Background: The basic region-leucine zipper protein c-Jun has been linked to cell proliferation, transformation, and apoptosis. However, a direct role for c-Jun in angiogenesis has not been shown. Methods: We used human microvascular endothelial cells (HMEC-1) transfected with a DNAzyme targeting the c-Jun mRNA (Dz13), related oligonucleotides, or vehicle in in vitro models of microvascular endothelial cell proliferation, migration, chemoinvasion, and tubule formation, a rat model of corneal neovascularization, and a mouse model of solid tumor growth and vascular endothelial growth factor (VEGF)–induced angiogenesis. All statistical tests were two-sided. Results: Compared with mock-transfected cells, HMEC-1 cells transfected with Dz13 expressed less c-Jun protein and possessed lower DNA-binding activity. Dz13 blocked endothelial cell proliferation, migration, chemoinvasion, and tubule formation. Dz13 inhibited the endothelial cell expression and proteolytic activity of MMP-2, a c-Jun–dependent gene. Dz13 inhibited VEGF-induced neovascularization in the rat cornea compared with vehicle control (Dz13 versus vehicle: 4.0 neovessels versus 30.7 neovessels, difference = 26.7 neovessels; P = .004; area occupied by new blood vessels for Dz13 versus vehicle: 0.35 mm 2 versus 1.52 mm 2 , difference = 1.17 mm 2 ; P = .005) as well as solid melanoma growth in mice (Dz13 versus vehicle at 14 days: 108 mm 3 versus 283 mm 3 , difference = 175 mm 3 ; P = .006) with greatly reduced vascular density (Dz13 versus vehicle: 30% versus 100%, difference = 70%; P <.001). Conclusion: DNAzymes targeting c-Jun may have therapeutic potential as inhibitors of tumor angiogenesis and growth.
The formation of new blood vessels from preexisting vasculature (i.e., angiogenesis) underpins normal biologic processes, such as wound healing and reproduction, as well as many pathologic conditions, including sight-threatening ocular disorders ( 1 ), restenosis following angioplasty ( 2 ), and solid tumor growth and dissemination ( 3 ). Angiogenesis is a complex multistep process that involves the proteolytic degradation of the basement membrane and surrounding extracellular matrix as well as microvascular endothelial cell proliferation, migration, tube formation, and structural reorganization ( 4 ). The efficacy of endogenous and synthetic inhibitors of angiogenesis is currently being evaluated in clinical trials ( 5 ).
Our understanding of the key transcription factors involved in the process of angiogenesis is limited at the present time. One potential factor is c-Jun. c-Jun is a member of the basic region–leucine zipper (bZIP) protein family that homodimerizes and heterodimerizes with other bZIP proteins to form the transcription factor activating protein-1 (AP-1) [reviewed in ( 6 ) ]. Results of studies conducted over the last few decades have linked c-Jun to cell proliferation, transformation, and apoptosis. For example, skin tumor promotion is blocked in mice expressing a dominant-negative transactivation mutant of c-jun ( 7 ). Microinjection of antibodies to c-Jun into Swiss 3T3 cells inhibits cell cycle progression ( 8 ). Compared with primary fibroblasts cultured from wild-type littermates, primary fibroblasts cultured from live heterozygous and homozygous mutant c-jun mouse embryos, which die in utero ( 9 , 10 ), have greatly reduced growth rates in culture that cannot be overcome by the addition of a mitogen ( 10 ). Impaired proliferation of immortalized c-jun null mouse embryo fibroblasts can be rescued by overexpression of wild-type exogenous c-jun ( 11 ). c-Jun has also been implicated in apoptosis. For example, c-jun null mouse embryo fibroblasts are resistant to apoptosis induced by UVC radiation ( 12 ). Despite this large body of work, c-Jun has not been linked directly to angiogenesis.
Insights into the function of a given gene product in a complex biologic process such as angiogenesis may be obtained with gene-targeting strategies that use DNA enzymes (DNAzymes). DNAzymes are synthetic, single-stranded DNA catalysts that can be engineered to bind to their complementary sequence in a target messenger RNA (mRNA) through Watson–Crick base pairing and cleave the mRNA at predetermined phosphodiester linkages ( 13 ). These catalysts have emerged as a potential new class of nucleic acid–based drugs because of their relative ease and low cost of synthesis, high stability, and flexible rational design features. Sequence-specificity of a DNAzyme for mRNA is determined by the sequence of deoxyribonucleotides in the hybridizing arms of the DNAzyme; the hybridizing arms are generally seven or more nucleotides long ( 13 ). A “general purpose” DNAzyme comprising a 15-nucleotide cation-dependent catalytic domain (designated “10-23”) that cleaves the phosphodiester linkage between an unpaired purine and a paired pyrimidine in the target mRNA ( 14 ) was developed some years ago using a systematic in vitro selection strategy. Here, we used such DNAzymes to examine the roles of c-Jun in determining microvascular endothelial cell phenotypes in vitro and in solid tumor growth and angiogenesis in vivo.
M aterials and M ethods
Cell Culture, DNAzyme Synthesis, and Transfections
The immortalized human microvascular endothelial cell line HMEC-1 (American Type Culture Collection, Manassas, VA) was grown in MCDB 131 medium (Invitrogen, Gaithersburg, MD) containing 10% fetal bovine serum (FBS) (Invitrogen), 2 m Ml -glutamine, 10 ng/mL epidermal growth factor, 1 μg/mL hydrocortisone, and 5 U/mL penicillin–streptomycin (all from Invitrogen). Murine brain microvascular endothelial (bEND-3) cells (American Type Culture Collection) and B16F10 cells (American Type Culture Collection) were cultured in Dulbecco's modified Eagle medium (Invitrogen) containing 10% FBS, 2 m Ml -glutamine, and 5 U/mL penicillin–streptomycin.
