Discovery of a Novel Control Element within the 5′-Untranslated Region of the Vascular Endothelial Growth Factor

The regulation of vascular endothelial growth factor (VEGF), a potent stimulator of angiogenesis, is controlled primarily through the interactions of control elements located within the 5′- and 3′-untranslated regions, many of which are yet to be described. In this study we examined the 5′-untranslated region of human VEGF for control elements with the aim of regulating expression both in vitro and in vivo using oligonucleotide gene therapy. A potential control element was located, two sense oligonucleotides (S1 and S2) were designed based on its sequence, and a third oligonucleotide (S3) was designed as a control and mapped to the 16 base pairs immediately upstream. Retinal cells cultured in the presence of S1 and S2 resulted in a 2-fold increase of VEGF protein and a 1.5-fold increase in mRNA 24 h post-transfection whereas S3 had no significant effect (p > 0.05) compared with controls. Subsequent reporter gene studies confirmed the necessity of this element for up-regulation by S1. Further in vivo studies showed that S1 and S2 mediated an increase in VEGF protein in a rodent ocular model that resulted in angiogenesis. In addition to providing insight into the regulation of the vascular endothelial growth factor, the use of these oligonucleotides to stimulate vascular growth may prove useful for the treatment of ischemic tissues such as those found in the heart following infarct.

The regulation of vascular endothelial growth factor (VEGF), a potent stimulator of angiogenesis, is controlled primarily through the interactions of control elements located within the 5-and 3-untranslated regions, many of which are yet to be described. In this study we examined the 5-untranslated region of human VEGF for control elements with the aim of regulating expression both in vitro and in vivo using oligonucleotide gene therapy. A potential control element was located, two sense oligonucleotides (S 1 and S 2 ) were designed based on its sequence, and a third oligonucleotide (S 3 ) was designed as a control and mapped to the 16 base pairs immediately upstream. Retinal cells cultured in the presence of S 1 and S 2 resulted in a 2-fold increase of VEGF protein and a 1.5-fold increase in mRNA 24 h post-transfection whereas S 3 had no significant effect (p > 0.05) compared with controls. Subsequent reporter gene studies confirmed the necessity of this element for up-regulation by S 1 . Further in vivo studies showed that S 1 and S 2 mediated an increase in VEGF protein in a rodent ocular model that resulted in angiogenesis. In addition to providing insight into the regulation of the vascular endothelial growth factor, the use of these oligonucleotides to stimulate vascular growth may prove useful for the treatment of ischemic tissues such as those found in the heart following infarct.
Vascular development is a fundamental requirement for all tissue growth, and the absence of adequate tissue vascularization results in cells becoming deprived of oxygen and nutrients. This fact provides the stimulus for cells to produce angiogenic factors, which function to recruit new blood vessels into the deprived tissue. The most important of the angiogenic factors involved in new blood vessel formation is vascular endothelial growth factor (VEGF), 1 which is highly regulated and consists of four isoforms encoded by a single gene via alternate splicing (1,2). A characteristic of all four isoforms is the presence of an unusually long and GC-rich 5Ј-and 3Ј-untranslated region (UTR) (1,2) that contains most of the important control and regulatory elements involved in the modulation of VEGF ex-pression (reviews in Refs. 3 and 4). These elements include several internal ribosomal entry sites (5,6), hypoxia response elements (7), and a number of stabilizing and destabilizing sequences (8,9).
The importance of VEGF-mediated vascularization in disease states makes it an attractive target for gene therapies. Several methods of down-regulating VEGF for the treatment of tumors and ocular neovascularization are currently being explored (10 -12). In addition, we have previously described a sense oligonucleotide (DS-085) that targets the 5Ј-UTR of the VEGF gene and has proven effective at down-regulating the transcription and subsequent translation of VEGF both in vitro and in vivo (13). The mechanism of action has been postulated to be due to Hoogsteen hydrogen bonding of the oligonucleotide (ODN) within the major groove of the duplex DNA, causing polymerase arrest (14 -18). Similar to the regulatory regions of other genes, DS-085 was found to be rich in GA purine residues (19,20). An examination of the 5Ј-UTR sequence was therefore made in an attempt to discover other potential homopurine regulatory sequences involved in VEGF expression. In this article we report on the discovery of a novel control element within the 5Ј-UTR of the VEGF gene that may represent the binding site of a destabilization protein.

