A Small Molecule Disruptor of Rb/Raf-1 Interaction Inhibits Cell Proliferation, Angiogenesis, and Growth of Human Tumor Xenografts in Nude Mice

The retinoblastoma tumor suppressor protein, Rb, is a vital regulator of the mammalian cell cycle, and its inactivation facilitates S-phase entry (1, 2). Rb is inactivated through multiple waves of phosphorylation during cell cycle progression, mediated by kinases associated with D and E type cyclins in the G₁ phase (3, 4). Rb is inactivated in most cancers, either by mutation or deletion of the gene, interaction with viral oncoproteins, or alterations in the levels and activity of upstream regulators of Rb function (5–8). Rb controls the G₁-S boundary by repressing the transcriptional activity of the E2F family of transcription factors, especially E2F1, E2F2, and E2F3 (9). Many genes necessary for DNA synthesis and cell cycle progression, such as cyclins A and E, cdc2, thymidylate synthase (TS), dihydrofolate reductase, ORC1, and DNA polymerase α require E2F for their expression (10–13). Whereas cyclins and cyclin-dependent kinases phosphorylate Rb in mid to late G₁ phase releasing transcriptionally active E2F (14–16), Raf-1 kinase binds and phosphorylates Rb early in the G₁ phase (17). Disruption of this Rb/Raf-1 interaction by an eight–amino acid peptide (corresponding to Raf-1 residues 10–18) prevented Rb phosphorylation even late in the G₁ phase, suggesting that the binding of Raf-1 is necessary for the eventual inactivation of Rb (18). Furthermore, the level of Rb/Raf-1 interaction was elevated in non–small cell lung cancer tissue compared with adjacent normal tissue (19), suggesting that this interaction might contribute to oncogenesis. These observations suggested that disruption of the Rb/Raf-1 interaction might have anticancer effects and raised the possibility that small molecules that can disrupt the Rb/Raf-1 interaction might be useful as anticancer drugs. Here, we report a potent and selective small molecule disruptor of Rb/Raf-1 interaction that significantly inhibits angiogenesis and tumor growth in vivo in an Rb-dependent manner.

Cell culture and transfection. The human myelomonocytic leukemia cell line U937 was cultured in RPMI (Mediatech) containing 10% fetal bovine serum (FBS; Mediatech). U2-OS, Saos-2, PANC1, CAPAN2, A375, SK-MEL-5, SK-MEL-28, and MDA-MB-231 cell lines were cultured in DMEM (Mediatech) containing 10% FBS. A549 cells and A549 short hairpin RNA (shRNA) Rb cell lines were maintained in Ham F-12K supplemented with 10% FBS; media for shRNA cells lines contained 0.5 μg/mL puromycin. shRNA cell lines were generated by stably transfecting A549 cells with two different shRNA constructs that specifically target Rb obtained from a shRNAmir library from Open Biosystems. H1650, PC-9, LNCap, and Aspc1 cell lines were cultured in RPMI (Life Technologies) containing 10% FBS. Human aortic endothelial cells (HAEC; Clonetics) were cultured in endothelial growth medium, supplemented with 5% FBS. U251MG and U87MG glioma cells were maintained in DMEM supplemented with nonessential amino acids, 50 mmol/L β-mercaptoethanol, and 10% FBS. The adenovirus constructs Ad-green fluorescent protein (GFP) and Ad-E2F1 were obtained from W.D. Cress. Ad-cyclin D was kindly provided by I. Cozar-Castellano.