DNAzymes were synthesized with an inverted thymidine at the 3′ position (Tri-Link Biotechnologies, San Diego CA) and purified by high-pressure liquid chromatography. Sequences of Dz13 (a c-jun–targeting DNAzyme bearing 9+9 nucleotide arms; cleavage junction in c-jun mRNA is G 1311 U), Dz13scr (sequence-scrambled counterpart of Dz13), As13 (antisense equivalent of Dz13, lacking the catalytic domain), and As13scr (sequence-scrambled counterpart of As13) are 5′-CGGGAGGAAggctagctacaacgaGAGGCGTTG-Ti-3′, 5′-GCGACGTGAggctagctacaacgaGTGGAGGAG-Ti-3′, 5′-CGGGAGGAACGAGGCGTTG-Ti-3′, and 5′-GCGACGTGACGTGGAGGAG-Ti-3′, respectively. Sequences of Dz13M, Dz13(11+11), Dz13(10+10), and Dz13(8+8) are 5′-CGGGAGGAAggctaCctacaacgaGAGGCGTTG-Ti-3′, 5′-GACGGGAGGAAggctagctacaacgaGAGGCGTTGAG-Ti-3′, 5′-ACGGGAGGAAggctagctacaacgaGAGGCGTTGA-Ti-3′, and 5′-GGGAGGAAggctagctacaacgaGAGGCGTT-Ti-3′, respectively. Sequences of Dz13(11+11)scr, Dz13(10+10)scr, and Dz13(8+8)scr are 5′-GAGCGACGTGAggctagctacaacgaGTGGAGGAGAG-Ti-3′, 5′-AGCGACGTGAggctagctacaacgaGTGGAGGAGA-Ti-3′, and 5′-CGACGTGAggctagctacaacgaGTGGAGGA-Ti-3′, respectively. Nucleotides in lower case represent the 10-23 catalytic domain. Ti represents a 3′-3′–linked inverted thymidine. Seventeen of the 18 nucleotides in the Dz13 target site in human c-Jun mRNA (5′-CAA CGC CUC G 1311 | UUC CUC CcG-3′) are conserved in murine and rat c-Jun mRNA (5′-CAA CGC CUC G | UUC CUC CaG-3′). The vertical line represents the DNAzyme cleavage junction in the mRNA.
We used FuGENE6 reagent (Roche Diagnostics, Castle Hill, Sydney, Australia) according to the manufacturer's instructions to transfect HMEC-1 and bEND-3 cells (at 60%–75% confluence) with DNAzymes, antisense oligonucleotides, or their sequence-scrambled counterparts (all of which had been resuspended in water at the concentrations indicated) 6 hours after the cells had been placed in serum-free medium to initiate growth arrest. Control cells were mock-transfected with vehicle alone. The transfected cells were incubated for 18 hours in serum-free medium and then transfected a second time with the same DNAzyme, antisense oligonucleotide, or vehicle alone and incubated in serum-containing medium for the times indicated. Where indicated, HMEC-1 cells were co-transfected twice with the indicated DNAzyme and 10 μg of pRSVjun, a c-jun expression vector ( 15 ). This double-transfection protocol resulted in a transfection efficiency of approximately 50% (data not shown).
Western Blot Analysis
Growth-arrested HMEC-1 cells (1 × 10 6 cells in 100-mm petri dishes) transfected twice with DNAzyme or mock-transfected with vehicle alone were incubated in medium containing 5% serum and 100 ng/mL VEGF 165 (the 165-amino-acids-long form of vascular endothelial growth factor; Promega, Annandale, New South Wales, Australia) or in serum-free medium for 2 hours and then washed extensively in phosphate-buffered saline (PBS) at pH 7.5. Total cell extracts were prepared by adding 1 mL of RIPA buffer (150 m M NaCl, 50 m M Tris–HCl [pH 7.5]), 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 1% Triton X-100, 5 m M EDTA, 10 μg/mL leupeptin, 1% aprotinin, and 2 m M phenylmethyl sulphonyl fluoride [PMSF]) to each dish, followed by a freeze–thaw cycle and centrifugation at 8000 g in a microfuge (Sigma, Germany) for 10 minutes at 4 °C to remove cell debris. Proteins in the supernatants were resolved (using 5 μg of total protein from each sample, as determined by a Bio-Rad protein assay; Bio-Rad, Hercules, CA) on 12% polyacrylamide gels and transferred to a polyvinylidene fluoride (PVDF) nylon membrane (Millipore, Bedford, MA). The membrane was then incubated with polyclonal c-Jun and Sp1 anti-peptide antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) at 1 : 1000 final dilution. Antibody binding was detected with the use of a Western Lightning chemiluminescence kit (PerkinElmer Life Sciences, Boston, MA) and quantitated digitally by using a CanoScan scanner (Canon, Beijing, China) and QuantityOne 4.1.1 software (Bio-Rad).
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assay
Growth-arrested HMEC-1 cells (1 × 10 6 cells in 100-mm petri dishes) that had been either mock transfected or transfected with DNAzyme (Dz13 or Dz13scr) and exposed to serum-containing medium for 2 hours were scraped into ice-cold PBS, pelleted by centrifugation at 200 g for 10 minutes at 4 °C, and resuspended in 100 μL of cold lysis buffer (10 m M HEPES [pH 7.9], 1.5 m M MgCl 2 , 10 m M KCl, 0.5% Nonidet P-40, 1 m M dithiothreitol [DTT], 0.5 m M PMSF, 4 μg/mL aprotinin, and 10 μg/mL leupeptin). After incubation on ice for 5 minutes, the suspension was spun at 8000 g for 1 minute at 4 °C, and the supernatants were discarded. The nuclear fractions (pellets) were resuspended in 20 m M HEPES (pH 7.9), 1.5 m M MgCl 2 , 420 m M NaCl, 0.2 m M EDTA, 1 m M DTT, 0.5 m M PMSF, 4 μg/mL aprotinin, and 10 μg/mL leupeptin and shaken for 20 minutes at 4 °C. The suspensions were spun at 8000 g for 5 minutes at 4 °C, and the supernatants (nuclear extract, 20 μL) were mixed with an equal volume of cold 20 m M HEPES (pH 7.9), 100 m M KCl, 0.2 m M EDTA, 20% glycerol, 1 m M DTT, 0.5 m M PMSF, 4 μg/mL aprotinin, and 10 μg/mL leupeptin. Electrophoretic mobility shift assay (EMSA) reactions were performed by incubating 2 μL (10 μg) of the nuclear extract mixture in a 20-μL reaction that contained 10 m M Tris–HCl (pH 7.5), 50 m M NaCl, 0.5 m M DTT, 0.5 m M EDTA, 1 m M MgCl 2 , 5% glycerol, 2.5 μg of poly dI-dC, and 150 000 cpm of probe for 20 minutes at room temperature. The double-stranded oligonucleotide probes used for EMSA were Oligo c-Jun (5′-CGCTTGATGAGATCAGCCGGAA-3′, sense strand) or Oligo A [30 base pairs (bp) of the proximal platelet-derived growth factor A-chain (PDGF-A) promoter containing a nuclear factor 1 (NF-1) binding site ( 16 ) ], each of which were labeled at the 5′-ends with [γ 32 P]ATP using T4 polynucleotide kinase (PerkinElmer, Wellesley, MA). Protein–oligonucleotide complexes were resolved by electrophoresis on non-denaturing 8% polyacrylamide gels that used a Tris–borate–EDTA buffer system. For studies of antibody inhibition of protein–oligonucleotide complex formation, nuclear extracts made from growth-arrested HMEC-1 cells previously transfected with DNAzymes and exposed to serum for 2 hours were incubated with 2 μg of rabbit polyclonal c-Jun antibodies (Santa Cruz Biotechnology) for 10 minutes before the 32 P-labeled probe was added.