EXPERIMENTAL PROCEDURES
Oligonucleotide Design-The 5Ј-UTR sequence of human VEGF (GenBank TM accession number NM_003376) was examined for the presence of homopurine regions that may represent potential regulatory sites. Sense ODNs 1 and 2 (S 1 and S 2 ) were subsequently designed to recognize the first and final 16 bp, respectively, of a homopurinehomopyrimidine sequence identified from base pair Ϫ265 to Ϫ223 from the ATG start codon. In addition, a third sense ODN (S 3 ) was designed and represented the 16 bp immediately 5Ј to S 1 and was used as a control. Oligo DS-085 has been described previously (13). Oligonucleotides were obtained from Proligo (Boulder, CO) and synthesized with a phosphorothioate (S) backbone.
In Vitro Oligonucleotide VEGF Inhibition Assay-A human retinal pigment epithelial (RPE) cell line (RPE 51) was grown in culture and used to assess the effect that S 1 , S 2 , S 3 , and DS-085 have on the production of VEGF protein and mRNA. Cells were seeded into 2 ϫ 6 well plates (35-mm diameter) at ϳ4 ϫ 10 5 cells per well and allowed to grow in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 0.5% streptomycin, and penicillin at 37°C and 5% CO 2 until 80% confluence was reached. Cytofectin (Gene Therapy Systems, San Diego, CA) was used per the manufacturer's instructions to deliver the ODNs into the cells at a final concentration of 1 M. Control groups consisted of cells transfected with cytofectin alone and null treated cells, which were not manipulated in any way. Following transfection, one of the plates was transferred to a CO 2 incubator and grown under normoxic condition at 5% CO 2 . The other plate was placed in a hypoxic incubator (2% O 2 and 5% CO 2 ), and each was grown for 24 h. After this time, the media were collected from both the normoxic and hypoxic grown cells for an enzyme-linked immunosorbent assay (ELISA) using a commercially available kit (CYTELISA™; Cytimmune Sciences, College Park, MD) to determine the level of VEGF protein expression. The ELISA was performed per the manufacturer's instructions using 100 l of undiluted culture media. The cells from each well were harvested by trypsinization and pelleted by centrifugation at 2000 ϫ g for 5 min. The cell pellet was washed twice in isotonic saline and resuspended in 250 l of the same. An A 600 reading was taken to determine cell density, which was used to normalize the VEGF concentration.
Levels of mRNA transcript were determined using RT-PCR. Cells were treated with 600 l of Trizol (Qiagen, Clifton Hill, Victoria, Australia) directly in the culture wells 24 h post-transfection. The lysed suspension was transferred to a microfuge tube where chloroform (200 l) was added, and the solution was subjected to a vortex to ensure complete mixing. The aqueous phase containing the total RNA was removed to a new microfuge tube, and 1 volume of isopropanol was added to precipitate the total RNA. The total RNA was pelleted by centrifugation at 20,000 ϫ g for 10 min, and the pellet was washed in 70% ethanol. The ethanol was aspirated, the pellet was air-dried and resuspended in 200 l of nuclease-free water, and the concentration was determined spectrometrically. An Omniscript™ RT kit (Qiagen) was used for the production of the first strand cDNA per the manufacturer's instructions starting with 200 ng of total RNA using an oligo(dT) primer in a final volume of 20 l. Directly from this reaction, 1 l was used as a template for the PCR of an internal VEGF fragment in addition to a ␤-actin fragment, which was used as an internal control. VEGF primers consisted of the sense 5Ј-CATCACGAAGTGGTGAAGT-T-3Ј and the antisense 5Ј-AACGCTCCAGGACTTATACC-3Ј. Primers used to amplify ␤-actin (GenBank TM accession number NM_007393) consisted of the sense 5Ј-AGGCACCAGGGCGTGAT-3Ј and the antisense 5Ј-TTAATGTCACGCACGATTTC-3Ј. Both sets of primers (Proligo) were included with the following reaction components in a final volume of 25 l: 2.5 l of 10ϫ reaction buffer, 2 mM MgCl 2 , 200 M each dNTP, 6 pmol of each primer, and 1 unit of Tth ϩ polymerase (Fisher Biotech, Perth, Western Australia, Australia). A touchdown cycling reaction was used and consisted of an initial denaturing step of 94°C for 2 min followed by seven cycles of 94°C for 10 s, 65°C for 10 s with a drop of 1°C per cycle, and 72°C for 30 s. This was then followed by 41 cycles of 94°C for 10 s, 58°C for 10 s, and 71°C for 30 s.
