Stretch-induced Retinal Vascular Endothelial Growth Factor Expression Is Mediated by Phosphatidylinositol 3-Kinase and Protein Kinase C (PKC)- but Not by Stretch-induced ERK1/2, Akt, Ras, or Classical/Novel PKC Pathways

Stretch-induced expression of vascular endothelial growth factor (VEGF) is thought to be important in mediating the exacerbation of diabetic retinopathy by systemic hypertension. However, the mechanisms underlying stretch-induced VEGF expression are not fully un-derstood. We present novel findings demonstrating that stretch-induced VEGF expression in retinal capillary pericytes is mediated by phosphatidylinositol (PI) 3-ki-nase and protein kinase C (PKC)- (cid:1) but is not mediated by ERK1/2, classical/novel isoforms of PKC, Akt, or Ras despite their activation by stretch. Cardiac profile cyclic stretch at 60 cpm increased VEGF mRNA expression in a time- and magnitude-dependent manner without altering mRNA stability. Stretch increased ERK1/2 phosphorylation, PI 3-kinase activity, Akt phosphorylation, and PKC- (cid:1) activity. Signaling pathways were explored using inhibitors of PKC, MEK1/2, and PI 3-kinase; adenovirus-medi-ated overexpression of ERK, PKC- (cid:2) , PKC- (cid:3) , PKC- (cid:1) , and Akt; and dominant negative (DN) mutants of ERK, PKC- (cid:1) , Ras, PI 3-kinase and Akt. Although stretch activated ERK1/2 through a Ras- and PKC classical/novel isoform-dependent pathway, these pathways were not responsible for stretch-induced VEGF expression. Overexpression of DN ERK and Ras had no effect on VEGF expression in these cells. In contrast, DN PI 3-kinase as well as pharma-cologic inhibitors of PI 3-kinase blocked stretch-induced VEGF expression. Although stretch-induced PI 3-kinase activation increased both Akt phosphorylation and activity of PKC- (cid:1) , VEGF expression was dependent on PKC- (cid:1) but not Akt. In addition, PKC- (cid:1) did not mediate stretch-induced ERK1/2 activation. These results suggest that stretch-induced expression of VEGF involves a novel mechanism dependent upon PI 3-kinase-mediated activation of PKC- (cid:1) that is independent of stretch-induced activation of ERK1/2, classical/novel PKC isoforms, Ras, or Akt. This mechanism may play a role in the well documented association of concomitant hypertension with clinical exacerbation of neovascularization and vascular permeability. non-normal p statistically

One in four American adults has hypertension, while 5.9% of the United States population (over 15 million people) have diabetes. Diabetic retinopathy is the leading cause of new onset blindness in the United States among working age individuals (1) and is exacerbated by coexistent systemic hypertension (2)(3)(4). Sight-threatening diabetic retinopathy is characterized by development of retinal neovascularization and/or retinal vascular permeability (5). Hypertension increases the risk of retinopathy progression, development of neovascularization (2,6,7), and retinal vascular permeability (8,9) by up to 3-fold. Blood pressure control reduces both retinopathy progression and severe visual loss (10). Even in normotensive diabetic patients retinopathy is associated with higher systolic blood pressure (11). Other vision-threatening conditions such as hypertensive retinopathy (12) and age-related macular degeneration are also aggravated by hypertension (13).
Although the mechanisms underlying the exacerbation of these conditions by hypertension are not fully understood, vascular endothelial growth factor (VEGF) 1 has been strongly implicated as a primary mediator of ocular complications in diabetes and age-related macular degeneration. VEGF is a hypoxia-induced, endothelial cell-selective mitogen (14 -16) also called vascular permeability factor after its potent ability to induce vasopermeability (17). VEGF is the principal stimuli for intraocular neovascularization and retinal vascular permeability in diabetic retinopathy, retinal vein occlusion, retinopathy of prematurity, age-related macular degeneration, and numerous other conditions (18 -27). VEGF exerts its action through the high affinity tyrosine kinase insert domain-containing receptor (KDR, VEGF-R2) (28,29). In vivo, hypertension can increase large artery (30) and retinal artery distention (31) as much as 15 and 35%, respectively. Mechanical stretch induces VEGF expression in rat ventricular myocardium (32), rat cardiac myocytes (33), human mesangial cells (34), and cultured retinal pigment epithelial cells (35). Recently we reported that mechanical stretch induced expression of VEGF and its receptors in retinal endothelial cells (36) and demonstrated that retinal expression of VEGF and VEGF-R2 was increased during hypertension in vivo.
The molecular mechanisms underlying stretch-induced VEGF expression have not been studied extensively. Stretch rapidly activates a plethora of second messenger pathways including tyrosine kinases, p21 ras , extracellular signal-regulated kinase (ERK), S6 kinase, protein kinase C (PKC), phospholipases C and D, and the P450 pathway (37,38). Mechanical stretch can also regulate protein synthesis and the activity of numerous factors including NO (39), endothelin-1 (40), platelet-derived growth factor (41), fibroblast growth factor (42,43), and angiotensin II (44). Cyclic stretch can increase nerve growth factor in cultured urinary tract smooth muscle cells, an effect blocked by prolonged exposure to phorbol ester resulting in down-regulation of multiple PKC isoforms including ␣, ␤, ␦, ⑀, and (45).