Cell Proliferation Assay
Growth-arrested HMEC-1 cells transfected with DNAzymes or mock-transfected with vehicle in 96-well plates were incubated in serum-containing medium for 3 days, harvested by trypsinization, resuspended in Isoton II (Coulter Electronics, Brookvale, New South Wales, Australia), and quantified with the use of a Coulter counter (Z series; Coulter Electronics). Transfections were performed directly in the 96-well plates.
Wound-Healing/Cell Migration Assay
HMEC-1 cells were incubated in serum-free medium for 6 hours, transfected twice with DNAzymes or mock-transfected with vehicle alone, and grown to confluent monolayers in 100-mm petri dishes for 18 hours in serum-free medium (three dishes per group). The medium was replaced with serum-containing medium and the monolayers were disrupted (i.e., wounded) by scraping them with a P200 micropipette tip. Two days after scraping, the cells were washed twice in PBS (pH 7.4), fixed in 4% paraformaldehyde (vol/vol), and stained with hematoxylin–eosin. The number of cells in the denuded (scraped) zone of each dish was counted at ×100 magnification in a blinded fashion. Each dish was counted three times.
HMEC-1 Cell Migration/Invasion Assay
Polycarbonate membranes (12-μm pore size) were incubated overnight with 1 mg/mL Matrigel (BD Biosciences, Chicago, IL) or 1 mg/mL collagen type I (Sigma, St. Louis, MO), air-dried, and placed in modified Boyden chambers (Nucleopore, Pleasanton, CA). A suspension of HMEC-1 cells (4 × 10 5 cells per milliliter) previously transfected with DNAzyme or mock-transfected with vehicle was placed in the upper chamber. The lower chamber contained medium supplemented with 20 ng/mL basic fibroblast growth factor-2 (FGF-2; Promega). The chambers were incubated for 24 hours at 37 °C and the membranes were fixed in methanol and stained with hematoxylin. Cells that had migrated to the underside of the membrane were quantitated at ×400 magnification.
Microvascular Endothelial Cell Tubule Formation Assay
HMEC-1 cells grown in 100-mm petri dishes were transfected twice with DNAzymes or mock-transfected with vehicle alone, harvested by trypsinization, resuspended in medium that contained 10% serum or 100 ng/mL VEGF 165 and plated (at 30 000 cells per well; 100 μL of suspension) into 96-well plates that were coated with 100 μL of 1 mg/mL Matrigel. Tubule formation was quantified 8 hours after plating by viewing the cells at ×400 magnification in a blinded manner.
Measurement of Matrix Metalloproteinase 2 mRNA by Reverse Transcription–Polymerase Chain Reaction
HMEC-1 cells were transfected with 0.4 μ M DNAzyme or mock-transfected with vehicle alone 6 hours after being placed in serum-free medium. Eighteen hours after transfection, 10 ng/mL transforming growth factor-beta 1 (TGF-β1; Promega) was added to the culture medium and the cells were transfected again with 0.4 μ M DNAzyme or mock-transfected with vehicle alone and incubated for 24 hours. We prepared total RNA from the transfected cells with the use of Trizol reagent (Invitrogen). We then synthesized single-strand complementary DNAs (cDNAs) in 20-μL reactions that contained 4 μg of total RNA, 200 U of Superscript II reverse transcriptase (Invitrogen), each deoxynucleotidetriphosphate (500 μ M ), and 0.5 μg of oligo (dT) 15 (Invitrogen). Next, we used reverse transcription–polymerase chain reaction to amplify matrix metalloproteinase 2 (MMP-2) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) RNAs in a 20-μL reaction that contained the reverse-transcribed cDNA template, 1 U of DNA polymerase (Promega), each deoxynucleotidetriphosphate (100 μ M ), 30 m M MgCl 2 (Invitrogen) and each of the four primers (0.1 μ M ). PCR analysis was performed by first determining the volume of cDNA template required to achieve equivalent levels of GAPDH (452-bp product) between samples and using that volume to amplify MMP-2 cDNA (446 bp). Primers were as follows: MMP-2 (forward primer, 5′-GGGACAAGAACCAGATCACATAC-3′; reverse primer, 5′-CTTCTCAAAGTTGTAGGTGGTGG-3′); GAPDH (forward primer, 5′-ACCACAGTCCATGCCATCAC-3′; reverse primer, 5′-TCCACCACCCTGTTGCTGTA-3′). The PCR conditions, optimized for amplification of MMP-2 RNA, were 22 cycles of 95 °C for 30 seconds, 57 °C for 30 seconds, and 72 °C for 40 seconds.
MMP-2 Enzyme-Linked Immunosorbent Assay
HMEC-1 cells transfected twice with DNAzymes or mock-transfected with vehicle control were incubated for 2 days in medium containing 10 ng/mL TGF-β1 and 0.1% FBS. The culture medium was harvested and centrifuged to remove debris, and the supernatants were normalized for equal amounts of protein. We used an enzyme-linked immunosorbent assay (ELISA) kit (Amersham Biosciences, Buckinghamshire, U.K.) to determine the level of MMP-2 in 100 μL of each supernatant.
Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis and Gelatin Zymography Assay for Metalloproteinase Activity
Culture supernatants obtained from HMEC-1 cells transfected twice with DNAzymes, co-transfected with DNAzyme and the c-jun expression vector, or mock-transfected with vehicle alone were resolved (equal amounts of protein per lane) at 4 °C on a 10% polyacrylamide resolving gel (with a 4% polyacrylamide stacking gel) that contained 1 mg/mL bovine type B gelatin (Sigma). The gel was soaked in 2.5% Triton X-100 (Sigma) for 1 hour and then incubated in substrate buffer (50 m M Tris–HCl [pH 7.6], 10 m M CaCl 2 , 0.02% NaN 3 ) overnight at 37 °C. The gel was stained by incubating in 0.2% Coomassie Blue R-250 (Bio-Rad) in water, methanol, and glacial acetic acid (5 : 4 : 1, vol/vol/vol) for 1 hour and destained in methanol and glacial acetic acid (4 : l, vol/vol) to reveal clear areas corresponding to protein bands with gelatinolytic (i.e., metalloproteinase) activity.
Rat corneal neovascularization assay.
We implanted the corneas (one eye per animal) of 7-week-old male Sprague-Dawley rats (five rats per treatment group; Animal Resources Centre, Perth, Australia) with 0.57-mm diameter nitrocellulose filter disks (0.45-μm pore size; Millipore, Bedford, MA) that had been soaked for 30 minutes in 30 μ M VEGF 165 in 82 m M Tris–HCl (pH 6.9) or 82 m M Tris–HCl (pH 6.9) alone (control) in covered petri dishes on ice. We used a modified von Graefe cataract knife to create a stromal pocket by cutting 1 mm in the proximal to distal direction from the corneal limbus and inserted one VEGF-soaked disk into each stromal pocket. Immediately following implantation, we used a 5-μL Hamilton syringe to inject 100 μg (9.6 nmol) of Dz13 or Dz13scr in a volume of 1.5 μL or 1.5 μL of vehicle (PBS, pH 7.5) into the conjunctiva directly opposite the disk implant over a period of 30 seconds. During implantation, the rats were given a general anesthetic (0.5% halothane, as needed), a local anesthetic (0.4% oxybuprocaine, 50 μL), and a general analgesic (buprenorphine, 0.025 mg/kg). Chloramphenicol ointment (1%) was applied to the implanted eyes after the procedure, and the rats were allowed to recover from anesthesia while lying on a heating pad, after which they were returned to clean cages.