In vitro library screening assays. ELISA 96-well plates (Nunc) were coated with 1 μg/mL of glutathione S-transferase (GST) Raf-1 (1–149 amino acids) overnight at 4°C. Subsequently, the plates were blocked and GST-Rb at 20 μg/mL was rotated at room temperature for 30 min in the presence or absence of the compounds at 20 μmol/L. GST-Rb with or without compounds was then added to the plate and incubated for 90 min at 37°C. The amount of Rb bound to Raf-1 was detected by Rb polyclonal antibody (Santa Cruz) 1:1,000 incubated for 60 min at 37°C. Donkey anti-rabbit-IgG-HRP (1:10,000) was added to the plate and incubated at 37°C for 60 min. The color was developed with orthophenylenediamine peroxidase substrate tablets (Sigma); the reaction was terminated with 3 mol/L H₂SO₄ and absorbance read at 490 nm. To determine disruption of Rb to E2F1, Phb, or HDAC1, the above protocol was used with the exception of coating GST-Rb on the ELISA plate and adding the drugs in the presence or absence of GST E2F1, Phb, or HDAC1. These interactions were detected with E2F1 monoclonal antibody (Santa Cruz; 1:2,000 dilution), prohibitin monoclonal antibody (NeoMarkers; 1:1,000 dilution), and HDAC1 polyclonal antibody (Santa Cruz; 1:1,000). For disruption of Mek-Raf-1 binding ELISAs, Raf-1 (1 μg/mL) was coated on the plate and GST-Mek (20 μg/mL) was incubated in the presence or absence of compounds for 30 min at room temperature. Mek1 polyclonal antibody (Cell Signaling) was used at 1:1,000 to detect the binding of Raf-1 to Mek1. The IC₅₀ concentrations for the Rb/Raf-1 inhibitors were determined by plotting with Origin 7.5 software.

Lysate preparation, immunoprecipitation, and Western blotting. Cell lysates were prepared by NP40 lysis, as described earlier (18). Tumor lysates were prepared with T-Per tissue lysis buffer (Pierce) and a Fisher PowerGen 125 dounce homogenizer. Physical interaction between proteins in vivo was analyzed by immunoprecipitation–Western blot analyses with 200 μg of lysate and 1 μg of the indicated antibody (18). Polyclonal E2F1, B-Raf, ASK1, and cyclins D and E were obtained from Santa Cruz Biotechnology. Monoclonal Rb and Raf-1 were obtained from BD Transduction Laboratories, and polyclonal antibodies to phosphorylated Rb (807, 811) and Mek1/2 were from Cell Signaling.

Chloramphenicol acetyltransferase assays. A549 cells were transfected by calcium phosphate and treated with drug for 24 h. Assays for chloramphenicol acetyltransferase (CAT) and β-galactosidase were performed using standard protocols (17).

Chromatin immunoprecipitation assay. A549 cells were serum-starved and restimulated with serum for 2 or 16 h in the presence or absence of Rb/Raf-1 disruptor 251 (RRD-251) at 20 μmol/L, and chromatin immunoprecipitation (ChIP) lysates were prepared (18). Immunoprecipitations were conducted using antibodies against E2F1, Rb, Raf-1, Brg1, HP1, and HDAC1, and the association with specific promoters was detected by PCR. Rabbit anti-mouse secondary antibody was used as the control for all reactions. The sequences of the PCR primers used in the PCRs were as follows: Cdc6 promoter forward primer, 5′-GGCCTCACAGCGACTCTAAGA-3′ and Cdc6 promoter reverse primer, 5′-CTCGGACTCACCACAAGC-3′; TS promoter forward primer, 5′-TGG CGC ACG CTC TCT AGA GC-3′ and TS promoter reverse primer, 5′-GAC GGA GGC AGG CCA AGT G-3′. The cdc25A and c-fos primers are described in ref. 18.

Real-time PCR. A549 cells were rendered quiescent by serum starvation and subsequently stimulated with serum in the presence or absence of RRD-251. Unstimulated serum-starved cells were used as control. Total RNA was isolated by an RNeasy miniprep kit from QIAGEN following the manufacturer's protocol. One microgram of RNA was DNase treated using RQ1 DNase (Promega), followed by first-strand cDNA synthesis using the iScript cDNA synthesis kit (Bio-Rad). A fraction (1/20) of the final cDNA reaction volume was used in each PCR (20). Primer sequences are as follows: 5′-CTG CCA GCT GTA CCA GAG AT-3′ (TS forward primer), 5′-ATG TGC ATC TCC CAA AGT GT-3′ (TS reverse primer), 5′-CCC CAT GAT TGT GTT GGT AT-3′ (Cdc6 forward primer), 5′-TTC AAC AGC TGT GGC TTA CA-3′ (Cdc6 reverse primer), 5′-CTC AAC ACG GGA AAC CTC AC-3′ (18S forward primer), and 5′-AAA TCG CTC CAC CAA CTA AGA A-3′ (18S reverse primer). Real-time PCR was performed on a Bio-Rad iCycler.