Of the numerous isoforms of PKC involved in the diverse signaling pathways of diabetes complications (46 -48) and tumor angiogenesis (49,50) PKC-has been implicated in the regulation of VEGF expression (49,50). PKC-is an atypical isoform lacking the Ca 2ϩ binding C2 domain and with only one cysteine-rich zinc finger-like motif in the diacylglycerol binding C1 domain (51). Thus, PKC-does not bind Ca 2ϩ and is not activated by diacylglycerol or phorbol esters (52). PKCis activated by several lipid mediators including phosphatidic acid (52) and phosphatidylinositol 3,4,5-trisphosphate (53). Nevertheless, PKCactivity is important in mitogenesis, protein synthesis, cell survival, and regulation of transcription (54,55).
Expression of VEGF in response to Ras (56), von Hippel-Lindau tumor suppressor gene (50,57), and transcription factor SP1 (49) is dependent upon PKCand subsequent ERK1/2 activation. Ras-induced VEGF expression in human fibrosarcoma and renal cell carcinoma cell lines is almost totally dependent on PKCactivity, which is mediated through both Raf-dependent and Raf-independent pathways (56). PKC-has also been reported to mediate the downstream proliferative effect of VEGF (58).
In this study, we examined the molecular mechanism of stretch-induced VEGF expression in retinal cells. These data are the first to demonstrate that stretch-induced VEGF expression is mediated by phosphatidylinositol (PI) 3-kinase and PKCin a manner independent of ERK1/2, Akt, or Ras. Thus, stretch-induced VEGF expression may be distinct from other pathways mediating VEGF expression, and theoretically, PI 3-kinase and PKCinhibitors may have therapeutic benefit in ameliorating the well documented exacerbation of ocular diseases by concomitant hypertension.
Cell Culture-Primary cultures of bovine retinal pericytes (BRPCs) were isolated by homogenization and a series of filtration steps as described previously (59). BRPCs were cultured in Dulbecco's modified Eagle's medium containing 5.5 mM glucose and 20% fetal bovine serum. The cells were maintained in 5% CO 2 at 37°C, and media were changed every 3 days. Cells were characterized for their homogeneity by immunoreactivity with monoclonal antibody 3G5 (60). Cells were plated at a density of 2 ϫ 10 4 cells/cm 2 and passaged when confluent. The media were changed every 3 days, and only cells from passages 2-5 were used for experiments.
Recombinant Adenoviruses-cDNA of constitutively active Akt (ca Akt; Gag protein fused to the N terminus of wild type Akt) was constructed as described previously (61). cDNA of dominant negative Akt (mt Akt) was constructed by substituting Thr-308 to Ala and Ser-473 to Ala as described previously (62). cDNA of ERK was constructed as described previously (63). cDNA of dominant negative mutant ERK (mt ERK) was constructed by substituting Lys-52 to Arg in the ATP-binding site as described previously (64). cDNA of dominant negative K-Ras (DN Ras; substituted Ser-17 to Asn) was kindly provided by Dr. Takai (Osaka University) (65). cDNA of ⌬p85 was kindly provided by Dr. Kasuga (Kobe University) (66). cDNAs of PKC-␣, -␦, and -were kindly provided by Dr. Douglas Kirk Ways (Lilly Laboratory, Indianapolis, IN). cDNA of dominant negative PKC-(mt PKC-) substituting Lys-273 to Trp in the ATP-binding site was constructed as described previously (67). The recombinant adenoviruses were constructed by homologous recombination between the parental virus genome and the expression cosmid cassette or shuttle vector as described previously (68,69). Adenovirus was applied at a concentration of 1 ϫ 10 8 plaqueforming units/ml, and adenovirus with the same parental genome carrying LacZ gene or enhanced green fluorescent protein gene (CLON-TECH, Palo Alto, CA) were used as controls. Expression of each recombinant protein was confirmed by Western blot analysis, and expression was increased ϳ10-fold with all constructs as compared with cells infected with control adenovirus.
Mechanical Stretch-Cells were plated on six-well flexible-bottom culture plates coated with collagen (Flexcell Corp., Mckeepsport, PA). After 2 days, media were changed to Dulbecco's modified Eagle's medium containing 1% calf serum, and the cells were incubated overnight. Cells were then subjected to uniform radial and circumferential strain in 5% CO 2 at 37°C using a computer-controlled, vacuum stretch apparatus (Flexcer Cell Strain Unit; Flexcell Corp.). A physiologic stretch frequency of 60 cpm and 3-20% prolongation of elastomer-bottomed plates were used as described previously (36).
RNA Extraction-RNA was extracted using the guanidinium thiocyanate method. RNA purity was determined by the ratio of optical density (OD) measured at 260 and 280 nm, and RNA quantity was estimated using OD measured at 260 nm.