Five days after disk implantation, the rats were killed by CO 2 asphyxiation, and slit lamp biomicrophotographs were taken of their corneas in a blinded fashion before the corneas were removed. While viewing the corneas with the use of a dissecting microscope, we measured the area occupied by neovascularization (according to the formula 0.2 × π × clock-hour surface area occupied by vascularization × maximum vessel length) and the number of vessels growing in the cornea. All measurements were made in a blinded fashion. No inflammatory response around the cornea was observed among rats in any of the treatment groups. A positive neovascularization response was recorded only if sustained directional ingrowth of capillary sprouts and hairpin loops toward the implant was observed. A negative response was recorded when no growth was observed or when only an occasional sprout or hairpin loop displaying no evidence of substantial growth was detected. Implantation of disks soaked in vehicle elicited no angiogenic response, suggesting that neovascularization associated with implantation of the VEGF 165 -soaked disks was not due to the surgical procedure per se.
B16 tumor growth study.
Six-week-old female C57BL/J6 mice (5–6 mice per group; Animal Resource Centre) were anesthetized with methoxyflurane and injected subcutaneously in the dorsal mid-back region with 5 × 10 4 B16F10 cells resuspended in 200 μL of vehicle (50% Matrigel [in saline] containing 2.5 μL of FuGENE6 transfection reagent) alone or 200 μL of vehicle containing 750 μg of Dz13, Dz13scr, As13, or As13scr. We measured body weights and the dimensions of palpable tumors (with the use of a digital caliper) at least twice per week and at study termination (i.e., 14 days after injection) in a blinded fashion. The mice were killed via cervical dislocation and their tumors were immediately removed and frozen on dry ice. The tumors were cryosectioned and stained with hematoxylin–eosin. The total number of blood vessels in three separate sections of each tumor was quantified by a blinded observer. Tumor volumes (in mm 3 ) were determined according to the formula length × width × height × (π/6). All animal procedures were in accordance with the guidelines put forward by the University of New South Wales Institutional Animal Care and Ethics Committee.
Sections of a formalin-fixed, paraffin-embedded cutaneous malignant melanoma (archival tissue from a 72-year-old man) were stained with rabbit polyclonal antibodies to c-Jun or CD31 (a vascular endothelial cell marker) and a goat polyclonal antibody to MMP-2 (1 : 1000 dilution; Santa Cruz Biotechnology), as previously described ( 17 ). Immunoreactivity was revealed by incubating the sections with biotinylated secondary anti-rabbit or anti-goat antibodies (Santa Cruz Biotechnology), as appropriate, followed by 3-amino-9-ethylcarbazole (ready-to-use AEC; DAKO, Carpinteria, CA), and counterstaining with Gill's hematoxylin solution (Sigma).
A test for unequal variance was performed on all data with a two-sample F test (that is, no assumption). Analysis of variance was used as a parametric means to statistically evaluate data from in vitro experiments; these data were found to be normally distributed according to a Rankit plot. The in vitro data were also analyzed using Student's t test to compare two groups with variable sample sizes using Microsoft Excel version X software (Redmond, WA) as previously described ( 18 ). A parametric t test was used to determine whether the mean values measured for two distinct samples were equal. In vitro experiments (i.e., cell proliferation, wound healing, cell migration, tubule formation, and ELISA) were performed at least twice and each sample was assayed in triplicate; the results of those experiments were used in calculations of the mean values and 95% confidence intervals (CIs). Data from in vivo experiments were not normally distributed, according to the Rankit plots. Therefore, we used the non-parametric Wilcoxon–Mann–Whitney test in PRISM version 3 software (GraphPad, San Diego, CA) to test for statistically significant differences between groups. All statistical tests were two-sided.
Effect of Dz13 on c-Jun Protein Expression and DNA-Binding Activity and on Microvascular Endothelial Cell Proliferation
Mock-transfected, growth-arrested HMEC-1 cells (i.e., cells cultured in serum-free medium) expressed a low level of c-Jun relative to the constitutively expressed transcription factor Sp1 (Fig. 1, A ). The relative level of c-Jun expression increased approximately twofold within 2 hours after the cells were placed in serum-containing medium (cells in serum-free medium: 0.263 relative band intensity units, 95% CI = 0.250 to 0.281 relative band intensity units; cells in serum-containing medium: 0.662 relative band intensity units, 95% CI = 0.574 to 0.750 relative band intensity units) (Fig. 1, A ). HMEC-1 cells were then transfected with Dz13, a 10-23 DNAzyme with a 3′-3′–linked inverted thymidine residue [a modification that confers DNAzyme stability in serum-containing medium ( 19 ) ] that targets the G 1311 U junction in the human c-jun mRNA. Dz13-transfected HMEC-1 cells that were cultured in serum-containing medium displayed only modestly higher Sp1-normalized c-Jun protein expression than mock-transfected, growth-arrested HMEC-1 cells cultured in the absence of serum (Dz13-transfected cells in serum-containing medium: 0.320 relative band intensity units, 95% CI = 0.300 to 0.339 relative band intensity units) (Fig. 1, A ). By contrast, HMEC-1 cells transfected with Dz13scr, a DNAzyme that has the same length and net charge as Dz13 and retains an intact catalytic domain but differs from Dz13 in the order of nucleotides in its hybridizing arms, displayed stronger relative induction of c-Jun protein expression in the presence of serum (Dz13scr-transfected cells in serum-containing medium: 0.750 relative band intensity units, 95% CI = 0.697 to 0.803 relative band intensity units). Cells transfected with As13, the exact antisense oligonucleotide counterpart of Dz13 (including the 3′-3′ inverted thymidine residue but lacking the catalytic domain of Dz13), displayed some reduction of serum-inducible relative c-Jun protein expression compared with relative c-Jun levels in the presence of serum, albeit to a lesser extent than cells transfected with Dz13 (As13-transfected cells: 0.468 relative band intensity units, 95% CI = 0.435 to 0.501 relative band intensity units), whereas, compared with As13, cells transfected with As13scr displayed less reduction of serum-inducible relative c-Jun protein expression (0.833 relative band intensity units, 95% CI = 0.715 to 0.951 relative band intensity units) (Fig. 1, A ). None of the oligonucleotides tested altered the level of the zinc finger transcription factor Sp1 in transfected cells compared with the mock-transfected control cells, regardless of whether the transfected cells were cultured in serum-containing or serum-free medium (Fig. 1, A ). Moreover, results of trypan blue exclusion assays revealed that none of the treatments were toxic to the cells (data not shown).