In vitro kinase assay. The kinase reaction for Raf-1 was carried out with 100 ng of Raf-1 (Upstate Biotechnologies), 0.5 μg of MEK1 (Upstate) as the substrate, 10 μmol/L ATP, 10 μCi of [γ-³²P]ATP in the kinase assay buffer in the presence or absence of the drugs at 30°C for 30 min. BAY-43-9006 (1 μmol/L) was used as a control, and RRD-251 (20 μmol/L) was used. Cyclins D and E kinase assays are described in ref. 18.

Proliferation assays. Bromodeoxyuridine (BrdUrd) labeling kits were obtained from Roche Biochemicals. Cells were plated in poly-d-lysine–coated chamber slides at a density of 10,000 cells per well and serum starved for 24 h. Cells were then stimulated with serum in the presence or absence of the indicated drugs for 18 h. S-phase cells were visualized by microscopy and quantitated by counting three fields of 100 in quadruplicate. For adenovirus experiments, A549 cells were serum starved for 48 h and subsequently infected with adenovirus in the presence or absence of RRD-251 for 36 h. Ad-GFP [6 × 10⁶ plaque-forming units (pfu)/μL], Ad-E2F1 (6 × 10⁶ pfu/μL), and Ad-cyclin D (7 × 10⁸ pfu/μL) were added at a multiplicity of infection of 150 particles per cell.

Soft agar colony formation assay. Assays were done in triplicate in 12-well plates (Corning). After allowing the bottom layer of agar (0.6%) to solidify at room temperature, the top layer of agar (0.3%) was mixed with 5,000 cells per well, and the indicated drug was added. Drugs were added twice weekly in complete media to the agar wells. Colonies were quantified by staining with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (1 mg/mL) for 1 h at 37°C.

Matrigel assays. Matrigel (Collaborative Biomedical Products) was used to promote the differentiation of human umbilical vascular endothelial cells (HUVEC) into capillary tube–like structures (18). A total of 100 μL of thawed Matrigel was added to 96-well tissue culture plates, followed by incubation at 37°C for 60 min to allow polymerization. Subsequently, 1 × 10⁴ HUVECs were seeded on the gels in EGM medium supplemented with 5% FBS in the presence or absence of 20 μmol/L concentrations of the indicated compounds, followed by incubation for 24 h at 37°C. Capillary tube formation was assessed using a Leica DMIL phase contrast microscope.

Ex vivo rat aorta ring angiogenesis assays. Forty-eight well tissue culture plates were coated with 200 μL of Matrigel and allowed to polymerize for 1 h at 37°C. Thoracic aorta was excised from 8-wk-old to 10-wk-old male Sprague-Dawley rats (250–300 g; ref. 21). After removing fibroadipose tissue, aortas were rinsed several times with EGM-2 (Clonetics), sectioned into 1-mm rings, and placed on the Matrigel-coated wells. The rings were covered with an additional 200 μL of Matrigel and allowed to polymerize. The rings were cultured in EGM-2 media in the presence or absence of 20 μmol/L of RRD-251. The media and drug were supplemented twice a week for 1 wk. The aortic rings were photographed on day 7 using a Leica phase contrast microscope. Quantitation of microvessel growth was done using Image Pro Plus (v.6.0) software, and values are reported as microvessel area.

In vivo Matrigel plug angiogenesis assays. In vivo Matrigel plug assays were carried out as previously described (22). Cooled liquid Matrigel (300 μL) was injected s.c. into both flanks of nude mice. One group of mice received the vehicle, and the second group received RRD-251 50 MPK daily by i.p. injection. At 7 d post-Matrigel injection, the mice were injected with 100 μL of 100 MPK FITC-Dextran (Sigma) through the tail vein. At 30 min later, the mice were euthanized and the Matrigel plugs were removed and fixed in buffered formalin. Samples were viewed with a Leica DMI6000 inverted microscope, TCS SP5 confocal scanner, and a 20×/0.7NA Plan Apochromat objective (Leica Microsystems). An Argon 488 laser line was applied to excite the samples, and tunable filters were used to minimize background fluorescence. Image sections at 2.0 μm were captured with photomultiplier detectors three-dimensional projections prepared with the LAS AF software version 1.6.0 build 1016 (Leica Microsystems). Quantification of intensity and angiogenesis was performed using Image Pro Plus 6.2 (Media Cybernetics, Inc.). Average intensity per pixel is plotted as percentage angiogenesis in each image (n = 12). Each image is representative of areas of vessel formation throughout entire Matrigel plug. After confocal imaging, samples were paraffin blocked and stained with H&E. H&E images display 1/4 of the Matrigel plug.