Northern Blot Analysis-Northern blot analysis was performed on 15 g of total RNA/lane after 1% agarose, 2 M formaldehyde gel electrophoresis and subsequent capillary transfer to Biodyne nylon membranes (Pall BioSupport, East Hills, NY). Membranes underwent ultraviolet cross-linking using a UV Stratalinker 2400 (Stratagene, La Jolla, CA). Radioactive probes were generated using Megaprime labeling kits (Amersham Biosciences, Inc.) and [ 32 P]dCTP (PerkinElmer Life Sciences). Blots were prehybridized, hybridized, and washed four times in 0.5ϫ SSC, 5% SDS at 65°C for 1 h in a rotating hybridization oven (Robbins Scientific Corp., Sunnyvale, CA). All signals were analyzed using a computing PhosphorImager with ImageQuant software analysis (Molecular Dynamics, Sunnyvale, CA). The signal for each sample was normalized by reprobing the same blot using 36B4 cDNA control probe. VEGF mRNA Half-life Analysis-BRPCs were cultured as indicated above and exposed to 9%/60 cpm mechanical stretch for 4 h. Actinomycin D (5 g/ml) was added, and RNA was isolated 0, 2, and 4 h later. Northern blot analysis of these samples was performed and quantitated as described above.
VEGF and PKC-Protein Detection-BRPCs were washed with cold phosphate-buffered saline and lysed in 1ϫ Laemmli buffer (50 mM Tris, pH 6.8, 2% SDS, 10% glycerol) containing protease inhibitors (10 mM sodium pyrophosphate, 100 mM NaF, 1 mM Na 3 VO 4 , 1 g/ml aprotinin, 1 g/ml leupeptin, and 2 mM phenylmethylsulfonyl fluoride). Protein concentrations were determined with the Bio-Rad protein assay. Total cell lysate (30 g) was subjected to SDS-PAGE under reducing conditions, and proteins were transferred to nitrocellulose membrane (Bio-Rad). The blots were incubated with primary antibodies followed by incubation with horseradish peroxidase-conjugated secondary antibody (Amersham Biosciences, Inc.). Visualization was performed using the Amersham Biosciences, Inc. enhanced chemiluminescence detection system (ECL) according to the instructions of the manufacturer.
ERK1/2 and Akt Phosphorylation-Cells were washed with cold phosphate-buffered saline and lysed in 1ϫ Laemmli buffer containing protease inhibitors as described above. Cell lysates were heated to 95°C for 2 min, and equal volumes of lysates were subjected to SDS-PAGE under reducing conditions. The blots were incubated with anti-phosphospecific ERK1(p44)/ERK2(p42) or anti-phospho-specific Akt antibody (New England Biolabs). Lane loading differences were normalized by reblotting with nonphosphorylation-specific (total) anti-ERK1 antibody (Santa Cruz Biotechnology, Inc.) or anti-Akt (total) antibody (New England Biolabs).
PI 3-Kinase Assay-PI 3-kinase activity was measured by in vitro phosphorylation of PI (70). Cells were lysed in ice-cold lysis buffer containing 50 mM Hepes, pH 7.5, 137 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 2 mM Na 3 VO 4 , 10 mM NaF, 2 mM EDTA, 1% Nonidet P-40, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 5 g/ml leupeptin, and 1 g/ml pepstatin. Insoluble material was removed by centrifugation at 15,000 ϫ g for 10 min at 4°C. PI 3-kinase was immunoprecipitated from aliquots of the supernatant with antiphosphotyrosine antibodies. After washing, the pellets were resuspended in 50 l of 10 mM Tris (pH 7.5), 100 mM NaCl, and 1 mM EDTA. 10 l of 100 mM MgCl 2 and 10 l of PI (2 g/l) sonicated in 10 mM Tris (pH 7.5) with 1 mM EGTA was added to each pellet. The PI 3-kinase reaction was initiated by the addition of 5 l of 0.5 mM ATP containing 30 Ci of [␥-32 P]ATP. After 10 min at room temperature with constant shaking, the reaction was stopped by the addition of 20 l of 8 N HCl and 160 l of chloroform:methanol (1:1). The samples were centrifuged, and the organic phase was removed and applied to silica gel TLC plates developing in CHCl 3 :CH 3 OH:H 2 O:NH 4 OH (60:47:11:2). The radioactive spots were quantitated by PhosphorImager (Molecular Dynamics).
Statistical Analysis-All experiments were repeated at least three times unless otherwise indicated. Results are expressed as mean Ϯ S.D. Statistical analysis used Student's t test or analysis of variance to compare quantitative data populations with normal distributions and equal variance. Data were analyzed using the Mann-Whitney rank sum test or the Kruskal-Wallis test for populations with non-normal distributions or unequal variance. A p value of Ͻ0.05 was considered statistically significant.

Characterization of Stretch-induced VEGF Expression in
Retinal Capillary Pericytes-Confluent cultures of BRPCs were subjected to a single instance of 5 or 20% static stretch for the durations indicated in Fig. 1A. Static stretch (20%) maximally increased VEGF mRNA expression 2.2-fold after 3 h (p ϭ 0.048). VEGF mRNA levels gradually declined thereafter returning to baseline values after 6 h. VEGF mRNA expression was increased 15 Ϯ 22%, 116 Ϯ 50% (p ϭ 0.048), 90 Ϯ 62%, and Ϫ4 Ϯ 23% after 1, 3, 6, and 9 h, respectively. VEGF mRNA expression in response to 5% static stretch was less pronounced with a tendency to increase within the first 3 h; however, this change was not statistically significant.