Sample Statistics-The transfections were performed in quadruplet sets for statistical analysis. Results were analyzed by one-way analysis of variance followed by a post hoc Fisher's least squares difference analysis with 95% confidence limits using the GB-Stat TM statistical software package (Dynamic Microsystems, Silver Springs, MD).
Reporter Gene Analysis-Plasmids were constructed to examine the role of various sections of the 5Ј-UTR in regulating VEGF expression both in the presence and absence of up-regulating ODNs. Initially, the entire 1039 bp of the human VEGF 5Ј-UTR was RT-PCR amplified from mRNA extracted from cultured human RPE 51 cells and subcloned into pGEM T Easy (Promega, Madison, WI) to produce pGEM T-UTR. The integrity of this clone was confirmed by DNA sequencing. Subsequently, the entire UTR was removed by digestion with ApaI and SalI (all restriction enzymes were sourced from New England Biolabs, Beverly, MA) and subcloned into the ApaI and XhoI site of the reporter plasmid (p⌬UTR) consisting of the cytomegalovirus strong promoter linked to the secreted alkaline phosphatase reporter gene to produce pUTR-W. Two further reporter plasmids were constructed by subcloning the proximal 703-bp ApaI-XmnI fragment and the distal 396-bp XmnI-EcoRI fragment of pGEM T-UTR into the ApaI-EcoRV and the EcoRV-EcoRI sites of p⌬UTR, respectively, to produce pUTR-L and pUTR-S.
To examine the effect on reporter gene activity, 0.8 g of each construct was used per well to transiently transfect RPE 51 cells cultured in 24-well plates using the transfection agent LipofectAMINE (Invitrogen) per the manufacturer's instructions. Transfection was allowed to proceed for 5 h, at which time the growth media were replaced with fresh media both with and without supplementation with a 1 M concentration of S 1 conjugated to the transfection agent cytofectin (Gene Therapy Systems). Media were sampled 24 h later and analyzed for secreted alkaline phosphatase activity by chemiluminescent detection on a TD 20/20 luminometer (Turner Designs, Sunnyvale, CA). Transfections were performed in quadruplet sets, and the results were expressed as a percentage of relative light units compared with the p⌬UTR control. Analysis of the results consisted of analysis of variance with post hoc Fisher's least squares difference analysis with 95% confidence limits.
Injections and in Vivo Analysis-All animal experiments were performed in accordance with the animal use guidelines of the Association for Research in Vision and Ophthalmology and were approved by the Animal Ethics Committee of the University of Western Australia. Oligonucleotide delivery to the anterior chamber was carried out on 6 -8-week-old non-pigmented RCS/rdy ϩ rats that had been anesthetized by an intramuscular injection of ketamine (50 mg kg Ϫ1 body weight) and xylazine (8 mg kg Ϫ1 body weight), followed by topical application of proparacaine hydrochloride to the eye. Two and one-half microliters of a 1 mM oligonucleotide solution or vehicle (phosphate-buffered saline containing 10% glycerol) were injected into the anterior chamber of both eyes of each rat via the temporal limbus using a 32-gauge needle attached to a 5-l Hamilton syringe after the same amount of aqueous humor was drained. Ophthalmologic examinations of the eyes were performed 7 days post-injection and photographed using a slit lamp camera.
Sub retinal injections were performed on 8-to 9-week-old non-pigmented RCS/rdy ϩ rats and C57 BL/6J mice of the same age. The injection technique used has been described previously (21). Briefly, the conjunctiva was cut close to the limbus to expose the sclera, which was then punctured with a 30-guage needle. A 32-guage needle was passed through this hole in a tangential direction under an operating microscope. Two microliters of oligonucleotide were delivered into the subretinal space of each eye. The needle was kept in the subretinal space for 1 min, withdrawn gently, and antibiotic ointment was applied to the wound site. Ophthalmologic examinations were performed 7 days postinjection and consisted of color fundus photography (CFP) and fluorescein angiography (FA).