Animal studies. A549 or H1650 cells were harvested and resuspended in PBS and implanted s.c. into the right and left flanks (10 × 10⁶ cells per flank) of 8-wk-old female athymic nude mice (Charles River) as described (18, 23). When tumors reached ∼100 to 200 mm³, animals were dosed i.p. or p.o. by gavage with 0.1 mL of the drug or the vehicle daily. Tumor volumes were determined by measuring the length (l) and the width (w) and calculating the volume (V = lw² / 2). Statistical significance between control and treated animals were evaluated using Student's t test.

Immunohistochemistry staining. Tumors were fixed in 10% neutral-buffered formalin before processing into paraffin blocks. Paraffin sections (5-μm thick) were rehydrated into PBS and processed using the following protocols. Sections were rinsed in dH₂O and then subjected to microwave “antigen retrieval” in 0.01 mol/L sodium citrate (pH 6.0). Sections were cooled, rinsed thrice in dH₂O and twice in PBS, and incubated in 5% normal goat serum. Sections were incubated in primary antibody in 5% normal goat serum and rinsed thrice in PBS. For color development, slides were treated with ABC kit from Vector Laboratories, rinsed in dH₂O, and developed with 3,3′-diaminobenzidine. Then, sections were lightly counterstained in hematoxylin, dehydrated, cleared, and cover-slipped. Tissue sections were stained with H&E using standard histologic techniques. Tissue sections were immunostained using Ki-67, CD31 phosphorylated Rb, and CD31 antibodies (BD Biosciences) using the avidin-biotin peroxidase complex technique. Mouse monoclonal antibody was used at 1:50 dilution after microwave antigen retrieval (four cycles of 5 min each on high in 0.1 mol/L citrate buffer). Stained slides were scanned on an Ariol SL-50 Automatic Scanning System, and whole tumor sections were quantitated using Image Pro Plus (v.5.1.0) software.

Statistical analysis. Statistical analysis was performed using one-tailed Student's t test. Values were considered significant when the P value was <0.01.

Identification of Rb/Raf-1 disruptor RRD-251. An ELISA was used to identify compounds that could inhibit the binding of GST-Rb to GST-Raf-1. Screening of the National Cancer Institute (NCI) diversity library of 1,981 compounds identified two compounds, NSC-35400 and NSC-35950, which inhibited Rb/Raf-1 interaction 100% and 95%, respectively, at 20 μmol/L concentration. NSC-35400 and NSC-35950 each contained a benzyl-isothiourea derivative and a phenyl-based counter ion (Fig. 1A); to establish whether the benzyl-isothiourea derivative is the active component, we synthesized RRD-251 (Fig. 1A), which contains chloride as the counter ion. ELISA analysis showed that NSC-35400 disrupted the Rb/Raf-1 interaction with an IC₅₀ of 81 ± 4 nmol/L, NSC-35950 with an IC₅₀ of 283 ± 46 nmol/L, and RRD-251 with an IC₅₀ of 77 ± 3.6 nmol/L (Fig. 1B), suggesting that benzylisothiouronium pharmacophore disrupts the Rb/Raf-1 interaction. ELISAs showed that these disruptors were highly selective for Rb/Raf-1 interaction over Rb/E2F1, Rb/HDAC1, Rb/prohibitin (Fig. 1C), and Raf-1/Mek (Fig. 1D) associations at a concentration of 20 μmol/L. Their selectivity in living cells was examined by immunoprecipitation–Western blot analysis. A549 cells were serum starved for 48 hours and subsequently serum stimulated for 2 hours in the presence or absence of 20 μmol/L of NSC-35400, NSC-35950, and RRD-251; Raf-1 peptide conjugated to penetratin (18) was used as a positive control, and a Raf-1 scrambled peptide was used as a negative control. It was found that the compounds inhibited the binding of Raf-1 to Rb (Fig. 2A), whereas the binding of Rb to E2F1 was not affected. To further confirm the selectivity of RRD-251, cyclin E was immunoprecipitated from lysates of quiescent cells or those serum stimulated for 8 hours in the presence or absence of RRD-251; Western blotting of the immunoprecipitates showed that RRD-251 did not inhibit the binding of Rb to cyclin E (Fig. 2B). A similar experiment was done on lysates from cells that were serum stimulated for 2 hours; RRD-251 did not inhibit the binding of B-Raf to Rb (Fig. 2C). Similarly, the binding of Raf-1 to Mek1/2 was not affected by RRD-251 (Fig. 2D). Examination of lysates from cells serum stimulated for 2 hours in the presence of RRD-251 showed a reduction in Rb phosphorylation, as seen by Western blotting (Fig. 2C,, row 5). Interestingly, in vitro kinase assays showed that RRD-251 did not affect the kinase activities associated with cyclin D (Fig. 3A,, left) or cyclin E (Fig. 3A,, middle) or Raf-1 (Fig. 3A , right). These results suggest that the reduction in Rb phosphorylation in cells treated with RRD-251 is due to a disruption of the association of Raf-1 with Rb and that Raf-1 has to physically interact with Rb to inactivate it.