The vasculature in vivo is continually exposed to repetitive stretch with pressure dynamics reflecting the cardiac cycle. To approximate this physiologically relevant condition, we evaluated whether cardiac profile cyclic stretch altered VEGF mRNA expression in BRPCs undergoing 9 and 3% cyclic stretch at a rate of 60 cpm with a dynamic stress contour reflecting that of the normal cardiac cycle. As shown in Fig. 1B, cardiac cycle cyclic stretch increased VEGF mRNA expression in a time-and dose-dependent manner. At 9% cyclic stretch, an increase in KDR mRNA expression was initially evident after 1 h, which continued to increase even after 9 h when expression was 3.1 Ϯ 0.2-fold greater than in control cells (p Ͻ 0.001). VEGF mRNA expression was increased 37 Ϯ 15%, 136 Ϯ 25% (p Ͻ 0.001), 168 Ϯ 10% (p Ͻ 0.001), and 206 Ϯ 17% (p Ͻ 0.001) after 1, 3, 6, and 9 h of cyclic stretch, respectively. Cyclic stretch of 3% also increased VEGF mRNA expression, although to a reduced extent with only a 1.7 Ϯ 0.6-fold increase observed after 9 h.
To determine whether stretch-induced VEGF mRNA expression resulted in increased VEGF protein levels, cells were exposed to 9% stretch at 60 cpm for 12 h. Cell lysates were evaluated by Western blot analysis ( Fig. 2A). VEGF protein expression was increased 2.7 Ϯ 1.0-fold (p ϭ 0.002) as compared with control cells. Since stretch-induced mRNA expres-

PKC-and PI 3-Kinase Mediate Stretch-induced VEGF Expression
sion could be the result of alterations in gene transcription or mRNA stability, BRPCs were exposed to 9%/60 cpm cyclic stretch for 4 h and then treated with 5 mg/ml actinomycin D, and RNA was harvested 2 and 4 h later (Fig. 2B). VEGF mRNA concentration declined at an equivalent rate in both control and stretched cells, suggesting that transcriptional regulation, rather than changes in mRNA stability, was primarily responsible for the stretch response.
Evaluation of Stretch-induced Signaling Pathways-Stretch stimulates several signaling pathways in retinal endothelial cells (36). To determine whether similar pathways were activated in retinal pericytes exposed to cardiac profile cyclic stretch, ERK phosphorylation, PI 3-kinase activity, and Akt phosphorylation were evaluated. As shown in Fig. 3, stretch induced a rapid increase in ERK1/2 phosphorylation that was initially evident after 2 min, maximal at 5 min (ERK1 ϭ 20-fold and ERK2 ϭ 8.9-fold increase), and still maintained above baseline even after 60 min (ERK1 ϭ 6.3-fold and ERK2 ϭ 4.5-fold). Both static (Fig. 3A) and cyclic stretch (Fig. 3B) resulted in similar ERK1/2 phosphorylation profiles. An excess of VEGF-neutralizing antibody had no effect on stretch-induced ERK phosphorylation, suggesting that VEGF does not mediate this initial effect (data not shown).
A potential mechanism underlying stretch-induced activation of PI 3-kinase could be the effect of stretch on PDGF receptor B (PDGFR-B) (41). Immunoprecipitation with antibody specific for PDGFR-B and subsequent immunoblotting with antibodies specific for phosphotyrosine or the p85 subunit of PI 3-kinase showed stretch-induced phosphorylation of PDGFR-B and increased association with p85 (Fig. 5A). Conversely, immunoprecipitation with phosphotyrosine-specific antibody and subsequent immunoblotting with antibodies specific for PDGFR-B or p85 showed similar stretch-induced phosphorylation of PDGFR-B and increased association with p85 ( Fig. 5B). Stretch greatly increased the PDGFR-B associated with p85 following immunoprecipitation with antibodies specific for p85 (Fig. 5C).
Mechanistic Evaluation of Stretch-induced VEGF Expression-To determine the mechanism by which stretch increased VEGF mRNA expression, inhibitors of MEK1 (PD98059, 20 M), classical/novel PKC isoforms (GF109203X, 5 M), tyrosine phosphorylation (genistein, 20 M), and PI 3-kinase (wortmannin, 100 nM; and LY294002, 50 M) were evaluated as shown in Fig. 6, A-D, respectively. In all experiments 9%/60 cpm cyclic stretch for 3 h induced VEGF mRNA expression (Fig. 6E, 2.3 Ϯ 0.3-fold, p Ͻ 0.01). As shown in Fig. 6, A and E, inhibition of ERK1/2 using PD98059 had little effect on either basal or stretch-induced expression of VEGF. Similarly, inhibition of PKC classical/novel isoforms using GF109203X did not alter VEGF mRNA expression (Fig. 6, B and E). In contrast, inhibition of PI 3-kinase using either the inhibitor LY294002 or wortmannin resulted in marked inhibition of stretch-induced VEGF mRNA expression without significantly altering basal expression levels (Fig. 6, D and E). LY294002 and wortmannin inhibited stretch-induced VEGF mRNA expression by 85 Ϯ 20% (p ϭ 0.039) and 96 Ϯ 25% (p ϭ 0.035), respectively. Addition of genistein inhibited stretch-induced VEGF mRNA FIG. 2. Cyclic stretch induces VEGF protein expression in bovine retinal pericytes without altering VEGF mRNA stability. A, confluent cultures of BRPCs were exposed to 9% cyclic stretch at 60 cpm for 12 h. Cell lysates were isolated and subjected to Western blot analysis using polyclonal antibody against VEGF. The VEGF signal and the location of an 18.5-kDa molecular mass marker are indicated in the figure (top). Quantitation of multiple experiments is also presented (bottom). B, cells were stretched as indicated above for 4 h, and control cells were treated similarly but were not stretched. Stretching was terminated, and actinomycin D (5 g/ml) was added to the cells at time 0. VEGF mRNA was evaluated after 2 and 4 h. Quantitation of multiple experiments is shown.