To determine whether DNAzyme inhibition of relative c-Jun protein expression was associated with changes in the DNA-binding activity of c-Jun, we performed EMSAs using a 32 P-labeled oligonucleotide bearing a consensus binding element for c-Jun ( 32 P-Oligo c-Jun) and nuclear extracts made from quiescent HMEC-1 cells that were transfected with DNAzymes or mock-transfected and cultured in serum-containing medium. We observed a faint band (representing a nucleoprotein complex) when we used extracts made from mock-transfected cells cultured in serum-free medium (Fig. 1, B , left panel, lane 2). The intensity of this band was higher when we used nuclear extracts made from mock-transfected HMEC-1 cells exposed to serum-containing medium for 2 hours (Fig. 1, B , left panel, lane 3). Nucleoprotein complex formation was diminished either by prior transfection with Dz13 (Fig. 1, B , left panel, lane 4) or by pre-incubation of extracts of serum-exposed cells to c-Jun antibodies (Fig. 1, B , left panel, lane 6), suggesting that the nucleoprotein complex consisted of c-Jun and its target sequence. By contrast, complex formation was not affected by Dz13scr (Fig. 1, B , left panel, lane 5). The binding of an unrelated transcription factor, NF-1, to its target DNA sequence was not disrupted by transfection with either Dz13 or Dz13scr (Fig. 1, B , right panel) thus demonstrating unbiased loading. These findings suggest that a DNAzyme targeting c-jun mRNA can block both c-Jun protein expression and subsequent c-Jun DNA-binding activity in a sequence-specific manner.
We next examined the effect of DNAzymes on microvascular endothelial cell proliferation by quantitating cell numbers in growth-arrested HMEC-1 cells that had been transfected with DNAzymes at two different concentrations or mock-transfected with vehicle alone and incubated in serum-containing medium for 3 days. HMEC-1 cell proliferation in the presence of serum was inhibited in a concentration-dependent manner by transfection with Dz13 (0.1 μ M Dz13: 70.8% [95% CI = 63.0% to 78.5%] versus vehicle: 100% [95% CI = 96.3% to 103.7%], difference = 29.2%; P <.001; 0.2 μ M Dz13: 31.7% [95% CI = 29.7% to 33.7%] versus vehicle: 100% [95% CI = 96.4% to 103.6%], difference = 68.3%; P <.001]) (Fig. 2, A ). However, Dz13scr had only a modest influence on cell proliferation in comparison with the vehicle at both the 0.1 μ M and 0.2 μ M concentrations (0.1 μ M Dz13scr: 97.6% [95% CI = 92.0% to 103.0%] versus vehicle: 100% [95% CI = 96.3% to 103.7%], difference = 2.4%; P = .470; 0.2 μ M Dz13scr: 90.8% [95% CI = 84.0% to 97.6%] versus vehicle: 100% [95% CI = 96.4% to 103.6%], difference = 9.2%; P = .014) (Fig. 2, A ). As13, like Dz13, inhibited cell proliferation, but only when cells were transfected with the higher concentration of oligonucleotide (0.1 μ M As13: 103.7% [95% CI = 101.5% to 106.0%] versus vehicle: 100% [95% CI = 96.3% to 103.7%], difference = 3.7%; P = .125; 0.2 μ M As13: 59.3% [95% CI = 51.5% to 67.0%] versus vehicle: 100% [95% CI = 96.4% to 103.6%], difference = 40.7%; P <.001) (Fig. 2, A ).
A G-to-C substitution in the sixth nucleotide of the catalytic domain of 10-23 DNAzymes has been shown to abolish DNAzyme cleavage activity ( 20 ). Dz13M, a variant of Dz13 bearing this substitution, inhibited HMEC-1 cell proliferation with efficacy similar to that of As13 (0.1 μ M Dz13M: 95.5% [95% CI = 87.5% to 103.5%] versus vehicle: 100% [95% CI = 96.3% to 103.7%], difference = 4.5%; P = .333; 0.2 μ M Dz13M: 56.6% [95% CI = 54.0% to 59.0%] versus vehicle: 100% [95% CI = 96.4% to 103.6%], difference = 43.4%; P <.001). This similarity indicates, first, that the catalytic motif contributes to the inhibitory potency of Dz13 and, second, that the G-to-C substitution in Dz13 renders Dz13M as ineffective as As13 (Fig. 2, A ).
To examine the effect of hybridizing arm length on the biologic activity of Dz13 (which bears 9+9 nucleotide hybridizing arms), we evaluated proliferation among cells transfected with DNAzymes that contained 10+10 nucleotide arms (Dz13[10+10]), 11+11 nucleotide arms (Dz13[11+11]), or 8+8 nucleotide arms (Dz13[8+8]), each with a 3′-3′–linked inverted T, or their sequence-scrambled counterparts Dz13(10+10)scr, Dz13(11+11)scr, and Dz13(8+8)scr, respectively. When cells were transfected with high concentrations of DNAzymes (0.4 μ M ), Dz13(11+11) and Dz13(8+8) inhibited HMEC-1 cell proliferation as effectively as Dz13 (0.4 μ M Dz13: 18.8% [95% CI = 16.7% to 20.9%] versus vehicle: 100% [95% CI = 90.3% to 109.7%], difference = 81.2%; P <.001; 0.4 μ M Dz13[11+11]: 17.3% [95% CI = 14.0% to 20.8%] versus vehicle: 100% [95% CI = 90.3% to 109.7%], difference = 82.7%; P <.001; 0.4 μ M Dz13[8+8]: 19.2% [95% CI = 17.5% to 20.7%] versus vehicle: 100% [95% CI = 90.3% to 109.7%], difference = 80.8%; P <.001] (Fig. 2, B ). However, when these DNAzymes were transfected at lower concentrations (i.e., 0.2 μ M ), Dz13(11+11) and Dz13(8+8) were only slightly less potent inhibitors of endothelial cell proliferation than Dz13 (0.2 μ M Dz13: 22.4% [95% CI = 19.6% to 25.2%] versus vehicle: 100% [95% CI = 93.0% to 107.0%], difference = 77.6%; P <.001; 0.2 μ M Dz13[11+11]: 34.0% [95% CI = 31.6% to 36.4%] versus vehicle: 100% [95% CI = 93.0% to 107.0%], difference = 66.0%; P <.001; 0.2 μ M Dz13[8+8]: 31.9% [95% CI = 27.0% to 36.8%] versus vehicle: 100% [95% CI = 93.0% to 107.0%], difference = 68.1%; P <.001]) (Fig. 2, B ). Dz13(10+10) also inhibited HMEC-1 cell proliferation, albeit to a lesser extent than Dz13, Dz13(11+11), or Dz13(8+8) (0.4 μ M Dz13[10+10]: 49.4% [95% CI = 39.4% to 59.4%] versus vehicle: 100% [95% CI = 90.3% to 109.7%], difference = 50.6%; P <.001; 0.2 μ M Dz13[10+10]: 66.3% [95% CI = 51.6% to 70.9%] versus vehicle: 100% [95% CI = 93.0% to 107.0%], difference = 33.7%; P = .002]) (Fig. 2, B ).