RRD-251 inhibits E2F transcriptional activity. We next reasoned that if the disruption of the Rb/Raf-1 binding has functional consequences on cellular physiology, then RRD-251 should affect the transcriptional activity of E2F1. To examine this, transient transfection experiments were done in control A549 cells, as well as A549 cells stably expressing two different shRNA constructs (sh6 and sh8) targeting Rb; these A549 cells had significantly less Rb protein compared with parental A549 cells (Supplementary Fig. S1A). Transfection of E2F1 induced the expression of an E2CAT reporter; treatment of the transfected cells with RRD-251 repressed E2F1-mediated transcription in a dose-dependent manner in wild-type (wt) A549 cells but not in the A549 cells lacking Rb (Fig. 3B); this suggests that the presence of Rb is necessary for RRD-251 to function. ChIP analysis of the transfected E2 promoter revealed a reduction in the amount of Raf-1 on the promoter with increasing doses of RRD-251 (Supplementary Fig. S1B). We had reported that Raf-1 can be detected on proliferative promoters upon serum stimulation, and these results indicate that RRD-251 probably affects E2F-mediated transcription by dissociating Raf-1 from the promoters. The effect of RRD-251 on the expression of two endogenous E2F-regulated proliferative promoters was next examined. A549 cells were serum starved for 72 hours and serum stimulated for 24 hours in the presence or absence of RRD-251 and the level of TS gene expression was assessed by real-time PCR. Inhibition of the Rb/Raf-1 interaction led to the silencing of the TS promoter (Fig. 3C); similar results were obtained on the cdc6 gene (Supplementary Fig. S1C). We had reported that the binding of Raf-1 to Rb resulted in the dissociation of the corepressor Brg-1 from E2F-responsive proliferative promoters (18); ChIP assays were carried out to examine whether RRD-251 affects this process. It was found that the association of Raf-1 to the above promoters upon serum stimulation was disrupted by pretreatment of cells with RRD-251 (Fig. 3D). Furthermore, dissociation of the corepressor Brg-1 from these promoters was also inhibited by RRD-251. This suggests that RRD-251 can modulate the transcriptional regulatory functions of Rb by modulating its phosphorylation status and affecting its interaction with chromatin remodeling proteins like Brg-1. The association of E2F1, HDAC1, and HP1 with these promoters was not affected by RRD-251, as seen by ChIP assays (Fig. 3D).