A second group of similarly injected rats were euthanized 7 days post-injection, and the eyes were enucleated and placed into 200 l of phosphate buffered saline containing protease inhibitors. The eyes were thoroughly homogenized and centrifuged at 3000 ϫ g for 15 min at 4°C. To determine the concentration of VEGF in the eyes, 50 l of supernatant was used in an ELISA specific for mouse/rat VEGF (Quantikine; R&D Systems). The concentration was normalized against total protein concentrations of the supernatant. ELISA results were analyzed using analysis of variance with a post hoc Dunnnett's procedure (GB-Stat™).

RESULTS
Regulation of VEGF Expression-S 1 , S 2 , S 3 , and DS-085 were transfected into the RPE 51 cell line, and the effects of VEGF translation and transcription were measured using ELISA and RT-PCR, respectively. ELISA (Fig. 1) revealed that both S 1 (1073 Ϯ 64 pg ml Ϫ1 ) and S 2 (969 Ϯ 60 pg ml Ϫ1 ) facilitated a statistically significant (p Ͻ 0.01) up-regulation of VEGF protein by ϳ2-fold as compared with the non-transfected control (578 Ϯ 63 pg ml Ϫ1 ). ODN S 3 (593 Ϯ 33 pg ml Ϫ1 ) had no significant effect (p Ͼ 0.05), whereas the transfection agent cytofectin mediated a slight decrease in VEGF expression (508 Ϯ 38 pg ml Ϫ1 ). However, this result was not found to be significant (p Ͼ 0.05).
To examine the effects on VEGF at the transcriptional level, total RNA was extracted from cells transfected with S 1 , S 2 , S 3 , and DS-085 and subsequently used as a template for RT-PCR (Fig. 2a). The profile for mRNA levels in the transfected cells, as determined by densitometry of the PCR products (Fig. 2b), reflected the protein concentration profile. Transfection with S 1 and S 2 mediated an increase in the levels of VEGF mRNA by a factor of 1.5 as compared with the non-transfected control 24 h after transfection. However, the increase in mRNA mediated by S 1 and S 2 was not found to be proportional to the increase in protein concentration. The previously described oligonucleotide DS-085 decreased the level of mRNA by 57.5%, which was directly proportional to the decrease in protein. This result indicated that the mechanism of down-regulation by DS-085 was separate to and distinct from the mechanism of up-regulation of protein by S 1 and S 2 . Transfection with the S 3 oligonucleotide resulted in no significant effect as compared with the effect on the control samples, which was the same in regard to protein concentration. Similarly, transfection with vehicle (cytofectin) alone produced a slight decrease (5%) in VEGF mRNA equivalent to that found for the protein reduction and may be reflective of the slight cytotoxic effect known to be associated with cytofectin (21).
Reporter Gene Analysis-Retinal cells were transiently transfected with the reporter gene constructs and cultured in both the presence and the absence of S 1 . Media were sampled and tested for alkaline phosphatase activity using a chemiluminescent detection method, and the results are summarized in Fig. 3. Plasmid p⌬UTR showed strong reporter gene activity and remained unaffected when cultured in the presence of S 1 (data not shown). Transfection with plasmid pUTR-W, which contains the entire VEGF 5Ј-UTR, resulted in a significant (p Ͻ 0.01) decrease (65%) in reporter gene activity compared with ␤-actin is a stably expressing gene and was used as an internal control to normalize the levels of VEGF mRNA. Cyt., cytofectin. Densitometry performed on the gel photograph was used to determine the relative quantities of mRNA, which were expressed as a percentage of the control sample (panel B). The VEGF mRNA levels were elevated up to 1.5-fold when transfected with S 1 and S 2 , whereas S 3 mediated no significant effect. p⌬UTR. However, when cultured in the presence of S 1 , reporter gene activity significantly (p Ͻ 0.01) increased 1.79-fold to be 62.9 Ϯ 5.4% of p⌬UTR activity. When transfected with pUTR-L in which the proposed destabilizing element was absent, the decrease in reporter activity as compared with p⌬UTR was significantly less (p Ͻ 0.01) than that of pUTR-W, indicating a more stable transcript. In addition, we only see a small increase in reporter activity when cultured in the presence of S 1 (63.5 Ϯ 4.3% to 73.9 Ϯ 3.3% compared with p⌬UTR), which was not found to be significant (p Ͼ 0.05). In cells transfected with pUTR-S, which comprises the 3Ј-distal end of the VEGF 5Ј-UTR and contains the destabilizing element, we again see reporter gene activity significantly (p Ͻ 0.01) increased by 1.39-fold when cultured in the presence of S 1 , indicating a stabilizing effect.