Western blot analysis revealed that HMEC-1 cells transfected with 0.4 μ M Dz13, Dz13(11+11), Dz13(8+8), or As13 and cultured in serum-containing medium expressed lower levels of c-Jun relative to Sp1 than HMEC-1 cells transfected with their sequence-scrambled counterparts Dz13scr, Dz13(11+11)scr, Dz13(8+8)scr, or As13scr (Dz13: 0.347 relative band intensity units [95% CI = 0.330 to 0.364 relative band intensity units]; Dz13scr: 1.240 relative band intensity units [95% CI = 1.223 to 1.309 relative band intensity units]; Dz13[11+11]: 0.580 relative band intensity units [95% CI = 0.560 to 0.599 relative band intensity units]; Dz13[11+11]scr: 1.340 relative band intensity units [95% CI = 1.242 to 1.438 relative band intensity units]; Dz13[8+8]: 0.699 relative band intensity units [95% CI = 0.601 to 0.797 relative band intensity units]; Dz13[8+8]scr: 1.280 relative band intensity units [95% CI = 1.258 to 1.302 relative band intensity units]; As13: 0.630 relative band intensity units [95% CI = 0.606 to 0.653 relative band intensity units]; As13scr: 1.083 relative band intensity units [95% CI = 0.993 to 1.173 relative band intensity units]) (Fig. 2, C ).
We found that Dz13(11+11) and Dz13(8+8), like Dz13, cleaved a 40-nucleotide 32 P-labeled synthetic RNA substrate in a time-dependent manner, whereas Dz13(11+11)scr and Dz13(8+8)scr were catalytically inactive (data not shown) . Dz13(10+10) was as catalytically active as Dz13 (data not shown) but differed from Dz13 in its capacity to inhibit HMEC-1 proliferation (Dz13 versus Dz13[10+10] at 0.4 μ M: 18.8% [95% CI = 16.7% to 20.9%] versus 49.4% [95% CI = 39.4% to 59.4%]) (Fig. 2, B ) and c-Jun protein expression (Dz13 versus Dz13[10+10] at 0.4 μ M: 0.347 relative band intensity units [95% CI = 0.330 to 0.364 relative band intensity units] versus 1.18 relative band intensity units [95% CI = 1.07 to 1.29 relative band intensity units]) (Fig. 2, C ). These data indicate that cleavage alone is not a reliable performance indicator of DNAzyme efficacy in a biologic system.
Effect of Dz13 on Microvascular Endothelial Cell Migration
We used two different in vitro assays to examine the role of c-Jun in microvascular endothelial cell migration. We subjected monolayers of HMEC-1 cells that were either transfected with DNAzymes or mock-transfected to a wound healing assay and quantitated the number of cells that had migrated into the denuded (i.e., wounded) zone after incubation for 2 days in serum-containing medium. Dz13 inhibited this reparative response to injury in a dose-dependent manner (0.2 μ M Dz13: 81.8% [95% CI = 75.3% to 88.3%] versus vehicle: 100% [95% CI = 92.7% to 107.3%], difference = 18.2%; P = .023; 0.3 μ M Dz13: 51.9% [95% CI = 47.5% to 59.3%] versus vehicle: 100% [95% CI = 92.7% to 107.3%], difference = 48.1%; P <.001; 0.4 μ M Dz13: 22.2% [95% CI = 18.2% to 26.2%] versus vehicle: 100% [95% CI = 92.7% to 107.3%], difference = 77.8%; P <.001) (Fig. 3, A and B ). These findings were confirmed by performing a cell migration and invasion assay that used modified Boyden chambers containing membranes coated with a reconstituted basement membrane (Matrigel). Microvascular endothelial cell invasion through the Matrigel to the underside of the membrane was blocked by Dz13 (Dz13: 44.1% [95% CI = 35.4% to 52.7%] versus vehicle: 100% [95% CI = 90.0% to 110%], difference = 55.9%; P <.001) but not by Dz13scr (Dz13scr: 80.7% [95% CI = 69.0% to 92.5%] versus vehicle: 100% [95% CI = 90.0% to 110%], difference = 19.3%; P = .232) (Fig. 3, C ). Cell migration through filters coated with collagen type I was also inhibited by Dz13 (Dz13: 55.2% [95% CI = 46.2% to 64.1%] versus vehicle: 100% [95% CI = 83.1% to 116.9%], difference = 44.8%; P = .037) but not by Dz13scr (Dz13scr: 128% [95% CI = 118% to 138%] versus vehicle: 100% [95% CI = 83.1% to 116.9%], difference = 28%; P = .175) (Fig. 3, C ).
Effect of Dz13 on Microvascular Endothelial Cell Tubule Formation
In addition to cell proliferation and migration, the complex process of angiogenesis involves the ordered assembly and alignment of endothelial cells. In vitro, endothelial cells spontaneously align and form a three-dimensional tubular capillary-like network within hours of plating on Matrigel ( 21 ). Dz13 blocked tubule formation on Matrigel in a dose-dependent manner (10 n M Dz13: 100.9% [95% CI = 88.8% to 113%] versus vehicle: 100% [95% CI = 87.5% to 112.5%], difference = 0.9%; P = .797; 30 n M Dz13: 88.2% [95% CI = 78.6% to 97.8%] versus vehicle: 92.7% [95% CI = 80% to 105.5%], difference = 4.5%; P = .683; 60 n M Dz13: 59.5% [95% CI = 51.7% to 67.3%] versus vehicle: 94.8% [95% CI = 78.0% to 111.7%], difference = 35.3%; P = .003; 0.1 μ M Dz13: 39.8% [95% CI = 33.3% to 46.3%] versus vehicle: 103.6% [95% CI = 95.4% to 111.8%], difference = 63.8%; P <.001) (Fig. 4, A and B ). Endothelial cell tube formation was not affected by Dz13scr at any concentration tested (10 n M Dz13scr: 108.4% [95% CI = 94.7% to 122%] versus vehicle: 100% [95% CI = 87.5% to 112.5%], difference = 8.4%; P = .300; 30 n M Dz13scr: 106% [95% CI = 95.6% to 116.5%] versus vehicle: 92.7% [95% CI = 80.0% to 105.5%], difference = 13.3%; P = .100; 60 n M Dz13scr: 100% [95% CI = 86.4% to 113.4%] versus vehicle: 94.8% [95% CI = 78.0% to 111.7%], difference = 5.2%; P = .658; 0.1 μ M Dz13scr: 103% [95% CI = 92.3% to 113.7%] versus vehicle: 103.6% [95% CI = 95.4% to 111.8%], difference = 0.6%; P = .618]) (Fig. 4, A and B ). These results suggest that c-Jun is required for endothelial cell tubule formation in vitro.