Inhibition of proliferation by RRD-251 is dependent on Rb status. Given the ability of RRD-251 to inhibit Rb phosphorylation and repress E2F transcriptional activity, it was examined if it could inhibit cell proliferation and whether such an inhibition required a functional Rb gene. RRD-251 (20 μmol/L) was effective at inhibiting serum-induced S-phase entry in parental A549 cells but had no effect on cells stably expressing sh6 and sh8, which lacked Rb (Fig. 4A). We next tested RRD-251 on cancer cell lines that carried natural mutations or deletion of Rb. Proliferation of Saos-2 osteosarcoma cells that have a loss of Rb (24) was not affected by RRD-251, whereas the U2-OS osteosarcoma cells carrying wt Rb were arrested (Fig. 4A). RRD-251 was unable to inhibit proliferation in the Rb mutant DU145 prostate cancer cells, yet could inhibit 50% of S-phase cells in PC3 cells (wt Rb; Fig. 4A). RRD-251 (20 μmol/L) did not inhibit proliferation of lung cancer cell lines H596 and H2172, both of which harbor mutations in Rb, yet treatment of H1650 and H1299 (wt Rb) with RRD-251 inhibited proliferation by 90% and 70%, respectively (Fig. 4A). It was next examined whether RRD-251 could inhibit the proliferation of cells that have mutations in the signaling pathways that regulate Rb function rather than in the Rb gene itself. RRD-251 could inhibit S-phase entry by 50% to 65% in pancreatic cancer cell lines Aspc1, PANC1, and CAPAN2 that harbor a nonfunctional p16INK4a gene (ref. 25; Fig. 4B). RRD-251 also inhibited S-phase entry of glioblastoma cell lines U87MG and U251MG, both of which are null for p16 and PTEN (26). The metastatic human breast cancer cell line MDA-MB-231 harbors a K-Ras mutation and overexpresses epidermal growth factor receptor (EGFR; ref. 27); 20 μmol/L RRD-251 inhibited its proliferation by 56% (Fig. 4B). The A375 melanoma cell line harbors the V600E B-Raf mutation (28), and RRD-251 inhibited S-phase entry by 58%. Prostate cell lines LNCaP and PC3 both contain mutations in K-Ras and PTEN genes (29), and RRD-251 inhibited proliferation at 86% and 35%, respectively (Fig. 4B). These results indicate that disruption of Rb/Raf-1 interaction could inhibit the proliferation of cell lines harboring a wide array of mutations in upstream signaling molecules, as long as Rb itself is not altered at the genetic level. Experiments were also carried out to examine the effect of RRD-251 in suppressing the adherence-independent growth of cells in soft agar; it was found that RRD-251 could significantly inhibit the growth of A549, H1650, and SK-MEL-5, SK-MEL-28, and PANC1 cells in soft agar (Fig. 4C).

We next reasoned that, if RRD-251 targets selectively the Rb/Raf-1 interaction, the forced expression of a downstream target of Rb, such as E2F1, but not of the upstream regulator cyclin D, would rescue the antiproliferative effects of RRD-251. To this end, A549 cells were serum starved for 48 hours and infected with Ad-E2F1 or Ad-cyclin D in the presence or absence of 20 μmol/L of RRD-251 for 36 hours. Ad-GFP infected cells were used as a control. BrdUrd incorporation assays showed that ectopic expression of E2F1 efficiently overcame the antiproliferative activity of RRD-251, whereas overexpression of cyclin D had only a partial effect (Fig. 4D), showing that the growth inhibition by RRD-251 occurs at the level of Rb and its downstream target, E2F1, can rescue the growth suppression.