In Vivo Analysis-To determine whether the in vitro observations would translate into an in vivo effect, S 1 , S 2 , and S 3 were injected into the anterior chamber of rat eyes. Subsequent ophthalmologic examination showed strong neovascularization in the iris of rat eyes 7 days following injection with S 1 and S 2 , (Fig. 4), but no effect was observed in rat eyes injected with S 3 or vehicle. This indicates that the oligonucleotides were able to mediate the up-regulation of VEGF in the eye and produce an angiogenic response.
The result observed in the iris was reflected in rats that were injected in the subretinal space with S 1 , S 2 , and S 3 . Eyes injected with S 3 remained clear of angiogenesis when viewed using CFP and FA (Fig. 5, a and b, respectively) for the duration of the experiment. However, a strong angiogenic response was observed in eyes 7 days post-injection with S 1 and S 2 when viewed using CFP. Neovascularization occurred some distance from the injection site and appeared as a distinct red band of blood vessels extending across the retina (Fig. 5c). Further examination using fluorescein angiography confirmed the formation of new vessels, which appears as hyperfluorescence (Fig. 5d) due to the "leaky" nature that blood vessels possess during angiogenesis. In addition, FA revealed the presence of microaneurisms (not shown) in eyes injected with S 1 and S 2 . Later examinations performed 14 days post-injection of S 1 and S 2 revealed the occurrence of intraretinal hemorrhage in eyes injected with S 1 and S 2 that appears as black spots using CFP (Fig. 5c) and as hypo-fluorescence using fluorescein angiography (Fig. 5f). The intraretinal hemorrhage increased in severity 21 days post-injection (Fig. 5g) and appears as a large hypofluorescent area using FA (Fig. 5h). The observations detailed above were closely paralleled in a similar mouse model following subretinal injection where leakiness, micro aneurysms, and intraretinal hemorrhage were also observed in eyes injected with S 1 and S 2 7 days post-injection, whereas eyes injected with S 3 retained a normal appearance (photographs not shown).
To correlate the angiogenic response seen in vivo to the results found in vitro, ELISA was used to assay the VEGF protein concentrations from the eyes of rats injected with S 1 , S 2 , and S 3 . The results (Fig. 6) show that both S 1 and S 2 mediated a significant (p Ͻ 0.05) increase in the concentration of VEGF protein within the eye (2952 Ϯ 193 fg mg Ϫ1 and 2404 Ϯ 124 fg mg Ϫ1 of total protein respectively) compared with the vehicle-injected control (1930 Ϯ 92 fg mg Ϫ1 of total protein). In addition, consistent to the in vitro results and in vivo observations, S 3 was unable to significantly change the concentration of ocular VEGF (1768 Ϯ 45 fg mg Ϫ1 of total protein).