Effect of Dz13 on HMEC-1 MMP-2 mRNA and Protein Expression and MMP-2 Proteolytic Activity
MMP-2 transcription is induced by c-Jun in non-endothelial cells ( 22 ). We hypothesized that Dz13 inhibition of endothelial cell growth might, at least in part, be mediated by its inhibition of c-Jun expression which, in turn, would inhibit MMP-2 mRNA expression. Compared with mock-transfected HMEC-1 cells, cells transfected with Dz13 had lower levels of MMP-2 mRNA (Dz13: 1.22 relative band intensity units [95% CI = 1.07 to 1.37 relative band intensity units] versus vehicle: 2.05 relative band intensity units [95% CI = 1.95 to 2.15 relative band intensity units], difference = 0.83; P = .018) (Fig. 5, A ) and MMP-2 protein (Dz13: 1.19 ng/mL [95% CI = 1.02 to 1.36 ng/mL] versus vehicle: 2.60 ng/mL [95% CI = 2.54 to 2.65 ng/mL], difference = 1.41 ng/mL; P = .025) (Fig. 5, B ). By contrast, HMEC-1 cells transfected with Dz13scr had levels of MMP-2 mRNA and protein similar to those in mock-transfected cells (Fig. 5, A and B ). Gelatin zymography analysis of MMP-2 activity secreted into the culture medium revealed that cells transfected with Dz13 had statistically significantly less MMP-2 proteolysis (i.e., metalloproteinase activity) of gelatin than mock-transfected cells (Dz13: 53.5% [95% CI = 50.0% to 57.0%] versus vehicle: 100% [95% CI = 94.2% to 105.8%], difference = 46.5%; P = .011) (Fig. 5, C ). By contrast, cells co-transfected with Dz13 and the c-Jun cDNA had a level of MMP-2 proteolytic activity similar to that of mock-transfected cells (Dz13/c-Jun: 91.8% [95% CI = 80.3% to 103.3%] versus vehicle: 100% [95% CI = 94.3% to 105.8%], difference = 8.2%; P = .308) (Fig. 5, C ). MMP-2 activity in cells transfected with Dz13scr was similar to that of mock-transfected cells (Fig. 5, C ). Thus, we conclude that Dz13 inhibits MMP-2 expression in human endothelial cells in a c-Jun-dependent manner.
Effect of Dz13 on VEGF 165 -Induced Neovascularization in Rat Cornea
We next examined the effect of Dz13 on angiogenesis in a rat model of corneal neovascularization, a process that requires MMP-2 ( 23 ). In this model, implantation of VEGF 165 -soaked disks in the normally avascular rat cornea stimulates new blood vessel growth from the limbus toward the implant within 5 days ( 24 ). We first performed experiments to examine the effect of VEGF 165 on endothelial c-Jun expression and the effect of Dz13 on VEGF 165 -inducible tubule formation. Western blot analysis demonstrated that VEGF 165 induced c-Jun expression in mock-transfected HMEC-1 cells within 2 hours of its addition to serum-free medium (serum-free medium versus serum-free medium containing VEGF 165 : 0.140 relative band intensity units [95% CI = 0.135 to 0.145 relative band intensity units] versus 0.234 relative band intensity units [95% CI = 0.223 to 0.245 relative band intensity units]) (Fig. 6, A ). Compared with cells exposed to VEGF 165 alone, HMEC-1 cells transfected with Dz13 and exposed to VEGF 165 had suppressed c-Jun levels (cells transfected with Dz13 and exposed to serum-free medium containing VEGF 165 : 0.144 relative band intensity units [95% CI = 0.141 to 0.147 relative band intensity units] versus mock-transfected cells exposed to serum-free medium containing VEGF 165 : 0.234 relative band intensity units [95% CI = 0.223 to 0.245 relative band intensity units]) (Fig. 6, A ). By contrast, Dz13scr had no influence on c-Jun expression in cells incubated with VEGF 165 as compared with untransfected cells exposed to VEGF 165 (cells transfected with Dz13scr and exposed to serum-free medium containing VEGF 165 : 0.233 relative band intensity units [95% CI = 0.223 to 0.243 relative band intensity units] versus mock-transfected cells exposed to serum-free medium containing VEGF 165 : 0.234 relative band intensity units [95% CI = 0.222 to 0.245 relative band intensity units]) (Fig. 6, A ).
Dz13 also inhibited VEGF 165 -inducible HMEC-1 tubule formation in vitro compared with the agonist-free mock-transfected control (Dz13-transfected cells exposed to VEGF 165 : 5.5% [95% CI = 4.6% to 6.4%] versus mock-transfected cells exposed to VEGF 165 : 100% [95% CI = 47.4% to 152.6%], difference = 94.5%; P = .039) (Fig. 6, B ). Results of these in vitro studies demonstrate that Dz13 inhibits VEGF 165 -inducible c-Jun expression and tubule formation.
Slit lamp biomicroscopic visualization of the corneas revealed that those implanted with VEGF 165 -soaked disks and injected with Dz13 had less neovascularization than corneas implanted with VEGF 165 -soaked disks and injected with vehicle (Fig. 6, C ) following assessment of both the number of new blood vessels (Dz13: 4.0 vessels [95% CI = 0 to 8.5 vessels] versus vehicle: 30.7 vessels [95% CI = 21.9 to 39.5 vessels], difference = 26.7 vessels; P = .004) and the area occupied by new blood vessels (Dz13: 0.35 mm 2 [95% CI = 0.00 to 0.74 mm 2 ] versus vehicle: 1.52 mm 2 [95% CI = 1.15 to 1.90 mm 2 ], difference = 1.17 mm 2 ; P = .005) (Fig. 6, D and E ).