RRD-251 inhibits angiogenesis in vitro and in vivo. Raf-1 kinase plays a role in facilitating angiogenesis (30, 31), and it has been suggested that Raf-1–mediated inactivation of Rb is involved in the process (18). To examine whether angiogenic tubule formation could be inhibited by RRD-251, HUVECs were grown in Matrigel in the presence or absence of 20 μmol/L RRD-251; RRD-251 significantly inhibited the angiogenic tubule formation (Fig. 5A). These results were confirmed in an ex vivo experiment using rat aortic rings. As shown in Fig. 5B, 20 μmol/L RRD-251 was able to inhibit angiogenic sprouting from rat aortic rings grown in Matrigel for 7 days. Quantitation of vessel area showed a significant reduction in angiogenesis (P = 0.000007; Supplementary Fig. S2A). We next examined whether RRD-251 could inhibit angiogenesis in Matrigel plugs in vivo (22). Aythmic nude mice were injected with cold Matrigel in both flanks. Mice were given either vehicle or RRD-251 50 MPK body weight by i.p. injection daily for 1 week. Mice were injected with 100 MPK FITC-Dextran via the tail vein before euthanasia, and Matrigel plugs were fixed in formalin. Angiogenesis in the entire plugs were assessed by confocal imaging. FITC images showed growth of angiogenic tubules in plugs from mice that received vehicle; in contrast, there was a remarkable inhibition of angiogenic vessel formation in the Matrigel plugs from mice treated with RRD-251 (Fig. 5C). Quantitation of vessel intensity plotted as relative angiogenesis per image shows significant inhibition; P = 0.0004 (n = 12; Supplementary Fig. S2B). Further examination of the Matrigel plugs by H&E staining showed a complete inhibition of cells migrating into the Matrigel for vessel formation (Fig. 5D). It was also examined if RRD-251 could inhibit vascular endothelial growth factor (VEGF) levels in cell culture. A549 cells treated with RRD-251 for 24 hours displayed decreased VEGF levels compared with control (Supplementary Fig. S2C). Furthermore, treatment with RRD-251 in the presence of VEGF for 12 hours could prevent Rb phosphorylation in HAEC cells (Supplementary Fig. S2D).

Antitumor activity of RRD-251. Given the ability of RRD-251 to inhibit cell proliferation, adherence-independent growth, and angiogenesis, we examined whether RRD-251 could inhibit tumor growth in vivo in nude mouse xenograft models. Athymic nude mice were implanted s.c. with 1 × 10⁷ A549 cells bilaterally, and the tumors were allowed to reach 200 mm³ in size before p.o. or i.p. administration of RRD-251 or vehicle (18, 23). Tumors from vehicle-treated mice grew to an average size of 1,040 ± 128 mm³; in contrast, tumors in mice treated with RRD-251 did not grow and even regressed slightly (50 MPK i.p. 145 ± 20 mm³, 150 MPK p.o. 148 ± 32 mm³; Fig. 6A,, left). P.o. dose response experiments were carried out on A549 xenografts, which resulted in RRD-251 at 100 MPK and 150 MPK completely inhibiting tumor growth (Fig. 6A,, middle). Tumors from vehicle-treated mice reached an average size of 996 ± 180 mm³; in contrast, tumors in mice treated with RRD-251 (p.o.) responded in a dose-dependent manner. Complete inhibition was seen in 100 MPK p.o. at 293 ± 44 mm³ and 150 MPK p.o. at 237 ± 67 mm³ (Fig. 6A,, middle). Similar results were observed with H1650 xenograft tumors; RRD-251 inhibited tumor growth significantly [2,185 ± 326 mm³ in vehicle-treated animals compared with 557 ± 76 mm³ in RRD-251 (50 MPK i.p.)–treated animals; Fig. 6A , right].

Tumors were harvested at the end of the treatment and analyzed by immunohistochemical staining with H&E and antibodies to Ki-67, phosphorylated Rb (807, 811), and CD-31. A significant inhibition of proliferation was observed in tumors from RRD-251–treated animals, as seen by Ki-67 staining (Fig. 6B); phosphorylation of Rb was also reduced as seen by staining with an antibody to phosphorylated Rb (Fig. 6B). Tumors also showed a reduction in microvasculature, as seen by CD31 staining (Fig. 6B). Quantitation of Ki-67 staining, phosphorylated Rb staining, and CD31 staining is shown in Supplementary Fig. S3A–C. To assess whether RRD-251 affected its target, tumors lysates were prepared from vehicle-treated and RRD-251–treated mice and Rb/Raf-1 interaction assessed by immunoprecipitation–Western blots. RRD-251 treatment had led to a reduction in Rb/Raf-1, but not Rb/E2F1 interaction in tumor xenografts (Fig. 6C).