FIG. 3. Static and cyclic stretch induce rapid phosphorylation of ERK1 and ERK2. Confluent cultures of bovine retinal pericytes were exposed to 20% static or 9% cyclic stretch for the times indicated in A and B, respectively. Phospho-ERK1/2 (pERK1/2) and total ERK1/2 were detected by chemiluminescent Western blot analysis using specific anti-phospho-ERK1/2 and anti-ERK1 (total) antibodies, respectively. Representative Western blots are shown. The experiment was repeated three times with similar results. expression 87 Ϯ 12% (p ϭ 0.041) also without altering basal VEGF expression (Fig. 6, C and E). These results suggest that tyrosine phosphorylation events and activation of PI 3-kinase are required for stretch-induced VEGF mRNA expression, whereas activation of classical/novel PKC isoforms and ERK1/2 are not major contributors to this response.
To determine whether Akt mediated stretch-induced VEGF expression, adenovirus infection using ca Akt or mt Akt was performed (Fig. 8C). Overexpression of constitutively active Akt did not increase basal or stretch-induced VEGF mRNA expression as compared with ␤-galactosidase control-infected cells. The effect of dominant negative Akt expression was variable and did not demonstrate a statistically significant effect. Further confirmation that PI 3-kinase was important in stretch-induced VEGF expression was obtained using adenoviral infection with the dominant negative mutant of the p85 subunit of PI 3-kinase (⌬p85), which inhibited stretch-induced VEGF mRNA expression by 130 Ϯ 24.5% (p Ͻ 0.01) without altering basal VEGF expression.
Role of PKC-in Stretch-induced VEGF Expression-Since the PKC inhibitors evaluated in this study effect novel and

FIG. 4. Cyclic stretch increases PI 3-kinase activity and Akt phosphorylation.
A, confluent cultures of bovine retinal pericytes were exposed to 9% cyclic stretch for the times indicated after which PI 3-kinase activity was measured as described under "Experimental Procedures." A representative TLC plate is shown (top) as is quantitation of multiple experiments (bottom). PIP, phosphatidylinositol phosphate. B, cells were exposed to 9% cyclic stretch for the times indicated. Phospho-Akt (pAkt) and total Akt were detected by Western blot analysis using specific antibodies and chemiluminescence. A representative Western blots is shown (top) as is quantitation of multiple experiments (bottom).

FIG. 5. Stretch increases PDGF receptor B tyrosine phosphorylation and association with p85
. Cells were exposed to 9% cyclic stretch for 15 min. Cellular protein was isolated and immunoprecipitated prior to immunoblotting. A, immunoprecipitation with antibodies specific for phosphotyrosine (PY) followed by immunoblotting with antibodies specific for PDGFR-B or p85. B, immunoprecipitation with PDGFR-B-specific antibody followed by immunoblotting with antibodies specific for phosphotyrosine (PY), PDGFR-B, or p85. C, immunoprecipitation with p85-specific antibody followed by immunoblotting with antibodies specific for PDGFR-B or p85. Experiments were repeated at least two times with similar results. IP, immunoprecipitation; IB, immunoblot.

PKC-and PI 3-Kinase Mediate Stretch-induced VEGF Expression
classical isoforms of PKC but not atypical isoforms and since PI 3-kinase has been reported to activate the atypical isoform of PKC (53,72), we evaluated the role of PKC-in stretch-induced VEGF expression. To determine whether PKC-was actually expressed in retinal pericytes, Western blot analysis using PKC--specific antibody was performed. As shown in Fig. 9A, retinal pericytes clearly expressed PKC-protein, and expression was greater than that observed in retinal endothelial cells. As shown in Fig. 9B, adenovirus-mediated overexpression of wild type classical PKC isoform ␣, novel PKC isoform ␦, or green fluorescent protein (GFP) control had no effect on either basal or stretch-induced VEGF mRNA expression. In contrast, overexpression of the wild type atypical isoform of PKC further increased stretch-induced VEGF mRNA expression 91 Ϯ 48% (p Ͻ 0.04), while dominant negative expression of PKCinhibited stretch-induced VEGF expression by 73 Ϯ 25% (p Ͻ 0.02) as compared with GFP control. Basal VEGF mRNA expression was not changed. In contrast, adenovirus-mediated expression of wild type or dominant negative mutant PKC-did not effect either basal or stretch-induced ERK1/2 phosphorylation (Fig. 9C) or Akt expression or phosphorylation (data not shown) as compared with GFP control-infected cells.