Sequence Comparison-Cross-species comparisons of VEGF 5Ј-UTR sequences between bovine (GenBank TM accession number NM_174216), murine (GenBank TM accession number NM_009505), and human have revealed a high level of conservation for the S 1 to S 2 region between human and bovine, but the murine 5Ј-UTR revealed a complete lack of the S 1 sequence (Fig. 7). No sequence information was available for the 5Ј-UTR of the rat and, therefore, no direct comparison could be made. DISCUSSION Controlled regulation of VEGF in vivo is important in maintaining the health of many tissues and cells types. However, increased levels of VEGF associated with ischemic conditions leads to a variety of angiogenic ocular diseases, including diabetic retinopathy and retinopathy of prematurity (22,23), in addition to promoting vasculogenesis in cancerous tissues (24,25). Central to the regulation of VEGF is the presence of both a 5Ј-UTR and 3Ј-UTR, both of which contain many regulatory elements, including hypoxia and glucose response elements (26), in addition to stabilizing and destabilizing elements (8). In this study we report on the discovery of a novel control element within the 5Ј-UTR of the human VEGF gene that may act as a target site for a destabilizing protein in addition to providing further insight into its regulation.  FIG. 3. Characterization of a destabilizing element within the VEGF 5-UTR. A schematic drawing of the plasmid constructs used in the transfections is shown. pUTR-W contains the entire UTR, pUTR-L contains the proximal ApaI-XmnI fragment, and pUTR-S contains the distal XmnI-EcoRI fragment from pGEM T-UTR. Plasmids were transfected into RPE 51 cells as described under "Experimental Procedures." On the right is a histogram representing the reporter gene activity as compared with that of the p⌬UTR control, with each value representing the average of four independent transfection experiments. CMV, cytomegalovirus.
Two sense oligonucleotides (S 1 and S 2 ) were designed to resemble a potential regulatory region within the 5Ј-UTR of the VEGF gene. A third ODN (S 3 ) was designed as a control and mapped to the 16 bp immediately 5Ј to S 1 . Results from the in vitro studies demonstrated that S 1 and S 2 mediated a 2-fold increase in protein production and up to a 1.5-fold increase in the level of a mRNA transcript. This indicates that the sequences in the 5Ј-UTR represented by S 1 and S 2 contain regulatory elements involved in the modulation of VEGF production. Possible mechanisms for VEGF protein up-regulation by S 1 and S 2 include competitive inhibition of either a mRNA destabilizing protein or a transcriptional repressor protein. In the case of the latter, transcriptional repressor proteins have been described previously (19,20) and share a common theme of recognizing the variations of a homopurine GA-type sequence consensus motif similar to the sequence found in S 1 and S 2 . However, our data suggest that S 1 and S 2 are competing for the recognition site of an mRNA destabilizing protein. If the mechanism of up-regulation were mediated by increased mRNA production through the inhibition of a repressor protein, we would see a proportional increase between protein and mRNA. However, this was not the case for S 1 and S 2 where protein levels were increased by 2-fold compared with the control, whereas mRNA was only increased by 1.5 and 1.25 times, respectively. Levels of mRNA are determined by the equilibrium that exists between synthesis and degradation; therefore, an increase in stability will reduce degradation and cause an equilibrium shift resulting in higher levels of mRNA being present without an increase in mRNA transcription. The improved mRNA stability and, hence, the increased half-life will result in a proportionally greater amount of protein produced per molecule of mRNA. In addition, stabilization/destabilization of mRNA has been shown previously to be the mechanism associated with increases in VEGF protein during periods of hypoxia (27) and has been well documented as playing a role in the regulation of other cellular elements such as transferrin receptors (28,29), elastin (30), and resistin (31). The possibility of destabilizing elements being present within the 5Ј-UTR of the VEGF mRNA transcript has been reported previously (8), but no definitive consensus sequence has yet been described.
