda A, Rawlinson L, Gaestel M molecules that enable or enhance cellular uptake: peptoid molecular et al. (2009). The p38 MAPK pathway inhibits tristetraprolin- transporters. Proc Natl Acad Sci USA 97: 13003–13008.

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

Oncogene