The effect of cyclic stretch on PKC-activity and its relation to PI 3-kinase activation was evaluated as shown in Fig. 10. PKC--specific activity was increased 2.6 Ϯ 0.7-fold by 15 min of 9% cyclic stretch, a response completely inhibited by the PI 3-kinase inhibitor wortmannin (p Ͻ 0.01).
In human fibrosarcoma and renal cell carcinoma cells, Ras can promote VEGF transcription by activating PKC- (56). To evaluate whether a similar mechanism was involved in stretchinduced VEGF expression, cells underwent adenoviral infection with DN Ras or ␤-galactosidase control. DN Ras did not effect basal or stretch-induced VEGF expression (Fig. 11A). In contrast, DN Ras inhibited stretch-induced ERK1 and ERK2 phosphorylation by 73 Ϯ 14% (p ϭ 0.003) and 70 Ϯ 20% (p ϭ 0.007), respectively (Fig. 11B). These data suggest that stretchinduced, PKC--mediated VEGF expression occurs via a mechanism not predominantly involving Ras or ERK1/2.
The time course of VEGF expression in response to static and cyclic stretch in retinal pericytes was similar to that observed in retinal endothelial cells, although the magnitude of the response was approximately one-third of that in endothelial cells (36). Cyclic stretch induced rapid increases in ERK1/2 phosphorylation, PI 3-kinase activity, Akt phosphorylation, and PKC-activity. However, the ERK1/2 independence of stretch-induced VEGF expression was substantiated by several findings. Stretch-induced VEGF mRNA expression was not suppressed by either PD98059 or adenovirus infection with dominant negative ERK. Overexpression of wild type ERK did not increase basal or stretch-induced VEGF expression. Furthermore, stretch-induced ERK1/2 activation was mediated by classical/novel isoforms of PKC and Ras (as evidenced by inhibition of the response by classical/novel PKC isoforms inhibitor GF109203X and overexpression of dominant negative Ras) but not mediated by PI 3-kinase, tyrosine kinases, or PKC-(as evidenced by lack of response to wortmannin and LY294002, lack of response to genistein, or overexpression of wild type and dominant negative PKC-, respectively). In contrast, the opposite results were obtained when evaluating these interventions on stretch-induced VEGF expression. These data demonstrate that, although stretch activates several signaling pathways, VEGF expression is mediated by PI 3-kinase and PKC-in an ERK-, Ras-and classical/novel PKC isoform-independent manner. In addition, direct modulation of ERK may not be adequate in itself to alter VEGF expression in these cells as evidenced by the lack of effect of ERK1/2 inhibitors and wild type or domi-FIG. 6. Stretch-induced VEGF mRNA expression is PI 3-kinase-and tyrosine phosphorylation-dependent but independent of ERK1/2 or classical/novel PKC isoforms. Confluent cultures of BRPCs were exposed to 9% cyclic stretch at 60 cpm for 3 h in the presence of MEK1 inhibitor PD98059 (20 M), PKC classical/novel isoform inhibitor GF109203X (5 M), tyrosine kinase inhibitor genistein (20 M), or PI 3-kinase inhibitors wortmannin (100 nM) or LY294002 (50 M). RNA was isolated and subjected to Northern blot analysis for VEGF and 36B4 control. Representative Northern blot analysis (top) and quantitation of multiple experiments following normalization at 36B4 control signal (bottom) are shown. ctl, control.

PKC-and PI 3-Kinase Mediate Stretch-induced VEGF Expression 1052
nant negative ERK expression. It should be noted, however, that overexpression of wild type ERK1/2 might not have a major impact on the basal state if ERK is not significantly activated.
The ERK independence of stretch-induced or basal VEGF expression is surprising. ERK has been reported as important in VEGF expression induced by starvation in human colon carcinoma cells (73), v-ras, v-raf, and c-myc transformation of rat liver epithelial cells (74), phorbol 12-myristate 13-acetate FIG. 7. Stretch-induced ERK1/2 phosphorylation is dependent on classical/novel PKC isoforms and independent of PI 3-kinase or tyrosine phosphorylation, while stretch-induced VEGF mRNA expression is not dependent on ERK1/2 activity. A, confluent cultures of bovine retinal pericytes were exposed to 9% cyclic stretch for 5 min and phospho-ERK1/2 (pERK1/2) and total ERK1/2 were detected by Western blot analysis. A representative Western blot is shown (top) as is quantitation of multiple independent experiments after normalization to total ERK1/2 (bottom). B, cells were infected with adenovirus containing ␤-galactosidase control (␤-gal), wild type ERK (wt ERK), or a dominate negative mutant ERK (mt ERK). Cells were exposed to 9% cyclic stretch for 3 h, and Northern blot analysis for VEGF and 36B4 control was performed. A representative Northern blot (top) is shown as well as quantitation from multiple independent experiments after normalization to 36B4 control probe (bottom). ctl, control; pp42, phospho-p42; pp44, phospho-p44.

FIG. 8. Stretch-induced Akt phosphorylation and VEGF expression are PI 3-kinase-dependent, but Akt does not mediate stretch-induced VEGF mRNA expression.