Reporter gene studies have been used to confirm that the activity of S 1 and S 2 was exerted at the transcriptional level. The presence of the many regulatory factors located within the 5Ј-UTR of VEGF makes it difficult to explain the overall effect that truncation has on reporter gene activity. However, we were able to show that removal of the 3Ј-distal end of the 5Ј-UTR (pUTR-L), which contained the proposed regulatory element, resulted in a reduced down-regulatory effect compared with the full UTR construct (pUTR-W) in addition to a loss of up-regulation of reporter product when cultured in the presence of S 1 . Conversely, cells transfected with reporter plasmids containing the complete 5Ј-UTR, and the fragment containing the proposed destabilizing element (pUTR-W and pUTR-S, respectively) showed significant increases in reporter gene activity when cultured in the presence of S 1 . It should be noted that the relative increase in reporter activity was less for pUTR-S (1.39-fold) than for the complete pUTR-W (1.79-fold), which was a comparable increase over that found in the in vitro FIG. 5. Effects on the retinas of rats injected with regulatory oligonucleotides. Eyes injected with S 3 remained free and clear of neovascularization using both CFP and FA (a and b, respectively), with the injection site being clearly visible (a, white arrow). Seven days post-injection with the oligos S 1 and S 2 resulted in angiogenesis, which appeared as a distinct red band under CFP (c, yellow arrows). Subsequent FA revealed hyperfluorescence (d, yellow arrows) due to the leaky nature of the new blood vessels. Fourteen days post-injection resulted in the development of an intraretinal hemorrhage, which appears as black spots using CFP (e, red arrows) and hypo-fluorescence using FA (f, red arrows). After 21 days the intraretinal hemorrhage had developed further (g, blue arrow) and appears as a large area of hypo-fluorescence (h, blue arrow) FIG. 6. Measurement of the VEGF concentration in rat eyes injected with vehicle, S 1 , S 2 , and S 3 . ODNs and vehicle were administered subretinally, and the eyes were enucleated 7 days later for VEGF ELISA analysis. VEGF concentrations were normalized against the total protein concentration of the eye extracts and expressed as fg (VEGF) mg Ϫ1 of total protein ELISA assays. Use of computer modeling has shown the 5Ј-UTR of VEGF to possess a complex secondary structure that is crucial for normal functioning of the gene (5,32). It is therefore conceivable that truncation of the 5Ј-UTR would cause an alteration in the folding pattern and affect the normal functions of the various control elements, resulting in reduced functioning such as that described above.
To study the effects of S 1 , S 2 , and S 3 on VEGF regulation in vivo, a rodent ocular model was chosen. VEGF isoforms are the same for all tissues, and the eye makes an attractive organ to use because the effects on ocular vascularization by changes in VEGF levels have been well described (review in Ref. 33). In addition, the vasculature of the eye can be readily studied through real time ophthalmologic examination. When introduced to the anterior chamber of the rat eye, a strong neovascular response in the iris was observed for both S 1 and S 2 . Likewise, subretinal injection of S 1 and S 2 in both rats and mice resulted in a similar response in the retina in addition to the formation of microaneurysms and leakage associated with the growth of new blood vessels. This pattern of neovascularization can also be observed in a rodent model with an elevated expression of a VEGF transgene (34) as well as in patients suffering from diabetic retinopathy (35). The injection of S 3 resulted in no observable response as was seen in the in vitro study. This provided a strong indication that the presence of S 1 and S 2 mediated an increase in the level VEGF protein with the effect of stimulating neovascularization. To test this hypothesis, the concentration of VEGF in rat eyes 7 days post-injection with S 1 , S 2 , and S 3 was calculated. Eyes injected with S 1 and S 2 showed significantly elevated VEGF protein levels (1.59-and 1.25-fold, respectively) compared with the vehicle-injected control when measured using ELISA, whereas no difference was recorded in S 3 -injected eyes. Although VEGF has been identified as a highly potent angiogenic factor, to date no data exists on the minimal increase required to promote an angiogenic response in the retina. Previous studies using RPE cells have shown that the up-regulation of VEGF due to hypoxic conditions to be 1.3-to 1.5-fold (36). In addition, VEGF was preferentially secreted to the basal side of the cell to yield a 2-to 7-fold higher accumulation of VEGF on the basal side as compared with the apical side, which may prove to be the more important factor involved in ocular angiogenesis.
Comparisons of published sequences of the 5Ј-UTR show some variation in the sequence region proposed for the presence of a destabilizing element. However, S 1 and S 2 both mediate a response in the mouse model, which lacks the S 1 sequence, thereby providing evidence that the inhibition was due to a shorter consensus sequence common to both S 1 and S 2 . Responses to hypoxia are dependent on the presence of the hypoxia response element, which consists of a 6-bp core consensus sequence (37). Similarly, low levels of glucose can mediate and increase in VEGF through the glucose response element (26). S 1 and S 2 both contain the element (T/A)GGGG, which may represent the core recognition sequence of a destabilizing protein.
This research provides the strongest evidence to date for the existence of a destabilizing element within the 5Ј-UTR of the VEGF gene. In addition, we have shown the potential location of these elements and discussed their importance in the regulation of VEGF protein production. Further research will evaluate the protein-mRNA relationship and how this may translate into a disease state. This understanding may lead to potential treatments for ischemic tissues such as that found in the heart following a myocardial infarct.