Effect of Dz13 on the Growth of Solid Melanomas in Mice
Aggressive growth of human melanoma is associated with a substantial increase in blood vessel density ( 25 ). Immunohistochemical analysis of a primary human cutaneous malignant melanoma revealed that c-Jun is strongly expressed in the CD31-positive endothelial cells that colonized the tumor as well as in surrounding melanoma cells (Fig. 7, A ). Intense cytoplasmic staining in both cell types was also apparent with antibodies to MMP-2 (Fig. 7, A ). This dependence on vascularization for aggressive melanoma growth has also been observed in mice. For example, the in vivo growth of solid murine-derived B16 melanomas is blocked by monoclonal antibodies to the VEGF receptor Flk-1 ( 26 ) and by MMP-2 inhibitors ( 27 , 28 ). These observations led us to hypothesize that melanoma growth in mice might be suppressed by a DNAzyme targeting c-Jun, particularly in light of the complementarity of Dz13 target sites in human and mouse c-jun RNA ( 17 ). Compared with c-Jun protein levels in mock-transfected murine bEND-3 cells exposed to serum, c-Jun expression was strongly inhibited in cells transfected with Dz13 (mock-transfected cells in serum-free medium: 0.184 relative band intensity units [95% CI = 0.172 to 0.196 relative band intensity units]; mock-transfected cells in serum-containing medium: 0.350 relative band intensity units [95% CI = 0.339 to 0.361 relative band intensity units]; Dz13 in serum-containing medium: 0.205 relative band intensity units [95% CI = 0.195 to 0.215 relative band intensity units]; Dz13scr in serum-containing medium: 0.358 relative band intensity units [95% CI = 0.344 to 0.372 relative band intensity units]) (Fig. 7, B ). Moreover, serum-inducible bEND-3 cell proliferation was suppressed by treatment with Dz13 compared with treatment with Dz13scr (Dz13: 31.9% [95% CI = 28.2% to 35.5%] versus vehicle: 100% [95% CI = 89.2% to 110.8%], difference = 68.1%; P <.001; Dz13scr: 87.9% [95% CI = 68.1% to 107.7%] versus vehicle: 100% [95% CI = 89.2% to 110.8%], difference = 12.1%; P = .326) (Fig. 7, C ). These findings demonstrate that murine microvascular growth could be blocked by a DNAzyme targeting c-Jun.
Dz13 inhibited solid B16 melanoma growth in C57BL/J6 mice in both a time-dependent and a sequence-specific manner (Fig. 7, D ). Fourteen days following tumor implantation, the c-Jun DNAzyme inhibited tumor growth by more than 60% compared with tumor growth in the vehicle-treated mice (Dz13: 108 mm 3 [95% CI = 56 to 160 mm 3 ] versus vehicle: 283 mm 3 [95% CI = 211 to 356 mm 3 ], difference = 175 mm 3 ; P = .006) (Fig. 7, D ). By contrast, Dz13scr-, As13-, or As13scr-treated mice had tumor volumes that were indistinguishable from those of the vehicle-treated mice (Dz13scr: 306 mm 3 [95% CI = 243 to 369 mm 3 ] versus vehicle: 283 mm 3 [95% CI = 211 to 355 mm 3 ], difference = 23 mm 3 ; P = .651; As13: 253 mm 3 [95% CI = 208 to 298 mm 3 ] versus vehicle: 283 mm 3 [95% CI = 211 to 355 mm 3 ], difference = 30 mm 3 ; P = .512; As13scr: 262 mm 3 [95% CI = 222 to 302 mm 3 ] versus vehicle: 283 mm 3 [95% CI = 211 to 355 mm 3 ], difference = 21 mm 3 ; P = .628) (Fig. 7, E ). None of the treatments was associated with a change in mean body weight compared with that of mice treated with vehicle control (Fig. 7, E ) or other symptoms (e.g., lethargy, ruffled fur, skin erythema, or soft feces) that would suggest the treatments were toxic. Tumor vascular density, assessed by quantitating the number of blood vessels in tumor cross-sections, was strongly inhibited by Dz13 at 14 days (Dz13: 30.0% [95% CI = 26.4% to 33.6%] versus vehicle: 100% [95% CI = 77.1% to 122.9%], difference = 70.0%; P <.001) (Fig. 7, F ). Thus, a DNAzyme that inhibited c-Jun protein expression and mouse microvascular endothelial cell proliferation in vitro also inhibited the growth of solid tumors in vivo as well as angiogenesis associated with the tumors.
Treatment strategies that target specific genes in a complex biologic milieu may be achieved by using synthetic agents such as ribozymes, antisense oligonucleotides, RNA interference, or DNAzymes ( 29 ). DNAzymes are versatile tools that can tease out the precise functions of the targeted gene in a variety of cellular processes ( 30 ) and that offer the advantages of catalytic activity with inherent stability and low cost of synthesis. We have previously shown that DNAzymes targeting Egr-1 and c-Jun can inhibit restenosis and/or in-stent restenosis, both of which involve vascular smooth muscle cell hyperplasia ( 17 , 19 , 31 , 32 ). Studies by other investigators have demonstrated several nonvascular applications of DNAzymes. For example, Iversen et al. ( 33 ) studied the effects of DNAzymes targeting tumor necrosis factor-alpha on hemodynamic performance in a rat model of post-infarction congestive heart failure. Grimpe et al. ( 34 ) used DNAzymes to demonstrate a key role for laminin in axon regeneration in the central nervous system.
Here we have shown that DNAzyme inhibition of c-jun expression suppresses microvascular endothelial cell proliferation, migration, and tubule formation in vitro. DNAzymes targeting c-jun also inhibited corneal neovascularization in rats and solid melanoma growth in mice. In each case, we established the sequence-specificity of inhibition by demonstrating that the sequence-scrambled counterpart of the active DNAzyme was inactive. DNAzyme inhibition of c-Jun suppressed MMP-2 expression and activity. MMPs cleave basement membrane and extracellular matrix molecules and are key to the process of angiogenesis ( 35 ). For example, mice deficient in MMP-2 (also known as gelatinase A) have compromised tumor-inducible angiogenesis and tumor progression ( 36 ).
There is presently no consensus on optimal arm length for DNAzymes. We report here that Dz13, which has 9+9 nucleotide hybridizing arms that confer target specificity, produced the most efficient inhibition of both c-Jun protein expression and endothelial cell proliferation of all the DNAzymes we tested. Design factors that influence the affinity and cleavage efficiency of a DNAzyme for its target mRNA (e.g., the sequence targeted and the length of the hybridizing arms) are poorly understood. We made an effort to choose target sequences that are likely to be located on the exterior of the folded mRNA (based on calculations of free energy). Target sequences that are located toward the 5′ end of the molecule may offer the advantage of inherent accessibility for the protein synthetic machinery to bind and translocate ( 37 , 38 ). Our observations comparing, for example, Dz13(9+9), Dz13(10+10) and Dz13(11+11), indicate that DNAzyme design, at least at the present time, needs to be evaluated on a case-by-case basis by examining the activities of a family of candidate molecules. We have, nonetheless, successfully used DNAzymes bearing 9+9 nucleotide arms in multiple other in vitro and in vivo applications, including those that establish positive regulatory roles for Egr-1 and c-Jun in intimal thickening in injured arteries. For example, we recently used Egr-1 DNAzymes to demonstrate that Egr-1–dependent angiogenesis involves endogenous FGF-2 (which is itself transcriptionally dependent upon Egr-1) but not VEGF ( 19 , 21 , 31 , 32 ). Our present findings provide the first direct evidence that c-Jun is a key mediator of angiogenesis and suggest the potential clinical utility of DNAzymes specifically targeting this bZIP protein in pathologies involving angiogenesis.