A, BRPCs were exposed to 9% cyclic stretch for 15 min in the presence of the inhibitors described in Fig. 5. Phospho-Akt (pAkt) and total Akt were detected by Western blot analysis. A representative Western blot is shown (top) as is quantitation of multiple independent experiments after normalization to total Akt (bottom). B, cells were stretched as above after infection with adenovirus containing a dominant negative mutant of the p85 subunit of PI 3-kinase (⌬p85) or ␤-galactosidase control (␤-gal). C, cells were infected with adenovirus containing ␤-galactosidase control (␤gal), constitutively active Akt (ca Akt), dominate negative mutant Akt (mt Akt), or a dominate negative mutant of the p85 subunit of PI 3-kinase (⌬85). Cells were stretched, and Northern blot analysis for VEGF and 36B4 control was performed. A representative Northern blot (top) is shown as well as quantitation from multiple independent experiments after normalization to 36B4 control probe (bottom). ctl, control. treatment in human glioblastoma U373 cells (57), Ras expression in human fibrosarcoma and renal cell carcinoma cell lines (56), endothelin stimulation of human vascular smooth muscle cells (76), and von Hippel-Lindau tumor suppressor gene action (50). Hypoxic induction of VEGF may also involve ERK since inhibition of Raf-1 markedly reduces VEGF induction (77); however, hypoxia can be additive to VEGF expression induced by ERK1/2 activation in hamster fibroblasts where a single inhibitor of ERK did not suppress hypoxia-induced VEGF expression (78). The ERK independence observed in our system suggests that VEGF expression in response to different stimuli may be mediated by a variety of signaling pathways and/or may reflect a potential uniqueness of retinal pericytes.
To our knowledge, the activation of PKC-by stretch has not been previously documented. The importance of the atypical PKC-isoform in mediating stretch-induced VEGF expression was underscored by several findings. PKC-protein expression was present in retinal endothelial cells and present in even higher amounts in retinal pericytes. PKC-activity was increased nearly 3-fold by cyclic stretch. Stretch-induced VEGF expression was inhibited by expression of dominant negative PKC-and increased by overexpression of wild type PKC-. In contrast, overexpression of wild type classical PKC-␣ isoform or novel PKC-␦ isoform did not effect VEGF expression. The activation of PKC-within 15 min of stretch onset is consistent with previous time course data for PKC-activation following exposure to insulin (10 -20 min) (79), nerve growth factor (9 -15 min) (80), or hypoxia-reperfusion (15 min) (81).
In other systems, including insulin-stimulated rat adipocytes (82), reoxygenation of rat cardiomyocytes (81), and endotoxin-treated human alveolar macrophages (84), PI 3-kinase activation induces ERK activity through a PKC--mediated pathway. However, our data suggest that stretch-induced activation of ERK1/2 in retinal pericytes is mediated by a different mechanism since inhibition of PKC-using dominant neg-(BREC) using PKC--specific antibody. B, bovine retinal pericytes were infected with adenovirus containing GFP control, wild type PKC-␣ (wt ␣), wild type PKC-␦ (wt ␦), wild type PKC-(wt ), or a dominate negative mutant of PKC-(mt ). After 2 days, cells were exposed to 9% cyclic stretch for 3 h, and mRNA was isolated for Northern blot analysis. Representative Northern blot results (top) and quantitation of multiple experiments following normalization to 36B4 control signal (bottom) are shown. C, BRPCs were infected with adenovirus containing GFP control, wild type PKC-(wt ), or dominate negative mutant PKC-(mt ) as described above, and ERK1/2 phosphorylation was evaluated. A representative Western blot of phospho-ERK1/2 and total ERK1/2 is shown (top) as is quantitation of multiple independent experiments normalized to total ERK1/2 (bottom). ctl, control.  10. Cyclic stretch-induced PKC-activity is PI 3-kinasedependent. Confluent cultures of BRPCs were exposed to 9% cyclic stretch for 15 min in the presence or absence of the PI 3-kinase inhibitor wortmannin after which PKCactivity was measured as described under "Experimental Procedures." The results of three independent experiments are shown. ctl, control.
Although these are the first studies to evaluate the role of PKC-in stretch-induced VEGF expression, PKC-has been previously implicated as a modulator of VEGF (49,50). Overexpression of PKC-in human glioblastoma U373 cells increased VEGF mRNA expression (57). The von Hippel-Lindau tumor suppressor gene has been shown to form cytoplasmic complexes with PKC-␦ and PKC-, preventing their translocation to the cell membrane and reducing the constitutive overexpression of VEGF characteristically observed in sporadic re-nal cell carcinomas (50). In addition, PKC-binds and phosphorylates transcription factor SP1 in renal cell carcinomas, resulting in VEGF expression. Ras-induced VEGF expression in human fibrosarcoma and renal cell carcinoma cell lines is almost totally dependent on PKC-activity (56). However, as discussed above, ERK was an important component of these pathways.