Tumor growth inhibition by RRD-251 is Rb-dependent. Because RRD-251 did not inhibit the proliferation of A549 cells lacking Rb in vitro, experiments were done to assess whether tumors generated from these cells can respond to RRD-251 in vivo. An experiment as in Fig. 6A was carried out on nude mice carrying tumors from A549 cells stably expressing shRNAs for Rb (sh6 and sh8). Interestingly, these tumors did not respond to RRD-251 and continued to grow at the rate of the vehicle treated tumors (Fig. 6D , left and right). To examine whether the sh6 and sh8 tumors maintained down-regulation of Rb, lysates were made from the sh6 and sh8 tumors at the end of the experiment and a Western blot was done for Rb. It was found that these tumors lacked Rb, further confirming that RRD-251 specifically targets the Rb/Raf-1 protein interaction to inhibit cell proliferation and tumor growth (Supplementary Fig. S3D).

The Ras/Raf/Mek/MAPK cascade is a proliferative pathway induced by a wide array of growth factors, is activated in many human tumors (32–34), and is an attractive target for the development of anticancer drugs (30, 31, 35, 36). Raf-1 kinase itself has been targeted for cancer therapy, and two clinical attempts have been made to inhibit Raf-1 activity in patients (37–39). It has been shown that signaling events involving the MAPK cascade do not proceed in a linear fashion and instead they have substrates outside the cascade (17, 40, 41). In this context, the Rb protein seems to be a cellular target of the Raf-1 kinase outside the MAPK cascade. In addition, Rb/Raf-1 binding was elevated in tumor compared with adjacent normal controls (19), suggesting that Raf-1–mediated inactivation of Rb might contribute to oncogenesis. Whereas it is established that Rb gene itself is mutated in cancers like retinoblastoma, osteosarcoma, and small cell lung carcinoma, majority of tumors harbor mutations in the upstream regulators of Rb function (7, 8). These include genes like Ras, PTEN, p16INK4, as well as receptor tyrosine kinases (42–44). Our results show that the disruption of the Rb/Raf-1 interaction can be fruitfully used to inhibit the proliferation of cells harboring such mutations in the Rb regulatory pathway. Thus, we believe that these molecules have the potential to target a wide variety of human cancers.

Whereas inhibitors of cell proliferation, DNA-damaging agents, and microtubule disruptors have widely been used as anticancer agents, developments in the past decade have shown that targeting angiogenesis is also an effective way of combating tumor growth (45). In this context, our results show that RRD-251 cannot only inhibit cell proliferation, but can also inhibit neoangiogenesis in vitro and in vivo. Given the published reports that Raf-1 kinase contributes to angiogenesis and that VEGF can induce Rb phosphorylation, it is likely that RRD-251 is inhibiting angiogenesis by affecting these molecules (31, 46). The ability of RRD-251 to inhibit both cell proliferation and angiogenesis might be acting in a two-pronged manner to inhibit the growth of tumors in vivo; as can be imagined, these are desirable features in anticancer drugs.

Whereas it has been difficult to generate small molecule inhibitors of protein-protein interactions that are clinically active (39), recent success in disrupting the hdm2-p53 (47) interaction shows that this is a viable strategy to develop novel anticancer drugs. Identification of RRD-251 as a cell-permeable, p.o. available, and highly selective inhibitor of the Rb/Raf-1 interaction is an example of the practicality of targeting protein-protein interaction for cancer therapy. Although we find that RRD-251 inhibits Rb/Raf-1 in vitro at nanomolar concentrations in an in vitro ELISA assay, higher concentrations are needed to inhibit cell proliferation, as well as growth of cells in soft agar; this finding is similar to what has been observed with other anticancer drugs, such as BAY-43-9006, R547, and Iressa (48–50). At the same time, our in vivo studies show that concentrations can be achieved in vivo where RRD-251 has a significant therapeutic benefit.

The finding that RRD-251 is effective at inhibiting the proliferation of cells harboring a wide variety of mutations in signaling cascades that inactivate Rb, but does not affect cells carrying mutated Rb or no Rb, shows the specificity of this agent. Rb protein has been reported to interact with ∼100 proteins in the cell; it can be imagined that small molecules that can maintain the tumor suppressor functions of Rb by disrupting its physical interaction with other proteins would be a fruitful avenue to develop novel anticancer drugs.

No potential conflicts of interest were disclosed.

Grant support: National Cancer Institute grants CA63136 and CA118210. R. Kinkade is a recipient of a predoctoral award from American Heart Association.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.