The role of PI 3-kinase in stretch-induced VEGF expression and Akt phosphorylation was supported by the inhibitory effect of two different PI 3-kinase inhibitors (wortmannin and LY294002) and dominant negative expression of the p85 subunit of PI 3-kinase. In addition, wortmannin completely inhibited stretch-induced PKC-activity. However, Akt did not appear to mediate stretch-induced VEGF expression as expression of dominant negative or constitutively active Akt had no effect. This finding differs from that observed in chicken cells where overexpression of myristylated Akt increased basal VEGF expression and restored VEGF expression in cells after PI 3-kinase inhibition (85). Thus, the role of Akt in mediating VEGF expression may be cell type-and/or stimuli-dependent. Our studies do not eliminate the possibility that stretch-induced Akt may be involved in late stages of VEGF expression (86) but do suggest that, at least for stretch-induced VEGF expression, the PKC-pathway, independent of Akt activation, predominates within the first several hours in retinal pericytes The upstream mechanism by which cellular stretch induces PI 3-kinase and PKC activation in retinal cells is not understood; however, stretch can induce the expression of numerous genes through activation of various intracellular pathways including membrane K ϩ channels, G proteins, intracellular Ca 2ϩ , cAMP, cGMP, inositol trisphosphate, protein kinase C, mitogen-activated protein kinase, protein tyrosine kinases, focal adhesion kinase, and alterations in intracellular redox state (87)(88)(89). Fluid shear stress can also mediate signaling through activation of heterotrimeric and small G proteins, resulting in ERK1/2 and phospholipase C activation with subsequent inositol 1,4,5-trisphosphate and diacylglycerol generation, Ca 2ϩ release, and PKC activation (37). However, this mechanism may not be involved in stretch-induced VEGF expression due to the noted ERK1/2 independence and involvement of PKC-, a Ca 2ϩ -independent isoform of PKC. Interestingly, mechanical stretch can directly induce growth factor receptor autophosphorylation presumably through changes in cellular morphology leading to altered receptor conformation and subsequent expo- FIG. 11. Stretch-induced VEGF mRNA expression is Ras-independent, whereas stretch-induced ERK1/2 phosphorylation is Ras-dependent. A, BRPCs were infected with adenovirus containing ␤-galactosidase control (␤gal) or a dominant negative mutant of Ras (DNras). After 2 days, cells were exposed to 9% cyclic stretch for 3 h, and mRNA was isolated for Northern blot analysis. Representative Northern blot results (top) and quantitation of multiple experiments following normalization to 36B4 control signal (bottom) are shown. B, BRPCs were infected with adenovirus as described above, and ERK1/2 phosphorylation was evaluated. A representative Western blot of phospho-ERK1/2 and total ERK1/2 is shown (top) as is quantitation of multiple independent experiments normalized to total ERK1/2 (bottom). ctl, control.
FIG. 12. Mechanism of stretch-induced VEGF mRNA expression in bovine retinal pericytes. Schematic representation of potential mechanism for stretch-induced VEGF mRNA expression as supported by the data presented. Despite stretch-induced activation of all pathways listed, stretch-induced VEGF expression in retinal capillary pericytes is not primarily mediated by ERK1/2, Akt, or Ras but rather involves PI 3-kinase-mediated activation of PKC-.

PKC-and PI 3-Kinase Mediate Stretch-induced VEGF Expression 1055
sure of the kinase domain (41). PDGF receptor can be activated by stretch independently of its ligand. Our data demonstrating stretch increases PDGFR-B tyrosine phosphorylation and subsequent p85 association suggests that such a response may mediate stretch-induced activation of PI 3-kinase. It is as yet unknown whether such stretch-induced receptor activation can mediate VEGF expression. Since mechanical stretch can regulate gene expression in a variety of ways (90,91) and since hypertension increases retinal arterial diameter up to 35% (31,92,93), it is possible that hypertension-induced stretch in vivo may increase VEGF expression enough to exacerbate ocular conditions characterized by endothelial proliferation and leakage such as diabetic retinopathy. Indeed, retinal expression of VEGF and VEGF-R2 are increased in spontaneously hypertensive rats (36). Although the magnitude of stretch experienced by the vasculature is likely to diminish as the internal capillary diameter becomes smaller (94), our studies did not identify a maximal VEGF mRNA accumulation as expression continued to increase after all durations of cardiac profile cyclic stretch. Thus, it is possible that even very small increases in cyclic stretch could eventually result in significantly increased VEGF expression.
This finding may also be important as retinal pericytes are characteristically lost early in the course of diabetic retinopathy (75,95). Thus, even with diminishing numbers, significant localized VEGF expression may be present. Retinal pericytes are an important cell type especially in early stages of retinopathy as they regulate retinal vascular tone and perfusion (94), mediate diabetes-induced alterations in autoregulation of retinal blood flow and vasoreactivity (83), and produce VEGF (19). In addition, retinal endothelial cells, which are not compromised until later stages of diabetic retinopathy, respond to stretch with very similar expression of VEGF as do pericytes (36). The applicability of these signaling pathways to other cell types remains to be determined.
In summary, we demonstrate that cardiac profile cyclic stretch induces VEGF expression via PI 3-kinase-mediated activation of PKC-. Furthermore, stretch-induced VEGF expression is independent of ERK1/2, Ras, classical/novel isoforms of PKC, and Akt despite stretch-induced activation of these molecules. In addition, PKC-activation does not mediate ERK1/2 activation. Since each of these molecules has been implicated as mediators of VEGF expression in response to other perturbations, these data suggest that a variety of pathways may be involved in mediating increased VEGF expression in response to diverse stimuli in various cell types. Furthermore, these studies identify new therapeutic targets with potential to ameliorate the well documented clinical exacerbation of ocular diseases, such as diabetic retinopathy, by concomitant hypertension.